Handbook of Ceramic Hard Materials

Ralf Riedel (Editor)
Handbook of Ceramic Hard Materials
@WILEY*VCH
Downloaded from https://onlinelibrary.wiley.com/doi/ by Technische Universität Chemnitz, Wiley Online Library on [14/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Related titles from WILEY-VCH
Structure and Properties of Ceramics
J. Bill, F. Wakai, F. Aldinger

Part 11: ISBN 3-527-29356-6
Precursor-Derived Ceramics
Part I: ISBN 3-527-26830-8
Processing of Ceramics
ISBN 3-527-29814-2
ISBN 3-527-26824-3
R. J. Brook (Ed.)
M. Swain (Ed.)
Ceramic Hard Materials
Weinheim . New York . Chichester . Brisbane . Singapore . Toronto

Handbook of
@WILEY-VCH
Ralf Riedel (Editor)
Editor:
Prof. Dr. Ralf Riedel
Fachgebiet Disperse Feststoffe
Fachbereich Materialwissenschaft
Technische Universitit Darmstadt
PetersenstraRe 23
64287 Darmstadt
Germany
This book was carefully produced. Nevertheless, authors, editor and publisher do not warrant the
information contained therein to be free of errorb. Readers are advised to keep in mind that state-
ments, data, illustrations, procedural details or other items may inadvertently be inaccurate.
Library of Congress Card No. Applied for
A catalogue record for this book is available from the British Library
Deutschc Bibliothek Cataloguing-in-Publication Data:

A catalogue record for this publication is available from Die Deutschc Bibliothek
ISBN 3-527-29912-6
C WILEY-VCH Verlag GmbH, D-69469 Weinheiin (Federal Republic of Germany), 2000

Printed on acid-free and chlorine-free paper
All rights reserved (including those of translation in other languages). N o part of this book may be repro-
duced in any form by photoprinting, microfilm, or any other means nor transmitted or translated into
~ ~
machine language without written permission from the publishers. Registered names, trademarks, etc. used
in this book, even when not specifically maked as such, are not to be considered unprotected by law.
Composition: Alden Bookset, Oxford
Printing: betz-druck, Darmstadt
Bookbinding: Buchbinderei Osswald, NeustadtiWstr.
Printed in the Federal Republic of Germany
Ute, Vincent, Lorenz and Marlene
This book is dedicate to
Preface
With increasing demand for improved efficiency of engines, plants and production
processes, ceramics have gained great importance as structural engineering
materials in recent years. Within the group of the so called advanced materials,
carbon in form of diamond or diamond-like structures, carbides, nitrides and bor-
ides have reached an outstanding position due to their excellent hardness and
thermo-chemical and thermo-mechanical properties. The distinct covalent bonding
of the aforementioned structures positively influences their hardness and their tribo-
logical behavior. Moreover, a series of oxides such as stishovite, a high pressure
modification of silica, or boron sub-oxides have been recently discovered to exhibit
high hardness apart from the well known alumina.
There is presently much effort in basic science and applied research to work on
novel ceramic hard materials denoted as super- or ultra-hard materials that can
compete with the hardness of conventional diamond. Aim and scope of the research
in this field is to develop hard materials with superior mechanical and chemical
properties and with similar hardness. Moreover, calculations of properties of
hypothetical carbon nitrides like C3N4 indicated that there might be compounds
exhibiting even higher hardness values than that of diamond. The low-temperature
synthesis of diamond and cubic boron nitride on the one hand as well as the success-
ful research on new carbon nitrides on the other hand have caused an enormous
impact around the world on both the basic science and the technological develop-
ment of these novel ultra-hard materials.
With the present book we wish to review comprehensively and concisely the state
of the art concerning the structure, synthesis, processing, properties and applica-
tions of ceramic hard materials in general. In particular, the synthesis, modeling
and properties of novel hard materials like binary carbon nitrides, ternary boron
carbonitrides and others are also addressed. It is the aim of this reference book
not only to reflect the state of the art and to give a sound review of the literature,
but to delineate the underlying concepts and bearing of this interdisciplinary field.
With the present edition we wish to show that the field of hard materials research
and development has to be recognized into the wider context of chemistry, physics
as well as materials science and engineering.
The book is organized in two volumes and three parts, covering the structure
and properties of ceramic hard materials (Volume 1, Part I), synthesis and pro-
cessing (Volume 1, Part 11) as well as the typical fields of applications (Volume 2,
Part 111).
Volume 1 starts with an introduction into novel ultra hard ceramics including
diamond and diamond-like carbon, carbon nitrides and silicon nitrides as well as
boron containing carbides, nitrides and carbonitrides. Here we wish to recognize
the great fundamental and technological challenge of developing new superhard
VIII Preface
materials which can compete with the hardest counterparts such as diamond and
cubic boron nitride.
In dealing with properties, the first Chapter in Part I is then devoted to the struc-
ture of crystalline and amorphous ceramic hard materials. The structural features
are responsible in particular for the intrinsic materials properties such as melting
point and hardness. It has been found that in many cases the hardness of a crystal-
line substance correlates with its melting point. Therefore, detailed knowledge of the
3dimensional arrangement of the atoms is required to understand the materials
behavior under certain conditions. More details of the individual crystal structures
with respect to a 3dimensional view can be found on our hard materials homepage
under the web address ‘www.hardmaterials.de’. Phase transitions and materials
synthesis under high pressure in laser heated diamond cells is the topic of the
continuing Chapter. The materials behavior under high pressure and temperature
is of fundamental interest for the synthesis of hard materials since many of the
ultra-hard substances like diamond, cubic boron nitride or stishovite are formed
naturally or synthetically under these harsh conditions. The next three Chapters
are concerned with the mechanical behavior and corrosion of ceramic hard
materials and their relation to microstructure. This correlation is an important
feature since hardness is not only governed by the intrinsic atomic structure of
the respective material but also to a great extend by its polycrystalline nature. There-
fore, the grain morphology and grain boundary chemistry play a decisive role in the
materials response under environmental or mechanical load. In the following
Chapter transition metal carbides, nitrides and carbonitrides are discussed with a
focus on their structure and bonding, thermodynamic behavior as well as on their
physical and mechanical properties. Part I is then completed by two Chapters
which deal with the theoretical design of novel sp2-bonded carbon allotropes and
novel superhard materials based on carbon and silicon nitrides. These Chapters
tribute to the fact that with proceeding computerization the number of calculated
novel solid structures that led to the prediction of new materials with hardness
comparable to or exceeding that of diamond has increased enormously in recent
years.
Part I1 continues with the synthesis and processing of ceramic hard materials.
Since the conventional powder technological synthesis and processing of ceramics
has been treated in a large number of published review articles here we concentrate
on novel synthetic routes that provide ceramic hard materials. Consequently, six
Chapters report on i) directed metal oxidation, ii) self-propagating high temperature
synthesis, iii) hydrothermal synthesis of diamond, chemical vapor deposition of
diamond (iv) and cubic boron nitride (v) films and finally vi) the polymer to ceramic
transformation. All these processes are particularly suitable for the formation of
refractories with high hardness. Part I1 is then closed by a Chapter on nano
structured superhard materials. In the course of this work high hardness is achieved
by microstructural control rather than by the synthesis of a distinct crystal structure.
In Volume 2 ceramic hard materials are highlighted in the light of their applica-
tions. Chapter 1 of Part I11 concisely reviews the history of diamond and diamond-
like super abrasive tools while Chapter 2 and 3 are concerned with the application of
chemical vapor deposited diamond and diamond-like carbon films. These sections
Prejace IX
include the synthesis of optical grade CVD diamond windows and discuss their
physical and mechanical properties. The most important and wide-spread ceramic
hard materials are based on alumina. Chapter 4 reports on the processing develop-
ments to increase the hardness of alumina based ceramics for grinding and cutting
applications. Silicon carbide and silicon nitride materials are the most technologi-
cally important non-oxide compounds and have gained great significance in the
field of cutting ceramics and are treated in Chapters 5 and 6. Boron-based ceramics
are a further group of either established or candidate materials with extreme hard-
ness. Therefore, Chapter 7 deals with boron carbide or transition metal borides like
titanium diboride and their distinct properties and applications. In Chapter 8, clas-
sical hard metals comprised of tungsten carbide as the hard phase and cobalt as the
binder phase are discussed. Volume 2 is finally completed by a data base (Chapter 9)
containing approximately 130 hard materials including carbides, nitrides, borides,
silicides and oxides. The data base references the crystal structure, physical proper-
ties like melting point and density, mechanical properties (Youngs modulus, micro
hardness) and oxidation resistance of the respective compounds. Future develop-
ments of novel hard materials such as the recently discovered intermetallic phase
A1MgBl4will be updated on our internet homepage ‘www.hardmaterials.de’.
In closing these introductory remarks, I would like to emphasize that the special
chance to place a summary of the outstanding expertise on the field of present hard
materials research and development would not have been possible without the great
enthusiasm and commitment of all the colleagues who contributed in the writing of
this two volume set. I am grateful for their enormous efforts in compiling a fascinat-
ing series of articles imparting depth insight into the individual fields of modern hard
materials research. Finally, I wish to thank the Wiley-VCH Editors Peter Gregory
and Jorn Ritterbusch for encouraging me in the preparation of this book and for
their continuous support throughout the editorial process.
Ralf Riedel
March 2000
Darmstadt
Foreword
One of the clearest hierarchies in materials science and engineering is provided by

the property of hardness. There are, of course, many properties where remarkable
differences exist between groups of materials. An example is provided by electrical
conductivity where a ratio of 10l8can be readily found; with electrical conductivity,
however, the different materials do not come into direct competitive opposition. In
the case of hardness, the very value of this property lies in the ability of one material
to demonstrate a higher place in the hierarchy than another; the one material is used
in effect to overpower the other.
The existence of this hierarchy, which has been long recognised in the traditional
measurement scale for the property, has direct relation to applications. In any use of
materials it is important to be able to shape them to be fit for purpose; where the
shaping process involves some type of machining, as it most commonly does,
then the property of hardness becomes the unambiguous figure of merit.
It is for these reasons that there has been long standing and productive interest in
hard materials, in their design, in their fabrication, in their use, and in the underlying
science and engineering. It is thorougLlv in keeping with this tradition of research
relevant to application that the present book brings together a set of authoritative
reviews of the progress which has been made.
The organisation of the book is a direct reflection of the logic which has been used
in developing hard materials. One of the great attractions of the subject has been the
close link that exists between hardness on the one hand and the bonding and struc-
ture of the material on the other. The link between these two has proved to be one of
the best foundations on which to base materials development. The link is a central
theme in the first part of the book where fine examples are given of the rich contri-
bution which has been made and which continues to be made by fundamental
studies of bonding and structure to materials performance.
It has long been recognised that the very aspect of their extreme resistance to
deformation would make it a particular challenge to manufacture hard materials
in reliable and cost-effective ways. It is here that the materials community has
shown itself to be imaginative and forward looking in seeking innovative
fabrication routes. These are well presented in the second part of the book
where specific attention is given to the paths which can be used to assemble
materials of precisely defined form without sacrifice of their characteristic mechan-
ical resilience.
The most striking aspect of hard materials, however, is the direct link to applica-
tions. This link has brought an unusual degree of purpose to materials development
which has enjoyed the benefits of being conducted in full recognition of the target to
be reached. It has also meant that the progress made in research can be rapidly
evaluated since the testing procedures relate so directly to the end use. The third
XI1 Foreword
part of the book accordingly gives close accounts of the performance of the different
classes of hard materials in the applications context.
The contributors to this text are to be congratulated on bringing their many dis-
ciplines to bear on this central theme. Materials science is well known to undergo
fashions as materials are developed and discarded and indeed as sectors of applica-
tion grow and decline. The one requirement which will remain is that the forming
and shaping of materials will always be necessary whatever the eventual sector of
application. We can accordingly be confident that the long history of hardness
studies, not least in the last two hundred years from the carbon tool steels, to
high speed steels, to stellite, to tungsten carbide, to cermets, to ceramics, and now
to diamond, boron, nitride and other special systems, will be continued with
informed imagination and with creative innovation. The present book is a splendid
platform on which to base such future development.
Richard Brook
January 2000
Oxford, UK
Contents
List of Contributors XXVII

List of Symbols XXXIII
List of Abbreviations XXXIX
Introduction: Novel Ultrahard Materials

A . Zerr and R. Riedel
Introduction XLV
Hard Materials XLVI
Hardness XLVII
Carbon-based Hard Materials L
Diamond LII
Diamond-like and Amorphous Carbon LV
Novel Hypothetical Three-dimensional Carbon Phases LVI
Fullerenes LIX
Carbon Nitride (C3N4) LIX
Boron-based Hard Materials LXIV
Boron Nitrides LXIV
Boron-rich Boron Nitrides LXVII
Nitrogen-rich Boron Nitride LXVIII
Boron Carbonitrides (B,C,N,) LXVIII
Boron Suboxides LXXI .
Silicon-based Materials LXXI
Concluding Remarks LXXII
Acknowledgement LXXIII
References LXXIII
Part I Structures and Properties
1 Structural Chemistry of Hard Materials

W . Jeitschko, R. Pottgen, and R.-D.
Hoffmann
1.1 Introduction 3
1.2 Diamond and Diamond-Related Structures 5
1.2.1 The Crystal Structure of Diamond 5
1.2.2 The Isoelectronic Compounds c-BN and S i c 6
1.3 Crystal Chemistry of Borides and Boron Carbides 8
1.4 The Structures of Transition Metal Carbides 12
1.5 Silicides and Silicide Carbides of Transition Metals 20
XIV Contents
1.6 Nitrides 23
1.6.1 Nitrides of Main Group Elements 24
1.6.2 Transition Metal Nitrides 25
1.6.3 Perspectives: Nitridosilicates 29
1.7 Oxide Ceramics 30
1.7.1 Hard Ceramics of Main Group Elements 30
1.7.2 Transition Metal Oxides 32
1.8 Amorphous Hard Materials 36
References 37
2 Phase Transitions and Material Synthesis using the C02-Laser

Heating Technique in a Diamond Cell
A . Zerr, G. Serghiou, and R . Boehler
2.1 Introduction 41
2.2 Technique of C02-Laser Heating in a Diamond Anvil Cell 42
2.2.1 Sample Assemblage in a Diamond Anvil Cell 42
2.2.2 Pressure Conditions in the Sample Volume 43
2.2.3 Experimental Set-up for C02-Laser Heating in a Diamond Anvil
Cell 44
2.2.4 Temperature Determination 45
2.2.5 Temperature Stabilization 45
2.2.6 Radial Temperature Gradients 48
2.2.7 Raman and Fluorescence Spectroscopic Analysis of Samples in a
Diamond Anvil Cell 48
2.3 Determination of Melting Temperatures at High Pressures 49
2.3.1 Melting of Cubic BN at 10 GPa 49
2.3.2 Melting Temperatures of Materials Relevant to the Earth’s Lower
Mantle 51
2.4 Phase Diagrams, Decomposition Reactions, and Stability of Solids at
High Pressures and Temperatures 54
2.4.1 Coesite-Stishovite Phase Boundary 55
2.4.2 High Pressure and Temperature Phase Diagram and Decomposition
Reactions in a Ternary System 56
2.4.3 Stability of a Perovskite Oxide with Respect to its Component
Oxides 59
2.5 C02-laser Heating Experiments on Organic Compounds 60
2.6 Conclusion 62
Acknowledgments 62
References 62
3 Mechanical Properties and their Relation to Microstructure

D. Sherman and D . Brandon
3.1 Introduction 66
3.1.1 Applications and Engineering Requirements 66
Contents XV
3.1.2 Bulk Components 68

3.1.3 Coatings 70
3.1.4 Engineering Requirements 70
3.2 Principal Mechanical Properties 7 1
3.2.1 Elastic Modulus 71
3.2.2 Strength 72
3.2.3 Fracture Toughness 74
3.2.4 Hardness 79
3.3 Mechanical Testing of Hard Materials 81
3.3.2 Fracture Strength 8 1
3.3.3 Fracture Toughness 83
3.3.4 Hardness 84
3.3.5 Indentation Toughness 86
3.3.6 Erosion, Wear and Scratch Tests 89
3.4 Microstructural Parameters and Mechanical Properties 9 1
3.5 Failure Mechanisms 94
3.5.1 Creep Behavior 94
3.5.2 Mechanical Fatigue 95
3.5.3 Ballistic Properties 97
3.6 Conclusions 98
References 99
4 Nanostructured Superhard Materials

S. Veptek
4.1 Introduction 104
4.2 Concept for the Design of Superhard Materials 109
4.2.1 Nanocrystalline Materials 110
4.2.2 Heterostructures 114
4.3 Preparation and Properties of Superhard Nanocrystalline
Composites 116
4.3.1 Preparation 116
4.3.2 Properties of the ncM,N/aSi3N4 Composites 119
4.3.3 Other Superhard Nanocomposites and the General Validity of the
Design Principle 124
4.4 Discussion of the Possible Origin of the Hardness and Stability of the
Nanostructure 128
4.4 Conclusions 133
Acknowledgments 134
References 134
5 Corrosion of Hard Materials

K . G. Nickel and Y. G. Gogotsi
XVI Contents
5.2 Corrosive Media 140

5.3 Corrosion Modes 141
5.3.1 Active and Passive Corrosion 141
5.3.2 Homogeneity and Location of Attack: Internal, External and
Localized Corrosion 141
5.4 Corrosion Kinetics 142
5.4.1 Physical Boundary Conditions 142
5.4.2 Active Corrosion Kinetics 143
5.4.3 Basic Passive Corrosion Kinetics 145
5.4.4 Kinetic Breaks 147
5.4.5 Complex Kinetics 148
5.5 Corrosion Measurement 150
5.5.1 Experimental Methods 150
5.5.2 Corrosion Data 151
5.6 Materials 154
5.6.1 Diamond and Diamond-like Carbons 154
5.6.2 Carbides 155
5.6.3 Nitrides 166
5.6.4 Carbonitrides 173
5.6.5 Titanium Diboride 176
References 177
6 Interrelations Between the Influences of Indentation Size, Surface

State, Grain Size, Grain-Boundary Deformation, and Temperature on
the Hardness of Ceramics
A . Krell
6.2 The Assessment of Residual Porosity and Flaw Populations: A
Prerequisite for any Hardness Investigation 184
6.3 Theoretical Considerations 185
6.3.1 The Role of the Lattice and of Grain Boundaries in the Inelastic
Deformation at an Indentation Site in Sintered Hard
Materials 185
6.3.2 Quantitative Understanding the Load Effect on the Hardness:
Theoretical Considerations Compared with Single Crystal Data 188
6.4 Influences of the Grain Size and the State of the Surface 191
6.4.1 The Grain Size Influence on the Load Effect of the Hardness:
Modeling Experimental Results 191
6.4.2 The Effect of the Grain Size and the Surface State in Ceramics when
Recorded by Different Measuring Approaches 193
6.5 Comparing the Grain Size Effect and the Indentation Size Effect: The
Role of Grain Boundaries at Room Temperature 195
6.6 The Effects of Temperature on the Hardness of Ceramics 198
6.7 Summary 199
References 20 1
Contents XVII
7 Transition Metal Carbides, Nitrides, and Carbonitrides

W. Lengauer
7.2 General Features of Structure and Bonding 205
7.2.1 General Structural Features 205
7.2.2 General Features of Bonding 206
7.3 Preparation 207
7.4 Characterization 2 10
7.4.1 Chemical Analysis 2 10
7.4.2 Physical Microanalysis 21 1
7.5 Thermodynamics 2 12
7.5.1 Stability of Carbides 212
7.5.2 Nitrogen Partial Pressure of Nitrides 212
7.5.3 Phase Equilibria of Important Carbide Systems 213
7.5.4 Transition Metal-Nitrogen Systems and Structure of Phases 216
7.5.5 Carbonitride Systems 221
7.6 Properties of Important Transition Metal Carbides, Nitrides, and
Carbonitrides 224
7.6.1 Melting Points 224
7.6.2 Color 224
7.6.3 Thermal and Electrical Conductivities 225
7.6.4 Thermal Expansion 228
7.6.5 Diffusivities 229
7.6.6 Elastic Properties 23 1
7.6.7 Microhardness 234
7.7 Industrial Applications 238
7.7.1 Cemented Carbides and Carbonitrides 238
7.7.2 Deposited Layers 241
7.7.3 Diffusion Layers 246
Acknowledgments 248
References 248
8 New Superhard Materials: Carbon and Silicon Nitrides

J . E. Lowther
8.2 Modeling Procedures 254
8.2. I Semi-empirical Approaches 254
8.2.2 Tight-binding Schemes 255
8.2.3 Ab initio Pseudopotential Approach 256
8.2.4 Transition Pressures and Relative Stability 256
8.3 Carbon Nitride 257
8.3.1 Crystalline Structures 258
8.3.2 Graphitic Structures 259
8.3.3 Amorphous Structures 261
XVIII Contents
8.3.4 Relative Stability 263

8.4 Silicon Carbon Nitride 264
8.4.1 j3SiC2N4 265
8.4.2 Near-cubic Forms of SiC2N4 266
8.4.3 Relative Stability 268
8.5 Conclusions 268
Acknowledgements 269
References 269
9 Effective Doping in Novel sp2 Bonded Carbon Allotropes

G. Jungnickel, P. K. Sitch, T. Frauenheim, C. R. Cousins, C. D. Latham,
B. R. Eggen, and M . I. Heggie
9.2 Lattice Description 274
9.3 Computational Methods 276
9.4 Static Properties 278
9.5 Electronic Properties 279
9.6 Conclusions 282
Acknowledgments 283
References 283
Part I1 Synthesis and Processing
1 Directed Metal Oxidation

V. Jayaram and D. Brandon
1.1 Historical Background 289
1.2 Oxidation and Oxide Formation 290
1.2.I Initial Oxidation 291
1.3 Related Ceramic Processing Routes 293
I .4 Directed Metal Oxidation Incubation 295
1.5 Directed Metal Oxidation Growth 300
1.5.1 Introduction 300
1.5.2 Directed Metal Oxidation Composites from Al-Mg Alloys 300
1.5.3 Directed Metal Oxidation Growth from other Aluminum Alloys 304
1.5.4 Microstructural Scale 305
1.5.5 Growth into Particulate Preforms 307
1.5.6 Growth into Fibrous Preforms 309
1.6 Mechanical Properties 310
1.6.2 Strength and Toughness 3 11
1.6.3 Thermal Shock 313
1.6.4 High Temperature Strength 313
1.6.5 Wear Properties 314
1.6.6 Mechanical Properties of Fiber-reinforced DMO Composites 3 14
Contents XIX
1.7 Corrosion of Directed Metal Oxidation Composites 3 16

1.8 Other Properties 316
1.9 Applications 3 I6
1.9.1 Wear Resistant Components 317
1.9.2 Ceramic Composite Armor 3 17
1.9.3 Thermal Barriers and Heat Sinks 318
References 3 18
2 Self-propagating High-Temperature Synthesis of Hard Materials

Z . A . Munir and U. Anselmi-Tumburini
2.2 Mechanistic Characterization of the Process 327
2.3 Effect of Experimental Parameters 33 1
2.3 Synthesis of Dense Materials 342
2.4 Synthesis by Field-Activated Self-propagating High-temperature
Synthesis 348
2.6 Selected Recent Examples of Synthesis of Hard Materials 356
Acknowledgment 368
References 368
3 Hydrothermal Synthesis of Diamond

K. G. Nickel, T. Kruft, and Y. G. Gogotsi
3.2 Evidence from Nature 376
3.3 Hydrothermal Synthesis 377
3.3.1 C-H-0 System 377
3.3.2 Hydrothermal Treatment of S i c 382
3.4 Outlook 387
Acknowledgments 387
References 387
4 Chemical Vapor Deposition of Diamond Films

C.-P. Klages
4.2 Preparation Methods for Diamond Films 391
4.2.1 Hot-filament Chemical Vapor Deposition 392
4.2.2 Microwave-plasma-based Methods 397
4.2.3 Preparation of Special Forms: Textured and
Heteroepitaxial Films 400
4.3 Thermochemistry and Mechanism of Chemical Vapor Deposition
Diamond Growth 407
4.3.1 Transformation of Graphite to Diamond at Low Pressures 407
xx Contents
4.3.2 Reactive Species in Diamond Chemical Vapor Deposition, the Role

ofCH3 408
4.4 Properties and Applications of Chemical Vapor Deposited
Diamond 410
4.4.1 Diamond Coated Cutting Tools 41 1
4.4.2 Thermal Conductivity of Chemical Vapor Deposited Diamond:
Thermal Management Applications 412
4.4.3 Electrical Properties and Electronic Applications 413
4.4.4 Electrochemical Use of Chemical Vapor
Deposited Diamond 415
4.5 Summary 417
References 4 17
5 Vapor Phase Deposition of Cubic Boron Nitride Films

K. Bewilogua and F. Richter
5.2 Empirical Results 421
5.2.1 Deposition Methods 421
5.2.2 Morphology and Structure of cBN Films 423
5.2.3 Film Adhesion 427
5.3 Models of cBN Formation 427
5.4 Sputter Deposition of cBN Films 429
5.4.1 Sputter Deposition with Conducting Targets 430
5.4.2 Deposition by d.c. Magnetron Sputter with a Hot Boron
Target 43 1
5.5 Discrimination between Nucleation and Growth Phase 433
5.5.1 Detection of hBNxBN Transition 433
5.5.2 RF Magnetron Sputtering 435
5.6 Properties of cBN Films 440
5.6.1 Mechanical and Tribological Properties 440
5.6.2 Optical Properties 440
5.6.3 Electrical Properties 441
5.6.4 Other Properties 441
5.7 Summary and Outlook 442
References 442
6 Polymer to Ceramic Transformation: Processing of Ceramic Bodies

and Thin Films
G. D. Soraru and P. Colombo
6.2 Processing of Monolithic Components 450
6.3 Preparation and Characterization of SiAlOC Ceramic Bodies by
Pyrolysis in Inert Atmosphere 452
6.3.1 Experimental Procedure 452
Contents XXI
6.4 Results 453

6.4.1 Characterization of the Pre-ceramic Precursors 453
6.4.2 Characterization of the Pre-ceramic Components 454
6.4.3 Characterization of the Ceramic Components 455
6.4.4 Mechanical Characterization at High Temperature 457
6.5 Discussion 458
6.6 Preparation and Characterization of SiAlON Ceramics by Pyrolysis
in Reactive Atmosphere 460
6.6.1 Experimental 460
6.7 Results and Discussion 460
6.8 Processing of Thin Ceramic Films 463
6.9 Experimental 463
6.10.1 Conventional Conversion Process: Annealing in
Controlled Atmosphere 464
6.10.2 Nonconventional Conversion Process: Ion Irradiation 467
6.1 1 Conclusions 472
Acknowledgments 473
References 473
Part 111 Materials and Applications
1 Diamond Materials and their Applications

Edited by R. J . Caveney
1.1 Superabrasive tools: A Brief Introduction 479
1.1.2 Early History 479
1.1.3 Synthetic Diamond 48 1
1.1.4 Cubic Boron Nitride 482
1.1.5 Polycrystalline Diamond and Cubic Boron Nitride 482
1.1.6 Chemical Vapor Deposited Diamond 484
1.1.7 Outline of Chapter 485
1.2 The Crystallization of Diamond 485
1.2.1 The Carbon Phase Diagram 485
1.2.2 Diamond Crystallization at High Pressure 487
1.2.3 High Pressure Apparatus 490
1.2.4 The Synthesis of Particulate Diamond Abrasives 49 1
1.2.5 Growth of Large Synthetic Diamonds 496
1.2.6 Novel Diamond Synthesis Routes 504
1.2.7 Cubic Boron Nitride Crystallization 5 10
1.3 Polycrystalline Diamond and Cubic Boron Nitride 5 12
1.3.1 Natural Polycrystalline Diamond 5 12
1.3.2 Synthetic Polycrystalline Diamond 512
1.3.3 Mechanisms involved in Polycrystalline Diamond Manufacturing
Process 513
XXII Contents
1.3.4 Polycrystalline Cubic Boron Nitride 5 18

1.4 New Ultrahard Materials 521
1.4.2 Hardness 521
1.4.3 C3N4 523
1.4.4 Boron Rich Nitride 526
1.4.5 Boron Carbonitrides 526
1.4.6 Boron Suboxides 526
1.4.7 Stishovite 526
1.5 Industrial Applications of Diamond and cBN 527
1.5.2 Abrasive Application 528
1.5.3 Machining of Stone and Concrete 540
1.5.4 Applications of Polycrystalline Ultra-hard Materials 548
1.5.5 Applications of Single Crystal Diamond 559
Acknowledgments 566
References 566
2 Applications of Diamond Synthesized by Chemical Vapor Deposition

R. S. Sussmann
2.2 Properties of Chemical Vapor Deposited Diamond 574
2.2.1 Material Grades 574
2.2.2 Optical Properties 576
2.2.3 Strength of Chemical Vapor Deposited Diamond 580
2.2.4 The Young Modulus 581
2.2.5 Thermal Conductivity 582
2.2.6 Dielectric Properties 583
2.3 Optical Applications 583
2.3.1 Chemical Vapor Deposited Diamond for Passive Infrared Windows in
Aggressive Environments 584
2.3.2 Windows for High-power Infrared Lasers 589
2.4 Windows for High Power Gyrotron Tubes 597
2.4.1 Window Requirements 598
2.4.2 The Development of Chemical Vapor Deposited Diamond Gyrotron
Windows 599
2.5 Thermal Management of Laser Diode Arrays 606
2.5.1 Laser Diode Arrays: General Issues 607
2.5.2 Modelling of Submount Heat Resistance 607
2.5.3 Flatness of Submount 610
2.5.4 Thermal Stress 610
2.6 Cutting Tools, Dressers and Wear Parts 61 1
2.6.1 Cutting Tools Trends 61 1
2.6.2 Cutting Tool Application of Chemical Vapor Deposited
Diamond 612
Contents XXIII
2.6.3 Chemical Vapor Deposited Diamond Dressers 616

2.6.4 Chemical Vapor Deposited Diamond Wear Parts 617
References 6 19
3 Diamond-like Carbon Films

C.-P. Klages and K . Bewilogua
3.2 Preparation Methods for Diamond-like Carbon Films 623
3.2.1 Hydrogenated Amorphous Carbon (a-C: H) 623
3.2.2. Hydrogen Free Amorphous Carbon (ta-C) 627
3.2.3 Metal-containing Amorphous Hydrocarbon 629
3.3 Microstructure and Bonding of Diamond-like Carbon 630
3.3.1 Amorphous Carbon and Hydrogenated Amorphous Carbon 630
3.3.2 Metal-containing Amorphous Carbon Films 634
3.4. Physical Properties of DLC Films 637
3.4.1 Electrical and Optical Properties 637
3.4.2 Mechanical Properties 639
3.5. Applications of DLC Films 640
3.5.1 Adhesion of DLC Films 640
3.5.2 Tribology of DLC Coatings 642
3.5.3 Tribological Applications 644
3.5.4 Other Applications 644
References 645
4 Ceramics Based on Alumina: Increasing the Hardness for Tool

Applications
A . Krell
4.1 Recent Trends in the Application of Ceramic Tool Materials 648
4.2 Technological Essentials for Producing Hard and Strong Tool
Ceramics 650
4.2.1 Typical Defects in Ceramics Tool Materials: The State of The
Art 651
4.2.2 Recent Trends in Ceramic Technologies Related to
Tool Ceramics 653
4.3 Tool Materials with Undefined Cutting Edge: Sintered Grinding
Materials 658
4.3.1 Technical Demands for Grinding Materials 660
4.3.2 Advanced Commercial Products: Sol/gel-derived Corundum 661
4.3.3 Sintered Alumina Grits Produced by Powder Processing
Approaches 665
4.4 New Trend for Cutting Hard Workpieces: Submicrometer Cutting
Ceramics for Tools with Defined Cutting Edge 666
4.4.1 Demands for Cutting Materials Used for Turning
Hard Workpieces 667
XXIV Contents
4.4.2 Carbide Reinforced Composite Ceramics Based on A1203 669

4.4.3 Single Phase Sintered Corundum 670
4.4.4 Comparative Cutting Studies with Submicrometer Ceramics: A1203
and Composites Reinforced with Ti(C,N) and Ti(C,O) 670
4.5 Summary 680
References 68 1
5 Silicon Carbide Based Hard Materials

K. A . Schwetz
5.1 lntroduction 683
5.1.1 History 683
5.1.2 Natural Occurrence [7] 684
5.2 Structure and Phase Relations of S i c 685
5.3 Production of S i c 688
5.3.1 The Acheson/ESK Process 688
5.3.2 Other Production Methods 69 1
5.3.3 Dense S i c Shapes 699
5.4 Properties of Silicon Carbide 719
5.4.1 Physical Properties 719
5.4.2 Chemical Properties 720
5.4.3 Tribological Properties 723
5.5 Quality Control 734
5.6 Toxicology and Occupational Health 736
5.7 Uses of Silicon Carbide 736
Acknowledgments 740
References 740
6 Silicon Nitride Based Hard Materials

M . Herrmann, H. Klemm, Chr. Schubert
6.2 Crystal Structure and Properties of the Si3N4Modifications 753
6.3 Densification 755
6.4 Microstructural Development 758
6.4.1 Microstructural development of P-Si3N4materials 758
6.4.2 Microstructural development of a’-SiALON materials 768
6.5 Properties of Si3N4Materials 771
6.5.1 Mechanical properties at room temperatures 771
6.5.2 High-temperature properties of silicon nitride materials 777
6.5.3 Wear resistance of Si3N4materials 782
6.5.4 Corrosion resistance of Si3N4 786
6.6 Conclusions/Further potential of silicon nitride materials 792
Acknowledgements 795
References 795
Contents xxv
7 Boride-Based Hard Materials
R. Telle, L. S . Sigl, and K. Takagi
7.2 Chemical Bonding and Crystal Chemistry of Borides 803
7.2.1 Chemical Bonding of Borides 803
7.2.2 The Crystal Structure of Borides 804
7.3 Phase Systems 812
7.3.1 Binary Phase Diagrams of Technically Important Systems 813
7.3.2 Ternary and Higher Order Systems 818
7.4 Boron Carbide Ceramics 837
7.4.1 Preparation of Boron Carbide 837
7.4.2 Sintering of Boron Carbide 839
7.4.3 Properties of Boron Carbide 851
7.4.4 Chemical Properties and Oxidation of Boron Carbide 855
7.4.5 Boron Carbide-Based Composites 857
7.5 Transition Metal Boride Ceramics 874
7.5.1 Preparation of Transition Metal Borides 875
7.5.2 Densification of Transition Metal Borides 876
7.5.3 Properties of Transition Metal Borides Ceramics 878
7.6 Multiphase Hard Materials Based on Carbide-Nitride-Boride-Silicide
Composites 888
7.7 Boride-Zirconia Composites 888
7.8 Cemented Borides 895
7.8.1 Boron Carbide-Based Cermets 895
7.8.2 Titanium Diboride-Based Cermets 897
7.8.3 Cemented Ternary Borides 919
7.8.4 Potentials and Applications 927
7.9 Future Prospects and Fields of Application 933
References 936
8 The Hardness of Tungsten Carbidecobalt Hardmetal 946

S . Luyckx 946
8.2 The Hardness of the Two Component Phases 947
8.2.1 The Hardness of Tungsten Carbide 947
8.2.2 The Hardness of Cobalt 948
8.3 Factors Affecting the Hardness of WC-Co Hardmetal 950
8.3.1 Cobalt Content and Tungsten Carbide Grain Size 950
8.3.2 Grain Size Distribution and Cobalt Mean Free Path 952
8.3.3 Binder Composition and Carbon Content 952
8.3.4 Porosity 953
8.3.5 Effect of Temperature 953
8.4 Relationships between Hardness and Other Hardmetal
Properties 960
XXVI Con ten /s
8.4.1 Relationship between Hardness and Toughness 962

8.4.2 Relationship between Hardness and Abrasive Wear Resistance 962
8.5 Conclusions 963
Acknowledgments 963
References 964
9 Data Collection of Properties of Hard Materials

G. Berg, C. Fviedrich, E. Broszeit, and C. Bevger
9.2 Profile of Properties 965
9.3 Organization and Contents of the Data Collection 966
Acknowledgement 967
Refercnces 99 1
Index 997
List of Contributors
U. Anselmi-Tamburini R. Bohler
Dipartimento di Chimica Fisica Max-Planck-Institute for Chemistry
Universita di Pavia Saarstrasse 23
27100 Pavia D-55020 Mainz
Italy Germany
M. W. Bailey D. Brandon
De Beers Industrial Diamond Department of Materials Engineering
Division Pty Ltd Technion - Israel Institute of
Diamond Research Lab Technology
PO Box 1770 Haifa 32000
Southdale 21 35 Israel
South Africa
J. R. Brandon
G. Berg De Beers Industrial Diamond
Fachgebiet und Institut fur Division Pty Ltd
Werkstofiunde der TU Darmstadt Diamond Research Lab
und Staatliche PO Box 1770
Materialpru fungsanhalt Southdale 2 135
Grafenstrasse 2 South Africa
D-64283 Darmstadt
Germany E. Broszeit
Fachgebiet und Institut fur
Werkstofiunde der TU Darmstadt
C. Berger und Staatliche
Fachgebiet und Institut fur Materialprufungsanhalt
Werkstofiunde der TU Darmstadt Grafenstrasse 2
und Staatliche D-64283 Darmstadt
Materialpriifungsanhalt Germany
Grafenstrasse 2
D-64283 Darmstadt R. C. Burns
Germany De Beers Industrial Diamond
Division Pty Ltd
K. Bewilogua Diamond Research Lab
Fraunhofer Institut fur Schicht und PO Box 1770
Oberflachentechnik (IST)
Southdale 21 35
Bienroder Weg 54 E South Africa
D-38 108 Braunschweig
Germany
XXVIII List of Contributors
R. J. Caveney G. J. Davies
De Beers Industrial Diamond De Beers Industrial Diamond
Division Pty Ltd Division Pty Ltd
Diamond Research Lab Diamond Research Lab
PO Box 1770 PO Box 1770
Southdale 2135 Southdale 2135
South Africa South Africa
S. E. Coe B. R. Eggen
De Beers Industrial Diamond School of Chemistry, Physics and
Division Pty Ltd Environmental Sciences
Diamond Research Lab University of Sussex
PO Box 1770 Falmer
Southdale 2135 Brighton BNl 9QJ
South Africa UK
J. L. Collins D. Fister
De Beers Industrial Diamond HC Starck Gmbh
Division Pty Ltd Kraftwerkweg 3
Diamond Research Lab D-79725 Laufenburg
PO Box 1770 Germany
Southdale 2135
South Africa T. Frauenheim
Fachbereich Physik
P. Colombo Universitat/Gesamthochschule
Universita di Bologna Paderborn
Dipartimento di Chimica Applicata e D- 33095 Paderborn
Scienza dei Materiali Germany
viale Risorgimento 2
1-40I36 Bologna C. Friedrich
Italy Fachgebiet und Institut fur
Werkstoffkunde der TU Darmstadt
M. W. Cook und Staatliche
De Beers Industrial Diamond Materialprufungsanhalt
Division Pty Ltd Grafenstrasse 2
Diamond Research Lab D-64283 Darmstadt
PO Box 1770 Germany
Southdale 21 35
South Africa Y. G. Gogotsi
Institut fur Angewandte Mineralogie
C. R. Cousins Universitat Tubingen
Department of Physics Wilhelmstrasse 56
University of Exeter D-72074 Tubingen
Stocker Road Germany
Exeter EX4 4QL
UK
List of Contributors XXIX
J. 0. Hansen G. Jungnickel
De Beers Industrial Diamond Fachbereich Physik
Division Pty Ltd Universitat/Gesamthochschule
Diamond Research Lab Paderborn
PO Box 1770 D- 33095 Paderborn
Southdale 2 135 Germany
South Africa
C. P. Klages
M. Hoffmann Fraunhofer Institut fur Schicht und
Fakultat fur Maschinenbau Oberflachentechnik (IST)
Institut fur Werkstoffkunde I1 Bienroder Weg 54 E
Universitat Karlsruhe D-38108 Braunschweig
Kaiserstrasse 12 Germany
Postfach 6980
D-76 128 Karlsruhe T. Kraft
Germany lnstitut fur Angewandte Mineralogie
Universitat Tubingen
R. D. Hoffmann Wilhelmstrasse 56
Westfalische Wilhelms-Universitat D-72074 Tubingen
Miinster Germany
Anorganisch-Chemisches Institut
Wilhelm-Klemm-Strasse 8 A. Krell
D-48 149 Miinster Fraunhofer Institute for Ceramic
Germany Technologies and Sintered Materials
Winterbergstrasse 28
M. I. Heggie D-0 1277 Dresden
School of Chemistry, Physics and Germany
Environmental Sciences
University of Sussex C. D. Latham
Falmer Department of Physics
Brighton BN1 9QJ University of Exeter
UK Stocker Road
Exeter EX4 4QL
V. Jayaram UK
Department of Metallurgy
Indian Institute of Science W. Lengauer
Bangalore Institute for Chemical Technology of
India Inorganic Materials
Vienna University of Technology
W. Jeitschko Getreidemarkt 9/161
Westfalische Wilhelms-Universitat A-1060 Vienna
Miinster Austria
Anorganisch-Chemisches Institut
Wilhelm-Klemm-Strasse 8
D-48149 Miinster
Germany
XXX List of Contributors
J. E. Lowther R. Pottgen
Department of Physics Westfalische Wilhelms-Universitat
University of Witwatersrand Munster
Johannesburg Anorganisch-Chemisches Institut
South Africa Wilhelm-Klemm-Strasse 8
D-48149 Munster
S. Luyckx Germany
School of Process and Materials
Engineering F. Richter
University of the Witwatersrand Technische Universitat Chemnitz-
Johannesburg 2050 Zwickau
South Africa Institut fur Physik
D-09107 Chemnitz
Z. A. Munir Germany
Facility for Advanced Combustion
Synthesis R. Riedel
Department of Chemical Engineering Fachbereich Materialwissenschaft
and Materials Science Technical University of Darmstadt
University of California Petersenstrasse 23
Davis CA 95616 D-64287 Darmstadt
USA Germany
K. G. Nickel K. A. Schwetz
Universitat Tubingen Advanced Ceramics Lab
Applied Mineralogy Elektroschmelzwerk Gmbh
Wilhelmstrasse 56 Max-Schaidhauf-Strasse 25
D-72074 Tubingen D-87437 Kempten
Germany Germany
S. Ozbayraktar P. K. Sen
De Beers Industrial Diamond De Beers Industrial Diamond
Division Pty Ltd Division Pty Ltd
Diamond Research Lab Diamond Research Lab
PO Box 1770 PO Box 1770
Southdale 2135 Southdale 2135
South Africa South Africa
C. S. J. Pickles G. Serghiou
De Beers Industrial Diamond Max-Planck-Institute for Chemistry
Division Pty Ltd Saarstrasse 23
Diamond Research Lab Mainz
PO Box 1770 Germany
Southdale 2135
South Africa
List of Contributors XXXI
D. Sherman K. Takagi
Department of Materials Engineering Toyo Kohan Co. Ltd.
Technion - Israel Institute of Tokyo
Technology Japan
Haifa 32000
Israel R. Telle
Institut fur Gesteinshuttenkunde
M. Sibanda RWTH Aaachen
De Beers Industrial Diamond MauerstraBe 5
Division Pty Ltd D-52056 Aachen
Diamond Research Lab Germany
PO Box 1770
Southdale 2135 S. Veprek
South Africa Institute for Chemistry of Inorganic
Materials
I. Sigalas Technical University Munich
De Beers Industrial Diamond Lichtenbergstrasse 4
Division Pty Ltd D-85747 Garching b. Munich
Diamond Research Lab Germany
PO Box 1770
Southdale 2 135 C. J. H. Wort
South Africa De Beers Industrial Diamond
Division Pty Ltd
P. K. Sitch Diamond Research Lab
Fachbereich Physik PO Box 1770
Universitat/Gesamthochschule Southdale 2135
Paderborn South Africa
D- 33095 Paderbron
Germany A. Zerr
Fachgebiet Disperse Feststoffe
G. D. Soraru Technical University of Darmstadt
Universita di Trento Petersenstrasse 23
Dipartimiento di Ingegneria dei D-64287 Darmstadt
Materiali Germany
Via Mesiano 77
1-38050 Trento
Italy
R. S . Sussmann
De Beers Industrial Diamond
Division Pty Ltd
Diamond Research Lab
PO Box 1770
Southdale 2135
South Africa
List of Symbols
a! absorption coefficient
a atomic attraction constant
a! growth parameter
a power absorption coefficient
a?P, x polytypes or phases
P atomic repulsion constant
P geometrical factor
Y rake angle degrees
Yi secondary ion yield
r(n) orientational surface energy
"/S surface energy J
r width of X-ray reflection
s microplastic deformability
tan 6 dielectric loss factor
Ac concentratation difference
AG!98 Gibbs free energy J
ASint interfacial entropy
AT temperature change K or "C
AX change of size or mass
& elastic strain
& emissivity
i strain rate
rl degree of conversion to nitride
rl fraction of reaction completed
0 angle
8 constant relating tensile strength and hardness
20 X-ray scattering angle
6 entering angle degrees
6 thermal conductivity J cm-2 s-I K-'
61 thermal conductivity of rectants
62 thermal conductivity of products
x empirical parameter relating bulk modulus and bond length
x inclination angle degrees
x layer thickness, mean free path m
x polarity of bond
x thermal conductivity
x wavelength m
x X-ray wavelength
A mean free path m
XXXIV List of Symbols
P C/Ti ratio
P coefficient of friction
v Poisson ratio
7r complementary energy J
P density kg m-3
P dislocation density
P resistivity
Pm theoretical density of product
U conductivity
U electrical conductivity
ff stress Pa
UO median failure stress Pa
ffb fracture strength in bending, modulus of rupture
ffij local stress field
ffS Stefan-Boltzmann constant J cm-* sC1 KC4
UY yield stress GPa
7 annealing time
7- help time S
4 angle between crack and tensile stress
4, w, 7- ternary phases
4 azimuthal angle between polarization vector and substrate direction
4 constraint factor
X electron affinity
W wear coefficient
(4 7 average diffusion distance in time r
0 vacancy in crystal structure
a indent size, half length of diagonal m

a crack length m
a depth of cut mm
a0 equilibrium bond distance m
acr critical flaw size m
ai depth of lateral crack on erosion m
a, b heat capacity coefficients
a, 6 , c crystal unit cell parameters nm
lattice constants
A area m2
A contact area m2
A material constant
ABAB,
ABCABC stacking sequences
A,B,H,HA stacking positions
b Burgers vector
B bulk modulus GPa
B designation of a dissolved species in, e. g., a liquid
(B4C) designation of a non-stociometric compound (solid solution)
List of Symbols XXXV
radius of radial crack m

velocity of light in vacuo m s-'
interfacial concentration
radius of lateral crack on erosion m
proportionality constant
specific heat per unit volume
Ci concentration of impurities
CP heat capacity J g-' K-'
d bond length A
d degree of dilution
d diameter of Brinell impression m
d diameter of Vickers impression m
d grain size m
d height of beam m
d layer thickness
d spacing of powder diffraction rings rad
dP pore diameter
D diameter of Brinell indenter m
D diameter of median crack m
D diffusivity
D size of particles
DO diffusion factor
DH, DC diffusion coefficients at high and low temperature
e unit electron charge
E activation energy J
E binding energy J
E Young's modulus
EO theoretical Young's modulus
E" E/(l-v2)
EC composite potential energy J
EF Fermi energy
4 band gap J
Ei ion energy J
Eirnax maximum ion energy J
EP potential energy J
f feed rate mm mine'
f frequency Hz
f volume fraction
f Weibull safety factor Hz
h volume fraction of Z
F force N
statistical failure probability
ion flux
external work (linear elasticity) J
vapor phase
shear modulus
XXXVI List of Symbols
strain energy release rate

toughness, fracture energy, work of fracture N m-l
convective heat transfer coefficient J cm-2 s - ~K-'
indentation depth
Planck constant
hardness
enthalpy
enthalpy of formation at 298 K
Brine11 hardness GPa
Meyer hardness GPa
Knoop hardness GPa
plastic hardness
Vickers hardness GPa
average applied compressive stress in hardness test
GPa
i, a fluxes of impinging ions and deposited atoms
I rank order of test result
IBB intensity of black body radiation
4 peak height
IRB intensity of real body radiation
ji current density
J mass flux
k Boltzmann constant
k constant for layer growth
k0 reaction constant
hog logarithmic rate constant
kL linear rate constant
k* parabolic rate constant m2s-l or kg2m-4s-'
kr reaction coefficient
K proportionality constant
K stress intensity factor
KIC KICO
J
fracture toughness
1 diffusion path length
1 length of sample
1 long diameter of Knoop impression
1 span of beam
L defined load
Lm,, maximum load
m Weibull modulus
M mode of deformation (I, 11, or 111)
M metal
M Mohs hardness
A4 weight
Mi molecular mass of impurity
n number
n order of reaction
Paris exponent for fatigue
refractive index
stress exponent
number of tests
average coordination number
no. of constraints for coordination number r
average coordination number
grain size exponent
momentum
porosity
fixed load N
gas pressure
porosity Yo
pressure GPa
ambient and high pressure Pa
confining pressure in powder
pressure of nitrogen Pa
partial pressure of O2 Pa
Porod scattering vector
scattering vector
activation energy J
heat of reaction J g-'
resonance factor
heat of transport J
radius
distance from crack tip m
interatomic distance m
radius of curvature
equilibrium interatomic distance m
gas constant
film growth rate
average roughness
radius of plastic zone m
solid phase
elastic recovery Yo
stoichiometric ratio
entropy Jmol-' K-'
thickness of window
time S
ChI4?
delay time S
time for wave propagation S
temperature K or "C
ternary phase
ambient temperature K or "C
temperature limits K or "C
XXXVIII List of Symbols
adiabatic combustion temperature K or "C

critical temperature and pressure
eutectic temperature
absolute melting temperature K or "C
substrate temperature K or "C
displacement m
stoichiometry factor
wave velocity
internal energy J
substrate bias voltage
elastic strain energy J
plasma potential v
surface free energy J
average velocity
machining velocity m min-'
volume m3
wavenumber cm-'
longitudinal velocity of sound m s-'
UP volume of pores
V volume m3
volume fraction
symbol for vacancy in a chemical formula
applied substrate bias voltage V
volume lost in erosion impact m3
molar volume of metal
orientational growth rate
molar volumes of product and reactants
sample thickness m
RF power W
carbon-to-metal ratio
coordinate
layer thickness m
liquid phase
lattice directions
Miller indices
List of Abbreviations
3PB three-point bend

4PB four-point bend
ACC amorphous covalent ceramics
AES Auger electron spectroscopy
AFM atomic force microscope
APW augmented plane wave
AR antireflection
ASEA Swedish company
ASTM American Society for Testing and Materials
b.c.c. body-centered cubic
b.c.t. body-centered tetragonal
BB black body
BET Brunauer-Emmett-Teller method for determining porosity
3C, 4H, 6R polytype notations of Sic (C: cubic, H: hexagonal, R:
rhombohedral)
CAD cathodic arc deposition
CALPHAD calculation of phase diagrams model
cBN cubic boron nitride
CCD charge coupled device
CED cutting edge displacement
CMC ceramic matrix composite
COOP crystal orbital overlap population
CN coordination number
CVD chemical vapor deposition
CVI chemical vapor infiltrated
cw continuous wave
d.c. direct current
DAC diamond anvil technique
DCC direct coagulation casting
DF-TB density-functional tight-binding method
DH methy ldiethoxysilane
DIN Deutsche Industrie Norm
dlC diamond-like carbon
DLC diamond-like carbon
DMO directed metal oxidation
DOS density of states
DTA differential thermal analysis
ECH electron cyclotron heating
ECR electron cyclotron resonance
XL List of Abbreviations
EDAX energy-dispersive analysis of X-rays

EELS electron energy loss spectroscopy
EP electroplated
EPMA electron probe microanalysis
EPR electron paramagnetic resonance
ERD elastic recoil detection
ERDA elastic recoil detection analysis
ESCA electron spectroscopy for chemical analysis
ESK Elektroschmelzwerk Kempten
EXAFS extended X-ray absorption fine structure
f.c.c. face-centered cubic
FEPA Federation Europeen des Fabricants de Produits Abrasifs
FTIR Fourier transform infrared
FWHM full width at half maximum
FZK Forschungzentrum Karlsruhe
GA-XRD glancing angle XRD
gc glassy carbon
GEC General Electric Company, USA
GFRP glass-fiber reinforced plastic
GGA generalized gradient approximation
h hexagon a1
hBN hexagonal boron nitride
HF-CVD hot filament CVD
HIP hot isostatic pressing
HK Knoop hardness
HOMO highest occupied molecular orbital
HOPG highly ordered pyrolytic graphite
HPHT high-pressure high-temperature
HPL high-pressure laminate
HPMS high-pressure microwave source
HR Rockwell hardness
HR-TEM high-resolution TEM
HSS high-speed steel
HV Vickers hardness
IBAD ion-beam assisted deposition
ICSD inorganic crystal structure database
ICDD international center for diffraction data
IED ion energy distribution
IR infrared
ISE indentation size effect
IS0 International Standards Organization
ITER international thermonuclear experimental reactor
JAERI Japan atomic energy research institute
JFM Johnson figure of merit
JIS Japanese Standards
KFM Keyes figure of merit
List of Abbreviations XLI
LAS lithium aluminosilicate

LDA local density approximation
LDA laser diode array
LIDT laser induced damage threshold
LPI liquid polymer infiltration
LPSSiC liquid-phase sintered S i c
LPSSS low-pressure solid-state source
LRO long-range order
LSF line spread function
LSI liquid silicon infiltration
LWIR longer wavelength infrared
MAK Maximal zulassige Arbeitsplatz Konzentration
MAS-NMR magic angle spinning NMR
Me-DLC metal-DLC hybrid
MMC metal matrix composite
MOR modulus of rupture
MOSFET metal-oxide silicon field effect transistor
MS mass spectroscopy
MSIB mass-selected ion beam
MTES methy ltriethoxy silane
MTF modulation transfer function
MW microwave
MWP-CVD microwave plasma CVD
NASA National Aeronautics and Space Administration (USA)
ncTi02 nanocrystalline titania
NDE nondestructive evaluation
Nd-YAG neodymium-yttrium-aluminum-garnet laser
NEA negative electron affinity
NICALON branded Si-C-0 composite fiber from Nippon Carbon
NIRIM National Institute for Research in Inorganic Materials (Japan)
NMR nuclear magnetic resonance
NRA nuclear reaction analysis
ORNL Oak Ridge National Laboratory
PA-CVD plasma assisted CVD
PAlC polyaluminocarbosilane
PCS polycarbosilane
p.p.m. parts per million
PBC periodic bond chain
PC potential cycling
pcBN polycrystalline boron nitride
PcD polycrystalline diamond .
PCS polycarbosilane
PCT Patent Cooperation Treaty
PTC polytitanocarbosilane
PTES phen y ltriethoxy silane
PTFE polytetrafluoroethylene
XLII List of Abbreviations
PP polymer pyrolisis
PVD physical vapor deposition
r rhombohedral
RBAO reaction bonded aluminum oxide
RBM reaction bonded mullite
rBN rhombohedral boron nitride
RBS Rutherford back-scattering
RBSN reaction bonded silicon nitride
RF radio frequency
RSF reduced spatial frequency
RSSC reaction sintered silicon carbide
S.C. simple cubic
s.c.cm. standard cubic centimeters
SAD small angle diffraction
SAXS small-angle X-ray scattering
scs Textron process Sic fibers with C core and C surface
SEM scanning electron microscope
SERR strain energy release rate
SHS self-propagating high-temperature synthesis
SIALON Si-A1-0-N (silicon aluminum oxynitride) fiber
SiCAlON SiC-AlN-Al20C composite fiber
Si-DLC Si-DLC hybrid
SIF stress intensity factor
SIMS secondary ion mass spectrometry
SNMS secondary neutron mass spectrometry
SP sintered powder
SRO short-range order
STM scanning tunnel microscopy
ta-C hydrogen-free amorphous carbon
taC tetrahedral amorphous carbon
TD theoretical density
TCNE tetracyanoethylene
TEM transmission electron microscope
TGA thermogravimetric analysis
TH triethoxysilane
TRS transverse rupture strength (= MOR)
TZP tetragonal zirconia polycrystals
UHP ultahigh purity
UPS ultraviolet photoelectron spectroscopy
uv ultaviolet
VAMAS Versailles Agreement on Materials and Standards
vc vapor phase formation and condensation process
VEC valence electron concentration
VLS vapor-liquid-solid process
vs vapor-solid reaction
wBN wurtzitic boron nitride
yttrium aluminium garnet, yttrium aluminate, Y2A15012
XLIII
List of Abbreviations
X-ray absorption near edge structure

X-ray photoelectron spectroscopy
yttrium-lithium-fluorite
X-ray diffraction
XANES
XRD
YAG
YLF
XPS
Introduction: Novel Ultrahard Materials
A. Zerr and *R. Riedel
Introduction
The synthesis of new materials with hardness comparable to or even harder than dia-
mond is of considerable fundamental and technological interest and is a great chal-
lenge to chemists, physicists, and materials scientists. Most of the known ultrahard
materials, including diamond and cubic boron nitride, were first synthesized in the
1950s and industrially manufactured using high pressure-high temperature processes
[I-31. Extensive research in this domain continues and recently a few new materials
have been synthesized or rediscovered as superhard ones (Si02-stishovite [4], cubic
Si3N4 [5]). New vapor deposition methods (CVD, PVD, laser ablation etc.) which
allow the deposition of diamond, cubic boron nitride and other hard materials
films at low temperature and low pressure (i.e. often under metastable conditions)
on a variety of substrates have been developed since the early 1980s [&lo].
Diamond and cBN (cubic boron nitride) combine excellent mechanical, chemical,
and physical properties. However, owing to its instability at high temperatures,
diamond cannot be used, for example, as a cutting tool for steel. Moreover, with
increasing temperatures diamond and cBN weaken due to the onset of the transforma-
tion to the graphite structure so that above 1100°C in a nonoxidizing atmosphere
boron carbide B4C(with a hardness of about 30 GPa) has been identified as the hardest
material [I I]. For this reason and because of the need to replace expensive diamond in
many other applications, new hard materials with comparable or even superior proper-
ties are required. Theoretical work on the carbon nitride C3N4predicted that this com-
pound could have a hardness comparable to or even greater than that of diamond
[12,13]. The synthesis of polycrystalline C3N4films was first reported in 1992 [I41
and in 1993 [ 151and has been under extensive investigationsince then, resulting in hun-
dreds of publications on this subject. However, in a few recent publications there has
been doubt the thermodynamic stability of theoretically predicted ultrahard C3N4
phases [ 161 and whether their hardness would be comparable to that of diamond
[ 171. In consequence, research on the low-temperature synthesis of diamond
and cubic boron nitride [18],the search for other possible candidate compounds as
well as microstructure design of known materials and their composites [ 19-21]
(heterostructures, whiskers, nanocomposites) remain the subject of experimental
and theoretical efforts. Microstructure design is addressed in this book by D. Sherman
and D. Brandon in Part I, by S. Vepfek in Part I1 and by A. Krell in Part 111.
This introduction deals with some of the latest experimental and theoretical
developments in the field of novel boron- and carbon-based ultrahard materials
as well as with new observations on a class of silicon-based compounds which
previously were not classified as ultrahard.
* This is a revised version of the paper published earlier [ 1521.

XLVI Introduction: Novel Ultrahurd Materials
Hard Materials
Generally, hard materials are solids with high hardness in the range 8-10 on the
Mohs scale of hardness, given by the sequence of minerals which can be scratched
by the next (Table 1). Usually, common hard materials are subdivided into
compounds with metallic (like TIN or WC), ionic (Al2O3>,or covalent bounding
(diamond, Si3N4)[22]. The definition of ultrahard materials is that their hardness
values are comparable to that of diamond. Another definition often used in the
literature is that the hardness of such materials exceeds 40GPa. Among all
known single phase compounds diamond, cBN and probably boron carbides
(Bl3C2-Bl2C3)satisfy the latter definition. Accordingly one can expect that novel
ultrahard compounds will be found in the isothermal ternary phase diagram B-
C-N given schematically in Fig. 1. In particular, the carbon nitrides (e.g. C3N4)
and the boron carbonitrides (B,C,NZ) have been discussed as substitute materials
Table 1. Hardness ranking of minerals and some prominent synthetic ceramic materials according
to F. Mohs. In the case of the synthetic materials microhardness values are given in units of the
Knoop scale. The microhardness variations result from variations in the grain size, the load of
indentation, the phase composition and the used densification techniques. Modified after [ 1521.
Minerals/ Formula Mohs Microhardness,' Microhardness

synthetic hardness Knoop 100 variations for
materials Wal polycrystals' [GPa]
Talcum Mg,[(OH)z/Si4O,oI 1
Hexagonal hBN 0.15-0.30
boron nitride*
Gypsum CaS04 2 H 2 0 2
Calcite CaCO, 3
Fluorite CaF2 4
Apatite Ca5[(F3OH)/(PO&I 5
Feldspar K[AlSi,O,] 6
Quartz Si02 7
Topaz A1z[F2/Si041 8
p-Silicon nitride* BSi3N4 17 14-26
Corundumt
Titanium nitride*
t g 3 } 9 21
14-29
Silicon carbide* Sic 26 20-38

%-Silicon nitride %Si3N4 26-35 2648
Titanium carbide* TIC 28 13-32
Boron carbide' B4C 20-38
Titanium diboride* TiBz } 30 19-35
Boron suboxides B,O 30-59
Stishovite* Si02 33
Cubic boron nitride' cBN 45
Diamond C 10 75-100
' The microhardness values are taken from published sources [7] and [102].
The microhardness varations are taken mostly from previous work [25].
* Synthetic material.
Synthetic material or natural mineral.
Hard Materials XLVII
Figure 1. Schematic of the isothermal ternary B C - N phase diagram at temperatures below the
decomposition of the stoichiometric compounds and the regimes of composition of some solid solu-
tions discussed. For explanation of the compounds see text. First published in [I 521 and reproduced
with permission.
for diamond. One can also expect that metastable dense high pressure phases of the
compounds based on silicon or oxygen can belong to ultrahard materials. One of
these is the high pressure-high temperature stishovite phase of Si02 with a hardness
exceeding 33 GPa [4], which is almost an order of magnitude above that of quartz,
the ambient pressure phase of Si02 (Table 1). The second one is cubic Si3N4with the
spinel structure, whose hardness may be comparable to that of Si02-stishovite [5].
Hardness
Hardness is one of the quantitative parameters that describe resistance of a material
towards plastic (irreversible) deformation. Plastic deformations begin when the
shear component of the stress applied to a material exceeds some value called the
yield stress. There are many ways to create a plastic deformation and consequently
many ways to define and to measure resistance of a material towards such deforma-
tions. Hardness can be determined in several ways:
- Scratching methods (Mohs, Martens),
- Grinding methods (Rosiwal).
- Indenting methods (Vickers, Knoop, Brinell, Rockwell, Shore).
For example, the Vickers hardness, Hv, is defined as the applied load P divided by
the surface area of the impression, while the Knoop hardness H K is derived from the
load P divided by the projected area of the impression:
1854.4P
Hv =
d2 ’
14 229 P
HK=-
12
XLVIII Introduction: Novel Ultrahard Materials
where d is the diagonal of the square-based diamond pyramid of the Vickers

indenter and I is the long diagonal of the rhombus-based diamond pyramid of the
Knoop indenter. In the case of applied test loads below 1.96N the determined
hardness is defined as microhardness and depends on the applied amount of load
[23]. Generally, the unit of the hardness measured is given in kgmm-* or in GPa
(1000 kgmmP2 = 9.81 GPa).
The Vickers or Knoop hardness can be also expressed in terms of Mohs hardness,
M , by the following expression [23]:
H v = 3.2M3. (3)
However, this expression cannot be applied for superhard materials like diamond.
Each testing method yields different hardness values for one material. Thus, in
order to compare hardness values of different materials, the specific test method
and the test conditions have to be described carefully. Moreover, the hardness of
single crystals depends for many compounds on the crystallographic plane tested
and for the Knoop indentor additionally on the orientation of the indentor’s long
axis relative to the crystallographic axis of the examined crystal. For example, in
experiments on single-crystalaSi3N4the Vickers hardness measured on different crys-
tallographic planes varied by more than 34% and the Knoop hardness measured on
the same crystallographic plane by 7% [24]. Hardness of polycrystalline materials is
strongly influenced by: residual stresses, toughening phases, microstructural textures,
the grain size, the applied load and the porosity as well as by the structure and
composition of grain boundaries. As a consequence, the hardness of polycrystals
may significantly differ from that of the single crystal counterpart. For example,
due to the above reasons the measured microhardness of polycrystalline TIC varies
between 1200 and 3250Hv or by more than 270% and that of polycrystalline
Si3N4between 1500 and 4800 (Hv)or by more than 300% [25]. In Table 1, the micro-
hardness values and their variations for some prominent synthetic ceramic materials
are listed ranked according to the Mohs scale. The above topics are discussed in more
detail in Part I by D. Sherman and D. Brandon, and by A. Krell.
There are few theoretical attempts to describe hardness quantitatively in terms of
the elastic bulk or shear moduli of an ideal solid. The elastic moduli and the
corresponding hardness of a compound or element are strongly related to its
molar volume, to its chemical bonding, and to its crystal structure. Basically, it
can be stated that the higher the hardness the lower the molar volume and the
more covalent the bonding of the material.
Liu and Cohen suggested in their paper on hypothetical PC3N4 [12] that ‘on the
microscopic level, for ideal systems, hardness is determined by the bulk modulus’.
To estimate the bulk modulus they used an empirical model developed earlier
[26], where the bulk modulus scales as a homopolar energy gap divided by the
volume of the bond charge. The resulting relation gives theodependence of the
bulk modulus B (in GPa) as a function of the bond length d (in A) and the empirical
parameter A:
1971 - 220X
B= d3.5 ’ (4)
Hurd Materials XLlX
Table 2. Calculated equilibrium volumes, bulk moduli, and cohesive energies of some postulated
ultrahard materials. The calculated and/or experimental values of diamond, lonsdaleite, cubic
BN, and !3Si3N4are given for comparison. The experimental volume and density values are derived
from crystal structure data. CH.6 = all sp2-bonded hexagonal carbon phase 26,73,74]; Cb,c,t,.4= all
1
sp*-bonded body-centered tetragonal carbon [26,72,76]; BNh.c.t.-4= all sp -bonded body centered
tetragonal BN [119]. First published in [152] and reproduced with permission.
Material Crystal Vo Density B, Ecoh Refs

symmetry nm3/ [g/cm3] [GPa] [eVI
atom]
Cdiarnond cubic 5.67 (exp) 3.51 (exp) 444 (calc) 8.17/atom (calc)
435 (exp) 7.37/atom (exp)
Clonsdalelte hexagonal 6.47 (exp) 3.52 (exp) 440 (calc) 8.14/atom (calc)
cBN cubic 5.90 (exp) 3.45 (exp) 367 (calc) 13.2/BN pair (exp)
369 (exp)
hexagonal 10.42 (exp) 3.19-3.20 265 (calc) 74.3/unit cell (calc)
(exp) 256 (exp)
hexagonal 6.25 3.49 427 (calc) 81.5/unit cell
hexagonal 6.29 3.17 372 (calc) 9.06/atom
tetragonal 6.72 2.97 362 (calc) 8.47/atom
tetragonal 7.37 2.82 268 (calc) 13.36/BN pair
X is 0 for homopolar solids of group IV elements, and 1/2, 1, and 2 for heteropolar
solids of group 111-IV, 111-V, and 111-VI elements, respectively. From this, it is
evident that increasing ionicity results in a decrease in B. This expression has
been further improved to account for non-octet compounds such as Si3N4 by
using the average coordination number N, # 4 [26]:
(N,.) (1971 - 220X)
B=-
4 d3.5
In Table 2, some experimental and calculated structural properties such as the

equilibrium volume, bulk modulus, and cohesive energies of candidate materials
with potentially ultrahigh hardness are summarized. The equilibrium volume
corresponds to the minimum of the total crystal energy as a function of volume [26].
The cohesive energy of a crystal represents the difference between the energy of the
isolated constituent atoms or molecules and the energy of the crystal at absolute
zero temperature [27]. It is evident from Table 2 that the higher the cohesive energy
and the lower the equilibrium volume the higher the bulk modulus of the material.
On the other hand materials deform plastically only when subjected to shear
stress. According to Frenkel analysis, strength (yield stress) of an ideal crystalline
solid is proportional to its elastic shear modulus [28,29]. The strength of a real
crystal is controlled by lattice defects, such as dislocations or point defects, and is
significantly smaller then that of an ideal crystal. Nevertheless, the shear stress
needed for dislocation motion (Peierls stress) or multiplication (Frank-Read
source) and thus for plastic deformation is also proportional to the elastic shear
modulus of a deformed material. Recently Teter argued that in many hardness
tests one measures plastic deformation which is closely linked to deformation of a
shear character [17]. He compared Vickers hardness data to the bulk and shear
L Introduction: Novel Ultrahard Materials
Figure 2. Scattering of the Vickers hardness for a common set of hard materials when compared
with bulk modulus (gray region) and shear modulus (black region). Following this comparison
Teter suggested that ‘the shear modulus is a significantly better qualitative predictor of hardness
than the bulk modulus’ [ 171.
moduli from over thousand measurements and found that the shear modulus of
polycrystalline aggregates is a significantly better qualitative predictor of hardness
than the bulk modulus (Fig. 2).
Experimental bulk moduli can be obtained from the measurement of lattice
parameters and volumes as a function of pressure [30]. The single crystal elastic
moduli can be measured using the Brillouin spectroscopy, inelastic neutron scatter-
ing, ultrasonic measurements or the Schaefer-Bergmann method [311. Once the
single crystal moduli are known one can derive the bulk B and shear G moduli of
a polycrystalline material [32].
Carbon-based Hard Materials

The most common allotropes of elemental carbon are graphite and diamond.
Graphite crystallizes in a sheet structure with hexagonal symmetry and sp2 hybridi-
zation of the trigonally coordinated C-atoms whereas diamond exhibits a cubic
lattice related to the zinc blende structure with sp3 hybridization of the tetrahedrally
coordinated C-atoms. Less common carbon polymorphs are lonsdaleite with
the hexagonal wurtzitic structure 1331, and the rhombohedral polymorph called p-
diamond (3R polytype of wurzitic structure) [34,35]. The rhombohedral poly-
morph, which was reported as birefringent lamellae in natural diamonds, has not
yet been obtained in significant amounts needed for detailed studies [35]. Graphite,
Carbon-based Hard Materials LI
Figure 3. The different carbon modifications: (a) hexagonal graphite; (b) cubic diamond; and (c)
hexagonal lonsdaleite. First published in [ 1521 and reproduced with permission.
diamond, and lonsdaleite are, however, well investigated and their structures are
shown in Fig. 3. In lonsdaleite, comprised of the same tetrahedral configuration
as in diamond, the planes of six-membered carbon rings are in the chair and boat
conformation and are stacked in an ABAB sequence while in diamond, these
planes exhibit the chair conformation exclusively and are arranged in an
ABCABC sequence. The recent discovery of the c 6 0 molecule shown in Fig. 4,
which is the most prominent representative of the fullerene family, led to a new
type of carbon allotrope. In fullerenes, the carbon atoms form spherical clusters
comprised of six- and five-membered rings of the general composition C20+2m
(m = 0 , 2 , 3 . ..) where m is the number of hexagons. The c 6 0 fullerene has been
predicted by Kroto and Smalley since 1985 [36-38] and was isolated first in
macroscopic quantities by Kratschmer et al. in 1990 [39-41]. At present, the
spherical carbon clusters and their derivatives are discussed with respect to their
application in material science [42,43]. Furthermore, a variety of new polymeric
carbon networks have been postulated to exhibit interesting materials properties
[41,44]. Recently, carbon nitrides such as the hypothetical compound C3N4 have
been also considered for the synthesis of ultrahard materials [ 14,151.
LII Introduction: Novel Ultrahard Materials
Figure 4. Molecular model of the fullerene C60.The gray carbon atoms illustrate one of the 12
pentagons present in C6,,. First published in [152] and reproduced with permission.
Diamond
Besides the aesthetic appearance of diamond in the form of gemstones, diamond is
the hardest material known, and is, therefore, on the top of the Mohs scale and has
the lowest molar volume (3.4cm3) of any material. Owing to this unique property,
diamond is used for many technological applications such as abrasion, cutting, and
polishing. Diamond also has the highest thermal conductivity (2000 W m-' K-') of
any material at room temperature, four times as high as the value of Cu or Ag, is a
good electrical insulator, it has a small dielectric constant and exhibits a high
electron/hole mobility. Boron doped diamond (Typ IIb) exhibiting a p-type con-
ductivity has been known for a long time. Production of diamond with a reasonable
n-type conductivity was one of the most difficult tasks in diamond synthesis.
Koizumi and coworkers recently obtained diamond films exhibiting n-type conduc-
tivity by using microwave enhanced plasma CVD with phosphine PH3 as a donor
source for homoepitaxial deposition of n-type diamond 1451. Therefore, diamond
will be a most important candidate material for future applications in electronic
devices. In addition, diamond has low coefficients of friction and thermal expansion,
high chemical and corrosive resistance towards most acids and oxidizing substances,
it is transparent to visible and infrared light, withstands ionizing radiation and can,
therefore, also be used as window or lens material or as a protective coating for this
application.
Conventionally, diamond is synthesized at high pressure and high temperature
according to the method developed by the General Electric Company in the 1950s
Carbon-based Hard Materials LIII
[l, 461. Synthetic diamond is now commercialized and is utilized for cutting, grinding,
and polishing. Recent developments in the high-pressure synthesis technique allow
growth of diamond single crystals up to 25 carats in weight [47]. The high-pressure
synthesis of diamond is described in more detail by M. W. Bailey et al. in Part 111,
while hydrothermal synthesis is discussed in Part I1 by K. G. Nickel et al.
Since the mid-1980s the vapor deposition synthesis of diamond films has
attracted increasing scientific and industrial interest. At present, vapor-grown
diamond products are commercially viable as thin-film-coated cutting tools and
freestanding thick-film cutting tools, as substrates for thermal management
application and radiation detectors, as optical windows and for production of
high-range audio-speaker diaphragms. Commercialization of the vapor-grown
diamond became possible due to the cost reduction below $5/carat as a result of
the dramatic increase in growth rate and of improvements in energy-use efficiency
[481.
Advanced diamond films are synthesized using the gas-phase decomposition of
volatile carbon sources such as methane (CH4), acetone (H3C-CO-CH3),
carbon monoxide (CO), acetylene (HC-CH), or adamantane (CI0Hl6)and is
conducted under temperature and pressure conditions where graphite is the stable
polymorph (Fig. 5). The most important methods for producing CVD-diamond
under metastable conditions are:
- Microwave Plasma Assisted Chemical Vapor Deposition (PACVD) [49,50],
- Heated Filament Assisted Chemical Vapor Deposition (HFCVD) [5 13,
Figure 5. Pressure and temperature conditions of the diamond synthesis: (a) shock wave production
of diamond; (b) high temperature, high pressure regime for the synthesis of diamond; (c) catalytic
region for diamond formation; (d) chemical vapor deposited diamond; and (e) transformation of
CG0into diamond. The most recent review of the P, T phase diagram of carbon can be found
elsewhere [151].
LIV Introduction: Novel Ultrahard Materials
- Direct Current (DC) Plasma Jet Deposition [52,53],and

- Oxygen-Acetylene Torch [54].
The chemical-vapor deposition of diamond films and their applications are reviewed
by C.-P. Klages and by R. S. Sussmann et al. in Parts I1 and 111, respectively. To date
the most effective CVD method (with the greatest mass deposition rate) is based on
the hydrogen/hydrocarbon gas mixtures. In this method diamond is formed for
-
kinetic reasons according to the simplified reaction:
H2
CH4 Cdiamond f 2H2.
In a typical process, the hydrocarbon precursor containing more than 95% H2-gasis
passed through a plasma (700-1000°C) or over a heating filament ( T M 2000") at
less than atmospheric pressure. The reaction product, solid carbon, is then deposited
on a substrate heated at 800-1000°C and contains both graphite and diamond. In
the presence of H2-gas, the formation of unsaturated carbon nuclei and hence the
growth of graphite is suppressed by atomic hydrogen formed under these conditions
and is due to the reaction of the solid carbon with hydrogen radicals (H') giving
volatile hydrocarbons. The reaction rate of Cgraphite with H' is about 20 times as
high as the rate of diamond. Therefore, the formation of diamond is promoted.
Growth rates of up to 0.9 mm h-' [55] or 20 carats per hour [48] have been achieved
on a variety of substrates making the CVD-diamond a highly interesting material
for technological applications.
Diamond films have also been deposited from hydrogen-free gas mixtures such
as C60/Ar using microwave apparatus. The deposition rates are, however,
significantly lower. Additionally, use of hydrogen-poor plasmas results in nanocrys-
talline (3-10nm) diamond films in contrast to micrometer sized crystals from the
hydrogen-rich plasmas [56].
There are continuous theoretical attempts to describe the mechanism of CVD-
diamond synthesis including mechanisms of surface reactions, diamond nucleation,
and film growth. To achieve this aim various phenomenological or first-principles
models, molecular dynamics and Monte Carlo simulations have been used [57,58].
While the dominant substrate for low pressure growth of diamond films is single
crystal silicon, several different materials such as Ta, Mo, W, Cu, Au, Ni, Sic, SO2,
and Si3N4have been coated with polycrystalline diamond or diamond-like deposits.
The nucleation rates and the adhesion of the generated films vary with the type of
substrate material, which is related to the formation of intermediate carbide layers
such as Sic or WC.
Identification of vapor grown diamond is accomplished by X-ray diffraction or
electron diffraction using a transmission electron microscope (TEM) and by
Raman spectroscopy. Diamond coatings show the characteristic first-order Raman
peak for diamond at 1332cm-' [59]. In many cases, an additional peak located at
~ 1 5 5cm-'
0 occurs in the Raman spectrum which is attributed to a highly disordered
carbon phase (diamond-likecarbon) [60]. This diamond-like carbon contains sp3 and
sp2 hybridized C-atoms which do not coincide with graphite.
Recently a new simple method for the synthesis of diamond, called metallic
reduction-pyrolysis-catalysis, was reported. In this method diamond powder was
Carbon-based Hard Materials LV
obtained as a product of reaction of carbon tetrachloride with metallic sodium in an

autoclave at 700°C, where CC14 served as a carbon source [18]:
CC14 + 4NaCl- Catalyst
700°C
C + 4NaCl.
As a metal catalyst an Ni-Mn-Co alloy (70 :25 :5 wt%) was used. About 2% of the
carbon yield was well crystallized diamond and the residue was most probably in
an amorphous form, as was evident from X-ray powder diffraction and Raman
spectroscopic measurements.
Diamond-like and Amorphous Carbon

In the course of the research into the synthesis of diamond under metastable condi-
tions, a new class of materials, diamond-like carbon and hydrocarbon phases, have
been discovered. The diamond-like hydrocarbons (aC :H) are generated by the R F
self-bias method, a technique derived from R F sputtering, developed by L. Holland
[61,62]. The molecular ions, C,H;, derived from the particular hydrocarbon used in
the plasma, disintegrate upon colliding with the substrate surface resulting in the
formation of diamond-like hydrocarbon films [63]. The main structural feature of
diamond-like hydrocarbons is the presence of both sp3- and sp2-carbon. Solid-state
NMR-investigations revealed that the material contains sp3-carbon atoms of the
form -C-H or H-C-H [6].
No quaternary carbon atoms could be detected while methyl groups, -CH3, were
found to be present in small amounts. In addition, resonant Raman spectroscopy
showed small 7r-bonded clusters of sp2 sites [64].
It was found that conducting the R F self-bias experiment with hydrocarbons
in the presence of H2 results in production of small diamond clusters. This experi-
mental result indicates that aC : H could be considered as an intermediate
compound formed during the transformation of hydrocarbons to diamond in the
plasma processes [65].
The mechanical properties of diamond-like hydrocarbon films strongly depend on
the hydrogen content. Increasing the ratio of sp3 sites results in decreased hardness
values and lower wear resistance [66]. This phenomenon is attributed to the fact that
hydrogen is monovalent and cannot contribute to formation of a covalently bonded
and highly cross-linked carbon network as found in the diamond structure. However,
microhardness values in the range 30-50 GPa have been measured in aC :H films,
values significantly higher than the hardness of Sic (20-38 GPa). In contrast, dia-
mond-like carbon (aC) is free of hydrogen or contains only little hydrogen and does
not represent microcrystalline diamond [67,68]. The preparation of aC has been
accomplished by magnetron sputtering [68]. Again, the structure and properties of
the aC phase is related to the ratio of the sp3/sp2 sites. However, hydrogen is not
responsible for the stabilization of the sp3-hybridizedcarbon atoms. Here, the mechan-
ical constraints resulting from random covalent networks, which can be calculated
using the constraint-counting method developed by J. C. Phillips [69], can be reduced
by the generation of medium-range or long-range order by clustering or crystallization.
LVI Introduction: Novel Ultrahard Materials
Another class of novel carbon material is amorphic carbon which can be obtained
by laser ablation [70]. This amorphous phase contains sp3 -bonded nodules of pure
carbon embedded in a matrix of other types of carbon phases. The mean particle size
of the nodules is 20 nm, the density of the films is between 1.85 and 2.89 g crnp3.The
internal stresses derived by distortions of the bonding angles in random networks
are counterbalanced by the high surface to volume ratios of the nanoparticles. Sub-
strate discs 30mm in diameter were coated with uniform layers of amorphous
carbon by laser ablation of a graphite feedstock. Maximum growth rates of about
0.5 pm h-' and film thicknesses of up to 5 ym have been achieved on different sub-
strate materials such as Si, Ge, ZnS, Cu, stainless steel, quartz, glass, and plastics.
One outstanding property of laser ablation in comparison with the CVD process
for the preparation of carbon films is that the substrate temperature does not
exceed 35°C during the whole deposition procedure. In contrast to the CVD dia-
mond, which develops polycrystalline columnar structures, the laser-ablated amor-
phous carbon is deposited in the form of self-seeding nodules. Diamond-like carbon
films are discussed in more detail in Part I11 by C.-P. Klages and K. Bewilogua.
Novel Hypothetical Three-dimensional Carbon Phases

A dense carbon phase with a calculated density of 4.1 g cmP3was predicted by N. N.
Matyusenko and V. E. Strel'nitzkii in 1979 [71]. Due to this high density value,
ultrahigh hardness of this carbon material is expected. In addition, several different
hypothetical three-dimensional polymeric carbon networks with interesting
materials properties have been proposed. The most relevant ones with respect to
the potential of high hardness are the following carbon networks.
(i) In 1983, R. Hoffmann published a metallic allotrope of carbon in which layers
of infinite polyene chains are connected by bonds parallel to the c-axis. Each
layer is rotated by 90" about the c-axis (Fig. 6a). The unit cell of the crystal
structure is primitive body centered tetragonal and contains four atoms
(b.c.t.-4 structure). In this network, carbon is present in the form of trigonal
sp2 atoms. However, in contrast to graphite where the carbon atoms are
arranged in a two-dimensional sheet structure, the trigonal carbon atoms of
the b.c.t.4 structure form a three-dimensional network [72].
(ii) A similar carbon allotrope was proposed by M. A. Tamor and K. C. Hass in
1990. In this structure, the chains of the trigonal sp2 carbon atoms are rotated
by 60" rather than 90" between the layers. The three-dimensional all sp2 phase
of carbon has a hexagonal Bravais lattice with six atoms in the primitive unit
cell and is known as the H-6 structure (Fig. 6b). However, theoretical calcula-
tions indicate that H-6 carbon is unstable with respect to the transformation to
diamond. This instability is probably due to the short distances between carbon
chains [73,74].
(iii) The 'super diamond' structure shown in Fig. 7 can be derived from tetraethynyl-
methane C(C2H)4. In principle, the polymeric network of the super diamond
could be obtained by oxidative coupling of the methane derivative [41,75].
Curbon-based Hard Materials LVII
Figure 6. (a) Model of a section of the hypothetical polymeric network of the body centered
tetragonal structure (b.c.t.4) of carbon and BN suggested by R. Hoffmann et al. [72]. (b) Model
of a section of the hypothetical hexagonal structure (H-6) of carbon postulated by M. Tamor
and K. Hass [73,74]. In both structures, each carbon atom is considered to be trigonally coordinated
and sp2-bonded, exclusively. First published in [ 1521 and reproduced with permission.
(iv) The carbon skeleton of allene, 2HC=C=CH2, could also be utilized for the
formation of a three-dimensional carbon phase [41]. In this case, the oxidative
coupling of allene could provide a carbon structure with orthogonally arranged
chains of sp2 carbon atoms (Fig. 8). A similar carbon phase could be obtained
by the coupling of tetraethynylallene, (HCGC)~C=C=C(C=CH)~[75].Here,
the carbon chains are connected by butadiene units instead of single bonds.
Both the b.c.t.-4 and the H-6 phase of carbon are discussed in terms of intermediate
structures formed during the chemical vapor deposition of diamond. Furthermore,
these types of structures could play an important role in diamond-like phases. The
calculations of the structural properties of the b.c.t.-4 and H-6 carbon phase using
LVIII Introduction: Novel Ultrahard Materials
Figure 7. Hypothetical super-diamond structure suggested earlier [41,75]. First published in [ I 521
and reproduced with permission.
the first-principles pseudopotential total-energy method revealed high bulk moduli

B of 350 GPa and 372 GPa, respectively [26,76]. The extraordinary high values of B
are in the range of cBN, the second hardest material known.
Presently, the synthesis of new carbon phases by the coupling of unsaturated
molecular organic compounds such as allenes or alkynes is being intensively
investigated in several laboratories.
Figure 8. Hypothetical allene structure of carbon postulated earlier [41,75]. First published in [I521
and reproduced with permission.
Carbon-based Hard Materials LIX
Fullerenes
It has been shown that fullerenes or their derivatives can exhibit very interesting
chemical, electrical, magnetic, and mechanical properties. Besides, for example,
the superconductivity, which has been experimentally verified for the alkali metal
fullerides M 3 C 6 0 with M = K or Rb [77], fullerenes can serve as a starting material
for diamond synthesis [78,79] and may exhibit high hardness themselves under high
pressure conditions [go].
In 1992 M. Regueiro et al. reported on the transformation of c 6 0 molecules into
polycrystalline diamond by nonhydrostatic compression at room temperature [78]:
P > 20 GPa/20"C
& c60 ' Cdiarnond.
In contrast c 6 0 withstands hydrostatic pressure up to 20 GPa [81]. However, the
football molecules seem to be unstable towards uniaxial or shear stresses whereas
they are stable under isotropic stress where the spherical molecules are homo-
geneously deformed. In a dense arrangement of c 6 0 spheroids, 48 of the 60
carbon atoms have a quasi-tetrahedral coordination which is required in the
diamond structure. Only small structural rearrangements are then necessary for
the transformation into diamond [78].
R. S. Ruoff and A. L. Ruoff proposed that c 6 0 is stiffer than diamond [go]. This
result has been estimated from the calculated bulk modulus of individual c 6 0
molecules. The calculations revealed a bulk modulus B = 843 GPa which is nearly
twice the experimental value of 441 GPa determined for diamond. Since solid c 6 0
forms a van der Waals crystal with f.c.c. lattice and the distance between the
single c 6 0 molecules is about l.Onm, B would be relatively small under normal
conditions. However, when the individual carbon spheres are compressed until
they touch each other, the bulk modulus of the crystal would become in the range
of that of the molecule. Taking into account a volume filling factor of 74% valid for
the f.c.c. structure, B is calculated to be 624 GPa using the volume fraction rule [80].
Therefore, diamond might be outranked by the hardness of c 6 0 crystals. This result
is of great fundamental interest, despite the fact that the extraordinary hardness
value will be only achieved at high pressures (20 GPa).
Carbon Nitride (C3N4)

Theoretical predictions that covalent solids comprised of carbon and nitrogen are
excellent candidates for extraordinary hardness [82,83] lead to a strong interest in
the scientific community which resulted in hundreds of theoretical and experimental
studies on this system. In this chapter we only depict some results on C-N com-
pounds without an attempt to be exhaustive.
The first of such predicted structures, namely PC3N4the structural analogue of
the well-known hexagonal PSi3N4, was suggested almost 10 years ago [12]. A
number of other structures of C3N4 were proposed later: graphite-like (P-6m2)
and cubic defect-zincblende (P-43m) [84], the hexagonal analogue of aSi3N4
LX Introduction: Novel Ultrahard Materials
(P31c), pseudocubic (P-42m), and cubic (I-43d) [13]. Bulk moduli (and hardness) of
most of these compounds were predicted to be comparable with and the last one
even to exceed that of diamond.
The hardness of the hypothetical carbon nitride, pC3N4, has been comprehen-
sively discussed by A. Liu and M. Cohen [12,83]. In the case of &N4, the bulk
modulus is estimated to be higher than that of diamond (483 GPa) by semi-empirical
methods using Eq. (4) with a bond length d = 0.147 nm and an ionicity factor of
X = 0.5. From ab initio calculations of the total structural energy of C3N4 using
the first principles pseudopotential total energy method, a somewhat lower value
of 427GPa for B is derived. This finding can be attributed to the structural
difference between this material and the completely tetrahedrally coordinated zinc
blende structure from which the scaling law (Eq. (4)) was developed [12,83]. In
contrast, the compound pC3N4 is considered to have the pSi3N4 structure and is
only a partially tetrahedrally coordinated structure. The pSi3N4structure is related
to the well known hexagonal structure of the beryllium silicate Be2Si04.Therefore,
the unit cell of PC3N4 is hexagonal and contains two formula units (Fig. 9). The
carbon atoms are sp3 hybridized and tetrahedrally surrounded by N atoms, whereas
nitrogen has a planar trigonal coordination (sp3 hybridization). The overall struc-
ture can be described by a network of CN4 tetrahedra and NC3 sites.
Teter and Hemley suggested the cubic high pressure phase of Zn2Si04denoted as
willemite-I1 as another possible structure of a superhard C3N4[13]. This suggestion
was based on the fact that the structure of the zero-pressure phase of Zn2Si04
(willemite-I) is isomorphous with the structure of Be2Si04. Using first principles
pseudopotential total energy techniques they calculated energetics, structure, and
some physical properties of the cubic willemite-I1 structure as well as of four
other candidate structures of C3N4. They found that aC3N4 (the analogue of
aSi3N4) is energetically favorable at zero-pressure and the cubic willemite-I1
structure might be a high pressure phase of C3N4. The predicted transition pressure
from a-to the cubic willemite-I1phase was 68 GPa. The calculated bulk modulus of
this cubic C3N4was 496 GPa, which is significantly higher than the measured bulk
modulus of diamond (442 GPa) or predicted moduli of a- and pC3N4 (425 GPa and
450 GPa, respectively). This result suggested that the hardness of cubic C3N4with
willemite-I1 structure may exceed that of diamond [13]. However, in his later
work Teter revised this conclusion to the contrary according to the almost linear
relation he has found between shear moduli of polycrystalline compounds and
their hardness [17]. The calculated shear moduli of the polycrystalline p- or cubic
C3N4 are only 60% of the shear modulus of diamond and smaller than that of
cubic BN. Accordingly the hardness of these carbon nitrides (once synthesized)
will probably not exceed that of cubic BN [17]. Further discussion of possible struc-
tures of C3N4can be found in Part I in the chapter of J. E. Lowther.
Synthesis of carbon nitrides has been attempted for example by the pyrolysis of
organic nitrogen-containing compounds such as melamine-formaldehyde resin
and subsequent shock compression of the obtained residues [85]. However, no tetra-
hedral carbon nitride could be detected in the reaction products. In a similar
approach, L. Maya et al. obtained a carbon nitride with trigonally coordinated
carbon by the pyrolysis of organic compounds at 700°C and 225 MPa pressure
Carbon-based Hard Materials LXI
a)
Figure 9. Theoretically suggested crystal structures of carbon nitride C3N4(filled circles represent
the carbon atoms): (a) Hexagonal structure corresponding to that of PSi3N4or phenacite, Be2Si04.
(b) Cubic structure corresponding to that of willemite-11, the high-pressure phase of Zn2Si04[ 131.
First published in [ 1521 and reproduced with permission.
[86]. Plasma decomposition of CH4 and N2 provided amorphous solids containing

C, N, and H, exclusively [87]. High pressure pyrolysis of tetracyanoethylene,
(CN)2C=C(CN)2, gave C4.66Nwith graphitic structure [88]. Similarly, the gas
phase decomposition of pyridine in the presence of chlorine at 800°C provided
the carbon nitride C5N with a hexagonal sheet structure [89].
Martin-Gil et al. reported on the synthesis of cubic carbon nitride with the zinc
blende structure by pyrolysis of N,N-diethyl- 1,4-phenylene-diarnmoniumsulphate
at 800°C in an N2 atmosphere using Se02 as a catalyst [90]. The material obtained
contained both an amorphous matrix and polycrystalline regions with crystallites of
5-50nm in size. The sample as whole had a composition CN, with 0.2 < x < 0.5.
The composition of the embedded crystallites was estimated using EELS (electron
energy loss spectroscopy) data to be C,N,,, with x and y between 3 and 4. The
LXII Introducfion: Novel Ultrahard Matericils
cubic zinc blende structure of the crystallites was derived from TEM measurements
and confirmed by extended energy loss fine structure analysis [90].
Besides the preparation of polycrystalline C- and N-containing thin films by using
direct current magnetron sputtering with nitrogen as the sputtering gas [ 141, the
synthesis of C3N4 was experimentally verified in the work of C. Niu et a/. [15]. In
this paper, the preparation of the carbon nitride was realized by pulsed laser
ablation of graphite targets in the presence of a nitrogen source. The C-N films
deposited on Si (001) and on polycrystalline Ni were characterized by Rutherford
backscattering (RBS). photoelectron spectroscopy (XPS, ESCA), and electron
diffraction measurements in the TEM mode. The analysis of the diffraction data
(convergent beam electron diffraction) clearly showed that the reflections were in
accord with that expected for the PC3N4 structure. Furthermore, the spectroscopic
analyses revealed the presence of C and N in the deposited films. The bonding in
these solids is covalent. The analysis of the chemical composition of the films deter-
mined by RBS exhibited nitrogen contents of up to 45% which is the highest amount
measured in any binary C-N solids to that time. The nitrogen content did not
decrease after thermal annealing in flowing N2 up to 800°C indicating the relative
stability of the C-N material. Preliminary investigations of the hardness showed
that the films could not be scratched by hard metals [15].
Thin carbon nitride films containing small crystals were deposited by Yu e t a / . on
Si wafers in the (100) orientation using R F diode sputtering of graphite target with
pure N2 [91]. The films were about 1 pm thick with a nominal composition of
Co6N0.300.1 and contained polycrystalline grains of 0.5-1 pm in size with higher
nitrogen content. Many of the lattice spacings obtained from the ring patterns
measured with TEM matched those of the theoretically predicted PC3N4 [91].
Analogous results were reported by Fernandez et al. who used a dual ion beam
sputtering system to deposit C-N films with a nitrogen content of about 44% on
KCI substrates [92].
Preparation of crystalline carbon nitrides was reported to be possible by use of
chemical vapor deposition techniques. Films containing a-,PC3N4 and an unknown
C-N phase were obtained on nickel substrates from the gas mixture of nitrogen and
methane in a hot-filament CVD apparatus [93]. The composition of the grown films
close to the stoichiometric value N / C = I .33 was measured using energy-dispersive
X-ray analysis and Auger electron spectroscopy. The lattice constants of synthesized
SI- and (jC3N4agreed with the theoretically predicted values. The authors have not
observed bands of diamond or graphite in Raman spectra of the synthesized films.
Presence of amorphous C-N compounds was ruled out after analysis of the films’
morphology [93]. Plasma enhanced CDV process was applied by He and Chang for
synthesis of PC3N4 on a silicon wafer as a substrate [94]. As a precursor they used a
mixture of methane, hydrogen, and nitrogen. The nitrogen concentration was varied
between 0 and almost 75%. The strongest evidence for synthesis of BC3N4 at the
highest N2 content was obtained from the TEM measurements. The lattice spacings
calculated from the powder diffraction rings agreed with the theoretically predicted d-
spacings of PC3N4. However, contamination of the deposited carbon nitride by
iron carbide or nitride most probably in amorphous form and possible presence of
amorphous CH, double bounds C=C or C=N was also reported [94].
Carbon-based Hard Materials LXIII
According to the theoretical calculations, the synthesis of dense C3N4structures

should be possible by applying high pressures and temperatures [12]. The simplest
possible starting materials for synthesis of carbon nitrides are elementary carbon
and nitrogen. Nguyen and Jeanloz used different forms of carbon (graphite,
amorphous carbon, and c 6 0 fullerene) and nitrogen for their high pressure synthesis
experiments where the samples were heated by an Nd :YAG laser [95]. After heating
+
of the c 6 0 N2 mixture at 30 GPa to 200CL2500 K they obtained a crystalline
material at ambient conditions, the X-ray powder diffraction pattern of which did
not match any of the known carbon phases, the carbon nitrides prepared by deposi-
tion techniques, or the theoretically predicted PC3N4.The measured d-spacings can
be better described by a cubic cell with lattice parameter a = 0.77 nm. Similar results
were obtained with the use of other carbon precursors. Presence of more than one
phases was not excluded. The chemical composition of the synthesized material was
not determined [95].
Molecular precursors containing carbon and nitrogen were used for synthesis
experiments in the Nd :YLF laser heated diamond cell. Tetracyanoethylene
(TCNE) with the stoichiometry C6N4 was investigated to 42 GPa [96,97]. EDX
and EELS measurements showed increase of the nitrogen content in the treated
sample with pressure: After heating at 20GPa the nitrogen content was 24%
and by 42GPa the nitrogen content reached 38%, which is the composition of
the starting material. However, at all pressures crystalline carbon was formed in
addition to the amorphous sp2-bonded carbon nitride. It was also found that
the amount of carbon increases with the heating time [96,97], indicating need of
higher pressure for synthesis of single phase carbon nitrides. Such experiment
was performed on a mixture of TCNE (C6N4) and cyanuric triazide (C3N12)
with a nominal composition C3N4 at 119GPa and 2500°C [97]. X-ray powder
diffraction spectra showed six new lines, which were indexed to a tetragonal cell
with a = 0.7815 nm and c = 0.6219 nm. The stability of another carbon nitride
precursor material paracyanogen (pCN) at high pressures and temperatures has
been investigated in detail [98]: The decomposition temperature of the precursor
was found to increase with pressure from 550°C at 3GPa to 750°C at 20GPa.
This work demonstrated that there is a well defined line for decomposition of
pCN and the decomposition temperatures are moderate. It was likely that the
carbon in pCN retained sp2 hybridization to the highest pressure of the work.
Additionally, above the decomposition temperature the authors obtained crystal-
line graphite even in a region of the carbon phase diagram where diamond was
stable. Thus the pCN precursor, with sp2 carbon bonding, favors the formation
of the carbon phase with sp2-bonded atoms instead of thermodynamically favored
sp3-bonding [98].
Successful synthesis of a bulk crystalline carbon nitrogen was reported by use of
an organic precursor containing carbon, nitrogen, and hydrogen in [99]. After heat-
ing this precursor to 400°C at 6 GPa the author obtained a three phase-agglomerate
consisting of poorly ordered graphite, a-carbyne and a hexagonal phase with lattice
spacings close to that of the predicted PC3N4.
The most significant problems in almost all experiments on synthesis of carbon
nitrides are as follows:
LXIV Introduction: Novel Ultrahard Materials
(i) The samples contain many phases often with different chemical composition.
(ii) Crystalline carbon nitrides have been obtained in small amounts and mostly X-
ray or electron powder diffraction data were presented. Such data do not
always allow unambiguous determination of the structure especially when
other phases were not identified.
(iii) Small sample amounts and presence of other phases is a serious hindrance for
accurate determination of chemical composition of the obtained crystalline
carbon nitrides.
Improvements in the synthesis techniques and analytic equipment may help to solve
these problems. It also has to be mentioned here that the stability of some of the
predicted carbon nitrides was questioned in a number of studies based on chemical
and thermodynamical grounds. A recent discussion of this subject and further
references can be found elsewhere [ 171.
Boron-based Hard Materials

Carbon-based materials such as diamond or diamond-like carbon have several
disadvantages. Among them, the low oxidation resistance of diamond is one of
the most important drawbacks for its technological applications. Boron-based
hard materials of the binary or ternary systems B-C, B-N, B-0, or B-C-N might
improve this property significantly. For example, diamond can be used up to about
600°C in air environment, whereas cubic boron nitride (cBN) withstands oxidation
up to 1100°C [loo, 1011. Elemental boron itself is a hard material. The high micro-
hardness of boron of 27 GPa [lo21 is related to the low molar volume of 5 cm3 and
the covalent bonding of the rhombohedral crystal structure comprised of B12-icosa-
hedra which are linked together by multiple-center bonds.
Besides diamond and cBN, the well known boron carbide B4C is among the
hardest materials and has been comprehensively reviewed by F. Thkvenot [1031.
In the present chapter, the latest developments concerning the binary and ternary
systems B-N, boron carbide nitrides (B-C-N), and boron suboxides are discussed.
Other hard materials based on boron are described by R. Telle et al. in Part 111.
Boron Nitrides
Boron nitride is isoelectronic with elemental carbon and is, therefore, found in
similar crystal structures:
- the hexagonal (hBN) [lo41
- the cubic zinc blende (cBN) [2,3]
- the wurtzitic (wBN) [lo51
- the rhombohedral (rBN) [lo61 structure
In contrast to rhombohedral carbon, rBN has been synthesized in significant
amounts by several methods and is well investigated [107,108]. In contrast to
Boron-based Hard Materials LXV
Figure 10. P,Tphase diagram of boron nitride: (a) catalytic conversion of hBN into cBN; (b) meta-
stable region for the vapor phase formation of cBN. An alternative phase diagram of BN suggested
previously [I 1I] is shown by dashed lines.
graphite, the hexagons of hBN are arranged on top of each other, while the
structures of cBN and wBN completely correspond to diamond and lonsdaleite,
respectively. The zinc blende polymorph of BN is also known as sphalerite or
PBN. Cubic boron nitride is the second hardest material and is produced conven-
tionally at high pressures and temperatures, analogous to those methods developed
for the fabrication of synthetic diamond. It has been commercially available since
1969 [loo]. Hot-pressed monolithic and polycrystalline BN, comprised of both the
cubic and the wurtzitic polymorphs, is already used in cutting tool devices [log].
The generally accepted P,T phase diagram of BN shown in Fig. 10 is based on the
work of Wentorf and colleagues, in which melting temperatures of hBN and the
phase boundary h B N - + c B N at high pressures were measured. This phase
boundary extrapolated to room temperature by analogy with the carbon phase
diagram intersects the pressure axis [110], thus implying that hBN is the most
stable modification at ambient conditions. In contrast, it was suggested in some
recent publications [l 1 I] that cBN is stable at ambient pressure at temperatures
up to 1300°C. This suggestion is based on measurements of the thermodynamic
properties of all modifications of BN, where the difference of enthalpies of forma-
tion of cubic and hexagonal BN was found to be negative [l l l]:
AHFBN(298K)- AHfhBN(298K)= -16kJmol-'.
Using these thermodynamic data an alternative phase diagram of BN (also shown in
Fig. 10) was developed.
The technological applications of cBN are similar to those of diamond. However,
because of improved oxidation resistance, cBN allows high-speed cutting or
LXVI Introduction: Novel Ultrahard Materials
grinding of steel-based materials which cannot be machined with diamond. The

discovery of diamond synthesis within the metastable regime of the carbon
phase diagram depicted in Fig. 5 resulted in an increased effort for the deposition
of cBN films under similar conditions. The main processes developed are given
below.
(i) Deposition of cBN by electron cyclotron resonance plasma using NH3 and BF3
as the reactant gases. The films were grown on single crystal Si (100) wafers at a
substrate temperature of 675°C and were characterized by infrared and Auger
spectroscopy. The film thickness was 100400 nm, the average growth rate was
10 nm min-' [I 121.
(ii) Formation of cBN by activated reactive evaporation of H3B03in the presence
of NH,. Cubic BN could be deposited on different substrate materials such
as steel, silicon, glass, or NaCl crystal with a rate of 150 nmmin-' at 450°C
substrate temperature [113].
(iii) Synthesis of cBN by R F plasma CVD under low pressure conditions and
negative self-bias. B2H6 and N2 were applied as the reactants. The cBN films
were obtained on single crystal Si wafers heated at 600°C with a deposition
rate of 6 nm min-' . Cubic BN was verified by transmission electron diffraction
(TED) and microscopy (TEM). The TEM investigations revealed crystallite
sizes of 10-20 nm [ 1141.
+
(iv) Plasma enhanced CVD from BC13 + NH3 + H2 Ar under irradiation of an
ArF excimer laser at 193 nm. The cBN films deposited were heteroepitaxially
grown on Si (100) surfaces at 500-900°C substrate temperatures. Molecular
orbital calculations of the cBN growth mechanism indicate an important role
of atomic chlorine formed during the deposition [I 151.
(v) Finally, cBN was synthesized in the form of films by the dual ion beam deposi-
tion technique. In this process, boron atoms are sputtered from a boron target
by an Ar' ion beam and subsequently deposited on a substrate (Si wafer) simul-
taneously bombarded by N i ions. The deposition rate is 2 .2 4 .6 nm min-' and
film thicknesses of up to 300nm could be achieved [116].
However, besides the formation of cBN, hexagonal hBN is also found in the
deposited films prepared by all of the different methods mentioned above. Vapor-
deposition of cBN films is reviewed in Part I1 by K. Bewilogua and F. Richter.
Additionally, it is also worth mentioning that cBN has been obtained by the
pressure pyrolysis of borazine, B3N3H6, in a hydrothermal apparatus at 25-
100 MPa and 250-700°C followed by a high temperature treatment at 6.5 GPa
pressure in a girdle type apparatus [117]. The decomposition of magnesium boron
nitride (Mg3BN3) in the presence of water and under high pressure ( P > 4GPa)
and high temperature (1400°C) conditions according to the reaction equation
Mg3BN3 + 3H20 * 3 Mg 0 + cBN + 2NH3
also provided cBN [117]. Magnesium boron nitride is a compound known in the
pseudobinary system BN-Mg3N2 and is synthesized from a mixture of hBN and
Mg3N2 at 4GPa and 800°C. The morphology of the cBN crystals formed is
strongly dependent on the temperature and pressure applied. In the case of the
Boron-based Hard Materials LXVII
decomposition at 1400°C and 4 GPa pressure, tetrahedral crystals 10 mm in

diameter were isolated [ 1 181.
By analogy to the body-centered tetragonal carbon, J. Corkill et al. postulated
also a b.c.t.-4 phase of BN with alternating B and N atoms (Fig. 6a) which has
been calculated by the pseudopotential total energy method [119]. While the
calculations for b.c.t.-4 carbon showed a reduced stability with respect to diamond
due to the occupation of antibonding states at the Fermi level, the structure of
b.c.t.4 BN has a small density of states at the Fermi level. Consequently,
b.c.t.-4 BN is estimated to have higher stabilization compared to b.c.t.-4
carbon. The calculated bulk modulus B (268GPa) of b.c.t.-4 BN is significantly
lower than that of the cBN structure (367GPa) owing to the relative openness
of the b.c.t.-4 phase.
Boron-rich Boron Nitrides

Crystalline boron-rich boron nitrides of the compositions B25N up to B53N as
well as amorphous phases of the composition B3N and B5N have also been
synthesized by CVD methods [120,121]. The formation of a rhombohedral struc-
ture of the boron-rich nitride, B4N, is described in [122]. The films deposited at
1200°C on hBN substrates by a tungsten hot filament assisted vapor-phase
reaction
x>l
were characterized by Raman and photoelectron spectroscopy as well as by X-ray

diffraction. The spectroscopic measurements indicate the presence of a B4N phase
with a crystal structure analogous to that of rhombohedral B4C, a compound
which also belongs to ultrahard materials. In this crystal structure, B l l N icosa-
hedra are linked by N-B-N chains as shown in Fig. 11 [122]. Due to this
Figure 11. Model of the unit cell of the B4N structure. The atoms marked in black represent the
nitrogen atoms within the N-B-N-chain. First published in [152] and reproduced with permission.
LXVIII Introduction: Novel Ultrahard Materials
structural feature, the new BN phase can be expected to exhibit interesting hard-
ness properties.
Nitrogen-rich Boron Nitride

Existence of a nitrogen-rich boron nitride denoted as h’BN was suggested by Yo0
et al., who observed appearance of a low-density hexagonal compound after laser
heating of boron in excess of nitrogen in a diamond cell [123]. This compound
was synthesized both in the stability fields of conventional hBN (2 GPa, 1300 K)
and of cubic BN (15 GPa, 1800 K). h’BN is highly transparent and can be recovered
as white polycrystals, which do not convert to other forms of boron nitride at
ambient conditions [ 1231. Unfortunately, chemical composition of the compound
obtained was not determined quantitatively.
Boron Carbonitrides (B,C,N,)

In addition to diamond and the compounds found in the binary phase diagrams of
C-N, B-C, and B-N (Fig. I), materials of the extended ternary system containing
boron, carbon and nitrogen can be also considered as potential candidates with
advanced hardness properties. These compounds are of broad interest because
hexagonal (BN),C, can be considered for applications as high temperature semi-
conductors, especially if they can be altered by doping or structural changes [ 1241.
Moreover, it is expected that the cubic form of ternary compounds or solid solution
with the general formula (BN),C, might exhibit hardness approaching that of
diamond, but will remain strong for steel cutting and withstand oxidation at
higher temperatures than diamond.
Low density B,C,N,, with hexagonal or turbostratic structure can be synthesized
in three different ways:
- nitriding of solid -phase precursors at high temperatures
- using CVD technique
- by pyrolysis of inorganic polymers containing boron, carbon, and nitrogen.
Boron carbonitrides of the general composition B,C,N, were first synthesized by
Kosolapova using the reaction of lampblack and amorphous boron in N2 atmosphere
at 1800-2000°C [ 1251. Most recently, B,CyN, with hexagonal graphite-like sheet
structure was obtained by CVD methods providing thin films at lower temperatures
(1000-1400°C). Typically, BC13/HC=CH/NH3 or BC13/NH3/CH4/H2gas mixtures
are used for the deposition of B-C-N films [124,126-1301. Boron carbonitride pow-
ders could be synthesized in high amounts by the pyrolysis of amine-borane derived
inorganic polymers [ 131, 1321. The molecular amine-boranes, pyridine-borane
- -
(C5H5N BH3) and piperazine-borane (C4H10N2)BH3, form highly cross-linked
polymers by annealing at moderate temperatures (100-200°C) with or without sol-
vents [1331. Subsequently, the polymeric materials can be thermally decomposed
Buron-bused Hard Materials LXIX
r! H \ I \ /
A
I c-c,
H
C N - i - H
H
-110 OC
Figure 12. Schematic representation of the decomposition of pyridine-borane into turbostratic

boron carbonitride, BC4N, with graphite like sheet structure. In the graphitic structure on the
right, only the sp2 o-bonds are shown. First published in [I521 and reproduced with permission.
giving boron carbonitrides, B,C,N2, with different carbon contents. The following
simplified equations represent the overall pyrolysis reactions providing BC4N and
BC2N (Fig. l), respectively:
1050"C/Ar
CN-B.3 BC4N + CH4 + 2H2,
n 1050"C/Ar
NH N-BH3 BC2N + 2CH4 + NH3 + H2.
U
Powder diffraction as well as photoelectron spectroscopy (XPS), electron energy
loss spectroscopy (EELS), and solid state NMR investigations of the synthesized
BC2N and BC4N revealed a graphite-like sheet structure with carbon substituted
for boron and nitrogen and with little sheet-to-sheet registry (turbostratic boron
carbonitride). The pyrolytic conversion of pyridine-borane to turbostratic B-C-N
via the formation of an intermediate polymeric compound is shown schematically
in Fig. 12. These boron carbonitride materials can be considered as solid solutions
of C and BN and are thermally stable up to 1800°C in an Ar atmosphere. At higher
temperatures, phase separation providing crystalline hBN, B4C, and elemental C is
observed (see Fig. 1) [ 1331. Microstructural investigation of the synthesized BC4N
by means of TEM showed also the presence of small amounts of isolated crystals
with cubic symmetry as determined by electron diffraction data [ 132,1341. The dif-
fraction data of the cubic crystals do not correspond to those of diamond or cBN.
Therefore, a cubic boron carbonitride phase is assumed for those single crystals
embedded in a turbostratic BC4N matrix. The most difficult issue in the synthesis
of hexagonal B,C,NZ compounds is proving of a single phase content. Indeed
there is no direct method of distinguishing between hexagonal B. C ,Nz, hBN and
..J
graphite, if all are turbostratically distorted. Andreev and Lundstrom pointed out
that in some of the above described work the available data are insufficient to
make unambiguous conclusions about single phase content [1351. They prepared
hexagonal B,C,N, materials by nitration of boron carbide B4C powder at 2100°C
LXX Introduction: Novel Ultrahard Materials
at different partial nitrogen pressures and found in all cases at least two coexisting
phases. For example, samples prepared at partial nitrogen pressures between 30 and
100% contained a carbon-poor three-dimensionally ordered phase and a carbon-
rich turbostratic phase [135]. Synthesis of ceramic materials from polymers is dis-
cussed in Part I1 by G. D. Soraru and P. Colombo.
The high-pressure and high-temperature formation of crystalline boron carbo-
nitride with a cubic lattice was reported in 1977 by various Russian authors using
hBN + C, hBN + B4C, or cBN + C as the reacting species [136,137]. High micro-
hardness values in the range between 60 and 80 GPa have been measured for these
cubic phase E C - N materials. High-pressure synthesis of the cubic B,C,N, phases
from the starting graphite-like (BN),C1-,x (with x = 0.15, 0.26 and 0.6) prepared
by CVD-method was performed at 14GPa and above 3300K [124]. Although
detailed chemical analysis was not performed, it was assumed that the obtained
material was a solid solution of cubic BN and diamond. X-ray powder diffraction
data suggested that this cubic material exhibits ideal mixing according to the Vegard’s
law and has the same composition as the starting hexagonal material. However, the
author also observed reflexes allowed only for cBN, which he explained by ordering of
the B-N pairs in the lattice of the synthesized cubic (BN),C1 -, materials. Addition-
ally, the author always observed the presence of graphite-like phases in a heat-treated
charge and small amounts of B4C in a few samples [124]. Ideal mixing in the cubic
(BN),C, compounds synthesized in this way was also reported 15 years later by
Kagi et al. [138]. In contrast, measurement of molar volume and of the bulk modulus
of cubic (BN),C1 --x phases synthesized in the YAG-laser heated diamond cell at pres-
sures of 30-50 GPa and temperatures of 2000-2500 K showed strong deviation from
ideal mixing [139]. According to these studies the molar volumes of cubic (BN),yCl
at ambient conditions are larger than that of the ideal solid solution and cubic
(BN),C1_, phases with x < 0.5 have a bulk modulus smaller than that of end-
member cBN [139]. If the first observation holds at pressures and temperatures of
synthesis then a mechanical mixture of diamond and cBN will appear at equilibrium
conditions. This is because high pressures favor formation of denser materials [ 1401.
It follows that this cubic (BN),C1 --x should be a metastable intermediate product of
transformation from hexagonal (BN),C, -, to the mixture of diamond and cBN.
The second observation suggests that the hardness of cubic (BN),C, phases with
x < 0.5 should be smaller than that of cBN, if the systematic approach of Liu
and Cohen [12] is applied. The conclusion of nonideal mixing made earlier [139]
can be called in question because hexagonal materials did not transform completely
into the cubic phases and quantitative chemical analysis of the synthesized cubic
phases was not performed. Nevertheless, metastability of cubic (BN),CI is
supported by the following results of Japanese authors. Sasaki et al. observed
crystallization of hexagonal BC2N in diamond and cBN in the presence of a Co
catalyst at relatively low pressures and temperatures (5.5 GPa and 1400-1600 K)
[141]. Nakano found that upon phase separation of metastable cubic BC2N to
diamond and cBN the composition of residual cubic (BN),C1 -, did not change
(x = 0.33). Accordingly, no other stable phases of type (BN),C, except diamond
and cBN exist at the temperature and pressure conditions of the work (7.7 GPa,
2300-2400°C) [142].
Silicon-based Materials LXXI
Because formation of cubic boron carbonitride is of great fundamental interest

with respect to superhard materials much additional effort is needed to succeed in
the preferential synthesis of the cubic B-C-N phase. As follows from the above
results the most promising way would be synthesis under nonequilibrium conditions
such as flash-heating at static pressures or shock-wave compression [ 1401. Successful
synthesis of cubic BC2,5Nsolid solution in 18% yield by shock-compression of
hexagonal BC2,5Nhas been reported [143]. The material obtained was a single
cubic BC2.5N phase with a diamond-like structure and crystals between 5 and
20 nm in size.
Boron Suboxides
In 1988, A. R. Badzian reported that boron suboxides exhibit exceptional high
hardness values [144]. The suboxides with compositions between B 2 0 and B220
were synthesized by the solid state reaction of elemental boron with B2O3 at
1600-2000°C. Subsequent melting of the reaction product under Ar provides
boron containing 4-5 at.% oxygen corresponding to B I 8 0and BZ20,respectively.
The crystal structure of the boron rich boron suboxides is related to that of a-
rhombohedra1 boron. The microhardness of boron suboxides was determined to
be between 31 GPa and 59 GPa, which is rather close to the hardness of cubic BN
[144, 1451.
Application of high pressure and high temperature allows synthesis of boron
suboxides with the high degree of crystallinity needed for studying the properties
of these compounds. For example, only fine powders with some secondary
amorphous products were obtained in the synthesis of boron suboxide with a
nominal composition B 6 0 at ambient pressure. Additionally, the synthesized
compounds were nonstoichiometric with compositions varying from B600,72to
B600.86.Hubert et al. showed that by reduction of B2O3 by boron at 1700°C and
5.5 GPa a highly crystalline material consisting of multiply twinned particles with
icosahedral symmetry can be obtained. The size of these icosahedral particles was
up to 20-30pm. The measured molar oxygen content was 0.96 f 0.08 (B600.96)
thus approximating the ideal value of 1 [146,147].
The a-rhombohedra1 structure of elemental boron has an electron deficiency that
can be reduced by the incorporation of interstitial atoms in structural voids. Filling
these voids with atoms can result in cross-linking of the boron icosahedra which is
realized for example in the structures of B4C (B12C3), B12P2, B4N, or boron rich
boron suboxides (see Fig. 11) and can contribute to a significant increase of the
rigidity of the solid state structure.
Silicon-based Materials
This section deals with high pressure phases of two silicon based compounds Si02
and Si3N4,which previously did not belong to the family of superhard materials.
LXXII Introduction: Novel Ultrahard Materials
While the a- and p-phases of silicon nitride were known for a long time as hard
materials, the common phases of Si02 (e.g. quartz) lie much further down the
hardness scale (Table 1). The high pressure phase of Si02 with the rutile structure
(stishovite) was first reported in 1961 [148]. Si02-stishovite is stable at pressures
above 11 GPa and persists metastably at ambient conditions. Its density
(4.287gcmP3) is almost 62% higher than that of the ambient pressure quartz
phase (2.648 g ~ m - [149].
~ ) The bulk modulus of Si02-stishovite (298 GPa) exceeds
that of B4C (200 GPa) which hardness is comparable with that of cubic BN 141. Such
a strong increase in density and bulk modulus is due to an increase of the co-
ordination number of Si-atoms from four to six. Recently LCger et al. measured
the Knoop hardness of polycrystalline single phase Si02-stishovite to be between
31 and 35GPa, which exceeds the literature values of 30GPa for B4C and B 6 0
[4]. However, one has to bear in mind that the measured hardness values depend
strongly on the state of the sample and on the specific test method used. Thus to
make reliable comparisons hardnesses of the materials have to be measured using
the same method and taking into account the above mentioned peculiarities.
A change to six-fold coordination of Si-atoms occurs upon transformation of
hexagonal c1- and pSi3N4 to the high pressure cubic phase with spinel structure
observed in the laser heated diamond cell at 15GPa [5]. The technique of C02-
laser heating in a diamond cell is described in Part I by Zerr et al. The density of
the cubic Si3N4is 23% above that of a- or p-phases. According to first-principles
calculations, the elastic bulk and shear moduli of cSi3N4exceed that of Si02-stisho-
vite. Consequently, the hardness of the cubic silicon nitride can be comparable to
that of Si02-stishovite [5]. Detailed discussion of silicon nitride based materials is
given in Part 111 by M. Herrmann.
These results on the high pressure phases of two silicon based materials show a
strong potential of the high pressure techniques in synthesis of new materials with
promising mechanical properties. Furthermore, there are many unexplored paths
for future work, even if some of these ways seem on first glance to be completely
worked out, as for the case of Si02-stishovite.
Concluding Remarks
The experimental and theoretical research activities in the field of materials with hard-
ness comparable to or higher than diamond have been stimulated in recent years by the
discovery and postulation of novel carbon allotropes and novel carbon- and boron-
based materials. Comparison of the experimental and theoretical results obtained in
the past shows that first principles quantum mechanical and semi-empirical calcula-
tions are very useful for the screening of the structural and electronic properties of
new materials. The fascinating variety of new possible carbon allotropes which have
been calculated to be stable relative to diamond or can be stabilized for kinetic reasons
reflects the high potential and the topicality of modern carbon chemistry. Novel
carbon allotropes and carbon and silicon nitrides are subjects of discussion in Part I
in the chapter written by G. Jungnickel et al. and J. E. Lowther.
However, the experimental evidence for the synthesis and structure of dense
superhard C3N4 phases reported in the literature is not unambiguous. Thus,
future research should be focused on the synthesis of sufficient amounts of these
new superhard materials and on their structural characterization. The structures
of all C3N4-polymorphsreported so far still remain to be verified by single crystal
X-ray or electron diffraction. Furthermore, the difficulties in the determination of
reliable hardness values of these new materials not only at room temperature but also
at elevated temperatures have to be overcome. Diamond will not lose its reputation
as the hardest material until the newly developed phases and compounds have been
characterized in detail and can be produced in reasonable amounts.
Acknowledgement
The authors thank J. Nicolich for preparation of crystal structure plots. This work
was supported by the Deutsche Forschungsgemeinschaft under the contract number
Ri 510/14-1.
References
1. F. P. Bundy, H. T. Hall, H. M. Strong, and R. J. Wentorf, Nature, 1955, 176, 51.
2. R. H. Wentorf, J . Chem. Phys. 1957, 26, 956.
3. R. H. Wentorf, J . Chem. Phys. 1961, 34, 809.
4. J. M. Leger, J. Haines, M. Schmidt, J. P. Petitet, A. S. Pereira, and J. A. H. da Jornada,
Nature, 1996, 383, 401.
5. A. Zerr, G. Miehe, G. Serghiou, M. Schwarz, E. Kroke, R. Riedel, H. FueR, P. Kroll, and R.
Boehler, Nature, 1999, 400, 340.
6. J. C. Angus and C. C. Hayman, Science, 1988, 241, 913.
7. K. E. Spear, J . Am. Ceram. SOC.1989, 72, 171.
8. W. A. Yarbrough and R. Messier, Science, 1990, 247, 688.
9. P. K. Bachmann, Adv. Muter. 1990, 2, 195.
10. P. K. Bachmann, Keramische Zeitschrift, 1992, 2, 93.
11. R. Telle, in Materials Science and Technology: A Comprehensive Treatment, R. W. Cahn, P.
Haasen, and E. J. Kramer (Series Eds), Vol. 11, N. V. Swain (Ed.), VCH, Weinheim, 1994,
p. 173.
12. A. Y. Liu and M. L. Cohen, Science, 1989, 245, 841.
13. D. M. Teter and R. J. Hemley, Nature, 1996, 271, 53.
14. J. Hoggin, Chem. Eng. News, May 25, 1992, 20.
15. C. Niu, Y. Z. Lu, and C. M. Lieber, Science, 1993, 261, 334.
16. A. J. Stevens, T. Koga, C. B. Agee, M. J. Aziz, and C. M. Lieber, J. Am. Chem. SOC.1996,118,
10900.
17. D. M. Teter, M R S Bull. 1998, 23, 22.
18. Y. Li, Y . Qian, H. Liao, Y. Ding, L. Yang, C. Xu, F. Li, and G. Zhou, Science, 1998,281,246.
19. U. Helmersson, S. Todorova, S. A. Barnett, J.-E. Sundgren, L. C. Markert, and J. E. Greene,
J. Appl. Phys. 1987, 62, 481.
20. I-W. Chen and A. Rosenflanz, Nature, 1997,389,701.
LXXIV Introduction: Novel Ultrahard Materials
21. S. VepTek, P. Nesladek, A. Niederhofer, F. Glatz, M. Jilek, and M. Sima, Surf. Coat. Technol.
1998, 108-109, 138.
22. H. Holleck, J. Vac. Sci. Technol. 1986, A4, 2661.
23. K.KuBmaul, in Lexikon Werkstoftechnik, H. Grafen (Ed.), VDI, Dusseldorf, 1991, p. 409.
24. K. Niihara, and T. Hirai, J. Muter. Sci. Lett. 1978, 13, 2276.
25. R. W. Rice, C. Cm. Wu, and F. Borchelt, J . Am. Ceram. Soc. 1994,77, 2539.
26. M. L. Cohen, J. Hard Muter. 1991, 2, 13.
27. B. Farid and R. W. Godby, Phys. Rev. B, 1991,43, 14248.
28. R. W. Hertzberg, Deformation and Fracture Mechanics of Engineering Materials, 3rd edn,
Wiley, New York, 1989.
29. A. Kelly and N. H. Macmillan, Strong Solids, 3rd edn, Clarendon Press, Oxford, 1986.
30. E. Knittle, R. M. Wentzcovitch, R. Jeanloz, and M. L. Cohen, Nature, 1989, 337, 349.
3 1. H. B. Huntington, in Solid State Physics: Advances in Research and Applications, F. Seitz and
D. Turnbull (Eds), Vol. 7, Academic Press, New York, 1958, p. 213.
32. 0. L. Anderson, in Physical Acoustics: Principles and Methods, W. P. Mason (Ed.), Vol. 111,
Part B, Academic Press, New York, 1965, p. 43.
33. R. E. Hanneman, H. M. Strong, and F. P. Bundy, Science, 1967, 155, 995.
34. H. Strunz, Mineralogische Tabellen, 7th edn, Akademische Verlagsgesellschaft Geest & Portig
K.-G., Leipzig, 1978.
35. A. Neuhaus, Proc. Intern. Mineralog. Assoc. 1960, 1, 33.
36. H. W. Kroto, J. R. Heath, S. C. O'Brien, R. F. Curl, and R. E. Smalley, Nature, 1985,318, 162.
37. H. W. Kroto, Science, 1988,242, 1139.
38. R. F. Curl and R. E. Smalley, Science, 1988, 242, 1017.
39. W. Kratschmer, L. D. Lamb, K. Fostiropoulos, and D. R. Huffman, Nature, 1990,347, 354.
40. H. W. Kroto, Angew. Chem. 1992, 104, 113.
41. F. Diederich and Y. Rubin, Angew. Chem. 1992, 104, 1123.
42. R. F. Curl and R. E. Smalley, Spektrum der Wissenschaft, 1991, 88.
43. W. Kratschmer and K. Fostiropoulos, Physik in unserer Zeit, 1992, 3, 105.
44. H. R. Karfunkel and T. Dressler, J . Am. Chem. Soc. 1992, 114, 2285.
45. S. Koizumi, M. Kamo, Y. Sato, H. Ozaki, and T. Inuzuka, Appl. Phys. Lett. 1997,71, 1065.
46. P. W. Bridgman, Sci. Am. 1955, 193, 42.
47. R. C. Burns, S. Kessler, M. Subanda, C. M. Welbourn, and D. L. Welch, in Proceedings of the
3rd NIRIM International Symposium on Advanced Materials ( I S A M '96), M. Akaishi et al.
(Eds), NIRIM, Tsukuba, 1996, p. 105.
48. J. E. Butler and H. Windischmann, M R S Bull. 1998, 23, 22.
49. M. Kamo, Y. Sato, K. S. Matsumoto, and N. Setaka, J . Cryst. Growth, 1983, 62, 642.
50. P. K. Bachmann, W. Drawl, D. Knight, R. Weimer, and R. F. Messier, in Diamond and
Diamond-Like Materials Synthesis, G. H. Johnson, A. R. Badzian, and M. W. Geis (Eds),
M R S Symp. Proc., Vol. EA-15, Materials Research Society 1988, 99.
51. S. Matsumoto, Y. Sato, M. Kamo, and N. Setaka, Jpn. J . Appl. Phys. 1982, 21, L183.
52. S. Matsumoto, J. Muter. Sci. Lett. 1985, 4, 600.
53. K. Kurihara, K. Sasaki, M. Kawarada, and N. Koshino, Appl. Phys. Lett. 1988, 6,437.
54. Y. Hirose and N. Kondo, in 35th Spring Meeting of the Japanese Appl. Phys. Society, 1988, p.
434.
55. N. Ohtake, H. Tokura, Y. Kuriyama, Y. Mashimo, and M. Yoshikawa, in Proc. 1st Int. Symp.
Diamond and Diamond-Like Films, J. P. Dismukes et a/. (Eds), The Electrochemical Society,
Pennington, NJ, Proceedings Vol. 89-12, 1989, p. 50.
56. D. M. Gruen, M R S Bull. 1998, 23, 32.
57. D. W. Brenner and J. H. Harrison, Am. Ceram. Soc. Bull. 1992,71, 1821.
58. A. M. Stoneham, I. J. Ford, and P. R. Chalker, MRS Bull. 1998, 23, 28.
59. S. A. Solin and A. K. Ramdas, Phys. Rev. B, 1970, 1, 1687.
60. C. Wild, N. Herres, J. Wagner, P. Koidl, and T. R. Anthony, in Proc. 1st Int. Symp. on
Diamond and Diamond-Like Films, J. P. Dismukes et al. (Eds), The Electrochemical Society,
Pennington, NJ, Proceedings Vol. 89-12, 1989, p. 283.
61. L. Holland, J . Vac. Sci. Technol. 1977, 14, 5.
62. L. Holland, Thin Solid Films, 1979, 58, 107.
References LXXV
63. 0. S. Selwyn and E. Kay, Plasma Chem. Plasma. Process, 1985, 5, 1983.
64. M. Ramsteiner and J. Wagner, Appl. Phys. Lett. 1987, 51, 1335.
65. K. Kobayashi, N. Mutsukara, and Y. Machi, J . Appl. Phys. 1986, 59, 910.
66. F. Jansen, M. Machonkin, S. Kaplan, and S. Hark, J. Vuc. Sci. Technol. 1985, A3, 605.
67. J. W. Rabelais and S. Kasi, Science, 1988, 239, 623.
68. N. Savvides, J . Appl. Phys. 1985, 58, 518.
69. J. C. Phillips, Phys. Rev. Lett. 1979, 42, 153.
70. C. B. Collins, F. Davanloo, T. J. Lee, J. H. You, and H. Park, Am. Cerum. Soc. Bull. 1992,71,
1535.
71. N. N. Matyusenko and V. E. Strel'nitzkii, J . Exp. Theor. Phys. Lett. 1979, 30, 199.
72. R. Hoffmann, T. Hughbanks, M. Kertesz, and P. H. Bird, J . Am. Chem. Soc. 1983,105,4831.
73. M. A. Tamor and K. C. Hass, J . Mater. Res. 1990, 5, 2273.
74. A. Y. Liu, M. L. Cohen. K. C. Hass, and M. A. Tamor, Phys. Rev. B, 1991,43, 6742.
75. A. H. Alberts and H. Wynberg, J. Chem. Soc., Chem. Commun. 1988, 748.
76. A. Y. Liu and M. L. Cohen, Phys. Rev. B, 1992,454579.
77. A. F. Hebard, M. J. Rosseinsky, R. C. Haddon, D. W. Murphy, S. T. Glarum, T. T. M. Palstra,
A. P. Ramirez, and A. R. Kortan, Nature, 1991, 350, 600.
78. M. N. Regueiro, P. Monceau, and J.-L. Hodeau. Nature, 1992, 355, 237.
79. M. N. Regueiro, Adv. Mafer. 1992, 4, 438.
80. R. S. Ruoff and A. L. Ruoff, Nature, 1991, 350, 663.
81. S. J. Duclos, K. Brister, R. C. Haddon, A. R. Kortan, and F. A. Thiel, Nature, 1991, 351,
380.
82. M. L. Cohen, Phys. Rev. B, 1985, 32, 7988.
83. A. Y. Liu and M. L. Cohen, Phys. Rev. B, 1990,41, 10727.
84. A. Y. Liu and R. M. Wetzcovitsch, Phys. Rev. B, 1994, 50, 10362.
85. M. R. Wixom, J . Am. Cerum. Soc. 1990, 73, 1973.
86. L. Maya, D. R. Cole, and E. W. Hagaman, J . Am. Ceram. Soc. 1991, 74, 1686.
87. H. X. Han and B. J. Feldman, Solid State Commun. 1988. 65, 921.
88. T. Sekine, H. Kanda, Y. Bando, M. Yokoyama, and K. Hojou, J . Muter. Sci. Lett. 1990, 9,
1376.
89. J. Kouvetakis, R. B. Kaner, M. L. Sattler, and N. Bartlett, J . Chem. Soc. Commun. 1986, 1758.
90. J. Martin-Gil, F. J. Martin-Gil, M. Sarikaya, M. Qian, M. Jose-Yacaman, and A. Rubio,
J. Appl. Phys. 1997,81, 2555.
91. K. M. Yu, M. L. Cohen, E. E. Haller, W. L. Hansen, A. Y. Liu, and I. C. Wu, Phys. Rev. B,
1994,49, 5034.
92. A. Fernandez, P. Prieto, C. Quiros, J. M. Sanz, J.-M. Martin, and B. Vacher, Appl. Phys. Lett.
1996, 69, 764.
93. Y. Chen, L. Guo, and E. G. Wang, Phil. Mag. Lett. 1997, 75, 155.
94. J. L. He and W. L. Chang, Surf. Coat. Technol. 1998,99, 184.
95. J. H. Nguyen and R. Jeanloz, Muter. Sci.Eng. A, 1996, 209, 23.
96. D. C. Nesting and J. V. Badding, Chem. Muter. 1996, 8, 1535.
97. D. C. Nesting, J. Kouvetakis, and J. V. Badding, in Proceedings ofthe 5th NIRIM International
Symposium on Advanced Materials (ISAM '98), M. Kamo et al. (Eds), NIRIM, Tsukuba,
Japan, 1998, 1.
98. A. J. Stevens, T. Koga, C. B. Agee, M. J. Aziz, and C. M. Lieber, J . Am. Chem. Soc. 1996,118,
10900.
99. V. P. Dymont, E. M. Nekrashevich, and I. M. Starchenko, JETP Lett. 1998,68,498.
100. L. Vel, G. Demazeau, and J. Etourneau, Mater. Sci. Eng. 1991, B10, 149.
101. B. P. Singh, Muter. Res. Bull. 1986, 21, 85.
102. F. Binder, Radex-Rundschau, 1975, 4, 531.
103. F. Thevenot, J . Eur. Cerum. Soc. 1990, 6, 205.
104. A. Lipp, K. A. Schwetz, and K. Hunold, J . Eur. Serum. Soc. 1989, 5, 3.
105. R. R. Wills, Int. J . High Technol. Serum. 1985, 1, 139.
106. T. Ishii, T. Sato, Y. Sekikawa, and J. M. Iwata, J . Cryst. Growth, 1981, 52, 285.
107. T. Matsuda, N. Uno, H. Nakae, and T. Hirai, J . Muter. Sci. 1986, 21, 649.
108. V. L. Solozhenko, V. A. Mukhanov, and N. V. Novikov, Dokludy Chem. 1990,312, 125.
LXXVI Introduction: Novel Ultrahard Materials
109. A. Nagel, Ceram. Tec AG, Plochingen, 1993, private communications.

110. F. R. Corrigan and F. P. Bundy, J . Chem. Phys. 1975, 63, 3812.
111. V. L. Solozhenko, High Press. Res. 1995, 13, 199.
112. S. Y. Shapoval, V. T. Petrashov, 0.A. Popov, A. 0.Westner, M. D. Yoder, and C. K. C. Lok,
New Diamond Sci. Tech. 1991, 1093.
113. K. L. Chopra, V. Agarwal, and D. Vankar, Thin Solid Films, 1985, 126, 307.
114. A. Chayahara, H. kokoyama, T. Imura, and Y. Osaka, Jap. J . Appl. Phys. 1987, 26,
L1435.
115. S. Komatsu, Y. Moriyoshi, M. Kasamatsu, and K. Yamada, J . Phys. D: Appl. Phys. 1991,24,
1687.
116. W. L. Lin, Z. Xia, and Y. L. Liu, Muter. Sci. Eng. 1990, B7, 107.
117. S. Hirano, A. Fujii, T. Yogo, and S. Naka, J. Am. Ceram. Soc. 1990,73, 2238.
118. S. Nakano, H. Ikawa, and 0. Fukunaga, J. Am. Ceram. Soc. 1992,75,240.
119. J. L. Corkill, A. Y. Liu, and M. L. Cohen, Phys. Rev. 1992, B45, 12746.
120. K. Ploog, P. Rauh, W. Stoeger, and H. Schmidt, J. Cryst. Growth, 1972, 13/14, 350.
121. K. Ploog, H. Schmidt, E. Amberger, G. Will, and K. H. Kossobutzki, J. Less-Common Met.
1972, 29, 161.
122. H. Saitoh, K. Yoshida, and W. A. Yarbrough, J. Muter. Res. 1993, 8, 8.
123. C. S. Yoo, J. Akella, and H. Cynn, Phys. Rev. B, 1997, 56, 140.
124. A. R. Badzian, Mat. Res. Bull. 1981, 16, 1385.
125. T. Y. Kosolapova, G. N. Marenko, T. I. Serebryakova, 0. Prilutskii, T. Khorpyakov, and
0. I. Chernysheva, Sov. Powder Metall. Met. Ceram. 1971, 10, 22.
126. A. R. Badzian, T. Niemyski, and E. Olkusnik, in Proc. Third Int. Con$ Chemical Vapor
Deposition, F. A. Glaski (Ed.), Salt Lake City, Utah, 1972, p. 747.
127. J. Kouvetakis, R. B. Kaner, M. L. Sattler, and N. Bartlett, J . Chem. Soc., Chem. Commun.
1986, 1758.
128. R. B. Kaner, J. Kouvetakis, C. E. Warble, M. L. Sattler, and N. Bartlett, Muter. Res. Bull.
1987, 22, 399.
129. T. M. Besmann, J. Am. Ceram. SOC.1990,73, 2498.
130. F. Saugnac, F. Teyssandier, and A. Marchand, J . Am. Ceram. Soc. 1992,75, 161.
131. R. Riedel, J. Bill, and G. Passing, Adv. Muter. 1991, 3, 551.
132. J. Bill, R. Riedel, and G. Passing, Z. Anorg. Allg. Chem. 1992, 610, 83.
133. J. Bill, M. FrieB, and R. Riedel, Eur. J. Solid State Inorg. Chem. 1992, 29, 714.
134. J. Bill, P h B . Thesis, University of Stuttgart 1993.
135. Yu. G. Andreev and T. Lundstrom, J. Alloys Comp. 1994,210, 311.
136. N. N. Sirota, M. M. Zhuk, A. M. Mazurenko, and A. I. Olekhnovich, Vestsi Akad. Nauk
BSSR, Ser. Fiz.-Muter. Nauk, 1977, 2, 111, or CA 1977, 87, 59009j.
137. A. K. Butylenko, G. V. Samsonov, I. 1. Timofeeva, and G. Makarenko, Pis’ma Zh. Tekh. Fiz.
1977,3, 186, or CA 1977,86, 175812.
138. H. Kagi, I. Tsuchida, Y. Masuda, M. Okuda, K. Katsura, and M. Wakatsuki, in Proceedings
of the Joint XV AIRAPT and XXXIII EHPRG International Conference, Warsaw, Poland,
W. A. Trzeciakowski (Ed.), World Scientific Publishing, Singapore, 1995, p. 258.
139. E. Knittle, R. B. Kaner, R. Jeanloz, and M. L. Cohen, Phys. Rev. B, 1995, 51, 12149.
140. V. L. Solozhenko, Eur. J. Solid State Inorg. Chem. 1997, 34, 797.
141. T. Sasaki, M. Akaishi, S. Yamaoka, Y . Fujiki, and T. Oikawa, Chem. Mat. 1993, 5, 695.
142. S. Nakano, in Proceedings of the 3rd NIRIM International Symposium on Advanced Materials
(ZSAM P6), M. Akaishi et al. (Eds), NIRIM, Tsukuba, Japan, 1996, p. 287.
143. T. Komatsu, M. Nomura, Y. Kakudate, and S. Fujiwara, J . Muter. Chem. 1996, 6, 1799.
144. A. R. Badzian, Appl. Phys. Lett. 1988, 53, 2495.
145. I. A. Bairamashvili, G. I. Kalandadze, A. M. Eristavi, J. Sh. Jobava, V. V. Chotulidi, and
Yu. I. Saloev, J. Less-Common Met. 1979, 67, 455.
146. H. Hubert, L. A. J. Garvie, B. Devouard, P. R. Buseck, W. T. Petuskey, and P. F. McMillan,
Chem. Mater. 1998, 10, 1530.
147. H. Hubert, B. Devouard, L. A. J. Garvie, M. O’Keeffe, P. R. Buseck, W. T. Petuskey, and P. F.
McMillan, Nature, 1998, 391, 376.
148. S. M. Stishov and V. Popova, Geokhimiya, 1961, 10, 837.
149. J. R. Smyth and T. C. McCormic, in Mineral Physics and Crystallography: A Handbook of
Physical Constants, T. J. Ahrens (Ed.), American Geophysical Union, Washington, DC,
1995, p. 1.
150. S. Fahy and S. G. Louie, Phys. Rev. B, 1987, 36, 3373.
151. F. P. Bundy, W. A. Bassett, M. S. Weathers, R. J. Hemley, H. K. Mao, and A. F. Goncharov,
Carbon, 1996, 34, 141.
152. R. Riedel, Novel ultrahard materials, Adv. Muter. 1994, 6, 549.
Structures and Properties
Part I
Edited by Ralf Riedel
Copyright Q WILEY-VCH Verlag GmbH, D-69469 Weinheim
(Federal Republic of Germany), 2000
1 Structural Chemistry of Hard Materials

W. Jeitschko, R. Pottgen, and R.-D. Hoffmann
1.1 Introduction
Most hard materials are formed from high-melting elements in the middle of the
periods of the transition metals and the main group elements. The stability of these
elements and compounds arises from a maximum of filled bonding bands and a mini-
mum of filled antibonding bands. The hardness of many elements and materials is
frequently not well known, but usually hardness and high melting points correlate,
and melting points are better known. Figure 1 shows the melting points of the
elements and some well-known high-melting and hard materials. At this point it is
worthwhile to note that refractory carbides like HfC or TaC have significantly
higher melting points than the frequently used crucible ceramic alumina, A1203.
Most well-known hard materials have surprisingly simple crystal structures. Here
we give an overview of the crystal chemistry of hard materials. Different aspects of
the various crystal structures are discussed, such as close-packed structures with
filled octahedral voids like titanium carbide, the tetrahedral arrangements in Sic
or the three-dimensional inorganic networks in a- and P-Si3N4. Also we briefly
mention the synthesis and some applications of various hard materials.
The basic crystallographic data of the most important hard materials are listed in
Table 1. Further access to crystallographic data can be gained through various com-
pilations. The older literature has critically been evaluated in the Strukturberichte
[ 13, the Structure Reports [2] and in the Landolt-Bornstein [3]. The non-oxidic
materials are well covered in several editions of Pearson’s Handbook [MI.The
various structure types of these materials have been critically compiled by Parthe
and coworkers [7]. Access to the more recent literature is best achieved through
4000 4000
- TaC
. =- ZrC
-* 3000
Figure 1. Melting points of most elements and some selected high-melting inorganic materials.
4 1 Structural Chemistry of Hard Materials
Table 1 Crystallographic data of hard materials
Compound Pearson Space Z a b C V Ref.

symbol group bm) (pm) bm) (nm3)
c-c cF8 Fdjm 8 356.7 0.0454
h-C hP4 P63/mmc 4 252.2 41 1.9 0.0262
c-Sic cF8 F43m 4 434.9 0.0823
6H-Sic hP12 P63mc 6 308.1 1511.7 0.1243
C-BN cF8 F43m 4 361.5 0.0472
B13C2 hR45 Rjm 3 563.3 1216.4 0.3343
B6° hR42 R h 6 538.6 1231.9 0.3095
TiB2 hP3 P6/mmm 1 303.1 322.9 0.0257
FeB oP8 Pnma 4 549.5 294.6 405.3 0.0656
CrB OC8 Cmcm 4 296.9 785.8 293.3 0.0684
R-MOB tI16 141lamd 8 310.5 1697.0 0.1636
YB6 cP7 PmSm 1 412.8 0.0703
wc hP2 ~6m2 1 290.6 283.7 0.0208
w2c hP6 P3ml 2 299 473 0.0366
TIC cF8 Fmjm 4 432.9 0.08 1 1
Fe3C oP16 Pnma 4 507.9 673.0 451.4 0.1543
cr3c2 oP20 Pnma 4 553.3 282.9 1147.2 0.1796
cr23c6 cF116 Fmh 4 1066.0 1.2113
cr7c3 oP40 Pnma 4 452.6 701.0 1214.2 0.3852
W3Fe3C cF112 Fdjm 16 1108.7 1.3628
MoSi2 t16 I4/mmm 2 320.2 785.2 0.0805
CrSi2 hP9 P6222 3 442.1 635.1 0.1075
TiSi, oF24 Fddd 8 826.7 480.0 855.1 0.3393
Ti3SiC2 hP12 P63/mmc 2 306.8 1766.9 0.1440
Mo5Si3C hP18 P63/mcm 2 728.6 504.6 0.2320
a-Si3N4 hP28 P31c 4 775.2 561.9 0.2925
p-Si3N4 hP14 p63 l m 2 759.5 290.2 0.1450
AIN hP4 P63mc 2 311.0 498.0 0.0417
5-Fe2N oP12 Pbcn 4 443.7 554.1 484.3 0.1191
&-Fe3N hP8 P6322 2 469.8 437.9 0.0837
y'-Fe4N cP5 Pm3m 1 379.0 0.0544
Ti2N tP6 P42lmnm 2 494.5 303.4 0.0742
Ti,AIN cP5 PmSm 1 411.1 0.0695
Ti2AIN hP8 P63/mmc 2 299.4 1361 0.1057
Zr3A1N oc20 Cmcm 4 336.9 1149.8 898.3 0.3480
a-A1203 hR30 R% 6 476.0 1299.3 0.2549
a-Be0 hP4 P63-mc 2 269.8 427.7 0.0270
MgO cF8 Fm3m 4 421.1 0.0747
MgA1204 cF56 Fdh 8 807.5 0.5265
Ti02, rutile tP6 P42/mnm 2 459.3 295.8 0.0624
Ti07. anatase t112 14, lamd 4 378.4 951.5 0.1362
TiO;; brookite oP24 Pbca 8 918.4 544.7 514.5 0.2574
y-Zr02 cF12 Fmh 4 506.5 0.1300
P-Zr02 tP6 P42/nmc 2 364 527 0.0698
u-Zr02 mP12 p21 1. 4 514.5 520.8 531.1 0.1404
p = 99..23"
CeOz cF12 Fm3m 4 552.6 0.1687
Tho2 cF12 Fmh 4 560.4 0.1760
the ICSD data base [8]. The X-ray powder patterns of most solid state materials are
stored in the ICDD files [9]. Knowing the crystal structure they can also be
generated by the program LAZY PULVERIX [lo].
1.2 Diamond and Diamond-Related Structures

1.2.1 The Crystal Structure of Diamond
Carbon in the structural form of diamond is the only element used industrially as a
hard material. Each year about ten tons of natural diamond and about twenty tons
of synthetic diamond (produced via high temperature high pressure synthesis) are
marketed as hard materials. While pure diamond is transparent, a yellow tint results
from the replacement of some carbon atoms by nitrogen, and a blue, yellow, or even
green tint through substitution of carbon by boron atoms. Polycrystalline diamond
with impurities, used as an abrasive, is often black.
The structure of cubic diamond (a = 0.3567nm) is presented in Fig. 2. Each
carbon atom is tetrahedrally coordinated by four other carbon atoms with the
C-C single bond distance of 0.1545nm [ll]. These tetrahedra are connected via
shared corners. Hexagonal puckered nets can be visualized (perpendicular to the
three-fold axes) where the condensed c6 rings are all in chair conformation.
These layers are stacked in the sequence ABC, ABC.
Starting from well-crystallized graphite with a large degree of AB, AB stacked
planar layers, a hexagonal modification of diamond (a = 0.2522 nm, c = 0.41 19 nm)
can be obtained under static pressure at high temperature [12]. This modification
has an AB, AB stacking sequence of the chair-like layers (Fig. 2). It is frequently
called hexagonal diamond or lonsdaleite. Small particles of lonsdaleite have been
discovered in the Canyon Diablo meteorite crater [12].
The C-C distance in lonsdaleite is equal to that in diamond [l 11. One quarter of the
c6 rings are in chair and three quarters are in boat conformation. The lonsdaleite
structure is possibly idealized. An imperfect ‘lonsdaleite’ could be formed by stacking
Figure 2. Crystal structures of cubic diamond (left-hand side, space group F d h ) and hexagonal
lonsdaleite (space group P6Jmmc). The tetrahedral environments of the carbon atoms and the
stacking sequences ABC, ABC and AB, AB, respectively, are indicated.
6 1 Structural Chemistry qf Hard Materiuls
Figure 3. Crystal structure of c-BN. Two unit cells of the cubic structure are shown at the right-hand
side. In the middle the structure is projected along one of the threefold axes (cube diagonal). At the
left-hand side the tetrahedral environments around one kind of atoms is emphasized. Boron and
nitrogen atoms occupy equivalent positions, hence the structure may be thought to be composed
of NB4/4 as well as BN4/4tetrahedra.
faults of the ABC stacking variant resulting in a significant degree of AB stacking.

Different stacking faults of such chair-like layers are frequently also observed in
thin films of diamond prepared by chemical vapor deposition (CVD) techniques.
The initial X-ray investigations of several samples of synthetic hexagonal diamond
indicated mixtures of both modifications. One problem of such X-ray powder
investigations is the overlap of several broad reflections from both modifications.
1.2.2 The Isoelectronic Compounds c-BN and S i c

Cubic boron nitride (c-BN) [6,13] is almost as hard as diamond and it is sometimes
called inorganic diamond. c-BN can be synthesized from hexagonal graphite-like
boron nitride (h-BN) under high pressure (50-90 kbar) and high temperature
(1500-2200°C) with the help of small amounts of a catalyst such as Li3N [14]. c-
BN crystallizes with a zinc blende type structure. The boron and nitrogen atoms
are ordered on the carbon positions of the diamond structure, resulting in a
three-dimensional network of corner-sharing BN414 and NB4/4 tetrahedra (Fig. 3)
with B-N distances of 0.157nm. Due to the ordering of boron and nitrogen
atoms, the space group symmetry is reduced: c-BN crystallizes in the noncentro-
symmetric space group F43m which is a translationengleiche subgroup of index 2
of Fdjm. The density of cubic boron nitride (3.45gcmp3) is slightly lower than
that of diamond (3.51 g ~ m - ~ At
) . higher temperatures c-BN is less oxidizable
than diamond. A wurtzite type modification of boron nitride corresponding to
the hexagonal lonsdaleite modification of diamond is also known. It is formed by
shock compression at higher pressure than necessary for c-BN [15].
An inexpensive diamond substitute is silicon carbide S i c which is produced in
megaton amounts per year. S i c can be synthesized from Si02 and carbon at high
+
temperatures: Si02 3C -+ Sic + 2CO. Between 2200 and 2400°C the com-
mercially sold 6H-modification [ 161 is obtained, while cubic 3C-Sic [ 171 (zinc
blende type) forms at lower temperatures (<2O0O0C) [18]. Pure S i c is colorless
c-Sic GH-SIC
Figure 4. Crystal structures of 3C- and 6H-Sic. The tetrahedral environments of the silicon atoms
and the stacking sequences are indicated.
and transparent, while technical Sic (carborundum) is black with a bluish tint, due to
a small excess of silicon. S i c is a very non-corrosive material with a remarkable
chemical and thermal stability.
The cubic modification of 3C-Sic (space group F43rn) is isotypic with c-BN. In this
modification the chair-like layers are stacked in ABC sequence (Fig. 4) as in cubic dia-
mond with a Si-C distance of 0.189nm in all polytypes. This is somewhat less than
the average distance of 0.195 nm, between the C-C (0.154nm) and Si-Si (0.235 nm)
single bond distances in the elements [ 1 11. For Sic a huge number of different stacking
sequences have been observed. In 1972 Parthe listed some 77 different polytypic forms
[19]. With the exception of the cubic 3C-Sic modification all these polytypes have
rhombohedral, trigonal, or hexagonal symmetry. The c dimensions of their respective
hexagonal cells are all multiples of 0.252 nm. They range from 0.505 nm to 149.5 nm,
all with the same a dimension of 0.308 nm. Growth by screw dislocations has been dis-
cussed as an explanation for the long-range order of the many polytypes [20,21]. As
an example the structure of 6H-Sic (space group P6gnc) is presented in Fig. 4. Here,
the stacking sequence is ABCACB.
Various symbols have been introduced to describe the many possible stacking
sequences of (not only) the modifications of Sic. As an example, the different
notations are presented in the following scheme for a structure with fourteen layers:
A B C B A B C B A B C A C B, A B C B A B C B A B C A C B A B C notation
14H Ramsdell
The simplest one is the ABC notation, where the letters A, B, C are used for the
three possible positions of the close packed layers, starting with the letter A for the
first layer. For more complex stacking sequences the Jagodzinski-Wyckoff notation
has been introduced. This notation is only applicable to structures which allow not
more than three stacking positions and where two subsequent layers cannot have the
same stacking position. B layers of a stacking sequence ABA, where the two
neighboring layers (A) are displaced by the same amount in the same direction
are designated with the letter h (hexagonal); and B layers of a sequence ABC
where the two neighboring layers (A and C) are displaced by opposite amounts
are assigned the letter c (cubic).
These sequences can be represented in a further abbreviated form by summing up
equal subsequences counting from one h to the next one, so the sequence ABC-
BABCBABCACB can be represented by ( h ~ ) ~ ( h c cUsing ) ~ . the Zhdanov notation
this can be further abbreviated by stating the number of layers and how often
they are repeated. Thus ( h ~ ) ~ ( h cis
c )represented
~ by 2432in the Zhdanov notation.
A more general notation was introduced by Ramsdell. This notation lists the
number of slabs in one unit cell and the respective symmetry with a capital letter
(C for cubic, H for hexagonal, T for trigonal, and R for rhombohedral). Since
there is sometimes more than one way to arrange a certain number of layers in a
trigonal or hexagonal cell this notation may be ambiguous.
1.3 Crystal Chemistry of Borides and Boron Carbides

The technically important boron carbide, usually designated with the formula B4C,
is synthesized from boron oxide and coke at about 2500°C [22]: 2B203 7C + +
+
B4C 6CO. The high costs for elemental boron make the direct synthesis
unprofitable. The product has a large homogeneity range [23] extending from
carbon-rich B8C2 to boron-rich B13C2; the latter may be considered as the ideal
composition. This material is almost as hard as diamond and cubic BN, but less
expensive. At temperatures above 1200°C its hardness even exceeds that of diamond
[24]. It has a surprisingly low density of only 2.51 gcmP3. B13C2itself shows an
extremely high chemical inertness. It is only attacked in alkaline media in the
presence of oxygen. Boron carbide then forms borates and carbonates, such as
+ +
B4C 40H- 4 0 2 + B40:- +
C0:- +
2H20. Boron carbide is widely used as
an abrasive. In this respect it is better than silicon carbide; however, it is more
expensive.
The unit cell of the ideal B13C2structure (space group R j m ) [25] is shown in Fig. 5.
B12icosahedra occupy the corners of the rhombohedral cell. They are linked directly
via B-B bonds and in addition by linear C-B-C units. Hence, the compound may
be represented by the formula (B12)CBC.The C-B distance of 0.143 nm indicates
some double bond character.
The homogeneity range of this boron carbide results from a substitution of some
boron atoms of the chains and/or of the icosahedra by carbon. Also, some related
Figure 5. The crystal structure of B13C2(‘B4C’)in the hexagonal setting. All boron atoms indicated
by filled circles form (empty) icosahedra. Other boron atoms (open circles) are located in the centers
of linear C-B-C units. At the left-hand side the stacking (ABC) of this rhombohedra1 structure is
indicated.
aluminum containing compounds, A1B40C4 [26] and A12,1BS1C8[27] are known,

which again contain BI2 icosahedra as the most important structural units. Such
compounds are reported to contain C-B-C, C-B-B, C-Al-B, or C-Al-C
chains between the icosahedra. A detailed list of these structures has been given
by Wells [28]. The structure of so-called ‘P-AlBl2’ was redetermined recently,
resulting in the composition A13B48C2[29]. This structure may be described as an
orthorhombically distorted variant of tetragonal boron. Besides aluminum, also
silicon, titanium, vanadium, copper, and iron can be incorporated in this structure
tYPe.
A considerable hardness was also reported for the boron suboxide B 6 0 [24]. The
structure of B 6 0 (B6P type, space group Rjm) was refined from X-ray powder data
[30]. Recently twinned crystals of B 6 0 with icosahedral morphology were grown at
high temperature and high pressure 1311. Again in this compound BI2icosahedra are
the most important structural unit. Within one layer, the B12 icosahedra are
connected via oxygen atoms. In this way each icosahedron has six oxygen neighbors
and each oxygen atom is connected to three icosahedra (Fig. 6). Subsequent layers
are stacked in ABC sequence and connected via boron-boron bonds.
Of the transition metal borides, the titanium boride TiB2 is the technically most
important compound. In the industrial process, TiB2 is synthesized by carbothermal
reduction of boron oxide in the presence of titanium dioxide [22,32]:
Ti02 + B 2 0 3 + 5C -+ TiB2 + 5CO. TiB2 crystallizes with A1B2 type structure [4-
6,331 (space group P6/mmm, Fig. 7), where the boron atoms fill trigonal prisms
formed by the titanium atoms. Each boron atom has three boron neighbors in a
trigonal planar arrangement, forming a two-dimensional honeycomb network
with B-B distances of 0.175 nm.
TiB2 has excellent thermal and chemical stability up to 1700°C. It has a relatively
low density of 4.52gcmP3 and a high melting point of 3225°C. ZrB2, HfB2, VB2,
NbB2, CrB2, MoB2, and WB2 are isotypic with TiB2 and A1B2; however, most of
10 I Structural Chemistry of Hard Materials
Figure 6. Crystal structure of B,O. On the right-hand side of the figure one layer of oxygen-bridged
(empty) B12icosahedra is outlined. The hexagonal cell of this rhombohedra1 structure and the stack-
ing of subsequent layers is presented at the left-hand side.
these diborides have a higher density and their production is more expensive. Never-
theless, some of them are used for special applications [22,32].
Apart from the diborides, also some monoborides are technically important.
Most of them are used as hard components in composite materials. One notable
compound is FeB. It has a remarkable adhesiveness on iron-based materials and
thus increases their hardness. Such hard coatings are produced via CVD techniques.
As an intermediate product elementary boron is obtained through the reaction
+ +
2BC13 3H2 -, 2B 6HC1 between 500 and 900°C, which subsequently reacts
with iron to form the FeB coating [32,34].
The transition metal monoborides crystallize with three different structure types.
The orthorhombic structures of CrB [35] and FeB [36] are presented in Fig. 8. The
boron atoms have a trigonal-prismatic metal environment and two near boron
neighbors, thus forming infinite boron zig-zag chains. The B-B distances are
comparable to the shorter B-B distances in the modifications of elemental boron.
The Cr-B and Fe-B distances within the trigonal prisms range from 0.212 to
0.223 nm. Each boron-centered trigonal prism shares two rectangular faces with
neighboring prisms forming infinite prism columns, which extend along the z
direction in CrB and along y in FeB. The difference between the two structures is
Figure 7. The AlB, type structure of TiB,. A projection along the hexagonal axis is shown on the
right-hand side and a perspective view is presented on the left. The two-dimensional boron network
is emphasized in both drawings.
X C - 4V
CrB FeB
Figure 8. The crystal structures of CrB (space group Cmcm) and FeB (Pnrna). Each boron atom in
these orthorhombic structures is located in a trigonal prism of metal atoms. Each prism shares two
rectangular faces with adjacent prisms. Thus, the boron atoms form infinite zig-zag chains.
the way in which the infinite columns are connected with each other (Fig. 8). The
monoborides VB, NbB, TaB, WB, NIB, and the high-temperature beta modification
of MOB also crystallize with CrB type structure, while TiB, MnB, and COB are
isotypic with FeB [6].
The structure of (low-temperature) a-MOB[37] is closely related to that of CrB. It
is also composed of infinite columns of boron-centered molybdenum prisms (Fig. 9).
However, in contrast to the CrB type, the boron zig-zag chains are now rotated by
90" with respect to each other (as a consequence of the four-fold screw axis). The
Mo-B distances within the trigonal prismatic units range between 0.223 and
0.265 nm. The infinite boron zig-zag chains have B-B distances of 0.174 nm. WB
is isotypic with a-MOB [37].
The rare earth hexaborides, well known for their superior electron emitting
properties [38], are also quite hard materials. The hardness of YB6 is reported to
be comparable to that of TiB2 [32]; however, such borides are quite expensive.
YB6 is isotypic with CaB6 and LaB6 [3840], where the boron atoms form B6
Mo
B
Figure 9. Crystal structure of tetragonal wMoB (space group I4,lumd).

12 1 Structural Chemistry of’ Hard Materials
X J-Y
Figure 10. Crystal structure of cubic YB6 (space group Pm3m). The three-dimensional network of
B, octahedra and the B24 coordination polyhedra around the yttrium atoms are emphasized.
octahedra which are bonded to six other octahedra forming a three-dimensional

cubic network (Fig. 10). The yttrium atoms occupy the large cavities formed by
this network. They have 24 boron atoms in their coordination shell. Alternatively,
the YB6 structure can be described as a CsC1-like packing of yttrium atoms and B6
groups.
1.4 The Structures of Transition Metal Carbides

Cemented carbides form one of the most important groups of hard materials
[41,42]. The carbides WC, Tic, and TaC are the technically most important ones.
They are produced in amounts of several thousand tons per year. VC, NbC, ZrC,
HfC, Mo2C, and the chromium carbides are somewhat less important. The extreme
hardness and high melting points of many transition metal carbides were already
recognized in the 1890s by Moissan [43].
The transition metal carbides, especially WC, are used as the hard components in
cutting tools, wire drawing dies, drills, and other wear resistant materials. For this
purpose they are embedded in a binder matrix of a lower melting metal of the iron
group, mainly cobalt, which supplies the necessary tenacity. The composition of
these composite materials is around 92% WC, 8% Co. Usually these cemented car-
bides are prepared by powder metallurgical techniques, starting from the elemental
components. Alternatively one can reduce the respective transition metal oxides
with pure graphite at about 2O0O0C, for example Ti02 + 3C + T i c 2CO. Very +
non-abrasive coatings of TIC can be obtained by chemical vapor deposition, e.g.
Tic& + CH4 + T i c + 4HC1 [22].
WC is the technically most important transition metal carbide. It adopts a very
simple structure (Fig. 11, space group P6m2) [44,45], which may be described as
a defect-A1B2 structure in which every other boron position is unoccupied. This
results in a symmetry reduction. In space group P6m2 (a translationengleiche sub-
group of index 2 (t2) of P6/mmm) the two-fold 2d position of P6/mmm is reduced
to two one-fold positions. This allows the introduction of ordered defects. The
I .4 The Structures of Transition Metal Carbides 13
OC
Figure 11. Crystal structure of WC. A projection along the hexagonal c axis is shown on the left-
hand side of the drawing and a view of the unit cell is presented on the right-hand side. The trigonal
W6C and WC6 prisms are outlined.
carbon atoms in WC are isolated (no C-C bonds) within the trigonal tungsten
prisms. This is a significant difference when compared with the various boride
structures discussed above. The filled prisms are stacked one upon each other by
sharing common triangular faces. Thus, filled and unfilled columns alternate. The
W-C and shortest W-W distances are 0.220 and 0.284 nm, respectively.
Besides the trigonal prismatic coordination of the carbon atoms, octahedral
coordination is observed, especially for the carbon atoms in the carbides of the
early transition metals [46]. Important examples are the NaCl type carbides TIC,
ZrC, HfC, VC, NbC, and TaC [42], where the metal atoms have a cubic close-
packed arrangement, and the tungsten carbide W2C, where the tungsten atoms
form a hexagonal close packed structure [47,48].
It is well known that close packed atoms form octahedral voids. The number of
these voids equals that of the number of close packed atoms. In the NaCl type struc-
ture of TIC all octahedral voids formed by the cubic close packed titanium atoms are
filled by carbon atoms, and the Ti& octahedra share only edges. In contrast, in
W2C the tungsten atoms form a hexagonal close packed arrangement and the
carbon atoms fill only half of the octahedral voids. The w6c octahedra are
linked not only via common edges, but also by common faces. The Parthe-Yvon
rule states that carbon atoms of transition metal carbides occupying octahedral
voids of face sharing transition metal octahedra, never occupy both of these adja-
cent octahedra [47]. This is demonstrated in Fig. 12 using the Jagodzinski-Wyckoff
notation for the close packed metal atoms. For every close packed transition metal
( T )layer with an h type packing only one of the adjacent face sharing octahedral
voids can be filled. For the carbide with the composition T2C this unoccupied
void is marked with an asterisk in Fig. 12. Examples for the carbide structures
shown in that figure are W2C with (anti-)Cd12 type structure Mo3C2 (hcc),
(Ta2V)C2 (hhc), V4C3, and TIC [47, 491. Perspective views of the TIC (space
group Fm3m) and W2C ( P h l ) structures are shown in Fig. 13 with an emphasis
on the octahedral carbon coordination. The Ti6C octahedra in TIC share all
edges, while edge-sharing only within the ab plane is observed for the W6C octa-
hedra in W2C.
It is interesting to compare the structures of WC and W2C. The main difference
between these two tungsten carbides is the carbon coordination: trigonal prisms of
tungsten atoms in WC and octahedra in W2C with W-C distances of 0.220 and
14 1 Structural Chemistry of Hard Muteriuls
I
h'
I
h'
*
'h
T2C T3C2 T3C2 T4C3 TC

h hcc hhc hhcc C
Figure 12. Stacking sequences of the hexagonal metal atom layers in several carbides with close
packed transition metals [47]. The structures are represented by cuts along the (1 10) planes of the
respective hexagonal cells. The transition metal (7') and carbon atoms are drawn as large open
and small filled circles, respectively. A 'forbidden' carbon position, discussed in the text, is
marked with an asterisk in the structure of T2C.
0.209 nm, respectively. The difference in the W-C distances may be rationalized by
geometrical considerations: in a hard sphere model (equal W-W distances) trigonal
prismatic voids are larger than octahedral voids. However, there are also differences
in the coordination numbers: the tungsten atom in WC has six carbon neighbors, in
W2C it has only three. In addition to the W-C bonding both compounds have
strong W-W bonds. Each tungsten atom has eight (WC) or twelve (W2C) tungsten
neighbors with W-W distances ranging between 0.284 and 0.300 nm, comparable
to the W-W distances in the cubic body centered structure of the element
Tic w2c
Figure 13. Crystal structure of cubic T i c (NaC1 type) and hexagonal W2C (Cd12 type). Some
carbon-filled Ti6 and W6 octahedra are emphasized.
(8 x 0.274 nm and 6 x 0.316 nm). W,C has a carbon content of only 3.2 weight-%.
It has a larger microhardness than WC, but it is also more brittle [42].
The transition metal carbides with the composition T2C are sometimes called
hemicarbides. In addition to tungsten, such carbides are also formed by molyb-
denum, vanadium, niobium, and tantalum. A large number of somewhat differing
crystal structures have been reported for these, all with very similar X-ray patterns,
corresponding to the hexagonal close packed arrangement of the metal atoms. The
differences arise through slight distortions and/or larger unit cells of lower sym-
metry, which result through differing order of occupied and empty carbon positions
[50]. Usually the Parthk-Yvon rule (of two face sharing metal octahedra only one
can be occupied by carbon atoms) is obeyed. The different structures may be
stable in differing ranges of temperature and they may also be stabilized by inter-
stitial impurities like nitrogen or oxygen.
Besides the systematic carbon defects due to the hexagonal stacking sequences,
carbon defects also occur by nonoccupancy of ‘allowed’ octahedral sites. Thus,
the compositions TC, T3C2, T4C3,and T2C discussed above are the upper limits
of the homogeneity ranges. Cubic titanium carbide for example has a homogeneity
range from TIC to about TiCo.5.At high temperature the defects are randomly
distributed, while an ordering of the defects is to be expected for thermodynamic
reasons at low temperature.
As examples, partial phase diagrams of the systems titanium-carbon and
vanadium<arbon are presented in Fig. 14. Most TCI-,y phases were not
intensively studied. The VC, system is by far the best investigated system. An
ordering of the carbon atoms was found for the compositions VsC7 [53-561 and
v6c5 [57,58]. The ordering of the defects influences the positions of the vanadium
atoms. The empty v6 octahedra are bigger than the filled ones [59]. In addition to
the long-range order established for the compositions V8C7 and v6c5 short-range
order has been studied by electron and neutron diffraction in the vicinity of the
ordered structures [52, 601. Long-range order has also been reported for Nb&5
[50,61], TIC,-, (x M 1/2) [50,62-641 and other defect carbides and nitrides
[50,65-691.
The transition metal carbides and nitrides have often been called interstitial
compounds [70]; however, this is somewhat misleading. The small boron, carbon,
or nitrogen atoms certainly occupy octahedral or trigonal prismatic voids of the
metal sublattice, but the arrangement of the metal atoms themselves is different
from that of the element. In the monocarbides the transition metal atoms show
cubic close packing. However, titanium, zirconium, and hafnium are packed
hexagonally and vanadium, niobium, and tantalum are body centered cubic [ 11.
Thus, these monocarbides are inorganic compounds with their individual crystal
structures and they should not be considered as an interstitial compound of a
transition metal host lattice.
Chemical bonding in several transition metal carbides was theoretically investi-
gated by quasi self consistent augmented-plane-wave (APW) calculations [71-73]
and by the extended Hiickel method [74]. These calculation indicated a charge
transfer from the early transition metal to the carbon atoms. A crystal orbital
overlap population analysis (COOP) revealed strong bonding T-T and T-C
-
9 3200
I
3200 4
-s
Y
3000
3
Liquid
2800 2800
2600 2600
2400 2400
2200 2200
2000 2000
1800 1800
1600 1600
1400 1400
1200 1200
1000 1000
-
800 800
600 600
0 10 20 30 40 50 60 70 0.70 0.75 0.80 0.85 0.90 0.95
Ti atomic % carbon C/V Ratio
42 43 44 45 46 47 48
atomic % carbon
Figure 14. Partial phase diagrams of the systems titanium-carbon [51] and vanadium-carbon [52].
In the latter diagram, regions with short-range (SRO) and long-range order (LRO) are indicated.
interactions. For these calculations usually full occupancy of all atomic sites of the
NaCl type structure in assumed, a supposition which may not be entirely correct. In
NbO with a defect-NaC1 type structure 25% of the niobium and oxygen positions
are vacant, and the vacancies are ordered [75].
Carbon plays an important role in steels, both in the form of solid solution and as
component of carbide phases. The cubic face centered modification of iron (y-Fe)
dissolves as much as 8 at-% (2 weight-%) carbon, which randomly occupies the
octahedral voids of the cubic close packed iron atoms. This cubic phase is called
austenite. On quenching, this phase distorts by a displacive (‘martensitic’) phase
transition to the corresponding tetragonal structure (martensite). The degree of
the distortion is proportional to the carbon content. For a carbon content of
zero, the distortion extrapolates to zero, that is, the lattice constants of the (tetra-
gonal) martensite extrapolate to those of pure y-Fe [76,77].
Cementite Fe3C [78] plays an important role in steels. This iron carbide is signi-
ficantly harder than pure iron. Its orthorhombic crystal structure is presented in
Fig. 15. The best structure refinement of a cementite type structure seems to have
been done for Fe2,7Mno.3C[79]. The carbon atoms occupy trigonal prismatic
voids of the iron atoms with Fe-C distances of 0.201 and 0.202nm. Three
additional iron neighbors at 0.237(2 x) and 0.28 1 nm complete the coordination
shell of the carbon atoms, resulting in tricapped trigonal prisms. The trigonal
prisms are condensed via common triangular faces. Columns of such trigonal
Figure 15. Crystal structure of cementite, Fe3C (space group Pnma). The carbon filled trigonal Fe6
prisms and one empty Fe6 octahedron are emphasized.
prisms are connected via Fe-Fe bonds forming empty Fe6 octahedra. The Fe-Fe
distances between the two crystallographically different iron atoms range from 0.250
to 0.268 nm. The shorter ones compare well with the Fe-Fe distance of 0.248 nm in
b.c.c. iron [l 11.
The various chromium carbides are relatively hard and brittle. They significantly
increase the hardness and pyrophoric stability of carbon rich hard materials. These
compounds are known as Stelliten. The hardness of alloyed steels [9] results from
several chromium-iron double carbides of compositions (Fe, Cr)3C2,(Cr, Fe)23C6r
and (Fe, Cr)7C3.These mixed carbides crystallize all in the structures of the respec-
tive pure chromium carbides with a mixed occupancy of the chromium positions by
chromium and iron atoms.
Cr3C2 [80] is the technically most important chromium carbide. It is used as a
cemented carbide with nickel as binder [41]. Its orthorhombic structure is shown
in Fig. 16. The striking structural motifs of Cr3C2 are carbon centered trigonal
prisms of the chromium atoms. These trigonal Cr6C prisms are connected by shar-
ing common faces as well as common edges. The two crystallographically different
carbon atoms in Cr3C2have Cr-C distances from the center to the corner of the
Figure 16. Crystal structure and coordination polyhedra of the chromium atoms of Cr3C2 (space
group Pnma). The carbon filled trigonal Cr6 prisms are also emphasized.
Crl CR
Cr3 C r4
Figure 17. Projection of the crystal structure of cubic Cr23C6(space group Fm3m). Some carbon
filled square antiprisms are emphasized at the left-hand side. The coordination polyhedra of the
chromium atoms are also shown.
prisms ranging from 0.202 to 0.213 nm. In addition, the C1 atoms have two further
chromium neighbors outside the rectangular faces of the prisms at 0.229 and
0.262nm, and the C2 atoms have a seventh chromium neighbor at 0.226nm. The
shortest Cr-Cr distances in Cr3C2 of 0.250nm are exactly as long as in b.c.c.
chromium [l 11. Again, the stability of this peculiar carbide structure arises from
strongly bonding metalkarbon and metal-metal interactions.
The cubic structure of the chromium carbide with the highest metal content,
Cr23C6[81] is shown in Fig. 17. Here, the carbon atoms occupy square antiprisms
of the chromium atoms with Cr-C distances of 0.21 1 and 0.213 nm. The square
antiprisms are condensed via common edges and common corners. The Cr-Cr
distances for the four crystallographically different chromium atoms range from
0.242 to 0.293 nm. Each chromium atom has at least four chromium neighbors at
short distances. The shortest Cr-Cr contacts are significantly shorter than in
b.c.c. chromium (0.250 nm) [ 1 11, indicating strongly bonding Cr-Cr interactions.
In the structure of Cr13C6 the carbon filled square antiprisms are built by the Cr3
and Cr4 atoms. Crl and Cr2 have exclusively chromium neighbors in their coordi-
nation shell. The Crl atoms have twelve chromium neighbors in a cuboctahedral
environment, while the 16 chromium atoms around the Cr2 atoms form a Frank-
Kasper [82,83] polyhedron.
The structure of Cr7C3[84] is presented in Fig. 18. Structural elements are carbon
filled trigonal prisms of chromium atoms which are condensed via common edges
and corners. The Cr-C distances within the trigonal prisms range from 0.197 to
0.21 1 nm, essentially the same range as in Cr3C2.All chromium atoms in Cr7C3
have at least three carbon neighbors and the Cr-Cr distances range from 0.249
to 0.276nm. The predominant features of chemical bonding in Cr7C3 are similar
to CrZ3C6and Cr3C2,that is, strong Cr-C and Cr-Cr interactions.
Besides the chromium-iron carbides, the so-called q-carbides, like W3Fe3C
[85,86] and W3C03C [87] play an important role in steels [42]. They are not
separately produced. The q-carbides (and nitrides) comprise a wide variety of
different compositions T3T$X,T 6 T i x , and T4TiX, where T = Mo, W; T' = Fe,
Figure 18. Crystal structure of orthorhombic Cr7C3(space group Pnma). The carbon filled trigonal
prisms are emphasized.
Co, Ni; X = C (and sometimes also nitrogen or even oxygen, e.g. Ti4Fe200.4[SS]).
In the cubic structure of the compounds T3T&Y and T6TLX the metal atoms
occupy the same atomic positions, but the structures differ in their X positions,
which are either sixteen- or eightfold [86,89]. In both cases, however, the C(N)
atoms occupy octahedral voids formed solely by the molybdenum or tungsten
atoms.
As an example, the structure of W3Fe3C is presented in Fig. 19. At first sight it
seems very complex; however, a description by two independent, interpenetrating
three-dimensional networks makes this structure type comprehensible. One network
consists of corner-shared stelh quadranguh [90] of iron atoms (right-hand part,
Fig. 19), and the second network is composed of corner-sharing w6/2c octahedra.
The latter network is different from that known for the Re03 structure. The two
Figure 19. Crystal structure of W3Fe3C.In the middle of the drawing all atoms of this cubic struc-
ture (space group Fd3rn) are shown. The carbon-filled octahedra of tungsten atoms are emphasized
on the left-hand side, only the condensed Fed tetrahedra are drawn on the right-hand side of the
figure.
crystallographically different iron atoms have six iron neighbors each, at Fe-Fe
distances ranging from 0.238 to 0.250nm. In view of the Fe-Fe distance of
0.248 nm in b.c.c. iron [l 13, these contacts in W3Fe3C certainly indicate strong
Fe-Fe bonding. In addition, there is W-W and W-Fe bonding.
As it turns out, in most ternary phases the carbon atoms are surrounded by those
metal atoms forming the more stable (i.e. higher melting) carbide. The W-C
distances of 0.206nm within the W6& octahedra of W3Fe3C, are essentially the
same as in the W6/3Coctahedra of W2C(0.209 nm) [48]. The remarkable segregation
of the W3Fe3Cstructure into an iron and a W6/2Csublattice parallels the stability of
the binary carbides; that is, high-melting WC and metastable Fe3C.
1.5 Silicides and Silicide Carbides of Transition Metals

The melting points and the hardness of transition metal silicides are lower than
those of the respective carbides [22]. Nevertheless, some of these silicides find
applications as hard materials, since they exhibit a remarkable resistance against
acids and they are also stable in air up to high temperatures because they form
dense coatings of silica and silicates. A large disadvantage of the silicides, however,
is their extreme brittleness. Today mainly MoSi, and the isotypic tungsten
compound WSi, have technical importance. They are used mostly for protective
coatings for molybdenum and tungsten wires and sheets. For this purpose MoSi2
+ +
can be prepared by CVD techniques: Mo 2SiC14 4H2 + MoSi, 8HC1. Their +
use as components of cermets is less important. TiSi2 has gained some importance
for the construction of turbines [91].
Three closely related structure types were found for the transition metal disilicides:
TiSi2 [92], CrSi2 [93], and MoSi2 [94]. The common structural motif of the three struc-
tures is a two-dimensional close-packed layer of composition 7Si2 as outlined in
Fig. 20. Within these layers each transition metal atom has six silicon neighbors.
The structures of TiSi2, CrSi2, and MoSi, then result from different stackings of
these 7Si2 layers. The stacking, however, is not close-packed. The layers are stacked
in such a way that the transition metal atoms of one layer are situated (in projection)
between two silicon atoms of an adjacent layer. Thus, the transition metal atoms fall
above the points B, C, or D, indicated as small dots in Fig. 20. In this way each transi-
tion metal atom obtains ten silicon and four transition metal neighbors. Each silicon
atom has five transition metal and five silicon neighbors. The stacking sequences are
AB, AB for MoSi,, ABC, ABC for CrSi2, and ABCD, ABCD for TiSi2. Again, it is
important to note that these structures should not be confused with the close-packed
layers described earlier, in which a stacking AB, AB or ABC, ABC is hexagonal or
cubic, respectively, for dense packed layers. With the disilicides these correspond to
tetragonal (AB, AB) and hexagonal (ABC, ABC) structures.
For a more familiar view of these structures we now turn to the unit cells. The
tetragonal crystal structure of MoSi2 (space group I4/mmm) is presented in
Fig. 21. It may also be considered as derived from the body-centered structure of
I S Silicides and Silicide Carbides of Transition Metals 21
MoSi,, AB
CrSi,, ABC TiSi,, ABCD

Figure 20. The crystal structures of MoSi2, CrSi,, and TiSi2. All of these structures contain
hexagonal close-packed layers TS4; one of these is shown in the upper left-hand corner. The
positions of the transition metal atoms of adjacent layers above and below are marked with small
black dots in this drawing. The three structures have the stacking sequences AB (MoSi,), ABC
(CrSi2), and ABCD (TiSi,), respectively. The hexagons connecting the silicon atoms around a
transition metal atom serve as guides for the eye.
tungsten by an ordered arrangement of the molybdenum and silicon atoms on the

tungsten sites. This leads to a tripling of the tungsten subcell. Due to the different
sizes of molybdenum and silicon the cubes are strongly compressed ( c / o = 2.45
instead of 3.0 for the ideal arrangement). The molybdenum atoms have ten silicon
MoSi, CrSi, TSi,
Figure 21. Crystal structures of tetragonal MoSi2, hexagonal CrSi,, and orthorhombic TiSi,. The
C N 10 polyhedra of silicon atoms around the transition metal atoms and parts of the two-
dimensionally close packed 7Si2 layers are outlined for each structure.
neighbors in the form of a bicapped square prism. These polyhedra are condensed
via common edges and faces.
The unit cells of the orthorhombic TiSi2 (space group Fddd) and hexagonal CrSi2
(space group P6222) structures are also presented in Fig. 21. For both structures, the
7Sil0 polyhedra and parts of the T-centered Si6 rings are outlined. The Si-Si
distances range between 0.247 and 0.279 nm. These bonding Si-Si distances are
all significantly longer than the Si-Si bond length of 0.235 nm in elemental silicon
[I 11. This has to do with the fact, that the coordination number (CN) of the silicon
atoms in the transition metal disilicides is considerably higher (CN 10) than in the
diamond structure of elemental silicon (CN 4).
Besides Ti5Si3C1- with partially filled Mn5Si3type structure (see below), Ti3SiC2
is the only other phase in the ternary system Ti-Si-C [95-971. Ti3SiC2crystallizes in
the form of thin platelets which can be plastically deformed [98]. It combines refrac-
toriness (m.p. M 3200°C) and stability against aggressive media with thermal shock
resistance. A high potential as a component of ceramics and composite materials
was discussed [99,100]. The electronic structure of Ti3SiC2 was investigated by
self-consistent full-potential linear muffin-tin orbital calculations [ 1011.
The structure of Ti3SiC2is shown in Fig. 22. The titanium and silicon atoms are
close packed with the stacking sequence ( h h h ~ )Thus,
~ . Ti3SiC2adopts a structure
typical for a metal-rich carbide. The titanium octahedra are filled by carbon
atoms, and these Ti6C octahedra are condensed via common edges, forming
double layers which are separated by the silicon atoms. No siliconxarbon bonds
are observed. In compounds of similar composition Si-C bonds (0.193 nm) have
been found only in U3Si2C2[102].
Figure 22. Crystal structure of Ti3SiC2 (space group P6,lmmc). The close-packed titanium layers
containing the carbon atoms are shown on the left-hand side. The edge-sharing Ti& octahedra
are emphasized. On the right-hand side a (110) cut through the hexagonal cell is shown. The
titanium and silicon atoms form close-packed layers with the stacking sequence hhhc where the
silicon atoms correspond to the second h.
1.6 Nitrides 23
Mol Si
Mo2 C
Figure 23. Crystal structure of Mo5Si3C(space group P6,lmcm). On the left-hand side of the draw-
ing a projection of the structure along the hexagonal z axis is presented. The carbon atoms (not
shown) are situated inside the emphasized Mo6 octahedra. These share faces, thus forming infinite
chains, which extend along the hexagonal axis. The near-neighbor coordinations of all atoms are
presented at the right-hand side of the figure.
The molybdenum silicide carbide Mo5Si3C [I031 belongs to a large family of

compounds with structures derived from the well known Mn5Si3 type structure
[104], also called D8* or Nowotny phases. The carbon atoms of Mo5Si3C occupy
octahedral voids formed by the molybdenum atoms. Today more than 300 com-
pounds [5] with unfilled, filled, or partially filled Mn5Si3 structure are known. As is
usually the case for compounds with a high content of metal atoms, in the structure
of Mo5Si3C all atoms have high coordination numbers. The Mo2 atoms (Fig. 23)
form strings, which extend along the c axis with rather short Mo2-Mo2 distances
of 0.252nm. In addition, the Mo2 atoms have 6 Si and 6 Mol neighbors. The silicon
atoms are coordinated by 9 Mo and 2 Si atoms. The Mol atoms form chains of face-
sharing octahedra, which are surrounded by the Si and Mo2 atoms. These octahedra
are filled with carbon atoms. According to the Parthe-Yvon rule [47] the carbon
positions should not be fully occupied and indeed the structure determination of
this compound showed considerable deviations from the ideal composition
corresponding to the formula Mo4.8Si3C0.6[ 1031. The structural chemistry of
compounds with filled Mn5Si3 structure is complex, because the octahedral voids
cannot only be filled by small impurity atoms like carbon, nitrogen, or oxygen, but
also by transition metal and main group elements, such as Hf5Sn3Cu [lo51 and
Ti5Ga4[106]. This gives many possibilities for deviations from the ideal composition.
1.6 Nitrides
In elemental form, carbon has many applications in both modifications as diamond
and as graphite. In contrast, the neighboring element nitrogen is a gas. Nevertheless,
a-Si3N4 P-Si3N4
Figure 24. Crystal structures of trigonal a-Si3N4 (space group P31c) and hexagonal P-Si3N4
(P63/m), emphasizing the corner-sharing SiN4 tetrahedra.
both elements form hard materials, which are structurally closely related, and both
form such refractory compounds with main-group elements (e.g. Sic, Si3N4)and
with transition metals (e.g. WC, TiN).
1.6.1 Nitrides of Main Group Elements

Silicon nitride, Si3N4,is the technically most important nitride of the main group
elements. It is dimorphic. Usually both forms are obtained together during the
preparation, for example by reaction of the elements at 1200°C. The low tempera-
ture a-modification is transformed to the P-modification by heating above 1650°C.
However, the inverse reaction has not been observed. At low preparation tempera-
tures the a-modification predominates. The proportions of a-Si3N4 and p-Si3N4in
mixtures can reliably be determined by X-ray powder diffraction. A high amount of
a-Si3N4is desired, since it significantly increases the sintering activity [ 107,1081.
The crystal structures of a-Si3N4 [lo91 and p-Si3N4 [7,110] are presented in
Fig. 24. Both modifications have a complex three-dimensional network of corner-
sharing SiN4 tetrahedra with an average Si-N bond length of 0.174nm. The
structure of the low-temperature modification has two crystallographically different
silicon atoms, while only one silicon position occurs in the high-temperature
modification. In both modifications the nitrogen atoms have more or less
trigonal-planar carbon coordination.
The high stability of silicon nitride is due to a complex interplay of predominantly
covalent bonding and a high degree of condensation of the corner-sharing SiN4
tetrahedra (three tetrahedra share one common corner). The high corrosion stability
of Si3N4results from the formation of a thin surface coating of SO2. In mixtures
with aluminum oxide, Si3N4 is used as Sialon mixed ceramic Si3N4-A1203.
Today, Si3N4 is the most frequently utilized nitridic material. It is used for the
construction of bearings, pistons, and turbines.
1.6 Nitrides 25
Another main group element nitride with considerable hardness is aluminum

nitride, AlN. It is important as a ceramic insulator with high thermal conductivity.
It has the hexagonal wurtzite type structure (h-ZnS) [l 111. In contrast to the high-
pressure compound c-BN, it is prepared by various methods at ambient pressure
[22, 1121.
Many attempts have been made in recent years to prepare carbon nitrides with the
ideal composition C3N4 in analogy to Si3N4 [113-1171. So far only thin off-
stoichiometric films with considerable nitrogen deficiency have been obtained. Ab
initio calculations have predicted that this nitride could be isomorphic with
(a-Si3N4 and/or b-Si3N4. A cubic modification of C3N4 with defect-zinc blende
type structure has also been discussed, and it was predicted that such a compound
could be even harder than diamond. This, however, is improbable for the simple
structure-chemical reason, that a compound containing only carbon and nitrogen
has too many electrons. In all of the structures proposed for C3N4 the nitrogen
atoms are three-coordinate with a nonbonding (lone) pair of electrons as the
fourth ‘ligand’, thus not utilizing one potential coordination place. In contrast,
the diamond structure has the ideal number of four electrons per atom, and all of
these can be used for bonding interactions.
1.6.2 Transition Metal Nitrides

Transition metal nitrides are used as hard surface coatings. They also play a role in
steels. The most important compounds are the NaCl type mononitrides of titanium,
zirconium, hafnium, niobium, and tantalum. They are isotypic with the correspond-
ing carbides (Fig. 13). Titanium nitride can be obtained by reduction of titanium
+ + +
dioxide: 2Ti02 N2 4C + 2TiN 4CO. For industrial applications several
composite ceramics like A1203/TiN/TiC and Si3N4/TiN have been developed.
Surface coatings of titanium nitride are produced via CVD techniques by the gas
+ +
phase reaction 2TiC14 H2 2NH3 + 2TiN + 8HC1. Such TIN coatings have
yellow color, while mixed nitrides like TiN/AlN are blue. The color of such coatings
can be varied by a substitution of some nitrogen by carbon or oxygen atoms. It also
depends on the concentration of unoccupied nitrogen sites x of TiNl --x.Thin films
of TIN are used as hard, abrasion-proof surface coatings with almost golden color,
also for decorative purposes, like for watches. An overview of the industrially
important nitrides was given by Benesovsky [118].
The mononitrides and nitride carbides have, like the carbides discussed earlier,
large homogeneity ranges, for example the composition of the NaCl type titanium
nitride TiN, p x extends from stoichiometric TiNl.o to TiNo,5[119]. At low tem-
peratures the vacant nitrogen sites can be expected to be ordered, and indeed a
tetragonal superstructure with a doubled c axis has been reported for this phase
for the composition (Ti2N) [120]. The degree of the nitrogen deficit strongly
influences the hardness and the melting point of these compounds. Also, the nitro-
gen atoms in TIN can to some degree be replaced by carbon and/or oxygen atoms
resulting in Ti(C, N, 0) materials. Chemical bonding in these nitrides is similar to
that of the corresponding carbides TIC and HfC [71-741.
A B A B A B
Figure 25. Crystal structure of c-Fe2N. The iron atoms form hexagonal close-packed layers which
extend along the yz-plane of this orthorhombic structure (space group Pbcn). In the x direction these
planes have the stacking sequence AB, AB. Hence, the iron atoms form a distorted hexagonal close-
packed arrangement where the nitrogen atoms fill half of the octahedral voids.
Nitrogen also plays a role in steel, not only as a component of solid solutions in
carbides, but especially in the form of hard surface coatings. The crystal structures
of the respective hard materials are not all well known. In the present section we
discuss some representative structures that were determined reliably. The nomen-
clature of the iron nitrides derives from the phase diagram of the binary system
iron-nitrogen. Thus, the a- and y-phases are solid solutions of nitrogen in the a-
and y-modifications of iron.
The structure of E,-Fe2N [I211 was established recently from a combination of
neutron diffraction and high-resolution synchrotron powder data. c-Fe2N crystal-
lizes with anti-a-PbO, type structure as outlined in Fig. 25. The structure consists
of edge- and corner-sharing Fe6N octahedra with Fe-N distances ranging from
0.189 to 0.201 nm. The iron atoms form a hexagonally close-packed array and the
nitrogen atoms fill half of the octahedral voids in an ordered manner.
The structures of E-Fe3N and y'-Fe4N are also based on close-packed arrange-
ments of iron atoms [122]. In &-Fe3Nthe iron atoms are hexagonal close-packed
and nitrogen atoms occupy only corner-sharing octahedra. The ordered arrange-
ment of the nitrogen atoms results in a & times larger a axis. In this structure
the deviations of the iron atoms from the atomic positions in an ideal hexagonal
close-packed structure are small.
The iron atoms in y'-Fe4N have a face-centered cubic arrangement, and the nitro-
gen atoms occupy octahedral voids (Fig. 26). This results in a three-dimensionally
infinite arrangement of all-corner-sharing Fe6N octahedra with Fe-N distances
of 0.190nm, slightly shorter than the average Fe-N distance of 0.195nm in
E,-Fe2N. Since the nitrogen atoms occupy only one of the four octahedral voids
formed by the face-centered cubic iron atoms, the space group symmetry is lowered
from Fmjm to Pm%.
In addition to the congruently melting (m.p. x 3300°C) NaCl type phase TiNl --x
(x = 0.5 to 1.O) the phase diagram of the system titanium-nitrogen contains at least
three other nitrides, all of which are stable at temperatures of less than 1300°C[119].
Of these the nitride Ti2N crystallizing with (anti-)rutile type structure (Fig. 27) is
known for some time [123]. In this structure the titanium atoms form octahedra
which are filled by nitrogen atoms. The Ti6N octahedra are linked via common
1.6 Nitrides 27
Fe4N1-x li3AIN
Figure 26. The crystal structures of y'-Fe4N and Ti,AIN (space group P m h ) . In the binary nitride
the nitrogen atoms occupy one quarter of the octahedral voids formed by the cubic close-packed
iron atoms. In the ternary compound the nitrogen atoms fill those octahedral voids which are
formed solely by the titanium atoms.
edges, thus forming chains which extend along the c axis of the tetragonal cell
(Fig. 27). These chains are further condensed by sharing corners, and in this way
a three-dimensional network is formed. For the other two titanium nitrides the
structure determinations resulted in the compositions Ti3N2- and Ti4N3- x.
These nitrides are stable at only very limited temperature ranges between 1050
and 1300°C [119,124,125].
Considering composites and thin-film coatings containing TIN and AlN, one has
to be aware of the ternary compounds Ti3A1N [126], Ti2AlN [127], and Ti3A12N2
[126]. Of these the latter is thermodynamically stable only between 1200 and
1300°C. In all three structures the metal atoms are close packed. While the structure
of Ti3A12N2has a large cell with a ten-layer stacking sequence, the metal positions in
Ti3AlN and Ti2AlN correspond to cubic (ABC) and hexagonal (AB) close packed
structures (Fig. 28). In both of these the nitrogen atoms are situated in octahedral
voids formed solely by the titanium atoms. The cubic structure of Ti3A1N is
known as perovskite (CaTi03 type) structure and Ti2AlN is an H phase (Cr2A1C
type). There are many representatives of these carbides T3MX and T2MX where
T are early transition metals, mainly from groups 14 and 15, M are main-group
elements, especially Al, Ga, In, Sn, Pb, and X = C, N [5, 128-1311.
Layers of zirconium and aluminum atoms can also be discerned in the structure
of Zr3AlN [132]. In this complex nitride the nitrogen atoms are again situated in
Figure 27. Crystal structure of Ti2N with tetragonal (unti-)rutile type structure (space group
P42/mnm).The edge- and corner-sharing Ti6N octahedra are emphasized.
Y41lOfl YL X
Ti,AIN Ti,AIN
Figure 28. Crystal structure of Ti3AlN (CaTiO,, perovskite type) and Ti2AlN (Cr2AlC, H phase
type structure). In both structures the metal atoms are close packed. In the cubic perovskite type
structure the positions of the close-packed metal atoms correspond to those of the Cu3Au structure
(space group Pm3m) with the stacking sequence ABC, ABC. In the hexagonal Cr2A1Ctype structure
of Ti2AIN the stacking sequence of the close-packed metal atoms is AB, AB.
octahedral voids formed solely by the transition metal atoms. These Zr6N octahedra
share edges and corners, thus forming two-dimensionally infinite sheets with
aluminum layers in between them (Fig. 29). It is striking that the carbon and
nitrogen atoms in complex carbides and nitrides, containing transition metal and
main-group elements, are usually situated in octahedral voids formed by the
“I “I ”‘I “I “I
i W
Figure 29. Crystal structure of Zr,AlN (space group Cmcm). The layers of edge- and corner-sharing
Zr6N octahedra are emphasized.
1.6 Nitrides 29
transition metals. Other examples already discussed above are the structures of
Ti3SiC2 [98] and Mo5Si3C [103]. Another example is the filled P-manganese type
structure of Nb3A12N and Mo3A12C [133]. Also, ,as already mentioned, in the q-
W3Fe3C type carbides and nitrides the carbon atoms are situated in octahedral
voids formed by those transition metal atoms which form the more stable (high-
melting) binary carbides. It can be assumed, that these octahedral Tgc and T6N
groups are very stable and that they possibly exist already in the melts.
1.6.3 Perspectives: Nitridosilicates

Nitridosilicates of the alkaline earth and rare earth metals have been investigated
intensively in the past five years. A recent review by Schnick and Huppertz [134]
summarizes the rich crystal chemistry and some potential applications for this
class of compounds. Such nitridosilicates can be synthesized via two different
routes. One possibility is the direct reaction of binary silicides under flowing
nitrogen at high temperature like 3CeSi2 + 1 1/2N2 + Ce3Si6NI1[135]. A more
sophisticated method is the reaction of silicium diimide with an alkaline earth
+
element like Ba 7Si(NH)24 BaSi7Nlo+ 2N2 + 7H2. The latter reaction is
carried out in tungsten crucibles under flowing nitrogen in a high-frequency furnace
[136] at temperatures up to 1600°C.
The nitridosilicates are highly stable compounds. They might find applications as
hard materials possibly in combination with Si3N4. In their structures the silicon
and nitrogen atoms form highly condensed networks of corner-sharing SiN4-
tetrahedra. The alkaline earth and rare earth atoms are embedded within the cages
of these tetrahedral networks. The structures of Sm3Si6N11type Ce3Si6NII [135]
and Ca2Si5N8[136] are shown as examples in Fig. 30. The Si-N distances in the
various nitridosilicates range from 0.164 to 0.180 nm. The different connectivities of
the SiN4 tetrahedra in this large family of structures are reviewed in reference 134.
e N
Ca2Si5N8 Ce,S,iN,
Figure 30. Crystal structures of monoclinic Ca2Si,N, (space group Cc) and tetragonal Ce3Si6NI
(Sm3Si6NII type, space group P4bm). The three-dimensional networks of comer-sharing SiN4
tetrahedra are emphasized.
1.7 Oxide Ceramics

1.7.1 Hard Ceramics of Main Group Elements
One technically very important hard ceramic is corundum, a-A1203.Due to its out-
standing resistance to abrasion, wear, and corrosion, its high thermal stability and
high electrical resistance, its compatibility with living tissue, and last but not least
its availability and relatively moderate price, it has many applications, for example
as polishing powder, material for crucibles, cutting tools, spark plugs, substrate for
electronic circuits, and artificial hip-joints [18,22, 137, 1381. The gemstones ruby
and sapphire are essentially a-A1203. The red color of ruby is due to a small
amount of Cr203 in solid solution. Blue sapphire contains minor amounts of
iron and titanium. Such crystals are grown also industrially by the Verneuil
technique. Large amounts of aluminum oxide are produced for the production
of aluminum. By heating A1(OH)3/A10(OH) mixtures to 400°C one obtains the
relatively soft y-modification of A1203,while hard a-A1203is produced by heating
above 800°C.
In the crystal structure of a-A1203[I391 the oxygen atoms form a hexagonal close
packed array in which the aluminum atoms occupy two-thirds of the octahedral
voids. In Fig. 31 the AIOs octahedra are emphasized. These octahedra share
faces, edges, and corners. Each oxygen atom has four aluminum neighbors in
approximately tetrahedral arrangement (2 x 0.185 nm, 2 x 0.197 nm).
Beryllium oxide is a high melting compound with an exceptionally high thermal
conductivity, very low electrical conductivity, and a remarkable chemical stability.
The breaking strength of Be0 is comparable to that of a-A1203, however, the
industrial production of Be0 is more expensive. The high toxicity of beryllium
compounds, for example abrasive dust of BeO, prevents a broad use of t h s material.
~ O ~group R3c). Pairs of face-sharing

Figure 31. The rhombohedra1 crystal structure of C L - A ~(space
octahedra extend along the threefold axis. Octahedra of adjacent pairs share corners and edges.
Two-dimensionally infinite layers of condensed octahedra may be discerned. Such packages are
indicated by braces. At the right-hand side such a layer is projected along the [110] direction.
Figure 32. The wurtzite type structure of cL-BeO, space group P63nrc. The three-dimensional net-
work of condensed Be04-tetrahedra is outlined. The stacking of these tetrahedra has the sequence
AB, AB. The corresponding stacking ABC, ABC is known to occur for the zinc blende (sphalerite)
structure (Fig. 3 ) .
Be0 is a dimorphic oxide. The low-temperature a-form (<2O0O0C) crystallizes

with the hexagonal wurtzite type structure [140]. Each beryllium atom has a slightly
distorted tetrahedral oxygen coordination (1 x 0.162 nm and 3 x 0,164 nm). These
tetrahedra share all corners with other tetrahedra, thus forming a three-dimensional
network of condensed Be04-tetrahedra. It can easily be seen (Fig. 32) that the z axis
of this structure is polar: the Be04 tetrahedra all point in the direction of the c axis.
The high-temperature p-form of Be0 is stable above 2000°C. It cannot be quenched.
In its tetragonal structure [I411 the arrangement of the oxygen atoms is the same as
in rutile, TiOz.
Magnesia, MgO, is industrially produced by thermal decomposition of mag-
nesium carbonate. It has a higher corrosion stability than a-A1203. Depending
on the thermal treatment two forms of MgO are obtained: sintered magnesia
through an annealing process between 1700 and 2000°C or molten magnesia pre-
pared in an arc-melting furnace between 2800 and 3000°C [18]. Sintered magnesia
is used as crucible material in metallurgical furnaces. Melted magnesia is very
expensive and used only for special applications as an insulating material at
high temperature.
MgO crystallizes with NaCl type structure [5, 1421 (space group Fmjm) which is
shown in Fig. 13 for titanium carbide. Each magnesium atom has octahedral oxygen
coordination with Mg-0 distances of 0.210 nm. These MgOh octahedra are
condensed via all edges.
Another relatively hard corrosion resistant material is the complex oxide
MgA1204 which is well known as spinel. Compared with a-A1203 spinel has a
lower thermal shock stability. The crystal structure of spinel [143] can be derived
from an almost ideal cubic close packed arrangement of oxygen atoms. The mag-
nesium cations fill one eighth of the tetrahedral voids and the aluminum cations
occupy one half of the octahedral voids. For electrostatic reasons the smaller
A13+ cations fill the larger octahedral voids. Alternatively, one can describe the
spinel structure with small building blocks (Fig. 33). Half of the magnesium
atoms build an f.c.c. cell which consists of eight smaller cubes. The latter are
alternatingly occupied by Mg04 tetrahedra and distorted A1404 cubes, denoted as
A and B in Fig. 33.
4Mg + 4Mg04 + 4A1404 = 8MgA1204

Figure 33. The crystal structure of cubic MgA1204(spinel, space group F d h ) . For details see text.
1.7.2 Transition Metal Oxides

Of the transition metal oxides, the highest melting and most stable ones are those of
the titanium group: Ti02, ZrOz, Hf02. Chemically related are the fluorite (CaF2)
type compounds Ce02 and Tho2. In this context it is worthwhile to remember
that for good reasons up to 1940 thorium was positioned directly under hafnium
in the Periodic Table. These oxides have many applications, for example Ti02 is
used as a white pigment, calcium- or yttrium-containing Zr02 as solid electrolyte,
or Zr02 and Thoz as high-melting materials. They are also quite hard, although
they are not used primarily as hard materials [18,221.
Titanium oxide TiOz occurs as a mineral in three well-known modifications:
rutile, anatase, and brookite. Today rutile is by far the most important white pig-
ment (others are anatase, ZnS, ZnO, BaS04). As a pigment rutile, when compared
to anatase, has superior quality. It has the highest density of the three forms of Ti02
and it has also certain applications as an inexpensive hard ceramic. To our
knowledge the p-T diagram of TiO, is not firmly established. However, it is
known that under normal conditions rutile is the most stable form. On heating
both anatase and brookite irreversibly transform to rutile. For comparison, we
briefly discuss the structures of all three modifications of Ti02. In all of these the
titanium atoms have somewhat distorted octahedral oxygen coordination and the
oxygen atoms have a trigonal, more or less planar titanium environment.
The most simple structure of Ti02 is the rutile structure (Fig. 34). All of the Ti06
octahedra share two opposite edges with adjacent Ti06 octahedra, in this way
forming chains, which extend along the z axis of this tetragonal structure [144]. In
the other directions the Ti06 octahedra share corners, thus forming an infinite three-
dimensional framework. Alternatively to the description by condensed Ti06 octa-
hedra, the rutile structure can also be derived from a distorted hexagonal close-
packed arrangement of oxygen atoms in which one half of the octahedral voids
are filled with titanium atoms in an ordered manner.
In contrast to rutile, anatase [145] and brookite [146] are derived from distorted
cubic close-packed oxygen arrangements in which again half the octahedral voids
are filled by titanium atoms. The voids are filled in such a way, that the octahedra
I . 7 Oxide Ceramics 33
21- +y Brookite
Figure 34. The crystal structures of orthorhombic brookite (space group Pbca), tetragonal anatase
(14,lamd), and tetragonal rutile (P4*/mnrn).The edge- and corner-sharing Ti06 octahedra and the
more or less distorted trigonal-planar environments of the oxygen atoms are emphasized.
share three common edges with other octahedra in brookite and even four common
edges in anatase. According to the Pauling rules for ionic crystals, the stability of
such an atomic arrangement decreases as the number of common edges increases,
so the repulsion of the titanium atoms increases. The Ti-0 distances in anatase
are 0.193nm (4x) and 0.198nm (2x), nearly identical with rutile (4 x 0.195nm,
2 x 0.198nm). In brookite on the other hand, they vary from 0.187nm to
0.204nm. As a consequence, one would expect the average Ti-0 distance of
0.196 nm in brookite to be significantly greater than the average Ti-0 distances
in rutile (0.1959nm) and anatase (0.1949nm), which is not the case, probably
because the structure refinement of brookite, carried out from film data, was not
that accurate.
In this context it is worth mentioning that the stishovite form of SiOz [147] with
rutile type structure is a very hard material. Recent investigations on polycrystalline
cubic Zr02
tetragonal Zr02 X--lY
L--
. I
- - - _ _ _ _ _ I
monoclinic Zr02
Figure 35. The crystal structures of the cubic, tetragonal, and monoclinic forms of Zr02. The
different zirconium-oxygen polyhedra are outlined. The cubic fluorite type (space group, Fm3m)
high-temperature y-modification transforms on cooling by a displacive phase transition to
tetragonal P-ZrOz (P4,/nmc), which on further cooling transforms, again by a displacive phase
transition, to monoclinic a-Zr02 (P2,/c).
stishovite have shown that this material is harder than alumina and almost equals
the hardness of c-BN [148].It is thus the hardest known oxide.
The high-temperature (y-) forms of Zr02 and HfOz [149] as well as CeOz [150],
Thoz, and UOz [I511 crystallize with the cubic fluorite type structure. The metal
atoms form a close-packed f.c.c. arrangement and the oxygen atoms fill all tetra-
hedral voids. This way each metal atom obtains a cubic environment of eight
oxygen atoms (Fig. 35).
Thoz and UOz are used as nuclear fuels. Tho2 has the highest melting point of all
oxides (>3200"C) and an extremely low oxygen vapor pressure. It is therefore used
as a corrosion-resistant ceramic whenever the less expensive oxides MgO, A1203,or

Zr02 are not suitable, for example as crucibles for high-melting metals [18, 1381.
Of the rare earth oxides, Ce02 is the most frequently used material. It is utilized
for polishing and abrasive purposes and it is obtained in large quantities during the
separation process of the early rare earth elements [152].
Zirconium dioxide has many applications mostly in its cubic, high-temperature
y-modification, which for pure y-Zr02 is only stable above 2300°C. However, it
can be stabilized by adding about 10% MgO, CaO, or Y 2 0 3 .In these yellow cera-
mics the metal atoms M2+ or M3+ substitute for Zr4+. As a consequence some
oxygen positions are not occupied in the solid solutions Zrl - y M ~ + 0 2 - , 0 , and
x , 0 indicates the void positions. In this form stabilized zir-
Zr2- 2 x M ~ ~ 0 4 p 1 0where
conia is used in the glass and steel industry as container material. It also has applica-
tions as solid electrolyte, like in fuel cells and as an oxygen sensor. The electric charges
are carried by the oxygen ions, which move from occupied to unoccupied sites.
Below 2300°C cubic y-Zr02 transforms to the tetragonal p-form which is thermo-
dynamically stable down to about 1 100°C. On further cooling, P-ZrOz transforms
to the monoclinic a-form, which also occurs as a mineral, called baddeleyite.
Both transformations are of the displacive type (as opposed to a reconstructive
phase transition), where the atoms move only by a few picometers. As a conse-
quence the y- and P-modifications cannot be quenched. Since the crystal systems
change from cubic to tetragonal to monoclinic, both transformations are ferroelas-
tic transformations, where the domain walls of the resulting twin domains are
planar. Even though the transformations can be classified as displacive, the
corresponding lattice constants change considerably from cubic a = 0.5065 nm
[149] to tetragonal ah = 0.515nm, c = 0.527nm [153] to monoclinic
a = 0.5145nm, b = 0.5208nm, c = 0.531 1 nm, P = 99.23' [154]. As a consequence,
the transformations can be expected to have large hystereses, especially as compact
polycrystalline ceramics. For unstabilized pure Zr02 the alp-transformation occurs
at about 1200°C on heating and at about 1000°C on cooling [155].
In comparing the corresponding cell volumes for Z = 4 formula units for the
(pseudo-) cubic cells of a-Zr02 (0.1404 nm3), P-Zr02 (0.1397 nm3), and y-Zr02
(0.1300 nm3) one notices the anomaly that the high-temperature cubic modification
has a cell volume, which is about 7% smaller than those of the other two modifica-
tions. In part this may be due to an oxygen deficiency of the black high-temperature
y-modification of the binary Zr02- ,r. There is also an increase in the volume for the
P/a (tetragonal to monoclinic) transformation. For stabilized tetragonal zirconia
this volume increase amounts to about 4%. In annealed ceramics of stabilized
zirconia the tetragonal and monoclinic modifications coexist in the two-phase
field with different compositions. Both of these may undergo phase transitions
not only on cooling or heating, respectively, but also because of the change in
volume under stress. When a crack arrives at an embedded grain of tetragonal
zirconia the grain may transform to the more voluminous monoclinic modification,
thus dissipating the stress to the whole surface of the embedded grain. This mechan-
ism is called 'transformation toughening' [156]. Furthermore, since the y/P- as well
as the B/a-transformations are ferroelastic, stress can be dissipated by movement of
domain walls. From a crystallographic point of view this is analogous to the change
in form as it is known for the low-temperature modification of alloys with shape-

memory.
In the crystal structure of tetragonal P-Zr02 the zirconium atoms have again (as
in the cubic y-modification) eight oxygen neighbors [153]. This distorted cubic co-
ordination can best be visualized as two interpenetrating distorted tetrahedra, one
compressed with Zr-0 distances of 0.2065 nm, the other stretched with Zr-0
distances of 0.2455nm. These may be compared to the eight oxygen atoms of a
zirconium atom in y-Zr02 at 0.2193 nm. In contrast to the y- and P-modifications,
in the monoclinic baddeleyite type structure of a-Zr02 [154] the zirconium atoms
have only seven oxygen neighbors at Zr-0 distances ranging from 0.205 to
0.228nm. Of the two kinds of oxygen atoms, one has three zirconium neighbors
in approximately planar configuration (Zr-0 distances of 0.205, 0.206, and
0.216 nm), the other has a distorted tetrahedral environment (Fig. 35) with Zr-0
distances between 0.215 and 0.229 nm.
1.8 Amorphous Hard Materials

A recent strategy is the synthesis of hard materials by pyrolysis (sometimes in the
presence of N2, NH3, HzO, H2, and/or Ar) from molecular precursors. Such hard
materials are frequently composed of inorganic networks which are amorphous to
X-rays. Examples are ceramics with the approximate compositions BSi3N40,
BSi3C4N,B3Si3N7,and BSiN3C [157-1591. These ceramics stay amorphous to X-
rays up to temperatures between 1500 and 1900°C. At higher temperatures they
decompose to composites containing BN, P-Sic, a-Si3N4, and P-Si3N4. For the
characterization of the reaction intermediates and the resulting amorphous solids
various thermal analysis and spectroscopic techniques are useful: differential
thermal analysis (DTA), thermal gravimetric analysis (TGA), mass spectroscopy
(MS), Fourier transform infrared (FT-IR) spectroscopy, l 1 B and 29Simagic angle
spinning nuclear magnetic resonance (MAS-NMR), and X-ray photoelectron
spectroscopy (XPS). These reveal that the atoms in the amorphous ceramics have
the coordination numbers known from the corresponding binary phases: four for
boron and silicon, three and four for carbon, three for nitrogen, and two for oxygen.
In characterizing a solid as 'amorphous' to X-rays one should be aware that this
does not rule out a high degree of order. The Scherrer formula D = k . X/Bcos 8 can
be used to determine the particle size of a crystallite by X-ray diffraction. In this
formula D is the diameter of a particle, k a constant close to k = 1, X the wavelength
of the X-rays, 0 the Bragg angle of the diffracted radiation, and B the line broaden-
ing (in radians). The latter is defined as the additional line width (apart from the
line width due to the instrumental set-up) at one half of the maximum intensity
of the considered reflection [16&162]. As a hypothetical example we may consider
a cube of 4 x 4 x 4 unit cells of diamond (8 atoms per cell, a = 0.357nm) at a
diffraction angle of 8 = 21.9" (i.e. the 111 reflection with Cu ka: radiation). Using
the Scherrer formula, this perfectly ordered arrangement of 8 x (4)3= 512 carbon
References 37
atoms results in a diffraction peak, which at half intensity has a width of

B = 0.116 rad = 6.7". With this width the 111 peak (at 21.9") practically overlaps
with the 200 peak (at 25.6" and a half-width of 6.9'). Thus a powder, made-up of
such hypothetical crystallites with 5 12 perfectly ordered carbon atoms each, looks
amorphous to X-rays!
References
1. P. P. Ewald and C. Herrmann, Strukturberichte Vol. 1-7, in Z . Kristallogr. Akademische
Verlagsgesellschaft mbH, Leipzig, 1913-1939.
2. A. J. C. Wilson, W. B. Pearson, J. Trotter, and G. Ferguson (Eds), Structure Reports Vol. 8-
58, International Union for Crystallography, N. V. A. Oosthoek's Uitgevers MIJ, Utrecht,
1956-1993.
3. Landolt-Bornstein, Numerical Data and Functional Relationships in Science and Technology,
K.-H. Hellwege (Ed.), Group III. Crystal and Solid State Physics, Vol. 6, Structure Data of
Elements and Intermetallic Phases, P. Eckerlin, H. Kandler, and A. Stegher (Eds), Springer,
Berlin, 1971.
4. W. B. Pearson, The Crystal Chemistry and Physics of Metals and Alloys, Wiley-Interscience,
New York, 1972.
5. P. Villars and L. D. Calvert, Pearson's Handbook of Crystallographic Data for Intermetallic
Phases, ASM International, Materials Park, OH, 1991.
6. P. Villars, Pearson's Handbook, Desk Edition, Crystallographic Data for Intermetallic Phases,
Vol. 1 and 2, ASM International, Materials Park, OH, 1997.
7. E. ParthC, L. Gelato, B. Chabot, M. Penzo, K. Cenzual, and R. E. Gladyshevskii, TYPIX-
Standardized Data and Crystal Chemical Characterization of Inorganic Structure Types, in
Gmelin Handbook of Inorganic and Organometallic Chemistry, 8th edn., Springer, 1993.
8. Fachinformationszentrum FIZ Karlsruhe, Inorganic Crystal Structure Data Base - ICSD,
Karlsruhe, 1998.
9. International Centre for Diffraction Data - ICDD, Powder Diffraction File, Newtown Square,
Pennsylvania, 1997.
10. K. Yvon, W. Jeitschko, and E. Parthe, J. Appl. Crystallogr. 1977, 10, 73.
11. J. Donohue, The Structures ofthe Elements, Wiley, New York, 1974.
12. F. P. Bundy and J. S. Kasper, J. Chem. Phys. 1967,46, 3437.
13. V. L. Solozhenko, V. V. Chernyshev, G. V. Fetisov, V. B. Rybakov, and I. A. Petrusha,
J . Phys. Chem. Solids, 1990,8, 1011.
14. L. F. Vereshchagin, E. V. Zubova, L. N. Burenkova, and N. I. Revin, Dokl. Akad. Nauk SSR,
1968, 13, 25.
15. T. Soma, A. Sawaoka, and S. Saito, Muter. Res. Bull. 1974, 9, 755.
16. A. H. Comes de Mesquita, Acta Crystallogr. 1967, 23, 610.
17. Z. Li and R. C. Bradt, J . Muter. Sci. 1986, 21, 4366.
18. W. Buchner, R. Schliebs, G. Winter, and K. H. Buchel, Industrielle Anorganische Chemie,
Verlag Chemie, Weinheim, 1984.
19. E. Parthe, Crystallochimie des Structures Tetraedriques, Gordon & Breach, Paris, 1972.
20. P. T. B. Shaffer, Acta Crystallogr. Sect. B, 1969, 25, 477.
21. M. Dubey, U. S. Ram, and G. Singh, Acta Crystallogr. Sect. B, 1973, 29, 1548.
22. H. Briehl, Chemie der Werkstofle, B. G. Teubner, Stuttgart, 1995.
23. H. L. Yakel, Acta Crystallogr. Sect. B, 1975, 31, 1797.
24. R. Telle, Chemie in unserer Zeit, 1988, 22, 93.
25. A. Kirfel, A. Gupta, and G. Will, Acta Crystallogr. Sect. B, 1979,35, 1052.
26. H. Neidhard, R. Mattes, and H. J. Becher, Acta Crystallogr. Sect. B, 1970, 26, 315.
27. A. J. Perrotta, W. D. Townes, and J. A. Potenza, Acta Crystallogr. Sect. B, 1969, 25, 1223.
28. A. F. Wells, Structural Inorganic Chemistry, 5th edn., Clarendon Press, Oxford, 1990.
29. H. Hillebrecht and F. D. Meyer, Z . Kristallogr. 1997, Suppl. 12, 11.

30. H. Bolmgren, T. Lundstrom, and S. Okada, AZP Conference Proceedings (American Institute
of Physics) 1991, 231, 197.
31. H. Hubert, B. Devouard, L. A. J. Garvie, M. O’Keeffe, P. R. Buseck, W. T. Petuskey, and P. F.
McMillan, Nature, 1998, 391, 376.
32. F. Benesovsky, Borcarbid und Boride, in Ullmanns Enzyklopudie der technischen Chemie, Band
8, VCH, Weinheim, 1974, pp. 657462.
33. M. Stiimke and G. Petzow, Z . Metallkd. 1975, 66, 292.
34. R. Fichte, M. Pistor, G . Knippschild, A. Meffert, and J. Rasch, Bor und Ferrobor, in Ullmanns
Enzyklopadie der technischen Chemie, Band 8, VCH, Weinheim, 1974, pp. 644-648.
35. R. Kiessling, Acta Chem. Scand. 1949, 3, 595.
36. T. Bjurstrom, Arkiv. Kemi Mineral. Geol. 1933, 11A, I .
37. R. Kiessling, Acta Chem. Scand. 1947, 1, 893.
38. J. Etourneau, J. Less-Common Met. 1985, 110,267.
39. P Blum and F. Bertaut, Acta Crystallogr. 1954, 7,81.
40. M. M. Koruskova, V. N. Gurin, T. Lundstrom, and L.-E. Tergenius. J . Less-Common Met.
1986, 117, 73.
41. F. Benesovsky and W. Schedler, Hartmetalle, in Ullmanns Enzyklopadie der technischen
Chemie, Band 12, VCH, Weinheim, 1974, pp. 5 15-52 1.
42. F. Benesovsky, Carbide, in Ullmanns Enzyklopadie der technischen Chemie, Band 9, VCH,
Weinheim, 1974, pp. 122-136.
43. See A. W. Frad, Advances in inorganic Chemistry and Radiochemistry, H. J. Emeleus and A. G.
Sharpe (Eds), Academic Press, New York, 1968, 11, 153.
44. E. Parthe and V. Sadagopan, Monatsh. Chem. 1962,93, 263.
45. E. Rudy, F. Benesovsky, and E. Rudy, Monatsh. Chem. 1962,93, 693.
46. W. Jeitschko, Structural Chemistry of Transition Metal-Metalloid Compounds, in M T P
international Review of Science, inorg. Chem., Series 2, Vol. 5, Butterworths, London, 1974,
pp. 219-281.
47. E. Parthe and K. Yvon, Acta Crystallogr. Sect. B, 1970, 26, 153.
48. Yu. B. Kuz’ma, V. I. Lakh, V. Ya. Markiv, B. I. Stadnyk. and E. I. Gladyshevkii, Poroshko-
vaya Metallurgiya, 1963, 2, 286.
49. K. Yvon and E. Parthe, Acta Crystallogr. Sect. B, 1970, 26, 149.
50. T. Epicier, in The Physics and Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.),
Kluwer, Dordrecht, 1990, pp. 215-248 and 297-327.
51. M. Hansen and K. Anderko, Constitution of Binary Alloys,Genium Publishing, Schenectady,
NY, 1985.
52. J. Billingham, P. S. Bell, and M. H. Lewis, Acta Crystallogr. Sect. A , 1972, 28, 602.
53. C. H. De Novion, R. Lorenzelli, and P. Costa, C. R. h a d . Sci. Paris, 1966, 263, 775.
54. C. Froidevaux and D. Rassier, J . Phys. Chem. Solids, 1967, 28, 1197.
55. D. Kordes, Phys. Stat. Sol. 1968, 26, K103.
56. A. W. Henfrey and B. E. F. Fender, Acta Crystallogr. Sect. B, 1970, 26, 1882.
57. J. D. Venables, D. Kahn, and R. G. Lye, Phil. Mag. 1968, 18, 177.
58. J. Billingham, P. S. Bell, and M. H. Lewis, Phil. Mag. 1972, 25, 661.
59. M. P. Arbuzov, V. G. Fak, and B. V. Khaenko, Sov. Phys. Crystallogr. 1970, 15, 164.
60. M. Sauvage, E. Partht, and W. B. Yelon, Acta Crystallogr. Sect. A , 1974,30, 597.
61. J. B. Venables and M. H. Meyerhoff, NBS Special Publication 364, Proceedings of 5th Mater.
Res. Symposium, 1972, p. 583.
62. H. Goretzki, Phys. Stat. Sol. 1967, 20, K141.
63. J.-L. Chermant, P. Delavignette, and A. Deschanvres, J . Less-Common. Met. 1970, 21, 89.
64. P. S. Bell and M. H. Lewis, Phil. Mag. 1971, 24, 1247.
65. B. V. Khaenko and V. V. Kukol’, Sov. Phys. Crystallogr. 1989, 34,905.
66. V. V. Kukol’, B. V. Khaenko, and 0. A. Gnitetskii, Crystallogr. Rep. 1995,40, 66.
67. Ch. H. de Novion and V. Maurice, J. Phys. (Paris), Colloq. 1977, 211.
68. Ch. H. de Novion, B. Benneu, T. Priem, N. Lorenzelli, and A. Finel, in The Physics and Chem-
istry of Carbides, Nitrides and Borides, R. Freer (Ed.), Kluwer, Dordrecht, 1990.
69. B. V. Khaenko, 0. A. Gnitetskii, and M. V. Karpets, Crystallogr. Rep. 1999, 44,281.
References 39
70. G. Hagg, Z . Phys. Chem. B, 1931, 12, 33.

71. A. Neckel, K. Schwarz, R. Eibler, P. Weinberger, and P. Rastl, Ber. Bunsenges. Phys. Chem.
1975,79, 1053.
72. A. Neckel, P. Rastl, R. Eibler, P. Weinberger, and K. Schwarz, J. Phys. C: Solid State Phys.
1976, 9, 579.
73. P. Blaha, K. Schwarz, F. Kubel, and K. Yvon, J . Solid State Chem. 1987, 70, 199.
74. S. D. Wijeyesekera and R. Hoffmann, Organometallics, 1984, 3, 949.
75. A. L. Bowman, T. C. Wallace, J. L. Yarnell, and R. G. Wenzel, Acta Crystallogr. 1966,21,843.
76. C. S. Barrett and T. B. Massalski, Structure of Metals, McGraw-Hill, New York, 1966.
77. C. S . Roberts, J. Metals, 1953, 5 (Trans. AIME 197), 203.
78. D. Meinhardt and 0. Krisement, Archivfiir das Eisenhiittenwesen, 1962, 33, 493.
79. E. J. Fasiska and G. A. Jeffrey, Acta Crystallogr. 1965, 19, 463
80. S. Rundqvist and G. Runnsjo, Acta Chem. Scand. 1969, 23, 1191.
81. H. L. Yakel, Acta Crystallogr. Sect. B, 1987, 43, 230.
82. F. C. Frank and J. S. Kasper, Acta Crystallogr. 1958, 11, 184.
83. F. C. Frank and J. S. Kasper, Acta Crystallogr. 1959, 12, 483.
84. M. A. Rouault, P. Herpin, and M. R. Fruchart, Ann. Chim. Paris, 1970, 5, 461.
85. A. Westgren, Jernkontorets Annaler, 1933, 111, 525.
86. Z . Bojarski and J. Leciejewiez, Archiwunz Hutnictwa, 1967, 12, 255.
87. P. Ettmayer and R. Suchentrunk, Monatsh. Chem. 1970, 101, 1098.
88. B. Rupp and P. Fischer, J. Less-Common Met. 1988, 144, 275.
89. J. M. Newsam, A. J. Jacobson, L, E. McCandlish, and R. S . Polizzotti, J . Solid State Chem.
1988, 75, 296.
90. B. G. Hyde and S . Anderson, Inorganic Crystal Structures, Wiley, New York, 1989.
91. U. Westermann, E. Lugscheider, and J. Wonka, Metal1 1993, 47, 741.
92. W. Jeitschko, Acta Crystallogr. Sect. B, 1977, 33, 2347.
93. B. Boren, Ark. Kem. Miner. Geol. A , 1933, lla(10), 1.
94. H. Nowotny, R. Kieffer, and H. Schachner, Monatsh. Chem. 1952,83, 1243.
95. J. J. Nickl, K. K. Schweitzer, and P. Luxenberg, J . Less-Common Met. 1972, 26, 335.
96. W. J. J. Wakelkamp, F. J. J. van Loo, and R. Metselaar, J . Eur. Ceram. SOC.1991, 8, 135.
97. Yong Du and J. C. Schuster, Ber. Bunsenges. 1998, 102, 1185.
98. W. Jeitschko and H. Nowotny, Monatsh. Chem. 1967,98, 329.
99. T. El-Raghy and M. W. Barsoum, J . Appl. Phys. 1998,83, 112.
100. S. K. Lee, B. R. Lawn, and M. W. Barsoum, J . Am. Ceram. SOC.1998,81,225.
101. N. I. Medvedeva and A. L. Ivanovskii, Russ. J . Inorg. Chem. 1998, 43, 398.
102. R. Pottgen, D. Kaczorowski, and W. Jeitschko, J. Muter. Chem. 1993, 3, 253.
103. E. Partht., W. Jeitschko, and V. Sadagopan, Acta Crystallogr. 1965, 19, 1031.
104. B. Aronsson, Acta Chem. Scand. 1960, 14, 1414.
105. W. Rieger, H. Nowotny, and F. Benesovsky, Monatsh. Chem. 1965,96, 232.
106. M. Potzschke and K. Schubert, Z . Metallkd. 1962, 53, 474.
107. W. Schnick, Angew. Chem. 1993, 105, 846.
108. H. Lange, G. Wotting, and G. Winter, Angeih,. Chem. 1991, 103, 1606.
109. M. Billy, J. C . Labbe, A. Selvaraj, and G . Roult, Mater. Res. Bull. 1983, 18, 921.
110. R. Griin, Acta Crystallogr. Sect. B, 1979, 35, 800.
111. H. Schulz and H. H. Thiemann, Solid State Commun. 1977, 23, 815.
112. A. W. Weimer (Ed.), Carbide, Nitride and Boride Materials, Chapman & Hall, London, 1997.
113. C. Niu, Y. Z . Lu, and C. M Lieber, Science, 1993, 261, 334.
114. T. Hughbanks and Y. Tian, Solid State Commun. 1995,96, 321.
115. A. K. Sharma and J. Narayan, Int. Mater. Rev. 1997,42, 137.
116. J. V. Badding, Adv. Muter. 1997, 9, 877.
117. D. M. Teter, Mater. Res. SOC.Bull. 1998, 22.
118. F. Benesovsky, Nitride, in Ullmanns Enzyklopadie der technischen Chemie, Band 17, VCH,
Weinheim, 1974, pp. 315-321.
119. W. Lengauer, J . Alloys Compd. 1992, 186,293.
120. A. N. Christensen, A. Alamo, and J. P. Landesman, Acta Crystallogr. Sect. C 1985, 41, 1009.
121. D. Rechenbach and H. Jacobs, J. Alloys Compd. 1996,235, 15.
122. H. Jacobs, D. Rechenbach, and U. Zachwieja. J. Alloys Compd. 1995, 227, 10.
123. B. Holmberg, Acta Chem. Scand. 1962, 16, 1255.
124. W. Lengauer and P. Ettmayer, J. Less-Common Met. 1986, 120, 153.
125. W. Lengauer, J. Less-Common Met. 1986, 125, 127.
126. J. C. Schuster and J. Bauer, J . Solid State Chem. 1984, 53, 260.
127. W. Jeitschko, H. Nowotny, and F. Benesovsky, Monatsh. Chem. 1963,94, 1198.
128. H. H. Stadelmaier in Developments in the Structural Chemistry of Alloy Phases, B. C. Giessen
(Ed.), Plenum Press, New York, 1969.
129. L. E. Toth, Transition Metal Carbides and Nitrides, Academic Press, New York, 1971.
130. H. Nowotny, Angew. Chem. Intern. Edit. 1972, 11, 906.
131. Th. M. Gesing, K. H. Wachtmann, and W. Jeitschko, Z. Naturforsch. 1997, 52b, 176.
132. J. C. Schuster, Z . Kristallogr. 1986, 175, 211.
133. W. Jeitschko, H. Nowotny, and F. Benesovsky, Monatsh. Chem. 1964,95, 1212.
134. W. Schnick and H. Huppertz, Chem. Ew. J. 1997, 3, 679.
135. M. Woike and W. Jeitschko, Inorg. Chem. 1995, 34, 5105.
136. T. Schlieper and W. Schnick, Z . Anorg. Allg. Chem. 1995,621, 1037.
137. H. W. Hennicke, U. Hofmann, W. Lehnhauser, and K.-J. Leers, Keramik, in Ullmanns Enzyk-
lopadie der technischen Chemie, Vol. 13, VCH, Weinheim, 1974, pp. 711-735.
138. K. Heldt and A. Reckziegel, Oxidkeramik, in Ullmanns Enzyklopadie der technischen Chemie,
Vol. 17, VCH, Weinheim, 1974, pp. 515-529.
139. J. Lewis, D. Schwarzenbach, and H. D. Flack, Acta Crystallogr. Sect A , 1982, 38, 733.
140. T. M. Sabine and S. Hogg, Acta Crystallogr. Sect. B, 1969, 25, 2254.
141. D. K. Smith, C. F. Cline, and S . B. Austerman, Acta Crystallogr. 1965, 18, 393.
142. E. Schiebold, Z. Kristallogr. 1921, 56, 430.
143. N. G. Zorina and S . S . Kvitka, Kristallografiya, 1969, 13, 599.
144. R. Restori, D. Schwarzenbach, and J. R. Schneider, Acta Crystallogr. Sect. B, 1987,43,251.
145. M. Horn, C. F. Schwerdtfeger, and E. P. Meagher, Z . Kristallogr. 1972, 136, 273.
146. W. H. Baur, Acta Crystallogr. 1961, 14, 214.
147. M. Sugiyama, S. Endo, and K. Koto, Mineral. J. Jpn. 1987, 13, 455.
148. J. M. Leger, J. Haines, M. Schmidt, J. P. Petitet, A. S. Pereira, and J. A. H. de Jomada, Nature,
1996,383,401.
149. L. Passerini, Gazz. Chim. Ital. 1930, 60, 762.
150. G. Brauer and H. Gradinger, Z. Anorg. Allg. Chem. 1954, 277, 89.
151. K. Clausen, W. Hayes, J. E. Macdonald, P. Schnabel, M. T. Hutchings, and J. K. Kjems, High
Temp. High Press. 1983, 15, 383.
152. I. McGill, J. Helgorsky, A. LevCque, Th. Petzel, and K. Reinhardt, Rare Earth Elements, in
Ullmann’s Encyclopedia of Industrial Chemistry, 5th edn., Vol. A22, VCH, Weinheim, 1993,
pp. 607-649.
153. G. Teufer, Acta Crystallogr. 1962, 15, 1187.
154. D. K. Smith and H. W. Newkirk, Acta Crystallogr. 1965, 18, 983.
155. G. Routschka and K.-E. Granitzki, in Ullmann’s Encyclopedia of Industrial Chemistry, 5th
edn., Vol. A23, VCH, Weinheim, 1993, pp. 1 4 8 .
156. L. Michalowsky (Ed.), Neue keramische Werkstofle, Deutscher Verlag fur Grundstoffindustrie,
Leipzig, 1994.
157. R. Riedel, Advanced ceramics from inorganic polymers, in Materials Science and Technology,
R. W. Cahn, P. Haasen, and E. J. Kramer (Eds.), Vol. 17B, Chapter 11, VCH, Weinheim,
1996.
158. R. Riedel, J. Bill, and A. Kienzle, Appl. Organomet. Chem. 1996, 10, 241.
159. H.-P. Baldus and M. Jansen, Angew. Chem. 1997, 109, 338.
160. B. D. Cullity, Elements of X-Ray Diffraction, Addison-Wesley, Reading, Massachusetts, 1959.
161. L. V. Aziroff, Elements of X-Ray Crystallography, McGraw-Hill, New York, 1968.
162. B. E. Warren, X-Ray DifSraction, Addison-Wesley, Reading, Massachusetts, 1969.
2 Phase Transitions and Material Synthesis

using the C02-Laser Heating Technique in a
Diamond Cell
A. Zerr, G. Serghiou, and R. Boehler
2.1 Introduction
At room temperature, pressures in excess of 500GPa can be attained using the
diamond anvil cell (DAC) technique [l]. This technique, described in a number of
comprehensive reviews [24], has found broad application in the high pressure
sciences, because diamond serves as an optical window from the far infrared to
the near ultraviolet wavelength regime and is transparent to X-rays [4,5].
Development of the convenient ruby pressure scale, where the red shift of the R1-
fluorescence line with pressure is used in situ [6,7], led to a much wider use of this
experimental technique.
A few high-pressure devices are used to achieve high temperatures at high static
pressures: Large-volume high-pressure devices (with characteristic sample dimen-
sions of 2-3mm) can be used at temperatures up to 3000K, and pressures of
about 25GPa (white field in Fig. 1). In an externally heated DAC temperature is
limited to about 1800 K. Above this temperature the diamond anvil transforms to
graphite, even in an inert atmosphere [2]. Moreover, these devices are limited to
pressures of about 30GPa due to weakening of the diamond anvils and of other
parts of the DAC upon heating. Additionally, pressure measurements are com-
plicated by the absence of accurate high-pressure-high-temperature equations of
state for any solid that could be used as a pressure gauge. Another possibility
involving use of an internally heated DAC, where the conducting sample is
heated by passing of an electric current through it [8,9], is constrained by its
range of applicability. The laser-heated DAC allows us to access much higher
static pressures and temperatures than the techniques listed above (gray field in
Fig. 1).
The technique of laser heating in a DAC is based on three main features: optical
transparency of diamond anviIs; the samples can be heated via the optical absorp-
tion of intense laser radiation, and the temperature can be determined from the
thermal radiation spectrum of the heated sample using the Planck formula [lo].
Laser radiation for heating of a sample in a DAC was first implemented by Ming
and Bassett [ll], who used a pulsed ruby laser, and a continuous-wave Nd-YAG
(yttrium-aluminum-garnet) laser to heat samples in a DAC above 3300 K, and up
to 2300K, respectively. Today two types of continuous wave infrared (IR) lasers
are extensively used in laser heating experiments: Solid state lasers (Nd-doped
YAG, or YLF (yttrium-lithium-fluorite) crystals with the most intense line at
42 2 Phase Transitions and Material Synthesis using the COI-Laser Heating Technique
s!? SUN
3
b 20 60
Pressure (GPa)
Figure 1. Pressure-temperature regime accessed to date using the C02-laser heating technique is
shown by the light gray field. The P-T values available for multianvil high pressure devices are
given by the white field. The temperature on the surface of the sun is shown for comparison.
1.06 pm), and C 0 2 gas lasers (most intense line at 10.6 pm). The Nd-YAG lasers
have been used for heating semiconductors [12], metals, and insulators containing
transition metals such as Fe, Ni, etc. [l 11. The absorption mechanism involves cou-
pling of the laser radiation with electronic excitations within the sample. The C02-
laser heating technique, first employed by Boehler and Chopelas [ 131, has been used
in experiments on numerous nonconducting inorganic (oxides, silicates, nitrides,
etc.) and organic materials (e.g. hydrocarbons C,H2, + 2 ) . Heating of these materials
with C02-laser radiation is possible due to the resonant absorption of the laser
radiation by lattice vibrations. The C02-laser heating technique is particularly
suitable for high-pressure-high-temperature experiments on hard materials (i.e.
cubic BN, Sic, Si3N4),since most of them are insulators.
In section two of this paper we describe the technical aspects of C02-laserheating
in a DAC. The third section focuses on the methods for measuring melting tem-
peratures at variable pressures, the fourth section on the determination of high
pressure and temperature phase diagrams, and in the fifth section some experiments
focusing on the synthesis of diamond and cubic BN from organic precursors will be
described.
2.2 Technique of C02-Laser Heating in a Diamond

Anvil Cell
2.2.1 Sample Assemblage in a Diamond Anvil Cell
Figure 2 shows schematically the sample assemblage in a DAC. The sample
container, which is made by drilling a hole in a pre-indented metal gasket, has
typical dimensions of about 150 pm in diameter and 50 pm in height. The sample,

which can be a single crystal, glass, or pressed powder pellet with typical dimen-
sions of about 70 x 70 x 15 pm, is separated from the highly thermally conducting
diamond anvils using small discs of the sample material or of an alkali halide (for
example NaC1, KBr). The sample is embedded in a pressure transmitting medium,
which provides thermal insulation of the sample from the diamond, and quasi-
hydrostatic pressure conditions in the sample volume (see below). Depending on
the chemical composition of the sample, different pressure media can be used:
Noble gases provide chemically inert conditions. Alkali halides, which are less
hydrostatic and less chemically inert, are better thermal isolators and do not
require use of gas loading devices. Nitrogen is used in specific experiments to
provide chemical stability of nitrides (for example BN [14]) at high temperatures
or as a reactant source for synthesis of nitrogen containing materials at high
pressures and temperatures [ 12,151.
2.2.2 Pressure Conditions in the Sample Volume

For pressure measurement we used the ruby fluorescence scale, calibrated under
quasi-hydrostatic load conditions to 80 GPa with the known equations of state of
two metals [7]. The pressure was measured at room temperature using micron-
sized ruby crystals, distributed on the edge of the sample volume (Fig. 2) to avoid
chemical reaction with the sample material during heating. The ruby fluorescence
was excited by Ar-laser radiation (Fig. 2).
At room temperature the pressure medium is solid, resulting in pressure gradients
in the pressure chamber by uniaxial compression. Use of soft pressure media (e.g.
argon) reduces these pressure gradients to about 10-15% of the average value.
It was found in C02-laser heating experiments to 20 and 43 GPa [16,17] that
COz-laser Ar-laser
Figure 2. Schematic drawing of a sample assemblage in the diamond cell. The sample is heated
with a C 0 2 laser. Nitrogen is used as a pressure medium in experiments on nitrides to preclude
decomposition. Argon is used as a pressure medium when chemically inert conditions are required.
Fluorescence of ruby crystals is excited using an Ar laser.
44 2 Phase Transitions and Material Synthesis using the C02-LaserHeating Technique
after heating of the sample the pressure gradients are reduced to 3% of the average
value.
During laser heating the pressure in the sample chamber increases owing to
thermal expansion of the sample and of the surrounding pressure medium. To
reduce this effect in our C02-laser heating experiments, the sample to pressure
medium volume ratio was kept below 1 :10. We found that upon heating silicates
up to 2600K at around 20 GPa the thermal pressure does not exceed 0.3 GPa,
which is less than 2% of the average pressure value [13].
2.2.3 Experimental Set-up for C02-Laser Heating in a Diamond

Anvil Cell
An experimental set-up for C02-laser heating in a DAC is shown schematically in
Fig. 3 . The incident angle of radiation of the C 0 2 laser (Melles Griot, 150W,
>90% of the intensity is in the single transverse electromagnetic mode TEM,,,) is
about 25" (see also Fig. 2). The laser beam is focused by a ZnSe-lens onto the
sample in a DAC. By changing the distance between the ZnSe-lens and the DAC
one can vary the size of the laser-heated spot. The heated area, which exhibits a
bell-shaped temperature distribution (see below), is imaged by a reflecting objective
on the plane of the entrance pin-hole of the monochromator. The magnification of
about 20, and a pinhole diameter of about 50 pm, result in a spatial resolution for
the sample temperature measurement of about 2.5pm in diameter. The thermal
l4 I I
L
13
Figure 3. Schematic drawing of the experimental set-up for the C02-laser heating experiments in a
DAC: 1 = C 0 2 laser, 2 = Si mirror, 3 = ZnSe lens, 4 =diamond cell, 5 =reflecting objective,
6 = monochromator, 7 = CCD detector, 8 = PC, 9 and 12 = beam splitter, 10 = optical diode,
11 =regulating unit, 13 = ocular, 14 =power supply of the C 0 2 laser. The dashed and dotted line
shows the path of the C02-laser beam, the dashed lines show the visible light path, and the solid
ones depict electrical connections.
radiation spectrum from this area is measured with a charge coupled device (CCD)
detector. To determine the temperature, the Planck formula is fitted to the measured
spectrum (see below). Approximately 50% of the incandescent radiation from the
sample is directed to an optical diode by a beam-splitter. The diode signal, after
amplification and conversion in the regulating unit, is used to stabilize the laser
power (see below). The second beam-splitter directs the sample radiation to the
ocular for visual observation.
2.2.4 Temperature Determination

The heated sample emits thermal radiation, which is used for temperature
determination. The spectrum collected was measured in the wavelength range
5 15-820 nm corresponding to the range of maximal quantum efficiency of our
CCD detector. To determine the temperature we fitted the Planck formula with a
wavelength independent emissivity to the measured spectrum. The Planck formula
[ 101 contains the temperature and the wavelength dependence of the thermal
radiation intensity I B B ( T , A) of the black body (BB):
where h is the Planck constant, k the Boltzmann constant, c the velocity of light in
vacuum, T the temperature of the BB, and X the wavelength.
The temperature and wavelength dependent emissivity E ( T ,A) is introduced
for description of a thermal radiation spectrum of a real body ZRB(T,X). Emis-
sivity is simply the ratio of the intensities of the real body to the BB at the
same X and T [5]:
In our experiments we assumed that the materials investigated are graybodies; that
is that their emissivities, which are smaller than unity, are wavelength independent
E ( T ,A) = e( T ) .The wavelength dependence of emissivity is only known for a few
materials in vacuum or at atmospheric pressure [18,19]. If we assume that E of
the investigated materials has the same wavelength dependence as tungsten [ 191
then the temperature correction would be about -100K at 3000K and about
-350K at 5000K. Figures 4a and 4b show the thermal radiation spectra (thick
solid lines) of S i c heated to about 6700K at 12GPa and of MgSi03 heated to
3000 K at 100 GPa, respectively. The temperature values correspond to the fits
(thin solid line) of the Planck formula with a wavelength-independent emissivity
to the measured spectra.
2.2.5 Temperature Stabilization

In C02-laser heating experiments in a DAC it is important to eliminate sample
temperature fluctuations due to fluctuations of the laser intensity. The peak-to-
peak amplitude of the COz-laser power fluctuations in our experiments exceeded
46 2 Phase Transitions and Material Synthesis using the COr-Laser Heating Technique
SIC
P=12 GPa
T=6700 K
.-
E
fn
C
9
-
c
550 650
I . 1 ’ I ’ I ’ I . I
0 600 700 750 800
a wavelength (nm)
E
Y,
C
d
-
c
500 550 600 650 700 750 800

b wavelength (nm)
Figure 4. (a) Thermal emission spectrum of S i c heated to 6700 f200 K at 12 GPa (thick line). The
thin line shows the fit of the Planck formula to the spectrum with a wavelength independent
emissivity. (b) Thermal emission spectrum of MgSiO, heated at 3000 K and 100GPa (thick line).
The thin line is the fit of the Planck formula.
5% of a chosen average value, when feedback stabilization was not employed. This
resulted in temperature fluctuations with a peak-to-peak amplitude exceeding 30%
of the average sample temperature. Figure 5a shows temperature fluctuations in a
sample heated for 60s with an unstabilized C 0 2 laser. This nonlinear relation
between laser power fluctuations and temperature fluctuations is most probably
due to the temperature dependent absorption of the sample material: For example,
experimental measurements of the absorption coefficient k(X) of MgO between 1
and 9 pm to about 1300K at atmospheric pressure showed that k(X) increases non-
linearly with temperature as well as with wavelength [20].
2.2 Technique ofC02-Las e r Heating in a Diamond Anvil Cell 47
3000 -. 34.5 GPa T=3000 K ~ , . ~ . s m ~
2800 -
.- ...
2600 - c
2400- -..
L
-
.-. . -%\
,L :
,,--
..'
I
2200 - -9
*. %
.;-
1
, 1
-
T=2140 K
2ooo
1800). , . , . I . , . I . . , . I
0 10 20 30 40 50 60 70 80 90 100
time (x0.6 s)
(Mg,Fe)SiO,
3000 3010+50 K 29 GPa
28001
2600
2000
0 5 10 15 20 25 30 35 40 45 50
time (min)
Figure 5. (a) Temperature fluctuations in the (Mg,Fe)Si03 sample heated in a DAC with an
unstabilized COz laser: 100 temperature measurements within a time interval of about 60 s from
the center of the heated area (squares). During this time interval the temperature fluctuated between
2140 and 3000 K. (b) Temperature fluctuations in the (Mg, Fe)SiO, sample heated in a DAC with
laser stabilization to average temperatures of 2090 K (diamonds), 2420 K (circles) and 3010 K
(squares) were monitored during 35, 20, and 17 min, respectively. The laser stabilization suppressed
the peak-to-peak fluctuations to below 120 K, or to below 5% of the average temperature. After [21].
To suppress sample temperature fluctuations a feedback stabilization circuit is

employed (see above): Figure 5b shows the sample temperature fluctuations
measured on a (Mg, Fe)Si03 sample heated at 29 GPa for 17-35 min when the
feedback stabilization system was activated. In these long-term heating runs the
peak-to-peak amplitude of the sample temperature fluctuations were kept below
5% of the average temperature.
48 2 Phase Transitions and Material Synthesis using the COz-Laser Heating Technique
I . I . I . I , I . I , l
2800 (Mg ,Fe)Si 0, P=29 GPa
h
2600 1
2400-
!?
3
2200-
2
.l-i
8
E
0
0
I- 2ooo:
1800 -
1600!, . I . , . . , . I !
-30 -20 -10 0 10 20 30

rad. distance (pm)
Figure 6. Radial temperature distribution in the (Mg, Fe)Si03 sample heated with a defocused COl
laser beam at 29 GPa. Temperatures were measured (symbols) in 4 pm steps with a spatial resolution
of 2.5 pm. In the central region of the heated area (with about 30 Fm in diameter) the temperature
gradients are below 10Kpm-'. After [21].
2.2.6 Radial Temperature Gradients

As mentioned above, the laser beam can be defocused in order to increase the heated
area (Figs 2 and 3). Owing to the high power of the C02-laser, and high absorption
of its radiation by the sample, the temperature gradients in the central portion of
the heated area can be reduced to below lOKpm-': Fig. 6 shows the radial tem-
perature gradients measured in an (Mg, Fe)Si03 sample heated at 29 GPa to peak
temperatures of about 2000 K and 2700 K, respectively.
2.2.7 Raman and Fluorescence Spectroscopic Analysis of Samples

in a Diamond Anvil Cell
Samples were characterized using Raman and fluorescence spectroscopy at room
temperature and high pressure up to 100 GPa. These techniques allow us to charac-
terize rapidly micron-sized samples which can be irregularly shaped and oriented. In
particular, with the use of available databases one can obtain detailed structural
information. Spectroscopic signals are sensitive to structural changes at the 0.5-
2.0nm level [22] and are less dependent on atomic number than X-ray signals.
Therefore, these techniques are very effective in monitoring structural changes in
crystals with small X-ray scattering cross-sections such as MgSi03 or Si02 and in
disordered solids [23,24]. Additionally, by using the shift of the vibrational modes
of the solids upon compression one obtains the pressure dependence of physical
properties such as the thermal expansion and the sound velocity [25,26]. In our
2.3 Determination of’ Melting Temperatures at High Pressures 49
experiments the samples were excited with an argon ion laser (A = 457.9 nm) with
powers varying from a few milliwatts to about 100mW. The spatial resolution of
the measurements was typically 10-15pm. Spectra (see below) were taken from
the center of the laser-heated spot at room temperature using CCD, or photon
counting detection systems.
2.3 Determination of Melting Temperatures at High

Pressures
Measurement of the melting temperatures at static pressures above 25 GPa or at
temperatures above 3000 K (at high pressures) first became possible after develop-
ment of the laser heating technique in a DAC. In our C02-laserheating experiments
we have used two melting criteria: The first involves a strong increase in the absorp-
tion of the laser radiation in the melt, resulting in a strong temperature increase. In
these experiments the melting temperatures were taken as the last temperatures of
the solid before the temperature increase due to melting. In the molten area of the
sample convective motion was observed, accompanied by strong temperature
fluctuations of about f 6 0 0 K. Characteristic textural features due to melting were
found on the recovered samples (see below). An increase in the absorption of the
laser radiation in the melt was usually observed for materials which melted
congruently. Changes in the absorption of laser radiation were also observed in
solid-solid phase transitions, for example when Si02 quartz transforms to coesite
at high pressure. In these transitions changes in the absorption are usually smaller
and the temperature fluctuations remain within the limits provided by the stabiliza-
tion system (below f 6 0 K).
A different criterion was employed when no changes in absorption of the laser
radiation in the melt were detected. This method takes advantage of the observed
change in the surface topography on the quenched sample, which is due to migration
of the melted sample material towards the laser beam (see below). Such migration is
possible when the melting temperature of the pressure medium is below the melting
temperature (or solidus) of the sample material. This effect was observed both in
congruently melting solids, where strong increases in the absorption of the laser
radiation in the melt are observed, and in partial melting of multicomponent
materials, where changes in the absorption of the laser radiation in the melt are
undetectable .
2.3.1 Melting of Cubic BN at 10GPa

An increase in the absorption of the laser radiation in the melt, as well as textural
changes on the sample surface, were clearly observed in melting experiments on
cubic BN at 10GPa described below. Based on previous estimates of the triple
point between hexagonal BN, cubic BN, and liquid [27], melting of cubic BN at
50 2 Phase Transitions and Material Synthesis using the C02-Laser Heating Technique
Laser power (arb. units)

Figure 7. Sample temperature of cubic boron nitride versus C02 laser power at 10GPa: 100
temperature measurements were taken within a time interval of about 5 s while the laser power
was steadily increased. The two discontinuous temperature increases at 3600 K are due to melting.
10GPa requires temperatures in excess of 3500K. In our experiment we used

nitrogen as a pressure medium to preclude decomposition of boron nitride upon
heating [14]. Long-term heating of the sample at the temperatures required for
melting of cubic BN (35004000K) led to diamond anvil failure. Therefore we
avoided long heating durations: The laser power was increased rapidly while
temperatures were measured continuously every 0.05 s. Sample temperatures
versus laser power (solid line) are shown in Fig. 7. This plot shows a typical
power-temperature dependence with two temperature jumps at about 3600 K.
Sudden temperature increases are due to an increase in the absorption of the laser
radiation in the melt. The subsequent temperature drops are most probably due
to changes in the surface topography which were observed on the quenched
sample (Fig. 8): The melted portion of the sample migrated toward the laser
beam in the melted nitrogen pressure medium [28], which decreased the thickness
of the insulating pressure medium between diamond and sample, leading to
higher heat conduction to the diamond anvil. As a result the sample temperature
decreased below the melting temperature and the second melting event could be
observed after further increase of the laser beam intensity. The melting temperature
was 3600 f 100 K, taken as the last temperature of the solid before the temperature
jumps. Raman spectroscopic analysis of the melted sample area showed the lines of
co2-laser \
Figure 8. Schematic of the cubic BN sample melted in a DAC at 10 GPa. Temperatures recorded for
this sample are shown in Fig. 7. The outgrowth in the center of the heated area is due to melting.
cubic BN only, no traces of the low pressure hexagonal BN phase were observed.
These results are consistent with the previous estimation of Corrigan and Bundy
~71.
2.3.2 Melting Temperatures of Materials Relevant to the Earth's

Lower Mantle
(Mg, Fe)O, (Mg, Fe)Si03-, and CaSi03-perovskite are considered to make up about
97 weight-% of the Earth's lower mantle, which extends from about 670 to about
2900km depth and makes up about 50% of the Earth's mass. Their melting
curves are important for modeling of chemical differentiation, rheology of the
lower mantle [29], and for interpretation of seismic wave attenuation [30]. In our
melting experiments on these compounds the experimental set-up and the sample
assemblage was the same as described above with the exception that argon was
used as a pressure medium.
Using C02-laser heating in a DAC we measured, for the first time, the melting
temperatures of MgO to about 32 GPa [31]. These measurements were not possible
previously because of the high initial melting temperature of 3063 K at 1 atm [32]. A
Nd-YAG laser is unsuitable for heating of MgO, because of the low absorption of
that radiation (A = 1.06pm) by MgO [5,20]. The results for MgO are shown in
Fig. 9. The initial slope of the MgO melting curve of about 36KGPa-' is
significantly smaller than theoretical estimates, where values between 100 K GPa-'
and 300 K GPa-' were reported [37-391. Using the theoretical [40,41] and empirical
1421 relations between melting temperature and pressure we extrapolated the high
pressure data back to 1 atm and obtained 3050K, which is very close to our
measurement at 1 bar of 3040 f l00K and to the data in the literature [32]. A fit
I ' I ' I ' I . I . I . I
5000 - (Mg,Fe)SiO,-pv
4500 - CaSi0,-pv ,
A,..
,*- multi-component 1
1500-.' multi-anvil -
~ - I - I ~ I - I - ~ ~
of the theoretical melting relation [40,41] to the data revealed that the Griineisen
parameter of MgO at 1 atm is 1.3, in good agreement with the literature value of
1.4 [43].
Melting temperatures of orthorhombic (Mg, Fe)Si03-perovskite and of cubic
CaSi03-perovskite were measured using the COJaser heating technique to
62 GPa [34] and 43 GPa [33], respectively (Fig. 9). Melting temperatures of both
perovskites (which are considered to be the most abundant lower mantle phases)
have similar strong pressure dependencies: The initial melting slope for both systems
is about 60 K GPa-', which is almost twice that of MgO. At the highest pressures
of these experiments the obtained melting temperatures exceeded 5000 K for
(Mg, Fe)Si03-perovskite and 4000 K for CaSi03-perovskite. Precise measurement
of such high melting temperatures is only possible if the radial and axial temperature
gradients are small, and if chemically inert pressure media (for example noble gases)
are used. These requirements were not fulfilled in earlier DAC experiments on
(Mg, Fe)Si03-perovskite [44-46]. In those experiments samples were heated with
YAG-laser radiation, which is only weakly absorbed by the sample, and no ther-
mally insulating pressure medium was used. In previous melting experiments on
CaSi03-perovskite [47], rhenium plates in contact with the sample were used to
absorb the Nd-YAG laser radiation, thus indirectly heating the sample. We have
Figure 10. (a) Back-scattering electron topographic image of the glass sample with the mantle
relevant composition heated just above the solidus to 2550 K at 20.6 GPa. Within a circular patch
of about 60 pm in diameter the starting material converted to a crystalline high pressure assemblage.
Cracks within this area became evident after pressure release. Another type of texture of about
lOpm in diameter in the center (arrow) is due to migration of molten sample material towards
the laser beam, as evident from quantitative measurements with an atomic force microscope in
the area indicated by dashed lines. (b) Three-dimensional surface topography of the 34 x 34pm
area, shown in Fig. 10(a) by dashed lines, measured with an atomic force microscope. Partial melting
of the sample within an area of about 10 pm in diameter caused height variations of up to 0.77 pm. In
contrast, height variations in the surrounding area are smaller than 0.2 pm. Reprinted with permis-
sion from [35]. Copyright [1998] American Association for the Advancement of Science.
54 2 Phase Trunsitions and Material Synthesis using the C02-Laser Heating Technique
found, however, that at high temperatures CaSi03 reacts with rhenium [33], which
probably depresses the melting temperatures by up to 650 K [47].
The solidus temperature of a multicomponent system (for example, the Earth‘s
mantle material) cannot be predicted even if the melting temperatures of all end-
member components of the system are known. Previous solidus temperatures of
mantle relevant materials were measured to about 25 GPa using multianvil high
pressure devices [36,48,49]. The solidus of the (Mg, Fe)O-(Mg, Fe)Si03 assem-
blage, which can be considered as a simplified model for the Earth’s mantle material,
was measured to about 33 GPa in a DAC using a Nd-YLF laser [33]. Additionally,
melting of (Mg, FehSi04, which breaks down to the same phase assemblage, was
estimated in shock-wave experiments to be 4300 K at about 130 GPa [50]. For the
measurement of the solidus temperature of a multicomponent material with a
realistic Earth’s mantle composition we used [35] a different melting criterion
involving changes of the surface texture due to melting. Textural changes on the
sample surface due to melting were verified using a scanning electron microscope
(SEM) and an atomic force microscope (AFM). Figure 10a shows a back-scattered
electron topographic image of the sample which was partially molten at 2550K
and 20.6 GPa. Characteristic melting features (about 10 pm in diameter) can be
seen in the center of the heated area. In the molten area the height variations are
almost four times larger than in the surrounding subsolidus material, as evident
from quantitative topographic measurements with an AFM (Fig. lob). At pressures
between 20 and 30 GPa our data (Fig. 9) agree well with previous multianvil data and
with the solidus of the (Mg, Fe)-(Mg, Fe)Si03 assemblage measured using an YLF-
laser. Our results indicated a strong depression in the melting temperatures of the
multicomponent system as compared with the end-member phases: At 60 GPa the
solidus is about 900 K below the melting temperature of MgO. Our data [35] extrapo-
late to the melting point estimation from the shock wave experiment at 130GPa [50].
This supports the assumption that the solidus of the Earth’s mantle material is
dominated by the melting behavior of the (Mg, Fe)@(Mg, Fe)Si03 assemblage.
2.4 Phase Diagrams, Decomposition Reactions, and

Stability of Solids at High Pressures and
Temperatures
The present C02-laserheating technique in a diamond anvil cell allows us to access a
large P-T range (see Fig. 1) for examining structural properties, establishing the
stability fields of solids, and for synthesizing new materials. Additionally, the
present method of laser heating in a diamond cell circumvents several difficulties
associated with prior large volume presses at high temperatures: Hydrostatic
pressures eliminate effects of shear on the transition pressures; no thermocouple
corrections as a function of pressure are required since the temperatures are
measured using the Planck radiation function; the temperatures are measured
2.4 Phase Diagrams, Decomposition Reactions, and Stability of Solids 55
from the same parts of the samples that serve for phase identification, thus, avoiding
problems associated with temperature gradient; the samples are characterized in situ,
avoiding reverse transitions upon pressure release; higher temperatures decrease
possible kinetic problems on phase transitions.
We illustrate the use of the laser heated diamond anvil cell for synthesizing high
pressure phases and establishing phase boundaries of solids with the following three
studies.
2.4.1 Coesite-Stishovite Phase Boundary

We examine first the high pressure and temperature phase boundary between the
tetrahedrally coordinated (coesite) and the octahedrally coordinated (stishovite)
SiOz polymorphs. The phase diagram of Si02 is the focus of intense interest,
because: it has been shown recently that stishovite is the hardest known quenchable
oxide and is a member of an emerging new family of superhard materials based on
the A 0 2 formula (A = Si, Os, Ru, Mn, Sn, Ge, Pb) [51,52]; the coesite-stishovite
transition is used for pressure calibration in multianvil devices. However, there
I ' I ' I ' I . 1 . 1
wesite (tetrahedral coordination)

after heating at 2600 K
10.0 GPa
stishovite (octahedral coordination)

10.9 GPa

12.0 GPa

13.0 GPa
30 500 600 700 800 90

rel. wavenumbers (cm -I)
Figure 11. Raman spectra of laser-heated SOz taken at room temperature which show that stishovite
converts to coesite at 10.0GPa and 2600K. After [17].
"0°1 Z (1993)
Se(1995)
-I
:
. ,'Z(1993)
I- 2000 :
. ,
,>-
1
. .,
: 3
3-..z'Z(1994)
1500 R
1000- C = Coesite
- .
8 1'0 15 1'4
Pressure GPa
Figure 12. Measurements of the coesite-stishovite transformation [ 171: Se. The phase boundary
(dotted line) is an extension of Yagi and Akimoto's results [54]: Y through our data points. Previous
large volume press data from [56]: A, [57]: S, [55,58]: Z are also presented in the figure (see text).
After [ 171.
are strong discrepancies (of order 2 GPa) between existing estimates of this bound-
ary [17,53-551.
The experiments were carried out in a C02-laserheated diamond cell using argon
as pressure medium. The starting materials were thin discs of Si02 glass. Samples
were typically heated for about 1-2min close to their melting temperatures to
avoid kinetic effects on the transition, and then temperature quenched by switching
off the CO, laser. Phase identification of the samples was carried out using Raman
spectroscopy.
The phase boundary was determined with three different experiments [17]. In the
first experiment Si02 glass was converted to stishovite and the phase boundary was
approached in 1 GPa increments until coesite was formed (Fig. 11). Analogously, in
the second experiment, coesite was heated at increasing pressures until stishovite
was formed. In the third experiment coesite or stishovite was synthesized directly
from the starting glass material in the vicinity of the phase boundary. These results
are summarized in Fig. 12 together with previous work on the phase boundary at
lower temperatures. We obtained a phase boundary that is represented by
P(GPA) = 7.8 + 0.0010 T(K). Our estimated slope is less than one half that
obtained by Zhang [ 5 5 , 5 8 ] using multianvil high pressure devices.
2.4.2 High Pressure and Temperature Phase Diagram and

Decomposition Reactions in a Ternary System
In our second study we examined the high P-T behavior of Mg3Al2Si3OI2garnet, a
major constituent of the Earth's upper mantle, and transition zone. Garnets are the
2.4 Phase Diugrams, Decomposition Reactions, and Stability of Solids 57
primary host minerals for A1203at these depths [59], containing structural elements
(polyhedra) of low density solids (tetrahedra), as well as structural elements of high
density phases, such as end-member MgSi03 ilmenite (octahedra), or perovskite
(octahedra and dodecahedra) [60]. It is known that end-member MgSi03 ilmenite
is a low-temperature, high-pressure phase, and transforms to the garnet structure
above 2100 K [61]. Both phases transform to MgSi03-perovskite above about
23GPa [61]. In order to investigate if the addition of aluminum changes the
transition pressure to the perovskite structure, and whether magnesian perovskite
can accommodate the available A1203 in the perovskite structure, we performed
C02-laser heating experiments on Mg3A12Si3012garnet to 45 GPa, and tem-
peratures close to the melting point. This doubles the experimental P-T range of
previous investigations [62,63]. To identify the phases we used Raman as well as
fluorescence spectroscopy. The latter allowed us to identify small amounts of
A1203, which is hard to detect using Raman spectroscopy. To do this, we used
Mg3Al2Si3OI2 : Cr3+ garnet as a starting material because a free A1203: Cr3+ phase
has a strong fluorescence signal. Moreover, the spectrum of A1203:Cr3+ can be
distinguished easily from that of Mg3Al2Si3OI2 :Cr3+. Typical Raman spectra of
pyrope, ilmenite, and perovskite are shown in Fig. 13. Figure 14a shows fluorescence
spectra of Mg3A12Si3012 : Cr3+ after heating in the perovskite stability field.
f i
30 GPa
I after heating at 3000 K

Ilrn
h
c
v)
26.4 GPa
C
3
.-
E
I after healing at 2730 K Ilrn
-
v)
C
0,
-
C
Ilrn a
22.1 GPa
P
P IT
&LA
P
PP
21.2 GPa
I . 1 . 1 . I I . 1 B 7 '
400 600 800 I000 1; I0

re\.wavenumbers (cm-')
Figure 13. Room temperature Raman spectra after heating pyrope glass between 21.2 and 30 GPa
showing pyrope (P), ilmenite (Ilrn), and perovskite (Pv) lines. The peak labeled with an asterisk is an
artifact. After [ 161.
-
2 Phase Transitions and Material Synthesis using the C02-Laser Heating Technique
I - I - 1 -
45.3GPa
R,/R,ruby
43.3 GPa
0 GPa
710 I I
I
708 -
706 -
R,fluorescence from ruby chip:

R,fluorescence from sample -
I I 1 I
0 10 20 30 40 50
b Pressure (GPa)
Figure 14. (a) Room temperature fluorescence spectra after heating Mg3Al2Si3OI2 : Cr3+ pyrope
crystals at about 2500 K at various pressures. R1 and R2 are the primary electronic peaks due to
Cr3+-A1203, the peaks labeled with N are N-lines (Nebenlinien) and those labeled with S are
vibrational sidebands of Al2O3 [64]; the feature labeled with an asterisk is residual fluorescence
from unconverted pyrope. (b) Comparison of the pressure dependencies of the R I fluorescence
lines of a ruby chip (located at the edge of the gasket) and A1203:Cr3+ produced after heating
Mg3A12Si3012 :Cr". Agreement of the R I peak positions of the ruby chips and the Cr3+-A1203
in the center of the sample shows that no pressure gradients are present in the sample chamber
after heating. After [16].
2.4 Phase Diagrams, Decomposition Reactions. and Stability of Solids 59
P(GPa)
Figure 15. Mg3A12Si3012-pyrope-ilmenitcperovskitephase relations. The diagram also includes
data points from earlier multianvil work by [63]: K and [62]: I. After [16].
Pyrope transforms to the ilmenite structure above 21.5 GPa, to perovskite plus
ilmenite above 24GPa, and to perovskite above 29GPa (Fig. 15). The stability
field of the ilmenite structure is expanded to higher temperatures with respect to
end-member MgSi03-ilmenite. The transition pressures to aluminous perovskite
is about 2GPa higher with respect to pure MgSi03. Furthermore, we found that
some A1203 exists as a free phase together with perovskite up to 43 GPa, because
we detected A1203:Cr3+fluorescence up to this pressure (Fig. 14). Our results
show that the perovskite structure can accommodate at least 25mole% A1203
above 43 GPa.
2.4.3 Stability of a Perovskite Oxide with Respect to its Component

Oxides
In the third study we investigated whether MgSi03-perovskite, considered to be the
most abundant mineral in the Earth, decomposes to its component oxides above
65 GPa, as suggested by recent studies [65]. We performed for the first time C02-
laser heating experiments on the stability of MgSi03 perovskite under hydrostatic
conditions between 73 and l00GPa and up to 3000K. In addition to using
MgSi03 glasses and crystals, we also employed a new approach using stoichiometric
mixtures of Si02 and MgO as starting materials. MgSi03-perovskite formed as the
single phase to the highest pressures and temperatures of our measurements
(lOOGPa, 3000K) [66].
Thus, at the present pressure-temperature conditions, which span most of the

Earth's lower mantle, silicate perovskite is denser than an assemblage of its compo-
nent oxides, MgO and SO2.
2.5 C02-laser Heating Experiments on Organic

Compounds
The C02-laser heated DAC can be used for in situ investigation of the high P-T
behavior of organic compounds containing carbon, boron, and nitrogen, as well
as the synthesis of new materials. Graphite and diamond, hexagonal and cubic
BN, molecular hydrogen, and nitrogen can be easily identified in the diamond
cell, or in the recovered reaction products by Raman spectroscopy.
It was shown [67] that diamond can be synthesized from a number of organic
compounds (like polyethylene or paraffin) at high pressures and temperatures.
Shock-wave experiments on methane [68] suggest that this alkane decomposes
into molecular hydrogen and diamond at pressures above 20 GPa and temperatures
of about 2500 K. In contrast, recent first-principles molecular dynamic simulations
[69] suggest that below 100 GPa methane polymerizes into a mixture of methane,
ethane, and butane. In order to examine if the decomposition products depend on
the chain length of the starting hydrocarbon material, we performed C02-laserheat-
ing experiments in a DAC on methane CH4, octadecane C18H38, and nonadecane
C19H4,, [70]. In all these experiments no pressure medium was used, resulting in
axial temperature gradients of several hundred K pm-' [21], thus the temperature
values given below are only approximate. After heating at around 20 GPa, we
observed formation of diamond for all three hydrocarbons, as evident from
Raman-spectroscopic analysis of the recovered samples at 1 bar. We note that
other decomposition products escaped from the sample volume after opening the
diamond cell. We found that diamond formed from longer chain alkanes (e.g.
octadecane) at lower temperatures than from short ones (e.g. methane). Figure
16A shows an SEM picture of the recovered diamond aggregate synthesized from
octadecane at 21 GPa and at temperatures exceeding 3000K. On the surface of
this diamond aggregate we found a few spheres with diameters of less than 1 pm,
which indicated melting of diamond (see Fig. 16B). The above results suggest that
alkanes can be used as a carbon source for diamond synthesis.
We also performed C02-laser heating experiments on borane-dimethylamine
BH3 NH(CH3)*to check if organic materials containing boron and nitrogen can
be used for synthesis of ternary BNC, compounds with diamond-like structure.
Borane-dimethylamine was heated at 23 GPa to about 3000 K. Raman spectra
from the heated sample area revealed the presence of cubic BN, and of liquid H2
at pressures below the solidification point of hydrogen (5.5 GPa at room tempera-
ture [71]) (Fig. 17). A Raman spectrum of the recovered white agglomerate
showed a weak line of diamond in addition to the Raman bands of cubic BN.
Thus, borane-dimethylamine transformed to cubic BN, diamond and hydrogen
2.5 COJaser Heating Experiments on Organic Compounds 61
Figure 16. (A) SEM-image of the diamond agglomerate recovered from a diamond cell after heating
of octadecane to above 3000K at 21 GPa. The area bounded by the dashed lines is enlarged in (B),
showing two spheres of molten diamond (indicated by arrows) on the agglomerate surface.
rn
600 800
I ' I . , . I . , .
400 1000 1200 1400 4200 4400
rel. wavenumber (cm-')
Figure 17. Raman spectra of liquid H2 and cubic BN at 0.6 GPa in a DAC after heating of borane-
dimethylamine to about 3000 K at 23 GPa.
after heating at 23 GPa. No evidence for the presence of ternary BNC, in the heated
sample volume was found.
2.6 Conclusion
The studies described above show that the C02-laser heated diamond anvil cell can
be used for investigating nonconducting materials under controlled conditions in a
very broad pressure and temperature range. Samples can be analyzed in situ using
Raman spectroscopy, fluorescence spectroscopy, and X-ray diffraction. For
structural, chemical, and textural analysis of the recovered samples techniques
such as atomic force microscopy, scanning- and transmission electron microscopy
can be employed. All of the above techniques can be used for determining melting
curves, developing high P-high T phase diagrams, evaluating the stability of solids,
and monitoring of chemical reactions, which are essential for controllable synthesis
and characterization of new materials.
Acknowledgments
We thank 0. Tschauner for fruitful discussions and the Editor of the book for
encouraging us to undertake this work. A.Z. gratefully acknowledges the financial
support of the Deutsche Forschungsgemeinschaft.
References
1. A. L. Ruoff, H. Xia, and Q. Xia, The effect of a tapered aperture on x-ray diffraction from
sample with a pressure gradient: Studies on three samples with a maximum pressure of
560GPa, Rev. Sci. Znstrum. 1992, 63, 43424348.
2 . M. Eremets, High Pressure Experimental Methods, Oxford University Press, Oxford, 1996.
3. R. M. Hazen and L. W. Finger, Comparative Crystal Chemistry, John Wiley & Sons Ltd.,
Chichester, 1982.
4. A. Jayardman, Diamond anvil cell and high-pressure physical investigations, Rev. Mod Phys.
1983,55, 65-108.
5 . W. L. Wolfe and G. J. Zissis, The Infrared Handbook, Office of Naval Research, Department of
the Navy, Washington, DC, 1985.
6. R. A. Forman, G. J. Piermarini, J. D. Barnett, and S. Block, Pressure measurement made by the
utilization of ruby sharp-line luminescence, Science, 1972, 176, 284-285.
7. H. K. Mao, J. Xu, and P. M. Bell, Calibration of the ruby pressure gauge to 800 kbar under
quasi-hydrostatic conditions, J . Geophys. Res. 1986, 91, 46734676.
8. R. Boehler, M. Nicol, and M. L. Johnson, Internally heated diamond anvil cell: Phase diagram
and P-V-T of iron, in High Pressure Research in Mineral Physics, Geophys. Monogr. Ser., M. H.
Manghnani and Y. Syono (Eds), AGU, Washington, DC, 1987, pp. 173-176.
Reterences 63
9. L. G. Liu and W. A. Bassett, The melting of iron up to 200 kbar, J . Geophys. Res. 1975, 80,
3777.
10. M. Planck, Entropie und Temperatur strahlender Warme, Ann. P h j x 1900, 1, 719-737.
11. L. Ming and W. A. Bassett, Laser heating in the diamond anvil press up to 2000°C sustained
and 3000°C pulsed at pressures up to 260 kilobars, Rev. Sci. Instrum. 1974, 45, 11 15-1 118.
12. C.-S. Yoo, J. Akella, and M. Nicol, Chemistry at high pressures and temperatures: In-situ
synthesis and characterization of P-Si3N4 by DAC X-ray/laser heating studies, in The 3rd
NIRIM International Symposium on Advanced Materials (ISAM'96), M. Akaishi et al. (Eds),
NIRIM, Tsukuba, Japan, 1996, pp. 175-179.
13. R. Boehler and A. Chopelas, A new approach to laser heating in high pressure mineral physics,
Geophys. Res. Lett. 1991, 18, 1147-1 150.
14. M. Eremets, K. Takemura, H. Yusa, D. Goldberg, Y. Bando, and K. Kurashima, Melting of
cubic and hexagonal BN, in The 3rd NIRIM International Symposium on Advanced Materials
(ISAM'96). M. Akaishi et al. (Eds), NIRIM, Tsukuba, Japan, 1996, pp. 169-174.
15. J. H. Nguyen and R. Jeanloz, Initial description of a new carbon-nitride phase synthesized at
high pressures and temperatures, Muter. Sci.Eng. 1996, A209, 23-25.
16. G. Serghiou, A. Zerr, A. Chopelas, and R. Boehler, The transition of pyrope to perovskite,
Phj:F. Chem. Min. 1998, 25, 193-196.
17. G. Serghiou. A. Zerr, L. Chudinovskikh, and R. Boehler, The coesite-stishovite transition in a
laser-heated diamond cell, Geophys. Res. Lett. 1995, 22, 441444.
18. F. Cabannes, Facteurs de reflexion et d'emission des metaux, J. Physique, 1967, 28, 235-248.
19. J. C. De Vos, A new determination of the emissivity of tungsten ribbon, Physica, 1954, 20,69&
714.
20. U. P. Oppenheim and A. Goldman, Infrared spectral transmittance of MgO and BaF, crystals
between 27" and IOOO'C, J . Optic. Soc. Am. 1964,54, 127-128.
21. A. Zerr, Schmelztemperaturen von (Mg, Fe)S03-Perowskit und von MgO bei hohen Drucken,
PhD thesis, University of Mainz, Mainz, 1995.
22. S. R. Elliot. Physics of Amorphous Materials, Longman, London, 1990.
23. R. J. Hemley, P. M. Bell, and H. K. Mao, Laser techniques in high-pressure geophysics,
Science, 1987, 237, 605-612.
24. G. Serghiou, A. Zerr, and A. Chopelas, Phase transitions and disordering of CaSi03 upon com-
pression, Rev. High Press. Sci. Techno/. 1998, 7, 359-361.
25. A. Chopelas, The fluorescence sideband method for obtaining acoustic velocities at high com-
pressions: application to MgO and MgA1204, Phys. Chem. Min.1996, 23, 25-37.
26. A. Chopelas and R. Boehler, Thermal expansivity in the lower mantle, Geophys. Res. Lett. 1992,
19, 1983-1986.
27. F. R. Corrigan and F. P. Bundy, Direct transition among allotropic forms of boron nitride at
high pressures and temperatures, J . Chem. Phys. 1975, 63, 3812-3820.
28. D. A. Young, C . 4 . Zha, R. Boehler, J. Yen, M. Nicol, AS. Zinn, D. Schiferl, S. Kinkead, R. C.
Hanson, and D. A. Pinnick, Diatomic melting curves to very high pressure, Phys. Rev. B, 1987,
35, 5353-5356.
29. P. E. van Keken, D. A. Yuen, and A. P. van den Berg, Implications for mantle dynamics from
the high melting temp,eratures of perovskite, Science, 1994, 264, 1437-1439.
30. %-I. Karato, Importance of anelasticity in the interpretation of seismic tomography, Geophys.
Res. Lett. 1993, 20, 1623-1626.
31. A. Zerr and R. Boehler, Constraints on the melting temperature of the lower mantle from high
pressure experiments on MgO and magnesiowustite, Nature, 1994,371,506-508.
32. B. Riley, The determination of melting points at temperatures above 2000" Celcius, Rev. Hautes
Temper. Refract. 1966, 3, 207-2 11.
33. A. Zerr, G. Serghiou, and R. Boehler, Melting of CaSi03-perovskiteto 430 kbar and first in-situ
measurements of lower mantle eutectic temperatures, Geophys. Res. Lett. 1997, 24, 909-912.
34. A. Zerr and R. Boehler, Melting of (Mg, Fe)Si03-perovskite to 625 kbar: indication of a high
melting temperature in the lower mantle, Science, 1993, 262, 553-555.
35. A. Zerr, A. Diegeler, and R. Boehler, Solidus of Earth's deep mantle, Science, 1998,281,243-246.
36. J. Zhang and C. Herzberg, Melting experiments on anhydrous peridotite KLB-1 from 5.0 to
22.5 GPa, J . Geophys. Res. 1994,99, 17729-17742.
64 2 Phase Transitions and Material Synthesis using the CO,-Laser Heating Technique
37. Z. Gong, R. E. Cohen, and L. L. Boyer, Molecular dynamics simulations of melting of MgO at
high pressures, in Annu. Rep. Director Geophys. Lab. Carnegie Instn. Wash., Washington, DC,
1991, pp. 129-134.
38. I. Jackson, Phase relations in the system LiF-MgF2 at elevated pressures: Implications for the
proposed mixed-oxide zone of the Earth’s mantle, Phys. Earth Planet. Inter. 1977, 14, 86-94.
39. E. Ohtani, Melting temperature distribution and fractionation in the lower mantle, Phys. Earth
Planet. Inter. 1983, 33, 12-25.
40. J. J. Gilvarry, The Lindemann and Gruneisen laws, Phys. Rev. 1956, 102, 308-316.
41. F. A. Lindemann, Uber die Berechnung molekularer Eigenfrequenzen, Phys. Z . 1910,11,609-
612.
42. E. A. Kraut and G. C. Kennedy, New melting law at high pressures, Phys. Rev. Lett. 1966,16,
608-609.
43. R. Boehler, Adiabats of quartz, coesite, olivine, and magnesium oxide to 50 kbar and 1000K,
and the adiabatic gradient in the Earth’s mantle, J . Geophys. Res. 1982,87, 5501-5506.
44. D. Heinz and R. Jeanloz, Measurement of the melting curve of Mgo 9Feo Si03 at lower mantle
conditions and geophysical implications, J. Geophys. Res. 1987,92, 11437-1 1444.
45. E. Knittle and R. Jeanloz, Melting curve of (Mg, Fe)Si03 perovskite to 96 GPa: evidence for a
structural transition in lower mantle melts, Geophys. Res. Lett. 1989, 16,421424.
46. J. S. Sweeney and D. L. Heinz, Melting of iron-magnesium-silicate perovskite, Geophys. Res.
Lett. 1993, 20, 855-858.
47. G. Shen and P. Lazor, Measurement of melting temperatures of some minerals under lower
mantle pressures, J . Geophys. Res. 1995, 100, 17699-17713.
48. E. lto and E. Takahashi, Melting of peridotite at uppermost lower-mantle conditions, Nature,
1987,328, 514-517.
49. E. Takahashi, Melting of a dry peridotite KLB-1 up to I4GPa: Implications on the origin of
peridotitic upper mantle, J. Geophys. Res. 1986, 91, 9367-9382.
50. K. G. Holland and T. J. Ahrens, Melting of (Mg, Fe)2Si04at the core-mantle boundary of the
Earth, Science, 1997, 275, 1623-1625.
51. J. Haines, J. M. Leger, and 0. Schulte, Pa3 modified fluorite-type structures in metal dioxides at
high pressure, Science, 1996, 271, 629-631.
52. J. M. Leger, J. Haines, M. Schmidt, J. P. Petitet, A. S. Pereira, and J. A. H. Jornada, Discovery
of hardest known oxide, Nature, 1996, 383, 401.
53. V. Swamy, S. K. Saxena, B. Sundman, and J. Zhang, A thermodynamic assessment of silica
phase diagram, J. Geophys. Res. 1994,99, 11787-1 1794.
54. T. Yagi and S. Akimoto, Direct determination of coesite-stishovite transition by in-situ X-ray
measurements, Tectonophysics, 1976,35, 259-270.
55. J. Zhang, B. Li, W. Utsumi, and R. Liebermann, Reversal of the coesite-stishovite phase trans-
formation, in AGU Spring Meeting, American Geophysical Union, Baltimore, 1994, p. 346.
56. S. Akimoto and Y. Syono, Coesite-stishovite transition J . Geophys. Res. 1969,74, 1653-1659.
57. K. Suito, Phase relations of pure Mg2Si04up to 200 kilobars, in High Pressure Research: Appli-
cation in Geophysics, M. H. Manghnani and S. Akimoto (Eds), Academic Press, Inc., New
York, 1977, pp. 255-266.
58. J. Zhang, R. C. Liebermann, T. Gasparik, and C. T. Herzberg, Melting and subsolidus relations
of Si02 at 9-14GPa, J. Geophys. Res., 1993,98, 19785-19793.
59. A. E. Ringwood, Composition and Petrology of the Earth’s Mantle, McGraw-Hill, New York,
1975.
60. P. McMillan, M. Akaogi, 1. Ohtani, Q. Williams, R. Nieman, and R. Sato, Cation disorder in
garnets along the Mg3A12Si3012-Mg4Si,0,2join: an infrared, Raman and NMR study, Phys.
Chem. Min. 1989, 16,428435.
61. H. Yusa, M. Akaogi, and 1. Eiji, Calorimetric study of MgSi03 garnet and pyroxene: heat
capacities, transition enthalpies, and equilibrium phase relations in MgSi03 at high pressures
and temperatures, J . Geophys. Res. 1993, 98, 64534460.
62. T. Irifune, T. Koizomi, and J. Ando, An experimental study of the garnet-perovskite transfor-
mation in the system MgSiO3-Mg3AI2Si3Ol2,Phys. Farth Planet. Inter. 1996, 15, 90-106.
63. M. Kanzaki, Ultra high-pressure phase relations in the system MgSiO3-Mg3AI2Si3Ol2,Phys.
Earth Planet. Inter. 1987, 49, 168-175.
References 65
64. L. Zhang and A. Chopelas, Sound velocity of A1203 to 616kbar, Phys. Earth Planet. Inter.
1994,87, 77-83.
65. S. K. Saxena, L. S. Dubrovinsky, P. Lazor, Y. Cerenius, P. Haggkvist, M. Hanfland, and
H. Jingzhu, Stability of perovskite (MgSi03)in the Earth’s mantle, Science, 1996,274, 1357-1359.
66. G. Serghiou, A. Zerr, and R. Boehler, (MgFe)Si03-perovskite stability under lower mantle
conditions, Science, 1998,280, 2093-2095.
67. R. H. Wentorf, The behavior of some carbonaceous materials at very high pressures and high
temperatures, J. Phys. Chem. 1965,69, 3063-3069.
68. M. Ross and F. H. Ree, Repulsive forces of simple molecules and mixtures at high density and
temperature, J Chem. Phys. 1980, 73, 614M152.
69. F. Ancilotto, G. L. Chiarotti, S. Scandolo, and E. Tosatt, Dissociation of methane into hydro-
carbons at extreme (planetary) pressure and temperature, Science, 1997, 275, 1288-1290.
70. A. Zerr, G. Serghiou, and R. Boehler, C02-laser heating in a diamond cell to 80GPa
and 5000 K: phase transitions and synthesis under controlled conditions, in The 5th NIRIM
International Symposium on Advanced Materials (ISAM’98), M. Kamo et al. (Eds), NIRIM,
Tsukuba, Japan, 1998, pp. 5-8.
71. S. K. Sharma, H. K. Mao, and P. M. Bell, Raman measurements of hydrogen in the pressure
range 0.2-630 kbar at room temperature, Phys. Rev. Lett. 1980, 44,886-888.
3 Mechanical Properties and their Relation to

Microstructure
D. Sherman and D. Brandon
3.1 Introduction
In this contribution we review current knowledge of the mechanical properties of
hard materials and the relation between these properties and their microstructure.
However, a quantitative relationship is only meaningful to the extent that test
methods are clearly defined and understood, while the test data are only useful
within the context of the engineering requirements which determine material
selection for a specific application. It is also important to define what is meant by
a ‘hard material’.
We therefore discuss briefly some typical applications of hard materials and the
engineering requirements that must be satisfied in these applications. In Section
3.2 we define the mechanical properties which need to be evaluated, while in Section
3.3 we review the test method which have been developed to ensure accurate and
reproducible measurements. We then examine the microstructure-property rela-
tions and the size effects that dominate the mechanical strength of brittle materials,
as well as some important failure mechanisms.
3.1.1 Applications and Engineering Requirements

Structural components are required to transmit a load, and their mechanical
strength and stiffness are the primary concerns of the design engineer [1,2].
Hard materials, on the other hand, are selected, as the term implies, for their hard-
ness and we should consider why ‘hardness’ is not a synonym for ‘strength’. A
good definition for hardness is ‘the resistance to penetration’ [3], emphasizing
that this property relates to the surface response to a mechanical load, and is
not necessarily a ‘bulk’ property. Hard materials may be used for structural com-
ponents, but they are also used as coatings [4,5], to improve the resistance of a
structural component to superficial mechanical damage. Thin, hard coatings
may be submicron, but are more commonly a few microns in thickness. Thick
coatings may approach or exceed a millimeter in thickness. All hard coatings are
dependent for their engineering performance on their adhesion to the substrate,
which is commonly achieved by combining a high-strength chemical bond with
control of the substrate roughness, in order to optimize mechanical interlocking
[6]. Coatings are susceptible to the presence of residual stress, either associated
with the mismatch in physical properties across the coating-substrate interface,
or due to the deposition mechanism. Compressive residual stress in the coating
3.1 Introduction 67
may be beneficial, inhibiting subcritical crack growth and delaying the onset of
damage, but residual tensile stresses are always detrimental. Mismatch in the
thermal expansion coefficient and the elastic moduli of the coating and the sub-
strate are common sources of residual stress.
It is important to distinguish between elastic and plastic penetration of an
indentor. An elastomer has a low resistance to elastic deformation, but releasing
the load will generally leave no mark on the rubbery elastomer. By contrast, a
refractory metal may have a poor resistance to plastic penetration, but show rela-
tively little elastic relaxation when the indentor load is relaxed [3]. Throughout
this review we assume the elastic response to be small (but not negligible) compared
to the plastic response, so that the high hardness of a ‘hard material’ is also accom-
panied by high stiffness, corresponding to a large tensile modulus.
The mode of loading and the rate at which load is applied at the surface is
important [3]. There is a clear distinction between a sliding bearing, in which
contacts between the components are predominantly subjected to shear parallel
to the contact surface, and a roller (or ball) bearing in which the mechanical con-
tacts are made and broken in tension (perpendicular to the contact surface). Most
mechanical contacts can be approximated to ‘point’ contacts, since the ‘perfectly
flat’ surface has yet to be invented. More exactly, the ‘true’ contact area depends
on the load applied, the rigidity of the components, and their yield strength. A high
hardness implies a minimum contact area at a given normal load. It follows that
contact damage is, in the first instance, localized and restricted to the contact
regions. Mechanical wear occurs by removal of material from points of contact,
followed by transfer of the load to new points of contact. Areas worn down by
mechanical wear will establish renewed contact once neighboring areas are
similarly eroded, leaving the original contact areas once more proud of the nom-
inal contact surface.
Since surface damage is localized, events taking place at the surface often occur
rapidly. True contact areas are often on the scale of the surface finish, less than a
micron for a well-polished surface. Sliding velocities of the order of meters s-’
then correspond to contact times of less than a microsecond, so that the localized
damage events are commonly adiabatic, and accompanied by intense ‘frictional’
heating [ 7 , 8 ] . Hence the importance of cooling the workpiece in all grinding,
polishing, and mechanical working operations. Susceptibility to brittle failure is
the penalty accompanying the desirable properties of hard materials [9]. Premature
brittle failure can be prevented, either by suitable engineering design or by ensuring
that brittle crack propagation is inhibited. In successful engineering design tensile
loading is minimized, since brittle materials are commonly strong in compression.
This may be achieved either by pre-loading in compression (assembly of the compo-
nent in constraint) or by ensuring that the operating loads are compressive. Fracture
toughness is the common parameter used to measure the resistance to brittle crack
propagation [9-111, but many hard materials exhibit R-curve behavior, in which the
critical stress intensity factor increases with crack length [9,12]. The mode I fracture
toughness determines the resistance to macroscopic crack propagation, but the
short-crack resistance or ‘damage tolerance’ [131 often determines the performance
in service, when damage is commonly localized at the surface.
68 3 Mechanical Properties and their Relation to Microstructure
Clearly, the primary engineering requirements for bearing components and cut-
ting tools include high stiffness and hardness, maintained to high temperatures
[14]. These are also the primary considerations in the selection of grit for grinding
wheels and polishing compounds. Erosion-resistant surface coating systems are
often based on hard and rigid materials, which may also employ elastomeric
materials to resist plastic but not elastic penetration. Armor systems are also
designed to resist penetration, at both sub- and supersonic velocities, and also
employ hard materials [15,161.
Finally, two other characteristics of hard materials are important: their density
and their chemical stability. The importance of these properties depends on the
application. Oxides are chemically more stable than nitrides, which are in turn
more stable than borides or carbides. The susceptibility to oxidation of boron
carbide prevents its application at high temperatures, but for ballistic protection,
where hardness, rigidity and low density take precedence, boron carbide is
unparalleled. Tungsten carbide, on the other hand, is the material of first choice
for cutting tools, because of its high hardness and stiffness, and high-temperature
oxidation resistance, but is far too dense for application in which weight or
inertial forces are important. Its use for ballistic protection is therefore out of
the question.
3.1.2 Bulk Components

Hard materials available for the production of bulk components fall into two major
groups: ‘hard metals’, used primarily in the manufacture of cutting tools and related
applications, and structural ceramics, primarily the oxides, carbides, borides, or
nitrides of the low atomic number cations. Relatively few materials are of engineer-
ing importance, and we will list these, explaining briefly why similar compositions
are less useful.
3.1.2.1 Tungsten Carbide-Cobalt, Tungsten Carbide-Nickel,

Vanadium Carbide
The bonding of tungsten carbide particles by cobalt (and, to a lesser extent, nickel)
has dominated the cutting tool industry for over 70 years [17,18]. There has been a
steady improvement in the compositions available, including additional carbide
particulates, control of the cobalt-nickel binder composition, and a reduction in
the tungsten carbide particle size. The commonest cutting tool formulations contain
6-12 weight-% binder. Production of these ‘hard metal’ components depends on
active wetting of partially soluble carbide particles by the molten binder phase.
The formation of deleterious secondary phases during liquid phase sintering and
cooling to room temperature must be avoided.
The success of hard metals in cutting tool applications is a result of their unique
combination of hardness and toughness, tailored by controlling the ratio of binder
to carbide phase, together with their unique microstructure. Both the ductile binder
3.1 Introduction 69
and the hard carbide phases are continuous. Since the ductile binder is continuous,
no crack can propagate without expending energy in ductile failure of the binder. On
the other hand, the three-dimensional continuity of the carbide skeleton maximizes
rigidity and stiffness of the composite for any given carbide content, so toughness is
obtained with a minimum loss of hardness.
Bulk cutting tools based on alternative materials, most notably alumina and
silicon nitride, have been developed [ 19,201, but, although niche applications have
been found, no substitute for tungsten carbide has yet been discovered. The closest
to a ‘breakthrough’ has been the development of polycrystalline diamond (which
lacks oxidation resistance) [21]. More recently, cubic boron nitride (cBN) has
found a large and growing market in high speed machining thanks to a combination
of hardness and oxidation resistance [22,23].
3.1.2.2 Ceramics: Oxides, Carbides, Borides, Nitrides

While ‘hard metals’ may have a fracture toughness of up to 20MPam1/2, most
engineering ceramics have a toughness well below 10 MPa m1I2 [ l ,3,9]. The sole
exceptions are the composite microstructures, associated with ‘directed metal
oxidation’ [24] and some other processing routes, and the transformation-
toughened zirconia compositions. Most of these materials lack the stiffness and
hardness required for wear-resistant components. The strength and rigidity of the
chemical bond in a ceramic depend on the valency and the radius of the cation
[25]. The bond strength (but not the chemical stability) generally decreases in
going from carbides to nitrides to oxides, while the highest elastic moduli are
expected for low atomic number cations. Any choice is a compromise in which
the primary requirements for hardness and stiffness (optimizing wear resistance)
are modified by secondary requirements for chemical stability, inertial response
and selected thermal properties (expansion coefficient, thermal conductivity or
thermal diffusivity).
Many ceramics have limited engineering applications because they fail to meet
one or other of the secondary requirements. For example, both magnesium
oxide and aluminum carbide are unstable in the presence of moisture. In other
cases an otherwise important engineering ceramic is unsuitable for a specific appli-
cation. Thus alumina has poor toughness and a relatively large thermal expansion
coefficient. Thermal stresses generated by adiabatic heating at the surface of an
alumina wear component can cause surface microdamage, leading to the formation
of hard wear particles which may cause abrasive damage of components far
removed from the site of the alumina bearing surface. Silicon nitride bearings,
on the other hand, have found a rapidly growing market, thanks to painstaking
development of high performance grades of silicon nitride, with high failure
strength and fracture toughness at temperatures of up to 1200°C. Silicon nitride
has a lower elastic modulus than alumina, a significantly smaller coefficient of
thermal expansion, and on oxidation forms soft, often protective, silica films.
This combination of properties gives better wear resistance than can be achieved
in the aluminas. By contrast, alumina is the material of first choice for the ball
joint in a hip prosthesis, mounted against a high density polyethylene cup. In
70 3 Mechanical Properties and their Rehtion to Microstrircture
this application lack of toxicity is a primary concern, while the wear path is
limited, so that wear is not a major concern.
3.1.3 Coatings
Hard coatings can solve the problem of maximizing the hardness of the surface, to
optimize wear resistance, without sacrificing the toughness of the bulk component
and risking brittle fracture [26]. A successful hard coating must be compatible
with the substrate, exhibiting good adhesion and minimizing residual stress in the
coating. A wide spectrum of coating technologies is available, encompassing
electrolytic coatings, anodizing, diffusion coatings, plasma spraying, chemical
vapor deposition, laser deposition, and more. The various coating technologies
have one thing in common: the extraordinary difficulty in determining their mechan-
ical properties. The problem can be reduced to three separate factors: The mismatch
in elastic properties between the coating and the substrate, which leads to interfacial
shear stresses at any discontinuity in the coating; the dependence of the mechanical
properties on thickness, both because of thickness-dependent microstructural
changes and owing to a size dependence of the properties; difficulties in standardiz-
ing and interpreting mechanical tests on coatings.
These considerations are independent of other problems: surface roughness
and adhesion, which are usually avoided by assuming, or requiring, that the
adhesive strength of the coating-substrate interface exceeds the cohesive strength
of the coating, and residual stresses in the coating, typically biaxial, which also
affect the critical load for microcracking, peeling, and (eventually) spalling
(flaking).
3.1.4 Engineering Requirements

It is the ability of a material to satisfy the engineering requirements for a given com-
ponent which is important, not the mechanical properties as such. Determining the
mechanical properties is, in many respects, just a way of grading the desirability of
candidate materials for a specific task. Specification of the value of specific proper-
ties (bend strength, elastic modulus, or fracture toughness) is generally insufficient
(by itself) to determine fitness for purpose. The engineering requirements for specific
applications (wear components, cutting tools, mechanical bearings, or armor) vary
dramatically. A wear component must exhibit high wear resistance, and good
performance is associated with a high elastic modulus, a high damage tolerance,
and good chemical stability (absence of attack by acid residues in a lubricating
oil, for example). A good candidate for composite armor is expected to have a
high Vso(the velocity at which a given projectile has a 50% chance of penetration).
This is clearly a function of the armor thickness and the rigidity of the mounting,
and it is usual to add that the weight per unit area should not exceed a given
value (placing a premium on a low density material such as boron carbide).
3.2 Principal Mechanical Properties 71
3.2 Principal Mechanical Properties

3.2.1 Elastic Modulus
The elastic moduli are an intrinsic mechanical property of a material dominated by
the strength of the chemical bond between atoms. The potential energy associated
with the bond between a pair of atoms is determined by a balance of the forces of
attraction and repulsion
where a and ,B are constants related to the attraction and repulsion components, r is
the distance between the atoms, and n > m. All four constants can be treated as
material properties. The net force between the atom pair is just
and is zero for r = ro, the equilibrium spacing in the absence of external forces. An
external driving force displaces the atoms by a distance u from the equilibrium
position, where u = r - ro, and the relationship between the external driving force
and the displacement is
F=-=-
dEp u. (3)
dr U=O
The second derivative, which determines the curvature, of the potential energy well
near y o , is constant for a given material and temperature, leading to Hooke's law:
*I
F = Ku. The constant K is proportional to the elastic, or Young's modulus
K o(
dU2 "=o
= f ( a ,p, n, m). (4)
K is therefore the proportionality constant between stress and strain in elastic

deformation. The modulus of elasticity is effectively independent of microstruc-
ture and minor alloying additions, but does depend on temperature, reflecting
changes in equilibrium spacing associated with thermal expansion. Ionic or
covalent bonds deform less than metallic bonds under a given external force,
and so have a higher modulus of elasticity. The high elastic modulus of hard
materials reflects the strong chemical bonds between the atoms. For comparison,
the elastic moduli of steel, alumina, and WC are 210, 400, and over 500GPa,
respectively [ 1,2,9].
Many hard materials of practical importance are polyphase materials, and it is
important to formulate their effective elastic modulus. Two bounds exist for
calculating the composite modulus, based on elastic deformation of the individual
phases. The upper, Voigt bound, assumes compatible displacements or strains.
The elastic modulus, Ec, is then given by the ‘rule of mixtures’:
Ec = cn
EiL,
whereh is the volume fraction of component i. The Reuss model gives a lower
bound for the elastic modulus, and assumes compatible loads or stresses:
1 J;:
E,=FE+
Most polyphase materials have elastic moduli which lie between these two bounds,
and more sophisticated models have been developed to narrow the gap. The
Hashin-Shtrikman model, in particular [27], has given better predictions for the
elastic modulus of a composite, while, more recently, Aboudi has used a cell
model [28] and finite element analysis has also been used to derive the relationship
between the volume fraction and effective modulus [29-321. These methods also
estimate the displacement and stress field in the composite.
3.2.2 Strength
The mechanical strength of hard materials is critical for load-bearing, structural
applications. These brittle materials only deform plastically at high temperatures,
or under severe hydrostatic constraint, since the Peierls stress for dislocation move-
ment is high. Failure is usually by unstable crack propagation under a tensile stress
that exceeds the tensile strength of the material. In terms of fracture mechanics, brit-
tle failure occurs when the Mode I stress intensity factor KI reaches the fracture
toughness of the material, K I c (see below).
An experimental uniaxial stress-strain relationship, determined in tension, would
provide the necessary design information for ceramics and hard materials: Young’s
modulus, Poisson’s ratio and the tensile fracture strength. However, tensile data for
brittle materials are often unreliable, due to parasitic bending stresses associated
with dimensional inaccuracy in machining tensile ‘dog-bone’ specimens and mis-
alignment of the specimen grips.
The test procedure most commonly used to validate the strength of ceramics and
hard materials is a flexure test, in either three-point (3PB) or four-point bending
(4PB). The specimens for bend tests are simple to machine and require less material,
while misalignment and parasitic bending can be avoided. The strength determined
in bending is variously termed the flexural strength, the transverse rupture strength,
TRS, or the modulus of rupture, MOR, and corresponds to the maximum tensile
stress developed in the surface ligaments of the sample at the point of fracture.
Several issues need to be addressed when a bend test is used to evaluate the
strength: The sources of errors that are responsible for statistical scatter in the
results; the statistical nature of the strength of a brittle material; the relationship
between the measured flexural strength and the strength derived from a valid
uniaxial tensile test, together with the effect of sample size on the strength; and
Loadpointfriction u
. - Twisted specimen
orfixture
n n
Reducedmomentofinertia
Figure 1. Source of errors in bending tests. Reproduced with permission of . . . . . . . . . . (Ref. 33).
the relationship between the properties measured on a laboratory sample or test

piece with those required of an engineering component under operating conditions.
A major cause of experimental scatter in determining the strength of brittle
materials is the deviation from ideal geometry [33]: nonprismatic specimens, a non-
parallel loading path, deviations from pure articulation, eccentricity in point-loading,
inaccurate measurements of load span and specimen dimensions, or inadequate
span :height ratio; all resulting in nonuniform bending and stress concentrations,
Fig. 1. 'Real' scatter of material properties is typically associated with processing
defects and microstructural inhomogeneity in the material, especially residual
porosity, impurity inclusions, and exaggerated grain growth.
Brittle materials fail by unstable crack propagation from a pre-existing flaw. The
size, orientation, and shapes of the flaws in a given flaw population determine the
variability of the failure stress; the statistical failure probability, F, first formulated
by Weibull [34] and described by the Weibull function:
F = I -exp[-jv[g]"du], (7)
where uoand m denote the median failure stress [35,36] and the Weibull modulus, u,
is the volume of the test piece. The material parameter uomeasures the amplitude of
the failure strength, while the parameter rn measures the strength variability. If the
failure probability function for N specimens is known, then the failure probability of
+
the ( N 1)th specimen can be predicted. The number of tests N required to ensure
predictive reliability is typically 20.
If a 3PB test piece is loaded, the volume element subjected to the maximum tensile
stress is very small. This volume is much larger for a 4PB specimen, but a uni-
directional tensile specimen will have the largest volume under maximum stress. It
follows that the median strength of a 3PB specimen may exceed that of a 4PB
specimen, which in turn should have a higher average strength than a tensile
specimen. The measured strength ratio between specimens of identical size subjected
to 3PB and to pure tension is predicted to be [34,37]:
which yields 1.73 for m = 10, a typical value for a structural ceramic. The predicted
strength ratio between a 4PB specimen (inner/outer span = 0.5) and a same-size
tensile specimen is:
(9)
ffT
which gives a strength ratio of 1.45 for same value of m = 10. Clearly, the loading
geometry must be considered when comparing the strength of ceramic components.
Actually, it was Leonard0 da Vinci (1452-1519) who first noticed that the
strength of steel wire increased for decreasing wire length. Indeed, assuming uniform
stresses in test pieces having different volume but the same failure probability,
Weibull statistics predicts a strength ratio:
which yields the value of 2 for V2 = 1000V1and m = 10, a typical situation for some
small test samples and the corresponding large engineering component.
The elastic modulus, the fracture toughness, the density, and the microstructure,
by contrast, are not usually sensitive to the size of the engineering component, and it
is principally the strength which is size-sensitive, and hence not a true material
parameter.
This has been demonstrated for a commercial glass ceramic [38], and for alumina
[39]. It has been suggested that the microstructure and the defect populations them-
selves may vary throughout the volume of a component, causing variations in
strength unpredictable by Weibull analysis [40,41]. An attempt to verify the minimal
specimen size required for strength determination was carried out lately [42].
3.2.3 Fracture Toughness

Griffith [43,44], who was the first to relate the strength of brittle materials to
material properties and crack length, and initiated the discipline of fracture
Figure 2. The Griffith cracked body A, with free surfaces S, and crack length 2a.
mechanics. Griffith calculated the balance between mechanical energy and internal
energy in a solid body containing a crack:
u = -FL UE us, (11)
where FLand UE are the external work and elastic strain energy stored in the body
under stress, respectively and Us is the surface free energy, Fig. 2. For linear elastic
bodies, FL = 2uE. UE was calculated from Inglis' [45] expressions for the stress field
in a body containing an elliptical hole. An instability occurs when the change in the
mechanical energy exceeds the surface energy required for a unit increase in crack
area:
dU
-= 0.
dA
This critical point marks the onset of crack propagation. Griffith calculated the
critical stress far from the crack location for a plate of unit width, and showed
that for a homogeneous, isotropic material, a relationship exists between the critical
stress, uc and half the crack length, a:
J.
I
Figure 3. The three basic modes of fracture, distinguished by the displacement driving force: I, the
opening mode; 11, the shear mode; and 111, the out of plane shear mode.
where ”is is the surface energy, E* = E for loading under plane stress conditions, and
E* = E/(1 - v2) for loading under plane strain, E and v are the Young modulus
and Poisson ratio respectively, assuming homogeneous, isotropic materials.
Eq. (13) does not involve the local stress field in the vicinity of the crack tip, and
that the only energy characteristic of the material is the surface free energy.
Three possible deformation ‘modes’can be defined, Fig. 3. Mode I, is a tensile mode
characterized by an identical displacement normal to the crack plane for any two
adjacent points on the free surface of the crack. Mode I1 is an in-plane shear mode,
in which the two adjacent points have the same (but with opposite sign) displacement
in the plane of the crack and parallel to the direction of crack propagation. Mode I11
corresponds to an out-of-plane shear mode; with shear displacements in the plane of
the crack but perpendicular to the direction of crack propagation. Crack propagation
seldom occurs under a pure tensile stress, and shear stresses are commonly present in
the early stages of crack propagation, so that most crack propagation takes place, to
some extent, under mixed-mode loading conditions.
Irwin [461and Orowan [47] generalized the Griffith energy balance, using the term
‘strain energy release rate’ (SERR), 9,to characterize a general relationship between
crack length, sample geometry and the mode of loading. In mode I deformation,
assuming unit thickness of the cracked body, is defined by:
or
P P
P,6
(4 (b) 6 = constant (c) P = constant
Figure 4. Strain energy release during constant displacement and constant load.
where U is the strain energy stored in the cracked body, 7r the complementary
energy, a the projected crack length, and da is crack increment, as shown in
Fig. 4a. For an imposed displacement (displacement control) the strain energy is
reduced for increasing crack length, Fig. 4b, and the crack advances stably, while
for a fixed applied load (load control), Fig. 4c, the strain energy increases for
increasing crack length, and the crack advances unstably, if brittle materials is
considered.
Increasing the external applied driving force will increase the strain energy release
rate until the strain energy release rate reaches a critical value gIC, which defines a
failure criterion for a brittle, flaw-containing component. The material property
qIc,is variously referred to as the toughness, the fracture energy, or the work of
fracture. gICis determined by the rate at which the stored strain energy is relaxed
when a crack advances. The critical ‘strain energy release rate’ per unit area of crack
advance has units of N m-’ or J mP2.The lower bound for the work of fracture must
be twice the surface energy, T ~ since
, two new surfaces are created by the advancing
crack. The actual value of the fracture energy will always be much higher than that
predicted from the equilibrium surface energy, since fracture is irreversible (and
approximately adiabatic) and additional energy dissipating mechanisms are
active: localized friction between microfacets, mechanical interlocking of grains
which bridge the crack, and some possible plasticity at the crack root.
An alternative stress analysis approach, is based on the local stress field near a
crack tip, Fig. 5 . The solution of the boundary value problem for a semi-infinite,
linear-elastic cracked body was found by Williams [48], and yields the Williams’
expansion of the stress field in a cracked body. The first term determines the local
stress field in the vicinity of the crack tip:
where r is the distance from the crack tip, 6’ is the angle (Fig. 5), and M denotes the
mode of deformation. The parameter Kis the stress intensity factor (SIF), which is a
function of geometry and loading only, and not a material property. The nature of
the near-stress field is that of square root singularity in distance from the crack tip
[9,10]. K is the amplitude of the stress field singularity near the crack tip.
As the external load increases, the stress intensity approaches the fracture tough-
ness, KIc , and crack propagation is initiated. This alternative failure criterion
/”
crack tip
X
Figure 5. Stresses at the vicinity of a crack tip.
becomes, for mode I: KI = KIc. The fracture toughness of a material K I c charac-

terizes the maximum stress field developed in the vicinity of a crack, and has the
dimensions of MPa m1/2.
Fracture toughness is a material property and measures the resistance to crack
propagation in the absence of generalized plastic yielding [49]. For an ideally brittle
material, the fracture toughness is independent of crack length, and crack initiation
under static loading is followed by crack propagation. In many hard materials the
fracture toughness increases with crack length, corresponding to an increased
resistance to crack propagation as the crack advances. In such cases crack initiation
is followed by a period of stable crack growth, with increasing stress intensity factor.
This is ultimately replaced by unstable crack propagation as the toughness
approaches the steady-state value of KIc characteristic of long cracks. This behavior
is summarized by the crack propagation resistance or R-curve (resistance curve)
[9,12]. The steady-state value of the toughness is significantly influenced by test
piece geometry: the initial ratio of crack length to specimen width and the length
of the uncracked ligament in the cross section.
Most static failures in brittle materials approximate to Mode I loading conditions,
while cracks that initiate in other in-plane loading modes, tend to propagate in
Mode I [49]. The Mode I crack is the most efficient relaxation path for the strain
energy stored in an elastic body. Plastic deformation can be ignored, since plasticity
is both limited and highly localized in hard materials. The fracture toughness of
hard materials generally ranges between 2 and 10 Mpa m1l2 [ 1,2,9], although
higher (and often invalid) values are sometimes reported, dependent on the
method of testing. This is an order of magnitude smaller than the toughness of
most engineering alloys [l, 2,101. It is the limited fracture toughness of hard
materials which limits their structural applications.
The fracture toughness determines the critical amplitude of the stress singularity
at the crack tip. For an ideal brittle continuum, provided the crack length, a, is
known, there is an exact relationship between the fracture toughness and the
remote stress:
where p is a function of the geometry of the component or test piece. Geometrical

functions for P have been published in the literature, most calculated by finite
element analysis [50-541.
Irwin [46] derived the relationship between the strain energy release rate and the
stress intensity factor from the stress field at the crack tip. For a pure mode I
deformation:
An analogous relationship holds for the fracture toughness and the fracture
energy:
Excellent texts on fracture mechanics are available [9-111.Related topics are failure
criteria for the onset of cracking in laminated systems [55],dynamic fracture
mechanics [56],and nonlinear fracture mechanics [9,1I].
3.2.4 Hardness
Hardness is commonly defined as the resistance of a material to static penetration by
a harder material [3].The qualifier, ‘static’, excludes dynamic effects associated with
erosion or ballistic penetration. Hardness can be quantified by determining the
depth of penetration of a hard indentor, usually diamond, loaded perpendicular
to a planar surface of the material under test [57].The penetration depth is related
to the contact area over which the load is transferred between the indentor and the
sample. The ‘hardness’ is equal to the load divided by the area of plastic contact.
There is some ambiguity in ignoring recoverable, elastic penetration, but, apart
from elastomeric materials, the effects of elastic recovery during unloading are
usually small. The fixed load, P , applied to the indentor and the known indentor
geometry are used to calculate the hardness from the contact area, A , between the
indentor and the sample:
P
H --, (20)
- A
where H , is an average applied compressive stress, generally measured in GPa. In
polycrystalline solids, plastic deformation beneath the indentor is associated with
dislocation multiplication and glide [58].In metallic solids the critical flow stress
or yield stress of the material, oY,governs the hardness [59].There is therefore a
direct relationship between the hardness and the plastic flow stress, often expressed
as:
H, = Cgy, (21)
where C M 3 for a perfectly elastic, plastic solid [60]. The large proportionality
factor reflects the elastic constraint exerted on the plastic zone beneath the indentor
by the surrounding elastic material. No such constraint exists in the unidirectional,
tensile or compressive tests used to determine the uniaxial flow stress. When the
constitutive relationship describing the mechanical response is more complicated,
or for a brittle material (when the plastic response depends on the hydrostatic
component of the constraint), the proportionality constant may be appreciably
higher [61]. Amorphous, glassy materials posses a clearly defined hardness, related
to the Young modulus [62]. The same elastic constraint inhibits cracking in the
vicinity of the indentor, ensuring that some plastic flow occurs beneath the area
of elastic contact and precedes crack nucleation and propagation, even in very
brittle materials.
From the engineering perspective, a hardness test is an ideal method for monitor-
ing the mechanical properties of hard materials, since minimal sample preparation is
required and the test can be performed on actual components using simple
apparatus operated at low loads. The hardness test may also be considered ‘non-
destructive’, since components can often be put into service after testing. The
mechanical performance of hard materials is often ranked by their indentation
hardness, which in itself accounts for both the popularity and the technological
success of this simple, cost effective test.
The measured hardness of any material depends on parameters associated with
the test method and indentor geometry (Brinell, Vickers, Knoop, etc.), so that hard-
ness is not an intrinsic bulk property, comparable to elastic modulus, yield strength
or fracture toughness. In general, the measured hardness varies with the applied
load and the indentor shape and dimensions, but also with the microstructure
and prior history of the material, as well as with the loading rate, the environment
and the test temperature [3,63-671. In crystalline materials the resistance to disloca-
tion glide determines the hardness, and glide is activated by the shear component of
the compressive stress generated beneath the indentor. The maximum strain beneath
an indentor (elastic and plastic) may be some two orders of magnitude greater that
the tensile strain to failure of a brittle solid [68]. Such conditions are seldom encoun-
tered by hard materials operating in engineering systems. The strain rate in an
indentation test is also high when compared to a static tensile test, since the velocity
of the indentor is much greater than the usual crosshead speed in tensile testing. A
high strain rate increases the flow stress ((~y),the effect increasing with indentation
load and rate. By contrast, at high homologous temperatures thermally activated
creep is possible, reducing hardness with increasing temperature and decreasing
loading rate [69].
Indentation testing dominates the determination of hardness, but some alter-
native tests are available: The scratch test ranks materials in order of their ability
to scratch materials lower on the scale [70], and assigns a ‘Mho’ hardness (1822)
to each material, with diamond having the highest value of 10 on the Mho scale.
Erosion testing is of considerable importance, especially for components subjected

to particulate bombardment. Dust and sand particles, but also rain, are all sources
of damage in supersonic aircraft and jet engines. Erosion is also common in pump-
ing systems required to transfer slurries and other solid-phase dispersions. The
rebound test (the ‘bounce’ of a steel ball dropped onto the surface) is of importance
in testing the resilience of elastomeric components, while damping losses, assessed
by measuring the decay of high frequency vibrations, are also used, primarily in
acoustic and vibrational damping systems.
3.3 Mechanical Testing of Hard Materials

It is possible to determine the tensile (Young) modulus by direct measurement of
stress and strain in unidirectional tensile or compressive loading. The average
stresses are monitored by a load cell, while the average strains are read from a
strain-gauge or extensometer. The elastic modulus is given by Hooke’s Law,
E = c/E.This method requires both a suitable test specimen and a uniaxial testing
machine that will guarantee unidirectional loading: Accurately machined ‘dog-
bone’ configurations are mandatory for tensile specimens, precision cylinders for
compressive tests. A measurement of the velocity of sound in the material, using
an ultrasonic transducer, is less prone to error. If the density p and longitudinal
sound velocity vL are known, then the elastic modulus is given by:
It is also possible to determine the lowest resonant frequency of a rectangular beam

statically loaded by a mass M in three-point bending. The tensile modulus is then
given by:
16 rM13f2
E=- (23)
3 d4 ’
where I is the span of the beam, d the height and f the resonant frequency.
3.3.2 Fracture Strength

Since the strength of brittle materials is stochastically distributed, a large number of
tests is required to characterize the mechanical properties accurately. The two most
important properties are the median strength and Weibull Modulus [34,37], and the
minimum number of (valid) test results needed to meet most engineering require-
ments is 20.
When the results of N tests of the fracture stress c are ranked in ascending order,
then the best statistical estimate of the failure probability, F, for each result is given
82 3 Mechanical Properties and their Relution to Microstructure
Inert strength, u,(MPa)

Figure 6. Typical ln(l/(l - F)) against In0 plots for three and four point bending bars. Repro-
duced with permission of . . . . . . . . . . (Ref. 9).
by (i + 0.3)/(N - 0.4), i being the rank order of the ith test result. The Weibull
modulus m is estimated from the slope of the graph of In(ln[l/(l - F ) ] ) vs. lnc,
as shown for various geometries and processing routes, Fig. 6 . A detailed discussion
of Weibull strength analysis is given in [37]. The determination of m should always
be made by graphical plotting of the test results, since alternative failure mechan-
isms can give rise to different median strengths and Weibull moduli. These may
be associated with the probability of failure initiation at a surface as opposed to a
bulk defect. Surface flaws commonly have a wider size distribution, giving a
lower Weibull modulus, while bulk flaws, with a narrower distribution, correspond
to a higher value of m [71-741. In consequence, the slope of the Weibull plot changes,
increasing for the higher strength values, dominated by the size and distribution of
the bulk defects. A full understanding of the In In[I / ( 1 - F ) ] vs. In c curve requires
fractographic analysis of the failed samples and correlation of the observed failure
origins (surface cracks, bulk porosity, hard aggregates, inclusions) with the
measured strength.
Various internationally recognized standards exist for the determination of flex-
ure strength (3PB or 4PB):ASTM C1161 and MIL STD 1942 (USA), DIN 51-1 10-1
(Germany), B41-104 (France), JIS R 1601 (Japan), GB 8489 (China). The specimens
are prismatic rectangular bars. Commonly specified outer spans are 20, 40, or
80 mm, with corresponding 4PB inner spans of 10, 20, and 40 mm, respectively.
The specified cross section (height x width, all dimensions in mm) varies from
1.5 x 2 (American) and 2 x 4 (French) for small-span specimens, to 3 x 4 for the
medium span, and 6 x 8 for the large span.
Other test samples and loading geometries are sometimes required for specific
components or applications. The biaxial strength of thin plate is commonly
evaluated using a ring-on-ring test or an alternative mounting suitable for biaxial

flexure (a hardened steel ball with an articulated tripod support, for example).
3.3.3 Fracture Toughness

A bewildering number of specimen geometries and experimental methods has been
proposed (and adopted) to determine the fracture toughness of brittle materials, and
values for fracture toughness of hard materials are routinely determined, published
and compared. However, at the time of writing, no universally agreed standard for
fracture toughness exists most internationally accepted methods specify a pre-
cracked specimen loaded to fracture under three point bending (for example
ASTM E-399) [75]. The fracture toughness is then calculated using an appropriate
function from the maximum recorded load and the specimen geometry.
The generation of a reproducible, sharp, and accurately positioned precrack in a
hard material is a technical problem which partially accounts for the variability of
reported fracture toughness results. While an individual laboratory may obtain a
reproducible result, different laboratories, employing nominally the same test
method for the same material, may find average values for fracture toughness
that lie outside the scatterband determined by their colleagues elsewhere [7&78].
Any fracture toughness test must approximate the square root singularity of the
stress field calculated for an atomically sharp crack. Deviations from the square
root singularity will reduce the local stress intensity and increase the apparent
value of fracture toughness. Unfortunately, since engineering applications for
hard materials usually specify a minimum toughness, it is not uncommon to wel-
come misleadingly high values of toughness. Any test procedure (or test laboratory!)
consistently giving unusually high values for fracture toughness should be viewed
with suspicion, and it is the lowest measured value of the fracture toughness
which should be used to estimate maximum allowed loads and specify the require-
ments for defect detection in the nondestructive evaluation of a component.
Unstable crack propagation often follows immediately on crack initiation in
brittle materials, and it is difficult to generate a straight, sharp precrack. It is
therefore common practice to cut a thin (typically about 100pm) notch, using a
diamond cutting disk [79], to act as a crack initiator. The presence of a square
root stress singularity at the notch tip is unlikely, although microcracking at the
root of the notch may approximate the square root singularity [SO], so the fracture
toughness determined from notched specimens is often an overestimate, while the
uncertainty in initial crack length, associated with the size of the microcracked
damage zone, introduces an additional error.
A precrack can be generated from microindentation cracks placed near the center
of the test beam prior to loading [81]. The indentation crack tip front is elliptical,
and the residual stress field associated with localized plastic flow beneath the inden-
tor has a significant effect on the square root stress singularity. A more successful
method for preparing a precracked beam is by applying a compressive load to a
notched or indented specimen held in a stiff fixture. Controlled crack growth
from the original defect gives the required sharp and straight crack front, and the
fracture toughness results obtained from such precracks have proved to be highly
reproducible [82-841.
Controlled precracking of a thin, notched specimen can be achieved by gluing the
sample to the side face of a ductile metal beam loaded in 3-point bending. Crack
arrest in the thin, brittle specimen occurs at the point where the stress in the thick
supporting beam becomes compressive [85]. Precracking of ceramic specimens by
cyclic fatigue of notched specimens at a slowly decreasing stress intensity factor
has also been reported [86]. Most applications of these test methods for
precracked-beams loaded by 3PB evaluate the plane-strain fracture toughness
using the methodology summarized in the ASTM standard E-399 [75].
The precise measurement of crack length is also difficult. Partial crack closure
after precracking, upon removal of the load, may preclude accurate crack length
measurement. The use of die penetrant as a visual crack marker is helpful but
inaccurate. Optical inspection at low magnifications is usually adequate, providing
the specimen finish is good. The fracture toughness of materials generally quoted in
the literature is the static or quasi-static fracture toughness, and is a poor guide to
dynamic behavior, slow crack growth, or mechanical fatigue strength.
Negotiations for an international fracture toughness standard are well advanced
within the framework of ASTM and VAMAS (Versailles Agreement on Materials
and Standards) [87]. The test geometry which is expected to become an ASTM
standard is the chevron-notched three-point bend beam. The chevron notch is cut
with a diamond saw, and the crack grows at low loads from the tip of the notch.
Stable crack growth continues until the crack front reaches the full cross-section
of the beam, when growth becomes unstable at the point of maximum load, the
critical load for fracture toughness evaluation. The fracture toughness determined
from a precracked beam may give values which differ significantly from those
determined from a chevron notched specimen. This difference is usually material-
dependent and it is not possible to predict which test will give the lower (more
dependable) value.
Alternative geometries and test methods to determine the fracture toughness of
ceramics and hard materials have been developed over the last decades: Herzian
indentation [88], the double torsion test [89], and compact tension tests [90-921.
3.3.4 Hardness
Microhardness tests in metals are performed under very low load [3], 0.01-2N,
which nevertheless activates dislocation glide. The higher values of microhardness
characteristic of hard materials require higher loads, typically 0.05-0.5 N, which
do not usually initiate microcracking (indentation cracking). At still higher loads,
2-100 N, microcrack initiation and propagation is expected in brittle materials,
resulting in higher values for the measured hardness, since microcracking increases
the compliance, and hence the elastic component of penetration, which is relaxed
(apart from frictional effects) when the load is removed. In addition to plastic
deformation, other viscous flow mechanisms may operate under high loads, such
as densification in both glasses and ceramics containing residual porosity [93]. It
Figure 7. A common load-hardness relationship. Reproduced with permission of . . . . . . . . . . (3).
follows that hardness tests should be performed over a range of load to ensure that a
load-independent value of this property can be defined [66],Fig. 7. In hard materials
a high-angle, pyramidal indentor, generates high residual tensile stresses on load
removal which may exceed the tensile strength, resulting in microcrack initiation
and propagation.
Valid hardness measurements require that the indentation be either very much
less than the grain size, in order to characterize the microhardness of individual
grains, or else appreciably greater than the grain size, to ensure that it is the hardness
of the polycrystal that is determined. Typical indent sizes, using standard micro-
hardness testing systems, are in the range 3-50pm, and up to 200pm in the low
load regime. A good optical microscope has a resolution of the order of 0.5pm
and should be capable of achieving 10% accuracy providing several measurements
are made to monitor the reproducibility.
The methods which have been developed to determine hardness vary in indentor
geometry and definition of the contact area. In Brinell’s method, a hard sphere of
diameter D is indented into the softer sample, and the Brine11 hardness is calculated
from the diameter of the circular indent which is left in the surface after removing
the load:
where d is the diameter of the impression. In the Rockwell method the depth of
penetration of the indentor, usually but not always a diamond cone, is used to
calculate the Rockwell hardness number on one of several scales (Rockwell A,
B, C , or D), depending on the applied load and the geometry of the indentor.
The Meyer hardness number is derived from the projected area of the residual
impression:
P
HM =4-.
?rd2
In the Knoop method an elongated diamond indentor with a specified ratio of the
two diagonals is used. Measurements taken from the long diagonal of the indent can
be used to assess hardness anisotropy by rotating the indentor about the loading
axis between measurements. The Berkovitz microhardness indentor employs a trian-
gular pyramid which avoids variations in hardness associated with anisotropy in the
plane of the surface and ensures that the indentor tip is truly a point (inaccuracy in
cutting a square pyramid results in a chisel edge at the apex). In the Vickers method
a square diamond pyramid with a standard apex angle of 136" is used and the
contact area is determined from the mean length of the two diagonals of the
indent, 2a, which is then translated into a Vickers hardness number:
P
Hv = 0.464-. (26)
a2
It is the Vickers test which is most commonly used to specify the hardness of hard
materials, with results often quoted in kgmm-2, but more correctly in SI units, GPa.
The repeated measurement of the size of indents, and the interpretation of indent
geometry for the purposes of calculation, may be tedious, and operator bias is
almost unavoidable. The edge of the impression is not always well defined, and mis-
leading edge effects may be associated with anisotropic plasticity or plastic recovery.
Faceted and elongated grains, or other microstructural features, together with the
limitations of contrast and resolution in the optical microscope, complicate the
interpretation, while the shape of the indent may differ in different materials: so-
called 'pin-cushion' or 'barreled' indents, associated with different constitutive
relations and frictional shear on the faces of the indentor in contact with the plastic
zone [3]. Mismeasurement of indent size is a major source of scatter in the experi-
mental data and the relative errors in the results of different operators.
3.3.5 Indentation Toughness

Palmqvist [94,95] has used the Vickers diamond pyramid indentor to initiate
and propagate cracks in hard materials, and Evans, Lawn and their coworkers
[96-1001 have related indentation cracking to fracture toughness by assuming a
simplified stress field. Since then, considerable effort has been made to characterize
fracture toughness using Vickers indentation cracking, Fig. 8. The primary advan-
tage seen for such a test is the limited volume of material needed and the inexpensive
test apparatus. Furthermore, such a test can be performed on an actual component,
with no need to machine a special test piece. There are several problems: the small
volume element of material under load, the variety of possible cracking mechanisms,
analysis of the stress field needed to define a calibration function, variations in the
shape of the crack front, and accurate measurement of crack length. These problems
reduce the significance of indentation toughness results, although this remains a
Figure 8. Typical plastic deformation and cracking after indentation toughness test. Reproduced
with permission of . . . . . . . . . . (3).
useful test for monitoring damage tolerance and the resistance to short crack propa-
gation. The measured values should not be interpreted as ‘fracture toughness’.
Four microcrack morphologies may develop in a brittle solid after formation of
the plastic zone beneath the indent: median cracks, Fig. 9a, radial cracks (a half-
penny shaped elliptical crack Fig. 9b, much larger than the median crack), lateral
cracks (in a plane normal to the median and radial cracks), Fig. 9c, and Palmqvist
i I_2c‘ *I
I
I I
! I
a b
\- Median
mqvlst radial
vent
I d
C
Figure 9. Cracks formed during an indentation toughness test: (a) median cracks; (b) radial cracks;
(c) lateral cracks; and (d) Palmqvist cracks. Reproduced with permission of . . . . . . . . . . (3).
radial cracks (originating near the edge of plastic zone beneath the indent), Fig. 9d.
In addition, Herzian cone cracks may develop, concentric with but outside the
plastic indent left by a hard sphere or cylindrical punch.
The median crack is a single, penny-shaped crack nucleated beneath the apex of
the plastic zone created by the indentor. The diameter of the median crack is com-
parable with the indent size, and a median crack is not visible in a polycrystalline,
opaque material. The driving force for nucleation of the median crack is the elastic
tensile stress developed normal to the indentation direction at the elastic-plastic
boundary when the external load is relaxed. Nucleation of a median crack depends
on the presence of a suitable flaw. Once nucleated, the median crack will propagate
spontaneously to a stable flaw size. The critical flaw size for growth is:
where 0 is a constant which relates the tensile strength to the hardness.

After nucleation, the median crack propagates away from the elastic-plastic zone
boundary. Stable propagation occurs on increasing the external load, and the ratio
of the diameter of the median crack, D,to the indent size is given by:
D
-m*.
a
When the diameter of the median crack exceeds the indent diameter, it can ‘pop’
out to the surface of the specimen to become a half-penny shaped, radial crack,
which is visible on the surface and whose length can be measured. A polished surface
may be necessary to identify accurately the position of the crack tip, while annealing
may be required to eliminate compressive surface stresses due to polishing. The
exact shape of the radial crack is unclear, but in a transparent specimen the ‘pop-
in’ of such cracks may observed on unloading the indentor. Multiple radial
cracks may occur, especially for indentors with a small pyramid angle. High fric-
tional forces may prevent sliding at the indentor contact surface and cause large
shear deformation adjacent to the indentor facets.
Upon unloading, the compressive stress normal to the surface reverses sign to
become tensile, due to the previous irreversible plastic deformation beneath the
indentor. If a suitable flaw is present, a lateral vent crack may be generated. This
shallow crack can ‘pop-in’ parallel to the surface and terminate subsequently at
the surface to remove a flake of material. The lateral crack is larger the plastic
zone, and may be larger that the radial cracks.
Finally, Palmqvist cracks are short, shallow cracks, initiated at the end of the
indent diagonals at relatively low loads, Fig. 9d. The plane of a Palmqvist crack
coincides with the plane of the radial cracks, and there is some evidence that the
Palmqvist cracks may, in some ceramics, initiate before the radial cracks, at the
same time as the median crack.
Given the variety of cracking mechanisms, both the stress field under load and the
residual stress field are clearly very complex [loll. Difficulties of stress analysis have
led to some inconsistency in the formulae derived to calculate indentation tough-
ness. Rather than review the large number of published equations [102-1071, a
3.3 Mechanical Testing of’ Hard Materials 89
basic parametric formulation may be given in the form of a nondimensional fracture

toughness [66]:
where Hv is the Vickers hardness, a, c, and R,, are the half-length of the diagonal
of the indent, the radius of the radial crack, and the radius of the plastic zone,
respectively, q5 the constraint factor, v and /-L are Poisson’s ratio and the coefficient
of friction between the indentor and the sample. Alternative’formulations exist!
The indent size, the surface crack morphology, and the extent of subsurface
damage are all influenced by the crystal structure, chemical composition, and micro-
structural morphology of the material.
3.3.6 Erosion, Wear and Scratch Tests

3.3.6.1 Erosion
Solid particle erosion is a major cause of damage in pipes, gas turbines, and com-
pressor components [108]. Erosion may be defined as the resistance of a material
to surface damage due to impinging particles of high kinetic energy [109]. Erosion
is a dynamic phenomenon, and the damage may be mechanical (loss of bend
strength due to surface microcracking) or physical (loss of transparency due to
roughening of a surface). Rain erosion, in which the impinging ‘particles’ are rain
drops arriving at sonic or supersonic velocities, may be as damaging as solid particle
erosion, and the loss of visibility due to surface roughening of a pilot’s cockpit
canopy can be as dangerous as any loss of strength. Erosion damage is commonly
measured as the change in a given property as a function of the number of impacts
or the total momentum transferred. The property of interest may be the weight loss
of an eroded sample, its residual strength, or some other measurable property
(residual transmittance, for example) [1lo].
In ductile materials the impinging particles gouge out fragments by micromachin-
ing of the surface, and the maximum weight loss is observed at an impingement
angle of about 20” [lll]. Erosion of hard materials is primarily associated with
the generation of lateral cracks beneath the point of impact, and the rate of
damage is a maximum at normal impact. The erosion rate depends on the fracture
toughness or damage tolerance of the target material and the hardness of the
impinging particles [112]. For an ideally brittle material, the volume of material
lost in a single impact, Vi, can be estimated from the radius of the lateral crack,
ci, and its depth, ai [9]:
where w is a ‘wear coefficient’. If additional energy absorbing mechanisms exist, and

the material exhibits R-curve behavior, then the value of the fracture toughness
should be that appropriate to short cracks of length ci. The above relation suggests
that the influence of fracture toughness should exceed the effect of hardness in
improving erosion resistance.
3.3.6.2 Wear Assessment

A good example of extreme wear is the attack that occurs at the top face of a cutting
tool when a fully-plastic, machined chip is sheared adiabatically across the tool face.
Adhesive and abrasive wear are major mechanisms [113], as is chemical wear,
especially at elevated temperatures. Adhesive wear occurs when two surfaces in
contact bond during shear, and is associated with large compressive stresses gener-
ated at points of contact formed during sliding under pressure [114]. If the local
compressive stresses are sufficient, subsurface microcracks propagate and fragment
near-surface material to form wear particles. This occurs when the toughness of the
bulk material is lower than that of the interface under shear. For a sufficiently low
interfacial toughness, the debond 'crack' will propagate at the interface without
wear. The distribution of defect-like, near-surface voids, impurities, or surface
microcracks determines the extent of the wear zone and the size of the wear particles.
Abrasive wear takes place when two hard materials are brought into contact
without bonding, when the primary damage is associated with mechanical interlock-
ing (shear of asperities) and the generation of Herzian cone-cracks surrounding the
contact points [115, 1161.
3.3.6.3 Scratch Testing

In a scratch test, a diamond or WC, conical or pyramidal indentor is sheared across
a specimen under a controlled normal load while monitoring the tangential force
[117]. The adhesive bond and the tangential plowing force determine the total
work performed by 'frictional' forces. The coefficient of friction is determined
from the ratio of the tangential to the normal load.
The results of a scratch test depend on the indentor material and geometry, the
normal load, the indentor velocity, the surface roughness, the lubricant (if any),
and the homogeneity and isotropy of the sample [118]. Crack damage in a scratch
test approximates that of a Herzian crack, and results from the tensile stress
generated near the trailing edge of the contact. Cracks are formed when the external
load exceeds a critical value. The trace of the scratch contains a set of cracks whose
density is a function of the load, the surface conditions and the indentor velocity.
Tensile stresses that generate cracks are enhanced by frictional forces between the
indentor and the sample in the contact area [119]. The presence of a lubricant
increases the threshold load for cracking [120]. A chemically active liquid, such as
water, may increase the friction load and decrease the threshold for cracking. The
fracture surface energy of glass may be reduced by the presence of scratches,
which decrease the threshold load for cracking. The shape of the indentor is impor-
tant. No damage was observed in soda-lime glass scratched by spherical indentor at
loads up to 50 g. Damage increased with load for a 90" square pyramidal indentor,
but only initiated at 15g load when the pyramidal indentor angle was 136". The
width and the depth of the scratch decreased with increasing indentor velocity.
3.4 Microstructural Parameters and Mechanical Properties 91
Scratch tests have been used to evaluate the adhesion of thin films to a ceramic
substrate.
3.4 Microstructural Parameters and Mechanical

Properties
Since the failure of hard, brittle materials is dominated by fracture originating from
sources of stress concentration, their performance in service is dominated by the
presence of processing defects as much as by the microstructure [121]. It is to
distinguish between the two, even though ambiguity remains when the scale of a
‘processing defect’ approaches that of the microstructure. Processing defects resol-
vable by available methods of nondestructive evaluation (NDE) and quality control
present no difficulties. These include porosity, microcracking, and those‘inclusions
whose density and elastic properties enable detection. Ultrasonic testing and X-ray
radiography are capable of a resolution well below 100 pm, providing the compo-
nent cross-section is not too large. Both techniques can be automated and have
tomographic capability, that is the defects can be located in three dimensions
with reasonable accuracy, not just in projection [122-1241.
Below the resolution limit of available NDE procedures, but for dimensions
which exceed the scale of the microstructure, it is still useful to talk of ‘processing
defects’, even though they may only be detectable by fractographic analysis,
which traces the origin of failure to an abnormally large grain, a hard aggregate,
a brittle inclusion, or other defect. The ‘undetectable’ processing defects are a func-
tion of processing history, and constitute the major obstacle to the improvement of
performance through improvements in the quality of raw materials for hard
materials. Reducing powder particle size below 0.1 pm is of little help if the resultant
finer grain size is accompanied by particle aggregation and residual porosity.
Similarly, the price of improved powder purity is only worth paying if contamina-
tion by milling media wear can be avoided during processing.
It is helpful to regard all ‘microstructural defects’ as features of the microstructure,
and avoid confusing them with processing defects, even though their presence is clearly
a result of processing history. Subgranular residual porosity and pockets of glassy
phase at grain boundary junctions, intergranular and transgranular microcracks,
and small second phase particles associated with impurities or precipitation, are all
microstructural defects [125, 1261. The effort to relate mechanical properties to the
microstructure of engineering materials has a long and respectable history, but with
many misconceptions. Before the analysis of crystal structure by X-ray diffraction
became generally accepted, it was thought that the mechanical fatigue of steel was
due to ‘crystallization’,while ductility was attributed to a ‘noncrystalline’microstruc-
ture, and t h s despite the clear metallographic evidence for a granular structure.
It is useful to distinguish three aspects of microstructure in hard materials, all of
which are related to mechanical properties. The first is the physical nature of the
features present, including the composition and crystal structure of the solid
phases, and the presence of porosity, microcracking, and impurity phases [121,127].
The second aspect is the microstructural scale of the features, including their size,
size distribution, volume fraction, and distribution in space (microstructural homo-
geneity). The third factor is the orientation of microstructural features, both their
morphological anisotropy (alignment of platelets, whiskers, and fibers), and their
crystallographic anisotropy (preferred orientation of crystalline phases with respect
to the component geometry, commonly termed ‘texture’).
As in metals, the scale of the microstructure is related to the hardness (and,
through the hardness, to the plastic flow stress), with the Petch law applying well
into the submicron range of grain size. Increasing the hardness for a given composi-
tion is readily achieved by reducing grain size, typically by reducing powder particle
size for the raw material together with the sintering temperature used to densify
(sinter) the component. Unfortunately, processing to full density becomes more
difficult as the powder particle size decreases, while any residual porosity reduces
the hardness. Grain growth during sintering is limited by appropriate dopant
(trace) additions. Fine zirconia particles in alumina pin grain boundaries [ 1281,
while MgO additions inhibit grain boundary migration without affecting densifica-
tion [1291. Commercial aluminas with submicron grain sizes and near-theoretical
densities are available with a hardness of the order of 20 GPa (compare the tensile
modulus of 380 GPa).
Fracture toughness has been improved by providing microstructural toughening
mechanisms, discussed more fully below. Direct melt oxidation (DMO) leads to
alumina products which contain a network of some 15vol-% residual metal
[130,131]. The metal fibers intersect a propagating crack front and dissipate
energy by deforming plastically, reducing the stress intensity factor at the crack
tip. Platelets or whiskers in the microstructure, present as additions to the powder
mix or, preferably, grown in situ as a result of trace additions, also bridge the propa-
gating crack front, supporting some of the load and generating frictional forces as
they detach from the matrix. Tetragonal zirconia grains or particles transform to the
lower density monoclinic phase in the stress field at a crack tip, generating crack
closure forces which lead to ‘transformation toughening’.
The high temperature strength and creep properties of hard materials are sensitive
to the presence of glassy phases at grain boundaries [132], which promote viscous
sliding at the boundaries, lead to enhanced stress concentration at the boundary
junctions and nucleate cavities. The cavities grow under the hydrostatic tensile
component of applied load, coalescing to form ductile microcracks which initiate
the third stage of creep failure (see below), leading to stress rupture. In some
grades of silicon nitride a glassy oxynitride phase responsible for the excellent
sinterability of the material, can be crystallized after sintering, greatly improving
the creep resistance. Since creep failure is commonly initiated by cavitation, which
results in a volume increase, creep in hard materials depends on the sign of the
uniaxial load. Creep rates in compression are generally lower than those in tension
[ 1331. The viscous flow of the glassy phase in the grain boundary channels, a perco-
lation process, commonly determines the overall strain history.
The availability of nanopowders has led to the development of submicrometer,
fine-grained ceramics, which exhibit improved mechanical properties. Many
3.4 Microstructural Parameters and Mechanical Properties 93
glass-free oxide ceramics with a submicron grain size are superplastic under con-
ditions of moderate temperature (= 0.5 T,) and low strain rate to lop2s-’)
[134-1 361. The exact mechanism of superplastic deformation is controversial, but
it is clear that the grain boundaries play an important role, and that diffusion is a
dominant contributor to plastic deformation. Tensile strains to failure of 100-
1000% are routinely observed [ 134-1 391, and should enable complex shapes to be
formed from dense ceramic preforms. The strain rates required for ceramic super-
plasticity are too slow for most forming operations.
Through trial and error, it has been found that grain sizes of less than 1 pm are
required for superplasticity, and the finer the grain size, the larger the superplastic
strain to failure of the ceramic [140,141]. A second advantage offered by a fine
grain size is an increase in the strain rates observed during superplastic defor-
mation. The effect has been rationalized using the following phenomenological
relation:
where C is the strain rate, A a material constant, k is Boltzmann’s constant, b the

Burgers’ vector, d the grain size, 0 an applied stress, E the Young modulus, Do a
pre-exponential factor for diffusion, Q the activation energy for superplastic flow
(usually Qsuperplastic
M Qdiffusion),and T is the temperature. The constant n is the
stress exponent, while p is the grain size exponent. The value of p is typically
between 2 and 3, and a 10-fold reduction in grain size should lead to a factor
of 100 to 1000-fold increase in strain rate at constant applied stress. Conversely,
the same grain size reduction should lead to a decrease in flow stress or lower
deformation temperature, at the same strain rate. The latter phenomenon may
explain the room temperature plastic deformation in nanostructured Ti02,
reported by Karch, Birringer, and Gleiter [142], who were the first to report sig-
nificant room-temperature ductility (M100% plastic deformation) in nanocrystal-
line ncTi02. By contrast, a compression test on a TiOz single crystal resulted in
brittle fracture. Recent experiments [ 1431 have shown that nanocrystalline
oxides may be plastically deformed under relatively low tensile stresses and at tem-
peratures below half the melting point.
The Hall-Petch relation relates the flow stress to the grain size:
0 = 00 + k(d)-”2, (32)
where 0 is the flow stress, go is a lattice friction stress, k a (positive) material
constant, and d is the grain size. This equation implies that any decrease in grain
size should increase the flow strength. The relation between hardness and grain
size is expected to be given by a similar equation, although this may not be true
of nanocrystalline materials, for which abnormal Hall-Petch behavior has been
reported for both nanocrystalline metals [144,145] and ceramics [I431 (a decrease
in grain size decreasing the hardness and increasing the ductility). Grain-boundary
sliding, grain boundary diffusion, and the generation of lattice dislocations from
boundary junctions have all been suggested as alternative mechanisms of plastic
shear, replacing dislocation glide, in nanocrystalline materials [ 146-1491. Plastic

deformation in crystalline ceramics can occur by dislocation glide, climb or
diffusion, usually at grain boundaries. The high Peierl’s stress and limited
dislocation density in ceramics usually restrict plastic deformation mechanisms to
diffusion-assisted processes.
The ballistic response of hard materials is also microstructure-dependent, pri-
marily through the hardness, although the density and elastic modulus determine
the mechanical impedance, and hence the shock-wave behavior. Light-weight, stiff
ceramics are effective armor modules. Boron carbide provides the best ballistic
performance, although the much cheaper alumina dominates the market. Since
the time scale of ballistic impact is so short, the failure processes are adiabatic,
and the presence of a glassy phase may be almost as deleterious as residual porosity
for ballistic performance. Adiabatic heating above the glass transition temperature
allows the glassy phase to flow, leading to cavitation failure under impact, not
dissimilar to that observed in creep!
3.5 Failure Mechanisms

3.5.1 Creep Behavior
Hard materials and ceramics are used in load bearing applications at high tempera-
tures and for extended times. Under these conditions the primary failure mode is
often by creep and stress rupture. ‘Creep’ is time-dependent, irreversible plastic
deformation under constant load or stress, terminating in time and temperature
dependent failure, termed ‘stress rupture’. Ceramics, like metals, begin to creep at
homologous temperatures of 0.4-0.5 T,. The melting temperature T, of a struc-
tural ceramic is usually well above 2O0O0C, and creep initiates at temperatures
well beyond the operating temperatures of heat resistant superalloys. The resistance
of ceramics to environmental attack is substantially better than the superalloys, and
certainly far superior to the refractory metals (molybdenum and tungsten, for
example).
Two engineering parameters are commonly used to rank the behavior of materials
under creep conditions (constant temperature and applied uniaxial stress): the
minimum strain rate and the time to rupture. The first parameter is related to the
useful life of components susceptible to shape-change in service (in heat engines,
for example), while the second estimates the time-dependent failure probability.
The deformation and damage mechanisms in creep of ceramics and hard
materials are similar to those in metals [150,151]. Under normal loading conditions
(in the absence of severe elastic constraint) ceramics fracture at room temperature
before any significant plastic flow. Dislocation glide in ionically bonded ceramics
is complicated by the presence of both anions and cations, which create electrostatic
(Coulombic) barriers to shear. As in metals, three creep regimes have been identi-
fied. The initial high strain-rate, observed on applying the load, decreases rapidly
as mobile slip dislocations are exhausted and boundary shear leads to grain
interlocking at the triple junctions. This primary, transient creep terminates at the
minimum creep rate, which may then persist for most of the creep life, correspond-
ing to the second, steady state, creep regime.
When an extended period of steady-state creep is present, this rate is denoted i.
and can often be described by the relation:
i = A 6 exp(-QIRT), (33)
where A and n are material parameters, and Q is an ‘activation energy’ for creep
[152]. The constant creep rate in the steady-state regime is usually associated with
strong work-hardening, arising from the blocking of secondary slip dislocations
by dislocation barriers on the primary slip system. The stress exponents close to 5
are due to dislocation climb and fully ductile behavior, and a stress exponent
close to 3 is due to dislocation climb [151, 1531.
The mechanisms responsible for fracture in structural ceramics at elevated tem-
peratures have been reviewed [ 1541. Sensitivity to flaws or microstructural inhomo-
geneities which nucleate microcracks are among the failure mechanisms. The flaws
which control failure under creep conditions are different from those responsible for
fast fracture at room temperature. A common feature is the development of cracks
through gradual damage accumulation, depend on the microstructure. The role
of cracks in the deformation and fracture behavior of polycrystalline structural
ceramics have been reviewed [1551.
3.5.2 Mechanical Fatigue

Mechanical fatigue under cyclic loading is a common cause of failure in engineering
systems. Fatigue cracks usually initiate at sources of stress concentration: re-entrant
angles in component design, hard inclusions and aggregates, residual porosity, or
shrinkage cracks. However, the parameter controlling crack advance in metals
and alloys is the stress intensity factor at the fatigue crack tip after nucleation,
and observed values of the SIF for fatigue crack propagation are significantly
below the fracture toughness ( K I c )of the material. In these ductile materials dis-
location mobility at the crack tip during loading and unloading in the fatigue
stress cycle determines the rate of crack growth. The semiempirical Paris’ law [lo]
describes this rate of crack advance as a function of the number of load cycles,
expressed in terms of the SIF amplitude:
da
-= C ( A K ) ” , (34)
dN
where C and IZ are material parameters. The predictive capability of this formula
permits the inclusion of allowed crack length or maximum working life in the
design of a component subjected to cyclic loading in service.
The Paris exponent n is in the range 2 4 for most metals and alloys, but is
often much larger in ceramics. This implies that crack advance in ceramics after
fatigue crack nucleation is much more rapid and occurs over a narrow range of
FRONTAL-WAKE BRIDGED-INTERFACE
.--.
Figure 10. Schematic presentation of toughening mechanisms. Frontal-wake mechanisms: (a) dis-
location glide, (b) microcracking, (c) phase transformation, (d) ductile second phase. Bridging
mechanisms: (e) grains, (f) fibers, (g) whiskers, (h) ductile second phase. Reproduced with permission
of . . .. . . . . . . (9).
AK before terminating in catastrophic failure at K = KIc. For example, crack

growth in Sic was found to depend on K,,, rather than AK, and was very
dependent on the stress ratio R [S6]. The damage mechanisms associated with
crack growth under cyclic loading were essentially identical to those in static load-
ing [156-1581.
Fatigue failure in brittle solids is often assumed to be absent, since no significant
plastic deformation occurs at the loaded crack tip. However, when toughening and
energy absorbing mechanisms exist [9,12], progressive changes in the stress field at
the crack tip occur during cyclic loading. Frontal, or crack-wake toughening
mechanisms include localized phase transformations (typical of zirconia) [ 1591,
and constrained plastic deformation of a ductile second phase (for example, metal
particles in a ceramic matrix), Fig. 1Oa-d. Extrinsic or crack-bridging mechanisms
are associated with oriented fibers whiskers [ 1601, grain interlocking, and the
presence of uncracked ligaments [161-1631, or bridging of the advancing crack by
ductile particles [164], Fig. 1O+h. These all reduce the SIF by shielding the crack
tip, and those toughening mechanisms that improve the mechanical performance
under static loading progressively increase the internal stress at the crack tip as a
result of the debris and damage generated during cyclic loading, leading to unstable
crack propagation.
3.5.3 Ballistic Properties

An effective armor module prevents penetration of a projectile at minimum weight
and cost per unit area. Steel armor has been preferred for over a century. Aluminum
and fibrous-composite armors have been developed for armored vehicles and per-
sonnel protection over the past 40 years. A thick armor plate may be required to
defeat a projectile. To minimize the weight of armor, damage to the armor must
be minimized and damage to the projectile maximized. Cracking and erosion of
the projectile in the first microseconds of impact will reduce the effective mass
and velocity of the projectile, increase the contact area over which energy is
dissipated, reduce the impact stress, and increase the volume of armor material par-
ticipating in energy dissipation. Destroying the projectile is thus a first priority, and
has led to the development of bilayer ceramic/metal or ceramic/fibrous composite
armor. The hardness of the ceramic component fractures and erodes the projectile,
but with only a limited absorption of kinetic energy due to the low fracture tough-
ness of the ceramic. The metallic or composite backing then absorbs the kinetic
energy of the projectile and ceramic fragments generated in the primary impact.
The backing also reduces the bending stresses in the module, preventing premature
failure of the ceramic. The efficiency of a ceramic/metal or ceramic/composite
assembly in defeating a given projectile is significantly better than that of a metallic
armor of the same weight/unit area.
Monolithic ceramics have been evaluated extensively for their ballistic perfor-
mance [15,16,165]. The failure modes observed during projectile penetration
include: deflection of the projectile, erosion of projectile and target [1661, shock-
wave propagation and reflection [ 166,1671, and crack initiation and coalescence
[168,169]. These mechanisms are affected by the rigidity and inertia of the armor
module [167], the constraints introduced by the mounting system, and the design
of the backing layer [167]. Each mechanisms may have a role in determining
the minimum weight of armor required to defeat a given threat. The damage
resulting from projectile impact on a hard, brittle target has been studied for over
70 years. Basic studies have used spherical projectiles [170-1721, planar impact
configurations, or Hopkinson bar tests [ 1731. The results are generally consistent,
and may be summarized in terms of three basic phenomena: cracking, crushing,
and erosion.
Damage to alumina tiles impacted by armor piercing rounds has been analyzed in
some detail [ 174-1761. Quasi-static, drop-weight tests have been correlated with armor
piercing tests, and similar damage mechanisms identified. The damage mechanisms,
both with and without lateral constraint, have been described for fully supported alu-
mina tiles and for tiles supported by a thin backing plate. Two types of damage were
identified:short-range damage due to quasi-static stresses generated by the stress front
propagating into the ceramic, and long-range, dynamic damage associated with multi-
ply reflected shock waves. The quasi-static damage is characterized by a wide size
distribution of large fragments, while the dynamic damage is characterized by much

smaller fragments and granules, down to an eroded ‘dust’ of alumina powder. The
projectile is fractured and eroded during the initial stages of impact, reducing projec-
tile mass and velocity, and the amplitude of the stress front decreases rapidly as it
advances in a semi-hemisphere from the contact zone.
The quasi-static damage is in the form of radial tensile cracks resulting from
biaxial bending and initiates at the bottom surface of the tile, opposite the point
of impact, propagating through the thickness as a semi-elliptical crack. A Herzian
cone crack is subsequently initiated at the edge of the tileprojectile contact zone
and propagates toward the bottom surface [177-1821. At high kinetic energies,
crushing occurs within the cone envelope beneath the impact, with microcrack
coalescence under the high compressive stress. Compressive cracks have been
termed ‘wing’ cracks by Ashby and coworkers and identified as a primary source
of damage in brittle materials subjected to compression [177,180]. Dynamic
damage in the form of spa11 cracks is generated when an elastic shock wave is
reflected from the tile-backing interface [176]. A high impedance, steel support
plate transmits the shock-waves through the ceramic-support interface, reducing
the damage in the impedance-matched ceramic.
Delaying the onset of failure in a ballistic ceramic tile (by only microseconds) can
promote erosion of the projectile. To achieve this, a delay in the onset of quasi-static
damage is required, primarily by reducing the bending moment. Limiting dynamic
damage is especially important for multiple impact performance, and is probably
best addressed by optimizing the lateral dimensions of the tile in order to limit
long-range damage without sacrificing overall protection.
3.6 Conclusions
The design of components for applications requiring high stiffness, hardness and
wear resistance under extreme conditions of load and temperature leaves little
choice for materials selection outside the realm of ‘hard materials’. The spectrum
of materials options available to the design engineer is by no means as broad and
varied as could be wished, but a combination of advances in materials and process
development, together with improved methods of inspection and testing, has led to a
significant reduction in the uncertainties and risks associated with the estimation of
mechanical failure probabilities and component for brittle materials in service.
Coating technology remains the least understood area in this domain, and the
improvements in sophistication of production technology (the development of
diamond coating, for example) have not necessarily been accompanied by improved
understanding of the factors that determine the performance of hard coating in
service. The optimization of coating thickness, the development of graded
(‘functionally graded’) coatings, the effect of substrate surface preparation prior
to coating, and the compatibility of the elastic and thermal properties of substrate
and coating, are all complex topics. They need to be addressed.
Ref&wms 99
References
I . M. F. Ashby and D. R. H. Jones, Engineering Materials I and 2, Int. Series on Materials
Science and Technology, Vol. 39, Pergamon Press, Oxford, 1988.
2. M. F. Ashby, Materids Selection in Mechanical Engineering, Pergamon Press, Oxford, 1992.
3. I. J. McColm, Ceramic Hardness, Plenum Press, New York, 1990.
4. F. S. Petit, in Coating f o r High Temperature Applicutions, E. Lang (Ed.), Applied Science
Publications, Essex, 1983, pp. 1-32.
5. C. Puret and P. Pichoir, in Coating,/or Hight Tempertrture Applications, E. Lang (Ed.), Applied
Science Publications, Essex, 1983, pp. 33-78.
6. D. Satas, Coating Technology Handbook, Dekker, 1991.
7. Y. Kimura, K. Okada. and Y. Enomoto, Wear, 1989, 133, 147-161.
8. 0. 0. Ajayi and K. C. Ludema, Wear, 1988, 124, 237-257.
9. B. R. Lawn, Fracture of’ Brittle Solids, 2nd edn., Cambridge Solid State Science Series,
Cambridge University Press, 1993.
10. D. Broek, Elementnry Engineering Fracture Mechanics, 4th edn., Kluwer Academic Press,
Dordrecht, 1986.
11. M. F. Kanninen and C. H. Popelar, Advanced Fructure Mechanics, Oxford Engineering
Science Series 15, Oxford University Press, Oxford, 1985.
12. A. G. Evans, J . Am. Ceram. Soc. 1990, 73, 187-206.
13. J. Gallagher, Damage Tolerance Design Handbook, Revised edn., Metals and Ceramics Infor-
mation Center, Wright-Patterson Air Force Base, Columbus, OH, 1983.
14. T. Nishioka et al., Sumitomo Elec. Tech Rev. 1990. 29, 262-269.
15. R. C. Laible, in Ballistic Materials and Penetrution Mechanics, R. C. Laible (Ed.), Elsevier
Scientific Publishers, Amsterdam, 1980, pp. 133-143.
16. D. G. Brandon, Ceramic Armor, in Concise Encj~clopcJdiaof Advanced Ceramic Materials, R. J.
Brook (Ed.), Pergamon Press, Oxford, 1991, pp. 22-24.
17. D. W. Lazenby and P. Phillips, Cutting,for Construction, Architectural Press, London, 1978.
18. H. J. Swinheart (Ed.), Cutting Tools Materials Selections, Am. Soc. of Tools and Manufactury
Eng., Dearborn, Michigan, 1986.
19. T. Tatsuno and M. Miyamoto, Kobelco Technol. Rev. 1994, 117, 11-15.
20. G. Lutz, Int. Ceram. Rev. 1993, 42. 367-371.
21. R. T. Coelho, S. Yamada, D. K . Aspinwall, and M. L. H. Wise, Int. J . Machine Tools Manuf:
1995,35, 761-774.
22. K. Przyklenk, Indust. Diamond Rev. 1993, 53. 192-195.
23. T. Ikeda, Y. Kawate, and Y. Hiral, Kobelco Technol. Rev. 1989, 6, 1 4 .
24. J. Bruhn, S. Schicker, D. E. Garcia, R. Jenssen. F. Wagner, and N. Claussen, Ke.v Eng. Muter.
1997, 121, 73-80.
25. W. D. Kingery, H. K. Bowen, and D. R. Uhlmann, Introduction to Ceramics, 2nd edn., A
Wiley-Interscience Publication, John Wiley & Sons, Inc., New York, 1976.
26. S. Chattergee and S. Chandrashekhar, Key Engin. Muter. 1996, 114, 67-104.
27. Z. Hashin and S. Shtrikman. J. Mech. Phys. Solids, 1963, 11, 163.
28. M. Paley and J. Aboudi, Comp. Sci. Technol. 1991, 41,411429.
29. T Christman, A Needelman, and S. Suresh, Actu Metall. 1989, 37, 3029-3050.
30. G. Bao, J. W. Hutchinson, and R. M. McMeeking, Acta Metall. 1991, 39, 1871-1882.
31. A. Maewal, Comput.Meth. Appl. Mech. Eng. 1986, 57, 191-205.
32. C. T. Sun and W. C. Liao, J . Cotnp. Mater. 1990, 24, 977-993.
33. G . D. Quinn and R. Morrell, J . A m . Ceram. Soc. 1991, 74,2037-2066.
34. W. Weibull, Sived. Ins. Eng. Res. Proc., 151, 1-45,
35. E. J. Gumbel, Statistics of’E.utremes, Columbia University Press, New York, 1958, p. 136.
36. L. G. Johnson, Theory and Technique of’ Variation Research, Elsevier, Amsterdam, 1964, p. 137.
37. R. W. Davidge, Mechanical Behavior of Ceramics, Cambridge University Press, Cambridge,
1979.
38. D. Lewis, A m . Ceram. Soc. Bull. 1982, 61, 1208-1214.
39. M. J. Hanney and R. Morrell, Proc. Br. Ceram. Soc. 1982, 13, 277-290.
40. D. Lewis 111 and S. M . Oyler, J. Am. Ceram. SOC.1976, 59, 289-307.
41. J. Seidel, N. Claussen, and J. Rodel, J. Europ. Ceram. SOC. 1995, 15, 395404.
42. T. Lube, M. Manner, Key Eng. Muter. 1997,132,488-491.
43. A. A. Griffith, Phil. Trans. Roy. SOC.Lond. 1920, A221, 163.
44. A. A. Griffith, Proc. First Znt. Cong. Appl. Mech., C. B. Biezeno, J. M. Burgers, and J. Waltman
Jr. (Eds), 1924, Delft, p. 35.
45. C. E. Inglis, Proc. Roy. SOC.Lond. 1913, A127,290.
46. G. R. Irwin, in Handbuch der Physik, Vol. 6, Springer-Verlag, Berlin, 1958, p. 551.
47. E. Orowan, Weld. Res. Supp. 1955, 34, 157s.
48. M. L. Williams, J. Appl. Mech. 1958, 24, 109.
49. T. T. Shih and C. Opoku, Eng. Frac. Mech. 1979, 12,479498.
50. D. P. Rooke and D. J. Cartwright, Compendium of Stress Intensity Factors, HMSO, London,
1976.
51. H. Tada, P. Paris, and G. Irwin, The Stress Analysis of Cracks Handbook, Del Research
Corporation, Hellertown, PA, USA, 1987.
52. Y. Murakami, Stress Intensity Factors Handbook, Vol. 1 and 2, Pergamon Press, Oxford, 1987.
53. D. P. Rooke, F. I. Baratta, and D. J. Cartwright, Eng. Fract. Mech. 1981, 14, 397426.
54. G. C. Sih, Handbook of Stress Intensity Factors, Lehigh University, 1973.
55. J. W. Hutchinson and Z . Suo, in Advances in Applied Mechanics, J. W. Hutchinson and T. Y.
Wu (Eds), Academic Press, New York, 1991.
56. L. B. Freund, Dynamic Fracture Mechanics, Cambridge University Press, Cambridge, 1990.
57. E. C. Combe and D. C. Smith, Br. Dent. J. 1964, 117, 237.
58. A. H. Cottrell, Dislocations and Plastic Flow in Crystals, Oxford University Press, Oxford, 1965.
59. D. Tabor, The Hardness of Metals, Clarendon Press, Oxford, 1951.
60. R. Hill, The Mathematical Theory of Plasticity, Clarendon Press, Oxford, 1950.
61. S. R. Witek, G. A. Miller, and M. P. Harmer, J. Am. Ceram. SOC.1989, 72, 469473.
62. L. A. Davis, c.-P. Chou, L. E. Tanner, and R. Ray, Scripta Metall. 1976, 10, 937.
63. H. Li and R. C. Bradt, J. Non-Cryst. Sol. 1992, 146, 197.
64. P. M. Sargent and T. F. Page, Proc. Br. Ceram. SOC.1987,26,209.
65. P. F. Becher, J. Am. Ceram. SOC.1974, 57, 107.
66. J. B. Quinn and G. D. Quinn, J. Muter. Sci. 1997, 32, 4331.
67. R. W. Rice, C. C. Wu, and F. Borchelt, J. Am. Ceram. SOC.1994.77. 2539-2553.
68. K. Zeng, E. Soderlung, A. E. Giannakopoulus, and D. J. Rowcliffe,’Acta Muter. 1996, 44,
1127-1141.
69. J. Lankford and D. L. Davidson, 3rd International Conference on Mechanical Behavior of
Materials, Pergamon, Oxford, 1980.
70. J. Valli, J. Vac. Sci. Technol. A , 1986, 4, 3007-3014.
71. S. Herada, N. Noda, T. Endo, and M. Nagano, J . SOC.Muter. Sci. Jpn. 1988, 37, 904-907.
72. Y. Le Petitcorps, M. Leheye, R. Pailer, and R. Naslain, Cornp. Sci. Technol. 1988, 32, 31-55.
73. B. Thomas, D. West, W. E. West, and K. Wyckoff, Powder Metall. Znt. 1988, 20, 39-44.
74. R. W. Davidge, Muter. Sci. Technol. 1986, 2, 902-909.
75. American Society for Testing and Materials, Standard Test for Plane Strain Fracture Tough-
ness of Metallic Materials, ASTM E-399, Philadelphia, Pennsylvania, 1987.
76. I. Hadley and M. G. Dawes, Fatigue Fract. Eng. Muter. Struct. 1996, 19, 963-973.
77. H. Awaji, T. Yamada, and H. Okuda, J. Jpn. Ceram. SOC.,Znt. Edn. 1991, 4, 403408.
78. G. D. Quinn, J. J. Kubler, and R. J. Gettings, VAMASTechnical Report No. 17, NIST,
Gaithersburg, Maryland, 1994.
79. T. Nishida, Y. Hanaki, and G. Pezzotti, J. Am. Ceram. SOC.1994,77, 606608.
80. D. Munz, in Fracture Toughness of Ceramics, Vol.6, R. C. Bradt, A. G. Evans, D. P. H. Has-
selman, and F. F. Lang (Eds), Plenum Press, New York, 1983.
81. B. R. Lawn and E. R. Fuller, J. Muter. Sci. 1975, 10,2016-2024.
82. R. Warren and B. Johnson, Powder Metall. 1984, 27, 25-29.
83. T. Sadahiro and S. Takatsu, in Modern Developments in Powder Metallurgy, Vol. 14, H. H.
Hauser, H. W. Antes, and G. D. Smith (Eds), Metal Powder Industries Federation, Princeton,
NJ, 1981, pp. 61-72.
84. T. Nose and T. Fujii, J. Am. Ceram. SOC.1988,71, 328-333.
References 101
85. D. Sherman, J . Am. Ceram. SOC.1997,80, 1904-1906.

86. C. J. Gilbert, R. H. Dauskardt, and R. 0. Ritchie, J. Am. Ceram. Soc. 1995,78,2291-2300.
87. J. A. Salem, G. D. Quinn, I. Bar-On, and M. G. Jenkins, A S T M Provisional Standard P S 070-
97, ASTM Committee C-28, 1997.
88. B. R. Lawn, J. Am. Ceram. SOC.1998,81, 1977-1994.
89. R. B. Tait, P. R. Fry, and G. G . Garrett, Exp. Mech. 1987, 27, 14-22.
90. M. Manoharan, S. v. Kamat, Znt. J. Fract. 1995, 73, R41LR45.
91. C. J. Gilbert, J. J. Cao, L. C. De Jonghe, and R. 0. Ritchie, J. Am. Ceram. Soc. 1997,80,2253-
2261.
92. B. R. Murphy, W. W. Predebon, and B. J. Pletka, J. Muter. Sci. Lett. 1994, 13, 13461348.
93. V. S. Krilov, V. I. Kovalenko V. G. Marinin, I. T. Ostapenko, and V. P. Podtyakn, Sov. J.
Superhard Mater. 1983, 5, 21.
94. S . Palmqvist, Jernkontorets Annalen, 1957, 141, 300.
95. S . Palmqvist, Jernkontorets Annalen, 1963, 167, 208.
96. B. R. Lawn and M. V. Swain, J . Mater. Sci. 1975, 10, 113.
97. B. R. Lawn and M. V. Swain, J. Muter. Sci. 1975, 10, 1049.
98. B. R. Lawn and A. G. Evans, J. Muter. Sci. 1979,12, 2195.
99. D. B. Marshall, B. R. Lawn, and A. G. Evans, J . Am. Ceram. SOC.1982,65, 561.
100. D. B. Marshall and B. R. Lawn, in Microindentation Techniques f o r Materials Science
and Engineering, P. Blau and R. Lawn (Eds), ASTM STP 889, Philadelphia, Pennsylvania,
p. 26.
101. J. R. Kelly, M. E. Cohen, and J. A. Tesk, J. Am. Ceram. SOC.1993,76,2665-2668.
102. S . G. Cook, J. E. King, and J. A. Little, Muter. Sci. Techno/.,995, 11, 1093-1098.
103. S. Srinivasan, R. 0. Scattergood, Li Zhuang, A. Grosh, A. S. Kobayashi, and R. C. Bradt,
J. Am. Ceram. Soc. 1991,74,887-890.
104. C. B. Ponton and R. D. Rawlings. Mater. Sci. Technol. 1989, 5, 961-976.
105. R. Kurth, B. F. Lawlor, Y. J. Stockmann, and R. W. Steibrech, Fract. Mech. Ceram. 1996,11,
151-162.
106. R. F. Cook and L. M. Braun, J. Muter. Sci. 1994, 29, 2192-2204.
107. S. I. Athabai, J. Bodel, T. Dabs, and B. R. Lawn, J. Mater. Sci. 1991, 26, 2313-2321.
108. J. Foster, Key Eng. Mater. 1996, 122, 247-278.
109. E. Ness and R. Zibbel, Wear, 1996, 196, 12e125.
110. J. E. Ritter, Key Eng. Muter. 1992, 171, 93-106.
111. Y. I. Oka, H. Ohnogy, T. Hosokawa, and M. Matsumura, Wear, 1997,203,573-579.
112. J. E. Ritter, S. R. Choi, K. Jakus, P. J. Whalen, and R. G. Rateick Jr., J. Muter. Sci. 1991,26,
5543-5546.
113. G. Barbezat, A. R. Nicoll, and A. Sickinger, Wear, 1993, 162-164, 529-537.
114. Chen Yin Qian et al., Wear, 1990, 137, 175-186.
115. K. Kato, Wear, 1992, 153, 277-295.
116. A. L. Ryzhikov V. M. Gropyanov, and L. G. Tarakanchikov, Sov. J. Frict., Wear, 1991,12,
143- 146.
117. J. Echigoya, Strength of Materials, Proceedings of the 10th International Conference, Japan
Institute of Metals, Sendai, Japan, 1994, pp. 853-856.
118. J. Echigoya, T. Hyakubu, Z. T. Liu, K. Sasaki, and H. Suto, Surf. Coat. Techno/. 1994, 67,
11 1-1 17.
119. D. A. ONeil and S. F. Wayne, J. Eng. Mater. Tecnol. 1994, 11, 471478.
120. V. Hajek, K. Rusnak, J. Vlcek, L. Martinu, H. M. Hawthorne, Wear, 1997,213, 80-89.
121. R. Raj and R. K. Boria, Tailoring Multiphase and Composite Ceramics, Proceeding of the 21st
University Conference on Ceramic Science, Plenum, New York, 1986 pp. 27-39.
122. P. Basetty, T. Jensen, and J. N. Gray, Review of Progress in Quantitative Nondestructive
Evaluation, Plenum, New York, 1997, pp. 2167-2174.
123. J. A. Medding and C. Duke Jr., Review of Progress in Quantitative Nondestructive Evaluation,
Plenum, New York, 1997, pp. 1503-1508.
124. D. J. Roth, D. B. Stang, S. M. Swickard, M. R. DeDuire, and L. E. Dolhert, Muter. Eva/.
1991,49, 883-888.
125. J. Lamon, Rev. Metall. 1995, 92, 265284.
126. G. Petzow, R. Telle, and R. Danzer, Muter. Churact. 1991, 26, 289-302.
127. A. Zimmermann, M. Hoffman, B. D. Flinn, R. K. Bordia, T. J. Chuang, E. D. Fuller, and
J. Rodel, J . Am. Ceram. SOC.1998,81, 2449-2457.
128. L. A. Xue, K. Meyer, and I. Wei Chen, J. Am. Ceram. SOC.1992,75,822-829.
129. Yet Ming Chiang and W. D. Kingery, J. Am. Ceram. SOC.1989,72,271-277; 1990,73, 1153-
1158.
130. V. Jayaram, R. Manna, M. G. Kshetrapal, J. Sarker, and S. K. Biswas, J . Am. Ceram. SOC.
1996,79,710-772.
131. B. S. S. Daniel and V. S. R. Murthy, Muter. Des. 1995, 16, 155-161.
132. R. W. Evans, D. P. Thompson, and B. Wilshire, Br. Ceram. Trans. 1993,92, 28-30.
133. W. E. Luecke, S. M. Wiederhorn, B. J. Hockey, R. F. Krause Jr., and G. G. Long, J. Am.
Ceram. SOC.1995,78,2085-2096.
134. Y. Maehara and T. G. Langdon, J . Muter. Sci. 1990, 25, 2275-2286.
135. I.-W. Chen and L. A. Xue, J . Am. Cerum. SOC.1990,73,2585-2609.
136. T. G. Nieh, J. Wadsworth, and F. Wakai, Int. Muter. Rev. 1991,36, 146-161.
137. F. Wakai, S. Sakaguchi, and Y . Matsuno, Adv. Cerum. Muter. 1986, 1, 259-263.
138. T. G. Nieh, C. M. McNally, and J. Wadsworth, J. Muter. 1989, 41, 31-35.
139. K. Kajihara, Y. Yoshizawa, and T. Sakuma, Scripta Metall. Muter. 1993, 28, 559.
140. T. G. Nieh and J. Wadsworth, J. Muter. Res. 1990, 5, 2613-2615.
141. F. Wakai and T. Nagano, J. Muter. Sci. 1991, 26, 241-2447.
142. J. Karch, R. Birringer, and H. Gleiter, Nature, 1987, 330, 556558.
143. Z. Cui and H. Hahn, NanoStructured Muter. 1991, 1, 419-425.
144. A. H. Chokshi, A. Rosen, J. Karch, and H. Gleiter, Scripa Metall. 1989, 23, 1679-1684.
145. A. M. El-Sherik, U. Erb, G. Palumbo, and K. T. Aust, Scripta Metall. 1992, 27, 1185-1188.
146. H. J. Hofler and R. S. Averback, Scriptu Metall. 1990, 24, 2401-2406.
147. R. 0. Scattergood and C. C. Koch, Scripta Metall. Muter. 1992, 27, 1195-1200.
148. C. S. Pande, R. A. Masumura, and R. W. Armstrong, NanoStructuredMater. 1993,2,323-331.
149. V. G. Gryaznov, M. Yu. Gutkin, A. E. Romanov, and L. I. Trusov, J . Muter. Sci. 1993, 28,
43594365.
150. W. R. Cannon and T. G. Langdon, J . Muter. Sci. 1988,23, 1-20.
151. E. Yasuda, K. S. Ramesh, S . Kimura, Report of the Research Laboratory of Engineering
Materials, Tokyo Institute of Technology, 1985, 10, 115-134.
152. A. Hynes and R. Doremus, Crit. Rev. Solid State Muter. Sci. 1996,21, 129-187.
153. A. H. Chokshi and T. G. Langdon, Muter. Sci. Technol. 1991,7, 577-584.
154. D. S . Wilkinson, Fruct. Mech. Ceram. 1992, 10, 349-366.
155. D. P. H. Hasselman, K. V. Donaldsod, and A. Venkateswaran, Fract. Mech. Ceram. 1992,10,
493-507.
156. G. Grathwohl, Muterialwiss. Werkstofftech. 1988, 19, 113-124.
157. E. D. Whitney, Powder Metall. Znt. 1983, 15, 201-205.
158. A. G. Evans, Znt. J . Fract. 1980, 16,485498.
159. D. B. Marshal, M. C. Shaw, R. H. Dauskard, R. 0 Ritchie, M. J. Readey, and A. H. Heuer,
J . Am. Ceram. SOC.1990,73,2659-2666.
160. R. H. Dauskardt, B. J. Dalgleish, D. Yao, R. 0. Ritchie, andP. F. Becher, J. Muter. Sci. 1993,
28, 3258-3266.
161. C. J. Gilbert, J. J. Cao, W. J. MoberlyChan, L. C. DeJonghe, and R. 0. Ritchie, Actu Muter.
1996,44, 3199-3214.
162. C. J. Gilbert, R. N. Petrany, R. 0. Richie, R. H. Dauskardt, and R. W. Steinbrech, J . Muter.
Sci. 1995, 30, 643.
163. C. J. Gilbert and R. 0. Ritchie, Eng. Fract. Mech. 1998, 60, 303-313.
164. Jian-Ku-Shang and R. 0. Ritchie, Metall. Trans. A , 1989, 20, 897-908.
165. M. L. Wilkins, C. F. Cline, and C. A. Hondol, Tech. Rept. UCRL-50694, Lawrence Radiation
Laboratory, 1969.
166. M. L. Wilkins, in Ballistic Materials and Penetration Mechanics, R. C . Laible (Ed.), Elsevier,
Amsterdam, 1980, pp. 225-252.
167. H. Kolsky, in Ballistic Materials and Penetration Mechanics, R. C. Laible (Ed.), Elsevier,
Amsterdam, 1980, pp. 185-252.
References 103
168. Y. J. Lee, Int. J . Frac. 1992, 57, 2546.

169. C. C. Ma and S. K. Chen, Int. J . Frac. 1992,58, 345-359.
170. A. S. Kobayashi, Muter. Sci. Eng. 1991, A143, 1 11-1 17.
171. Z. Rosenberg, Y. Yeshurun, and G. Brandon, J . Physique, 1985, 8, 331-341.
172. D. A. Shockey, A. H. Marchand, S. R. Skaggs, G. E. Cort, M. W. Burkett, and R. Parker, Int.
J . Impact Eng. 1990,9, 263-275.
173. J. E. Field, S. M. Walley, N. K. Bourne, and J. M. Huntly, Journalde Physique IV, 1994,4, C8/
2-22.
174. D. Sherman and D. G. Brandon, J. Muter. Res. 1997, 12, 1335-1343.
175. D. Sherman and T. Ben-Shushan, Int. J. Impact Eng. 1998,21,245-265.
176. D. Sherman, Int. J . Impact Eng., accepted, 1999.
177. B. R. Lawn, J. Am. Ceram. SOC.1998, 81, 1977-1994.
178. C. Kocer and R. E. Collins, J. Am. Cerurn. Soc. 1998,81, 1736-1742.
179. R. Mouginot and D. Maugis, J. Muter. Sci. 1985, 20, 4354.
180. M. F. Ashby and S. D. Hallam, Acta Metall. 1986, 34, 497.
Nanostructured Superhard Materials

S. Vepfek
4.1 Introduction
Hardness of a material is a measure of its ability to resist deformation [l, 21. The
upper limit of the hardness is determined by the rigidity of the crystal structure
upon isostatic pressure. This ‘theoretical’ hardness is proportional to the reciprocal
value of the bulk modulus, B. From the Hooke’s law B = da/dE (CTis the applied
stress and E the resulting elastic strain) and from the relationship between the
stress (a force) and the first derivative of the binding energy, E, with bond distance,
(2)
a, at the equilibrium position ao, CT = (dEb/da)o, one obtains [3]
B= 0 .
Thus, the theoretical hardness increases with the increasing curvature of the inter-
atomic potential curve at the equilibrium bond distance, ao, that is, with increasing
bond energy and decreasing bond length. A high bond energy means a high electron
density between the atoms as found for a nonpolar, covalent bond between atoms of
small radii of the first period. In addition, a high coordination number is required in
order to maximize the value of B. For these reasons carbon in its metastable, four-
fold coordinated sp3-hybridization, diamond polymorph, is the hardest material
known, followed by cubic boron nitride, cBN [2].
Based on such general considerations C. M. Sung predicted in 1984 that C3N4
should have a hardness comparable to diamond [2]. Later on, Cohen [4] and
coworkers [5-71 derived Equation (2), which reproduced very well the bulk
moduli of many materials and confirmed the prediction of C. M. Sung regarding
the theoretical hardness of C3N4,
Here, ( N c ) is the average cpordination number and X is the polarity of the bond, B is
in GPa and a. is given in Angstroms [6]. For nonpolar, covalent bonds in diamond
X = 0, whereas in other compounds, such as cBN, Si3N4and C3N4 X > 0 which
decreases the value of the elastic modulus. The expected high theoretical hardness
of C3N4 is based on the small bond distance and relatively small polarity A.
However, like diamond and cBN, C3N4is a thermodynamically unstable com-
pound whose preparation requires a high degree of dissociation of nitrogen in
order to achieve the desired stoichiometry, a medium-energy ion bombardment to
promote the carbon into the metastable fourfold coordination and a relatively
high temperature of about >800°C in order to evaporate the paracyanogen,
(CN),, which is always formed during the deposition of the stoichiometric C3N4(see
[S] for further details). However, the Vickers hardness reported for stoichiometric
amorphous C3N4 films reached only about 30 GPa, far below that of diamond
(70-100 GPa [2,5,6]). Using similar synthetic principles, other authors succeeded
in the deposition of crystalline, stoichiometric C3N4 films [9-131 but they could
not measure the hardness because of an insufficient density of the polycrystalline
films [lo]. A large number of papers report on the ‘synthesis of carbon nitride’
but those films are highly substoichiometric with a large nitrogen deficiency even
below that of paracyanogen and have a lower hardness. Nevertheless, CN, may
find some ‘niche’ applications such as protective, low friction coatings on magnetic
storage discs [14-16].
In contrast to the C3N4 and substoichiometric CN, thin films of turbostratic
graphite with a relative low concentration of nitrogen, CN0,25p0,35 and three
dimensional cross-linking deposited at a relatively high temperature of 400°C
have a hardness of 40-50GPa [17-191. Also the hardness of diamond crystals can
be increased by nitrogen impurities 120,211, thus reaching the upper end of the
hardness scale in Fig. 1. These are two examples of the well known rule that the
microstructure determines the practically achievable strength (and hardness) of
engineering materials, which is orders of magnitude smaller than the theoretical
strength. The latter is calculated for the slip of crystal planes as c,(theor)
G/2n, where cT,(theor) is the theoretical critical fracture stress and G the shear
modulus [3,22]. The reason is that the deformation and fracture of materials
occurs due to the multiplication and movement of dislocations in crystalline
materials and growth of microcracks in glasses and ceramics, all of which require
a much smaller stress than the theoretical value a,(theor) mentioned above [3,22-
241. Therefore, the practical strength and hardness of materials is determined by the
6000{ (nc-M,N/a-Si3N, 160
4000
0
Figure 1. Vickers hardness of selected materials in comparison with that of the heterostructures and
nanocrystalline composites. Notice that the hardness of diamond depends on the quality and purity
of the crystal (see text).
106 4 Nanostructured Superhard Materials
9000 I I I . I r n I . I I
8000 -
0
-
7000 - -
N
-
E 6000- 0 -
0 O . 0
E -
3 5000-
25
(I) 4000- 0 -
(I)
3000-
0
-
E -
nJ 2000-
0
r
1000 - -
0 I ' I ' I ' I ' I .
0 20 40 60 80 100
Figure 2. (a) An example of the indentation curve measured on a =5pm thick layer of
ncTiN/aSi3N4.L,,, and h,,, are the maximum applied load and indentation depth, hplastic is
the depth of the remanent, plastic deformation. The arrows -+ and t denote the loading
and unloading part. (b) Measured values of the hardness against the maximum applied load. (c)
Indentation depth at a constant applied load of 50mN against time which shows the absence of
creep.
c-
'ma, , I , I , I
I . I . I .
0.26
H, = 7529 [N/mm2]
0.24
- El(1-v2) = 520.28 [GPa]
.....................
sl
sf$0.22- -
-0 - -
C
0
.g 0.20
- -
c
E
9,
TI
E 0.18
- -
c
0.16z0
0
time [sec]
(C)
microstructure which hinders the multiplication and movement of dislocations and

growth of the microcracks. Of course one always should use a material with a pos-
sibly high cohesive energy and optimize the design of its microstructure depending
on the conditions of its applications and on the nature of the load imposed on that
material (tensile, compressive, shear, static, dynamic, . . .).
The same applies for the hardness. Besides the 'theoretical' hardness as the
resistance of a material against isostatic compression, depending on the reciprocal
value of the bulk modulus mentioned above (see Equation (2)), three categories of
hardness are distinguished in practical measurements: Scratch hardness (e.g. Mohs
scale used by mineralogists), static indentation hardness (e.g. Brinell, Vickers,
Knoop) and rebound or dynamic hardness. In the latter measurement the indentor
is usually allowed to fall under gravity from a given height which, together with the
indentor mass, defines the impact energy [ 11. Obviously, the theoretical hardness is
of little use in considering a material for an application. Instead one has to
consider the practically measured type of hardness depending on the particular
applications.
Here we shall limit ourselves to the static indentation measurements using the
Vickers method for the simple reason that this, and the Knoop technique, are the
only ones which can be simply applied to the measurement of the superhard
materials. The Vickers indentor is a regular pyramid made of diamond with an
angle of 136" between the opposite faces. The choice of this angle is based on an
analogy with the Brinell test because both methods yield similar values of hardness,
however only for relatively soft materials with H 5 1000kg mm-2 (see [ 11for further
details). The diamond pyramid is pressed into the material to be tested under a
defined load L (kg) and, after unloading, the average value of the two diagonals d
(mm) of the resultant plastic deformation is measured with a microscope. The

Vickers hardness is proportional to the ratio of the applied load and the area of
the plastic deformation H = const L / d 2 and given in units of kgmm-2 or GPa
[ 1,2,25-281. (For historical reasons, the values of Vickers hardness are usually
given in kgmmP2, the conversion factors to other units used are
1000kgmmP2 = 9.807 GPa = 9807NmmP2.) Figure 1 shows the Vickers hardness
of some materials including the heterostructures and nanocomposites to be dis-
cussed here. The Knoop hardness is measured in a similar way, but the diamond
pyramid has two different angles of 172’30’ and 130’, thus yielding an elongated
plastic deformation. The hardness is calculated from the longer diagonal whereas
the shorter one yields information on the elastic properties (see [l, 24,291 for further
information).
Modern indentometers for the measurement of the hardness of thin films use
a computer controlled step-wise increase of the load up to a chosen maximum
L,,, followed by a step-wise unloading. Instead of measuring the diagonal of the
plastic deformation, the indentation depth, h, is measured electronically and the
indentation curve evaluated (see Fig. 2a). The ‘plastic hardness’ is calculated from
the maximum load, L,,,, and the depth of the plastic indentation, hplastic,as
The ‘universal hardness’, H,, is obtained from the same formula if h,,, is inserted
instead of hplastic. The universal hardness includes both the elastic and plastic
deformation. The linear part of the unloading curve corresponds to the elastic
recovery when the diamond pyramid is in a constant area contact with the material.
Therefore it represents Hooke’s law and allows one to calculate the corresponding
‘elastic modulus’ E/(1 - v 2 )which is a complicated function of the bulk, shear, and
tensile moduli (v is the Poisson ratio). The details of the apparatus, the measuring
procedure and possible errors are given in the relevant papers to which we refer here
[25-281. If done correctly, the plastic hardness measured by the indentation agrees
within about 10-1 5% reasonably well with that from the classical Vickers method at
least in the range H 5 1500 kgmmp2 [25].
The so called ‘nanoindentation’, which is frequently used to measure < 1 pm thin
films, is subject to a number of possible errors [27,28]. For example, when
measuring soft materials, such as aluminum or pure iron with a small load and
indentation depth, the dislocations are pinned in the surface contaminant layer
(oxides, carbides) and, consequently, unrealistically high values of hardness are
found. The plastic deformation may also need a certain time to reach equilibrium
under the given load because of a finite velocity of dislocation movement. This
can be seen as creep (increase of the indentation depth) when a constant load is
applied for 10-30 s.
Therefore, in order to obtain reliable values of the hardness, one should always
use at least >4pm thick films, make sure that the maximum indentation depth,
h,,,, does not exceed 10% of the film thickness, check for possible dynamic
response of the material by sustaining the indentor at the maximum load L,,,
4.2 Concept for the Design of Superhard Materials 109
for at least 2&30 s and measure the hardness (always at a different position on the
layer) as a function of the maximum applied load. Figure 2b shows the typical
dependence of the measured plastic hardness on L,,, and the typical scattering
of individual measurements. In the case of superhard materials, one typically
finds a smaller hardness and large scattering of the individual measurements
when using small loads of SlOmN because the indentation depth of 50.1 pm is
comparable with the surface roughness. Figure 2c shows the time dependence of
A,, under the constantly applied maximum load L,,,. One notices that there is
no creep seen, at least on the time scale available with the given instrument. We
emphasize these problems here to make sure that in the measurements to be
reported hereafter, all these precautions have been taken in order to make the
results as reliable as possible. However, one should keep in mind that the Vickers
technique, which was originally developed for the conventional alloys and
materials with hardness not exceeding 2000-2500 kg mm-*, is not suitable for
the measurements of materials whose hardness approaches that of diamond.
Therefore we consider values of hardness of 7000-10 000 kg mmp2 (and even
higher) frequently found in our measurements only as an indication that these
materials approach the hardness of diamond, but the Vickers hardness scale is
probably not linear in that range. As it will be shown later, there is no dislocation
activity in the superhard nanocomposites and, therefore, no ‘plastic deformation’
in the classical sense. Strictly speaking one should use the term ‘remaining’, ‘per-
manent’ or ‘pseudoplastic’ deformation because it is a brittle fracture as found, for
example when indenting into a glass. Because the current literature does not differ-
entiate between these terms we adopt the term ‘plastic deformation’ keeping, how-
ever, in mind that it is in fact a brittle fracture whose microphysics and
micromechanics are not yet understood.
Last but not least, one should not forget that for practical applications the
hardness at room and at a high temperature is only one of many properties which
determine the applicability of a material. Further mechanical properties, such as
fracture toughness under static and dynamic load, coefficient of friction, corrosion
resistance, and reactivity with the material to be machined at room as well as at
elevated temperature are decisive for most applications.
4.2 Concept for the Design of Superhard Materials

Most researchers agree on the definition accordin to which ‘superhard’ materials
are those with Vickers hardness 24000 kg mm-’ (about 40 GPa), and that the
‘theoretical hardness’ (see Equation (2) and related discussion) is only of an
academic interest but hardly of any use. There are various possible approaches
to achieve such a goal. A strong material must have a high cohesive energy, be
free of flaws (e.g. voids), and its microstructure must be designed in such a way
as to efficiently hinder the multiplication and movement of dislocations and
growth of microcracks. The fracture stress, af,,can be increased in various ways
known from the metallurgy, which are briefly summarized in Equation (4) (see e.g.
~31):
Here, the first term describes the Peierls-Nabarro stress which determines the dislo-
cation movement (slip within the crystallographic plane) in a crystal, the second
term corresponds to the solution hardening (C is the concentration of the impuri-
ties), the third term is the Hall-Petch relationship [3,30,3I], which describes grain
boundary strengthening (d is the crystallite size, see below), the fourth term
describes the strain hardening due to dislocation pile up ( a is a material constant,
G the shear modulus, b the Burgers vector, and p the dislocation density). The
last term describes the overall dependence of the crack growth on the size of
microcracks (voids) of the size a. ( E is the Young modulus, ys the surface
energy). Any of these processes allows the engineer to increase the strength of a
material by a different amount. For example, the hardness of TIN films deposited
in glow discharge plasmas under energetic ion bombardment which causes strength-
ening due to the formation of dislocations or due to a high biaxial compressive stress
is typically higher than that of a single crystal.
4.2.1 Nanocrystalline Materials

The decrease of the grain size d strengthens metallic alloys (Hall-Petch relationship,
third term)
as well as ceramics (last term in Equation (4), Griffith theory [3]). There are several
different mechanisms and theories which describe Hall-Petch strengthening
[3,24,30-331: Pile-up models, work hardening and grain-grain boundary composite
models which yield different relationships for a. and kgb [32]. The work hardening
1
models and the core-mantle composite model of Koks 34 yield also a more com-
I .
plicated dependence of ofr on d . In ceramics, the d-'I dependence can originate
from the fact that the size a. of the possible flaws such as voids and microcracks,
which are formed during the processing of the material, also decreases with decreas-
ing grain size. For these reasons, the Hall-Petch relationship (5) should be consid-
ered as a semiempirical formula which is valid down to a certain minimum grain
size, for example of d > 450b = l00nm for the pile-up models and
d > 60b x 15 nm for a dislocation source model [32].
There are many examples which show that the strength and hardness of an alloy
can be increased by a factor of 4-7 if the grain size is decreased into the nanometer
range. One such example is the rapidly solidified A19&r3CelCo, (x = 1-2%) alloy
containing as a main component a nanosized quasicrystalline icosahedral phase [35].
A similar strengthening is found in many metals such as TiAl[36], Cu [37], and other
4.2 Concept ,for the Design of Superhard Materials 111
nanocrystalline metals and intermetallics [38,39]. However, when the dislocations

activity decreases with grain size decreasing below about 20-50 nm (which causes
the strengthening) grain boundary sliding becomes active. This results in a softening
of such materials when the crystallite size decreases below 6-10 nm [36-391. A simple
phenomenological model describes the softening in terms of an increasing volume
fraction of the grain boundary material, fgb, with crystallite size, d, decreasing
below 4-6nm [40]:
1
H ( d ) = (1 -fgblHc +fgbHgb with f g b - (6)
d’
Because the hardness of the grain boundary material, H g b , is smaller than that of the
crystallites, H,, the overall hardness of the material decreases with d decreasing
below 10nm. The amount of such decrease depends on the nature of the grain
boundary material.
The decrease of the hardness in the grain size range 56-20 nm, sometimes called a
‘negative Hall-Petch slope’ was first reported by Chokshi et al. [41] and, afterwards,
subjected to many studies with controversial conclusions [38, 39,42-501. In a recent
paper Sanders er al. [50] have shown, that in the grain size range between about 10
and 100 mn the Hall-Petch slope depends on the sample preparation. If nanocrystal-
line Cu and Pd samples, prepared by inert gas condensation, were compacted at ele-
vated temperature, the slope followed nearly that extrapolated from coarse-grain
samples. However, even in this work one can see that the slope of the experimental
points in the lower grain size range of 20-8 nm decreases and levels off (see Figs 6
and 7 in [50]). The authors emphasize that even such a compaction is probably
not sufficient to remove all flaws from the grain boundaries which lead to softening.
In a more recent paper, Musil and Regent [51] have found such a softening even for
dense nanocrystalline NiCr thin films prepared by argon ion sputtering. These
results, which are shown in Fig. 3, clearly demonstrate that the softening in the
I’
y1600-
.
E
m
251400-
cn
u)
g1200-
E
m
r
.g 1000-
-cnmn
800;,
€ I
d’
r I I I
crystallite size [nm]

Figure 3. Hardness of ncNiCr (80/20 weight-%) alloy prepared by argon ion sputtering against
crystallite size (from [51] with permission).
1 12 4 Nanostructured Superhard Materials
range d < 5-10 nm is most probably an intrinsic property of simple nanocrystalline

materials which have inherent flaws in the grain boundaries. Although many details
are still not understood there is little doubt that grain boundary sliding is the reason
for the softening in this range of the crystallite size [38,39]. Various mechanism of
grain boundary creep and sliding were discussed and are described by deformation
mechanism maps in terms of the temperature and stress [3,52-571. The theories have
been critically reviewed [32]. More recent computer simulations have shown that the
grain boundary sliding in nanocrystalline materials is a universal phenomenon
which can occur without any significant activation at room temperature [58,59].
Based on our understanding of these mechanisms one can expect that an
appropriate design of a nanocrystalline/amorphous composite, which would
avoid the grain boundary sliding, may cause the hardness of such material to
continue increasing even for grain sizes below 6-10 nm and at elevated temperature.
The absence of grain boundary sliding in a nanocrystalline material at room tem-
perature does not exclude its appearance at elevated temperature which may
occur under the conditions of their particular application, such as the leading
edge of cutting tools which may reach a temperature of 700-1000°C.
An idealized microstructure of such a composite is shown in Fig. 4. In reality, one
might also evoke an interwoven bi-phase system as observed by Milligan et al. for
ncAu/aSi (see Fig. 6 in [60]) or a bicontinuous two-phase system. However, as will be
discussed below, we have not observed any such microstructure in our detailed high
resolution transmission electron microscopic study but only isolated nanocrystals
imbedded within the amorphous matrix [61,62]. In the interwoven bi-phase
system ncAu/aSi, Milligan et al. have observed a significant improvement of the
toughness of the composite as compared to a pure ncAu [60].
If such a composite consists of strong materials, such as hard nitrides with a hard-
ness of about 2000 kg mmP2, even a relatively modest increase of the hardness by a
factor of four would result in a hardness comparable to that of diamond. In order to
a-Si,N,
Ol -
Figure 4. Idealized structure of the nanocrystalline-amorphous composite.
4.2 Concept ,for the Design OjSuperhard Materials 113
form such a bi-phase system, both materials must be immiscible (i.e. display a
thermodynamically driven segregation during the deposition) and the cohesive
energy at the interface between both phases must be high. Last but not least, the
amorphous phase must possess a high structural flexibility in order to accommodate
the coherency strain without forming dangling bonds and other flaws. In order to
ensure that such a composite will show no grain boundary sliding even at a high
temperature, both materials should be refractory in order to ensure that the opera-
tional temperature of the tools will be less than half the melting or decomposition
temperature, Tm,decomp 2 2500 K.
It is important to note that, although the strengthening caused by the decreasing
dislocation activity in the range of crystallite size of d 5 lOOnm finally reaches a
saturation due to the absence of dislocations in ultrasmall nanocrystals of
d 5 10 nm [38,39], the strength and hardness of such material can still increase
upon a further decrease of the crystallite size when another mechanism of
strengthening becomes dominant. In the absence of dislocations and grain boundary
sliding, the nanocomposites show a brittle behavior which means that the fracture
strength (and hardness) are proportional to the elastic modulus of the material. This
behavior has been confirmed in our nanocrystalline superhard composites [63-651.
The fracture stress of such material is then determined by the critical stress for the
growth of microcracks (see the last term in Eq. (4)). The size of the 'microcrack', ao,
that is, of a possible flaw and voids, is in a well compacted material always smaller
than the size of the nanocrystals, uo 5 d . Moreover, even if the critical stress for the
crack growth
should be exceeded by the applied (e.g. uniaxial) stress, the crack could hardly grow
to a size larger than d because it would have to bend to pass around the nano-
crystals. If p is the angle between the plane of such a bent crack and the direction
of the tensile stress gappl, the component ( T = ~ oapplsin p < rapplis acting in the
direction normal to the crack plane. Only this component can cause the crack to
grow. For randomly oriented nanocrystals embedded in a 51 nm thin matrix, as
found in our nanocomposites with the smallest crystallite size and highest hardness,
the cracks have to bend so strongly that cp + 0 and ( T <<~ oappl. Furthermore, upon
the bending also crack branching ('meandering') occurs in a three-dimensional iso-
tropic nanocomposite which further decreases the ability of such nanocracks to pro-
pagate [3, 1701. These are the reasons why the hardness of our composites increases
when the crystallite size decreases below 5-7 nm [64,66,67].
The obvious first (but not the only) choice of such an ncja system which meets the
above discussed criteria for achieving the superhardness is straight forward: Many
transition metals M, such as Ti, Zr, Hf, V, Nb, Ta, Cr form stable (refractory) and
hard nitrides which crystallize upon deposition even at relatively low temperature of
>lOO"C [68-701. Silicon nitride, on the other hand, grows amorphous even at
1100°C and it possesses the desirable structural flexibility due to the fourfold
coordination of silicon combined with the threefold coordination of nitrogen.
1 14 4 Nanostructured Superhard Materials
TiN-Si3N4 are immiscible up to a relatively high temperature of 1000°C [71-731. A

simple thermodynamic estimate [63] shows that similar applies to the majority of the
transition metal nitrides mentioned above [74]. An even better choice of the amor-
phous phase should be boron nitride because the boron atoms can be three- and
fourfold coordinated and the B-N bond is stronger than the Si-N one [6,7,74].
Because silicides and borides are refractory materials, also the cohesive energy at
the ncM,N/aSi3N4 and ncM,N/(a and nc)BN interface should be large. We shall
see later that this concept indeed works and superhard materials can be obtained
in the M,N/aSi3N4 and ncM,N/(a and nc)BN systems. Because silicon dissolves
in many metals at high temperatures, the choice of boron nitride, aluminum nitride
and others offers advantages with respect to many applications such as machining.
In the next section we shall briefly summarize another approach towards the design
of nanostructured superhard materials which has been successfully developed in
recent years.
4.2.2 Heterostructures
In a theoretical paper published in 1970 [75] Koehler suggested a concept for the
design of strong solids which are now called ‘heterostructures’ or ‘superlattices’.
Originally, he suggested depositing epitaxial multilayers of two different metals
M(l) and M(2) having as different elastic constants as possible, EM(2) > EM(,),
but a similar thermal expansion and strong bonds. The thickness of the layers
should be so small that no dislocation source could operate within the layers. The
dislocations then would form within a layer M(l) with the smaller elastic modulus.
If, under applied stress, the dislocation would move towards the M( 1)/M(2)
interface, the elastic strain induced in the second layer M(2) with the higher elastic
modulus would cause a repelling force which will hinder the dislocation movement
across the interface. Thus, the strength of such multilayers should be much larger
than that expected from the rule of mixtures.
His prediction was further developed and experimentally confirmed by Lehoczky
who deposited AI-Cu and Al-Ag heterostructures and measured their mechanical
properties [76,77]. According to the rule of mixture, the applied stress, ca,which
causes elastic strain, E , is distributed between the layers proportionally to their
volume fractions VMcX)and elastic moduli EM(.)
ga = (VM(l)Eh4(l) + VM(2)EM(2))&. (7)
Lehoczky has shown that the tensile stress-strain characteristics measured on
multilayers consisting of two different metals displayed a much higher Young
modulus and tensile strength, which both increased with decreasing thickness of
the double layer (‘layer period’). For layers thickness of 570 nm the yield stress of
Al/Cu multilayers was 4.2 times larger and the tensile fracture stress was 2.4 to 3.4
times larger than the values given by the rule of mixture.
This work was followed by a number of researchers who confirmed the experi-
mental results on various metallic multilayer systems (e.g. Cu/Ni [78-801 as well
as on hard epitaxial and polycrystalline superlattices of nitrides, such as TiNjVN
4.2 Concept f o r the Design of Superhard Materials 1 I5
[81], TiN/NbN [82,83], TiNV,Nbl -xN [84-861, Ti/TiN and WjWN [87], oxides
[88,89] and superlattices consisting of transition metal nitride and CN, [90]. In all
these cases, the hardness of the heterostructures increases with decreasing lattice
period, reaches a maximum of about 4000-5000kgmm-2 at 5-7nm and, after-
wards, decreases again.
The theoretical work on the heterostructures was summarized in several excellent
review papers [88,89,91]. Therefore, only a brief summary will be given here. For
a large lattice period, where the dislocation multiplication source can still operate,
the increase of the hardness and tensile strength (most researchers measured the
hardness because it is simpler than the measurement of tensile strength as done
by Lehoczky [76,77]) with decreasing layer thickness is due to the increase of the
critical stress needed to multiplicate dislocations, a, = Gb/lpp (Ipp is the distance
between the dislocation pinning sites [3]). Usually one finds strengthening depen-
dence similar to the Hall-Petch relationship, Eq. (9,but with a somewhat different
dependence on the layer period X-" (instead of the crystallite size, d , in Eq. ( 5 ) ) with
n = for layers with different slip systems and n = 1 for layers with a similar slip
systems [91]. In thin layers where no dislocation source can operate and the
Koehler's and Lehoczky's model applies, the maximum value of the critical shear
stress a,(X) which can cause a movement of a dislocation from the weaker layer 1
into the stronger layer 2 is given by [75-771
Here G M ( I ) are the shear moduli of the two materials, XI is the thickness of the softer
layer, b is the Burgers vector of the dislocation, and cp is the angle between the dislo-
cation slip plane in the layer M(1) and the interface M(l)/M(2). This theory predicts
that the strength (and hardness) depends mainly on the relative difference between the
+
shear moduli ( G M ( 2 ) - G M ( ~ ) ) / ( G M ( ~ ) G M M ( I ) )and the angle cp. For small period but
still X1 >> 46 the enhancement reaches an asymptotic value of
GM(2) - GM(l) GM(l) sincp
a,(max) = (9)
GM(2) + GM(l) 8.ir
The decrease of the hardness experimentally observed at small values of X < 5-
7nm is due to two effects: For XI smaller than about 15-20 times the Burgers
vector, b, the repelling mirror forces induced by the strain in the layer of stronger
materials on both sides of the layer M(l) begin to cancel out (these forces have
opposite sign with respect to each other) and, therefore, the enhancement of the
strength decreases. The other effect is a roughness of the interface due to the
interdiffusion during the preparation. These effects were discussed in a number of
papers ([82,92-951). In particular, Chu and Barnett [93] have shown that an
interface roughness of about l n m will cause a strong decrease of the hardness
and probably dominates the experimentally observed softening for X < 5-7 nm.
Recent experimental data on superlattices with a high immiscibility which form
sharper interfaces support this conclusion [94].
In a remark added in proofs, Koehler mentioned that the ideas described in his
paper should also be valid if one of the layers is amorphous. In spite of that note,
the researchers studied only the epitaxial and, later on, polycrystalline heterostruc-
tures. Only recently, several papers appeared in which one of the layers consist of
amorphous CN,, the other of a transition metal nitride such as TIN [88,96,97] or
ZrN 1901. However, with decreasing layer thickness the layered structure vanished
and a nanocrystalline composite structure appeared [90]. Such films also show a
high hardness exceeding 4000 kg mm-2.
In summary, significant progress has been achieved in the preparation and under-
standing of the hardness and strength enhancement in heterostructures and it is best
described in the recent theoretical papers by Chu and Barnett [93] and Anderson and
Li [94]. The application of heterostructures for cutting tools has been well developed
[98] and cutting tools with superior properties as compared to single layer coatings
are available. However, the relatively complicated preparation technique, which
requires a periodic change of the source for the subsequent deposition of the multi-
layers, makes the investigation into the novel superhard nanocrystalline composites
very attractive because they can be deposited in a simple plasma chemical vapor
deposition (CVD), plasma vapor deposition (PVD) or a combined PVD and
CVD process. Therefore, the remaining part of this chapter will concentrate on
these nanocomposites.
4.3 Preparation and Properties of Superhard

Nanocrystalline Composites
4.3.1 Preparation
As already discussed, the formation of a nanocrystalline/amorphous (or an ncjnc)
composite schematically illustrated in Fig. 4 and its thermal stability require a
high immiscibility of the components. This is fulfilled, for example, in systems
consisting of a stable transition metal nitride and silicon nitride. A simple estimate
shows that in such a system, the immiscibility is assured if the activity of nitrogen is
sufficiently high at the given temperature to shift the equilibrium of reaction (10) far
to the left hand side [63,73,74]
4mi-3
M,N + (m/3)Si3N4= Me,Si, + ___
6 N2.
For most of the stable nitrides, such as TIN and others, this applies at a nitrogen
pressure above about 1 mbar and temperature below 600°C. A similar estimate
shows that the segregation should be much stronger in the TiN/BN and many
other M,N/BN systems as compared to the TiN/Si3N4 one. It should also occur
in the TiN/AlN and many other systems.
We are primarily interested in superhard coatings for cutting tools, which have to
be deposited on appropriate substrates, such as cobalt cemented carbide or steel.
Thus, the preparation technique should assure a high nitrogen activity and a
sufficient activation energy for the thermodynamically driven segregation of the
4.3 Preparation and Properties of Superhard Nanocrystalline Composites 117
phases to occur at such a low temperature which is compatible with these materials.
The problem is mainly the steel, because the maximum tolerable temperature is
about 550°C or even lower for high speed steel. Therefore, an activated chemical
or physical vapor deposition technique which allows one to coat simultaneously a
large number of nonplanar substrates (i.e. a large total area) at a sufficiently high
deposition rate of 2 3 pmh-' (about 1 nms-') such as plasma CVD and PVD
appear to be most appropriate. Because plasma CVD usually operates at a some-
what higher pressure of about 1-5 mbar whereas a lower pressure of <lo-* mbar
is used in plasma PVD (reactive sputtering or activated reactive evaporation), the
plasma CVD was used in our initial work. It provides also a somewhat better
flexibility regarding the choice of the ncM,N/aA,Nb system. However, plasma
CVD still suffers problems with scaling from a small experimental reactor to a
large scale production units. Also the use of volatile transition metal halides as
sources of those metals makes it less attractive for industrial applications because
of the problems associated with the corrosive nature of the chlorine to the
vacuum pumps and the hydrophilic nature of the deposits at the reactor walls
and in the vacuum tubings. For these reasons, a recently developed combined
plasma PVD and CVD technique [99] is more appropriate for future industrial
applications whereas plasma CVD will remain the preparative tool for basic studies
on a variety of nc/a and nc/nc systems.
Both, high frequency (HF) [63,64,66,67,100] and direct current (d.c.)
[65,1101,102] glow discharge were used for the deposition. The apparatus used
for the deposition in an H F discharge has been described [63]. Here we shall briefly
describe the plasma CVD apparatus which operates under conditions of an
abnormal d.c. glow discharge.
A schematic of this apparatus is shown in Fig. 5 [65,102]. It consists of a reactor
made of Inconel 600 which is heated by means of an oven whose temperature is
controlled via a thermocouple (T.C.) inserted into the substrate holder and a
thermoregulator (not shown). Because the thermocouple is at the potential of the
cathode, 350-700 V, the whole thermoregulation system is electrically insulated
from the ground. The details of the gas manifold and of the pumping system are
obvious from the figure. The delivery of the metal halides (in Fig. 5 TiC14 for the
deposition of ncTiN/aSi3N4 is shown) occurs via a thermostatized evaporator or
bubbler. Typically, the total pressure is maintained at 1-5 mbar and the flow rates
of HZ,N2 and TiC14 are in the ratio of about 20 : (1-5) :0.1 sccm. The SiH4 flow is
adjusted between 0 and about 51 sccm depending on the required Si3N4 content
in the films. The substrates are mounted at the cathode of the d.c. discharge
which is operated in the abnormal glow regime [103-1051. In this regime, the
whole surface of the cathode is covered by a uniform glow because of the high resis-
tivity of the discharge region between the cathode and the negative glow. This glow
also follows fairly uniformly the surface of nonplanar substrates. For industrial
r
ap lications, a pulsed d.c. discharge with a repetition frequency of the order of
10 Hz and variable duty time is more appropriate in order to obtain a better control
of the deposition conditions, the substrate heating, and to avoid arcing. Of course, a
uniform discharge over nonplanar substrates can be maintained also in the H F
regime of 10-80 MHz if a sufficiently high power density is applied [106,107].
feedthreugh ,/ T'C* R=30kQ

HV-supply
Pirani
%-
\themostate
' XCI*
Figure 5. Schematics of the apparatus for the deposition of ncTiN/aSi3N4 coatings in an abnormal
direct current glow discharge (see text).
TiN prepared from TiC14 usually suffers from a relatively large chlorine content
which, for thermal CVD films deposited at 1000°C, can approach almost 1 at.%,
and may be much higher in films prepared by means of plasma CVD at lower tem-
peratures. Following our earlier studies about the shifts of the chemical equilibrium
in intense glow discharge plasmas [log, 1091we have shown that the chlorine content
of TIN deposited at 500°C can be significantly reduced by increasing the discharge
current density at the cathode to 2 1 mA cm-2 [ 1101. Using this principle and a large
excess of nitrogen and hydrogen, ncTiN/aSi3N4 films deposited at 500-550°C
contain 50.5 at.% of chlorine, which is fairly tolerable.
Another question is how to achieve the desirable stoichiometry, that is, a
sufficiently high activity of nitrogen, a(N2),in order to achieve the segregation in
reaction (10). The phase diagram of the ternary Ti-Si-N system at 1273K shows
that this is achieved at loga(N2) 2 -6 which corresponds to a nitrogen pressure
+
of lop6atm [71,73]. At a lower value of a(N2) the TiN, Si phases coexist and
+
for loga(Nz) 5 -7.5TiNx TiSi2 appear. The phase diagram becomes more
complicated at an even lower nitrogen activity, but this is not of any interest here.
The important conclusions are: First, for a nitrogen pressure of 2 1 mbar as used in
our experiments, fully stoichiometric TIN + Si3N4 phases should be formed at
550°C if the activation energy for N2 reaction with Ti and Si (i.e. a sufficient
4.3 Preparation and Properties of Superhard Nunocrystalline Composites 119
degree of dissociation) is provided by an intense discharge; second, decomposition

of Si3N4is expected upon a long term annealing at T 2 1100°C under vacuum with
a partial pressure of nitrogen of mbar [68,71,73,74]. This will be of impor-
tance later in this discussion.
Thus, any nonstoichiometry observed experimentally during the plasma CVD
must be due to either an insufficient activation of N2 or a preferential sputtering
of nitrogen from the films during the deposition. It is therefore not surprising that
in most cases to be reported here the films are fully stoichiometric and have shown
gold-like color typical of TiN as long as the Si3N4 content was below about
15mol.% 1671. A small substoichiometry which we observed at a relatively high
current density of 22.5 mA cm-2 might be due to preferential sputtering of nitrogen
during the deposition. When the Si3N4content is increased significantly above the
percolation threshold the color becomes darker (for Si3N4content of 230 mol.%)
and finally gray (>40mol.%) [67]. The small fraction of elemental silicon observed
by Li Shizhi et al. [lo11 and J. Patscheider et al. [ l l l ] was probably caused by
insufficient nitrogen activity during the deposition.
More recently, stoichiometric and superhard films of ncTiN/aSi3N4 and of
ncTiN/ncBN coatings were prepared by means of a chlorine free, combined
plasma PVD (vacuum arc evaporation of Ti) and CVD (silane and borazine used
as a source of silicon and boron, respectively) technique by the SHM company
[99] which is compatible with large scale industrial production. Also reactive
sputtering was demonstrated to yield similar very hard ncTiN/aSi3N4 [l 11,1121,
ncTiN,/CN,x [113], ncNbN/CN, [I 141, and TiC/aC [115] nanocomposites.
Voevodin and Zabinski reported the deposition of ncTiC/aC nanocomposites by
means of laser ablation [116,117]. Although the hardness of this material reached
only about 32 GPa, the coatings have shown a remarkably high toughness. Because
of the preliminary stage of this work it is too early to discuss details of these prepara-
tion techniques here. It should be just stated that this more recent development
shows that it is possible to prepare the superhard nanocomposites also by techniques
which are compatible with industrial large scale processing.
4.3.2 Properties of the ncM,N/aSi3N4 Composites

Figure 6 shows the dependence of the average TiN crystallite size in the ncTiN/
aSi3N4 composites prepared in an H F discharge [63] on the silicon content as
determined by energy dispersive analysis of X-rays (EDX). The squares correspond
to the results obtained from the X-ray diffraction using the Warren-Averbach
analysis [l IS] which in the range of the crystallite size of L1Onm yields the same
values as the Scherrer equation if the integral width of the Bragg reflection is used
[119].This is because the line broadening due to the random stress (50.5 GPa in our
films) is small compared to that due to the finite crystallite size. Only Bragg reflec-
tions from the f.c.c. lattice of TIN were found in the XRD [63,64,66, 1011 and elec-
tron diffraction study [61,62] investigations of all ncTiN/aSi3N4 composites. The
solid circles in Fig. 6 show the data obtained from direct lattice image in high reso-
lution transmission electron microscopy (HR-TEM) [62]. Both series of data show a
1 I
very good agreement. The somewhat larger value of the crystallite size determined
from the HR-TEM is probably due to the fact that the experimentalist easily over-
looks small crystallites in the HR-TEM micrographs, in particular if these are in the
shadow of the larger ones.
XPS studies have shown that the silicon is bonded as Si3N4[63,101] and is always
amorphous [61,62]. The available EDX does not allow us to measure the nitrogen
content. Thus, the content of silicon nitride [Si3N4]corresponds to (7/3)-times the
measured content of silicon [Si]. This scale is used in Fig. 7 which shows a similar
dependence of the average TiN crystallite size on the Si3N4 content for films
deposited in an abnormal d.c. glow discharge (see apparatus in Fig. 5) where the
surface of the growing film is exposed to energetic ion bombardment. The measured
hardness of the films is shown as well. One notices that the TIN crystallite size
initially decreases with increasing content of Si3N4,passes a minimum, and increases
afterwards again. The hardness shows exactly the opposite behavior reaching a
maximum at the Si3N4content of about 20mol.% when the TiN crystallite size
approached its minimum. Such behavior has also been found for other systems
studied so far, ncW2N/aSi3N4[64] and ncVN/aSi3N4 [66,67] and it appears to be
a universal phenomenon in these systems.
For the understanding of the development and stability of the nanostructure,
which will be discussed in the following section, it is important to realize that all
other deposition parameters, such as temperature, discharge current density, partial
pressure of all gases including TiC14 but the SiH4 (which was the only variable
parameter) are kept constant in this series of experiments. Thus, all these ncM,N/
aSi3N4 systems choose the minimum crystallite size at the Si3N4content of about
20 mol.% because this minimizes their total Gibbs free enthalpy.
[ 2000
01 I
0 10 20 30
a - Si,N, content [mol %]
Figure 7. Dependence of the average TIN crystallite size and of the hardness on the Si3N4content in
composites deposited in a d.c. discharge.
One might speculate whether the initial decrease of the crystallite size is a result of
a decrease of the TiN (or more generally M,N) growth rate because of a decrease of
the supply of the Ti gaseous source when the SiH4 flow is increased. However, this
can be almost certainly ruled out because the nucleation rate remains high and
almost constant since the large supersaturation changes only by less than I % in
these experiments when the small flow rate of SiH4 is varied between 0 and
5 1 sccm as compared to 20-100 sccm. of the total gas flow. This is supported by
the fact that the deposition rate remains fairly constant even if the Si3N4 content
is varied from 0 to almost 40mol.% (see Fig. 3a in [63]). Moreover, even if this
speculative explanation should hold, it can never explain the increase of the
crystallite size with increasing the Si3N4content above about 20 mol.%, which is
close to the percolation threshold in three dimensional f.c.c. system [120]. The
crystallite size of TIN should continuously decrease with increasing content of
Si3N4if the development of the microstructure were kinetically controlled.
Therefore, we conclude that the minimum crystallite size together with the
concomitant maximum hardness of the ncTiN/aSi3N4 composites is a result of a
(relative?) thermodynamic stability of such a nanostructure at the percolation
threshold (see remark added in proof and ref. [170]). We recall that at this percola-
tion threshold the HR-TEM and XRD data show that the nanostructure consists of
isolated TiN nanocrystals with a nearly spherical shape and only very few nanocrys-
tals touching each other [61,62], that is, there is no indication of bicontinous or an
interwoven bi-phase systems. Such a nanostructure with the minimum crystallite
size has a maximum specific area of the interface. If, as experimentally observed,
the system adjusts the minimum crystallite size at the percolation threshold resulting
in the maximum specific area of the interface, this interface must possess an unusual
stability. As such a behavior was also found with the other systems (ncW2N/aSi3N4
and ncVN/aSi,N4) it must be a result of some universal phenomenon. The possible

nature of this phenomenon will be discussed later.
The absence of dislocation activity in the nanocrystalline material, together with
the absence of grain boundary sliding, as discussed above, should result in brittle
behavior, so the measured hardness should be proportional to the elastic modulus.
Both the hardness (Figs 6 and 7) and modulus (not shown here) of the composites
were varied via the variation of the Si3N4content and the resulting hardness was
plotted against the elastic modulus. For all three systems a linear relationship was
found regardless of whether the deposition was done in the H F or d.c. discharge
[63,64,66,67,102]. These results lend further support to the universality of this
concept for the design of superhard nanocomposites.
The films also show a fairly high elastic recovery of 280% upon indentation (see
Fig. 2a and [64]), which is the result of the absence of dislocation activity and grain
boundary sliding. Such a property is of great importance for applications as protec-
tive coatings on machining tools, because a local stress which arises when the cutting
edge hits a grain of a hard precipitate in an alloy is distributed over a larger volume
of the coatings thus decreasing the local load and increasing the material's resistance
against fracture. A semiquantitative figure-of-merit is the ratio H" : Ey where the
exponents x and y depend on the geometry of the tool and the machining conditions
as well as on the theoretical model used for such an analysis [121]. Some researchers
prefer to use the ratio H 3 :E 2 . This ratio approaches the value of 0.9 in our nano-
composites and is (compared with other hard materials) relatively high.
In many applications the coating reaches a high temperature. For example, the
temperature of the leading edge of cutting tools can be more than 800°C. With
increasing cutting speed and decreasing consumption of coolant, which both
increase the economical performance of the machining, this temperature increases.
Therefore it is important to investigate the resistance of the coatings against oxida-
tion in air and the stability of the mechanical properties at high temperatures. In our
earlier study we found that the ncTiN/aSi3N4coatings are resistant to oxidation in
air up to about 8OO"C, which is significantly higher than that for T i c (500°C) and
TiN (600°C) and fairly comparable with Til -,Al,N [63,66,67].
The absence of creep and grain boundary sliding at room temperature does not
exclude its occurrence at elevated temperature (see the deformation mechanisms
maps [3,52-541). Because the equipment available for the measurements of hardness
cannot be used at such high temperatures, and other measurements of the
mechanical properties require much more time and labor, we have studied so far
only the stability of the nanostructure and hardness upon isochronal annealing of
the films in vacuum for 0.5 h at a given temperature. After each annealing of the
sample the X-ray diffraction pattern and the hardness were measured, and the
crystallite size was evaluated. Because such experiments are very time consuming,
only preliminary data obtained from eight samples with average crystallite size
chosen between about 3.5 and 9nm could be systematically studied. For films
deposited on a steel substrate the maximum temperature is limited to about
900°C because at 1000°C the films peel off.
In this series of measurements with three different samples with an average
crystallite size of about 3.5 nm, both the crystallite size and the hardness of about
4-1 F
1-
O0O0
-(u3
(I)
v)
4000
0 600 700 800 900
annealing temperature ["C]

(a)
4
61 r 6000
-....
-5000
E E
-4000 $
0 Y
0 -3000 ;
o-, o o o o-,
5
a,
0 -2000
m
+
Tdep 09.07.97, [Si] 10.3%, [Cl] 0.9%
H = 2650 kg/rnrn2, d = 3.4prn
silicon - substrate
'-1000
044 I ' I ' I ' I 0

0 600 800 1000 1200
annealing temperature ["C]

(b)
Figure 8. (a) Dependence on the crystallite size and hardness of ncTiN/aSi3N4films deposited on a
stainless steel substrate on the annealing temperature (0.5 h for each step). Note that after annealing
at 1000°C for 0.5 h the sample peeled off. (b) The crystallite size and the hardness of ncTiN/aSi3N4
films deposited on an Si wafer against the annealing temperature (0.5 h each step), see text.
5000 kgmm-* remained unchanged up to the highest temperature before the sample
peeled off. One example of such measurements is shown in Fig. 8a. For reasons
which are not fully understood yet [122] the hardness of the composites deposited
on Si-wafers is always somewhat lower. With this in mind we show in Fig. 8b
that for such samples both the crystallite size and the hardness do not change
upon isochronal annealing for 0.5h for each step up to 1100°C and only a small
coarsening is observed at 1150°C. Because Si3N4 decomposes under vacuum at
this temperature (saturation pressure of about lop6atm [74]), the absence of
coarsening (Ostwald ripening) is surprising. The lack of any change of the hardness
is just a consequence of the stability of the nanostructure. Nevertheless, this clearly
indicates that the grain boundaries and the ncTiN/aSi3N4interface possess a sur-
prising stability. In view of the fact that the maximum annealing temperature of
1100°Cwhere no coarsening occurred is about 0.7 Tdecat 1 atm of Si3N4,this finding
merits more detailed discussion (see below).
In contrast to these results, ncTiN/aSi3N4composites with a larger crystallite size
of 2 6 nm show recrystallization accompanied by a decrease of the hardness already
at 800°C. This suggests that the stabilizing effect of the ncTiN/aSi3N4 interface
increases with decreasing crystallite size.
4.3.3 Other Superhard Nanocomposites and the General Validity of

the Design Principle
In this section we shall discuss further nanocrystalline composites in which the
superhardness has been observed with emphasis on their practical applicability
for industrial applications and on the general nature of the design principle. Let
us start with the latter question.
The fact that the same development of the nanostructure and hardness exceeding
5000 kgrnmp2 with increasing content of Si3N4as shown in Figs 6 and 7 was found
also in the ncW2N/aSi3N4 system strongly supports the prediction that super-
hardness will be achieved in any other thermodynamically stable and hard transition
metal nitride in a combination with Si3N4.The reason is easily seen by considering
the thermodynamically driven segregation described by Eq. (10). WN is thermo-
dynamically unstable at >5OO0C and 30 bar of N2, and W2N decomposes under
vacuum above about 600°C [68]. Therefore, the occurrence of the segregation and
the superhardness in the ncW2N/aSi3N4system shows that it will occur also with
any other M,N, because for all thermodynamically stable metal nitrides with a
strongly negative Gibbs free enthalpy of formation the equilibrium of reaction
(10) will be shifted farther to the left hand side. This principle can be further
generalized for any other nitride system which meets these conditions, and it will
also apply to other systems, such as oxides, borides, carbides etc. showing the
segregation. This is of great importance because this large variety of the possible
materials and their combinations allows one to optimize the properties of the
coatings, such as hot hardness, toughness, friction coefficient, chemical stability in
air and against the material to be machined, as required for the given applications.
This is illustrated by the following example.
An apparent disadvantage of silicon nitride with respect to the applications of the
superhard nanocomposites for machining tools, is the solubility of silicon in many
materials including ferrous and aluminum alloys, superalloys [24], and others at
high temperatures. In view of the surprisingly high stability of the ncTiN/aSi3N4
composites one cannot exclude the possibility that ncM,N/aSi3N4 will show
sufficient stability against silicon dissolution in such applications. Nevertheless,
both the question of the applicability of the superhard composites as well as of
the universality of their design principle call for the investigation into further
ncM,N/(a and nc)A,Bb systems.
4.3 Preparation and Properties of Superhard Nunocrystalline Composites 125
The choice of boron nitride instead of Si3N4 is one of the most attractive
possibilities for several reasons: BN (in particular cBN) does not react with
ferrous-, aluminum-, and many other alloys and is fairly stable against oxidation
in air at elevated temperatures. Moreover, the equilibrium in the system described
by reaction (1 1) is shifted much more to the left hand side than in the reaction (10).
x+l
M,N + xBN 1 M,B, +-2
N2. (11)
This is due to the much higher stability of BN (AGf(1300K) = -68 kJ/mole and
atom) at high temperatures as compared with Si3N4 (AGf(1300K) = -44.9kJ/
mole and atom) [74]. The boron nitride matrix possesses also a high structural flexibil-
ity because of the coordinational flexibility of boron (threefold in hBN, fourfold in
cBN, and mixed in aBN). Furthermore, because TiB2 and many other transition
metal borides are refractory materials, also the ncM,N/(a and nc)BN interface
should have a high cohesive energy which is required for blocking the grain boundary
sliding and achieving the superhardness. These general considerations were recently
confirmed by experiments in which we have demonstrated that the hardness in the
ncTiN/(a and nc)BN coatings reached 7000 kgmm-2. This is important particularly
in comparison with the earlier results of other researchers to be briefly discussed here.
Using a conventional plasma PVD (e.g. sputtering) at a low pressure where such
conditions are not fulfilled leads to the formation of homogeneous phases (meta-
stable solutions), such as Til_,A1,N (see e.g. [123-1251) or Ti-B-N [126-1281.
Other authors have found phase segregation with the formation of binary phases
TIN and TiB2 in the Ti-B-N coatings with relatively little nitrogen [129-1361
corresponding to ‘region 4’ of the equilibrium phase diagram [73,131]. These appar-
ently contradictory results of different research groups can be probably attributed to
different plasma conditions at the surface of the substrate during the deposition, that
is, the above mentioned activation as well as to different film composition (mainly
nitrogen content). Although these data are not quite clear from the published papers
one can find there data supporting this hypothesis. Gissler, Mitterer, and coworkers
found that the deposited Ti-B-N films with a relatively low nitrogen content formed
two segregated binary phases TIN (or TiNl - J + TiB2 as expected from the equili-
brium phase diagrams [129,131-1361. The hardness of the films depended on the
composition showing a maximum for a composition of about Til B0.6p1N0,5 where
the coatings had a very fine, almost isotropic nanocrystalline microstructure [ 131-
1331. The value of the maximum hardness depended on the substrate temperature
and applied negative bias. For Tsubstrate of 3OW0O0C, ion energy of 126eV and
the ratio of the ion to atom flux to the substrate of 0.7, the hardness reached
about 50 GPa or slightly more for a nitrogen content of 5 15 at.% [134-1361. In gen-
eral, the hardness as a function of the composition did not follow the rule-of-mix-
ture but displayed a maximum at the given composition where the microstructure of
the films was very uniform (no columnar growth) and the crystallite size was 3-5 nm.
If the films were deposited at a nominal temperature of 20°C and post-annealed at
600-800°C, the hardness increased from 25-30 GPa to about 40 GPa after 30 min
[132]. Similar values of superhardness were obtained if the films were deposited at
about room temperature as ultrathin TiN/BN multilayers and post-annealed at
400°C for 20 h [131]. However, in spite of the significantly higher hardness as com-
pared with TiN, the performance of these films in cutting tests was slightly worse
than TIN and (TiA1)N coatings [ 131,132,1351. With increasing nitrogen content,
when the hexagonal BN (hBN) phase was formed, the hardness of the films strongly
decreased to 15-20 GPa for an overall composition of TiB2N2[135]. This decrease
was attributed to the formation of the soft hBN.
Andrievski has also found a strong hardness enhancement up to 52GPa in
TiBo.7No.5 thin films with a grain size of about 3 nm [137,138]. The microstructural
information regarding the possible segregation is unfortunately not available in his
papers. However, the development of the morphology of the films as a function of
the TiB2:TiN ratio (see Fig. 1 in [137]) is very similar to that found in our ncTiN/
aSi3N4and ncW2N/aSi3N4films (see Figs. 5 and 6 in [64]) thus suggesting that the
microstructure in his TiB2-TiN films corresponds to a similar two phase system.
In view of these results and the above discussed thermodynamic and kinetic con-
siderations we have chosen deposition conditions which promote both, the phase
segregation and the formation of cubic cBN which alone is ultrahard. This means
using a high plasma density, high deposition temperature and high ion bombard-
ment. The plasma CVD using halides, which can easily provide these conditions,
has the advantage of a large flexibility for fundamental research but is not the
best choice for large scale industrial processing (see the foregoing section). Therefore
a chlorine free, combined plasma CVD and PVD technique was developed for the
deposition of superhard ncTiN/BN coatings whose hardness reaches 70 GPa [99]
(see also remark added in proof and ref. [171]).
The list of possible superhard nitride nanocomposites can be further extended
including further immiscible systems. For example, the above mentioned metastable
solid solution Til -,Al,N coatings prepared by plasma PVD show a high stability
and corrosion oxidation at elevated temperature (see [123-125,1391 and references
therein). However, if deposited under conditions of a high activity of nitrogen, these
coatings will undergo segregation into the TIN and AlN phases which, under an
appropriate volume ratio, should also form a nanocrystalline superhard composite.
The TIN may act as template for the pseudomorphic growth of the metastable cubic
A1N [88,89,14&142] if the thickness of the TIN is sufficiently small. Similar effects
have been found also in the TiN/NbN, TiN/CrN, and TiN/CN, heterostructures
[88,89] and will surely be found in many others. Because CrN has similar tribo-
mechanical properties as TiN but a significantly higher oxidation resistance (see
e.g. [143,144] and references therein), nanocomposites based on this material in
combination with BN, AlN, Si3N4 or other immiscible nitrides have a large techno-
logical potential.
A very interesting template effect of TIN resulting in the development of a
superhard nanocomposite microstructure was reported by Sproul, Y.-W. Chung
and their co-workers for the TiN/CN, [96,97,145] and ZrN/CN, [90] systems.
The researchers were depositing heterostructures by unbalanced magnetron sputter-
ing from two opposing targets (Ti and graphite), between which the substrate was
mounted on a turntable. The substrate temperature was not controlled, but the
authors give 200°C as the upper limit. The thickness of the heterostructures was
varied by changing the rotation speed of the turntable. When the apparent ‘layer
4.3 Preparution and Properties of Superhard Nanocrystalline Composites 127
thickness’ (calculated from the deposition rate) decreased to 2 4 nm, the hardness
suddenly increased to almost 5000 kg mmP2 and the small angle X-ray diffraction
peaks from the layered structure vanished. The XRD showed a (1 11) texture of
the films and the HR-TEM micrographs revealed that the material consisted of
ncM,N/CN,x nanocomposite (M = Ti and Zr) with CN, pseudomorphic growth
on the M,,N template.
It is interesting to compare these TiN/CN, heterostructures and composites
prepared by reactive sputtering at a relatively low pressure of about 0.01 mbar
and temperature of 1200°C with the results obtained by plasma CVD at signi-
ficantly higher pressure of 1-5mbar and temperature of 560°C [67]. Because
carbon can substitute nitrogen in the TIN crystal lattice, TiC,N1-, forms a
homogeneous solid over the whole region of 0 < x < 1. The higher deposition
temperature and a high activation energy provided by the intense glow discharge
plasma during plasma CVD resulted in the formation of a well mixed titanium car-
bonitride TiC,Nl --I. The hardness of the films plotted against the composition
parameter, x, did not show any extremum, but followed monotonically the rule-
of-mixture increasing from the hardness of TIN of about 2000 kg mm-2 to that of
TIC of about 4000 kgmm-2 [67].
Knotek and coworkers [ 1151 reported on the preparation of ‘superstoichiometric
carbides’ MCI+x, which, in fact, consist of nanocrystals of stoichiometric carbides
imbedded in a aC:H matrix. With increasing excess, x,of the carbon, the hardness
of the films increases, reaches a maximum of about 4000 kgmmP2 and, afterwards,
decreases again upon a further increase of x in a similar manner as in our nano-
composites (Fig. 7). Similar results were recently reported by Voevodin and Zabinski
who used laser ablation for the deposition of ncTiC/aC films [I 16,1171. The maxi-
mum hardness of about 32GPa was found for the total carbon content of about
65at.% [116]. Although this is below the range of the ‘superhard’ materials, the
very high toughness of 70 N and low coefficient of friction of 0.15 [ 1171 are impressive.
The hardness of stoichiometric amorphous boron carbide of 25000 kg mm-2
[ 1461 exceeds significantly that of the crystalline B13C3 (often incorrectly written
as ‘B4C’ [147,148]) of about <3500kgmm-2. The fundamental unit of the crystal-
line BI3C3and of elemental boron is the B12boron icosahedron with a five-fold sym-
metry, which is incompatible with a closed packed three-dimensional crystalline
lattice. Therefore, the icosahedra are connected via bridging carbon (in BI3C3) or
boron (in cB) atoms [148]. The amorphous boron carbide consists also of the
boron icosahedra which are, however, randomly oriented and bridged. Thus,
aB,C (x M 4) can be considered as a composite consisting of randomly oriented
‘quasicrystals’ of the B-icosahedra which are imbedded in a carbon (or carbon
rich) amorphous matrix. Figure 9 shows the dependence of the hardness of aB,C
on the composition of the films [148,149].
The significant limitation regarding the applicability of these carbon containing
superhard nanostructured materials is the high solubility of carbon in ferrous
alloys (however, not in Al-alloys) and the low thermal stability of transition metal
and boron carbides against oxidation in air. The stability of BI3C3is limited by the
low melting point of the oxide of about 460”C, which allows a fast diffusion of
oxygen and resulting oxidation of the film. In the case of composites containing
[C]:[B] in the gas inlet

0 1 2 3 4
0 6
x in B,C(s)
Figure 9. Hardness of amorphous boron carbide deposited from diborane and methane by plasma
CVD against the composition (upper scale) and the C : B ratio in the gas inlet.
aC and CN, one has to keep in mind that these materials decompose at a tempera-
ture of 30O-50O0C, which limits their possible applications.
These examples show that superhard nanocomposites of various composition can
be prepared and their properties tailored according to the desired applications. Many
of these nanocomposites may have limited applications due either to their chemical
and structural instability at elevated temperature or their solubility (or chemical
reactivity) with the machined material. However, the stable nanocomposites, such
as ncTiN/aSi3N4,ncTiN/BN or more generally ncM,N/A,Nb will surely find inter-
esting applications. Preliminary results from the dry, hard- and high-speed turning
which are currently being done at our institute in collaboration with the SHM
company are very promising (for more recent results see [171]).
4.4 Discussion of the Possible Origin of the Hardness

and Stability of the Nanostructure
As already discussed the practical hardness (and strength in general) of a material is
determined by its microstructure. An appropriate design of the microstructure can
easily result in an increase of the hardness by a factor of 4-7. The experimentally
found superhardness of the ncM,N/aSi3N4 and ncTiN/BN nanocomposites of
5000-7000 kgmm-2 corresponds to an increase by a factor of 2.54, rather in the
lower range. Also the hardness of 10000-13 000 kgmm-2, which we often measure
on our composites, corresponds to an increase by a factor of 5-6.5. For the reasons
4.4 Discussion of the Possible Origin of the Hardness and Stability of the Nanostructure 129
outlined above, we do not consider these values of hardness as quantitatively

unambiguous. It is, however, difficult to dispute the measured value of the elastic
modulus of 1500 GPa (which was calculated from the linear part of the unloading
curve and corrected for the elastic deformation of the diamond indentor; see
Fig. 4 in [102]) as irrelevant because the linear part of the load-indentation-depth
dependence is equivalent to the linear part of any stress-strain characteristic, that
is, to the Hooke's law which defines the elastic modulus. Thus, the superhardness
of the nanocomposites is due to the nanostructure, which efficiently eliminates the
dislocation activity and grain boundary sliding. The question of the origin of the
superhardness is thus reduced to the question of the formation and stability of
the nanostructure and the high cohesive energy of the interface which avoids
grain boundary sliding.
The comparison of the ncTiN/CN, composites prepared by reactive sputtering
with the TiC,Ni, + I carbonitride films deposited by plasma CVD clearly shows
that, in this case, the nanocomposites are stabilized kinetically and they do not
form under conditions of intense plasma with a sufficient availability of activation
energy. On the other hand, the absence of Ostwald ripening in our nanocomposites
with the smallest crystallite size and largest value of the hardness (see Fig. 8)
strongly suggests that these materials may be thermodynamically stabilized. Such
a hypothesis is surely provoking because, to best of our knowledge, there are no
known grain boundaries with a negative Gibbs free energy of formation. The struc-
tural disorder within the grain boundaries and the incommensurability of the grain
boundaries consisting of materials with a different crystal structure results in an
excess lattice energy due to strain [150] and a positive Gibbs free energy of the
grain boundaries, that is, lower thermodynamic stability of the nanocrystalline
metals [151, 1521. As a result, all these systems show coarsening (Ostwald ripening)
upon annealing to a temperature where the diffusion becomes fast enough, which is
typically T 2 OST,,,, (Tm,"is the melting or vaporization temperature). Let us
therefore discuss the possible explanation of the absence of the Ostwald ripening
in our superhard nanocomposites as described above.
The Ostwald ripening is a thermodynamically driven process due to a difference in
chemical potential,
&A, = 2vM 7 0 (rl - r 2 ) / r 1 r2 (12)
( V , is the molar volume and "yo is the specific interfacial Gibbs free energy) between
particles of different radii rl > r2. The smaller the particle, the higher is its Gibbs
free energy of formation [I 53,1541. Upon the Ostwald ripening the system decreases
its total Gibbs free energy due to the growth of larger particles at the expense of the
smaller ones. In the quantitative treatment of the rate of this process two limiting
cases can be distinguished: Either the transport (diffusion between the particles)
or the reaction at the phase boundary is rate limiting.
Let us first assume that the observed structural stability of the nanostructure, the
absence of Ostwald ripening, is due to a slow diffusion [153,154]. In other terms, we
assume, in contradiction to the experimentally found percolation behavior of these
systems (see Figs. 6, 7, Fig. 10 in [155], and the discussion above), that the nano-
structure is thermodynamically unstable and it forms due to the lack of diffusion
during the deposition of the films. Consequently, the experimental result that
Ostwald ripening does not occur during the 30min annealing at 1100°C means
that the mean diffusion length of Ti in Si3N4must be much smaller than the thick-
ness of the aSi3N4tissue for the films at the percolation threshold, with the smallest
crystallite size and highest hardness. From the Einstein formula [153,154]
(x), = J207 (13)

we can estimate the upper limit of the diffusion coefficient D(1000°C)at the annealing
temperature of 1100°C assuming (in accord with the absence of the Ostwald ripening)
that during the annealing time 7 = 30 min the average diffusion distance (x), is much
smaller than the average thickness of the aSi3N4tissue, about 0.5 nm. Thus, we obtain
as an upper limit for the diffusion coefficient of Ti in aSi3N4 at 1100°C a value of
D(1100"C) 5 10-22-10-20cm2s-1. Such a value is far too small for any system
reported in the literature. As a rule, the lower limit of the coefficient for bulk diffusion
in crystals at a temperature T = 0.57', is of the order of >10-15-10-'4cm2 s-' and
much larger (lO-'cm s
the order of ci2 i" for the grain boundary diffusion [23,156]. A value of
was reported for the selfdiffusion of nitrogen in TIN
(N15 exchange technique) at 0.5Tdecomp in [157] and 2 x lo-'' cm2 s-* for the interdif-
fusion between TIN and Si couple at 1100°C [158]. Because these values are many
orders of magnitude larger than the above calculated upper limit of the diffusion co-
efficient D(1100°C) 5 lop2' cm2s-' we conclude that the experimentally observed
absence of the Ostwald ripening due to a slow diffusion can be with certainty ruled
out.
Consequently, the reaction at the nc/a-interface must be rate controlling. This
means that the titanium atoms do not dissolve in the amorphous Si3N4 tissue.
Because many metals, such as Ni, Au, Ag, Al, Ga, and others diffuse easily through
Si3N4 at elevated temperatures, and in view of the large number of vacancies in
aSi3N4 at 1100°C (which is close to 0.77'") the lack of dissolution of Ti must be
due to the stability of the ncTiN/aSi3N4 interface. This is in accord with the
observation of the percolation threshold discussed above. Moreover, the increase
of the TiN-crystallite size beyond the percolation threshold means that the thermo-
dynamic driving force, which is causing the maximum specific area of the ncTiN/
aSi3N4 interface at the percolation, decreases with increasing thickness of the
Si3N4tissue in that range.
The results of annealing experiments with samples deposited either with a smaller
or a higher Si3N4 content, that is, having a larger crystallite size of 6-7nm, show
that such films coarsen already at a temperature of about 800°C. This result is par-
ticularly important for the sample with a higher Si3N4content of about 40mol.%,
because the diffusion length of Ti is larger than in the stable films (Si3N4content of
about 17-23 mol.%). If the systematic study, which is presently being done at our
institute, confirms these results, there will be hardly any doubt left that the ncja
interface of the material with the minimum crystallite size and maximum hardness
is thermodynamically stabilized. With this limitation in mind let us hypothesize
what could be the physical origin of such stabilization.
The strong increase of the hardness occurs at a TiN-crystallite size of 545 nm
and thickness of the aSi3N4-tissue of <1 nm, that is, in the range where surface
4.4 Discussion of the Possible Origin of’the Hardness and Stability of the Nunostructure 131
0 . 0 1 4 4 4
.-K 0.008-
z 0.006-
0.004-
0.002-
20 40 60 80 100 120
crystallite size [A]
Figure 10. Dependence of the measured tensile strain on the crystallite size for ncSi and ncTiN/
aSi3N4,see text.
energy strongly influences the mechanical and thermal properties (e.g. lattice
dilatation or contraction [159,160] and a decrease of the melting point [160]) and
the quantum localization phenomena dominate the optoelectronic properties of
solids [ 155,1601. In thin films of nanocrystalline materials lattice dilatation has
been found to increase with decreasing crystallite size for nanocrystalline silicon
[159] as well as for ncTiN [63] and ncW2N [64]. These data are reproduced in
Fig. 10. The lattice dilatation in films which do not have any significant biaxial
stress has been explained due to the density defect in the grain boundaries and
the anharmonicity of the interatomic potential [161]. (Our ncTiN/aSi3N4 films
have a very small compressive stress of 0.2-0.5 GPa [ 1621. Energetic ion bombard-
ment during the deposition can increase this stress to 5GPa [163].) Because the
elastic lattice energy increases with the second power of the strain [22] it
represents a destabilizing contribution to the total free energy of the system
regardless whether it is compressive or tensile. Thus, there has to be some stabilizing
contribution. Considering the Gibbs free enthalpy of the nanocomposite given by
Eq. (14), the only possibility seems to be to assume that the stabilizing effect is
associated with a high cohesive energy Eintof the ncM,N/aSi3N4 interface
A G j ( d ) = A G $ ( C O ) + + N ~ T E E-
* ~N~r d 2 ( E $ +A&,,). (14)
Here, A G j ( d ) and AG;(co) are the Gibbs free energy of the nanocrystalline and
coarse grained systems, respectively, the second term describes the destabilizing
effect of the lattice strain (see Fig. 10) and the last term the stabilizing effect of
the grain boundaries. N is the number of nanocrystals per mole, E$ > 0 is the
increase of the cohesive energy (dependent on d ) and is the entropy term of
the interface.
The decrease of the surface energy of the ionic transition metal nitride by ‘wetting’
it with aSi3N4 may account for the initial decrease of the crystallite size with
increasing silicon nitride content, but it cannot explain the increase of the crystallite
size at high Si3N4 content above the percolation threshold. Therefore, Eint must
reach a maximum stabilizing effect at the minimum crystallite size and the thickness
of the Si3N4interface at the percolation into separate ncTiN and aSi3N4 phases.
With a further increase of the Si3N4 content this stabilizing effect must decrease
and vanish, that is, E i t 5 E i y ) for d > d(min) M 3-3.5nm. Thus, for larger
crystallite size of d > 5-8nm the destabilizing effect of the grain boundaries
dominates and the thermodynamically driven Ostwald ripening (Eq. (12)) occurs.
We can not offer any complete theoretical explanation of this phenomenon now,
but only a hypothesis which may explain the experimentally observed results.
The possible explanation is associated with the changes of the band structure due
to quantum confinement phenomena in the nanocrystals and resultant increase of
the cohesive energy at the interface. Indeed, a strong increase of the band gap
was theoretically calculated [164,1651and experimentally observed in silicon [1661
as well as in compound semiconductors [160]. In metallic nanoclusters deposited
on a carbon substrate, the conduction band narrows and the Fermi level shows a
downshift [160,167]. This means, that the bonds which are formed at the nc/a
interface by a combination of the states near the Fermi level of the nanocrystalline
transition metal and the upper edge of the valence band of the 5 1 nm thin Si3N4
should also show downshift with decreasing crystallite size as long as the aSi3N4
tissue is thin. In such a case the system gains stability by increasing the specific
area of that interface. However, when the fraction of Si3N4 increases above the
value corresponding to the percolation threshold, its thickness increases and
the band gap decreases towards the bulk value. Thus, the stabilizing effect at the
nc/a-interface is lost and there is no driving force which would maximize its area.
Consequently, the crystallite size increases again. The observed minimum of the
crystallite size at the percolation threshold corresponds to a minimum of the total
free energy of the system, that is, to a balance between this stabilizing effect at the
interface and the destabilizing increase of the elastic lattice energy associated with
the strain.
This hypothesis, of course, needs further theoretical development and experi-
mental verification. One may imagine other effects which could explain the one or
the other observation. For example, the stability of the nanostructure might be
explained by a strong immiscibility of the TIN and Si3N4phases in a similar way
as it is believed to apply for heterostructures prepared of materials which show
high immiscibility [89,95,168,169]. However, such an explanation is in contra-
diction with the phase diagram of the TiN/Si3N4 at high temperature and low
pressure of nitrogen. Furthermore, it cannot explain why the ncTiN/aSi3N4 films
with a large crystallite size coarse at much lower temperature and why, during the
deposition, the crystallite size passes a minimum at the Si3N4 content
corresponding to the percolation threshold. For similar reasons also a strain
induced decrease of the diffusion in the ncTiN/aSi3N4 is unlikely to play any
important role.
Of course, one should keep in mind that the high structural stability of the nano-
structure at high temperature is not necessary condition for the occurrence of the
superhardness. Any phenomenon which avoids the grain boundary sliding will
4.4 Conclusions 133
result in the superhardness of nanostructured materials. Recently, we have found

the superhardness also in ncTiN/aSi3N4 films with crystallite sizes of about 6-
10 nm, when they were prepared under different plasma conditions (lower current
density and lower voltage at the substrates during the deposition). These films,
although superhard, recrystallize at about 800°C. We mention these observations
here just to emphasize that there may be also other phenomena which avoid the
grain boundary sliding, at least at room temperature. One example is the template
effect which was observed, for example, in the ncTiN/ncAlN and ncTiN/ncCN,
systems mentioned above. Any such effect can easily result in an enhancement of
the strength and hardness by a factor of 4-8 and thus in superhardness of nano-
composites (ncja as well as ncjnc) with a crystallite size of the order of lOnm or
less. The above discussed explanation of the high stability against coarsening at
elevated temperatures applies probably only to ncja nanocomposites, where the
percolation behavior as seen in Figs 6 and 7 is found. The high thermal stability
of such nanocomposites may lead to a high hot hardness and toughness whereas
that of superhard nanocomposites with d = 6-10 nm or more may deteriorate due
to the thermally activated grain boundary sliding. Future research will, hopefully,
help to clarify these questions.
4.4 Conclusions
The practical strength and hardness of materials can be increased by an appropriate
control of the microstructure, which hinders the multiplication and movement of
dislocations and growth of microcracks. One such possibility is the decrease of
the grain size which leads to a decrease of the dislocation activity and of the size
of microcracks and an increase of the strength and hardness by a factor of 3-8.
However, a new deformation mechanism, grain boundary sliding, becomes
important in nanocrystalline materials when the grain size decreases below about
10-20 nm. Grain boundary sliding results in a softening of nanocrystalline metals
typically for a grain size less than 6-10 nm.
Grain boundary sliding can be avoided by an appropriate design of nanocrystal-
line composites, such as ncTiN/aSi3N4, ncVN/aSi3N4, ncW2N/aSi3N4,ncTiN/BN,
and others (Si3N4and BN content of 17-23 mole.%), which have a small crystallite
size of 3-9 nm and a high cohesive energy of the ncja interface. These nanocompo-
sites show a high elastic recovery of 280% and Vickers microhardness exceeding
5000 kg mm-*. Presently available data indicate that these nanostructures are
thermodynamically stabilized and do not show Ostwald ripening up to a tempera-
ture of 1100°C. Films with either a larger or smaller content of Si3N4 than that
corresponding to the percolation threshold of about 20 mole.% have a larger
crystallite size, smaller hardness and show coarsening (Ostwald ripening) by
800°C. A hypothesis based on a strengthening of the interface due to quantum
confinement phenomena is suggested to explain these observations.
For many practical applications, such as tribological coatings of machine tools
for ferrous metals, aluminum alloys, superalloys, and others, nanocomposites which
do not contain silicon (which is soluble in such alloys), new ncM,N/a or ncA,Bb
nanocomposites were suggested and their preparation investigated. Among
others, ncTiN/a and ncBN nanocomposites, which were prepared by a combined
plasma PVD and CVD technique without any halides, showed a very high hardness
of 6000-7000 kg mm-2.
Some theoretical aspects were discussed to show the generality of the principle
for the design of superhard nanocomposites. The present review was complemented
by a brief summary of the recent development in the field of heterostructures
which represent another approach to the design of superhard nanostructured
materials.
Acknowledgments
I should like to thank the coworkers from my institute for their enthusiastic
collaborations, my colleagues and friends Profs. Li Shizhi, H. Warlimont, V.
Vitek, and many others for illuminating discussions, and my wife Dr. Maritza
Vepiek-Heijman for her valuable comments on the manuscript. I am very much
indebted to Drs. M. Jilek, M. Sima, and P. Holubar from the SHM Company for
a very fruitful collaboration towards the industrialization of our superhard nano-
composite coatings. Partial financial support of our work by the German Science
Foundation (DFG) and the Bavarian Ministry for Education and Science is
thankfully acknowledged.
References
1. D. Tabor, The Hardness of Metals, Clarendon Press, Oxford, 1951.
2. Ch.-M. Sung and M. Sung, Muter. Chem. Phys. 1996,43, 1-18.
3. R. W. Hertzberg, Deformation and Fracture Mechanics of Engineering Materials, 3rd edn.,
Wiley, New York, 1989.
4. M. L. Cohen, Phys. Rev. B, 1984, 32, 7988-7991.
5. A. Y. Liu and M. L. Cohen, Science, 1989,245, 841-842.
6. M. L. Cohen, Solid State Commun. 1994, 92, 45-52.
7. M. L. Cohen, Mater. Sci. Eng. A , 1996, 209, 1 4 .
8. S. Vepfek, J. Weidemann, and F. Glatz, J . Vac. Sci. Technol. A , 1995, 13, 2914-2919.
9. Y. Zhang, Z. Zhou, and H. Li, Appl. Phys. Lett. 1996,68, 634-636.
10. Y. Zhang, private communication (1997).
11. Y . Chen, L. Guo, and E. G. Wang, Mod. Phys. Lett. B, 1996, 13, 615-617.
12. Y. Chen, L. Guo, and E. G. Wang, J . Phys.: Condens. Muter. 1996,8, L685SL690.
13. Y. Chen, L. Guo, and E. G. Wang, Philos. Mag. Lett. 1997, 75, 155-162.
14. E. C. Cutiongco, D. Li, Y.-W. Chung, and C. S. Bhatia, J . Tribol. 1996, 118, 543-548.
15. A. Khurshudov, K. Kato, and S. Daisuke, J . Vac. Sci. Technol. A , 1996, 14, 2935-2939.
16. X. Yun, R. C. Hsiao, and D. B. Bogy, IEEE Trans. Magn. 1997, 33, 938-943.
References 135
17. H. Sjostrom, S. Stafstrom, M. Boman, and J.-E. Sundgren, Phys. Rev. Lett. 1995, 75, 1336-
1339; see also errata 1996, 76, 2205.
18. H. Sjostrom, L. Hultman, J.-E. Sundgren, S. V. Hainsworth, T. F. Page, and G. S. A. M.
Theunissen, J . Vac. Sci. Technol. A , 1996, 14, 56-62.
19. W. T. Zheng, H. Sjostrom, I. Ivanov, K. Z. Xing, E. Broitman, W. R. Salaneck, J. E. Greene,
and J.-E. Sundgren, J. Vac. Sci. Technol. A , 1996, 14, 2695-2701.
20. G. B. Bokii, N. F. Kirova, and V. I. Nepsha, Sov. Phys. Doklady, 1979, 24, 83-84.
21. A. F. Goncharov, Sov. Phys. Uspechy, 1987, 30, 525-534.
22. Ch. Kittel, Introduction in Solid State Physics, Wiley, New York, 1971.
23. E. Hornbogen and H. Warlimont, Metallkunde, Springer, Berlin, 1996.
24. A. Kelly and N. H. MacMillan, Strong Solids, 3rd edn., Clarendon Press, Oxford 1986.
25. H.-H. Behnke, Harterei Technische Mitteilungen, 1993,48, 3-10.
26. M. F. Doerner and W. D. Nix, J . Muter. Res. 1986, 1, 601-609.
27. T. F. Page and S. V. Hainsworth, Surf. Coat. Technol. 1993,61, 201-208.
28. W. C. Oliver and G. M. Pharr, J . Muter. Res. 1992, 7, 613-617 and 15641583.
29. K. G . Schmitt-Thomas, Metallkundejiir das Maschinenwesen, Vol. I & 11, Springer, Berlin, 1990.
30. E. 0. Hall, Proc. Phys. SOC.B, 1951, 64, 747-753.
31. N. J. Petch, J . Iron Steel Inst. 1953, 174, 25-28.
32. A. Lasalmonie and J. L. Strudel, J . Muter. Sci. 1986, 21, 1837-1852.
33. N. Hansen, Metall. Trans. A , 1985, 16A, 2167-2190.
34. U. F. Koks, Metall. Trans. 1970, 1, 1121-1143.
35. A. Inoue, H. M. Kimura, K. Sasamori, and T. Matsumoto, Muter. Trans. Jpn. Inst. Metals,
1994,35, 85-94.
36. H. Hahn and K. A. Padmanabhan, Philos. Mag. B, 1997,76, 559-571.
37. R. W. Siegel, Spektrum der Wissenschaft, 1997, 3, 62-67.
38. R. W. Siegel and G. E. Fougere, Muter. Res. Soc. Symp. Proc. 1995, 362, 219-229.
39. R. W. Siegel and G. E. Fougere, NanoStructured Muter. 1995, 6 , 205-216.
40. J. E. Carsley, J. Ning, W. W. Milligan, S. A. Hackney, and E. C. Aifantis, NanoStructured
Muter. 1995, 5, 441448.
41. A. H. Chokshi, A. Rosen, J. Karch, and H. Gleiter, Scripta Metall. 1989, 23, 1679-1684.
42. G. W. Nieman, J. R. Weertman, and R. W. Siegel, Scripta Metall. 1989, 23, 2013-2018.
43. J. S. C. Jang and C. C. Koch, Scripta Metall. Muter. 1990, 24, 1599-1604.
44. K. Lu, W. D. Wie, and J. T. Wang, Scripta Metall. Muter. 1990, 24, 2319-2323.
45. G. W. Nieman, J. R. Weertman, and R. W. Siegel, Scripta Metall. Muter. 1990, 24, 145-150.
46. G. W. Nieman, J. R. Weertman, and R. W. Siegel, J . Muter. Res. 1991, 6, 1012-1027.
47. G. E. Fougere, J. R. Weertman, R. W. Siegel, and S. Kim, Scripta Metall. Muter. 1992, 26,
1879-1 883.
48. R. Z. Valiev, F. Chmelik, F. Bordeaux, G. Kapelski, and B. Baudelet, Scripta Metall. Muter.
1992,27, 855-860.
49. A. M. El-Sherik, U. Erb, G. Palumbo, and K. T. Aust, Scripta Metall. Muter. 1992,27, 1185-
1188.
50. (a) P. G. Sanders, C. J. Youngdahl, and J. R. Weertman, Muter. Sci. Eng. 1997, A234, 77-82;
(b) P. G. Sanders, J. A. Eastman, and J. R. Weertman, Acta Muter. 1977, 45, 40194025;
(c) C. J. Youngdahl, P. G. Sanders, J. A. Eastman, and J. R. Weertman, Scripta Muter.
1997,37, 809-813.
51. J. Musil and F. Regent, J . Vac. Sci. Technol. 1998, A16, 3301-3305.
52. M. F. Ashby, Acta Metal. 1972, 20, 887-897.
53. H. Luthy, R. A. White, and 0. D. Sherby, Mafer. Sci. Eng. 1979, 39, 211-216.
54. F. A. Mohamed and T. G. Langdon, Metall. Trans. 1974,5,2339-2345.
55. J. Crampon and B. Escaig, J . Am. Ceramid Soc. 1980, 63, 686686.
56. R. C. Gifkins, Metall. Trans. 1976,7A, 1225-1232.
57. J. W. Edington, Metall. Trans. 1982, 13A, 703-715.
58. S . Yip, Nature, 1998, 391, 532-533.
59. J. Schiotz, F. D. Di Tolla, and K. W. Jacobsen, Nature, 1998, 391, 561-563.
60. W. W. Milligan, S. A. Hackney, M. Ke, and E. C . Aifantis, NanoStructured Muter. 1993, 2,
267-276.
136 4 Nunostructured Superhard Materials
61. S. Vepiek, S. Christiansen, M. Albrecht, and H. P. Strunk, Muter. Res. SOC.Symp. Proc. 1997,
457,407412.
62. S. Christiansen, M. Albrecht, H. P. Strunk, and S. Vepfek, J. Vac. Sci. Technol. B, 1998, B16,
19-22.
63. S. Vepfek and S. Reiprich, Thin Solid Films, 1996, 268, 64-71.
64. S. Vepiek, M. Haussmann, and S. Reiprich, J. Vuc. Sci. Technol. 1996, A14, 4 6 5 1 .
65. S. Vepfek, Proc. of the 10th Znt. Conf. on Thin Films, Salamanca, Spain, September 1996, Thin
Solid Films, 1998, 317,449-454.
66. S. Vepiek, M. Haussmann, and Li Shizhi, Electrochem. SOC.Proc. 1996,96-5,619-628.
67. S. Vepfek, M. Haussmann, S. Reiprich, Li Shizhi, and J. Dian, Surf. Coat. Technol. 1996,86/
87, 394401.
68. G. W. Samsonow, Nitridi, Naukova Dumka, Kiev, 1969 (in Russian).
69. H. Holleck, J. Vuc. Sci. Technol. A , 1986, 4, 2661-2669.
70. H. Fischmeister and H. Jehn, Hurtstofschichten zur Verschleissminderung, DGM-Informa-
tionsgesellschaft, Stuttgart, 1987.
71. S. Sambasivan and W. T. Petuskey, J. Muter. Res. 1994, 9, 2362-2369.
72. T. Hirai and S. Hayashi, J. Muter. Sci. 1982, 17, 1320-1328; 1982, 17, 3336-3340; 1983, 18,
240 1-2406
73. P. Rogl and J. C. Schuster (Eds.), Phase Diagrams of Ternary Boron Nitride and Silicon Nitride
Systems, ASM International, Materials Park, OH, 1992.
74. M. W. Chase, C. A. Davies, J. R. Downey, D. J. Fruip, R. A. McDonald, and A. N. Syverud,
JANAF Thermochemical Tables, 3rd edn., Phys. Chem. Re$ Data, 1985, 14, No. 1.
75. J. S. Koehler, Phys. Rev. 1970, B2, 547-551.
76. S. L. Lehoczky, J. Appl. Phys. 1978,49, 5479-5485.
77. S. L. Lehoczky, Phys. Rev. Lett. 1978, 41, 1814-1818.
78. C. A. 0. Henning, F. W. Boswell, and J. M. Cobert, Acta Metall. 1975, 23, 193-197.
79. R. F. Bunshah, R. Nimmagadda, H. J. Doerr, B. A. Movchan, N. I. Grechanuk, and E. V.
Dabizha, Thin Solid Films, 1980, 72, 261-275.
80. S. Menezes and D. P. Anderson, J. Electrochem. SOC.1990, 137, 440-444.
81. U. Helmersson, S. Todorova, S. A. Barnett, J.-E. Sundgren, L. C. Markert, and J. E. Greene,
J. Appl. Phys. 1987, 62, 481-484.
82. M. Shinn, L. Hultman, and S. A. Barnett, J. Muter. Res. 1992, 7, 901-911.
83. X. Chu, M. S. Wong, W. D. Sproul, S. L. Rohde, and S. A. Barnett, J. Vac. Sci. Technol. 1992,
A10, 1604-1609.
84. K. M. Hubbard, T. R. Jervis, P. B. Mirkarimi, and S. C. Barnett, J. Appl. Phys. 1992,72,4466-
4478.
85. P. B. Mirkarimi, L. Hultman, and S. A. Barnett, Appl. Phys. Lett. 1990, 57, 2654-2656.
86. P. B. Mirkarimi, S. A. Barnett, K. M. Hubbard, T. R. Jervis, and L. Hultman, J .Muter. Res.
1994,9, 1456-1467.
87. K. K. Shih and D. B. Dove, Appl. Phys. Lett. 1992,61, 654-656.
88. W. D. Sproul, Surf. Coat. Technol. 1996,86/87, 170-178.
89. W. D. Sproul, Science, 1996, 273, 889-892.
90. M. L. Wu, X. W. Lin, V. P. Dravid, Y. W. Chung, M. S. Wong, and W. D. Sproul, J. Vac. Sci.
Technol. 1997, A15, 946950.
91. S. A. Barnett, in Physics of Thin Films, Vol. 17, Mechanics and Dielectric Properties, M. H.
Francombe and J. L. Vossen (Eds), Academic Press, Boston, 1993, p. 2.
92. J. E. Krzanowski, Scripta Metall. Muter. 1991,25, 1465-1470.
93. X. Chu and S. A. Barnett, J. Appl. Phys. 1995, 77,44034411.
94. S. A. Barnett, Invited Lecture presented at the TSM Annual Meeting, San Antonio, Texas,
February 1998 unpublished (see also [168, 1691).
95. P. M. Anderson and C. Li, Nunostructured Muter. 1995,5, 349-362.
96. D. Li, X. Chu, S. C. Cheng, X. W. Lin, V. P. Dravid, Y. W. Chung, M. S. Wong, and W. D.
Sproul, Appl. Phys. Lett. 1995,67,203-205.
97. D. Li, X. W. Lin, S. C. Cheng, V. P. Dravid, Y. W. Chung, M. S. Wong, and W. D. Sproul,
Appl. Phys. Lett. 1996, 68, 1211-1213:
98. T. L. Selinder, M. E. Sjostrand, A. Ostlund, M. Larsson, S. Hogmark, and P. Hedenqvist,
References 131
Proc. 14th Int. Plansee Seminar, G. Kneringer, P. Rodhammer, and P. Wilharditz (Eds), Plan-
see AG, Reutte, Austria, 1997, p. 44.
99. S. Vepiek, P. Nesladek, A. Niederhofer, F. Glatz, M. Jilek, and M. Sima, 25th Int. Conf.
on Metallurgical Coatings and Thin Films, San Diego, April 1998, Surf. Coat. Technol. 1998,
108-109, 138-147.
100. S. Vepiek, S. Reiprich, and Li Shizhi, Appl. Phys. Lett. 1995, 66, 2540-2642.
101. Li Shizhi, Shi Yulong, and Peng Hongrui, Plasma Chem. Plasma Proc. 1992, 12, 287-297.
102. S. Vepiek, A. Niederhofer, P. Nesladek, and F. Glatz, Electrochem. SOC.Proc. 1997, 97-25,
3 17-324.
103. A. von Engel, Ionized Gases, 2nd edn., Clarendon Press, Oxford, 1965.
104. S . C. Brown, Introduction to Electrical Discharges in Gases, Wiley, New York, 1966.
105. G. Francis, Ionization Phenomena in Gases, Butterworth, London, 1960.
106. S. Vepfek, J. Th. Elmer, Ch. Eckmann, and M. Jurcik-Rajman, J. Electrochem. SOC.1987,136,
2398-2405.
107. S. Vepiek, Ch. Eckmann, and J. Th. Elmer, Plasma Chem. Plasma Proc. 1988,8,445466.
108. S. Vepi-ek, C. Brendel, and H. Schafer, J. Cryst. Growth, 1971, 9, 266-272.
109. S. Vepiek, in Current Contents in Materials Science, E. Kaldis (Ed.), North-Holland, Amster-
dam, 1979, pp. 151-236.
110. J. Patscheider, Li Shizhi, and S. Vepiek, Plasma Chem. Plasma Proc. 1996, 16, 341-363.
11 1. M. Diserens, J. Patscheider, and F. Levy, 25th Int. Conf. on Metallurgical Coatings and Thin
Films, San Diego, April 1998, Surf. Coatings Technol. 1998, 108-109, 241-246.
112. F. Vaz, L. Rebouta, S. Ramos, M. F. da Silva and J. C. Soares, 25th Int. Conf. on Metallurgical
Coatings and Thin Films,San Diego, April 1998, Surf. Coat. Technol. 1998, 108-109,236240.
113. H. Jensen, J. Sobota, and G. Sorensen, Surf. Coat. Technol. 1997, 94/95, 174178.
114. H. Jensen, J. Sobota, and G. Sorensen, 25th Int. Conf. on Metallurgical Coatings and Thin
Films, San Diego, April, 1998, Surf. Coat. Technol. unpublished.
115. 0 . Knotek, private communication, 1997.
116. A. A. Voevodin and J. S. Zabinski, J . Muter. Sci. 1998, 33, 319-327.
117. A. A. Voevodin and J. S . Zabinski, Diamond Rel. Mater. 1998, 7, 463467.
118. H. P. Klug and L. E. Alexander, X-Ray DzfSraction Procedures, Wiley, New York, 1974.
119. M. Riickschloss, PhD Thesis, Technical University, Munich, 1993.
120. R. Zallen, The Physics of Amorphous Solids, Wiley, New York, 1983. See chapter 4, in parti-
cular Figure 4.5for a two dimensional network and Table 4.4. Notice the following important
facts: ( I ) The structural percolation corresponds to the situation when the percolating phase
(here Si3N4)forms a continuous connection (‘cluster’) through the matrix phase (here TIN). It
occurs at a critical fraction of the percolating phase (here aSi3N4)Pc which depends on the
nature (site or bond percolation), structure see e.g. the f.c.c., b.c.c., and diamond lattices in
Table 4.4) and dimensionality of the network (here TIN). Pc = 1 for one dimensional
chain, 0.354.7 for two and 0.124.4 for three dimensional network. For the f.c.c lattice of
TiN Pc N 0.2 for the site percolation as observed in the present work. (2) Because of different
near-threshold scaling laws, the percolation phenomena of electrical and optical properties,
such as conductivity and optical reflection, commence only at a fraction of the percolating
phase which is significantly larger than P, (see e.g. Figure 4.5 and Table 4.3 in [120] and
the change of the reflectivity of the ncTiN/aSi3N4 films reported in [67]). (3) Even above the
percolation threshold (P(Si3N4)> Pc) there still may be some local connectivity within the
matrix (nc-TiN). In our particular case it means, that some of the TIN nanocrystals can be
in contact with each other.
121. T. E. Fischer, private communication, 1998.
122. So far, only films deposited on Si show a lower hardness. Films deposited on hard metal (Co
cemented tungsten carbide) or on such a substrate precoated with TiN or Ti, -.Al,N show the
same or somewhat larger hardness as those deposited on steel. For both steel and silicon sub-
strates the crystallite size and the Si3N4content are equal. The possible reasons of this effect
are under investigation.
123. 0. Knotek, M. Bohmer, and T. Leyendecker, J. Vac. Sci. Technol. 1986, A4, 2695-2700.
124. T. Leyendecker, 0. Lemmer, S. Esser, and J. Ebberink, Surf. Coat. Techno/. 1991, 48, 175-
178.
125. Y. Tanaka, T. M. Giir, M. Kelly, S. B. Hagstrom, T. Ikeda, K. Wakihira, and H. Satoh, J .

Vac. Sci. Technol. 1992, A10, 1749-1756.
126. 0.Knotek, R. Breidenbach, F. Jungblut, and F. Loffler, Surf. Coat. Technol. 1990,43144. 107-
115.
127. 0. Knotek and F. Loffler, J . Hard Muter. 1992,3, 29-38.
128. 0.Knotek, A. Schrey, J. W. Schultze, and B. Siemensmeyer, WerkstofSKorros. 1992,43,511-
519.
129. C. Mitterer, M. Rauter, and P. Rohdammer, Surf. Coat. Technol. 1990,41, 351-363.
130. M. Tamura and H. Kubo, Surf. Coat. Technol. 1992,54155,255-260.
131. W. Gissler, Surf. Coat. Technol. 1994, 68/69, 556563.
132. P. Hammer, A. Steiner, R. Villa, M. Baker, P. N. Gibson, J. Haupt, and W. Gissler, Suyf.
Coat. Technol. 1994, 68/69, 194-198.
133. T. P. Mollart, M. Baker, J. Haupt, A. Steiner, P. Hammer, and W. Gissler, Surf. Coat. Tech-
nol. 1995, 74/75, 491496.
134. P. Losbichler, C. Mitterer, P. N. Gibson, W. Gissler, F. Hofer, and P. Warbichler, Surf. Coat.
Technol. 1997,94195, 297-302.
135. W. Gissler, M. A. Baker, J. Haupt, P. N. Gibson, R. Gilmore, and T. P. Mollart, Diamond
Films Technol. 1997, 7, 165-170.
136. C. Mitterer, P. Losbichler, F. Hofer. P. Warbichler, P. N. Gibson, and W. Gisler, Vacuum,
1998,50, 313-318.
137. R. A. Andrievski, J . Solid State Chem. 1997, 133, 249-253.
138. R. A. Andrievski, J. Muter. Sci. 1997, 32, 4463-4484.
139. J. G. Han, J. S. Yoon, H. J. Kim, and K. Song, Surf. Coat. Technol. 1996,86-87, 82-87.
140. I. Petrov, F. Adibi, J. E. Greene, L. Hultman, and J.-E. Sundgren, Appl. Phys. Lett. 1993,63,
3638.
141. F. Adibi, 1. Petrov, J. E. Greene, L. Hultman and J.-E. Sundgren, J. Appl. Phys. 1993, 73,
858G8589.
142. (a) M. Setoyama, A. Nakayama, T. Yoshioka, T. Nomura, A. Shibata, M. Chudou, and
H. Arimoto, Surnitomo Electric Industries, 1995, 146, 91-95; (b) M. Setoyama, A. Nakayama,
M. Tanaka, N. Kitagawa, and T. Nomura, Surf. Coat. Technol. 1996,86-87,225-230.
143. G. Berg, Ch. Friedrich, E. Broszeit, and Ch. Berger, Surf. Coat. Technol. 1996,86-87,184-191;
1996,3, 1597-1602.
144. J. Vetter, R. Knaup, H. Dwuletzki, E. Schneider, and S. Vogel, Surf. Coat. Technol. 1996,86-
87, 739-147.
145. Y.-W. Chung, Surf. Rev. Lett. 1996,3, 1597-1602.
146. S. Vepfek and M. Jurcik-Rajman, Proc. of 7th Int. Symp.on Plasma Chem., C. J. Timmermans
(Ed.), Eindhoven University of Technology, Eindhoven, 1985, pp. 90-94.
147. G. Will and K. H. Kossobutzki, J. Less Common Metals, 1976, 47, 4348.
148. N. N. Greenwood and A. Earnshaw, Chemisty of the Elements, Pergamon Press, Oxford, 1984.
German translation, VCH-Verlag, Weinheim, 1990.
149. S. Vepfek, Plasma Chem. Plasma Proc. 1992, 12, 219-235.
150. H. P. Aubauer and H. Warlimont, Z . Metallkunde, 1974, 65, 297-303.
151. F. Gartner, R. Bormann, R. Birringer, and A. Tschope, Scripta Muter. 1996, 35, 805-
810.
152. I recommend reference 11511for more details because it shows that some of the earlier papers
predicting a higher stability of the nanocrystalline metals were incorrect.
153. H. Schmalzried, Chemical Kinetics of Solids, VCH, Weinheim, 1995.
154. H. Schmalzried, Solid State Reactions, Verlag Chemie, Weinheim, 1981.
155. S. Vepfek, Thin Solid Films,1997, 297, 145-153.
156. D. Gupta and P. S. Ho (Eds), Diffusion Phenomena in Thin Films and Microelectronic
Materials, Noyes, Park Ridge, NJ, 1988, p. 43.
157. Hj. Matzke, in Difusion in Metals, A. L. Laskan, J. L. Bocquet, G. Brebec, and C. Monty
(Eds), Kluwer, Dordrecht, 1990, pp. 429455.
158. W. Wakelkamp, PhD Thesis, Technical University, Eindhoven, 1991.
159. S . Vepfek, Z. Iqbal, and F.-A. Sarott, Philos. Mag. B, 1982,54, 137-145.
160. A. P. Alivisatos, Science, 1996, 271, 933-937.
References 139
161. S. Vepiek, F.-A. Sarott, and Z. Iqbal, Phys. Rev. B, 1987, 36, 3344-3350.
162. P. Nesladek and S. Vepiek, unpublished results, 1998. See also [170].
163. S . Vepfek, in Proc. of Muter. Res. Soc. Europe, Strasbourg, 1984, P. Pinnard and S. Kalbitzer
(Eds), Les Cditions de physique, Les Ullis, France, 1984, pp. 425444.
164. C. Delerue, M. Lannoo, and G. Allan, J . Lumin. 1993, 57, 249-256.
165. B. Delley and E. F. Steigmeier, Phys. Rev. B, 1993, 47, 1397-1400.
166. S. Vepiek and Th. Wirschem, in Handbook of Optical Properties Vol. ZZ: Optics of Small
Particles, Znterfaces and Surfaces, R. E. Hummel and P. Wissmann (Eds), CRC Press, Boca
Raton, 1997, pp. 129-143.
167. S. B. DiCenzo and G. K. Wertheim, in Clusters of’ Atoms and Molecules ZZ, H. Haberland
(Ed.), Springer Verlag, Berlin, 1994, pp. 361-383.
168. S. Barnett and A. Madan, Physics World, 1998, 11, 4548.
169. A. Madan, Yun-yu Wang, S. A. Barnett, C. Engstrom, H. J. Ljungcrantz, L. Hultman, and
M. Grimsditch, J. Appl. Phys. 1998, 84, 776-779.
170. A. Niederhofer, P. Neslidek, H. Mannling, S. Vepfek and M. Filek, Surf. Coatings Technol.
(1999) in press (Proc. Int. Conf. Metallurgical Coatings and Thin Films, San Diego, April
1999).
171. P. Holubai, M. Jilek and M. Sima, Suyf. Coatings Technol. (1999) in press (Proc. Int. Conf.
Metallurgical Coatings and Thin Films, San Diego, April 1999).
172. A. Voevodin, ibid.
Remarks added in proofs.

Since the submission of the manuscript in 1998 many new publications appeared
about the industrialization of the (TiA1)N/aSi3N4nanocomposites [ 1711, thermal
stability [ 1701 of the ncTiN/aSi3N4 and nc(TiAC)N/aSi3N4, on the ultra-tough
nanocomposites ncTiC/aC and ncTiC/aC + MoS2 [ 1721 and others. Microhardness
of 80 to 2 l00GPa which is in the range of diamond, was achieved in ultrahard
nanocomposites consisting of nc-TiN/a-Si3N4/a-2nc-TiSiz [ 1701 when the surface
of the TIN nanocrystals is covered with about one monolayer of ‘Si3N4’[170].

K. G. Nickel and Y. G. Gogotsi
5.1 Introduction
Corrosion may be defined as the physical and chemical alteration of a material due to
its interaction with the environment of interest. It must be emphasized that corrosion
resistance is not a material property but a system property and real environments have
a high variability in both chemical and physical boundary conditions.
Corrosive alteration leads to a change in the desired original properties and so
corrosion usually, but not exclusively, has adverse effects and causes a decrease in
strength, toughness, hardness, and increases wear. There are many reasons for the
degradation of properties by corrosion. Decrease of the cross-section carrying the
load, pitting, stress induced by phase changes, and grain separation are some of
the common features.
Some hard materials, namely Sic and Si3N4,are designed to serve at high tem-
peratures, where corrosion resistance is an essential feature. This is the reason
why a large proportion of the ceramic literature on corrosion, including recent
textbooks [ 1-51, deals with these materials.
In order to react with the environment, the material must not be in equilibrium
with it. The hard materials covered in this chapter are non-oxides and hence this
condition is fulfilled for all of them in almost every application environment, includ-
ing ambient air. The applicability of hard materials thus rests on the kinetics of the
processes.
Before discussing individual materials we will introduce important terms and
concepts in corrosion like attack modes, media classes, location of attack, kinetics,
and their parameters. We feel that this is necessary for understanding of the dis-
cussion of the materials, because there is no generally accepted standard procedure
that allows us to compare the corrosion behavior of materials as easily as other
properties like mechanical strength or hardness.
5.2 Corrosive Media

The compositional variability of environments is so vast that only very few attempts
have been made to classify corrosion behavior from the point-of-view of the discrete
environment at hand. A noteworthy exception is the Dechema handbook [6],in
which valuable information may be found if the corrosive agent is simple and
known (e.g. acetates, chlorine, steam). However, its main thrust is on the corrosion
of metals.
5.3 Corrosion Modes 141
The more common classification scheme is to divide the corrosive media into their
state of aggregation, that is to subdivide into corrosion by solids, liquids and gases.
While solid state corrosion is rarely dealt with, we have vast amount on literature on
hot gas corrosion. The case of corrosion by liquids is commonly further subdivided
into more specific cases, such as aqueous corrosion (e.g. acids and water), corrosion
by glasses, metal melts and salt melts. The last case is for historic reasons known in
the form of a rather misleading expression: hot corrosion. A special case, which
spans from the liquid into the gaseous state is given by the corrosion in hot water
systems: hydrothermal corrosion.
At first glance this may not make too much sense, but the experience is that the
different names mentioned above correspond to main forms of the attack mode and
often refer to different application temperature levels.
5.3 Corrosion Modes

5.3.1 Active and Passive Corrosion
A fundamental distinction between corrosion modes is the division into active and
passive corrosion. The active mode is characterized by the loss of material to the
environment and results in the decrease of size and weight of the specimen. The
loss may be in form of gaseous or dissolved species.
The passive mode denotes a process, where the material reacts with the environ-
ment to yield a new condensed phase, often in form of a layer or scale on the surface.
In most cases it is associated with a weight and/or component size gain.
In the strict sense, the passive mode implies that the newly formed phase is retard-
ing the process, that is corrosion is slowed down with time (protective scale), but
there are cases in which the scale is nonprotective: a scale with cracks, low viscosity,
or foamy texture may not hinder the access of the corrosive agent to the substrate.
5.3.2 Homogeneity and Location of Attack: Internal, External and

Localized Corrosion
If there is a perfectly homogeneous surface attack and a removal of material or a
scale growth is taking place solely and uniformly on the surface of the material,
the process is denoted external corrosion.
This is most likely if we have a perfect single-phase material. As soon as grain
boundaries in general or secondary phases appear as discrete particles or grain-
boundary phases the process is likely to become more complex, because each
phase of the assemblage will react differently to the environment at hand.
An easily envisaged form of the problem is the attack of oxygen on a S i c contain-
ing discrete free carbon particles. At high temperature the carbon reacts in an active
mode (formation of CO,T), while the Sic forms condensed Si02.Inhomogeneous or
142 5 Corrosion of Hard Materials
local corrosion with bubble and pit formation is the result, at least in the early stages
of this oxidation [7].
If the secondary phase is a grain-boundary phase, which allows a relatively fast
transport of the corroding agent into the material via the grain boundaries, we
will have an attack of the main material and/or the grain-boundary phase not
only on the surface, but in a zone stretching out into the bulk of the specimen.
This is known as internal corrosion.
Despite its localized character on a microscopic scale, intergranular corrosion
may be quite homogenous on the macroscopic scale. Additionally, the process of
internal corrosion is also capable of slowing down with time, so it may offer some
protective quality.
It must be emphasized that those processes do not exclude each other. Internal
corrosion may occur while a scale is formed on the outside. We will discuss the
problem in the context of the individual materials.
5.4 Corrosion Kinetics

5.4.1 Physical Boundary Conditions
In modeling the kinetics of corrosive processes and quoting numbers of kinetic
constants it is usually inherently implied that we have steady-state conditions
with a constant activity of the corrosive agent in the environment and an infinite
reservoir of the material to be corroded.
The first condition is met in many technical applications, where the corrosive
medium is flowing, the second one may be true or not. But even the first condition
is violated for some corrosion conditions, such as when a substance is deposited on
the surface of the material by a physical process and can hence be treated as an
attack by a substance which was implanted at a given amount without replenishing.
Exact corrosion kinetics must be modeled by solving the second law of Fick for
the geometry of the case at hand. However, in some cases a net effect may be
calculated from simple thermodynamics, as for closed system conditions in active
corrosion [8]. For the case of diffusion through scales it has been demonstrated
that quasi-steady-state modeling is often a good approximation for an exact
solution, at least for conditions t D / x 2 > 2 [9] (where t = time, D = diffusivity,
x = layer thickness). Some basic solutions for situations with instant singular
corrosion can also be found in the literature [lo].
But even under steady-state conditions there is a profound influence of physical
boundary conditions on corrosion behavior. The most widely known example of
this is the boundary between active and passive oxidation of silica-formers. The
classic modeling has been done by Wagner (1 11 for silicon.
The line of argument is as follows. Active oxidation of silicon occurs by the
formation of gaseous SiO and passive oxidation by the formation of silica:
Si + 402H SiOt (1)
and
Si + O2 H Si02.
Calculations of the oxygen partial pressure P(0,) for the boundary condition both
by simply taking the values from the JANAF-tables [12] or by using a modern
thermochemical program [ 131 yield similar results with an equilibrium P(0,) as
low as bar at 1000°C. The prediction from the equilibrium would thus be,
that at conditions with P ( 0 2 ) > lop2*bar we should find the formation of a silica
layer on a bare Si surface. Experimental observation shows that this only takes
place at P(02)-levels more then 20 orders of magnitude higher.
Wagner’s solution was to transform the problem into a physically controlled one.
Silica is formed at conditions, which allow its condensation from the gas phase in
equilibrium with the substrate:
4Si + Si02 * SiO. (3)
If reaction (1) is fast, a concentration gradient of O2 inwards and SiO outwards must
exist. Then the interdiffusion coefficients and the effective thickness of the boundary
layer will control whether the critical P(Si0) of the condensation reaction (3) is
reached or not. From tables and estimations of those physical constants we can
relate back to the oxygen pressure of the bulk gas necessary to induce the critical
P(Si0) on the surface. The calculation yielded reasonable results for the active-
passive boundary of Si in streaming atmospheres with low oxygen contents and
accordingly the theory was later applied to other silica-formers [ 14,151.
This success story demonstrates that equilibrium calculations disregarding physi-
cal boundary conditions can be misleading, and that thermochemical calculation is
capable of giving good results if those boundaries are taken into account.
5.4.2 Active Corrosion Kinetics

Under steady-state conditions the loss of material to the environment by decom-
position into gaseous species or dissolution is usually a reaction or a diffusional
problem. For reaction control we have the flux J as
J = krci, (4)
where J = mass flux, k , = reaction coefficient, ci = concentration of agent at inter-
face, and for first-order reactions a linear law follows under steady-state conditions.
The basic shape of the kinetics for diffusion control can be seen by looking at
Fick’s first law
J= -D-dc
dl ’
where I = diffusion path length. With constant conditions, the gradient from the
surface to the bulk of the corrosive medium becomes constant, 1 stays constant
and accordingly the flux, J , becomes constant with time. Therefore the change in
10
- 5
h
f
Y
c
.-0
In
In
a
0
0
2
v
-8
-5
0 1 2 3 4 5 6 7 8
log (time [s])
Figure 1. Predicted behavior of a model case for active corrosion under different external conditions.
size or mass (Ax) is also linear with time t:

AX = kit. (6)
This expected linearity is usually experimentally confirmed. A theoretical prediction
of the size of the h e a r rate constant kl in E q . (6) for a given temperature and
composition of medium and material is possible, when thermodynamic data are
available to calculate the reaction, and boundary parameters such as the geometry
of the sample, viscosity, stream velocity, Schmidt and Reynold’s numbers are
known or can be estimated [16].
Other approaches to active corrosion prediction utilizing thermochemical calcu-
lations [8,17,18] require the experimental determination of effective parameters.
They show both the importance of physical boundary conditions and the extremely
low level of partial pressures at which active corrosion is potentially dangerous.
In Fig. 1 a model case applying such simplified kinetic equations [17] is illustrated.
Thermodynamic calculations for Sic in H2 yield partial pressures of about lo-’ bar
for SiH4 and CH4 even at moderate temperatures of 1 100°C, so that the reaction
Sic + 4H2 w SiH4 + CH4 (7)
is used as the dominant reaction. Geometric boundary conditions (a tube of 1 m
length with an inner diameter of 30 cm) allow us to calculate a volume flow through
such a pipe from a given stream velocity. The calculated material recession after a
given time ranges from less than a monolayer in the closed system to more then
70mm per year for ideal, and hence gas velocity independent, evaporation. The
model for intermediate cases indicates that both mild and conservative applicability
limits for high speed civil transport systems [18] are not met even at low gas
velocities.
It is interesting to note that different approaches [17,18] result in applicability

limits, which specify partial pressure limits of <lop7 bar to be tolerable in applica-
tions with strong security requirements in the active corrosion mode unless reaction
control retards the process.
5.4.3 Basic Passive Corrosion Kinetics

Passive corrosion with scale formation on the substrate is often more complicated.
The simplest form is that of reaction control. If a reaction is slow compared to the
delivery of the agent to the reaction site, the concentration of the agent at the
interface is as high as the external concentration of the corrosive medium. For a
given temperature this reaction rate constant, k , dictates then the linear rate con-
stant k~ in Eq. (6) and linear kinetics prevail. Only this time Ax is positive and
denotes a layer growth or a mass gain.
Likewise Eq. (6) is correct, if the chemical reaction rate constant is high (= fast
reaction) and the growing scale is not hindering the agent from access to the
substrate. This situation is most easily envisaged in a scale with plenty of short-
circuits like cracks or open porosity. Here the concentration gradient develops
from the substrate into the bulk of the environment. Accordingly the analysis is
identical to that for active corrosion kinetics above, which gave the linear kinetics.
Another simple case is a homogeneously growing single-phase layer, which allows
only a slow diffusional transport of corrosive medium to the substrate. Here the
steady state means for Eq. (5) that a layer, and hence the diffusion path length 1,
is growing with time while the absolute concentration difference Ac stays constant.
At the substrate-oxide interface, the equilibrium with the material is achieved or the
concentration of the agent can be approximated to zero, while the concentration at
the surface of the scale is constant at the solubility limit of the scale material.
The law appropriate for this situation as derived from Fick's first law is called the
parabolic law due to diusional control and introduces the parabolic rate constant k,,
which has units of m2 s-' or kg2 m-4 s-' , depending on the definition of Ax as scale
growth or mass gain:
(ax)2
= k,t (8)
or
ax= fi. (9)
The derivation of the parabolic law above is only a good approximation for a situa-
tion with a sufficiently thick pre-existing scale. It has been shown in a classic paper
by Deal and Grove [19] on the oxidation of silicon, that the correct parabolic law
should recognize the chemical reaction rate and the gas transport coefficient along
with the diffusion coefficient. Their analysis demonstrated that the parabolic rate
law is rather
+ axA = B ( t +).
(ax)2 (10)
with a linear ‘short-term’-solution for t << T and a parabolic ‘long-term’ solution

identical to Eq. (8) for t >> T ( A , B, T = constants). It should be noted that all
those constants vary with temperature and that the so called ‘help time’, T , can be
of appreciable length: For silicon at 1700°C we have a time of at least several
tens of hours, during which the kinetics are basically linear! A nice review of the
complexity of the ‘simple’ parabolic oxidation of Si, particularly in its initial
stages, has been given by Irene [20].
A third basic law for passive corrosion comes from the analysis of the situation,
where the reaction product formed on the material is completely blocking the
further access of the corrosive agent. This law was derived from the study of
metals at moderate temperatures. It is easily envisaged: If every particle formed
protects the substrate completely, we have a situation, in which the surface area
available for reactions is diminished rapidly. The mathematical form following is
the logarithmic law
with the logarithmic rate constant klog.The same equation is valid for a more
complex situation, pertinent to ceramic systems. If the diffusion through a scale is
rate controlling and the scale crystallizes a phase, which is either completely block-
ing for the agent or has a diffusion coefficient much smaller than the original scale,
the effective cross section for the agent transport is likewise diminished. For a con-
stant crystallization rate we will end up with Eq. (1 1).
However, natural nucleation and crystal growth is complex, dependent on many
variables and usually not constant with time. The energy requirements are often so
strongly temperature dependent that we see either none or full crystallization. Thus
in pure systems we will often have kinetic breaks with time or temperature rather
-
h
u)
._
c
=I 300
P
2
c
e
2.200
ln
u)
Q)
c
Y
.-0
100
Jt>.
-m
0
0 50 100 150 200 250 3 I0
Time (arbitrary units)
Figure 2. Basic passive laws.
then a nice curve following Eq. (11) exactly. This may be different for systems in
which phase formation or crystallization is linked to chemical changes of the
scale, induced by diffusion of components from the interior.
The shapes of curves with behavior according to the basic laws is illustrated in
Fig. 2. It is obvious from Fig. 2 that data from a single point (here: t = 28 and
x = 100) would not allow prediction of the behavior at any other time unless the
law is specified.
5.4.4 Kinetic Breaks

The basic laws of Section 5.4.3 are for constant growth under constant conditions.
However, a scale growing according to the parabolic law may change to a different
behavior only after some time has elapsed.
There are a number of reasons for such kinetic breaks, shown in Fig. 3 . Again
the most easily envisaged case is a crystallizing scale. Crystallization is often
accompanied by a volume expansion, which may induce stresses in the scale and
lead to spalling. The renewed fast access of the medium to the substrate accelerates
reaction rates. If spalling is connected to a critical thickness, we find pseudo-linear
growth built from individual parabolic segments (Fig. 3). If the first break is not
repaired by the following growth, for example if the scale is not spalling but
permanently cracks, we find break-away behavior, where parabolic behavior
changes to linear.
We may also find breaks which slow down corrosion, such as if a scale crystallizes
only after some critical thickness or time is reached without disintegration. A
parabolic-logarithmic behavior may follow. Note that this break is not apparent
to the naked eye unless very long times of observation are involved.
Figure 3. Common kinetic break types.

*Oo0 I
n 1500-
=!
cu
P
Y
Y 1000-
.-0
5
5
- 500-
0 20 40 60 80 100 10
Time (a.u.)
Figure 4. Growth change due to developing diffusion coefficients, D.
5.4.5 Complex Kinetics

The real corrosion behavior may be far more complex then the ideal laws suggest,
because of changes in properties of the scale with growth and the operation of
differing types of corrosion at the same time.
Changes in the properties may be microstuctural changes like the development of
bubbles, pits, or internal porosity, where effective surfaces or effective diffusion path
lengths are changed. Other common changes are those of the effective diffusion
coefficient. If a scale is not built from a phase of constant composition like the
SiOz formed on Si, but contains elements from external impurities or internal
secondary phases, which are concentrated or diluted with time, the diffusion co-
efficient will change with these changes. Accordingly the overall behavior must
deviate from parabolic behavior.
Such changes are illustrated in Fig. 4. Note that the increase of D with time makes
the shape more linear with time, while the decrease in D brings it to a curve more like
the logarithmic law.
The development of scales on highly porous media has been reported to show
a typical break of growth speed with increasing temperature. The model case
shown in Fig. 5 involves indeed no true kinetic break, the parabolic rate constant
increases regularly with temperature and is identical for short and long times. The
accepted reason for the shape of the 'high-T' curve in Fig. 5 is the pore closure with
the associated breakdown of the effective surface. A typical factor is of the order of
300 if a sample with a geometric surface of 1 cm2 has a specific surface of 0.1 m2/g-'
in the porous state.
A relatively simple analytical treatment for complex growth patterns has been
suggested by Nickel [21]. It works on the assumption that the basic laws of
30
Time (a.u.)
Figure 5. Typical development of layer growth with increasing temperature in porous media with
pore closure due to passive corrosion.
growth can operate simultaneously. Growth according to

Ax = k,t + + klogIog(t)
is evaluated by stepwise multiple linear regression. The stepwise procedure was
adopted to give fit constants with more physical meaning. The fit allows quantifica-
tion of individual contributions of the basic laws and is illustrated in Fig. 6.
Other attempts to model complex kinetics have been put forward by Ka11 et al.
[22]. Both complex modeling methods do not incorporate true kinetic breaks and
thus need improvement in order to be used as a general growth law.
0 500 I000 1500

Time (a.u.)
Figure 6. Model for complex growths kinetics after Eq. (1 1) [21] with individual contributions.
- 4
?
m
Y
% 2
c
m
c
0 0
E
.-0)
a
3 -2
-4
0 20 40 60 80
Time (a.u.)
Figure 7. Weight change curves obtained from the simultaneous action of parabolic weight gain and
two different linear weight loss rates (‘para-linear behavior’).
An important special case of complex kinetics is the simultaneous action of basic

passive laws together with active corrosion. A typical case for this is when a scale
grows but is consumed at the same time by evaporation. The resulting shape of
curves is shown in Fig. 7 and has been described as ‘para-linear’ [23] behavior. It
may be analyzed with Eq. (12) omitting the logarithmic term.
It should be noted that small variations in the linear loss rate constant give
macroscopically different behavior: the curve labeled ‘data 1’ in Fig. 7 might
wrongly be interpreted as ‘hardly any change after a short period and ‘data 2’ as
‘active corrosion after an incubation period’.
5.5 Corrosion Measurement

5.5.1 Experimental Methods
The most common type of corrosion experiment is the after-treatment test, that is
investigations on property changes are done after the end of the test by any of a
variety of means (weight, dimensions, penetration, phase change, retained strength
etc.). The limitations are discussed in the section on corrosion data (5.5.2).
For the corrosion of ceramics by ordinary liquid media, the testing is usually done
by immersion tests: the sample is placed in a heated retort or autoclave inside a
stirred excess of corrosive medium. Common test conditions are at the boiling
point of the medium (e.g. 10% H2S04, 30% NaOH) for a week.
The corrosion by melts is classically studied by the methods of the refractory
industry like drop-, crucible-, and finger tests [24]. While those methods allow an
easy ranking they are hardly capable of producing quantitative data outside the
narrow conditions set by the standardized experimental set-up [25].
For hot corrosion testing burner rigs can be applied [26].The agent is injected into
the flame of the burner and the sample is held or agitated at a distance, which allows
deposition of the corrosive agent from the gas phase at a constant rate.
In the ordinary furnace test for gas corrosion the samples are placed in a furnace
for a given time in the atmosphere of interest. The advantage of the furnace test is
that many different samples can be tested in one set of conditions, the drawback that
it gives only one data point in time per experiment. The pitfalls are apparatus
related: Temperature and atmosphere constancy have to be guaranteed, cross-
contamination must be avoided, and interaction with furnace linings and crucible
material minimized. The danger of those points is often underestimated; there are
reports of different results from a mere change in the furnace size [27].
Contamination by furnace linings is a source of error for all corrosion experi-
ments. There is ample evidence that the commonly used linings of A1203-ceramics
do contaminate samples with A1 and alkalis via gas phase transport, even if high-
purity A1,03 is used. The contamination is strongly enhanced in wet atmospheres
[281.
In situ measuring of corrosion is mostly done by thermogravimetric measure-
ment, that is the monitoring of a weight change during a isothermal period of
time. The advantage here is that the true shape of changes is recorded, which
allows proper kinetic modeling. The limitations of the method come firstly from
the apparatus: Resolution, accuracy, precision, and the often unchecked long-
term stability are major points of concern. The pitfalls of contamination by furnace
interior and crucibles, and local atmosphere changes are identical to those men-
tioned above.
For coarse estimations of danger levels in corrosion nonisothermal thermo-
gravimetry or differential thermal analysis (DTA) can be used.
5.5.2 Corrosion Data

From the descriptions above the reader will accept that there are plenty of problems
of defining data and parameters for the characterization and comparison of
corrosion resistance. Standardization is under discussion [29], but is not generally
accepted. Only for very specific corrosion cases are the first norms and recommen-
dations available [30-341.
This is the reason why even excellent data collections on advanced ceramics
[35], become very fuzzy when corrosion is concerned. We will discuss here briefly
the different data types.
For a general case we first look for the changes in dimensions (Fig. 8). There are
three principal changes: a thickness of a scale growing on the substrate, a change in
the overall size of the component, and penetration depths. The most interesting
parameter from the point-of-view of an engineer would be the penetration depth,
because it would allow evaluation of the effective cross section of the component
after corrosive attack and hence prediction of the remaining strength.
layer thickness
corroded size
Figure 8. Principal dimensional changes during corrosive attack.
For homogenous surface attack by active corrosion the case simplifies to the
recession as there is no scale growing and the remaining cross section is of virgin
material.
The parameter of interest is then the recession rate, which is identical to kl of
Eq. (6) for linear kinetics. In corrosion science the SI-units ms-' or the more
vivid unit mmyear-' of engineering practice are useful. The relation to mass loss
is obtained by multiplying by the density of the lost material to yield kgmp2s-'
or equivalent units.
It should be emphasized again that this parameter is useless unless the physical
boundary conditions are specified. The quotation of partial pressures of outgasing
species is often more meaningful.
A problem arises when the attack is localized, because the attack is then different
for different phases and the outer surface is no longer identical to the virgin material.
The evaluation of penetration depth and rate may be performed in some cases using
dye tests. If penetration goes with formation of porosity, dyes like methylene-blue or
phosphorescing agents can be applied to show the depth of corrosion in cross
sections.
More complex situations, such as with chemical modification of the grain-
boundary phase, may be measured in optical thin sections, but they need usually
more careful evaluation by chemical profiling with microanalytical techniques.
Note that the kinetics of active corrosion may then deviate strongly from linear
kinetics because leaching by liquid media or evaporation out of channels often
involves diffusional problems.
A k1 is then not the parameter for ranking. A better choice would be the reference
to critical effects, such as the 'time needed for 0.1 mm penetration at 0.7 T,' [36].
Regrettably, such parameters are extremely scarce in the literature.
As the experimental determination and the investigations afterwards are highly
time-consuming and expensive, an often followed method of characterization is
by the effect of corrosion on mechanical properties, usually strength or wear proper-
ties [37,38], in addition to thermogravimetric data.
Changes (a.u.)
0.0 -
\
\
\
\
Moving to the more complex cases with scale growth, Fig. 9 shows the range of
kinetic curves one may find. Figure 9 shows a parametric study for a single case of
para-linear behavior. Not even the sign of the curves is constant. We must therefore
discuss the individual contributions.
For linear scale growth the discussion above is valid. For processes of retarded
scale growth with time we describe the layer growth by k, (units of m2 s-l or equiva-
lents) for simple parabolic behavior (Eq. (8)).
The calculation of the appropriate mass change unit is not as straightforward as
in the linear case. This is because the mass change recorded by weighing or thermo-
balance is measuring the difference in masses and not the mass of the layer.
The latter may be inferred if the chemical reaction and the properties of the scale
material are known, introducing a stoichiometric factor u. As an example the
reaction
Sic + 1 4 0 2 = SiO2 + CO (13)
means that I mole of the condensed product Si02 (= 1Msio2= 60.09g with a
density psio2 of 2 . 2 g ~ m -in
~ the amorphous state) is produced from 1 mole of
S i c (= 40.09 g). Thus a recorded mass change of Ax = nMproduct- nkfmaterial
(= 19.99g) in Eq. (12) represents the production of a full mole of S O 2 . In general
it follows that the factor u is defined from the reaction equation as
(nM)condensed product
U = (14)
(nM)condensed product - (nM)material '
which is 3 for reaction (13). Because of the form of Eq. (8) the two k, are hence
[ y1 [$1
related by
3
Pproduct
2
k,(layer-thickness) - = kp(mass-change)
U2
Neglecting the factor u has lead to occasional inconsistencies in the literature. The
discussion of it seems trivial but a close look at Eq. (14) shows that for other cases,
for example
SiBCN3 + 2.25 O2 H S O 2 + 4B203+ CO + 1.5N2, (16)
u can become very large with the result that substantial scale production may occur
almost invisible to thermogravimetry.
The layer thickness derived from any of the kinetic equations may be recalculated
to give a solution for homogeneous recession:
From this discussion one might conclude that layer thickness measurements are the
better choice for monitoring passive corrosion. While this is true from the point of
resolution (ellipsometry allows the determination of scales down to the nm scale) it
is difficult for high-temperature in situ recording and it implies dense transparent
layers, which may not be the case.
Thus a safe characterization will ask for both quantities to be measured indepen-
dently. Inconsistencies between measured and calculated values from Eq. (15) will
then immediately point to problems concerning scale structure and/or internal
corrosion.
Additional recording of mechanical property changes has the advantage of
including information on potentially dangerous new flaw populations derived by
corrosion. The overlay of several parameters (flaw geometry, cross-section changes,
stress field changes) will not allow us to use mechanical testing as the only criterion.
Thus, at the present time we prefer to compare and rank materials in terms of time
needed for penetration to a certain depth at the temperature of interest to make any
type of kinetic equation accessible. The fixed ‘critical condition’ is taken to be
0.1 mm, following Ashby and Jones [36], the time unit is in years to give a measure
for engineering practice. An engineer may then be able to decide that a material is
not useful for long-term applications if it drops to log t, < 0 at the temperature of
interest where tc is the critical time in years for a penetration of 0.1 mm. In the
following sections on the materials this is calculated but the uncertainties have to
be kept in mind.
5.6 Materials
5.6.1 Diamond and Diamond-like Carbons
Diamond is known to resist all acids and most other chemical substances in the
low-temperature range to 100°C. However, transformation of diamond into
other carbon phases may occur under local contact stress at room temperature
[39,401
5.6 Materiuls 155
Diamond will be etched by caustic alkalis and some oxidizing salts, for example
by NaClO and KOCl at 380°C [41], and K and Na nitrates above 400°C. Diamond
reacts with metals, which form carbides (e.g. W, Ta, Ti, and Zr) or which dissolve
carbon (e.g. Fe, Co, Mn, Ni, and Cr) [42]. Detailed knowledge of the temperatures
of reactions with metals is reviewed in the context of experiments with metals for the
catalysis of diamond synthesis [43].
Diamond corrosion is usually a problem of oxidation and graphitization. Graphi-
tization in inert atmospheres was observed at about 1500°C and becomes very fast at
2100°, where sizable diamonds are converted to graphite within minutes [44]. An
extrapolation of the data gives a T M 1450°C for t, = 0. It has been reported that
graphitization is prevented in pure H2 up to 2200°C [45].
Surface graphitization associated with concurrent oxidation, which may be found
at T > 600°C, is not thought to be a true conversion but an autocatalytic surface
reaction involving CO, species.
The oxidation of diamond is clearly an active corrosion process. At least up to
700°C diamond has a fast reacting { 11 1)-plane, an intermediate { 1 lo} and a slow
{ 100)-plane, which indicates reaction control. At higher temperatures and/or
lower oxygen pressures gas diffusion becomes rate determining in analogy with
graphite [46],and this is indicated by a more even attack [47]. Hence corrosion
rates are faster or start at lower temperatures for fine powders compared to films
and the corrosion in air is faster than in low-oxygen environments [48].
At 720°C diamond will burn in an oxygen jet. It is often stated in the literature
that oxidation starts at 600°C. The theoretical onset of oxidation should be well
below 600"C, because from the etching rates of Evans [44] for the { 111)-plane in
1 bar oxygen a t, well below zero ( ~ 0 . 0 2 )is calculated. Linear extrapolation
would give T M 340°C for tc = 0. However, in analogy with graphite there may
be an accumulation of C-0 complexes at the surface, which is faster then their
removal between 300°C and 500°C [49]. Surface graphitization at T > 600°C
points in this direction.
Also in analogy with other carbon forms, impurities may act as catalysts for the
oxidation, which can decrease oxidation temperatures by more then 100°C and
increase rates by orders of magnitude [50].
To a first approximation the temperature sensitivity of the oxidation rate at
T < 800°C is less dependent on the structure of diamond (films) then on crystallo-
graphic orientation, the data for the diamond faces bracket those of thin layers [47].
As described in more detail in the contribution on hydrothermal synthesis of
diamond in this book, hydrothermal reactions start for fine powders at 7OO0C,
but a significant etching and dissolution of large diamond grains is observed only
at T 2 800°C [51].
5.6.2 Carbides
The carbides considered as hard materials include B4C, WC, and Sic. In oxidizing
environments, in particular, their behavior is so different that they can not be com-
pared directly.
5.6.2.1 Silicon Carbide

Compilations of corrosion data on Sic-based materials have been given in some
hand- and textbooks [2,35,52,53], many papers on it are included in the proceed-
ings of specific meetings [3,5,54].
5.6.2.1.1 Corrosion by Liquid Media There is only limited information about the
behavior of pure Sic in low-temperature water, acids, and bases. A relatively
pure sintered S i c is applied in many pumping systems for the chemical industry
[%].This provides evidence that it is hardly affected by chemical attack.
Significant attacks can occur via a grain-boundary phase in other Sic types, and
this is usually accompanied by a decrease in strength and Weibull modulus. Thus it
is clear that liquid-phase sintered S i c will show less corrosion resistance with a
behavior depending on the chemistry of the additives. For Y/Al-sintered S i c signi-
ficant corrosion was reported for a treatment in boiling 20% HC1 and 50% NaOH;
concentrated acids (HZSO4, H3P04)and bases (KOH) had much less effect [56].
Sic with free Si (Si-Sic) is affected in particular by hot strong bases (10M
NaOH) and hot diluted acids (H2SO4 at pH = 3) [57].
5.6.2.1.2 Hydrothermal Corrosion In contrast to the good behavior of S i c at lower

temperatures a strong attack can be obtained in supercritical water. At pressures of
100MPa there is some indication from powder experiments that initial reactions
start at 300°C [58, 591, strong corrosion is observed at T > 500°C [60,61]. The
lower temperature onset is supported by tribological studies [62].
The main mechanism is the formation of silica from Sic along with CH4,
COXf C. The silica is then dissolved in HzO. The dissolution rate of silica will
play a vital role in the kinetics of the process. Basically the attack should have
active corrosion character (Eq. (6)).
Chemical modification of the water to increase the pH will enhance the dissolu-
tion of silica [63] and hence further increase corrosion of Sic.
5.6.2.1.3 Corrosion by Metals Sic can be attacked by a number of metals, because

often silicides and in some cases other carbides are stable phases and eutectic
temperatures are low. From [52] we have evidence of reaction with Bi (600"C),
Li (815"C), Mg (8OO0C), A1 (750"C), Ti (575"C), Ta (12OO0C), Nb (1300"C),
Cr (lOOO"C), Mo (1200"C), W (1500"C), Ni (500"C), Co (1150"C), Fe (lOOO"C),
Cu (950"C), Pd (600"C), and Au (950°C). Resistance to attack was noted for Na
(350"C), Ca (1 180"C), Cd (500"C), Sn (600"C), Pb (815"C), and Ag (962°C).
These data have to be treated with some caution, because many results are
obtained from treatment in vacuum without a protective scale present. The kinetics
of the reaction have been described as parabolic for a number of cases. However,
they depend also on physical conditions like the wetting behavior of the liquid
metal or a formed melt [64].
Thus, it is obvious that Sic with free Si will react more readily than Sic. Tempera-
tures of possible reactions may be evaluated with the aid of phase diagrams for the
appropriate Si-metal system.
5.6 Materials 157
From own experience we know that Pt can be used as crucible material in

oxidation studies of Sic. However, in vacuum the formation of melt and silicides
at the eutectic of the system (= 1400°C) is so fast that individual Pt grains drill
deep holes into the material within seconds.
The use of Sic as a hard material in wear application, can thus be endangered by
a reaction with the metal if the silica scale is eroded by the process and contact
temperatures are high enough.
5.6.2.1.4 Corrosion by Salt Melts (Hot Corrosion) Data and reviews of the
mechanisms of the hot corrosion behavior of Sic have been presented in a
number of papers by Jacobson and coworkers [65-681.
A representative member of the hot corrosion agents is Na2S04. Its generation is
explained in combustion environments by
2 NaCl + SO2 + 0.5 O2 + H 2 0 H Na2S04+ 2 HCl. (18)
Since the action in the liquid state is regarded as most effective, the first condition
for the window of hot corrosion is given by the melting and dewpoint of Na2S04.
The melting point of 884°C does not vary much, the dewpoint depends on pressure,
S-content of the fuel and NaC1-concentration in the atmosphere. For conditions
with NaCl > 1 p.p.m. and combustion pressures of 100 bar it is calculated to be
1100-1200°C from Eq. (18).
However, the effective corrosion action is generally believed not to be by the
sulfate itself. It stems from the activity of N a 2 0 arising from the decomposition
Na2S04 * N a 2 0 + SO3 (19)
because only the N a 2 0 is attacking the protective scale of Si02 by
-
x Si02 + N a 2 0 H N a 2 0 (SiO2).,. (20)
This is the reason why other compounds releasing alkali oxide on decomposition,
such as Na2C03,have similar hot corrosion effects. It should be noted that complex
salts, in particular those containing vanadium, can have an even stronger hot
corrosion effect, as is well known from metals and oxides [69,70].
From Eq. (19) it can be calculated that only for conditions with a low P(SO3)will we
have a dissociation sufficient to drive reaction (20) to the right side. Consequently the
window of hot corrosion has a third axis, the P ( S 0 3 )of the combustion, and it has been
computed [71] that the window should only open for low-S fuels (<0.5%, a condition
met by many modern fuels) and at comparatively high temperatures (FZ>IOOO"C).
In reality the attack already starts at the melting point of Na2S04. This is
attributed to internal controlling factors. Many Sic-based materials contain free
carbon, which promotes the dissociation of Na2S04 and makes the silicate melt
more basic. The dramatic attack of Na2S04 on S i c with free carbon has been
confirmed experimentally [72].
As is described in more detail for Si3N4,recent studies suggest that free carbon is
an enhancing factor but not a necessary condition [73,74]. S i c may act as the
reducing agent to promote sulfate dissociation either directly or via an oxygen
gradient in the silica scale consumed by reaction (20).
It is very difficult to quantify hot corrosion in laboratory experiments with pre-

loaded samples, because the salt melts propagate rapidly both by spreading and
gas phase transport, resulting in uneven coverage of the sample and changes with
time.
A much better constancy of conditions is achieved in burner-rig methods [75-781.
However, even here a quantification is difficult as the usually highly porous
scales containing sulfate, silicate and silica are liable to spalling. Furthermore, the
attack is often in the form of pitting, and so there is a change in exposed surface
and the attack is nonuniform. The statistical nature of pit distribution is probably
not due to a distribution of secondary phases but the result of the action of surface
tension effects, because two immiscible liquids (sulfate and silicate) with differing
surface stresses are present [74].
Therefore, while weight changes of chemically cleaned corroded samples are good
indications for the strength of attack, they remain unsatisfactory and a widely used
criterion is by the influence of corrosion on mechanical properties. Even for cases
with good control of chemical changes, a prediction of the effect on strength can
not follow directly, because the fractures occur usually from the newly formed
pits (Fig. 10 [78]).
None the less, as seen in Fig. 10, the relation between pit size and failure is evident
for most Sic materials. A model for the prediction of properties from corrosion
conditions would thus have to give an expression for the formation and growth
of pits with time, temperature, and external properties. The author is not aware
of a fully consistent model to yield this result.
Thus we are left with individual studies. A study on commercial Sic [78] yielded
the typical result that a reduction in strength of 20-30% is obtained after 4 0 h
exposure at 1000°C.
1 1
(ac)"', pm
500 200 100 50
2 400
z --0.75K,c
-
d (a~)"~
300
5E
c
;200 0 Type A
0 Type B
A Type C
3
c
0
m
t 100 0 Type D
0.1 0.2 0.3

(Pit dimensions)-'", (ac)-'", pm-"'
Figure 10. Relation of fracture stress and pit dimension from a study of four different Sic materials
1781.
5.6 Materials 159
Figure 10 also shows that the relation with pit size is not true for all Sic-materials:
Type D shows no correlation and has a much stronger decrease in strength
(= -50%). It was a material with free silicon. It follows that Si-Sic has less hot
corrosion resistance and that the attack and its consequences are due to the
distribution of the preferentially corroded Si.
To improve the hot corrosion resistance current developments aim for a protec-
tion by mullite-based coatings [79,80].
5.6.2.1.5 Corrosion by Hot Gases Halogens do not form condensed species with Si
at the temperatures of interest. Thus, if they are pure (no oxygen or water present)
they will automatically induce active corrosion behavior and calculated pressures of
species like SiF4 or SiCI4over Sic will be as high as the concentration of F and C1 in
the gas.
In the presence of enough oxygen to form silica, these pressures will be signi-
ficantly reduced for the case of attack by C12 and accordingly lower corrosion
rates are measured [81]. However, pressures of SiC14 over Si02 can exceed
lop6bar at temperatures above 700°C in high-C12 environments.
Calculated pressures for an attack by H2 are likewise significant even at low
temperatures [8]. The calculated main low-T species is CH4, but the kinetics are
so unfavorable for its formation that in reality the beginning of significant active
corrosion is at temperatures above 13OO0C,as is known from etching studies [82].
This is only true for pure Sic: grain boundaries and secondary phases in sintered
S i c are attacked at temperatures as low as 1000°C [83].
For the survival of S i c it is therefore of great importance whether silica is formed
or not. The boundary for the active-to-passive oxidation is described in Fig. 1 1. It
shows the literature data and a Wagner-type theoretical model [15] based on the
H i m 76
Gulbranran
66
Vaughn
*
so
I 1 I 1
-8
1,200 1,400 1,600 1,800 2,000
Temperature ("C)
Figure 11. Transition temperatures and oxygen pressures for active and passive modes of oxidation
of Sic after [15].
influence of carbon activity and the presence of condensed SiO. Even though the
model might be challenged in view of new thermodynamic data, arguing against
condensed SiO, the general position of the boundary between active (I) and passive
oxidation is known from the experiments.
The region denoted active (11) in Fig. 11 is for a special situation, when the
pressures of the produced gas species (CO) exceed 1 bar, which can lead to bubble
formation and spalling, and hence quasi-active behavior, even though nominally
silica is produced. Experimental evidence for this was obtained at temperatures of
about 1760°C [84].
It should be noted that the oxygen pressures for the transition in Fig. 1 I should
not be confused with those obtained by COX-equilibria.Despite low oxygen partial
pressures of, for example CO in equilibrium with carbon, we have passive oxidation
of S i c up to x 1400°Cin CO, because CO is a reactive and oxidizing gas species for
S i c [52,85,86]! Consequently active corrosion in COX-environmentsis measured
only at very high temperatures [87].
In passive oxidizing environments S i c forms silica (Eq. (13)). For pure S i c a
parabolic behavior is evident [88]. For SSiC there is evidence that oxygen does not
penetrate deeply into the interior [89], leading to a relatively homogeneous surface
oxidation.
Therefore SSiC has often parabolic oxidation rate constants nearly as good as
chemical vapor deposition (CVD)-Sic and powder. From a data collection [53]
plus some more recent data [88] the critical time t, for 0.1 mm recession is computed
and shown in Fig. 12. We have omitted data from experimental changes, which were
deliberately made with high additive contents to study their influence on oxidation.
It is evident from Fig. 12 that basically all cases meet a long-term stability criter-
ion up to very high temperatures.
Some cautionary remarks have to be added: At very high temperatures the
kinetics may not be adequately described by the parabolic law; at T > 1700°C the
problem of active (11) oxidation will become important; SSiC can show bubble
and pit formation at much lower temperatures, which is usually attributed to the
additives B and/or A1 [90].
This bubble-and-pit formation is sometimes only present after a ‘incubation
period’ of several lOOh at T > 1600°C. Most oxidation studies run only for times
< 100 h. Therefore long-term studies may be necessary for true qualification of Sic.
Industrial utility of Sic is not guaranteed by a sufficient resistance to oxidation
by O2 or air. The influence of other components of the atmosphere or transferred
impurities has to be evaluated. Often this is difficult because strong variations in
temperature and gas composition are encountered.
None the less it has been shown that SSiC can resist flue gases of aluminum
remelting furnaces for up to 1000h without much change in strength [91], although
this is certainly not true for all plants of this type [92]. Similarly nitrogen-based gases
of the system N2-H2-C0 did not decrease the strength of SSiC, while endothermic
gases had an adverse effect (-20-50%) at 1200-1300°C [93,94] in long-term
experiments. Typically B-doped Sic behaved better then Al-doped or those with
free Si. A good resistance of Sic-based materials to SO2 or H2S has also been
reported [95].
5.6 Materials 161
I " 4
*
* *
* ssic
A powder&SC
A CVD
800 1000 1200 1400 1600 1800

900 1100 1300 1500 1700
Temperature ("C)
Figure 12. Calculated critical times (years) for 0.1 mm recession of S i c materials, using the data
compilation of [53].
The influence of water on the corrosion of Sic seems to be threefold: It enhances

the oxidation rate slightly; it transports impurities (e.g. Al and alkalis) which can
strongly enhance oxidation; and it is able to evaporize the silica scale inducing
para-linear behavior under appropriate flow conditions [23]. This may impair
some applications.
5.6.2.2 Boron Carbide

Boron carbide is among the hardest materials yielding only to diamond and boron
nitride. It is also one of the most corrosion-resistant compounds at room or moder-
ate temperatures. When considering the corrosion resistance of boron carbide
materials, it is important to remember that they are rarely stoichiometric, with
the carbon content varying from 9.88 to 23.4% [96] Many of them contain free
carbon or sintering aids. Thus their behavior depends on the chemical composition.
5.6.2.2.1 Corrosion by Liquid Media Pure boron carbide is insoluble in HCl,

H2S04, and HN03, even when boiling. Resistance of hot-pressed boron carbide
to sulfuric acid solution has been investigated [97]. The material was produced
from a powder containing 77.8-79% boron, 21.6-20.4% carbon, 0.25-0.7Y0
boron oxide, and 0.25-0.7% free boron. Dense and additive-free B4C specimens did
not interact with 10% sulfuric acid either at room temperature or on boiling. A
small mass loss (<O. 1 mg cm-2) in the initial period of boiling can be explained by
surface impurities. As has been shown [98],the corrosion resistance of boron carbide
powder in water, mineral acids and alkalis is largely dependent on its purity and the
ordering of the crystalline structure. The studies on the electrochemical behavior of
hot-pressed boron carbide in H2SO4 and NaOH solutions [99] have demonstrated
that the corrosion resistance of B4C depends to a large extent on the content of
additives and impurities.
However, during heating, strong oxidizing acids and oxidant mixtures such as
+
Cr03 H2SO4, KIO3 + +
H3P04, K2Cr207 KIO3, Cr03 + H2SO4 + H3P04,
+ + +
KMn04 H2SO4, H2SO4 HC104 K2Cr207[98] oxidize free carbon present
in many boron carbide materials. To remove free carbon from B4C, alkaline
solutions of hydrogen peroxide, bromine, and others were also used [98].
5.6.2.2.2 Hydrothermal Corrosion and Corrosion in Water Vapor The interactions of

boron carbide with water vapor starts at 250°C. The following reactions are possible:
+
B4C + 8 H20 + 2 B 2 0 3 C02 + 8 H2, (21)
B2O3 + H20 + 2 HBO2, (22)
and
B203 + 3 H20 + 2 H3B03. (23)
Gaseous boric acid removes a boron oxide film. The rates of formation and removal
of the B2O3 film are equal at 550-600°C in air with a dew point of 25-70°C and at
650°C with a dew point of 88°C. At higher temperatures, B203 is formed at a higher
rate than it is removed by the interaction with water vapor. Therefore, at low
temperatures boron carbide is oxidized with water vapor more rapidly than with
dry air, at high temperatures the situation is quite the opposite [2].
Hydrothermal corrosion of B4C has been studied theoretically and experimen-
tally [loo]. Reaction (21) as well as reactions leading to the formation of CH4 and
CO are possible. Reactions that lead to the formation of carbon oxides dominate
only at low pressures and carbide:water ratios. Under all other conditions, the for-
mation of methane has been predicted. HB02, H3B03and H3B306gas molecules,
which were predicted by thermodynamic simulation under various temperatures
and pressures, are very stable thermodynamically. They may condense to some
type of boric acid on cooling.
Unlike Sic, boron carbide did not show any evidence of free carbon formation
under hydrothermal conditions. Neither thermodynamic simulation, nor experiments
show the formation of elemental carbon under hydrothermal conditions. Even the
traces of carbon that were present in the as-received B4C powder disappeared after
a hydrothermal treatment.
Tests of a hot-pressed B4C-lO%SiAlON composite demonstrated high corrosion
resistance up to 500°C. Even at 500"C, almost no changes of the sample surface
occurred under the experimental conditions.
5.6 Materials 163
5.6.2.2.3 Corrosion by Melts Hot-pressed boron carbide is attacked by fused

alkalis. The use of boron-carbide-based materials at high temperatures is limited
at present, which probably results from the fact that they have been investigated
primarily at room or moderate temperature [ 1011.
5.6.2.2.4 Corrosion by Hot Gases Boron carbide is mainly used at relatively low
temperatures [96], therefore its oxidation has not been studied as thoroughly as
that of the Si3N4and Sic ceramics. There are data on the oxidation of B4C powders
[lo21 and several papers on oxidation of boron carbide materials produced by
different methods [103-1071.
In [ 1081 different reactions in the system B4C-02 were calculated thermodynami-
cally at oxygen pressures of 105 and 1.3x lop3kPa. It was shown that the oxidation
of B4C may result both in condensed (B203,B, C) and gaseous (CO, COz, B202,
BO, BOz) products. At high oxygen partial pressures the reaction
B4C + 4 0 2 ++ 2 B 2 0 3+ C02 (24)
should dominate. This reaction was confirmed by the results of experimental
investigations on powders and sintered specimens [ 1091.
Oxidation of boron carbide powders starts at about 600°C. Studies on hot-pressed
B4C containing iron, aluminum, magnesium, calcium, titanium, silicon, and other
impurities at a level of up to 4% and oxidized in air [lo41 demonstrated that upon
nonisothermal heating of specimens exothermal effects due to oxidation can be
registered above 600°C. A detectable mass gain starts above 700°C. At 1000°C the
oxidation process intensifies and above 1200°C the material starts oxidizing consider-
ably, with a high thermal effect. At temperatures above 12OO0C,the specimen weight
decreases due to the B203vaporization, the rate of which increases significantly at this
temperature [ 1091. The X-ray diffraction (XRD) analysis of oxidized specimens did
not reveal any solid oxidation products other than B203.This is in agreement with
the thermodynamic calculations [98] and the experimental investigations showing
that reaction (24) dominates B4C oxidation.
The kinetic curves obtained under isothermal conditions (Fig. 13) follow the
paralinear law (Fig. 7) with competing B4C oxidation and B 2 0 3 vaporization.
3, 1
.
N
6 1
F o
@
E -10
-20
-300 < 1 2 3 4 5
t, h
Figure 13. Oxidation kinetics of hot-pressed B4C at: (1) 800°C; (2) 900°C; (3) 1000°C;(4) 1100°C;
and (5) 1250°C; reproduced from [2].
Thus, the oxidation kinetics due to the boron oxide vaporization at temperatures
above 1200°C result in weight loss of specimens (Fig. 13), which becomes linear
in time after the first 30min of oxidation. On oxidation at temperatures below
1100°C a liquid boron oxide layer (T, of amorphous B2O3 is 450°C [llO])
uniformly covers the specimen surface.
Under the conditions of active B203 vaporization the diffusion of impurities
apparently does not exert great influence on the oxidation process, unlike in the
case of oxidation of silica formers. One may suppose that up to 1200°C the
oxidation rate of B4C is controlled by the oxygen diffusion through the B203
layer.
Differencesin the thermal expansion coefficients of B203 and B4C lead to crack-
ing of the oxide layer after cooling from high temperatures. The XRD analysis
revealed B203 and H3B03 in the surface layer of oxidized specimens. However,
H3B03 is not formed during the oxidation of materials, but during the exposure
of specimens to air at room temperature for several days due to the hydration of
boron oxide according to reaction (23). The oxidized specimens kept in air were
covered with a white film due to the hydration of boron oxide.
It is important to note that the specimens of ceramics oxidized at 1400°C had a
graphite layer of up to 30 pm thickness under the B203film [107]. If the oxidation
process was accompanied by friction, the graphitized layer was formed at much
lower temperatures. The formation of free carbon was also observed upon oxidation
of B4C-Sic composites [l 111.
Active oxidation of B4C at a low partial oxygen pressure leads to the formation
of gaseous boron oxides and removal of them from the surface of the specimen.
Similar to other materials, porosity increases the reactive surface of boron carbide
specimens and weight gain on oxidation [1061.
All the above primarily refers to materials containing relatively small amounts of
impurities. However, boron carbide ceramics usually contain 7-20% Al or A1203to
increase hardness and 6-1 3% silicon to improve mechanical characteristics [98].
Aluminum- and silicon-doped materials often exhibit structural nonuniformity.
They have areas of pure boron carbide, and silicon- and aluminum-doped areas
[107].
In contrast to the oxidation of the additive-free material, weight gain instead of
its loss is observed above 1200"C, that is the rate of oxide formation is higher than
the rate of its vaporization. Electron microscopy investigations of specimen sur-
faces of Si- and Al-doped materials heated to 1430°C demonstrated glassy oxide
layers containing boron, oxygen, silicon, and aluminum. Thus, the oxide layer is
a borosilicate glass. Crystallization of aluminum borates forming needle-like
crystals also has been found in our work. The dissolution of silica and alumina
in boron oxide retards the rate of its vaporization. In general, addition of silicon
and aluminum to boron carbide facilitates the formation of borosilicate glass on
oxidation and decreases the rate of boron oxide vaporization, thus increasing
the oxidation resistance of the material [ll 11 [2]. However, to form a continuous
protective layer on the specimen surface, it is necessary to provide a uniform
distribution of additives in the material excluding the existence of pure boron
carbide areas.
5.6 Materials 165
B4C reacts with C 0 2 to yield B203 and CO or free carbon [96]. Boron carbide
neither interacts with sulfur and phosphorus vapors, nor with nitrogen up to
1200°C. BN can be formed upon reaction with nitrogen at higher temperatures,
or when ammonia is added. With chlorine it reacts above 1000°C to form BC13
and graphite. Bromine and iodine do not react with B4C [98].
5.6.2.1 Tungsten Carbide

Hard materials on the basis of WC have usually a metallic binder phase of Co
and/or Cr, Fe, Ni, and additions of other carbides (Tic, TaC, NbC), which deter-
mine the corrosion resistance. Accordingly, these composites are attacked by many
acids and bases. The authors refer the reader to previous work [112,113] for more
details.
Tungsten carbide has a low resistance to oxidation. It is the result of low
protective properties of tungsten oxides. Dissolution of oxygen in the carbide
structure starts at low temperatures, and W 0 2 and W 0 3 form upon oxidation at
500°C and higher. The oxidation rate increases drastically at 800°C due to vaporiza-
tion of W 0 3 [108].
According to thermodynamic simulation [loo], at a carbide:water ratio of 1: 10,
tungsten carbide should be completely oxidized producing H2W04or W 0 3 at lower
temperatures and W 0 2 at higher temperatures. Carbon is oxidized to CO and C 0 2
in the case of a low W C : H 2 0 molar ratio. Formation of free carbon upon hydro-
thermal corrosion of WC was predicted to be possible at a high WC:water ratio
upon hydrothermal corrosion of WC. The stability of WC increases with increasing
pressure.
Raman spectra obtained from the hydrothermally treated WC powders demon-
strate the presence of graphite. Hydrothermal treatment of a WC-G%Ni commercial
cemented carbide demonstrated a rather low corrosion rates below 500°C. Only a very
thin layer of corrosion products covered the surface of the samples after a treatment at
350°C. At 500"C, the corrosion rate increased and all WC grains from the surface
layer of the sample were dissolved in the supercritical fluid. This led to a slight
decrease of the sample weight. Ni does not interact or interacts very slowly with
water under these conditions. Therefore, a network of Ni grain-boundary layers
remained on the surface. Ni grain-boundary phase serves probably as a diffusion
barrier and protects the material from further corrosion. At 700°C, the corrosion
rate increases and the surface of the sample was covered with a thick layer of
corrosion products.
No changes in the phase composition of the samples were found by XRD after
the treatment at 350°C. WC, Ni, and traces of W 0 3 were found after hydro-
thermal treatments in the temperature range 400-500°C. W 0 3 dissolves in water
and a weight loss of the samples was observed. At 700"C, oxidation of Ni occurs
and nickel oxide interacts with W 0 3 (or H2W04) leading to the formation of
NiW04.
Also, hydrothermal corrosion of WC-Zr02 composites has been studied, but at
lower temperatures when less significant interaction of WC with water occurred
[114, 1151.
5.6.3 Nitrides
5.6.3.1 Silicon Nitride
In many cases the references quoted for Sic can be used to access corrosion data on
Si3N4-basedmaterials [2,3, 5, 35, 53, 54, 116, 1171. Si3N4ceramics are much more
complicated in their behavior because sintered types come with a significant amount
of grain-boundary phase, which determines the overall behavior.
5.6.3.1.1 Corrosion by Liquid Media The higher susceptibility of Si3N4ceramics to

corrosion by liquid media is already apparent from reports on hydroabrasive wear,
which is much higher then the wear of S i c [55].
It is not the attack of the matrix Si3N4,which controls aqueous corrosion but that
of the grain-boundary phase. Often this grain-boundary phase is a oxidic silicate
glass with or without small nitrogen contents. To predict the corrosion resistance
in water, acids, and bases the reader is thus referred to the literature on glass and
glass corrosion [118,119].
Rules of thumb would imply that a glass with more network-modifiers is less
corrosion resistant to corrosion, materials with less glass content are better and
that a greater crystallinity of the grain-boundary phase is of advantage.
The kinetics of the leaching processes are often linear for H F and strong bases,
while selective acid etching can be parabolic. Leaching is always associated with a
loss in strength and tends to correlate linearly with degree of dissolution from the
glass matrix as has been shown for H F and HCl with Y - and Al-leaching [120].
At 70°C a treatment for 240 h with 1 M HCl can reduce the bending strength by
z 50%.
5.6.3.1.2 Hydrothermal Corrosion The onset of reactions of Si3N4 powder with

water at 100 MPa according to
+
Si3N4+ 6 H 2 0 H 3 Si02 4 NH3 (25)
starts at temperatures below 200°C [61], probably at about 130°C [121]. At 500°C
the powder is completely oxidized within a few hours.
Reaction (25) would suggest a passivating behavior but the solubility of silica is
favored even at low temperatures in the alkaline water present due to the dissolution
of NH3. Hence basic reaction kinetics are linear, it is a form of active corrosion.
The grain-boundary phase of Si3N4-ceramicsmay be even stronger affected by
hydrothermal corrosion. This leads to pit formation at conditions of 300°C and
8.6MPa [122]. The shape and depths of pit formation depend on the additive
type. It is interesting to note that a hot isostatically pressed (HIP) Si3N4 ‘without
sintering additives’ was attacked more strongly then those with additives, because
the grain boundaries with silica from the oxygen content of the powder were leached
and resulted in grain dislodgment [ 1231.
It should be noted that hydrothermal corrosion should not be confused with cor-
rosion by wet atmospheres and that the behavior changes at higher temperatures
when N2 + H2 is the product gas rather then NH3 [60].
5.6 Materials 167
5.6.3.1.3 Corrosion by Metals Reactions of Si3N4 with metals and alloys are a
complex issue, because the atmosphere both for oxygen and nitrogen partial
pressures plays a role and silicides, nitrides, or ternary compounds can form. A
good compilation of the partly conflicting observations has been provided [1161.
For the interaction of the related SiAlONs and Fe-based alloys see published
work [124].
Predictions for reactions can be made on the basis of Ellingham-type diagrams. In
noble gas atmospheres or under vacuum, Si3N4reacts readily with most transition
elements below 1000°C and with many other metals at varying temperatures.
Whether this is true for a given Si3N4 ceramic under application conditions
depends on many parameters like a silica layer protection, diffusion coefficients,
wetting behavior etc. The reader is here referred to the reference mentioned above
[116].
5.6.3.1.4 Corrosion by Salt Melts Hot corrosion of Si3N4 has been extensively
studied and reviewed by researchers at NASA labs [67,68,77, 125, 1261. The basic
attack is that on the protective silica layer, which makes the behavior similar to
that of Sic and basic mechanisms can be taken from the chapter on S i c above.
Recent investigations [74] gave more insight into the details of the kinetics, which
is shown in Fig. 14. The surface tension differences between sulfate and silicate melts
result in a fractal pattern of hillocks of silicate melt beneath the sulfate, hereby
exposing unaltered silica with a strong oxygen potential gradient to the sulfate.
On consumption of this silica the decomposition of the sulfate is enhanced up to
a gas bubble formation. The exposed areas are the sites for stronger corrosion,
which can ultimately produce pits. However, the process does slow down, when
the sulfate is used up or a critical thickness of silicate layer is reached.
There is a strong indication that the grain-boundary phases modify the develop-
ment. Wide shallow pits and/or exposure of individual Si3N4-grainsare observed
[127,128], and Na advances into the grain-boundary phase.
The resulting degradation of mechanical properties is of similar order as for S i c
(20-50%) [26,127,129]. However, it seems that the degradation is not linear with
time but levels out with time [129]. Furthermore, in pitted samples there is an
increase in the Weibull modulus, with the result that the failure stress at low failure
probabilities is decreasing much less than the average strength [ 1301.
Taking into account the higher level of initial strength of Si3N4-basedceramics
the remaining strength after hot corrosion is usually higher compared to SIC.
5.6.3.1.5 Corrosion by Hot Gases Active corrosion has usually an even more
pronounced effect on strength, compared to Sic, as the phase between the grains
may vanish leading to grain dislodgment and inhomogeneous internal corrosion
[ 1311. All gases which produce high partial pressures of Si-bearing species (namely
halogens) are therefore dangerous, particular in low-oxygen environments [ 132-
1341.
In mixed and oxidizing atmospheres Si3N4 has a superior resistance to C1-
containing environments relative to Sic materials [ 135-1 371. A small water content
in H2S-H2 mixtures is also capable of producing a silica scale, which slows down
Figure 14. Schematic model for the hot corrosion kinetics of silica-formers after [74].
Ess
SiO,
3J
(not watersoluble)
Na,O * @i02
-
immiscible at 9OOOC
1
(watersoluble)
n
of the Na-sulfate
Na,SO,
with amounts of
168
5.6 Materials 169
corrosion [ 1341. As for Sic the plain influence of a water vapor content in oxidizing
environments is probably moderate [ 1381 unless it transports impurities.
The active(1)-to-passive transition has been discussed with some controversy but
most models [14,139,140] and experimental data [141-1431 point to very similar
conditions as outlined in Fig. 11 for Sic.
The passive oxidation of pure Si3N4has been intensively studied. Thermodynamic
data suggest that a two-step oxidation
2 Si3N4+ 1; O2 H 3 Si2N20 N2 + (26)
and
Si2N20 14 O2 + 2 SiOz+ + N2 (27)
and accordingly a duplex layer in a oxygen potential gradient, developing in a scale,
is favored. However, the discussion of the correct kinetics, the presence or absence
of a sublayer of Si2N20,the activation energy, and other issues are still not settled
[14,144-1511. Extremely low oxidation rates up to high temperatures are, however,
clear.
This contrasts strongly with the behavior of sintered Si3N4. In Fig. 15 we have
calculated tc for pure and sintered qualities, assuming parabolic growth and using
the data compiled earlier [53].
-1
0
... 1 -0
0
.
::'"""": .: ,
~
0 .
0 .
-21
HPISSN
, I , I 0 , , ,,,
-3
800 1000 1200 1400 1600 1800
900 1100 1300 1500 1700
Temperature ("C)
Figure 15. Calculated critical times (years) t, for 0.1 mm homogeneous recession by parabolic silica
scale growth on Si3N4-basedmaterials using data from the collection of [53].
A number of things can be learned from Fig. 15. First, sintered Si3N4 never
reaches the level of oxidation resistance of pure Si3N4. Typically a temperature
difference of 250°C exists to have comparable behavior for even the best qualities.
Second, at low to moderate temperatures pure Si3N4oxidizes much slower then
S i c (Fig. 12), but sintered Si3N4is comparable to Sic. Third, there is a very wide
scattering of results for sintered Si3N4.And last, but not least, at about 1400°C
there is a rapid breakdown in log t, to values below and much below zero, indicating
very rapid oxidation of sintered Si3N4.
The reasons for these features are manifold and again related to the presence of
grain-boundary phases. We often have a very complex passive oxidation behavior
with leaching of grain-boundary constituents to the scale, crystallization, internal
oxidation, bubble formation, scale cracking, and other features. All these
phenomena are different for different systems and vary with the amount and
exact composition of the additives, For detailed information the reader is referred
to the textbooks and proceedings already mentioned and some reviews, which cite
papers from a vast amount of literature [140,152, 1531.
We have to point out some facts, which should make the reader cautious in inter-
preting Fig. 14 too strictly. At low temperatures some Si3N4 types (sintered with
pure Y or rare earth additives) contained secondary phases, which oxidize rapidly
and with large volume increases to yield the oxide phase. Since at low temperatures
(<lOOO°C) the outer scale becomes ridged, it leads to cracking of the specimen [154]
and breakaway or pseudolinear behavior. This is part of the reason why there is such
a large spread in data from Si3N4ceramics even at low temperatures.
Particularly at T > 1300"C, we commonly have deviations from the parabolic law
and all the forms described in Sections 4.3.2 and 4.4.3 can be observed. Since the
extrapolation to t, is done on the basis of a parabolic approximation it will not
give very reliable t,.
Perhaps the most disturbing characteristic of sintered Si3N4 oxidation is the
appearance of internal oxidation accompanied by outer scale growth. Leaching of
grain boundary constituents into the scale (Fig. 16) is observed for Y, Al, Mg,
REE, and others to varying degrees. The leaching is connected to internal oxidation
and is for some cases the rate limiting step [156]. In the leached zone, which extends
into the sample with the same order of magnitude as the outer scale thickness, we
often find increased porosity and hence decreased hardness [155]. At least this
zone must be incorporated into the value for penetration but can not be gained
from measuring layer thickness or mass change.
There is furthermore recent evidence from transmission electron microscopy
(TEM) studies for the dissolution of matrix grains in the grain-boundary phase
during oxidation at high temperatures, which widens the grain boundary films as
a function of depth into the material [157,158]. This type of penetration will be
even more difficult to detect and quantify.
Therefore, for most Si3N4 ceramics it will be necessary to have mechanical
property measurements along with corrosion to evaluate their applicability. None
the less there are rules of thumb, which may help in materials selection. The most
important one is that the level at which oxidative corrosion becomes the prime
problem, is connected to the eutectic of the system sintering additive(s) + Si02.
5.6 Materials 171
Figure 16. EDX map for Y and SEM picture of SSN oxidized at 1400°Cfor 990 min [ 1551, showing
a leaching zone beneath the scale.
This may well apply also to the pure Si3N4 qualities, because here the eutectic
Si3N4-Si02 in the vicinity of 1600°C may be the ultimate temperature for long-
term use [1591.
The challenge for sintered Si3N4to be applicable at T > 1400°C has found first
success in studies which show that, for example, additions of MoSi, and surface
treatments are capable of enhancing the oxidation resistance [ 160, 1611.
5.6.3.2 Boron Nitride

Hard materials based on dense crystalline modifications of boron nitride find wide
application in cutting tools [162]. Since these materials are produced at high
pressures, the size of samples is usually small. Corrosion has been studied primarily
in view of abrasive and tool applications of BN and the data are limited.
However, there exists much information on corrosion of hexagonal graphite-like
BN. Similar to the graphite and diamond phases of carbon, the reaction products
are the same for all modifications of BN. However, hard modifications of BN typi-
cally show a higher corrosion resistance. Thus the available information on graphitic
BN can be used for evaluation of the lowest corrosion resistance limit of hard BN.
Although materials based on the high-pressure phases possess very high hardness
and strength, they are unstable at high temperatures, which imposes restrictions on
their application. When they are heated in air, along with oxidation, the wurzite and
sphalerite modifications can be transformed into the graphitic one [ 1621.
5.6.3.2.1 Corrosion by Liquid Media The resistance of pure boron nitride to attack
by diluted acids is lower than that to attack by concentrated ones and decreases in
the order H N 0 3 -+ HC1 + H2S04. It interacts slowly with concentrated sulfuric

acid to form ammonium sulfate and boric acid [I631 and the corrosion rate of
dense specimens is extremely low. Boron nitride dissolves on interaction with con-
centrated hydrofluoric acid (94% H F solution) by the reaction
BN + 4HF H NH4BF4 (28)
and on heating with ammonium fluoride and concentrated sulfuric acid by the
reaction
+ +
BN 3 NH4F 2 H2S04 % 2 (NH4)2S04 BF3. + (29)
Boron nitride specimens are resistant to attack by hydrochloric, sulfuric, and
phosphoric acid solutions with additions of oxidizing agents (KMn04, K~Cr207,
KC104), but decompose most rapidly in a 5% H2S04solution with KC104 addition
[ 1631. The corrosion resistance of powders and bulk specimens in alkaline solutions
is rather high and almost independent of the concentration.
5.6.3.2.3 Corrosion by Melts Boron nitride fused with potassium carbonate

decomposes by the reaction
BN + K2C03 H KB02 + KOCN. (30)
Nitride excess leads to a certain amount of KCN. If carbon is present, the reaction
proceeds as follows:
4 BN + 3 K2CO3 + 2 C H K2B407 + C02 + 4 KCN. (31)
Boron nitride reacts with sodium formate [163].
It must be emphasized that boron nitride has an exceptionally good resistance to
most molten salts and glasses, probably due to its resistance to wetting. It is inert to
a number of corrosive molten salts, such as alkali halides, lithium borate and
cryolites, and has a good resistance to molten glasses. However, it does react with
molten alkali carbonates and hydroxides.
5.6.3.2.2 Corrosion by Hot Gases Thermodynamic calculations [1081 show that

the oxidation of boron nitride at atmospheric and low pressures over the whole
temperature range to 2000°C is dominated by
4BN+302~22203+2N2. (32)
The oxidation of boron nitride as well as boron carbide is distinguished by an oxide
layer remaining in a liquid state and vaporizing considerably over the whole
temperature range. However, the study of the mechanisms of oxidation of boron
nitride materials is complicated by the existence of several BN modifications with
considerable differences in their structure and properties. The oxidation of a
stable hexagonal a-modification of BN with a graphite lattice has been studied
most thoroughly [104].
The studies on Hexanite-R ceramics based on the wurzite modification of boron
nitride [164] demonstrated that the oxidation kinetic curves (Fig. 17) were very
similar to that for B4C (Fig. 13). A rather high oxidation resistance is also exhibited
5.6 Materials 173
a
E -1 -
-2 -
I I
3O 1 2
t, h
Figure 17. Oxidation kinetics of wurzite BN Hexanite-R at: ( I ) 900°C; (2) 1000°C; (3) 1IOo"C, and
(4) 1200°C; reproduced from [2].
by a cubic boron nitride-based material having sphalerite structure (Elbor-R) and

materials that are a mixture of wurzite and sphalerite BN [165].
Graphitic boron nitride is hydrolyzed with humid air to form ammonia and boric
acid:
BN + 3 H 2 0 @ H3B03+ NH3. (33)
Boron nitride is resistant to hydrogen and sulfur dioxide; it reacts with fluorine at
room temperature by the reaction
2 BN + 3 F2 @ 2 BF3 + N2. (34)
The absence of interaction with other halogens and some halides can be used for
removal of boron, boron carbide and other impurities from boron nitride [98].
5.6.4 Carbonitrides
5.6.4.1 Silico-carbonitrides
The few investigations available for the new precursor-derived silico-carbonitrides
do not permit their corrosion behavior to be assessed in detail. Most studies of
interest here deal with thermal stability only.
The issue of thermal stability is important for these phases, because precursor-
derived ceramics are amorphous and have stoichiometries where detrimental
structural and chemical changes may occur on crystallization. The highest
temperatures recorded so far come from phases in the system Si-B-C-N
(2000°C [166]).
For the carbonitrides, the stoichiometry of the amorphous product is
Si3+.TN4Cs+p [167] and thus in terms of possible crystallization products there is
always C in addition to Si3N4 and Sic. In precursor-derived Si-C-N-ceramics,
micro/nanocrystalline Si3N4 and Sic appear at M 1400-1500°C [168]. Thermal

stability experiments show that in Si-GN-ceramics the reaction
+
Si3N4 3 C @ 3 Sic + 2N2 (35)
takes place at about the thermodynamically predicted temperature (M 1405°C at
1 bar N2). The effectivity of this reaction is well known from the sintering of
Si3N4-SiC composites [ 1691. Its suppression by oxidation [I 701 indicates an internal
pressure build-up of several bar.
In ceramics from polyborosilazanes, on the other hand, ordering effects are
demonstrated only at 1700°C and strong crystallization at 1800°C [171]. The
crystallization of Si3N4 at T 2 1750°C in this system was also confirmed by
Baldus et al. [ 1721 who reported even higher crystallization temperatures, some-
times exceeding 1900°C.
In boron-containing systems the initial Si3N4-crystallization without decomposi-
tion may indicate that the carbon is not present in the form of a reactive free carbon:
boron may keep carbon in some other (amorphous?) form to suppress or to retard
the kinetics of reaction (35).
It is a very old experience that the crystallization of amorphous substances is
influenced by numerous parameters [ 1731. Therefore we consider it very dangerous
to take the observed crystallization behavior as a fixed value.
Despite a lack of data on active oxidation behavior some thermal stability tests
give good hints. It relates to the oxygen impurities of the precursor material. The
mass loss observed and clearly attributed to oxygen removal from silicocarbo-
boronitrides [166] starts from T M 1300°C. Thus in an environment with low
P ( 0 2 ) , at which no protective layer is formed, we will have a substantial loss by
active oxidation.
In passive oxidation some very promising data for the new ceramics have been
given [170,174, 1751. It is indicated that silicocarbonitrides behave as good as
pure Si3N4 [170] and that the boron-containing types are even more resistant
[175]. A t, value calculated from the k, given by [174] would be over 300 years at
1550"C!
We are not aware of hot corrosion testing of precursor-derived ceramics. A
prediction rests upon the argument that for those which contain boron a top
scale compositions in the Na-B-Si-0-system should form, that is a classical glass
forming system with low melting points and viscosities [176]. Apart from the hot
corrosion degradation by itself it is completely unknown whether this will induce
crystallization of the amorphous precursor-derived ceramics.
5.6.4.2 Titanium Carbide, Nitride, and Carbonitrides

Carbides and nitrides of transition metals are usually hard and some of them are
used as hard materials. Tungsten carbide, which was briefly described above, is a
major component of cutting tools and wear-resistant materials. Titanium carbide,
TIC, and nitride, TiN, are two other major hard materials that are widely used in
tools. They also find application as hard coatings (in particular, TiN [177]) and
components of ceramic matrix composites [178]. Both TIC and TiN show a very
5.6 Materials 175
wide stoichiometry range and almost never exist in the stoichiometric composition.
Due to their structural identity, they also form solid solutions in the Ti-C-N system.
Oxygen can be dissolved in the material until saturation of carbon or nitrogen
vacancies [179] with no changes in the f.c.c. (NaC1) crystal structure. In all cases,
whether we consider TIN, Tic, and TiC,N, materials, the oxidation behavior is
similar and these rutile-formers will be analyzed together.
The peculiarities of the defect structure of TIC and TiN allow for oxygen dissolu-
tion on the first stage of oxidation. Titanium oxides appear later, when the material
structure is saturated with oxygen. Anatase was observed at lower temperatures
and/or short oxidation times ( 5 min). With increasing time and temperature, rutile
is formed as the only oxidation product [108].
Oxidation of TIN powders [180], thin films [177], and dense bodies [I081 was
extensively investigated. It was shown that oxidation of TiN, following a logarith-
mic law, starts at room temperature. Above 500°C the oxidation rate increases and
the kinetics follow a parabolic law. Oxidation of a TiN powder prepared by plasma
vapor phase synthesis and having the specific surface area of 46 m2 g-' is noticeable
above e2250"C and results in the formation of Ti02 as anatase and brookite at lower
temperatures, and as rutile at higher temperatures [181].
The scales formed at oxidation of sintered TIN in air can be divided into two
major types with the following layer systems: At low temperatures ( 5800°C)
TiN-dense Ti02-air; at high temperatures (1000°C) TiN-porous Ti02-dense
Ti02-air.
The latter can develop into a multilayered scale during the long-term oxidation
above 1000°C [181]. Growth of such rutile scales was based on the simultaneous
diffusion of titanium and oxygen ions through the oxide layer in opposite directions.
We can suppose that the fine grain inner layer grows by inward diffusion of
oxygen, while the coarse grain outer layer grows by outward cation diffusion.
Oxidation of TIC is similar to that of TIN. Many carbides of transition metals
form carbon at the carbide-oxide interface. In the case of TIC, no carbon was
formed [182]. Oxidation of titanium and carbon at 1000°C occurs with the same
rate [183].
Both TiN and TIC are widely used in ceramic matrix composites for improving
electrical conductivity and mechanical properties [ 178, 184,1851. The formation of
a rutile scale was observed on the surfaces. The growth of such a scale should be
related to the diffusion of titanium to the surface of the composite and its oxidation
according to the reaction:
2TiN + 2 0 2 = 2Ti02 + N2. (36)
The protective properties of rutile are fairly low compared to those of alumina or
silica. Thus, the presence of TiN or TIC limits the high-temperature applications
of the above composites. At small amounts and particle size of TIN in silicon nitride
ceramics, a continuous silicate film can be formed, covering TIN particles and
protecting them from further oxidation [ 1781.
Thermodynamic analysis of the chemical equilibria in water-carbide systems
[loo] indicated that carbon, corresponding oxides, CH4, C 0 2 , CO, and H2 are the
main products of hydrothermal corrosion of metal carbides in the temperature
range 300-1000°C. There exists a possibility for selective etching of T i c and carbon
layers formation. A low H 2 0 to carbide molar ratio results in the formation of
carbon layers. Lower pressure is beneficial here. A high H 2 0 to carbide molar
ratio means that under equilibrium conditions carbon can be oxidized to CO and
C02. Experimental studies showed that no protective oxide layer was formed
upon hydrothermal corrosion of carbides of transition metals. Raman spectra
obtained from the hydrothermally treated TIC powders demonstrated the presence
of graphite. Anatase crystals were also observed on the surface of TIC. With grow-
ing temperature, anatase was partly transformed to rutile [loo]. This is in agreement
with other published data on corrosion of TIC and TiN [186].
5.6.5 Titanium Diboride

TiB2 is widely used as a high-temperature material, particularly in coatings [187] and
as a component of composites [ 1881. According to thermodynamic calculations for
the Ti-B-0 system under oxygen partial pressures close to the ambient one and suf-
ficiently high partial pressures of B203, Ti02 is a stable reaction product. Lower
titanium oxides are stable only at reduced oxygen pressure, and TiB only at
extremely low values of PBl0,. Thus, the main products of TiB2 oxidation in air
or oxygen are Ti02 and B2O3. As the oxygen partial pressure is reduced, Ti305
becomes more stable. It was experimentally detected at the scale-boride interface
at an oxygen partial pressure of 14.66 kPa [189].
TiB2 oxidation above 700°C was approximated by the parabolic rate law (9) [190].
Deviations from the parabolic oxidation start at temperatures of 950-1 100°C when
vaporization of B2O3 becomes noticeable. A cubic law was suggested [191,192], but
we assume that a paralinear law (Fig. 7) should better describe the kinetics, similar
to the case of BN and B4C.
However, unlike for boron carbide and nitride, the kinetics becomes parabolic
again at higher temperatures (>12OO0C). This can be explained by the fact that
when B203 is completely evaporated the process is controlled by the diffusion
through the titanium dioxide film.
Impurities and metal binders were found to improve the oxidation resistance of
TiB2 [193]. This was explained by the fact that they accelerate the sintering of a
rutile scale, thus increasing its protective properties.
The oxidation behavior of an equimolar TiB2-A1N composite has been studied in
the temperature range 700-1300°C in flowing oxygen [194]. Throughout this
temperature range, the oxidation rate is governed by diffusional mechanisms, but
three temperature domains can be identified. In the range 7O0-85O0C, only the
oxidation of titanium diboride occurs. Calculations assuming a cubic diffusional
kinetic law result in the apparent activation energy of 180 kJ mol-' . In the tempera-
ture range 900-1000°C, the oxidation of TiB2 still dominates over oxidation of AlN,
but the formation of aluminum borates A14B209and A118B4033occurs and leads to
a more complex behavior. An apparent activation energy of 80kJmol-' was
determined for this stage. Above 1050"C, aluminum nitride oxidation becomes
References 177
more significant and the oxidation kinetics can be described by a parabolic law with
an apparent activation energy of 472 kJ mol-' .
References
1. R. J. Fordham, D. J. Baxter, and T. Graziani (Eds), Corrosion of advanced ceramics. Key Eng.
Muter. 1995, 113, 256.
2. Y . G. Gogotsi and V. A. Lavrenko, Corrosion o f High-Performance Ceramics, Springer
Verlag, Berlin, 1992, p. 181.
3. K. G. Nickel (Ed.), Corrosion of Advanced Ceramics Measurement and Modelling. NATO
-
AS1 Series E. Vol. 267, Kluwer, Dordrecht, 1994, p. 467.

4. R. Telle and P. Quirmbach (Eds), Korrosion und VerschleiJ von keramischen Werkstoffen,
Deutsche Keramische Gesellschaft, Koln, 1994, p. 359.
5. R. E. Tressler and M. McNallan (Eds), Corrosion and corrosive degradation of ceramics,
Ceram. Trans. 1990, 10, 493.
6. D. Behrens, Dechema Corrosion Handbook, Vol. I , VCH, Weinheim, 1987.
7. D. M. Mieskowski, T. E. Mitchell, and A. H. Heuer, J . Am. Ceram. SOC.Comm. 1984, 67,
17-18.
8. K. G. Nickel, H. L. Lukas, and G. Petzow, in The SGTE Casebook. Thermodynamics at Work,
K. Hack (Ed.), Institute of Metals, London, 1996, pp. 163-175.
9. J . R. Frade, in Corrosion of Advanced Ceramics - Measurement and Modelling, K. G. Nickel
(Ed.), Kluwer, Dordrecht, 1994, pp. 47-58.
10. G. H. Frischat, Glastechn. Ber. 1979, 52, 143-154.
11. C. Wagner, J . Appl. Phys. 1958, 29, 1295-1297.
12. M. W. Chase et al., J . Phys. Chem. Re$ Data, 1985, 14 (Suppl. I).
13. K. Hack, Metall. Trans. 1990, 21B, 1013-1023.
14. S. C. Singhal, Ceram. Int. 1976, 2, 123-129.
15. K. G. Nickel, J. Europ. Cerum. Soc. 1992, 9, 3-8.
16. H. E. Kim and D. W. Readey, in Silicon Carbide Transactions, Ohio State University, Wester-
ville, OH. 1989, pp. 301-312.
17. K. G. Nickel et a/., Powder Metall. Int. 1989, 21, 29-34.
18. N. S. Jacobson, High- Temperature Durability Considerations f o r HSCT Combustor, Report:
3162, NASA, Cleveland, OH, 1992.
19. B. E. Deal and A. S. Grove, J . Appl. Phys. 1965,36, 3770-3778.
20. E. A. Irene, CRC Crit. Rev. Solid State Muter. Sci. 1988, 14, 175-223.
21. K. G. Nickel, in Corrosion of Advanced Ceramics - Measurement and Modelling, K. G. Nickel
22. P. 0. Kall, M. Nygren, and J . Persson, in Corrosion oj'Advanced Ceramics - Measurement and
Modelling, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 73-84.
23. E. J. Opila and R. E. Hann, J . Am. Ceram. Soc. 1997,80, 192-205.
24. R. Schober, G. Konig, and D. Grimm, in Technische Keramische Werkstofe,J. Kriegesmann
(Ed.), Deutsche Keramische Gesellschaft, Koln, 1993, Chapter 6.1.6.2 (16 pp.).
25. M. Cable, in Corrosion of Advanced Ceramics - Measurement and Modelling, K . G. Nickel
26. D. J. Baxter, in Corrosion of Advanced Ceramics Measurement and Modelling, K . G. Nickel
-

27. M. Maeda, K. Nakamura, and A. Tsuge, J. Muter. Sci. Leu. 1989,8, 195-197.
28. E. Opila, J . Am. Ceram. Soc. 1995, 78, 1107-1 110.
29. R. Telle, K. G. Nickel, and P. Quirmbach, in Korrosion und Verschleg von keramischen Werk-
stoffen, R. Telle and P. Quirmbach (Eds), Deutsche Keramische Gesellschaft, Koln, 1994,
pp. 347-352.
30. G. C. Padgett, in Proc. of 1st ECERS, Maastricht, 1989.
178 5 Corrosion of Hard Material.)
31. J. S. Association, in Japanese Industrial Standard JIS R 1609-1990, Japanese Standard

Association, Japan, 1990.
32. J. J. Kubler and R. Morrell, A S M E , 1993, 93-GT-158, 1-5.
33. K. G. Nickel et al., Ceram. Forum Int. 1994, 71, 624631.
34. K. G. Nickel and Y. G. Gogotsi, Key Eng. Muter. 1996, 113, 15-30.
35. R. Morell, Handbook of Properties of Technical & Engineering Ceramics, HMSO, London,
1987.
36. M . F. Ashby and D. R. H. Jones, Engineering Materials I , Pergamon, Oxford, 1980, p. 279.
37. H. Hoppert, H. Rasch, and E. Litzow, in Korrosion und VerschleiJ von keramischen Werkstofen,
R. Telle and P. Quirmbach (Eds), Deutsche Keramische Gesellschaft, Koln, 1994, pp. 98-106.
38. T. Graas et al., Ceram. Forum Int. 1997, 74, 723-726.
39. A. Kailer, Y. G. Gogotsi, and K. G. Nickel, in Proc. EuroMat ’97,Maastricht, 1997, pp. 171-174.
40. Y. G. Gogotsi, A. Kailer, and K. G. Nickel, Muter. Res. Innovat. 1997, 1, 3-9.
41. Y. L. Orlov, The Mineralogy of the Diamond, Wiley, New York, 1977.
42. J. E. Field, in Diamond and Diamond-like Films and Coatings, R. E. Clawing et al. (Eds),
Plenum, New York, 1991, pp. 17-35.
43. R. J. Wedlake, in The Properties qf Diamond, J. E. Field (Ed.), Academic Press, New York,
1979, pp. 501-535.
44. T. Evans, in The Properties of Diamond, J. E. Field (Ed.), Academic Press, New York, 1979,
pp. 403424.
45. S. Kumar et al., Diamond Rel. Muter. 1996, 5, 12461248.
46. E. A. Gulbransen and J. S. A,, in Heterogeneous Kinetics and Elevated Temperatures, G. R.
Belton (Ed.), Plenum, New York, 1970, pp. 181-208.
47. K. Tankala and T. Debroy, in Proc. New Diamond Science and Technology, R. Messier et al.
(Eds), MRS, Pittsburgh, 1991, pp. 827-831.
48. W. Zhu et al., in Proc. New Diamond Science and Technology, R. Messier et al. (Eds), MRS,
Pittsburgh, 1991, pp. 821-826.
49. F. Schroder (Ed.), Chemisches Verhalten von Graphit. Gmelins Handbuch der Anorganischen
Chemie, Vol. Kohlenstoff, Teil B, Lieferung 3, Verlag Chemie, Weinheim, 1968, pp. 789-809.
50. F. Marcuccilli, Etude exphrimentale et thhoretique de la structure et de la rhactivite‘ de suies
issues de moteur diesel, Dissertation, Universiti de Haute-Alsace, 1993, p. 212.
51. Y. G. Gogotsi and K. G. Nickel, Ceram. Eng. Sci. Proc. 1997, 18, 747-754.
52. V. Haase, in Gmelins Handbuch der Anorganischen Chemie, F. Schroder (Ed.), Springer, Berlin,
1986, pp. 325-395.
53. K. G. Nickel and P. Quirmbach, in Technische Keramische Werkstofie,J. Kriegesmann (Ed.),
Deutscher Wirtschaftsdienst, Koln, 1991, Chapter 5.4.1.1, pp. 1-76.
54. J. R. Blachere and F. S. Pettit, High Temperature Corrosion of Ceramics, Noyes Data Corpora-
tion, Park Ridge, NJ, 1989, pp. 188.
55. W. Prechtl, Keram. Z. 1993, 45, 197-201.
56. E. Gehrke et al., Mechanical, Thermal and Chemical Properties of’a Sic-material, Report VI l /
94, IWM, Freiburg, 1994.
57. K. Wendland et al., in Korrosion und Verschleij von keramischen Werkstoffen, R. Telle and
P. Quirmbach (Eds), Deutsche Keramische Gesellschaft, Koln, 1994, pp. 107-1 13.
58. Y. G. Gogotsi and M. Yoshimura, M R S Bull. 1994, 10, 3945.
59. Y. G. Gogotsi et al., Diamond Rel. Muter. 1996, 5, 151-162.
60. M. Yoshimura, J. Kase, and S. Somiya, in Proc. 2nd Int. Symp. Ceramic Materials and
Componentsf o r Engines, W. Bunk and H. Hausner (Eds), Germany, 1986, pp. 529-536.
61. M. Yoshimura et al., in Corrosion and Corrosive Degradation ojCeramics, R. E. Tressler and
M. McNallan (Eds), American Ceramic Society, 1990, pp. 337-354.
62. S. Kitaoka et al., J. Am. Ceram. Soc. 1994, 77, 1851-1856.
63. R. 0. Fournier and J. J. Rowe, Am. Min. 1977,62, 1052-1056.
64. R. Sangiorgi, in Corrosion of Advanced Ceramics, K. G. Nickel (Ed.), Kluwer, Dordrecht,
1994, pp. 261-284.
65. N. S. Jacobson, J. Am. Ceram. Soc. 1986, 69, 7482.
66. N. S. Jacobson, J. L. Smialek, and D. S. Fox, in Handbook qf’ Ceramics and Composites, N . P.
Cheremisinoff (Ed.), Dekker, New York, 1990, pp. 99-136.
References 179
67. N.S. Jacobson, J. Am. Ceram. SOC.1993,76, 3-28.

68. N. S.Jacobson, J. L. Smialek, and D. S. Fox, in Corrosion of Advanced Ceramics Measure-
-
ment and Modelling, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 205222.
69. H.J. Grabke, in Corrosion of Advanced Ceramics - Measurement and Modelling, K. G. Nickel
70. F. S. Pettit, G. H. Meier, and J. R. Blachere, in Corrosion of Advanced Ceramics - Measure-
ment and Modelling, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 235-248.
71. N.S . Jacobson, Oxid. Metals, 1989, 31, 91-103.
72. N. S.Jacobson, J. Am. Ceram. SOC.1985,68,432-439.
73. C. Berthold and K. G. Nickel, Key Eng. Muter. 1997, 132-136, 1588-1591.
74. C. Berthold and K. G. Nickel, J . Europ. Ceram. SOC.1998, in press.
75. J. Schlichting, Werkstof Korros. 1975, 26, 753-758.
76. N. S. Jacobson, C. A. Stearns, and J. L. Smialek, Adv. Ceram. Muter. 1986, 1, 154-161.
77. D. S. Fox, N. S. Jacobson, and J. L. Smialek, Hot Corrosion of Ceramic Engine Materials,
Report DOE/NASA/50111-2, DOE/NASA, Cleveland, 1988.
78. D. S. Fox and J. L. Smialek, J . Am. Ceram. Soc. 1990, 73, 303-31 1.
79. K. N. Lee e f al., 1995.
80. K. N. Lee and R. A. Miller, J . Am. Ceram. SOC.1996, 79, 62G-626.
81. S. Y.Ip, M. J. McNallan, and D. S. Park, J . Am. Ceram. SOC.1992, 75, 1942-1948.
82. G. S. Fischman, A. Zangvil, and S. D. Brown, Adv. Ceram. Muter. 1988, 3, 66-70.
83. G. W. Hallum and T. P. Herbell, Adv. Ceram. Muter. 1988, 3, 171-175.
84. G. A. Schneider, K. G. Nickel, and G. Petzow, in Physics and Chemistry of Carbides, Nitrides
and Borides, NATO ASI Series E: 185, R. Freer (Ed.), Kluwer, Dordrecht, 1990, pp. 387401.
85. E. Fitzer and R. Ebi, in Silicon Carbide, K. L. Marshall, T. W. Faust, and C. E. Ryan (Eds),
South Carolina University Press, Columbia, SC, 1974, pp. 320-328.
86. K. Schwab, Korrosion poroser Sic-Keramiken unter Rujjilterbedingungen. Dissertation, Uni-
versity Tiibingen Mineralogy, 1996, p. 170.
87. T. Narushima et al., J . Am. Ceram. SOC.1993, 76, 2521-2524.
88. L. U. J. T. Ogbuji and E. J. Opila, J . Electrochem. SOC.1995, 142, 925-930.
89. J. A. Costello and R.E. Tressler, Ceram. Int. 1985, 11, 3944.
90. J. W. Fergus and W. L. Worell, in Corrosion and Corrosive Degradation of Ceramics, R. E.
Tressler and M. McNallan (Eds), Martinus Nijhoff, Boston, 1990, pp. 43-51.
91. R. M. Persoons, J. Luyten, and A. D. Stalios, in Corrosion of Advanced Ceramics, K. G. Nickel
92. M. McNallan, M. Van Roode, and J. R. Price, in Corrosion and Corrosive Degradation of
Ceramics, R. E. Tressler and M. McNallan (Eds), Martinus Nijhoff, Boston, 1990, pp. 445-
467.
93. D. P. Butt, R. E. Tressler, and K. E. Spear, J. Am. Ceram. Soc. 1992, 75, 3257-3267.
94. D. P. Butt, R. E. Tressler, and K. E. Spear, J. Am. Ceram. Soc. 1992, 75, 3268-3277.
95. H.Nickel and R. Forthmann, in Corrosion of Advanced Ceramics, K. G . Nickel (Ed.), Kluwer,
Dordrecht, 1994, pp. 99-1 16.
96. P. S. Kisly et al., Naukova Dumka, Kiev, 1988.
97. G. I. Postogvard, Y. G. Parkhomenko, and I. T. Ostapenko, Poroshkovaya Metall. 1983,8,
pp. 28-30.
98. T. Y.Kosolapova (Ed.), Nonmetallic Refractory Compounds, Metallurgiya, Moscow, 1985.
99. A. M.Pivak et at., in New Technological Processes in Powder Metallurgy, I. D. Radomyselsky
(Ed.), Institute for Problems of Materials Science, Kiev, 1986, pp. 31-35.
100. N. S. Jacobson, Y. G. Gogotsi, and M. Yoshimura, J . Muter. Chem. 1995.
101. F. Thevenot, in Proc. 1st ECERS, G. De With, R. A. Terpstra, and R. Metselaar (Eds),
Elsevier, Maastricht, 1989, pp. 2.1-2.25.
102. Y. L. Krutsky, G. V. Galevsky, and A. A. Kornilov, Poroshkovaya Metall. 1983,2, pp. 47-50.
103. V. A. Lavrenko et at., Oxid. Metals, 1976, 10, 85-95.
104. V.A. Lavrenko, Y. G. Gogotsi, and I. N. Frantsevich, Dokl. Akad. Nauk SSSR, 1984,27,114-
117.
105. V. A. Lavrenko and Y. G. Gogotsi, Oxid. Metals, 1988, 29, 193-202.
106. L. N. Yefimenko, E. V. Lifshits, and I. T. Ostapenko, Poroshkovaya Metall. 1987, 4,56-60.
107. Y. G. Gogotsi and V. V. Kovylyaev, Poroshkovaya Metall. 1989, 4, 71-75.

108. R. F. Voitovich, Oxidation of Carbides and Nitrides (in Russian), Kiev, 1981.
109. A. P. Pomytkin, Kinetics of High Tempetrature Oxidation of Titanium, Niobium, Chromium,
Boron and Silicon Carbides in Oxygen, Thesis, Academy of Science of USSR, Institute for
Problems of Materials Science, Kiev, 1976.
110. R. C. Weast (Ed.), Handbook of Chemistry and Physics, CRC Press, Boca Raton, 1987.
111. T. Narushima et al., Key Eng. Muter. 1996, 113, 99-104.
112. I. M. Low and X. S. Li (Eds), Key Eng. Muter. 1996, 114, 284.
113. I. M. Low (Ed.), Key Eng. Muter. 1998, 138, 696.
114. T. Yoshio et al., in Proc. Fall Meeting of the Ceramic Society of Japan, 1993, p. 206.
115. T. Nishida et al., in Proc. 10th Japan-Korea Seminar on Ceramics, 1993, pp. 385-388.
116. R. C. Sangster and G. Bar, Silicon, Suppl. B 5dl,8th edn., F. Schroder (Ed.), Vol. 15, Springer-
Verlag, Berlin, 1995, p. 255.
117. R. C. Sangster, P. Kampf, and U. Nohl, Silicon, Suppl. B 5d1, 8th edn, F . Schroder (Ed.), Vol.
15, Springer-Verlag, Berlin, 1995, p. 304.
118. D. E. Clark and B. K. Zoitos, Corrosion of Glass, Ceramics and Ceramic Superconductors,
Noyes Publications, Park Ridge, NJ, 1992, pp. 672.
119. H . Scholze, Glas, 3rd edn, Springer-Verlag, Berlin, 1988, pp. 342.
120. M. Shimada and T. Sato, in Corrosion and Corrosive Degradation of Ceramics, R. E. Tressler
and M. McNallan (Eds), American Ceramic Society, 1990, pp. 355-365.
121. K. G. Nickel, U. Daumling, and K. WeiBkopf, Key Eng. Mater. 1994,89-91,295-300.
122. T. Yoshio and K. Oda, in Corrosion and Corrosive Degradation of Ceramics, R. E. Tressler and
M. McNallan (Eds), American Ceramic Society, 1990, pp. 367-386.
123. K. Oda et al., J . Am. Ceram. SOC.1993,76, 1365-1368.
124. J. Vleugels and 0. Van der Biest, Key Eng. Muter. 1998, 138-140, 127-176.
125. D. S . Fox and N. S. Jacobson, J. Am. Ceram. SOC.1988,71, 128-138.
126. N. S . Jacobson and D. S. Fox, J. Am. Ceram. SOC.1988,71, 139-148.
127. D. S. Fox, N. S . Jacobson, and J. L. Smialek, in Corrosion and Corrosive Degradation of
Ceramics, R. E. Tressler and M. McNallan (Eds), Martinus Nijhoff, Boston, 1990, pp. 227-249.
128. C. Berthold, Degradation von Si,N, als Ventilwerkstoffdurch Oxidation und Salzschmelz-
korrosion bei Ein&tztemperaturen, Dissertation, University Tubingen, Applied Mineralogy,
1997, D. 142.
129. T. E. Strangman and D. S. Fox, in Proc. Erosion, Corrosion and Foreign Object Damage Eflects
in Gas Turbines, Rotterdam, 1994, pp. AGARD36/1-36/11.
130. C. Berthold and K. G. Nickel, in Proc. Werkstoff-Woche '96, F. Aldinger and H. Mughrabi.
(Eds), DGM, Frankfurt, 1997, pp. 577-582.
131. D. R. Clarke, J . Am. Ceram. SOC.1983,66, 92-95.
132. H. E. Kim and A. J. Moorhead, J. Am. Ceram. SOC.1990, 73, 3007-3014.
133. F. A. Oliveira, R. A. H. Edwards, and R. J. Fordham, Werkstoff.Korros. 1990,41,73&742.
134. F. A. Costa Oliveira, High Temperature Gaseous Corrosion of Silicon Nitride Ceramics,
Dissertation, TU Delft, JRC Petten, 1992, pp. 271.
135. J. E. Marra et al., J. Am. Ceram. SOC.1988,71, 1067-1073.
136. M. McNallan et al., in Corrosion and Corrosive Degradation of Ceramics, R. E. Tressler and
M. McNallan (Eds), Martinus Nijhoff, Boston, 1990, pp. 309-333.
137. P. Hsu, C. Park, and M. J. McNallan, J. Am. Ceram. SOC.1993,76, 1621-1623.
138. M. Maeda, K. Nakamura, and M. Yamada, J. Muter. Sci. 1990,25,379&3794.
139. A. H. Heuer and V. L. K. Lou, J. Am. Ceram. SOC.1990,73,2785-3128.
140. K. G. Nickel, Z. Fu, and P. Quirmbach, J . Eng. Gas Turb. Power, 1993, 115, 76-82.
141. J. E. Sheehan, J . Am. Ceram. SOC.Comm. 1982,65, 11 1-1 13.
142. W. L. Vaughn and H.G. Maahs, J . Am. Ceram. SOC.1990,73, 1540-1543.
143. T. Narushima et al., J. Am. Ceram. SOC.1994,77, 2369-2375.
144. H. Du, R. E. Tressler, and K. E. Spear, J. Electrochem. SOC.1989, 136, 3210-3214.
145. L. Ogbuji and J. Smialek, J. Electrochem. SOC.1991, 138, 51-53.
146. L. U. J. T. Ogbuji, J. Am. Ceram. SOC.1992, 75, 2995-3000.
147. L. U. J. T. Ogbuji, J. Am. Ceram. SOC.1995,78, 1272-1278.
148. L. U. J. T. Ogbuji, J . Am. Ceram. SOC.1995, 78, 1279-1284.
References 181
149. L. Ogbuji, in Proc. Microscopy of Oxidation 3, 1996, Cambridge, UK

150. K. L. Luthra, J . Am. Ceram. SOC.1991,74, 1095-1103.
151. K. L. Luthra, in Corrosion ofAdvanced Ceramics - Measurement and Modelling, K. G. Nickel
152. M. Billy, in Progress in Nitrogen Ceramics, F. L. Riley (Ed.), Martinus Nijhoff, Boston, 1983,
pp. 403419.
153. K. G. Nickel, Ceram. Int. 1997, 23, 127-133.
154. J. K. Patel and D. P. Thompson, Br. Ceram. Trans. J. 1988,87, 70-73.
155. Z. Fu-Rudolph, Oxidationsverhalten von additivhaltigen Si3N4-Werkstoffen,Dissertation, Max
Planck Institut fur Metallforschung, PML, 1992, p. 159.
156. D. R. Clarke, in Progress in Nitrogen Ceramics, F. L. Riley (Ed.), Martinus Nijhoff, Boston,
1983, pp. 421426.
157. M. Backhaus-Ricoult and Y. Gogotsi, Key Eng. Muter. 1996, 113, 81-90.
158. C. O'Meara and J. Sjoberg, J . Am. Ceram. SOC.1997, 80, 1491-1500.
159. E. Butchereit et al., in Proc. Werkstoflwoche '96, F. Aldinger and H. Mughrabi (Eds), DGM
Frankfurt, 1997, pp. 569-575.
160. H. Klemm et al., J . Am. Ceram. SOC.1996, 79,2429-2435.
161. H. Klemm, M. Herrmann, and C. Schubert, Ceram. Eng. Sci. Proc. 1997, 18, 615-623.
162. A. V. Kurdyumov and A. N. Pilyankevich, Phase Transformations in Carbon and Boron
Nitride, Naukova Dumka, Kiev, 1979.
163. G. V. Samsonov, 0. P. Kulik, and V. S. Polishchuk, Processing and Methods of Analysis of
Nitrides, Academy of Science of USSR, Kiev, 1978.
164. V .A. Lavrenko, T. G. Protsenko, and A. V. Bochko, Dokl. Akad. NaukSSSR,1975,224,877-
879.
165. A. V. Bochko et al., Sverkhtverdye Materialy, 1986, 1, 16-18.
166. T . Riedel et al., Nature, 1996, 382, 796-798.
167. T. Riedel, in Processing of Ceramics, Part 2, R. Brook (Ed.), VCH, Weinheim, 1996, Chapter
11, pp. 1-50.
168. Friess, M. et al., Key Eng. Muter. 1994, 89-91, 95-100.
169. K. G. Nickel et al., Adv. Ceram. Muter. 1988, 3, 557-562.
170. T. Riedel et al., Nature, 1995, 374, 526-528.
171. Funayama. 0. et al., J . Am. Ceram. SOC.1993,76,717-725.
172. H.-P. Baldus, M. Jansen, and 0. Wagner, Key Eng. Muter. 1994, 89-91, pp. 75-80.
173. W . Vogel, Struktur und Kristallisation der Glaser, VEB Deutscher Verlag fur Grundstoff-
industrie, Leipzig, 1965.
174. H. P. Baldus and G. Passing, Muter. Res. SOC.Symp. Proc. 1994, 346, 617422.
175. M. Jansen et al., in Proc. Werkstoff-Woche'96, F. Aldinger and H. Mughrabi (Eds), DGM,
Frankfurt, 1997, pp. 669-674.
176. G. W. Morey, J . SOC.Glass Technol. 1951, 35, 27G283.
177. H. G. Tompkins, J . Appl. Phys. 1991,70,3876.
178. Y .G. Gogotsi, J . Muter. Sci. 1994, 29, 2541-2556.
179. E. M. Fryt, in Microscopy ofOxidation, S . B. Newcomb and J. A. Little (Eds), Institute of
Materials, London, 1997, pp. 739-747.
180. Y. Sakka, S. Ohno, and M . Uda, J . Am. Ceram. SOC.1992,75,244248.
181. Y .G. Gogotsi, F. Porz, and G . Dransfield, Oxid. Metals, 1993, 39, 69-91.
182. S. Shimada and M. Kozeki, J. Muter. Sci. 1992, 27, 1869-1875.
183. M. Munster, Z. Elektrochem. 1959, 3, pp. 806-819.
184. A. Tampieri and A. Bellosi, J . Am. Ceram. SOC.1992,75, 1688-1690.
185. J. Mukerji and S. K. Biswas, J . Am. Ceram. SOC.1990, 73, 142-145.
186. M. Yoshimura, M. Hayakawa, and S . Somiya, Nihon Ceramikkusu, 1989,97, 1339-1347.
187. J . Desmaison, in Corrosion of Advanced Ceramics - Measurement and Modelling, K. G. Nickel
188. A. Bellosi, in Corrosion of Advanced Ceramics - Measurement and Modelling, K. G. Nickel
189. V. V. Svistunov, Investigation of Preparing and Oxidation Processes of Transition Metals
Melted Borides, Thesis, Sverdlovsk, 1975.
190. A. Lebugle and G. Montel, Rev. Int. Hautes Temp. Refract. 1974, 11, 231.
191. R. J. Irving and I. G. Worsley, J . Less-Common Metals, 1968, 16, 103.
192. A. Munster and G. Schlamp, 2. P h p . Chem. 1960,25, 116.
193. V. B. Voitovich, V. A. Lavrenko, and A. M. Adejev, Oxid. Metals, 1994, 42, 145-161.
194. V. Schneider et al., in Corrosion of Advanced Ceramics, R. J. Fordham, D. J. Baxter, and
T. Graziani (Eds), TransTech, 1996, pp. 49-58.
6 Interrelations Between the Influences of

Indentation Size, Surface State, Grain Size,
Grain-Boundary Deformation, and Temperature
on the Hardness of Ceramics
A. Krell
6.1 Introduction
This contribution describes the most important influences on the hardness of
materials from the viewpoints of the measuring approach and of probable chances
for further progress in the development of new materials. Therefore, the discussion
will be focused on questions that arise when precise measurement has to be achieved
in the investigation of hard materials.
- Which testing load should be applied? Is there a real size effect to be
expected from basic properties of the material (or is it an artifact of the measure-
ment?)?
- In investigations of new hard materials, the size of available samples is often
small, and many people prefer microhardness approaches. Can grain size effects
be recorded independent of the applied load?
- How does the hardness depend on the surface preparation? Do ground ceramics
(which are important in technical use) exhibit the same hardness as polished
samples (often preferred for tests in development procedures)?
- In ceramics, the dependence of the strength on flaws is a well-known
phenomenon associated with stress concentrations that cannot relax by plastic
deformation in these brittle materials. Does hardness testing and the development
of sintered materials with an increased hardness also have to consider flaw
populations?
- Is the intensively investigated activity of lattice processes inside the grains
(dislocations, twins) the one significant contribution to the inelastic

deformation at an indentation site at room temperature? What is the role of
grain boundaries?
- Which kind of temperature dependence is to be expected in different temperature
ranges? Is there a different impact of the microstructure on the hardness at, for
example, 400°C or at 1400°C? Which elements of the microstructure are most
important for obtaining a high hardness in the different ranges of the
temperature?
It is not intended here to present new evidence for special micromechanisms.
Instead, the available knowledge about such processes is used to answer the
questions outlined above.
184 6 Interrelations Between Influences
6.2 The Assessment of Residual Porosity and Flaw

Populations: A Prerequisite for any Hardness
Investigation
Whereas the influence of the porosity on the hardness is not an object of general
controversy, it is often neglected both in characterizing and in materials development
investigations. In fact, the grain size effect on the hardness of a material gives little
information when samples with relative densities of 96 and 99% are compared.
Figure 1 shows that the grain size effect may be not only obscured but offset by
porosity phenomena: a small porosity of 1-2% can affect the hardness more than
a decrease in the grain size from 2 to 0.5ym. Therefore, it is imperative to use
specimens with relative densities 299% if data are required to characterize the
material independent of porosity influences.
With the large influence of even small amounts of porosity shown in Fig. 1 and
with a surprisingly large contribution from grain boundaries to the inelastic defor-
mation at an indentation site (demonstrated below), it is not surprising to find a
strong influence of many kinds of defects on the hardness. Individual pores which
Figure 1. The Vickers hardness (testing load 1 kg) of fine-grained sintered alumina ceramics
prepared by a sol/gel approach starting with boehmite (y-seeding).
are so small that they are invisible in the optical equipment of hardness testers,
clusters of pores, microcracks, and other defects located within the subregion of
the material affected by the indentation, all promote the penetration of an indenting
body in one way or another.
Therefore, the development of new hard materials requires the preparation of
phases with small microplastic deformability of the crystal lattice, and reduced
grain sizes in the sintered polycrystals, and highly perfect microstructures. For
example, sintered alumina ceramics which have a submicrometer microstructure
and associate this small grain size with a macrohardness of 20-24 GPa (compared
with 17-19 GPa in conventionally sintered corundum) were obtained only
when new processing approaches became available, which could reduce the flaw
population [ 11.
6.3 Theoretical Considerations

6.3.1 The Role of the Lattice and of Grain Boundaries in the
Inelastic Deformation at an Indentation Site in Sintered Hard
Materials
On a microscale, the permanent plastic deformation at an indentation site in brittle
ceramics is associated with similar deformation structures as in metals. The dis-
location density is only 106-107 cm-* in virgin pressureless sintered alumina
(A1203),an extremely low value which should make microplastic deformation on
indentation very difficult. However, transmission electron microscopy (TEM)
investigations in sa phire (single crystal A1203) revealed a very high dislocation
density of 101'-10''cm-2 and even more after indentation at only 50g [2]. This
high degree of plasticity under the tip of the indenter in the brittle material is
possible because the highly localized stress field exceeds the high yield stress of
the ceramic (Table 1: 4 8 G P a for single and polycrystalline A1203 at room
temperature [3-5]) while macroscopic fracture is avoided due to the extreme spatial
concentration of the stress field.
It is important to understand that small finite crack systems generated beyond
some critical stress increase the elastic compliance at the site; contrary to incorrect
interpretations they do, however, not significantly change the area of the material
exposed to the (constant) load of the indenter. Hence, the size of the permanent
indent and the conventionally measured hardness remain unaffected.
It is generally assumed that the higher hardness of most polycrystals compared
with the single crystalline state, and the increase of the hardness at reduced grain
sizes, has to be attributed to hindrance of the movement and the growth of dis-
locations and twins by grain boundaries. By qualitative observations, the generation
of dislocations and twins at the indentation site has been investigated repeatedly, but
no attempt has been made to describe theoretically the relationship between
deformation structures in the lattice, its hardness, and the grain size dependence
of these processes. Without t h s knowledge, it is not surprising that even basic

questions like the dependence of the hardness on the testing load (i.e. on the size of
the indent) are still a subject of controversy. Considering the very complex dislocation
and twin structures at an indentation site in ceramics, it is not possible to describe
precisely this situation theoretically, and probably any model has to be a phenomeno-
logical one. Such a model describing the indentation size effect (ISE) independent of
special micromechanisms, is introduced in Section 6.3.2 and may be used, for
example, for extrapolation purposes when very small indents have to be discussed
(e.g. on investigating surface properties). Even with small loads of 1&100mN (1-
log), the length of Vickers diagonals in A1203 is 1-3pm, and the size of the
associated plastic zone is rather large (x4-12 pm) compared with today’s most inter-
esting Jine-grained microstructures. At further reduced loads, unfortunately, even
advanced indentation techniques often do not give reliable data, and a theoretically
supported extrapolation towards smaller sizes can be useful.
Surprisingly, even less experimental and theoretical attention has been paid to
another mechanism which should contribute to the inelastic deformation addition-
ally to the role of dislocations: the activity of grain boundaries. Table 1 compares
the critical stresses for microplastic deformation of the crystal lattice and for
grain boundary fracture. There is no doubt that the compressive yield stress and
the grain boundary strength are similar at least within an order of magnitude. There-
fore, stresses at an indentation site which exceed the yield stress of 4-8 GPa should
also exceed the grain boundary strength of 2-8 GPa.
The key idea for investigating the question about the amount of grain boundary
contribution to the generation of a permanent indent at room temperature is out-
lined in Fig. 2: it is assumed that the grain size effect in the hardness of polycrystals
and the influence of the testing load (the indentation size) are analogous features.
Considering a finite volume within which dislocations or twins can grow, move,
or multiply, a limitation of this space should reduce the microplastic deformability
of this segment of material (increasing its hardness). It seems unimportant whether
this limitation is due to close spacing of grain boundaries (lower part of Fig. 2) or a
small spatial extension of the plastically deformed zune (limitation of the stress field)
at the tip of the indenter (demonstrated in the upper part of Fig. 2 for a single crystal
where the ISE does not interfere with other mechanisms that might affect the
mobility of the dislocations).
Table 1. Comparison of critical stresses for microplastic deformation of the crystal lattice and for
grain boundary fracture in AI2O3at room temperature [6].
Compressive yield stress estimations from Grain boundary strength, ngb,

for sintered alumina (grain size D = 1-
5 pm) with grain boundary fracture
toughness [7] FZ 1.5 MPam’’’, size of
defects (e.g. pores at triple junctions) =
a
Dynamic shock wave Indentation analysis
m-
I
investigations
Sapphire [4] 4 GPa Sapphire [5] 4.5 GPa U
Sintered A 2 0 3[3] 6 GPa Sintered A 2 0 3 [5] 7.5 GPa f7gb = 2-8 GPa
I I
I---
grain boundary contrit,,,,. 1Id.nr :tri I?51 7e 6 necr
II
-
-
% -
,
to deformat grain size effect

-- --_
at constant load ( d a m zone size)
Grain Size in Sintered Polycrystals

L I
Figure 2. The limitation of the free space available for dislocation or twin activity by the spatial size
of the plastic zone in a single crystal (upper curve/upper x-axis), and by the grain size of polycrystal-
line microstructures (lower curve/lower x-axis) [6].
Under this condition, the size effect (load effect) as measured in single crystals
simulates the grain size influence of the hardness in polycrystalline materials.
In the mean time, this simulation of the grain size influence in the hardness of poly-
crystals by the indentation size effect (load influence) in a single crystal applies under a
very interesting condition: it excludes, in the single crystal, all possible contributions
from grain boundaries. Therefore, the dzflerence between the plastic zone size effect
(load effect) measured in a single crystal (e.g. in sapphire) and the real grain size
effect observed in the sintered polycrystals (e.g. sintered corundum) gives a measure
that characterizes the amount of grain boundary contributions to the deformation
that governs the hardness at room temperature (experimental data c.p. ##6.5, Fig. 8).
To be able to compare grain size and indentation size influences, the size of the
plastically deformed zone must be known. This information comes from TEM
investigations where an approximately constant ratio 2RP1/2aM 4-5 between the
plastic zone size 2RP1and the length of the Vickers diagonals 2a can be derived for
single as well as polycrystalline ceramics and for hard materials with fundamentally
different bonding like (ionic) alumina [2] and (covalent) S i c [8].
Together with the measurements of the ISE in sapphire and of the grain size
dependence of the hardness in available sintered microstructures with grain sizes
20.2 l m , the theoretical extrapolation will give a forecast of the probable hardness of
nanostructured alumina ceramics. The model also explains the observed influence of
the grain size on the measured load effect and demonstrates why reliable information
about the grain size dependence of the hardness requires the use of larger testing loads.
6.3.2 Quantitative Understanding the Load Effect on the Hardness:

Theoretical Considerations Compared with Single Crystal Data
Considering the very complex dislocation and twin structures at an indentation site
in ceramics, it is neither possible to describe precisely this situation nor would it be
straightforward to ask for a model that depends on special configurations. Hence,
the consideration has to be a phenomenological one, and it cannot be intended to
justify one or another special deformation mechanism. Instead, the limited scope
of the question to be answered here is whether the dislocation idea will generally
give a size effect which agrees with the experimental observation, or if some ideas
have to be rejected because of obvious disagreements.
For example, people are often upset when sometimes a strong size (load) effect is
observed - which disappears in other microstructures of the same material (or
simply on measuring on surfaces prepared in another way). Therefore, it seems
interesting to ask whether dislocation activity (or an associated deformability that
changes with increasing load) can introduce an indentation size effect that possibly
depends on the microstructure.
With the observed high dislocation and twin densities at the indentation site [2], an
idea proposed by Upit and Varchenya [9] is adopted here: microscopic forces (or
energies) required to generate and to move dislocations depend on the size of (e.g.)
dislocation half loops (and, therefore, on the space available for their formation),
and cause the observed size effect of the hardness. A definition of the size of the plas-
tically deformed zone 2RpIbeneath indentations in ceramics was given by Perrott [lo]
together with an analysis of the local stress field assuming a ratio 2Rp1/2aM 4
(2a = length of Vickers diagonals). It is important to understand that, neglecting
grain boundary contributions in polycrystals, the microplastic deformability at the
indentation site is governed by the local compressive yield stress of each of the
grains. This yield stress of individual grains in a polycrystalline microstructure
should be similar to the yield stress of a single crystalline volume when, at small testing
loads, the plastic zone size 2RP1in the single crystal equals the size of the grains in the
considered polycrystal.
The typical 'as sintered' structure of grains in sintered alumina consists of aplurality
of grains without any lattice defects (elements of plastic deformation); only in a few
grains can single dislocations, dislocation loops, and dislocation networks be
observed [l l-121. Therefore, as has already been mentioned, at the beginning of an
indentation process the average dislocation density is 106-107 cm-2 or even less,
and it will be rather hard to initiate microplastic deformation. However, at a
small load of only 0.5 N Hockey [2] observed regions with a dislocation density at
the upper limit of resolution (10'1-10'2 cm-*), and it is probable that the deforma-
tion process has arrived at a state where strain hardening (if existing at all) has
already been passed through.
Then, on increasing the load further, fundamental principles of solid state physics
suggest that the further extension of the plastically deformed zone in a single crystal
should occur much more easily by growth or by multiplication of pre-existing
elements of lattice deformation (dislocations, twins) than by the more difficult
generation of new dislocations in the virgin material around. Hence, on increasing
the load (the size of the indent), the microplastic deformability increases (the hard-
ness decreases) because the more easy growth of pre-existing dislocations becomes
more important and the role of the (difficult) generation of new dislocations
decreases.
To simplify the mathematical procedure, a characteristic area A e (2Rp1)2is
introduced. At some load PI, the microplastic deformability S can be defined as
an inverse local compressive yield stress:
S = (dP/dA)-' (la)
the smaller the increase in the load required to produce some incremental growth
dA = A - A l of the plastic zone, the larger is the deformability.
With different dislocation structures in the virgin ceramic and generated during
indentation as outlined above, the deformability may change with proceeding
penetration. Without hypothesizing any mechanism or specific character of this
change, two different symbols Si and 6,distinguish the following situations.
For the preceding penetration at loads P < P I , the average deformability Si is
simply expressed by
Note that Si characterizes an initial deformability only when PI is very small, and
that 6i gives dzflerent averages for increasing values of P I . With Eq. (la), the latter
integral equals Jt' dA = A , , and the expression for Si integrates to
Si = AI/P1. (1b)
On the other hand, with Eq. (la) the deformability 6,for the further extension of
the plastic zone at PI is
6, PI, A , ) (dP/dA)ibp,,A=Al,
= (lc)
where the considerations discussed above suggest 6, 2 Si.
~ = testing load, a = 1.8544 for the
In the hardness definition H = a P / ( 2 ~ 2 )( P
Vickers pyramid), 2a can be replaced by the size of the plastically deformed zone
2Rp1as far as the ratio n = 2Rp1/(2a)is approximately constant. With the character-
istic area A = (2Rp1)2,the hardness becomes
H = o'P/A, Pa)
where the parameter a' = n 2 0 depends on the value of n = RpI/a. In the associated
differential equation
dH -dP dA
H P A
dA/A can be substituted by means of Eq. (lb) by an expression that depends on the
load P only:
dA--
-
A
-dA
A1
- Se(P1)dP
A1 = (;)p=p,(E)
=
(Se/Si> d P
P (2c)
+
For finite load ranges P I -+ PI AP, and Eq. (2b) can be integrated if a slowly
varying ratio Se/Si is regarded as a small perturbation within this load interval.
The result is
H(P)= (g) H [(6c/6i)-ll

,
(3)
with H , = H(P,) and a scaling parameter P, that generally can be chosen arbi-
trarily. The choice of a large P,, however, offers the advantage that H , acquires
the character of an asymptotic (constant) level H , = Hasympt (P 4 00).
It is only here that the knowledge about the processes of microplastic deformation
becomes physically important for the model: if the deformability 6,for the further
extension of the plastic zone is larger than the average deformability Si for the
preceding penetration at any PI, the ratio Se/Si > 1 attributes a specific character
to the load influence in Eq. 3, and the hardness decreases with increasing load in
a way that agrees qualitatively with the measured data obtained for a Czochralski
grown sapphire crystal in the load range 3-100 N (Fig. 3 , interrupted line).
To obtain reliable data which are really characteristic for the single crystalline
material (independent of special features associated with one or another orientation),
I I I I I I I 1
1.o 2.0 4.0 7.0 10.0 20.0 40.0 70.0 100.0
indenter load (N)
Figure 3. Conventionally me2sured Vickers hardness on sapphire ((1510) surface, Vickers diagonals
parallel with the traces of (1012) and (1102) p 1 a n e s ) m T h e interrupted line is the fit accord-
ing to Eq. (3) with a constant (average) ratio (Se/Si) resulting in H , = 15.40GPa and
(Se/S,) = 1.119. An only slightly different fit results with the varying (S,/Si) of Eq. (3a)
( H , = 16.20GPa, ( & / c ? ~=) ~1.115).
6.4 Influences of the Gruin Size and the State of the Surface 191
investigations were performed on (1210) sapphire surfaces which offer the

opportunity of comparing hardness values associated with most difficult and most
easy indentation fracture. One series of measurements was performed with the
diagonals of the Vickers pyramid parallel with the traces of crystallographic (lOT0)
and (0001) planes, and another one with the diagonals nearly parallel with two
(1072) planes (the basis (0001) shows the highest resistance to cracking, whereas
{ 1072) planes exhibit a minimum fracture toughness [14,15]). When the diagonals
are aligned with (1070) and (OOOl), the cracks only start with this orientation but
change to the (1072) system a few micrometers beyond the boundaries of the
indent [13]. No significant differences in the hardness were observed for the two
orientations.
With a constant ratio Se/Si > 1, Eq. (3) gives a hardness that decreases infinitely
for very large testing loads (interrupted line in Fig. 3). To understand the asymptotic
character of the function H ( P ) . the (slow) variation of (Se/Si) with P needs to be
taken into consideration: for very large values of the load, the deformability 6,
which characterizes the further extension of the plastic zone approaches the average
deformability Si for the preceding penetration (due to A l >> AA at P I + 00). Then,
the exponent in the denominator approaches zero, the denominator becomes unity,
and H ( P ) approaches a constant value H,. It has been shown [16] that the exponent
of the denominator in Eq. (3) may be described, for example, by
with = (he/Si)p+o. Since (S,/Si) as given by Eq. (3a) is a very slowly varying
function, fits according to (3a) or using an average (constant) value show only small
differences in practically interesting load ranges (Fig. 3), and it is obvious that use
and the form of Eq. (3a) has little practical meaning for the derived load effect
over the experimentally investigated range.
The expressions (3) and (3a) give a load influence that compares well with
measurements in single crystalline A120, (Fig. 3). However, it should be emphasized
again that with the phenomenological character of the model this agreement is not
an evidence for some specific micromechanism, it just means that there is no general
discrepancy between the dislocation idea and the observed indentation size effect.
6.4 Influences of the Grain Size and the State of the

Surface
6.4.1 The Grain Size Influence on the Load Effect of the Hardness:
Modeling Experimental Results
Previous conventional and depth sensing Vickers hardness measurements performed
on sapphire and with sintered traditional A1203 ceramics (grain sizes 2-10 pm,
30 Sintered Al 203
with different grain sizes
Sapphire
10 : I I I I I
1.o 3.0 6.0 10.0 30.0 60.0 100.0
indenter load (N)
Figure 4. Load dependence of the Vickers hardness of sapphire (same data as in Fig. 3) and of fine-
>
grained sintered alumina ceramics (relative density 99%, purity > 99.7% A120,) [16].
99.8-99.9% A1203)[ 131 were supplemented by measurements with advanced more

fine grained (0.4-1.5 pm) and highly pure (99.99%) microstructures [17]. Figure 4
gives the results of conventional load-dependent measurements of the Vickers hard-
ness in different fine grained alumina ceramics. Generally, the hardness of the poly-
crystals is (of course) higher than the hardness of sapphire, but it is also obvious that
the increasing hardness of more fine-grained microstructures is significant at larger
indentation sizes only, associated with a decreasing load (size) influence.
With Eq. (3/3a), this decreasing influence of the load in more fine-grained micro-
structures is expressed by smaller values of the ratio (Se/Si)o. Table 2 displays fitting
parameters for the experimental data of Fig. 4: for smaller grain sizes, increasing
asymptotic hardness values H , are associated with decreasing parameters
( C ~ ~ / SIn~ submicrometer
)~. alumina microstructures, the increase of the microplastic
deformability (the decrease of the hardness) becomes smaller and smaller and
approaches zero already at small indent sizes of 10-20pm (see Fig. 4), and 6,
which characterizes the extension of the plastic zone approaches the ‘initial’ deform-
ability Si at rather small loads of about 1 N.
Table 2. Parameters of a least squares fit (Eqs (3) and (3a)) for the load effect in the hardness of
alumina ceramics with different grain sizes (Fig. 4).
Average grain size (pm) (bel4)O H , = VH (P= 98.1 N) (GPa)
0.2 1.04 20.5

0.5 1.04 19.7
3 1.09 18.0
Single crystal (sapphire) 1.115 16.2
6.4 Influences of the Grain Size and the State of the Surface 193
The measurements in Fig. 4 demonstrate that any polycrystal exhibits a hardness

like the single crystal if the load is small enough (if only a couple of grains are
affected).
A difference appears only when, on increasing the load, the indentation size
becomes much larger than the grains: whereas in sapphire the deformability
increases with the growing plastic zone, in the polycrystals this size effect is partly
offset by the hindrance of dislocation activity due to the close spacing of the grain
boundaries.
This deviation from the single crystalline behavior starts earlier, the smaller the
grain size is (limiting at higher testing loads the increasing deformability, i.e. the
decrease of the hardness).
6.4.2 The Effect of the Grain Size and the Surface State in
Ceramics when Recorded by Different Measuring Approaches
Figure 4 clearly demonstrates the reason of the declining grain size effect at decreas-
ing testing loads (smaller indentation size). A more conventional presentation is
given in Fig. 5. A similar grain size influence is recorded by Vickers indents at
10 kg (HVlO) and 1 kg (HV1; HV1 > HVlO due to the ISE), and close results are
obtained by HV1 and by a Knoop approach at 100 g (HKO. 1) because the longer
of the two Knoop diagonals exhibits a similar length at a load of 100 g to the Vickers
diagonal size at HVl. With further decreasing sizes of the plastic zones in the
0 0
22 -
-
0-
a
m s9!
g20- 4
m
m
al
C
T?
2 18-
-
2 HKO.l
.
t - - - ~
al
0
Y *--- HV1
5 HVlO
polished surfaces
high-purity Al (no doping additives)
14
0.1 0.2 0.4 0.6 0.8 1.0 2.0 4.0
D (w)
Figure 5. Investigations of the grain size effect in alumina ceramics by using different approaches [6].
Each data point indicates the average value of 6-15 indents.
24-1 ground
22 -
polished
i'
420-
0
2
$18-
:
C
16-
14 I I I I I 1 1 1 1 I I )
0.1 0.2 0.4 0.6 0.8 1.0 2.0 4.0

D (PC.Lm)
Figure 6. Influence of the surface preparation on the hardness of sintered alumina ceramics with
different grain sizes. In the submicrometer range, grinding results in a higher hardness than
measured on polished surfaces.
Vickers microhardness (400 g, HV0.4), however, the grain size effect becomes
smaller, and a similar hardness is derived for the single crystal (sapphire) as for
an even extremely fine-grained 0.2 pm polycrystal.
Figures 6 and 7 show that the hardness of ground surfaces may be higher than
measured on polished surfaces (unfortunately, no detailed investigation of this
effect is known in ceramics). The reason for this seems to be similar to the grain
size influence. Whereas no significant difference exists in alumina ceramics with
grain sizes >1 pm, an increasing advantage of the ground state is observed for
fine-grained materials with HV10 > 20 GPa (submicrometer A1203, A1203-
Ti(C, 0,N) composites). TEM has revealed extended dislocation activity in both
ground and in polished alumina surfaces [18], but the depth of the affected surface
layer was very different: the maximum depth of damage was about 20pm in
ground (40pm diamond) but less than 2pm in polished (0.25pm diamond)
surfaces.
Therefore, the most probable reason for the different hardness in ground and in
polished ceramics is that a high density of grinding-induced dislocation and twin
networks in the machined surfaces reduces the microplastic deformability (increase
the hardness) at an indentation site, in a similar way to small grain sizes.
Within the framework of the understanding outlined above it is clear that after
grinding, a subsequent further microplastic deformation (on indentation) must be
the more difficult the less mobile the grinding-induced dislocations are, that is the
closer the spacing of grain boundaries, the smaller the grain size is. As a
6.5 Comparing the Grain S i x E j e c t and the Indentation Size Effect 1%
T - raw mixture synthesized via Al-titanate

14 I ' I ' l l 1 I I I
consequence, the influence of surface preparation on the hardness is stronger in

hard, very fine-grained microstructures (Figs 6 and 7).
6.5 Comparing the Grain Size Effect and the

Indentation Size Effect: The Role of Grain
Boundaries at Room Temperature
Figure 8 compares the Vickers hardness HVlO (P= 98.1 N) of alumina ceramics
with the experimentally observed size effect in sapphire described by Eq. (3)
(including the load-dependent exponent given by Eq. (3a)). The curves for n = 1
and n = 4 represent extrapolations of the experimentally measured size effect on
the hardness of single crystalline alumina.
As additional information, the curve of the indentation size effect of sapphire in
Fig. 8 was complemented by data measured and analyzed by Oliver and Pharr [I91
on a (0001) plane. It is very remarkable that neither the different crystals and
orientations, nor the different (depth sensing) measurements (Vickers-Berkovich)
and the very different data analyses affect the character of the size effect observed
in sapphire: it can be described by one curve according to Eq. (3)/(3a) for available
data in an extremely wide range of testing loads between 1 mN and 100N.
With the two parameters H , and (6e/Si)o = (6e/6i)p+o (Table 2), it is possible
to analyse the experimentally recorded dependence of the load effect in sintered
For indentation size effect of sapphire: 2a (pm) for n=l ,2R,, (pm) for n=4
0.1 1.o 10.0 100.0 1000.0
I I t I I I l l I I I l l I I I l l I I I l l L
I
4 n=l n=4 (curves derived from measured load effect in sapphire)
HVlO - grain size effect measured in AI,O,

--- - - = - - - - - +-
10 ' I I I I I I I 1 I I I l l I I I l l I I I I [ I
0.1 1.o 10.0 100.0 1000.0

For HV10: Grain size D (prn)
Figure 8. Grain size effects in the hardness of alumina ceramics (HV-10, triangles), and indentation
size effect (load influence) in the hardness of sapphire (squares). For HVlO values, the lower x-axis
gives the average grain size (data for sapphire indicated at 1OOOpm). For the curves describing the
size effect in sapphire, the upper x-axis represents the length of the Vickers diagonal (n = 1 ) or the
plastic zone size (n = 4) [6,16]. See Fig. 2 (p. 166) for theoretical background.
alumina ceramics on the grain size. Note that the deformabilities 6, and Si are
phenomenological parameters that describe the microscopic deformability without
assuming specific mechanisms. For the use of these parameters within the frame-
work of the model introduced in Section 6.3.2 it is, therefore, unimportant whether
inelastic microdeformations are generated by lattice processes only (dislocations,
twins), or if additional grain-boundary deformation takes place.
If grain-boundary processes are active, they will increase the deformation and
reduce the hardness. Therefore, it is not a surprise to find in Fig. 8 a quite different
grain size dependence for polycrystalline alumina ceramics (HV 10) compared
with the (also experimentally measured!) size effect in the single crystal (sapphire,
represented by the curve derived under the assumption of a plastic zone size
2R,1 = 4 x 2a). The degree of this difference, however, is much larger than might
be expected: even microstructures with coarse grain sizes of the order of 10ym
exhibit a much higher microplastic deformation (lower hardness) than isolated
sapphire subregions of the same size.
It should be emphasized that this conclusion is independent of theoretical
speculations: all data points in Fig. 8 are measured values. Only the extrapolation
6.5 Comparing the Grain Size Effectand the Indentation Size Effect 191
of the indentation size effect in sapphire towards smaller sizes depends on the model,
but this extrapolation was performed in a way that agrees well with published
nanoindentation results.
Comparing in Fig. 8 microstructures with average grain sizes of about 10 and
0.2 pm, respectively, the dzfeerence between the hardness of these plycrystalline
microstructures on the one hand, and of plastically deformed single crystalline
volumes that exhibit the same size as the grains in the polycrystals on the other
hand, is almost doubled when fine-grained materials are compared with coarser
sizes, and it will further increase for nanoceramics. Therefore, with the data in
Fig. 8 it seems rather improbable to find a sharp increase of the hardness in
future nanoscale corundum microstructures with grain sizes of 50nm or less.
Additionally to efforts intended to reduce the grain size, grain-boundary engineering
is, therefore, a second approach to the development of hard polycrystals, which
becomes the more important the more fine-grained the material is.
For example, the ratio between the contributions of lattice processes and of grain
boundaries to the inelastic deformation at an indentation site may be quite different
if different types of bonding (ionic, covalent) are considered. In fact, first
investigations indicate an improved resistance against inelastic grain boundary
microdeformation in submicrometer microstructures with a covalent bonding
character if these materials are produced without ionic or metallic binder phases.
Figure 9 compares highly-pure sintered alumina (>99.9% A1203) and binder-
free (!) ceramics based on WC and (Ti, W)C [20]. Whereas all these materials exhibit
a similar hardness when microstructures with average grain sizes 1 4 p m are
30
\v
WC\\
25 -
\? (Ti,W)C
\ *
\ *'
m
a \T\ ***
s?
0 pure AI,O,
0
\ *.
-
I 20-
>
(I)
acl
(I)
\ . - m
0 *=
\
E
c
\
\ *
+\
\
\
\
T \
15- \
blished surfaces] \
I I I l ~ l ~ l l I 1 1 1
0.1 0.2 0.4 0.6 0.8 1.0 2.0 4.0

considered, a difference appears in the submicrometer range. With grain sizes

<1 pm, the slope of the curve decreases for alumina, indicating an increasing
contribution of grain-boundary deformation. The carbides, however, show a
steady increase of the hardness towards smaller grain sizes.
6.6 The Effects of Temperature on the Hardness of

Ceramics
This section considers the direct influences of the temperature on the hardness,
excluding secondary consequences when phases become thermodynamically unstable.
Figure 10 compares the temperature effect in the hardness of two single crystalline
materials with a different bonding character (A1203,Sic) with the behavior of poly-
crystalline alumina ceramics. The authors did not discuss the unusual result of a con-
stant hardness of Sic between room temperature and about 400°C. It is, however, a
general observation that in single crystals the influence of the temperature on the
hardness decreases at high temperatures >lOOO°C, whereas a much larger effect is
observed at room temperature and associated with a further increase of the hardness
below 0°C due to the thermally activated character of dislocation movement (contrary
to the strength, which is almost constant up to about 600-1000°C in most ceramics).
Below 1000°C, Fig. 10 displays a similar temperature effect on the hardness of
single and polycrystalline A1203. The one difference is that at room temperature
I I I I I I I I
0 200 400 600 800 1000 1200 1400 1600
Temperature (OC)
Figure 10. Temperature dependence of the hardness in highly ionic (A1203)and highly covalent
(Sic) single crystals compared with polycrystalline (sintered) ceramics [21].
6.7 Sunitnary 199
Figure 11. Intergranular damage at the edge of a 4 kg indent in a 2.5 pm alumina ceramic at 1600 C
P11.
the 9 pm material is less hard than the fine-grained ceramic due to the usual grain
size effect and less hard than the sapphire crystal (probably associated with some
defects). In the range 500-1 000°C the disadvantage of the coarser microstructure
compared with the 2.5 pm ceramic is offset because it is less susceptible to inelastic
grain boundary deformation due to its lower frequency of grain boundaries per
unit volume. Altogether, however, the influence of the microstructure on the
effect of the temperature is small at temperatures < 1000°C.
The situation changes when at some critical temperature > 1000°C weakening
grain boundaries cause a global instability of the microstructure at the indentation
site (Fig. 11). This temperature range is of special interest for ceramic tools, because
both cutting ceramics and grinding materials have to resist temperatures of more
than 1000°C at processing zones (cutting edges and tips). The consequence of the
general instability of the grain boundaries at high temperatures is the decrease of
the hardness to values which are much smaller than observed for the single crystal.
It is obvious that the critical temperature for this drop of the hardness depends
strongly on the grain boundary structure (amorphous interfaces, segregating
species) and on the grain size.
6.7 Summary
The measured size effect on the hardness of single and polycrystalline alumina can
be fitted assuming an increasing deformability, S,, when the extension of the plastic
zone proceeds by growth and multiplication of pre-existing elements of plasticity
instead of the more difficult generation of new dislocations, dislocation loops, or
twins. With increasing testing load (size of the indent), there is a decreasing hardness
due to the involvement an increasing number of pre-existing dislocations but, on the
other hand, with increasing size of the plastic zone the value of 6, approaches the
average deformability Si of the preceding stages of the penetration process, and
the degree of the load effect decreases. As a consequence, the hardness approaches
an asymptotic value H , at larger testing loads.
Smaller grain sizes restrict the microplastic deformation to smaller volumes, and
the process approaches H , at a smaller size of the indent. Therefore, the influence of
the load (size) on the hardness decreases in more fine-grained microstructures. The
grain size effect is larger when larger testing loads are applied, it disappears at small
loads. Hence, microhardness data should not be used to study grain size effects in the
hardness.
Comparing the hardness of polished and of ground surfaces, there is a tendency
for higher values in the latter. Whereas no significant difference exists in alumina
ceramics with grain sizes >1 pm, an increasing advantage of the ground state is
observed for fine-grained materials with HVlO > 20 GPa (submicrometer A1203,
A1203-Ti(C,0,N) composites). The most probable reason is that grinding-induced
dislocation and twin networks in the machined surfaces reduce the microplastic
deformability (increase the hardness) at an indentation site in a similar way to the
effect of small grain sizes.
A remarkably large discrepancy between the hardness of plastically deformed
single crystalline volumes and of polycrystalline microstructures with grains of
the same size, indicates a significant contribution of grain-boundary deformation
at the indentation site in polycrystalline alumina, even at room temperature and
even for coarse microstructures.
The increase of grain-boundary contributions to the deformation at small grain
sizes makes a sharp increase of the hardness in nanoceramics like alumina highly
improbable. Additionally to efforts that reduce the grain size, grain boundary
engineering is, therefore, a second approach to the development of hard poly-
crystals, which becomes more important the more fine-grained the material is.
First investigations indicate an improved resistance to inelastic grain-boundary
microdeformation in submicrometer microstructures with a covalent bonding
character if these materials are produced without ionic or metallic binder phases.
Under this condition, an increase of the hardness with decreasing grain sizes may
still exist at grain sizes of lOOnm or less.
Different mechanisms control the temperature effect in the hardness below
1000°C and at higher temperatures when weakening grain boundaries induce a
loss of the global stability of the microstructure. At lower temperature, thermally
activated processes result in a similar influence of the temperature in polycrystalline
ceramics to that in single crystals, and the influence of the microstructure on the
temperature effect is small. A steady decrease of the hardness with increasing
temperature is observed between room temperature and about 1000°C, with a
decreasing slope the higher the temperature is. The influence of the temperature
again increases when, beyond some critical temperature, weakening grain boun-
daries induce a drop of the hardness to much lower values than observed for the
single crystal at same testing load and temperature. The influence of the micro-
structure (e.g. of amorphous grain-boundary phases and of the grain size) on this
critical temperature is technically important because many ceramic tool materials
Rejerences 20 1
run under conditions where the local temperature at the processing edge or tip
exceeds 1000°C.
References
1. A. Krell and P. Blank, A1203-Sinteringmaterial, process for the production of said material and
use thereof, European Patent EP-756 586 BI, Int. CL6 C04B 35/11 I , published 29 October 1997.
2. B. J. Hockey, Plastic deformation of aluminum oxide by indentation and abrasion, J . Am.
Ceram. SOC.1971,54, 223-231.
3. D. E. Munson and R. J. Lawrence, Dynamic deformation of polycrystalline alumina, J . Appl.
Phys. 1979,50,6272-6282.
4. R. A. Graham and W. P. Brooks, Shock-wave compression of sapphire from 15 to 420kbar,
J. Phys. Chem. Sol. 1971,32,2311-2330.
5. A. Krell, Gezielte Mikrostrukturentwicklung in Sinterkorund, PhD Thesis, Academy of
Sciences/Central Institute for Solid State Physics and Materials Science, Dresden, 1982.
6. A. Krell, Effects of load, grain size, and grain boundaries on the hardness of alumina, Proc.
22nd Cocoa Beach Conference (2&24 January 1998), American Ceramic Society, Westerville,
OH, 1998, pp. 159-168.
7. A. Krell and P. Blank, Inherent reinforcement of ceramic microstructures by grain boundary
engineering, J. Europ. Ceram. Soc. 1992, 9, 309-322.
8. J. Lankford and D. L. Davidson, The crack-initiation threshold in ceramic materials subjected
to elastic/plastic indentation, J . Muter. Sci. 1979, 14, 1662-1668.
9. G. P. Upit and S. A. Varchenya, The size effect in the hardness of single crystals, in The Science
of Hardness Testing and its Research Applications, J. H. Westbrook and H. Conrad (Eds), The
American Society for Metals, Metals Park, OH, 1973, pp. 135-146.
10. C. M. Perrott, Elastic-plastic indentation: Hardness and fracture, Wear, 1977,45,293-309.
I I . A. Krell, J. Woltersdorf, E. Pippel, and D. Schulze, On grain boundary strength in sintered
A120,, Philos. Mag. A , 1985, 51, 765-776.
12. A. Krell, Improved hardness and hierarchic influences on wear in submicron sintered alumina,
Muter. Sci.Eng. A , 1996, 209, 156163.
13. A. Krell, Vickers hardness and microfracture of single and polycrystalline AI2O3,Cryst. Sci.
Technol. 1980, 15, 16667-16664.
14. S . M. Wiederhorn, Fracture of sapphire, J . Am. Ceram. SOC.1969, 52, 485-491.
15. P. F. Becher, Fracture-strength anisotropy of sapphire, J . Am. Ceram. SOC. 1976, 59, 59-65.
16. A. Krell, A new look at grain size and load effects in the hardness of ceramics, Muter. Sci. Eng.
A245, 1998, 277-284.
17. A. Krell and P. Blank, Grain size dependence of hardness in dense submicrometer alumina,
J. Am. Ceram. SOC.1995,78, 1118-1120.
18. B. J. Hockey, Observations by transmission electron microscopy on the subsurface damage
produced in aluminum oxide by mechanical polishing and grinding, Proc. Br. Ceram. SOC.
1972,20,95-115.
19. W. C. Oliver and G. M. Pharr, An improved technique for determining hardness and elastic
modulus using load and displacement sensing indentation experiments, J . Muter. Res. 1992,
7, 15641583.
20. V. Richter, M. v. Ruthendorf, and J. Drobniewski, Perspektiven feinkorniger Hartmetalle, in
Pulvermetallurgie in Wissenschaft und Praxis, Vol. 13, Proc. Hagen Symposium Powder-
metallurgy, 13-14 November 1997, R. Ruthardt (Ed.), Werkstoff-Informationsgesellschaft,
Frankfurt, 1997, pp. 29-45.
21. A. Krell and 0. V. Bakun, High-temperature hardness of A1203-base ceramics, Acta Metall.
1986,34, 1315-1319.
7 Transition Metal Carbides, Nitrides, and

Carbonitrides
W. Lengauer
7.1 Introduction
Transition metal carbides, nitrides, and carbonitrides belong to a class of materials
that show an unusual combination of outstanding properties, amongst which are
exceptional hardness, high melting point (e.g. TaC and HfC have among the highest
known melting points of all materials), metallic lustre combined with vivid colors, and
simple metallic structures. In addition, they show high electrical and thermal conduc-
tivities. This combination of properties has attracted considerable attention and has
resulted in numerous technical applications [ 11 and theoretical investigations. Many
binary transition metal carbides and nitrides especially the face centered cubic 6
phases of the IVB group, exist over broad ranges of composition with an upper
limit of the nonmetal/metal ratio near 1. Practically all solid state properties show
a gradual change with this ratio. Other carbides with a different structure such as
tungsten carbide and chromium carbides have a very narrow homogeneity region.
Transition metal carbides and nitrides often form solid solutions resulting in
carbonitrides in which the solid-state properties also change with the [C]/([C] +
[Nl) ratio. This behavior offers a large field for tailoring carbonitrides.
The carbides, and to a minor extent also nitrides and carbonitrides, are easily
wetted by liquid metals, which has found application in hardmetals and cermets.
Since the basic invention of preparing composite materials by liquid phase sintering
of carbides with iron group metals, where a very successful combination of hardness
with toughness was achieved, these materials are applied as components in hard-
metals for a variety of cutting and drilling operations. In steels, carbide and carbo-
nitride precipitates cause hardening known as dispersion hardening.
The chemical stability of transition metal carbides, nitrides, and carbonitrides is
high, they are not readily attacked by dilute acids (except oxidizing acids and
hydrofluoric acid) or alkaline solutions. The thermal stability is connected with
their free energy of formation, the stability decreasing with increasing group
number. While the group IV transition metal nitrides TIN, ZrN, and HfN and
carbonitrides Ti(C, N), Zr(C, N), and Hf(C, N) can be melted without decomposi-
tion, the nitrides and carbonitrides of the other groups decompose before melting
points are reached. As a result nitrogen is released.
Tables 1 and 2 give a short overview of the properties important for technical
applications of the most well characterized and important binary transition metal
carbides and nitrides. Some specific properties which cannot easily be listed in a
table, as well as the properties of transition metals carbonitrides) especially those
of the most commonly used group IVB carbonitrides) are described in detail later.
Table 1. Bulk properties at room temperature of some important transition metal carbides (disordered state). For f.c.c. carbides (Pearson symbol
cF8) the composition is near 50 at-% C, except for VC0.88.The melting points do not apply to this composition.
Compound Pearson Lattice Density Micro- Melting Young's Heat Linear thermal Electrical
symbol parameter(s) hardness point modulus conductivity expansion coefficient resistivity
(nm) (gcm-') (GPa) ("C) (GPa) (Wm-' K- (IO-~)K-' (FQ cm)
TiC cF8 0.4328 4.93 28 3067 450 28.9 8.5 100

ZrC cF8 0.4700 6.46 25 3420 350 24.6 7.5 75
HfC cF8 0.4638 12.3 20 3930 420 25.1 6.1 67
VCO.88 cF8 0.4164 5.36 26 2650 430 26.8 7.2 69
NbC cF8 0.4470 7.78 18 3610 340 27.0 6.6 20
TaC cF8 0.4455 14.48 16 3985 290 22.1 6.3 15
Cr3C2 0P20 a=0.28290 6.68 27 1810 380 14 10.3 75
h = 0.55329
c = 1.14719
Mo~C hP3 u = 0.30233 9.18 17 2520 530 15 7.8 51
c = 0.47344
WC hP2 a = 0.29058 15.72 23 2776 707 19 3.9 17
c = 0.28365 u
k
6'
h,
0
P
u
3
a
z.g.
3
Table 2. Bulk properties of important transition metal nitrides. Values for room temperature and for stoichiometric composition.
5k
?I
Phase Pearson Lattice Color Density Micro- Melting Young’s Heat Linear thermal Electrical T, $.
symbol parameter( s) hardness point* modulus conductivity expansion coefficient resistivity 5
2
(nm) ( g ~ m - ~ (GPa)
) (“C) (GPa) (Wm-’ K- ) (10-6K-’) (PQcm) (K) .
3
?
TiN cF8 0.4242 golden yellow 5.39 17 3050 420 29 9.9 27 5.8
ZrN cF8 0.4570 green yellow 7.32 15 3000 460 11 7.8 24 10.47 $
HfN cF8 0.4516 dark yellow 13.83 18 3330 380 11 8.5 27 6.92 3
VN cF8 0.4138 brown 6.04 5.7 2350 380 11 10.8 65 8.9
NbN cF8 0.4392 light yellow 8.16 11 d 360 3.8 10.2 60 17.2 ?I
~ ~
TaN cF8 0.4238 gray yellow 15.9 32 d ~ 8.0 - 8.9
~ ~ -
TaN hP3 a = 0.2936 dark gray 14.3 11 d 9.5 128 2.
-
c = 0.2885 2.
CrN cF8 0.4148 gray brown 6.14 13 d 450 11.7 - 640 -
* d = decomposes
7.2 General Features of Structure and Bonding 205
7.2 General Features of Structure and Bonding

7.2.1 General Structural Features
Transition metal carbides, nitrides, and carbonitrides belong to the family of
interstitial alloys or compounds. The common features of this class of materials
are the very simple metallic structures with small carbon and/or nitrogen atoms
in the interstitial voids of the often densely-packed host lattice. They generally
have face-centered cubic (f.c.c.) or hexagonal close packed (h.c.p.) metal lattices
with nonmetal atoms more-or-less randomly distributed on the interstitial sites.
The metal atoms are arranged in such a way that they form close packed arrange-
ments of metal layers with ABAB.. . (hexagonal, h) or ABCABC.. . (cubic, c)
stacking sequence (Fig. 1) or with a mixture of these. The nonmetal atoms in
these phases occupy the octahedral interstitial sites. According to a crystallo-
chemical rule [2,3] the phases of pure h type can have a maximum nonmetal/
metal ratio of 1/2 and the c type phases have a maximum carbon content of 1.
The phenomenon of vacancy ordering is often observed especially in the transition
metal carbides [4,5]. In the carbides of the group IVB and VB transition metals the
common structural element is the T6X octahedron. With increasing radius ratio
r\./rT the trigonal prismatic group T6X is favored over the octahedral group. There-
fore the transition metal carbides of groups VIB-VIIIB feature trigonal prismatic T6X
elements in different crystallographic arrangements.
Deviations from the simple close packed structure are more frequently encoun-
tered with nitrides where T5X quadratic pyramids are observed as another structural
element. This is the case particularly in group VB compounds. These prevail even in
the complex nitrides, such as the Z-phases NbCrN, TaCrN, NbMoN, and TaMoN,
with the filled-up BCuTi-type structure. With this quadratic pyramid, which has
lower symmetry than an octahedron, such nitrides are more complex. Because of
the introduction of structural elements other than T6N octrahedral, the f.c.c.
structure of 6 nitrides is destabilized increasingly with increasing group number,
and (within a group) with increasing periodic number. The nitrides of group IVB,
E h
c
Figure 1. Hexagonal (h) and cubic (c) stacking sequence of metal atoms of close-packed h.c.p. and
f.c.c. transition metal carbides and nitrides. Open large circles designate metal atoms, filled small
circles designate interstitial nonmetal. In h sequences only 50% of the interstices can be occupied.
Nitrogen and carbon atoms are randomly distributed among the nonmetal sublattice.
206 7 Transition Metal Curbides, Nitrides, and Carbonitrides
FTiNl --x, GZrN1--x, and 6HfN1px, crystallize in the sodium chloride structure and
no low-temperature modifications of these stoichiometric nitrides are known. In
the VB group, also all three metals form stoichiometric f.c.c. nitrides, but GVNl.oo
transforms at 205K into a tetragonal modification [6], FNbNl,ooat around 1310-
1320°C into hexagonal qNbN [7] and GTaNI.ooat around 1800°C into hexagonal
ETaN [8], the transformation temperatures depending on the nitrogen equilibrium
pressure.
In the VIB group these f.c.c. nitrides are increasingly less stable. GCrN1--x requires
a relatively high nitrogen pressure of preparation and transforms at below ca. 280 K
into a tetragonal antiferromagnetic compound [9] and yMoNl --x occurs only at up
to 38 at-% N using flowing ammonia, a value that could be increased if the nitriding
potential is additionally increased. This stability feature continues further to the
Mn-N system, where the f.c.c. phase has the stoichiometry T4N and to the Fe-N
system. In the latter only the f.c.c. solid solution of nitrogen in yFe is observed
[lo] (the Fe4N phase has f.c.c. structure with respect to the metal positions only).
7.2.2 General Features of Bonding

A mixture of metallic, covalent and ionic components prevails in the bonding of
transition metal carbides, nitrides, and carbonitrides. The metallic character is
shown by the high electrical conductivities of these compounds. The bonding
mechanism has been described extensively by a variety of approaches for calculating
the density of states (DOS) and hence the electron density in f.c.c. transition metal
carbides, nitrides, and oxides [l 11. In the DOS of these compounds there is a
minimum at a valence electron concentration (VEC) of 8, which corresponds to
the stoichiometric composition of the group IVB carbides TIC, ZrC, and HfC.
Transition metal carbides have a lower DOS at the Fermi level than the correspond-
ing transition metal nitrides, hence the electrical properties such as electrical and
thermal conductivity and the superconducting transition temperature, T,, are
lower than those of the nitrides.
Although transition metal nitrides and carbides show sometimes complete mutual
solubility, differences due to the different bonding character of nitrogen as com-
pared to that of carbon can be encountered. Figure 2 compares the decomposed
and the total DOS of TiN with T i c [12]. At low energies the s band is characterized
by a large contribution of nonmetal 2s state. At intermediate energies the p band
follows and there is a significant contribution of d states to this band. At higher ener-
gies the d band composed of d states follows with the Fermi energy, EF, located well
within this band. The DOS at EF is higher for TiN than for Tic, corresponding well
with the observed superconducting properties (TIN: T, = 5.4 K, TIC: T, < 1.2 K)
and room-temperature electrical conductivities. The covalent contribution to bond-
ing in these compounds can be drawn in the molecule orbital (MO) scheme [13],
where atomic orbitals of atoms are combined to form bonding and antibonding
states. It could be shown that eg bonding symmetry in T i c changes to t bonding
26
symmetry in TIN. Thus the change in bonding strength from a strong Ti-C inter-
action in TIC towards a more pronounced Ti-Ti interaction in TiN is to be expected.
7.3 Preparation 207
-0.6 -0.4 u 0.4 I ?

E (Ryd)
Figure 2. Density of states (decomposed into the different contributions and total) of Tic and TIN [12].
The ionic contribution to the binding mechanism in these compounds can be esti-
mated from the charge transfer which occurs from the metal atom to the nonmetal
atom. It is about half an electron, which contributes to the electrostatic interaction
of metal and nonmetal.
7.3 Preparation
Transition metal carbides can be prepared by direct combination of the elements,
preferably as powder mixtures, at elevated temperatures:
Ti + C --+ GTiC.
If heat is generated by the reaction such as with Ti and C, the reaction is exothermic
enough to keep up the reaction temperature. The process can be carried out in form
of a self-sustaining high-temperature synthesis (SHS). The reaction front, once
ignited, travels through the powder bed producing the nitride and dissipating the
heat over a certain period of time. Also nitrides can be produced by SHS techniques.
208 7 Transition Metal Carbides, Nitrides, and Carbonitrides
Likewise carbon containing gases such as methane or benzene can be utilized:

Ti + CH4 -+ TIC + 2H2.
The preparation of TIC, VC, NbC, TaC, Mo2C, WC, and chromium carbides are
important technical processes for the production of carbide powders for hardmetals.
Generally, the carbides are prepared by the reduction of oxides with carbon, while
Mo2C and WC are made by reacting the metal with carbon such as graphite or
carbon black.
Solid solution of transition metal carbides and carbonitrides, which also play an
important role in the hardmetal industry, can be advantageously manufactured by
the simultaneous reduction of oxide or oxide-carbide mixtures in the presence of
carbon
Ti02 + WC + 2 C --f +
(Ti,W)C 2CO.
The generally volatile chlorides of the transition metals can be employed to obtain
transition metal carbides via the reaction scheme
+
TiC14 CH4 -+ T i c + 4 HC1.
This reaction proceeds in the gas phase at temperatures above about 600°C and is
usually applied either in chemical vapor deposition of carbide layers on solid sub-
strates or to produce carbide powders with very fine (submicron) grain sizes. In
the latter case the nucleation of the carbide must proceed in the homogeneous gas
phase.
The oxides of the transition metals can be reduced by carbon according to the
following overall reaction
+
w O 3 4 c -+ wc + 3 co.
Coarse chromium carbide powders can be produced by reducing Cr203 with
aluminum powder in the presence of carbon:
3 Cr203+ 6A1+ 4 C -+ 3 A1203+ 2 Cr3C2.
Bulk transition metal nitrides can be prepared by reacting the metal or the metal
hydride in powdered or compact form with molecular or atomic nitrogen or flowing
ammonia. This can be expressed by following examples:
Ti + i N 2 -+ TiN,
+
ZrH2 4N2 ZrN + H2,
-+
MO+ NH3 -+ MoN + +H2.

The pressure of molecular nitrogen at a given temperature that is necessary to obtain
a specific nitride phase is defined by the nitrogen potential (partial free energy of
nitrogen) of the nitride.
The formation of nitrides from oxides is performed by the aid of carbon as a
reducing agent, according to the overall reaction equation
Ti02 +2C +iN2 -+ TiN + 2CO.
7.3 Preparation 209
This reaction usually proceeds over several intermediate products and the resulting
nitride can contain oxygen and carbon.
For the preparation of large crystals (1 cm3) zone-annealing techniques were
developed [14],which are similar to those zone melting already known from material
purification processes. The starting materials are either solid-state sintered nitride
powders or metal powders pressed into rods. The applied temperatures must be
very high (around 3000°C) in order to overcome the low diffusivities of nitrogen
in nitrides, but they must not exceed the melting temperatures of the compounds
in order to avoid incongruent melting and decomposition.
Carbonitride powders for industrial applications are prepared by mixing carbide
and nitride powders intimately together and heating the mixtures to high temperatures,
TIC + TIN + 2 Ti(Co,5No.5),
whereas for the case of the Ti-C-N system any proportion leads to a single phase
product Ti(C,N1 -J provided that sufficient time or a sufficient high temperature
is provided for equilibration. For this reaction the nitrogen equilibrium pressure
is important either for preparation of Ti(C,N) or if this compound is heated at
high temperatures such as in sintering procedures. Figure 3 gives experimental as
1800 1400
-300 otm
-3 0 otm .x’ .
-I otm/
-5k - our thwmodvnomic estimation

our experimental determination
+ ZELIKMAN 8 GOROVITS
PORTNOI 8 LEVINSKII
-7 ---X KiEFFER 8 o l
0 0I 0 . 2 0 3 0 4 0.5 0.6 0.7 0.8 0.9
x
Figure 3. Nitrogen equilibrium pressure of Ti(C, N) as a function of composition from [16] with lit-
+
erature data. The composition is x in Tic, -,yN,, which is 1 - [C]/([C] [Nl).
well as calculated results [15, 161 for the nitrogen equilibrium pressure as a function
of the carbon/nitrogen ratio in Ti(C,N, -~r).
A great variety of methods exists for the preparation of nitride and carbonitride
layers. These can be divided into the groups CVD, PVD [17], and diffusion layer and
thermal surface modification techniques.
(i) Chemical vapor deposition (CVD): a volatile transition metal compound such
as a chloride is brought to reaction with ammonia, nitrogen/hydrogen, or
ammonia/hydrogen/nitrogen mixtures. For carbonitrides a carbon-containing
gas such as CH4 or other hydrocarbon is used [18]. Also single metal-organic
precursors can be used which make the construction of the furnaces more
simple. An example is Bis(benzene)vanadium, V(C6H6). which is solid at
room temperature, and can be used for deposition of V(C, N ) layers [ 191. The
nitride or carbonitride that forms as a reaction product in the gas phase
nucleates on solids, like hardmetals, and forms solid, strongly adhering
layers at temperatures around 1000°C. Thus hardened steels cannot be
coated unless a hardness decrease is accepted.
(ii) Physical vapor deposition (PVD): The transition metal is evaporated or
sputtered by glow discharge or arc heating, and reacted with low-pressure.
Numerous modifications of PVD methods have been developed [20-231. The
nitride layers that are deposited on the substrate surfaces often have high lattice
defect concentrations [24] and hence exhibit significant deviations from gener-
ally accepted bulk values (lattice parameter, hardness, composition) [25]. PVD
processes applied to coat hardmetals generally result in a smoother interface
than the CVD process because of the lower temperatures.
(iii) Diffusional techniques: Carbide, nitride, and carbonitride formation, which is
used for surface hardening, can be achieved by subjecting the material surface
to the action of various salt baths, ammonia, methane, molecular or atomic
nitrogen, or nitrogen ions [26]. Reactive plasma techniques and high-pressure
techniques were developed to keep the workpiece temperature low and to
avoid thermally induced changes in the workpiece. Laser irradiation together
with an N2 jet directed on the heated spot is employed for limited local heating
of the sample [27].
7.4 Characterization
7.4.1 Chemical Analysis
Carbon analysis of transition metal carbides is nearly exclusively performed by
combustion of the carbide in oxygen and measurement of the reaction products
C 0 2 or CO.
A common technique is vacuum or carrier-gas hot extraction where CO is
formed and measured with an IR spectrometer or with a thermal conductivity
7.4 Characterization 21 1
detector. The sample is heated to more than 2000°C in the presence of a bath
metal such as platinum in a stream of a carrier gas such as He. Due to the high
temperatures involved, the method is suitable for powder samples with larger
grain size or even for small solid pieces. The accuracy of the method is on the
order of 1-2 rel-% C.
In another technique measurement of the change in electrical conductivity of an
alkaline solution resulting from the C 0 2 absorption is performed. An oxidizing aid
such as V 2 0 5 or copper oxide is used to facilitate the oxidation.
Several procedures have been applied for the analysis of nitrogen in nitrides and
carbonitrides.
The Kjeldahl analysis involves the dissolution of nitrides in acids whereby the
nitrogen is converted into ammonium ions. The dissolution process is time con-
suming since some nitrides, such as GTiN,-,, are chemically very stable against
acids and must be dissolved under rigorous conditions by the action of hydrofluoric
acid in PTFE containers at 100-200°C.
Also vacuum or carrier-gas hot extraction such as for the carbides can be applied.
If carbon is present the CO is absorbed after IR measurement and the remaining
nitrogen is measured usually in a thermoconductivity cell. Temperature-controlled
furnaces can yield a fractional nitrogen analysis which resolves nitrogen differently
bonded or different nitride phases.
In another combustion analysis the sample is flash-heated at temperatures of up
to 1800°C.The evolved N2 is either measured volumetrically in an azotometer or via
a gas chromatograph [28,29] equipped with a thermoconductivity detector. N2,
COz, and H 2 0 can also be determined simultaneously and hence the C and the H
contents. The accuracy of the method is better than 1 rel-% N for finely powdered
samples.
7.4.2 Physical Microanalysis

Among the microphysical methods for determining nitrogen in nitrides are: elec-
tron probe microanalysis (EPMA), Auger electron spectroscopy (AES), ultravio-
let and X-Ray spectroscopy (UPS, XPS), secondary ion or neutron mass
spectrometry (SIMS, SNMS) and Rutherford backscattering (RBS). EPMA has
the highest reproducibility and accuracy, provided that the lateral resolution of
not less than a few microns is sufficient. The main problem in EPMA are the
rather low energies of the CK, and NK, radiations which requires windowless
or ultra-thin window detector systems and results in a low background-to-peak
ratio together with low net intensities [30,31]. High-accuracy EPMA requires
external or internal standardization [32] together with a wavelength-dispersive
spectrometer equipped with high-quality crystals. Recent development of
analyzing crystals have yielded multilayer crystals of the type WjSi, NijC and
Mo/B4C, which show excellent peak-to-background properties. The background
signal obtained with these crystal is more smooth, the sensitivity for high-order
intensities is low which facilitates background measurements and avoids line
coincidences.
A problem of EPMA of titanium nitrides and carbonitrides is the close overlap of

the Ti L1 line with the N K, line which cannot be separated by diffraction methods.
Therefore, one procedure to overcome this problem is the measurement of the Ti K,
+
line and to subtract the appropriate contribution from the Ti L1 N K, peak for
determination of the net N K, peak. If this method is made iteratively the accuracy
of the method is within 1 at-% N.
7.5 Thermodynamics
7.5.1 Stability of Carbides
The thermodynamic stability of the carbides is different between heavier and lighter
metals. Stoichiometric composition of the carbides can more easily be attained with
the heavier elements of the IVB and VB groups the than with the carbides of the
metals of the first long period. Vanadium carbide shows a maximum carbon content
of VCo.89.Also chromium does not form a carbide with 50 at-% C, instead a carbide
of stoichiometry Cr3Cz.While the monocarbides of the IVB and VB groups can be
melted congruently the carbides of the VIB group metals decompose peritectically
before their melting points are reached. The monocarbide MoC decomposes in a
peritectoid manner already at about 1230°C into Mo2C C. +
7.5.2 Nitrogen Partial Pressure of Nitrides

The nitrogen equilibrium pressure of the transition metal nitrides is a function of
temperature and increases substantially with increasing group number. This is
reflected in the Richardson diagram [l]. The lower (the more negative) the free
energy of formation, the lower the equilibrium nitrogen pressure of the nitride is.
Group IVB nitrides can easily be obtained by nitridation with molecular nitrogen
under low pressures even at high temperatures, whereas for group VB nitrides the
nitrogen pressure must exceed normal pressure at temperatures above 1400"C,
and even higher pressures or more energetic nitriding conditions must prevail for
the preparation of group VIB nitrides. The nitrogen pressure determines the
composition of the nitride if equilibrium is attained.
Since nitrogen pressure determines the phase stability, nitride phases with a non-
metal/metal ratio greater than 1 can be obtained if high-pressure nitrogen or high
nitrogen potentials, such as those prevailing in flowing ammonia or in atomic or
ionic nitrogen atmospheres, are applied. Either an extension of the mononitride
phase field to compositions [Nl/[Me]> 1 occur, such as in the N b N , Hf-N, and
Ta-N systems, and/or distinct nitride phases, such as in the Zr-N, Nb-N, and
Ta-N systems are obtained.
Ion implantation techniques can lead to nitrides with substantially more than
50 at-% N, but these phases are probably not in thermal equilibrium.
7.5.3 Phase Equilibria of Important Carbide Systems

A lot of effort has been made to model phase equilibria by the CALPHAD
(CALculation of PHAse Diagrams) method 1331. These methods, however, need
experimental results, the least information being that a phase occurs or not. Thus
a variety of calculated phase diagrams were published which are no longer relevant
because other phases have been found. In the following presentation phase diagrams
or portions mainly of such experimental investigation are presented and discussed if
they have appeared recently. Ordered phases [4,5] are generally not discussed, unless
they are interesting for industrial applications.
7.5.3.1 The Group IVB Carbide Systems

The group I V B transition metal carbon systems are very similar to each other con-
cerning the carbide phases: the f.c.c. monocarbides T i c l --x, ZrCl --x, and HfCl --x
show extensive homogeneity ranges and have high congruent melting temperatures.
Carbon does not stabilize the h.c.p. low-temperature modifications of the IVB metals
(in contrast to nitrogen) and the solubility of C in the CL phases is much smaller when
compared to the IVB transition metal nitrides (see below).
7.5.3.2 The Group VB Carbide Systems

The VB transition metakarbon systems (phase diagrams in Fig. 4) are again very
similar to each other and contain structurally identically subcarbide phases
PMe2C and <Me4C3pxas well as the f.c.c. monocarbide phases GMeC,-,. All
carbides of the composition pVzC have high- and low-temperatures modifications
which differ only by the arrangement of the carbon atoms [2] on the interstitial
positions. The exact composition (homogeneity range) and stability of the
c phases, (V4C3- . ~ (Nb4C3
, -.~,and cTa4C3- x, with a composition of around
42 at-% C and has been elucidated only recently [8]. The most interesting ordered
phase in this group occurs in the V-C system because a substantial increase in
microhardness occurs upon ordering [34] (Fig. 5 ) .
7.5.3.3 The Group VIB Transition Carbide Systems

The crystal structures of the phases in these systems are characterized by
destabilization of the octahedral structure elements and the appearance of trigonal
prismatic structural elements. F.c.c. monocarbides MeCl --x in the Mo-C and W-C
system exist only at high temperatures and with substoichiometric compositions.
The stoichiometric monocarbides WC and MoC are hexagonal.
The Cr-C system is characterized by the absence of a monocarbide phase. Three
carbide phases Cr23C6,Cr7C3,and Cr3C2with crystal structures that deviate from
the simple concept of a densely packed metal atom host lattice and interstitial
carbon atoms. A recent investigation on the homogeneity ranges have confirmed
the accepted values, except for the Cr23C6phase which has found to be lower in
carbon. Cr3C2has trigonal prismatic structural elements in an orthorhombic lattice.
L
2000
T/OC / "b I 2200
1800 T/ K
2000
1600
1800
1400
1600
1200 40 50
0 10 20 30
at% C
2000
2200
T/OC
T/ K
1800
2000
1600
1800
1400
0 10 20 30 40 50
Nb at% c
(b)
Figure 4. Recently reinvestigated portions of phase diagrams of the group VB carbide systems
(a) V-C, (b) N b C , (c) Ta-C [8].
Each carbon atom has eight metal neighbors in the form of a quadratic antiprism. In
that respect the structure differs basically from all other carbides.
7.5.3.4 Ternary Transition Metal Carbides

There are several families of ternary carbide phases with transition metals and main
group elements several of them being of structural interest such as the H-phases, the
filled p manganese type carbides and the K carbides. Of technical importance are
however only phases which are used as hard constituents in cemented carbides
2300
T/OC 2500
2200
T/ K
2400
21 00
2300
2000
2200
1900
2100
1800
2000
1700
I 1'0 20 30 40 50
Ta at% C
(C)
such as (Ti, W)C and the q phases [35] such as W3C03Cof which the occurrence in
decarburized compacts (e.g. near interfaces of CVD coated cutting tools) is
unwanted causing embrittlement.
F.c.c. transition metal carbides often form complete solid solutions with each
other particularly at high temperatures. They undergo phase separation at lower
temperatures [36,37]. Miscibility gaps have been observed in the quasibinary
systems Tic-ZrC, Tic-HfC, VC-NbC, and VC-TaC. At high temperatures also
33
T 31
m
a
(3
29 i
Q)
c
P
27 r"0
L
0
2
25
23
60 0 70 0.80 0 90
Composition [ C ] / M
Figure 5. Microhardness of ordered (B) and disordered (A) VC, --I as a function of composition
1341.
complete miscibility is observed in the systems where the f.c.c. phase is a high-
temperature phase at least for one binary constituent such as in the group IVB
carbide (f.c.c. WC and group VB carbide) f.c.c. MoC quasibinary systems. Even at
low temperatures the solid solubility of MoC and WC in the f.c.c. carbides of the IVB
and VB groups is substantial and is exploited for use in cutting tools.
7.5.4 Transition Metal-Nitrogen Systems and Structure of Phases

Several of the transition metal-nitrogen systems were established only recently and
there is still some lack of knowledge for systems for the investigation of which a high
nitrogen partial pressure is necessary. As for the carbides systems only recent results
will be presented here (see also [lo]).
7.5.4.1 The Group IVB Nitride Systems

The industrially very important Ti-N system features a high solubility of nitrogen in
aTi, three subnitride phases, qTi3N2-x,(Ti4N3-x and ETi2N, and the f.c.c. phase
GTiN, --x with the NaC1-type structure [38]. qTi3N2-x and cTi4N3-x have been
found to be isostructural with the hafnium subnitride phases and are characterized
by metal atom layers comprising cubic and hexagonal stacking sequences. Figure 6
shows the phase diagram of the Ti-N system. The system was recently modeled by
the CALPHAD method [39] where good agreement with the experimental results
was obtained.
The Zr-N system is characterized by the presence of only one nitride phase,
FZrN, --x, up to a composition of [N]/[Zr]= 1. It should be mentioned that none
3500 I I
T/OC 3500
3000 T/K
3000
2500
2500
2000
2000
1500
1500
1000
1000
500
0 10 20 30 40 50
Ti at% N
Figure 6. Phase diagram of the Ti-N system [38].
2500
T/OC 2500
2000 T/ K
2000
1500
1500
1000
1000
500
500
0
0 10 20 30 40 50
v at% N
Figure 7. Phase diagram of the V-N system [lo] with reinvestigated homogeneity ranges in the
temperature range 1100-1700°C (thick lines) [46].
of the subnitride phases which exist in the neighboring systems Ti-N and Hf-N
occur here. The Zr3N4phase [40] is obviously not of technical importance.
Besides the f.c.c. 6HfNl the Hf-N system contains two subnitride phases [41],
qHf3Nz-, and <Hf4N3-x.Only a few phase equilibria studies have been performed
for this system and they are cited in the ASM Handbook [lo]. The decomposition
temperatures of the subnitride phases need to be re-investigated.
7.5.4.2 The Group VB Nitride Systems

In the vanadium-nitrogen system the phases BV2N and 6VN1--x have h.c.p. and
f.c.c. structure, respectively [6,4245]. A recent investigation of the homogeneity
ranges by diffusion couples [46] has yielded a revised version of the phase diagram
mainly with respect to the subnitride phase of which the homogeneity was found to
reach up to 33.3 at-% N, the most probable value also from a crystallographic point
of view (Fig. 7).
The Nb-N system has already been subjected to many investigations [47,48].
Near the composition NbN at least three phases have been observed. For qNbN
two different structures have been proposed with respect to the nitrogen positions.
It contains trigonal prismatic and octahedral T6N structure elements and features a
metal layer sequence AABB. qNbN undergoes a phase transition at about 1320°C
(depending on the nitrogen pressure) into f.c.c. GNbNI --x with an NaC1-type
structure. Another phase, 6'NbNo.95,crystallizes in the anti-NiAs-type structure
which is closely related to qNbN but is most probably not stable and occurs only
transiently within the transformation of 6NbNl --x into qNbN. After recent in situ
investigations by high-temperature XRD [49] and differential scanning calorimetry
[7] as well as diffusion couple investigations [7,50] the phase reactions in the Nb-N
system seem to be clarified (Fig. 8). The GNbN1-, 4 yNb4N3*xtransition occurs
2000 j j
1400/ P-Nb2N
lzyyl
1000
0 10 20 30 40 50 60
Nb at% N
Figure 8. Phase diagram of the Nb-N system with reinvestigated homogeneity ranges and phase
reactions (thick lines and data points from electron-probe micoanalysis and thick broken line for
the yNb4N3+tx+ GNbNI --x transition [7]). The thin solid lines are previous data.
in the temperature range 112O-122O0C, depending on the composition. The

transition is of quasi-continuous character without a two-phase field between
yNb4Ns*, and GNbNI-,. By reacting Nb powder with nitrogen under high pres-
sure the composition of the GNbNI-, phase can be extended beyond [N]/[Nb]= 1
and the phase Nb5N6can be prepared.
The lower stability of the tantalum nitride phases as compared to the nitrides of
the IVB metals is reflected in the substantially higher equilibrium pressures [51]. In
order to prepare the high-temperature f.c.c. phase GTaN, - ,the nitrogen pressure
must exceed 10 bar. It is the hardest known transition metal nitride phase. The
pTa2N phase belongs to the &Fe2Ntype at low temperatures and undergoes an
order-disorder transition to an L’3 type between 1600 and 2000°C. TaN is expected
to melt at about 2800-3000°C and nitrogen equilibrium pressures of 1-3 kbar. The
BTaN phase with a WC-type structure is often observed in the reaction products
between Ta powder and nitrogen or ammonia at temperatures below 1100°C and
is believed to be a high-pressure phase which is metastable under normal pressures
[52]. ETaN has a structure not encountered in carbides, featuring an arrangement of
quadratic Ta5Npyramids. There are three hyperstoichiometric phases in the Ta-N
system with decreasing metallic character. The one with the highest nitrogen
content, the red Ta3N5,is nonmetallic and has been synthesized by the reaction
of TaC15 with NH3. In industrial applications these nitrides occur in the deposition
of tantalum nitrides for electronic devices (see 7.7.2).
Recent investigations by diffusion couples have yielded a revised version of a
phase diagram of the Ta-N system [8], which has also been modeled recently [53]
(Fig. 9a and 9b).
2000
- 2200
T/OC
1800
T/ K
- 2000
1600
- 1800
1400
0 10 20 30 40 50
Ta at% N
(a)
3500 *
3000
2500
s!3
E 2000
c
a
0.
+E, 1500
E-TaN t
,
1
1000
500 5 0.6 0.7 0.8

0 0:l 0:2 0:3 0:4 C
Mole-Fraction N
(b)
Figure 9. (a) phase diagram of the Ta-N system from experimental investigations [8]. (b) Phase
diagram modeled by the CALPHAD method [53].
7.5.4.3 The Cr-N and Mo-N Systems

The Cr-N system is characterized by the nitride phases pCr2N and GCrNI --x.The
former has an arrangement of the metal atoms is h.c.p. and the nitrogen atoms
are distributed in an ordered arrangement in the interstitial voids corresponding
to the &Fe2N-typestructure. CrN, --x crystallizes in the NaC1-type structure. It
has been found only recently [54]that this phase has a large homogeneity range
at high temperature and has only a line compound character at temperatures
below 1150°C. The phase diagram of this system (recently investigated on the
basis of diffusion couples) is given in Fig. 10. The substantial increase of the
Figure 10. Phase diagram of the Cr-N system [lo] with recent results on the homogeneity ranges of
Cr,N and CrN, - X . The homogeneity range of the latter broadens substantially upon increasing
temperature. The nitrogen-rich phase boundary of CrN,-, is for 30 bar N2 and extends most
probably at least up to 50 at-% N at higher pressures.
homogeneity range of CrNl --x is interesting and it is not completely clear whether
the line compound occurring below about 1200°Chas the same structure as CrNl - x
at higher temperature.
Cr2N is readily dissolved by dilute acids, whereas CrN appears to withstand acid
attack for prolonged periods. CrN is paramagnetic at room temperature and under-
goes a paramagnetic-antiferromagnetic first-order transition at temperatures of
276-286 K, depending on the composition [9].
In the Mo-N system three nitride phases have been reported 1551. An f.c.c. phase,
originally named yMo2N because it has a composition around 33 at-% N, should
instead be written as yMoNl - x , to be consistent with the notation for other f.c.c.
nitrides. yMoNlPx has been reported to also exist at nitrogen contents above
33 at-% N. It can be assumed that it is a phase with a very broad homogeneity
range. This is also indicated by the results obtained from high-pressure experiments.
yMoNl - x undergoes an ordering transition between 400 and 850"C, depending on
the composition, into the so-called pMo2N. Another phase, GMoN, with a structure
similar to that of WC, has been prepared by various methods, including nitridation
of Mo powder with flowing ammonia, or reaction of MoC15 with NH,. It has been
structurally characterized to have several types of ordering arrangements of nitro-
gen in the centers of trigonal prismatic structure elements [56].
Molybdenum nitrides are studied for their interesting properties in catalytic
reactions such as for the desulfurization and denitridation of oil products which
is of increased importance for use in cars where sulfur can destroy the activity
of the catalyzer. These powders can be produced with a high surface area by
reaction of MOO, or MoS2 with NH3. Also layers of molybdenum nitrides, pre-
pared by various techniques such as reactive sputtering or ion beam deposition,
were studied for application of wear-resistant coatings as well as for diffusion bar-
riers [57,58].
7.5.4.4 Ternary Transition Metal Nitrides

One of the most important ternary nitride system is the Ti-Al-N system, however,
the only technically important phase (Ti, A1)N is not thermodynamically stable.
Applications are described in 7.7.2.
The most important transition metal nitride, TIN, has complete miscibility with
ZrN, HfN, VN, and NbN at high temperatures but miscibility gaps occur upon low-
ering the temperature [36]. The two-phase mixtures which form upon demixing have
increased hardness and are thus interesting for commercial applications.
The Ti-Nb-N system is interesting because of the superconducting properties
of the (Ti, Nb)N phase having a slightly higher superconducting transition tem-
perature, T, [59] than the NbN phase (once the world record holder of T,).
Thus, Ti/Nb alloys (used for commercial superconducting applications dealing
with high magnetic fields such as nuclear magnetic resonance spectrometers in
medicine and for accelerator tubes in high-energy physics because of their excellent
processing properties) can be modified by nitrogen implantation or in-diffusion to
form a (Ti,Nb)N layer [60]. The TIN and NbN phases are completely miscible,
however, recently an indication for a spinodal-like decomposition in (Ti, Nb)N
was found to occur in the composition region around Tio.75Nbo.~sN-Tio.6~Nbo.37N
at high temperatures [61].
7.5.5 Carbonitride Systems

7.5.5.1 The Group IVB Carbonitride Systems
Because of the evident similarities between transition metal carbides and transition
metal nitrides, carbon can be replaced completely by nitrogen in the IVB and VB
group carbides. without changing the structure of the binary phases. Within the
IVB-VIB group, only one ternary phase Cr2(C,N)2 has been reported [62]. Solid
solubility in the VIB group carbonitride systems Cr-C-N and Mo-C-N is not
complete because in the boundary systems the crystal structures of the carbide
and nitride phases are different.
The most important ternary carbonitride system is the Ti-C-N system, followed
by the carbonitride systems of the other group IVB elements Zr-C-N and Hf-C-N.
The phase diagrams of the Ti-C-N and Zr-C-N systems are presented in Fig. 11
[63]. It can be seen that the binary f.c.c. carbides and nitrides show a complete quasi-
binary solid solution. The carbon and nitrogen atoms occupy all the octahedral
interstitial lattice sites randomly so that also the voids are randomly distributed
on these interstitial sites. Because of the wide range of homogeneity a large variation
of composition is possible for the preparation of carbonitrides powders and layers.
Most of these powders and layers contain 50 at-% C + N. The excellent properties
of Ti(C, N) have been exploited for cermets and hardmetals for a long time; and also
Zr(C, N) was recently introduced in industrial coating processes (see 7.7.2).
Ti
Tic TIN
at% N
(a)
Zr
ZrC 10 20 30 40
ZrN
at%N
(b)
Figure 11. Phase diagrams of (a) the Ti-C-N and (b) Zr-C-N systems for T = 1120°C [63].
7.5.5.2 The Ti-Mo-C-N and Ti-W-C-N Systems

The quaternary systems Ti-Mo-C-N [64] and Ti-W-C-N [65] are very important
because the quaternary phases (Ti, Mo)(C, N) and (Ti, W)(C, N) are used as a
7.5 Thermodynamics 223
I I I I I I
I \ I
0.3 X
LL
a 0.2
0.1
fcc
I
01 I I I I I I
0 0.1 0.2 0.3 0.4 0.5 0.6
Tic Site Fraction Mo
(a)
TIN G+graphite/G+WC+graphite WN
wc
Figure 12. Calculated phase diagrams of (a) the Ti-Mo-C-N [64] system for 1600°C with experi-
mental data of [67] and of (b) the Ti-W-C-N [65] system for 1423°C with experimental data of
[68]. Both for 50 at-% nonmetal in the representation of a reciprocal salt pair.
hard constituent in cermets [66]. These systems are technically most important for
50 at-% carbon + nitrogen and were described in terms of a section TiN-Tic-
MoN-MoC and TiN-Tic-WN-WC. Figure 12a, b presents the calculated version
of these phase diagrams for a nonmetal/metal ratio = 1 together with experimental
data. A homogeneous f.c.c. phase (Ti, Mo)(C, N) separates into two f.c.c. phases, of
which one is richer in Ti and N than the other (Fig. 12a). The two-phase field
f.c.c. + f.c.c.2 becomes smaller at higher temperatures.
A more complicated situation exists for the Ti-W-C-N system because of the pre-
sence of a hexagonal WC phase. Figure 12b shows the calculated phase diagram of
this system, also for a nonmetal/metal ratio of 1 [65] together with experimental data
[68]. Also two f.c.c. phases form and the two phase field is even larger than in the
corresponding system with Mo because of the higher nitrogen equilibrium
pressure. Upon increasing carbon concentration graphite and WC form together
with the two f.c.c. phases.
The phase equilibria in the Ti-Mo-C-N and Ti-W-C-N systems have influence
on the microstructures of cermets (see 7.7.1), at least partially responsible for a
favorable core-rim type microstructure of the hard particles [69].
7.6 Properties of Important Transition Metal Carbides,

Nitrides, and Carbonitrides
7.6.1 Melting Points
Amongst the transition metal carbides, one of the highest known melting tempera-
tures of any material has been measured: 6TaCo,89melts congruently at 3985°C and
6HfCo,94congruently at 3930°C (see Table 1 for the melting temperatures of other
carbide phases).
Melting points of various nitrides [70] have been measured under nitrogen pres-
sures up to 300 bar. The nitrogen pressure has to be chosen in such a way as to
be equal to the nitrogen equilibrium pressure of the nitride at its melting point.
The melting point temperatures of the various nitrides can reach more than
3000°C (e.g. for TIN), see Table 2.
7.6.2 Color
Transition metal carbides generally show a gray color, the exceptions are TaC and
NbC, which have dark-brown/bronze color.
Many mononitrides exhibit an intense color of bright metallic appearance which
depends on the nonmetal/metal ratio. Table2 lists the colors of the mononitrides
near the composition indicated by their formula. These colors can be observed
not only for bulk samples but also play an important role when thin films [71] are
produced for ornamental purposes (see 7.7.2).
Very interesting is the color change of the stoichiometric group IVB carbonitrides
+
as a function of the [C]/([C] [“I) ratio (Fig. 13). The higher number of the period
the more the vivid violet color extends into the ternary region starting from the
yellow colored nitrides.
7.6 Properties of Important Transition Metal Carbides, Nitrides, and Curbonitrides 225
+
Figure 13. Colors of the group IVB carbonitrides as a function of the [C]/([C] [N])ratio (which is x
in MeC,NI -.J. Note that the colored region extends wider into the carbon-rich region the higher is
the period number [73].
7.6.3 Thermal and Electrical Conductivities

The thermal conductivity of transition metal carbides increases with increasing
temperature, an unexpected phenomenon that has been investigated extensively
on titanium and zirconium carbides [72]. Previous studies have reported a linear
increase of the thermal conductivity with temperature, but more recent investi-
gations have detected a nonlinear relationship [73]. Carbon deficiency increases
the low temperature thermal conductivity of &TicI--x but decreases the high-
temperature thermal conductivity. Transition metal carbides have generally a
much lower thermal conductivity than isotypic transition metal nitrides. The
thermal conductivity of WC is much higher than that of the f.c.c. transition metal
carbides which can probably be explained by the stronger covalent bonding in
WC and hence heat transport by phonons.
Figure 14a illustrates the thermal conductivities for the group IVB nitrides,
carbides and carbonitrides. The substitution of nitrogen by carbon in group IVB
transition metal nitrides significantly reduces the thermal conductivity (compare
7.2.2).
A variety of other ternary and quaternary carbides and carbonitrides is especially
interesting for use in metal cutting tools in the form of layers, and powder particles
have been investigated recently for their specific heat, temperature and thermal con-
ductivity [66]. The behavior of the heat conductivity of most of these compounds is
similar to that of the group IVB carbonitrides for (near) stoichiometric composition
showing a positive curvature of heat conductivity as a function of temperature.
Interestingly, however, Nb- and Mo-containing ternary titanium carbides and
nitrides and quaternary titanium carbonitrides show a linear behavior in the same
temperature interval.
The similarity between f.c.c. transition metal carbides and nitrides is also valid for
room-temperature electrical resistivity which decreases with increasing carbon or
nitrogen content. Generally the transition metal carbides have higher electrical resis-
tivities than the transition metal nitrides (Tables 1 and 2). This is also reflected in the
50
45
F
5 40
-0,79
iij 25
a,
L - 0,4
20
-..- 0,oi
15 1 , 1 , 1 , ( , 1 , 1
0 200 400 600 800 1000 1200 1400

temperature [“C]
50
45
_----------
0.61
0,43
20 0,25
_ ,1.1.
0
15 I I I I I , , , , , , ,
0 200 400 600 800 1000 1200 1400

temperature [“C]
(b)
Figure 14. Thermal conductivity of group IVB carbonitrides with a nonmetal content close to
50 at-% as a function of temperature [73]: (a) Ti(C, N), (b) Zr(C, N), (c) Hf(C, N), The numbers
denote the [C]/([C]+ [N]) ratio.
resistivity of group IVB carbonitrides (Fig. 15) which shows a general increase of resis-
tivity with increasing [C]/([C] + [q) ratio. Only at around [C]/([C] [N]) = 0.5 the +
resistivity increase is much less than at high and low carbon contents, respectively.
Some transition metal carbides are superconducting. The highest T, value has
been reported for NbC (1 1.1 K) and TaC (10.3 K) and hexagonal WC (10.OK).
7.6 Properties of Important Transition Metal Carbides, Nitrides, and Carbonitrides 227
50
45
z-
i.$ 40
35
.-
c
U
$ 30
8
g 25
c
20
15
0 200 400 600 800 1000 1200 1400
temperature ["C]
(C)
Tic, VC, HfC, and ZrC are not superconducting above 1 K. The electrical con-
.
ductivities of transition metal carbides and nitrides are greatly influenced by the
nonmetal/metal ratio; as this ratio approaches unity the electrical conductivity
reaches a maximum. This is shown for FTiNl --x in Fig. 16 [74]. A similar behavior
can be observed for other carbides and nitrides. The electrical conductivities of these
compounds decrease linearly with increasing temperature [75].
Superconductivity is common among all f.c.c. transition metal nitrides of the
IVB and VB groups as well as yMoN, -.x and GMoN. GNbN, -.x shows the highest
7
0.06
0.05
E 0.04
u
R
F
b
m
0.03
0
0.02
O.O10
0 0.2 0.4 0.6 0.8 1
COMPOSITION, [C]I([Cl+[NI)
+
Figure 15. Electrical conductivity of group IVB carbonitrides as a function of the [C]/([C] [N])
+
ratio [74]: Ti(C,N), Zr(C, N), 0 Hf(C, N).
0,4
0,50 0,60 0,70 0,80 0,90 1 ,oo
composition [N]/p]
Figure 16. Electrical conductivity of TiN, --x as a function of composition [74].
superconducting transition temperature of 17 K, which was the highest known of

any material in the early superconductivity research. It is noteworthy that at
around 15 K, hexagonal GMoN also has one of the highest transition temperatures
of these materials [56] and f.c.c. yMoNl,oo,although never prepared with a stoichio-
metric composition, was predicted theoretically to have a T, of up to 30K. All
transition metal nitrides are type I1 (hard) superconductors. For several f.c.c. nitrides,
such as GTiN1--x,GZrN1--x, and 6HfNl --x (data reported for 6HfN1--x are, however,
contradictory) T, increases linearly with increasing nitrogen content up to [Nl/[T]= 1
[76]. Table 1 gives the highest T, for each compound in absolute temperature.
7.6.4 Thermal Expansion

Thermal expansion is a very important property of transition metal carbides,
nitrides and carbonitrides because they are practically never used in pure form
but mostly in contact with other materials (metals). Upon thermal load different
thermal expansion of the carbide phase and the substrate may cause degradation
of the composite. Generally their thermal expansion is higher than that of the
corresponding metals.
The thermal expansion of WC has recently been compiled [77]: the volume expan-
sion increases steeply up to 15 x lop6 K-' at 1000 K. Figure 17 gives the data
calculated also for higher than ambient pressure.
The thermal expansion of various transition metal nitrides has been determined
for both low temperatures [78] down to 8 K and for high temperatures [79] up to
around 1500K. 6VN1--x exhibits a very low or even negative thermal expansion
7.6 Properties of Important Transition Metul Carbides, Nitrides, and Carbonitrides 229
25
20
.-'Y
h
15
% v
v
d lo
0
100 1100 2100 3100 4100
T (K)
Figure 17. Thermal expansion data of WC for ambient and higher pressures [77]. Reprinted by per-
mission of The American Ceramic Society.
at temperatures below than 70 K. The high-temperature thermal expansion is given in

Fig. 18 [79]. Interestingly, considering its low-temperature thermal expansion,
6VN1 has the highest high-temperature expansion coefficient up to
1 I x lop6K-I among these compounds.
For the group IVB carbonitrides also data for well characterized f.c.c. phases
exist [80]. The average linear thermal expansion (which is the linearized thermal
expansion as a function of temperature) for the three carbonitrides is:
GTi(C,N1_,y): LY,,
= (9.9 - 1.4 x [C]/([C]+ [N]))x 10-6K-1
GZr(C,NI -.y): a,, = (7.8 - 0.3 x + [N])) x lop6K-I
[C]/([C]
GHf(C,NI Q,, = (8.5 - 2.4 x [C]/([C]+ [N])) x loP6 K-',
where [C] and [N] are the carbon and nitrogen contents, respectively. These
+
expressions are valid for 50at-% [C] [N]. It can be realized that Zr(C,N) has
the smallest thermal expansion coefficient at nitrogen-rich compositions, and also
the smallest increase as a function of increasing carbon content. For the (nonaver-
aged) linear thermal expansion fitted by polynomial expressions see [go].
7.6.5 Diffusivities
In transition metal carbides, nitrides, and carbonitrides, nitrogen diffusion occurs via
a vacancy mechanism in the nonmetal sublattice. Because the vacancy concentration
12.0 -
1
11.0 -
-
h
id
L 10.0 -
v
8
e
.-
9.0 -
-
2
0
8.0-
-3
.-
L
-
0
C
I! 1.0 -
f
8
U
6.0 -
5.0 -
, 1 1 1 1 1 1 1 1 1 1 1 , 1 1 1
300 500 700 900 1100 1300 I500 1700
Figure 18. High-temperature thermal expansion of various transition metal nitrides [79].
is a function of composition within the homogeneity range of the f.c.c. phases, the
nitrogen diffusivity is often composition-dependent as well. This has only been
recently started to be investigated in transition metal carbides and nitrides for
which a detailed analysis of the nitrogen diffusion profile is a necessary pre-requisite.
Many diffusivity studies were conducted by investigating nitride or carbide layer
growth as a function of time, which gives the average nitrogen diffusivity in the
various phases [81]. A parabolic relationship, d = kt’/*, holds for this layer
growth, where d is the layer thickness, k is a constant, and t is the diffusion time.
The parabolic behavior indicates that the nitride layer growth is indeed diffusion-
controlled, provided that the sample is semi-infinite. Recent efforts for investigation
this layer growth has shown that the thickness of the sample can have crucial
influence on the layer thickness [82,83]. Methods were developed using wedge-
shaped diffusion couples in order to exploit this behavior for diffusivity
measurements [84]. A wedge shaped diffusion couple is presented in Fig. 19,
which shows the thickness enhancement of diffusion layers upon decreasing
sample thickness.
7.6 Properties of Important Transition Metal Carbides, Nitrides, and Carbonitrides 23 1
Figure 19. Diffusion layers of hafnium nitrides formed in a wedge-type sample. Note the thickness
enhancement upon decreasing sample thickness [82].
A quantitative description of the dependency of diffusivity on the composition in

f.c.c. carbides and nitrides can be made by the expression
D =f3dexp(-E/(kBT) exp(a(c+ - c)) [cm2/s],
wherefis 1/12 for f.c.c. compounds, v is the jump frequency of the nonmetal atoms
NM, d is the mean distance between lattice planes, E is the activation energy, kB is
the Boltzmann constant, and a is a parameter for the steepness of concentration
dependency. The concentration c is in mol NM x cm-3 and c+ is the maximum
concentration (in [NM]/[M] units this value is 1). Hence the diffusivity term is
split into the activation energy term and a concentration-dependent term. The
product f v 2 d is often expressed through a single factor, Do, the so-called pre-
exponential factor and is in fact an extrapolation to infinite temperature, so that
the accuracy is poor, especially if only a small temperature interval is investigated.
A detailed study of concentration-dependent diffusivities was conducted for
GTiCI--y and GZrCI--x [85], GNbCI-,x and GNbNI--x [86], GTaC1--x and GTaN1--x
[87], and GCrN, --x [54]. These compounds show a large concentration-dependent
diffusivity the higher the more nonmetal deficient the compound is and is a result
of the vacancy diffusion mechanism.
Further concentration-independent diffusion coefficients were measured for
vanadium nitrides [46], hafnium nitrides [88], and chromium carbides, GTiNI--x
[89] and GZrNI--x [90]. Although the metal diffusivity in transition metal nitrides
has not yet been investigated, the activation energy of that process is much higher
than for nitrogen diffusion and can be estimated to be of the order of about 8eV.
Tables 3a and 3b summarize some nonmetal diffusivity data of transition metal
carbides and nitrides.
7.6.6 Elastic Properties

Transition metal carbides, nitrides and carbonitrides exhibit high Young’s moduli,
appreciably higher than that of the transition metals. Many of the carbide and
Table 3. (a) Concentration dependency of the nonmetal diffusivity of transition metal carbides and
nitrides (for compounds with a = 0 no significant concentration dependency is present). If two
values are listed the first value corresponds to the nonmetal rich and the second to the nonmetal
poor composition.
&Tici 0.058 3.12 112

GZrCi 0.167 3.59 144
GNbCI - 0.065 3.53 98
GTaCi - x 0.004 3.28 99
GNbNi - 0.43 3.13 20
PTa2N 1.40 3.3 0
GCrN, 0.018 2.92 184
Table 3. (b) Concentration independent non-metal diffusivities in

transition metal carbides and nitrides.
2.9 3.95
6.5
20 5.0
1.28 3.2
0.718 2.9
0.0121 2.5
0.02 2.70
0.14 2.14
0.54 2.73
12.7 2.93
13.6 2.92
3.4 3.6
120 4.0
3.51 2.68
nitride phases were investigated for their elastic properties, mainly the Young’s
modulus E, and the Poisson’s ratio v. Both, bulk samples as well as thin film samples
were measured and generally polycrystalline material was studied [911.
Among these studies, many of the data suffer from lacking or inappropriate
characterization of composition and microstructure. The porosity of bulk samples
as well as the substrate of thin films can influence the results substantially. Correc-
tion procedures for accounting for such influences are sometimes inappropriate [92].
Depending on the correction function, 5% porosity can cause a drop in E of about
50 GPa.
For T i c most values for the Young’s modulus group around 450-460 GPa for a
composition around 50 at-% carbon, the Poisson ratio is 0.17-0.19. The Young’s
modulus of the important hexagonal phase WC is very high at room temperature
[77] and drops by about 20% upon reaching 2000 K (Fig. 20).
Although the measured Young’s modulus of TIN1-, show, quite a scatter
because of the above-mentioned problems, it can be stated that E increases with
L
0 Kauk &Kockelmanna
a
28
0 Kreimer
24
0 Lee 8 Gilmre (polyuys.)
v1 0 Koster L Rauscher ”
3
n
3 0 Bukatov et al.
W
0 + Brown et al.
25
v) 0
X Neshpor 8 Sarnsonov
C
ET)
c + C
3
0
600 X
550
0 1000 2000 3000 4000
Figure 20. Young’s modulus of WC as a function of temperature of various authors compiled by

[77]. Reprinted by permission of The American Ceramic Society.
increasing nitrogen content. Many of the thin film samples show a higher Young’s
modulus than bulk samples (see Fig. 21). Poisson’s ratios of 0.18-0.321 were
reported, the most probable value is likely to be around 0.22. Similar statements
can be made for ZrNl-, and H f N - , with values reaching up to 500 and
459 GPa, respectively (both values are for different kind of bulk samples: diffusion
annealed or hot-pressed) and with Y = 0.25 for both compounds. Young’s modulus
data on group VB transition metal nitrides are scarce but were measured for single
crystals of 6VN and 6NbN [93], yielding the maximum values in the (100) direction
of 487 and 490 GPa, respectively. For CrN, values up to 520 GPa were reported.
For Ti(C,N) an increase of E with carbon content was measured on several
samples (see citations in [92]), the porosity of which was, however, rather larger
and different in different samples. A well characterized sample Ti(C0.20N0.80) with
a porosity of only 3.5% [63] yielded the highest value of 451 GPa (and v = 0.212),
which probably comes very near to the value for an ideal zero-porosity samples. An
investigation of the E and G behavior as a function of temperature was recently
conducted on polycrystalline samples of TIN [94], where a linear relationship was
found for both quantities (see Fig. 22).
For an extensive review on the elastic properties ( E and Y) of transition metal
nitrides, carbonitrides (and carbides) as well as on the porosity correction routines
the reader is referred to the tables given in [92]. Upon application and development
of high-temperature investigation techniques, such as Brillouin scattering [95,96] a
800 I I I I I -
0
a 600 -
(3
Y
Y,
3
-
3 0
400 -
0
tn
c3)
c
13
0
>- a
n
"
0.0 0.2 0.4 0.6 0.8 1 .o
0 hot-pressed 0 sintered A not stated # diffusion annealed

0 zoneannealedfilm 0 film A single crystal
Figure 21. Room temperature Young's modulus of TiN, - as a function of composition and for
different sample types [92].
gain in knowledge on the high-temperature properties of these materials (even in

single crystalline form) can be expected in the near future.
7.6.7 Microhardness
Sometimes large differences are encountered between the microhardness of bulk
material and thin films. This is most probably due to the different dislocation density
and the grain size of both types of materials. Also the difference in testing (micro- or
nano-indentation for films, heavier load up to several N for bulk samples) most
probably yield differences. In addition, the composition (as for any of the properties
discussed here) influences the microhardness. Because of these peculiarities the
values are difficult to compare.
Many of the transition metal carbides such as T i c and ordered VCI --x are very
hard compounds. In GTiC1- the microhardness increases with increasing carbon
content [97], a phenomenon that is probably closely related to the valence electron
concentration (VEC) with a maximum stability at VEC = 8 at the composition TIC.
435 I I I I 180
430 - 175
425 - 170 F
- 9
420 165 0
rn
41 5 - 160 53
U
41 0 - 155
Q
405 - 150 f
0)
400 - 145
395 I I I I
140
Figure 22. Young’s modulus ( E ) and shear modulus (G) of TIN as a function of temperature for a
composition of 50 at-% N [94].
Hardness data for transition metal carbides near the composition indicated by the
formula are contained in Table 1.
Single crystal investigations and orientation-dependent measurements on poly-
crystalline material of transition metal carbides show that the microhardness is
orientation dependent [98] (e.g. for WC and NbC [99]). For T i c the Knoop hardness
in the (1 10) plane is 27 GPa whereas that in (100) plane is 31 GPa, for example.
Upon ordering the microhardness can change significantly (compare Fig. 5). Gen-
erally the hardness values of annealed crystals are lower than as-grown crystals
due to their lower dislocation density. Polycrystalline material has higher values
by about 1-2GPa than single crystals due to the grain boundary influence on
hardness (Hall-Petch relation).
Single-crystal and polycrystalline transition metal carbides have also been
investigated with respect to creep, plasticity and slip systems. The f.c.c. carbides
show slip upon mechanical load within the (111) plane in the (110) direction
[98,100]. The ductile-to-brittle transformation temperature of TIC is about 800°C
and is dependent on the grain size. The yield stress in TIC obeys a Hall-Petch
type of relation, that is, the yield stress is inversely proportional to the square
root of the grain size. TIC and ZrC show plastic deformation at surprisingly low
temperatures around 1000°C.
Figures 23a-e show the hardness vs. composition for a variety of transition metal
nitrides [43,74,79,101] and carbonitrides [88]. The hardness decreases with
increasing nitrogen content for the group VB nitrides [43,79,101]. This behavior
can probably be explained by an electronically induced lattice softening as a
result of the generation of antibonding states. It is known from band structure
26
25
24
23
19
18
17
16
0,50 0,60 0,70 0,80 0,90 1,oo
composition [N]/[Ti]
(a)
35
30
0
0,70 0,80 0,90 1,oo
composition "]/[Me]
(b)
Figure 23. Microhardness of nitrides and carbonitrides as a function of composition. (a) TiN [74],
(b) VN [43], NbN and TaN [loll, (c) Cr-N [54], (d) Ti(C, N) +, and Ti(C, B, (e) Zr(C, N) +
and Hf(C, N) [73].
231
7.6 Properties of Important Transition Metal Carbides, Nitrides, and Carbonitrides
1
1
50
0,8
03
composition [C]/([C]+[N])
40
composition [C]/([C]+[N])
Composition [at% N]
0,6
0,6
1
30
p-Cr,N
I
0,4
0,4
20
1
I0
02
02
I
0
0
0
12
18
16
14
-m1 4 1
28
26
20
E 18
16
24
22
14
24
22
g 20
a 12
(e)
(d)
E
2
E
(C)
P
E
.-0
gc
rn
ln
ln
.-
calculations that in f.c.c. transition metal compounds with more than eight valence
electrons the antibonding states are occupied. An increasing valence electron con-
centration due to increasing nitrogen content would then lead to a decrease in
hardness when the number of valence electrons exceeds eight. Indeed in GTiN, --x
[74] a maximum in microhardness was found for the compound STiN0.67
corresponding to 7.3 valence electrons. For 6HfNl --x, however, a smooth increase
in microhardness with increasing nitrogen content was observed. This cannot be
explained in the same manner, and is probably a result of the difference in electronic
states due to 5f electrons. The microhardness of chromium nitrides [54] is shown in
Fig. 23c as obtained from diffusion couples. Upon use of diffusion couples, which
exhibit thick diffusion layers, bulk values of the microhardnesses can be measured
as a function of composition, provided that the nitrogen diffusion profile is
measured.
The microhardness of the IVB group carbonitrides (Fig. 23d, e) increases with
increasing [C]/([C]+ [Nl) ratio. For Zr(C, N) and Hf(C, N) the increase is steep on
the nitrogen-rich side and flattens on the carbon-rich side, so that a significant
positive deviation from linearity can be observed. Contrary, for Ti(C, N) this devia-
tion is only very small for 50 at-% nonmetal but significant for a substoichiometric
composition of 45 at-% nonmetal.
7.7 Industrial Applications

There is a widespread field of applications of transition metal carbides, nitrides and
carbonitrides. The most important applications are:
- powders for hardmetals, cermets and spray coatings,
- deposited layers made by various techniques such as PVD and CVD, as well as
- diffusion layers for surface-modified components made by high-temperature and
plasma techniques.
The layers are prepared for enhancing the abrasive resistance of the components, for
ornamental purposes and layers in electronic devices, most of them for preventing
the interaction of Si with other metals.
The three application fields are discussed in more detail later. It should be noted
that a variety of potential or just beginning applications mainly of nitride films are
reported in the literature and discussed below.
7.7.1 Cemented Carbides and Carbonitrides

The most widely used transition metal carbide is tungsten carbide, hexagonal WC,
which is employed as the hard constituent in WC-Co hardmetals. Such hardmetals
are sintered composite materials with 80-90% of hard particles such as WC embedded
in a ductile binder phase such as Co. For these applicationsWC combines a number of
P Atomic fraction of tungsten

Figure 24. Phase diagram of the W-C-Co system for 1423°C[102]. The top C corner is not shown.
favorable properties such as high hardness and wear resistance, good electrical and
heat conductivities and is readily wetted by the binder metals Co and Ni.
The ternary system W-C-Co is shown in Fig. 24 at 1423°C [102]. WC coexists with
the Co phase and forms a pseudobinary lower melting eutectic between WC and Co.
This eutectic facilitates full densification during liquid phase sintering. The carbon
content must be kept close to the theoretical value because a decrease in total
carbon content will lead to the formation of -q carbides (W, Co)& which are brittle
and harmful to the performance of these materials. In modern hardmetals, also
f.c.c. carbides such as TaC, NbC and T i c are admixed which increase the high-tem-
perature performance. Figure 25 shows a microstructure of a modern WC/Co-based
hardmetal containing about 10weight-% of f.c.c. carbides Tic, TaC, and NbC.
Shortly after the invention of WC-Co hardmetals, T i c based hardmetals have
been proposed as cutting tools but have not found extended use because of their
brittleness. Only in the last two decades Tic-based hardmetals have found wider
use when titanium nitride, GTiN, had been introduced as a further constituent.
GTiN is used as one of the hard components in sintered carbonitride compositions
containing TIC, Mo2C, WC, TaC, NbC, and other compounds [69,103]. Carboni-
trides for such materials can be produced either by blending the various powders
before liquid phase sintering with Ni and/or Co as a binder phase, or directly
within the liquid phase sintering cycle. The modern titanium carbonitride cermets
[ 103,1041 are characterized microstructurally by a structured hard phase featuring
core-rim structures embedded in a tough Co-Ni binder. These carbonitrides are
often called cermets or Ti(C, N) hardmetals, a microstructure of which is given in
Fig. 26. The hard particles a roundly shaped as compared to the faceted WC
grains (Fig. 25). In the microstructure of Fig. 26, which is a SEM image, core-
rim type grains with a dark core and a gray rim as well as grains with a white
core and a gray rim can be seen.
240 7 Transition Metal Carbides, Nitrides, trnd Carbonitrides
Figure 25. Microstructure (light-optical microscope) of a WC/Co-based hardmetal (WIDIA TTR)

containing hexagonal WC and cubic carbides (right: cubic phase, left hexagonal phase enhanced by
specific etching).
Figure 26. Microstructure (SEM image) of a cemented carbonitride (WIDIA TTI 25) showing
round hard particles with core-rim type structure.
The boundary between hardmetals and cermets is not strict because many of these
compacts resemble microstructure features of both type of materials [ 1061 faceted
WC crystals together with round-shaped titanium carbonitride-based hard particles.
Generally, these titanium carbonitride hardmetals are comparable with respect to
properties and microstructure to WC-based hardmetals. The powders of these
materials are liquid phase sintered with Ni or Ni-Co binder metal alloys. The
core-and-rim structure of the hard phase usually exhibit a molybdenum- and
carbon-rich (Ti, Mo)C rim and a titanium- and nitrogen-rich Ti(C, N) but can
also be inverted (compare Fig. 26). The metallurgy of the phase reactions is (because
of the complexity of the multicomponent system) not yet fully understood [69].
According to the theoretical background of the phase reactions within the
system Ti-Mo-C-N [67] the quaternary compound (Ti, Mo)(C, N) with a random
distribution of the metals Ti and Mo on the metal sites and the nonmetals C and
N on the nonmetal sites undergoes a decomposition with the formation of isotypic
f.c.c. phases but with a preferential arrangement of N around the Ti atoms and Mo
around the C atoms. This decomposition is of a spinodal type, which is a decompo-
sition induced by local composition fluctuations without necessary activation energy
rather than a classical nucleation and growth process. The question of whether the
spinodal decomposition is a necessary prerequisite of core-and-rim formation is still
unsettled because core-and-rim type structures may develop via reactions of the
hard constituents with the liquid binder metal as well. The core-rim structure has
turned out to be most probably stable [64, 1051. It has been claimed to be favorable
because the molybdenum-containing rim is much better wetted by the binder alloy
than the core. Modern Ti(C, N) hardmetals have been successfully applied for finish-
ing operations of high-strength steel grades and ductile cast irons at high cutting
speeds but with moderate cross section of the chips. The main advantages of
Ti(C, N) hardmetals are the high chemical resistance and the low frictional welding
tendency giving longer life of the cutting edge and good workpiece surface.
Hardmetal-like microstructures are also produced by spray coating processes in
which material is deposited by high velocity. Sprayed parts are used in a variety
of processes where sliding wear occurs (aerospace industry, stone industry), advan-
tage being that the parts can be refurbished. Various methods have been developed,
such as detonation gun spraying and high-velocity oxygen fuel spraying. With these
techniques layers of WC-Co, Cr3C2-Ni/Cr, and TiC/Ni mixtures are deposited with
a thickness up to the mm range [107].
7.7.2 Deposited Layers

Transition metal carbide, nitride and carbonitride layers of the IVB group made
by CVD (chemical vapor deposition) and PVD (physical vapor deposition) are
produced in order to increase the service life of hardmetal tools [18,21,108] for
cutting and milling operations. About 90% of cutting tools are coated.
The surface of such an insert must resist interaction with the workpiece (e.g. steel)
and environment (oxidation), the adherence to the sintered compact must be
excellent and the material in itself must have high hardness, strength and fracture
Figure 27. Milling and turning inserts with various coatings. The top coating is TiN (golden yellow),
Zr(C, N) (violet) and A1203 (black). From WIDIA, Germany.
toughness. A number of binary, ternary, and quaternary compounds were studied

and are actually in use, a large number of them contain titanium. Figure 27
shows various milling an turning inserts, which are multilayers coated with TiN,
Ti(C, N), Zr(C, N) together with A1203.
After the early established monolayers such as of TIN, multilayers composed of
TIN, TIC, Ti(C, N), and A1203were developed. Frequently used layer systems are
TiC/Ti(C, N)/TiN and TiN/Ti(C, N)/TiN, which have better wear and thermal
cracking resistance than TiN monolayers [109]. Although Ti(C, N) is suitable for
interaction with the workpiece the top layer is often TIN because of its appealing
golden yellow color. Figure 28 shows a microstructure of such a TiN/Ti(C,N)/
TiN coating produced by WIDIA for a hardmetal grade used to turn steels.
A1203is often used to provide a ceramic material in contact with the workpiece or
to apply several multilayers of the type A1203/TiN.The thicknesses of the sublayers
are on the order of about 1 pm and the total layer thickness is about 10-12 pm. Such
a structure is shown in Fig. 29 for the WIDIA grade TN150. Most of these layers are
deposited by CVD. Also a combination of PVD and CVD (‘duplex’) techniques was
studied and has yielded better performing layers than other titanium nitride and
carbonitride coatings [110].
The wear mechanism of multilayer coatings is assumed to occur layer by layer and
the cracks are deflected by the interfaces between the sublayers. Besides cutting test
the performance of coatings can be tested (at least compared) by scratch tests, in
which a critical load is obtained at which the coatings are removed. The critical
loads are typically 50-IOON and are influenced by the type of the deposition
process. Also the obtained hardness is dependent on the preparation method;
magnetron sputtered Ti(C, N) layers reach up to about 37 GPa [108].
For ornamental layers bright golden yellow TIN and yellow-green ZrN, as well as
Figure 28. Microstructure of a coated WC/Co-based cutting tool for turning steel (WIDIA TN250)
where the layer is composed of three sublayers.
(Ti,Al)N gold + dark blue, depending on the Ti/AI ratio, Ti(C,N), Zr(C,N)
(colors: see Fig. 13), CrN (metallic to brown) are used [l 1I, 1121 for example on
watch cases, lighters, frames for glasses, and bathroom furnishings where the
combination of color with increased scratch and corrosion resistance is exploited.
The color of these materials stems from the absorption behavior, which is the
result of the response of the material to the incident light, in turn a function of
Figure 29. Microstructure of a multilayer-coated steel-turning hardmetal (WIDIA TN150), with

various TiN/A1203 sublayers a TIN and a Ti(C, N) layer.
the electron density of states. The CIE system is widely applied for defining colors.
This is a plot of in the coordinates L* (lightness), a* (red-green value), and b*
(yellow-blue). The color can be also dependent on the surface roughness, the
layer thickness and the structural disordering. For example, in TiN the color
appearance is compositionally dominated whereas in ZrN it is microstructure domi-
nated. Thus, a certain variability exists for establishing colors apart from composi-
tional point of view and related materials can behave differently. However, this
behavior brings some difficulties with respect to the reproducibility of colors
which in turn is influenced by the color perception of the human eye [113].
The deposition of TIN and ZrN on glass for windows represents a combination of
ornamental application with the economic constraint of energy saving in buildings.
ZrN layers, especially, have been found to exhibit favorable properties for transparent
heat mirrors, i.e. windows which transmit visible light but reflect heat radiation [114].
Chromium nitride layers (fabricated by, e.g. cathodic arc plasma deposition) are
interesting because of their corrosion properties as well as because of their excellent
adhesion properties and fine-grained structure. They are applied for die-casting
moulds where excellent edge properties are necessary [115,116]; some of these
layer can have a multiphase character composed of Cr(N), Cr2N, and CrN1-,
[117]. Sputter deposited ternary chromium nitrides such as Cr,Me, -,N with
Me =Ti, Nb, Mo, and W additions and with grain sizes of up to 25 nm have been
found [118] to show either a hardness minimum (Me = Mo, Ti) or a maximum of up
to 27 GPa (Me = W, Nb).
The Ti-A1-N system contains three ternary nitride phases but the most important
phase for industrial applications is the thermodynamically unstable f.c.c. phase
(Ti, A1)N [119,120], which can be prepared by magnetron sputtering and has
found application for hardmetal tools [121-1231. (Ti,Al)N was found to exhibit a
better oxidation resistance than TiN, depending on the Ti/Al ratio which can
vary over a broad range. At high A1 contents a hexagonal phase forms probably
related to or identical with the wurtzite-type AIN. Drill bits coated with (Ti, A1)N
show a better performance than coated with TIN. The best results (about two-
fold lifetime) were obtained with (Tio,5Alo.5)N. Film hardnesses are on the order
of 20-26 GPa, depending on the composition.
Layers composed of (Ti, Zr)N show a higher hardness (up to 33.5 GPa) but the
performance is not as good as of (Ti,Al)N [121]. In (Ti,Zr)N layers a spinodal
decomposition around the composition (Tio.sZro.7)Nwas found [ 1241 which was
also described for the phase equilibria of bulk samples [36,125].
Zr(C,N) layers together with a variety of other layers were successfully intro-
duced for cutting tools for turning steel. A microstructure of such a multilayer is
shown in Fig. 30 where Zr(C, N) represents a thick top layer. The favorable perfor-
mance of such Zr(C, N) containing layers stems for the thermal expansion behavior
of Zr(C,N), which is the lowest amongst the group IV carbonitrides, and which
reduces tensile stress upon thermal load. Such a behavior enhances the performance
of the coatings because of increased crack resistance and adhesion.
While the first and second generation of coatings contained monolayers of TiN or
T i c and multilayers composed of TiN, Ti(C, N), TIC, TiB2, and (Ti, A1)N (together
with A1203)there is a general trend in research towards multicomponent/multiphase
+ Zr(C,N)
+ AI,O,
+ Ti(C,N)
6 TIN
f- Substrate
10 pm
H
Figure 30. Microstructure of a Zr(C, N) top-layer coated steel turning hardmetal (WIDIA TN7015).
coatings. Recently, layers composed of (Ti, - ,- y - zA1,CryY,)N, with y = 0.03 and

z = 0.02, to were found to have an even higher oxidation resistance than (Ti, A1)N
layers, with an onset of rapid oxidation about 80°C higher than of the latter [126].
The Knoop hardness is dependent on the composition and reaches values around
27 GPa.
A further example of such thin films are the superlattices, which are in principle
composed of two different nitrides, such as TiN/NbN and TiN/VN, which are
successively deposited by PVD techniques such magnetron sputtering with a thick-
ness in the nm range [127,128]. Hardnesses of up to about 50 GPa were measured on
such layers depending on the thickness of the individual sublayers, called wave-
length, which reach a maximum at around 5 nm.
The most recent development which has, in addition, already be applied in
industrial scale, is the preparation of superhard nano-crystalline layers by a combined
plasma-CVD/PVD method. Two-phase layers of the type ncTiN/cBN and ncTiN/
aSi3N4were prepared, where nc denotes nano-crystalline, c cubic, and a amorphous
[ 129-1 3 13. Hardnesses of up to 70 GPa were measured. The grain size of the crystalline
phases of these materials is of the order of a few nm, whereas the maximum hard-
nesses are observed at the lower end of grain sizes studied (about 2 nm).
The hardness increase of such multilayers or nanocrystalline coatings as compared
to conventional bulk nitrides has not yet completely clarified. Three different phenom-
ena were discussed [ 1321, which are known to increase the hardness of a material. The
+
well-known Hall-Petch type grain size hardening ( H = Ho k / d ’ ” , where H =
hardness, Ho = single-crystal hardness, k = constant, d = grain size), is probably
only responsible for a minor extent, the coherency strain and the dislocation line
energy effect. From the experimental finding that also a combination of layers prac-
tically without a lattice parameter mismatch (e.g. TiN/Nbo,4Vo,6N; mismatch: <0.2%
[127]) also show a hardness increase, coherency strain effects (in which strain is caused
due to the mismatch) can also be ruled out at least for a part of the superlattices. The
presence of dislocation line energy effects in which the shear modulus is important to
move a dislocation (line energy= Gb2/2,where G is the shear modulus and b is the
Burgers vector) can also contribute to the hardness increase but the observed increase
is much higher than calculated from this theory (there is, however, some uncertainty
in these calculation due to the lack of data for G ) .
While a Hall-Petch type hardening is probably not the most prevailing
mechanism in the above cited multilayers, it has been made responsible in nano-
crystalline layers composed of two phases rather different in nature (crystalline,
amorphous or different crystal structure). Although extremely fine-sized powder
can show a negative Hall-Petch effect (nanosoftening), films composed of two
such different phases having strong interphase boundaries suppress such effect
resulting in a drastic increase in hardness.
Films of the Ti-B-N system [133,134] consisting of a thermodynamically
unstable f.c.c. phase Ti(B, N), where the B/N ratio can be varied, were also found
to show increased hardness of up to 50 GPa. These films can be made single crystal-
line, the hardness of which is dependent on the ion bombardment intensity. Thus a
dislocation density mechanism seems to be responsible for the hardening effect.
Nitrides in the form of thin layers have an important application in electronic
devices such as VLSI (very large scale integrated circuits) as Ohmic interface
materials to prevent the formation of intermediate phases between Si and A1 or,
in the most recent developments, where A1 is replaced by Cu because of the better
electromigration properties upon continuing miniaturization, between Si and Cu.
This application uses the chemical inertness, their high electrical conductivities
and the ease with which nitrides can be deposited as layers. Mainly TiN deposited
by PVD techniques is used in these circuits but also tungsten and molybdenum
nitrides [58]. Also Tio.8A10,2N was studied and has shown a better performance
than TIN [135]. For the new Si/Cu interconnection, TaN has proved to be of super-
ior performance [136]. Films of tantalum nitrides are also used or of potential use for
a variety of electronic applications ranging from thin-film resistors as well as for
thermal printing heads and for wear and corrosion resistant coatings [137]. These
films can be deposited by CVD with and without plasma [I 381 and PVD methods
[136,137]. For CVD methods organic precursors [139,140], TaC1, [141], or TaBr5
[138] can be used. If a low resistivity of the tantalum nitride films is wanted, then
the f.c.c. TaN,_, is desired without presence of the insulation Ta3NS phase,
+
which is difficult if TaCl, NH3 as starting substances are used. TaNl --x films
(deposited by both by PVD and CVD) have a resistivity of 130-250pRcm [142].
Niobium nitride films show good performance for vacuum microelectronic devices
for displays because of a high work function and resistance to sputtering [143].
7.7.3 Diffusion Layers

Diffusion layers of transition metal nitrides are (similar to case hardened steels,
which should not be discussed here) prepared by heating metals and alloys in
nitrogen or ammonia, occasionally diluted with an inert gas. Mainly titanium and
its alloys (e.g. TiA16V4) are of interest in order to circumvent the strong cladding
behavior of these materials [26]. Also plasma nitriding techniques are used for the
preparation of surface layers on the order of several microns thickness. The
microstructure of these diffusion layers depend strongly of the alloy composition,
the temperature and the nitrogen pressure [ 1441.
A technique which is related to this surface refining method, is the surface laser
heating of alloys. In this method a strong laser is used to melt titanium alloys at
the surface with simultaneous supply of nitrogen via a jet or by operating in nitrogen
atmosphere [27,145]. By such a method nitrides (though high-melting) form by
surface melting together with an intermediate diffusion layer form in the solid
state. Nitride layer thicknesses of the order of millimeters can be prepared. These
layers are, for example, applied in large turbine blades in order to prepare a sur-
face withstanding the intensive cavitation of water droplets reaching supersonic
velocity.
One of the most interesting developments in the field of hardmetals and cermets
for cutting tools is the creation of a graded surface which makes use of the favorable
properties of functional-gradient materials by dissipating the thermal and mechan-
ical load onto a larger part of the cutting insert in order to increase the lifetime or to
withstand extreme conditions of cutting [146]. Figure 31 shows surface-near
microstructures of such functional-gradient cermets composed of (Ti, W)(C, N)
with additions of TaC and NbC and a Co binder metal. Upon sintering, the
hexagonal phase WC forms (which can also be present in the starting formulation).
Depending on the composition and the sintering cycle a smooth variation of the
Figure 31. Microstructures of functional-gradient cemented carbonitrides based on the system

(Ti, W)(C, N F C o which are surface modified by applying a reactive gas phase during sintering in
a one step fabrication process [146].
phase content as a function of the distance from the surface can be seen. This
morphology is caused by the action of reactive gases such as nitrogen and carbon
monoxide introduced at specific points of the sintering cycle. The action of reactive
gases can be such modified that different gradients can be fabricated and even layers
of different chemical and phase composition can be prepared together with gradi-
ents. This functional-gradient hardmetal has a Ti(C, N) top layer, an intermediate
WC-Co layer and a gradient in which the (Ti, W)(C, N) phase is enriched at the
outer side and continuously reaches bulk composition. In view of the rather high
production costs of coated materials (the coating process is about 15-20% of the
total production costs of a cutting insert) such modified cemented carbonitrides
could be lower the production costs significantly because the diffusion layers and
gradients are established during the sintering cycle and (provided similar perfor-
mance can be obtained) an additional coating cycle is then unnecessary.
Acknowledgments
The author would like to thank Dr. K. Dreyer, WIDIA GesmbH, Essen (Germany) for
support with figures and microstructures of tools parts. The help of Mr. W. Prohaska
with the preparation of the phase diagram drawings is gratefully acknowledged.
References
1. P. Ettmayer and W. Lengauer, Nitrides in Ullmann’s Encyclopedia of Industrial Chemistry,
Verlag Chemie, Weinheim, 1991, A17, p. 341.
2. E. Parthe and K. Yvon, Acta Crystallogr. B, 1970, 26, 153.
3. K. Yvon and E. Parthe, Acta Crystallogr. B, 1970, 26, 149.
4. T. Epicier, in The Physics and Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.),
Kluwer, Dordrecht, 1990, p. 215.
5 . C. H. de Novion, B. Beuneu, T. Priem, N. Lorenzelli, and A. Finel, in The Physics and
Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.), Kluwer, Dordrecht, 1990, p. 329.
6. F. Kubel, W. Lengauer, K. Yvon, A. Junod, and K. Knorr, Phys. Rev. B, 1988,38, 12908.
7. W. Lengauer, M. Bohn, B. Wollein, and K. Lisak, Acta Muter. 1999, in press.
8. H. Wiesenberger, W. Lengauer, and P. Ettmayer, Acta Muter. 1998,46,651.
9. J. D. Browne, P. R. Lidell, R. Street, and T. Mills, Phys. Stat. Solid, 1970, Al, 715.
10. T. B. Massalski (Ed.), Binary Alloy Phase Diagrams, 2, ASM International, Metals Park, OH,
1990.
11. K. Schwarz, CRC Crit. Rev. Solid State Muter. Sci. 1987, 13, 211.
12. A. Neckel, in The Physics and Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.),
Kluwer, Dordrecht, 1990, p. 485.
13. P. Blaha and K. Schwarz, Int. J. Quantum Chem. 1983,23, 1535.
14. A. N. Christensen and S. Fregerslev, Acta Chem. Scand. A , 1978, A31, 861.
15. R. Kieffer, H. Nowotny, P. Ettmayer, and M. Freudhofmeier, Monatsh. Chem. 1970,101,65.
16. H. Pastor, Muter. Sci. Eng. 1988, A105/106, 401.
17. J.-E. Sundgren and L. Hultman in: Materials and Processes f o r Surface and Interface Engineer-
ing, Y .Pauleau (Ed.), Kluwer, Dordrecht, 1995, p. 453.
References 249
18. W. Schintlmeister and 0. Pacher, J . Vac. Sci. Technol. 1975, 12/4, 743.
19. S. Abisset and F. Maury, Surf. Coat. Technol. 1998, 108-109, 200.
20. W. D. Sproul, J . Vac. Sci. Technol. A , 1994, 12, 1594.
21. J. Smolik and K. Zdunek, Vacuum, 1998, 51/3, 441.
22. V. Craciun, D. Craciun, C. Ghica, L. Trupina, C. Flueraru, and N. Nastase, Appl. Surf: Sci.
1999, 138-139, 593.
23. K. Obata, K. Sugioka, H. Takai, and K. Midorikawa, Applied Surf. Sci. 1999, 138-139,
335.
24. A. J. Perry and J. R. Treglio, J . Vac. Sci. Technol. A , 1995, 13/3, 1067.
25. S. Xiao, C. P. Lungu, and 0. Takai, Thin Solid Films, 1998, 334, 173.
26. J. Magnan, G. C. Weatherly, and M.-C. Cheynet, Metall. Muter. Trans., A , 1999, 30A, 19.
27. C. Hu, H. Xin, L. M. Watson, and T. N. Baker, Acta Muter. 1997, 45/10, 431 1.
28. W. Lengauer, Talanta, 1991, 38, 659.
29. R. Tiubler, S. Binder, M. Groschner, W. Lengauer, and P. Ettmayer, Mikrochim. Acta I, 1992,
107, 337.
30. G. F. Bastin and H. J. M. Heijligers, J . Microsc. Spectrosc. Electron. 1986, 11, 215.
31. G. F. Bastin and H. J. M. Heijligers, Quantitative Electron Probe Microanalysis of Nitrogen,
University of Technology, Eindhoven, 1988.
32. W. Lengauer, J. Bauer, M. Bohn, H. Wiesenberger, and P. Ettmayer, Mikrochim. Acta, 1997,
126, 279.
33. K. Frisk, A Thermochemical Evaluation of the Cr-N, Fe-N, M e N and Cr-Mo-N Systems,
Report TRITA-MAC-0393, Materials Research Center, Royal Institute of Technology,
Stockholm, 1989.
34. V. N. Lipatnikov, W. Lengauer, P. Ettmayer, E. Keil, G. Groboth, and E. Kny, J . Alloys
Comp. 1997,261, 192.
35. R. Ward in M T P International Review of Science, Inorganic Chemistry Series One, D. W. A.
Sharp (Ed.), Butterworths, London 1972, p. 5.
36. H. Holleck, Binare und ternare Carbid- und Nitridsysteme der Ubergangsmetalle, Borntrager,
Berlin 1984.
37. H. Holleck, J . Vac. Sci. Technol. A , 1986, A6, 2661.
38. W. Lengauer, Acta Metall. Muter. 1991, 39, 2985.
39. K. Zheng and R. Schmid-Fetzer, Z . Metallk. 1996, 87, 540.
40. R. Juza, A. Rabenau, and I. Nitschke, Z. Anorg. Allgem. Chem. 1964,332, 1.
41. E. Rudy, Metall. Trans. 1970, 1, 1249.
42. 0. N. Carlsson, J. F. Smith, and R. H. Nafziger, Metall. Trans. A , 1986, A17, 1647.
43. W. Lengauer and P. Ettmayer, J . Less-Common Met. 1985, 109, 351.
44. W. Lengauer, J . Phys. Chem. Solids, 1988,49, 59.
45. N. Kieda, N. Mizutani, and M. Kato, J . Less-Common. Met. 1988, 144, 293.
46. C. Teichmann, W. Lengauer, P. Ettmayer, J. Bauer, and M. Bohn, Metall. Muter. Trans A ,
1997,28A, 837.
47. G. Brauer and W. Kern, Z. Anorg. Allgem. Chem. 1983,507, 127.
48. G. Brauer and W. Kern, Z. Anorg. Allgem. Chem. 1984,512, 7.
49. R. Berger, W. Lengauer, and P. Ettmayer, J . Alloys Comp. 1997, 259, L9.
50. M. Joguet, W. Lengauer, M. Bohn, and J. Bauer, J . Alloys Comp. 1998,269,233.
51. J. Gatterer, G. Dufek, P. Ettmayer, and R. Kieffer, Monatsh. Chem. 1975, 106, 1137.
52. G. Brauer, E. Mohr, A. Neuhaus, and A. Skokan, Monatsh. Chem. 1972, 103,794.
53. K. Frisk, J. Alloys. Comp. 1996, 278, 216.
54. W. Mayr, W. Lengauer, D. Rafaja, J. Bauer, and M. Bohn, J . Phase Equil. 1999, 20, 35.
55. P. Ettmayer and A. Vendl, in Science of Hard Materials, R. K. Viswanadham, D. J. Rowcliffe,
and J. Gurland (Eds.), Plenum, New York, 1983, p. 47.
56. A. Bezinge, K. Yvon, J. Muller, W. Lengauer, and P. Ettmayer, Solidstate Commun. 1987,63,
141.
57. S. L. Roberson, D. Finello, and R. F. Davis, Muter. Sci. Eng. A , 1998, 248, 198.
58. V. P. Anitha, A. Bhattacharya, N. G. Patil, and S. Major, Thin Solid Films, 1993, 236, 306.
59. W. Mayr, W. Lengauer, V. Buscaglia, J. Bauer, M. Bohn, and M. Fialin, J. Alloys Comp. 1997,
262-263, 521.
- A. Martinelli, R. Musenich, W. Mayr, and W. Lengauer, J . Alloys Comp. 1999,

60. V. Buscaglia.
283, 24.
61. B. Wollein, V. Buscaglia,
- W. Lengauer,
- R. Musenich, P. Ettmayer, and M. Bohn, ECASZA 99,
Sevilla, Spain, 1999.
62. P. Ettmayer, G. Vinek, and H. Rassaerts, Monatsh. Chem. 1966, 97,1248.
63. S. Binder, W. Lengauer, P. Ettmayer, J. Bauer, J. Debuigne, and M. Bohn, J . Alloys Comp.
1995, 217, 128.
64. H.-J. Chung, J.-H. Shim, and D. N. Lee, J. Alloys Comp. 1999, 282, 142.
65. S. Jonsson, Z . Metallkd. 1996,87, 788.
66. L. Chen. G. Groboth, E. Kny, W. Lengauer, and P. Ettmayer, Proc. I l l hPlansee Seminar '97
1997,2,498, G. Kneringer et al. (Eds.), Reutte (A).
67. E. Rudy, J. Less-Comm. Met. 1973, 33, 43.
68. A. Doi, T. Nomura, M.-A. Tobioka, K. Takahashi, and A. Hara, High Temp.-High Press,
1986, 18,443.
69. P. Ettmayer, Ann. Rev. Muter. Sci. 1989, 19, 145.
70. P. Ettmayer, R. Kieffer, and F. Hattinger, Metall. (Berlin), 1974, 28, 1151.
71. A. J. Perry, M. Georgson, and W. D. Sproul, Thin Solid Films, 1988, 157, 255.
72. W. S . Williams, in The Physics and Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.),
Kluwer, Dordrecht 1990, p. 625.
73. W. Lengauer, S. Binder, K. Aigner, P. Ettmayer, A. Guillou, J. Debuigne, and G. Groboth,
J . Alloys Comp. 1995,217, 137.
74. W. Lengauer, J . Alloys Comp. 1992, 186,293.
75. R. E. Taylor and J. Morreale, J . Am. Ceram. Soc. 1964,47, 69.
76. W. Lengauer, Surf. Znterf. Anal. 1990, 15, 377.
77. R. R. Reeber and K. Wang, J . Am. Ceram. SOC.1999,82, 129.
78. W. Lengauer, P. Ettmayer, and K. Yvon, J. Less-Common Met. 1991, 168, L7.
79. W. Lengauer and P. Ettmayer, Monatsh. Chem. 1986, 117, 275.
80. K. Aigner, W. Lengauer, D. Rafaja, and P. Ettmayer, J . Alloys Comp. 1994, 215, 121.
81. W. Lengauer, H. Wiesenberger, M. Joguet, D. Rafaja, and P. Ettmayer, in Transition Metal
Carbides and Nitrides: Preparation, Properties and Reactivity, S . T. Oyama (Ed.), Blackie,
London, 1996, p. 9 1.
82. W. Lengauer, D. Rafaja, R. Taubler, C. Kral, and P. Ettmayer, Acta Metall. Muter. 1993,41,
3505.
83. W. Lengauer, J . Alloys Comp. 1995, 229, 80.
84. D. Rafaja and W. Lengauer, Model. Simul. Muter. Sci. Eng. 1998, 6, 141.
85. T. C. Wallace Sr., D. P. Butt, in Transition Metal Carbides and Nitrides: Preparation, Proper-
ties and Reactivity, s. T. Oyama (Ed.), Blackie, London, 1996, p. 53.
86. D. Rafaja, H. Wiesenberger, W. Lengauer, and M. Joguet, Metall. Muter. Trans A, 1998,29A,
439.
87. D. Rafaja and W. Lengauer, Acta Muter. 1998, 46, 3477.
88. W. Lengauer, D. Rafaja, G. Zehetner, and P. Ettmayer, Acta Muter. 1996, 44, 3331.
89. F. W. Wood and 0. G. Paasche, Thin Solid Films, 1970,40, 131.
90. H. Matzke, in The Physics and Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.),
Kluwer, Dordrecht 1990, p. 357.
91. E. Tiirok, A. J. Perry, L. Chollet, and W. D. Sproul, Thin Solid Films, 1987, 153, 37.
92. C. Kral, W. Lengauer, D. Rafaja, and P. Ettmayer, J . Alloys Comp. 1998, 265,215.
93. H. S. Kim, G. Bugli, and G. Djega-Mariadassou, J. Solid. St. Chem. 1999, 142, 100.
94. F. Li, K. Fatrdla, and W. Lengauer, to be published 1999.
95. P. R. Stoddart, J. D. Comins, and A. G. Every, Physica B, 1996, 219-220, 717.
96. A. G. Every, W. Pang, J. D. Comins, and P. R. Stoddart, Ultrasonics 1998, 36, 223.
97. J.-L. Chermant, P. Delavignette, and A. Deschanvres, J. Less-Common Met. 1970, 21, 89.
98. F. W. Vahldiek and S. A. Mersol, J . Less-Common Met. 1977, 55, 265.
99. Y . Kumashiro, E. Sakuma, Y . Kimura, H. Ihara, and S. Misawa, J . Cryst. Growth, 1981, 52,
597.
100. D. W. Lee and J. S . Haggerty, J. Am. Ceram. SOC.1969,52, 641.
101. C. Politis and G. Rejman, Kernforschungszentrum Karlsruhe, KfK Report Ext6178-I, 1978.
References 25 1
102. L. Akesson, Thesis, Royal Inst. Technol. Stockholm, Sweden 1982.

103. K. Dreyer, P. Drohmann, H. Holzhauer, M. Miiller, H. van den Berg, and W. Weith, Proc.
PM World Congress, Granada, 1998.
104. P. Wally, P. Ettmayer, and W. Lengauer, J . Alloys Comp. 1995, 228, 96.
105. P. Ettmayer, H. Kolaska, W. Lengauer, and K. Dreyer, Int. J . Refract. Met. Hard. Mat. 1995,
13, 343.
106. T. Viatte, S. Bolognini, D. Mari, and W. Benoit, Proc. PM World Congress, Granada
1998.
107. L.-M. Berger, W. Hermel, P. Vuoristo, T. Mantyla, W. Lengauer, and P. Ettmayer, in Thermal
Spray: Practical Solutions for Engineer. Problems, C. C. Berndt (Ed.), ASM International,
Materials Park, OH, 1996, p. 89.
108. J. Deng, M. Braun, and I. Gudowska, J . Vac. Sci. Technol. A , 1994, 12/3, 733.
109. Y. L. Su and W. H. Kao, Wear, 1998, 223, 1 19.
110. T. Bell, Met. Muter. 1991, 7, 478.
11 1. U. Kopacz and R. Riedl, Z. Metallkd. 1992, 83, 492.
112. G. Reiners. U. Beck, and H. A. Jehn, Thin Solid Films, 1994, 253, 33.
113. G. Reiners, H. Hantsche, H. A. Jehn, U. Kopacs, and A. Rack, S u r - Cout. Technol. 1992,54/
55, 273.
114. K. E. Anderssson, M. Veszelei, and A. Roos, Solar Energ. Muter. Solur Cells, 1994, 32, 199.
115. 0. Knotek, F. Loffler, and B. Bosserhoff, SUI$ Coat. Technol. 1993,62, 630.
116. S. S. Kim, J. G. Han, and S. Y. Lee, Thin Solid Films, 1998, 334, 133.
117. K. Volz, M. Kiuchi, and W. Ensinger, Surf. Coat. Technol. 1998, 108-109, 303.
118. P. Hones, R. SanjinCs, and F. Levy, Thin Solid Films, 1998, 332, 240.
119. W.-D. Miinz, J. Vuc. Sci. Technol. 1986, A4, 2717.
120. 0. Knotek, W. D. Miinz, and T. Leydendecker, J . Vac. Sci. Technol. 1987, AS, 2173.
121. H. Randhawa, P. C. Johnson, and R. Cunningham, J. Vac. Sci. Technol. A 1988, 6/3,
2136.
122. C. Jarms, H.-R. Stock, and P. Mayr, Surf. Coat. Technol. 1998, 108-109, 206.
123. B.-J. Kim, Y.-C. Kim, J.-W. Nah, and J.-J. Lee, J . Vuc. Sci. Technol. A , 1999, 17/1, 133.
124. 0. Knotek and A. Barimani, Thin Solid Films, 1989, 174, 51.
125. R. Kieffer, H. Nowotny, P. Ettmayer, and G. Dufek, Metall. (Berlin) 1972, 26, 701.
126. L. A. Donohue, I. J. Smith, W.-D. Miinz, I. Petrov, and J. E. Greene, Surf: Cout. Technol.,
1997,9495, 226.
127. M. Shinn, L. Hultman, and S. A. Barnett, J . Muter. Res. 1992, 7/4, 901.
128. M. Larsson, P. Hollman, P. Hedenqvist, S. Hogmark, U. Wahlstrom, and L. Hultman, Surf.
Cout. Technol. 1996,8647, 351.
129. S. Veprek and S. Reiprich, Thin Solid Films, 1995, 268, 64.
130. S. Veprek, M. Haussmann, S. Reiprich, L. Shizhi, and J. Dian, Surf. Coat. Technol. 1996,86-
87, 394.
131. S. Veprek, P. Neslidek, A. Niederhofer, F. Glatz, M. Jilek, and M. &ma, Surf Coat Technol.
1998, 108-109, 138.
132. X. Chu and S. A. Barnett, J. Appl. Phys. 1995,77/9, 4403.
133. P. Losbichler, C. Mitterer, P. N. Gibson, W. Gissler, F. Hofer, and P. Warbichler, Sure Coat.
Technol. 1997, 94-95, 297.
134. C. Mitterer, P. Losbichler, F. Hofer, P. Warbichler, P. N. Gibson, and W. Gissler, Vacuum,
1998, SO/-, 313.
135. I. Petrov, E. Mojab, F. Adibi, J. E. Greene, L. Hultman, and J.-E. Sungren, J . Vuc. Sci. Tech-
no/. 1993, A l l , 11.
136. M. H. Tsai, S. C. Sun, C. E. Tsai, S. H. Chudng, and H. T. Chiu, J . Appl. Phys. 199679,6932.
137. K. Radhakrishnan, N. G. Ing, and R. Gopalakrishnan, Muter. Sci. Eng. B, 1999, 57, 224.
138. X. Chen, H. L. Frisch, A. E. Kaloyeros, B. Arkles, and J. Sullivan, J . Vac. Sci. Technol. 1999,
B17, 182.
139. R. Fix, R. Gordon, and D. Hoffman, Chem. Muter. 1993,5, 614.
140. H. Chiu and W. Chang, J. Muter. Sci. Lett. 1992, 11, 92.
141. K. Hieber, Thin Solid Films, 1974, 24, 157.
142. X. Sun, E. Kolawa, J. Chen, J. Reid, and M.-A. Nicolet, Thin Solid Films, 1993, 236, 347.
143. Y. Gotoh, M. Nagao, T. Ura, H. Tsuji, and J. Ishikawa, Nucl. Znstr. Meth. Phys. Res. B, 1999,
B148, 925.
144. J. K. Gregory, W. Lengauer, and W. Mayr, in Surface Performance of Titanium, J. K.
Gregory, H. J. Rack, and D. Eylon (Eds.), TMS, Warrendale, PA, 1996, p. 65.
145. H. Xin, S. Mridha, and T. N. Baker, J . Muter. Sci. 1996, 31, 22.
146. W. Lengauer, L. Chen, J. Garcia, V. Ucakar, K. Dreyer, D. Kassel, and H. W. Daub,
Proc. PM2tec 99, Znt. Congress on Powder Metallurgy and Particulate Materials, Vancouver
1999.
8 New Superhard Materials: Carbon and

Silicon Nitrides
J. E. Lowther
8.1 Introduction
Novel procedures continuously being introduced into the synthesis of new
materials have lead to the speculation that it could be possibie to produce a
material with hardness exceeding that of diamond, the hardest material presently
known to man. As the chemistry of diamond is based upon understanding the
structural bonding of carbon, a modification of known carbon structures through
application of high pressure or temperature would seem a promising route. Indeed
there has been much speculation about the hardness of carbon found in fullerene
complexes, yet production of ‘post-diamond’ phases has not met with any success,
at least through known synthesis procedures. Even so the commercial synthesis of
diamond (and more recently cubic boron-nitride) has lead to much speculation in
the field and with it enormous excitement as there is potential for new superhard
materials.
Properties of diamond, its hardness, unique electrical and thermal properties,
arise from a three-dimensional structure of very strong covalent bonds. Similar
properties are also associated with cubic boron-nitride. That other similar highly
covalent three-dimensional structures may exhibit similar properties to that of
diamond is highly likely. A possible material that first stimulated interest in such
structures was PSi3N4 [I], a hard material with quite interesting properties. A
possible replacement of one (or several) of the Si atoms with C would tentatively
suggest the synthesis of a material that is harder than PSi3N4 and even compete
with diamond. There is therefore a rich possibility of structures in the tertiary
Si,C,,N, system that afford unique potential for synthesis. But problems are
always encountered when dealing with nitrogen, the N2 dimer has a very strong
covalent bond and does not react with carbon readily when it competes with
either carbon or silicon in forming stable Si,C,,Nz structures. Graphite is very
likely more stable than diamond, and in a tertiary material like Si,C,,N, the forma-
tion of a SirC, structures is also possible. Quite clearly sophisticated techniques are
needed for a successful synthesis [2,3].
Given the cost and complexity of a commercial synthesis procedure, even when
identified, an unknown synthesis route requires considerable investment. For this
reason computer modeling procedures have been introduced that attempt to
predict new structures and examine phase transitions connecting them. Such
procedures are continuously being applied [4] and, bearing in mind that funda-
mentals of the modeling procedures are also being debated, indications in the
last few years have shown that reliable calculations of material properties can be
254 8 New Superhard Materials: Carbon and Silicon Nitrides
achieved at little expense. Liu and Cohen [5] applied computer modeling to
investigate a hypothetical phase of PC3N4 and suggested that such a material
would have a large binding energy and may be metastable under ambient condi-
tions. More importantly, the calculations implied that this material would have
a hardness comparable with (or even greater than) diamond. The computational
technique used had already been applied to diamond with considerable success
[6] and its reliability simulated enormous interest and effort in investigating
other materials.
This chapter examines some potential structures that have been proposed for
different phases of carbon-nitride and also the tertiary material silicoti-carbon-
nitride. First we briefly examine some basic modeling procedures that have been
employed to date when investigating these materials and then discuss properties
of the predicted structures.
8.2 Modeling Procedures

Several approaches are essentially used in the computational modeling of solids.
These are classified as being either semi-empirical or ab initio. The term semi-
empirical is appropriate to those systems where significant knowledge exists about
the system under specific conditions of temperature or pressure. Subsequently the
system is modeled under other conditions. Semi-empirical modeling is really an
interpolation from one system to other, probably unknown or experimentally
inaccessible, conditions. On the contrary the term ab initio is appropriate for systems
where little or no knowledge exists about the system. The ab initio procedure is
applicable for predictive modeling of new phases of materials and is far more
reliable than semi-empirical methods. Molecular dynamics procedures can be
implemented at both levels of the formalism. Such techniques endeavor to examine
the trajectories of atoms under different temperatures and pressures. Presently there
are many excellent algorithms readily available to achieve this task when the atomic
interactions can be described in analytic form. Unfortunately for covalent solids,
which is likely to be the character of superhard materials, evaluation of the
atomic interactions is complex and computationally very demanding.
8.2.1 Semi-empirical Approaches

This category of approaches assumes that interactions in the solid can be
represented by two or three-body potentials or force-fields between atoms. The
interaction is usually obtained by fitting models to results on molecular systems.
This is a convenient approach for ionic materials where long range Coulomb
forces are strong and for a typical material like SiOz at least eleven parameters
are needed to characterize the system. Such a choice of potential, but in the form
+
of V(r) = D{exp[-a(r - req)]- 1}2 ( k / 2 ) ( 8- Oe,)2 with empirical parameters
8.2 Modeling Procedures 255
D,a, re,, k, and Be, fitted to small molecular systems was applied to structures of
carbon-nitride by Guo and Goodard [7]. However, these results were in conflict
with many other calculations for carbon-nitrides highlighting that use of such a
simple interatomic interaction has danger when the potential is transferred to
materials with a far higher density. In the case of highly covalent materials like
diamond and (most likely) related superhard materials the potential is likely to be
far more complicated than used by Guo and Goodard. In order to study fracture
and related degradation processes in silicon nitride a similar effective potential
has been obtained [8] representing an average of the interactions. This allows for
simulations of up to lo6 effective atoms. However, the drawback of such an
approach is that it apparently disregards the possibility of atomic segregation.
One of the most promising analytic potentials that has been introduced in this
regard is the Tersoff potential [9] that has the form V ( r ) = f c ( r ) ( f R ( r )+ / f A ( v ) )
where f ( r ) and b are complicated functions containing angular as well as a
radial (v) dependences. For a typical binary solid some 24 empirical parameters
are now needed to adequately describe the potential, with considerable success
being reported for C and Si structures. Suitable parameters for the Tersoff
potential have now been obtained for cubic BN [lo] and recently for PSi3N4
[l 11 but earlier results [12] had indicated that parameters in the Tersoff potential
may not be unique. Further work is now being undertaken to try and find a
transferable Tersoff potential for the Si,C,,N, system, but in all likelihood such
a potential may have a limited range of transferability amongst closely related
structures.
Quite clearly the semi-empirical procedure is very dependent upon the choice of
parameters, although it has been possible to obtain sensible parameters from fixed
systems. The shortcoming with the approach is that the parameters may not be
transferable to other (especially unknown) situations. At the same time once a reli-
able potential is obtained there are many excellent molecular dynamics algorithms
available to compute the subsequent motion of the system quite efficiently and with
low computer cost.
8.2.2 Tight-binding Schemes

Although various tight-binding schemes have proven successful in studies of carbon
and silicon surfaces, the problem of obtaining a successful interaction with N has
not been satisfactorily solved. And even if such approaches such as the virtual
crystal approximation could be employed it is doubtful whether such a potential
could be truly transferable, a similar problem as with the semi-empirical
approaches. Recently an approach using density functional concepts has been intro-
duced [ 131 and applied to high density structures of carbon-nitride [ 141 with various
stoichiometries. The tight-binding interactions are only evaluated once for the
system and stored numerically making the method rather similar to conventional
tight-binding methods yet avoiding an interpolative scheme for their evaluation at
different distances. Even though the method still retains some reliance on molecular
rather than solid state concepts for computing electronic interactions it has proven
quite useful in examining various stoichiometric forms of carbon-nitride.
8.2.3 Ab initio Pseudopotential Approach

The pseudopotential approach employs a quantum mechanical description for the
electronic interactions as specified by either through density functional or Hartree
Fock techniques [15]. In the case of materials modeling the density functional
approach appears to be extremely accurate in predicting ground state structures
of systems [16] and it is this approach that presently appears to be gaining
ground in the modeling of superhard materials. Even within the density functional
formalism two approaches, the local density approximation (LDA) [17] or the
generalized gradient approximation (GGA) [ 181, are presently receiving attention,
although in the case of carbon based systems the LDA approach seems more
promising for structural properties. Once the electronic interaction is described,
forces acting on the atoms can be calculated and in turn their positions established
using well developed techniques [19]. Overall the procedure does not require any
input apart from elementary chemical details of the interacting atoms and the
crystalline structure of the system. The operational shortcoming with this
approach is that it is computationally expensive requiring very long computer
runs as well a being memory intensive in that large arrays are necessary to describe
the electronic degrees of freedom. Without doubt however this procedure is now
regarded as state-of-the-art and is most reliable for predictive modeling. The
quantum mechanical basis of the approach very often consist of simple plane
waves which together with fast Fourier transform techniques and a suitable
pseudopotential give accurate results. One of the more successful pseudopotentials
is a soft pseudopotential that attempts to limit the number of planes waves needed
in the basis, with the pseudopotential generated from the Troullier-Martins [20]
method.
8.2.4 Transition Pressures and Relative Stability

The equation of state of the material is usually represented by the Morgeneau [21] or
Birch [22] form and the material modeled through a unit cell structure in which the
individual atoms have fractional (Wykoff) parameters optimized as the cell changes
in volume. From the fitted equation of state we then find the pressure on the system,
P = - d E / d V . However this is not the full story. Ideally there are two important
factors specifying a phase transformation between two phases. First two (or
more) phases can coexist when the separate phases have the same energy. Accord-
ingly an investigation of the full thermodynamic energy of the system against pres-
sure would reveal the coexistence of distinct phases. But energy is also needed when
compressing one phase to another: the energy required to surmount the energy bar-
rier between the phases. The latter value needs to know specific details of phases
involved. Thus there are two ways of deducing the pressure for a phase transforma-
tion. Either from the slope at the common energy of E / V ,or by the intersection of
E/P.The former is specific in that it refers to distinct phases whereas the latter
makes no assumption of the path followed.
For example, if we look at the transformation from graphite into diamond we
calculate from E/V that a pressure of P = 80 GPa is needed to transform graphite
into diamond by overcoming the energy barrier. The commercial synthesis process
involves pressures far less than this, although high pressure treatment of graphite (or
CG0type material) leading to diamond is consistent with application of such large
pressures. However, the calculated relative energies of graphite and diamond are
very close to each other. This means that the pressure for the transformation
found from the E/P intersection is calculated to be very small, and this would be
the situation if the graphite to diamond transformation itself was continuous. Of
course the transformation process of graphite into diamond itself is not continuous
but rather involves changes from graphite into other (rhombohedral) graphitic
forms through to hexagonal diamond and then finally diamond. Chemical
influences (such as metal flux) as well as temperature factors are important in the
synthesis. Collectively these latter features are influencing the relative energetics
of the various phases.
Thus we must conclude that whereas the calculated pressure from E/P has some
guiding significance, the pressure obtained from E / V identifies a chosen synthesis
route. The pressure calculated in this way will be an upper value needed for the
transformation between the two phases. Accordingly, an in depth modeling of the
synthesis procedure needs an investigation of the stability of the intermediate
phases if they are known. Yet the important physical properties of the system are
essentially contained in relative values of the energies of the two phases and how
the energy changes under compression.
8.3 Carbon Nitride

The difficulty of producing large enough samples of C3N4 for a precise structural
characterization is proving problematic. It has been observed to exist in apparent
amorphous or disordered phases [23-251 and there are several reports of crystalline
phases [26,27] being dispersed in the amorphous matrix. Laser ablation [28,29]
appears to be a promising technique for the synthesis of crystalline carbon nitride
and vapor deposition has recently proven to be promising [30,31] as has ion-
beam deposition [32,33].
Graphite is an important starting point for the synthesis of diamond. It exists in
several forms and as such through applied pressure leads to other structures such as
hexagonal diamond and then finally diamond. Graphite structures for carbon
nitride have been considered theoretically as possible forms of the material
[6,34,35]. These all consist of C3N4layers in a hexagonal planar lattice in which
a C atom is missing. This form is not as hard as diamond but the location of the
258 8 New Superhard Materials: Carbon and Silicon NitridcJ
Table 1. Structural details of some predicted superhard crystalline phases of C3N4.
Phase Z Unit cell (nm) Wykoff parameters Bulk modulus (GPa)
PC3N4 (P6,lrn) a = 0.641-0.644, C: (0.178, 0.772, 0.250)

2 425 [37], 437 [5], 451 [35]
c = 0.24040.2467 N: (0.331, 0.033, 0.250)
N: (0.333, 0.667, 0.250)
Cubic-C3N4 (Pz3rn) 1 a = 0.342-0.344 C: (0.500, 0.500, 0.000) 425 [37], 430 [26], 448 [35],
(zinc-blende) N: (0.254, 0.254, 0.254)
Cubic C3N4 (1534 4 a = 0.540 C: (0.875, 0.000, 0.250) 480 [26], 496 [35],
N: (0.281, 0.281, 0.254)
a G N 4 (P31c) 4 a =0.6466 C: (0.773, 0.178, 0.250) 425 [35]
c = 0.4710 C: (0.227, 0.821, 0.750)
N: (0.333, 0.667, 0.750)
N: (0.033, 0.331, 0.250)
N: (0.033, 0.331, 0.250)
N: (0.967, 0.670, 0.750)
missing C atom in the C3N4layer suggests there could be several ways of stacking in
much the same way as, for example, graphitic BN [36]. A rhombohedra1 [37] form of
the material in graphitic form is predicted to be stable at a material density lower
than the 0-phase and more recently other graphitic forms with still lower density
have been suggested as being possible precursors to the formation of the superhard
phase [34,35]. High energy amorphous structures have also been investigated [14,38]
and these have shown competition between the C-C, C-N, and N-N bonds in
forming such a structure. Possible crystalline structures with stoichiometries different
from C3N4[6,39] have also been examined.
8.3.1 Crystalline Structures

To date, several forms of C3N4have been suggested for crystalline structures with
superhard forms being BC3N4[37,40,41] (hexagonal structure and space group P3),
cubic or defect zinc-blende C3N4 [37] (P43rn) and another cubic form [35] (1434
suggested to be the hardest as measured by the calculated bulk modulus. In each
of these structures, carbon atoms are in four-fold coordination by nitrogens, the
nitrogens in three-fold coordination by carbons. All structures are predicted to be
stable with unit cells given in Table 1.
The harder structures are shown in Fig. 1, as can be seen each C atom is
surrounded in four-fold coordination by N atoms. The C-N bond length is small:
0.145nm in 0C3N4and 0.148nm in cubic (zinc-blende) C3N4.
Some idea of the nature of the bonding in phases of C3N4can be obtained from a
mapping of the electronic charge density.
In Fig. 2 we show such a plot for the 0 structure, as expected it is quite directional,
yet there are quite unique features about N associated with the N(2p) orbits directed
away from the C-N bond. It is possible that such a feature could lead to sites for
passivation (i.e. hydrogen association) leading to sources of brittleness in such a
material.
Figure :1 Some superhard crystalline phases of C3N4:(a) P-phase (P63/m),(b) cubic (P43m),(c)
cubic (1434 structures. Black spheres represent C atoms.
8.3.2 Graphitic Structures

Graphitic phases consist of hexagonal planes of bonded C-N atoms with a missing
C on one of the conventional planar hexagonal lattice sites. Various sites are
possible for the ordering of the graphitic planes depending on the relative location
of the missing C atom from one plane to another [42]. First is the case where empty
C sites are aligned relative to each other along the hexagonal c-axis. This structure
emphasizes C-C and N-N bonding between the adjacent layers so that all bonds
along the c axis are saturated though there is a rather large region where bonding
is probably weak. Such a structure was the first type of graphitic structure consid-
ered for C3N4and named p-C3N4 by Ortega and Sankey [34]. We shall refer to the
structure as p-C3N4(even-phase). If we consider that the planes may move relative
to one another, another structure can arise with mixed C-N bonding between the
Figure 2. Charge density of P(P6,Irn) C3N4plotted perpendicular to the c-axis.
layers. This structure we shall name p-C3N4(odd-phase). Probably there are many
possibilities between. There is also the possibility that one of the planes may have no
missing C atoms with the adjacent plane having two missing C atoms: a C2N4-C4N4
structure. The three possibilities are shown in Fig. 3.
The other type of graphitic structure considered recently by Teter and Helmley [35]
also contains 14 atoms in the unit cell but is somewhat different to the p-C3N4 struc-
tures as here two possible C-N bond lengths are in the hexagonal plane. This structure
is shown in Fig. 4 and emphasizes C-N bonding between layers.
As the stability of the graphitic phases probably relates to the intralayer structure
than the interlayer spacing we investigate the layer charge density for the phases. In
Fig. 5 we show the charge density of the P63/m phase; rather surprisingly most of
the charge on N is directed away from the C-N bond toward the vacant site, as is the
case for some superhard structures.
8.3 Carbon Nitride 26 1
Figure 3. p-structures of hexagonal graphite C3N4: (a) odd-phase structure, (b) the even-phase
structure, and (c) the CZN4-C4N4planar structure. Carbon atoms are shown in black.
The low energy P63/m graphite phase appears to change to one of the higher
energy p-C3N4 graphite phases following compression well before the superhard
(3-phase is reached. However this p-C3N4 graphite phase seems slightly softer
than the P63/m phase and as such requires less pressure to surmount the energy
barrier that leads to the p-phase. If a graphitic phase of C3N4 can be established
it would therefore appear quite fruitful to use pressure (especially uniaxial) tech-
niques to change between the graphitic phases and possibly attain a superhard
phase. Of the graphite structures stacking involving C4N4-C2N4 sheets was
unlikely but each of the p-graphite phases were found quite stable and with relative
energies lying very close to each other. Forces between adjacent planes in the graphi-
tic structures are very small and so graphitic carbon nitride is extremely soft along
the c-axis. This weak interaction probably leads to dispersed layers of graphitic
sheets in the material as recently observed using high resolution spectroscopy of
carbon nitride thin films [43].
8.3.3 Amorphous Structures

To model disorder, a 56-atom unit cell as described in Fig. l(b) was used [38] and
heated to l0000K using a very approximate Tersoff potential [12]. We chose
Figure 4. P6m2 structure of hexagonal graphite C3N4.

Figure 5. Charge density of the graphitic C3N4.
standard Tersoff parameters for C and for N those that give best results specifically
for the cubic form of C3N4,bearing in mind (as pointed out by Kroll [12]) that the
Tersoff potential has limitations in describing N-N bonds. The subsequent compu-
tational procedure was quite time consuming for convergence, this indicating that
the initial starting point was indeed a highly metastable configuration. Relaxation
was monitored throughout and we noted importance of the various bonding config-
urations, in particular the build up of C-N bonds and competition with C-C and N-
N bonds. Although the number of atoms in the simulation was small we did observe
that N-N bonds were often not connected into the complete network as much as C-
C bonds were. Instead the N atoms tended to group into relatively loosely bonded
structures often isolated from the main continuous network. This can be seen in
Figure 6. Calculated structure of amorphous C3N4.
Fig. 6 which shows the distribution of atoms in the amorphous 56-atom cell at a
volume very near the stable volume for such a structure.
This equilibrium volume corresponds to a material density of p = 2.9 g ~ m - still
~,
relatively high but far lower than the crystalline hard material p = 3.8 gcmP3.
Specifically at p = 2.9 gcmP3 there are a larger number of C-N bonds than either
C-C or N-N bonds and this is typical of most volumes considered up to a material
density of 4.5 g ~ m - Thus
~ . the modeling has shown that C-N bond formation is
preferred over others.
8.3.4 Relative Stability

The equilibrium results for the various crystalline, graphitic and amorphous C3N4
systems are given in Table 2 and the relative energies in Fig. 7.
Only modest pressure is needed to overcome the energy barrier separating the
graphitic phases and induce a transition between them. Estimating this pressure
from the slope at the common energy/volume intersection gives a hydrostatic
Table 2. Structure, bulk modulus of some phases of C3N4.

Phase Shortest C-N Interplanar B Relative energy
distances (nm) distance (nm) (GPa) (eV/atom)
P 0.145 432 0.000

Cubic (defect zinc-blende) 0.148 428 $0.200
P63/rn graphite 0.132-O.145 0.329 253 -0.002
p-graphite (odd-phase) 0.137 0.337 205 +0.029
p-graphite (even-phase) 0.137 0.372 198 +0.038
Amorphous 128 +0.564
-220.4t 1
-221.0'. I
40 50
, ' '
60
'
vo~ume(a.u.~)
Figure 7. Energy-volume relationships for various phases of C3N4.
transition pressure necessary to transfer between the various graphite phases to be less
than 5 GPa: a uniaxial compression is far less than this. Hydrostatic pressures that lead
to the formation of the superhard P-phase calculated in the same way for graphite
phases however are much higher. For example again from the slope at the common
energy intersection we calculate that a pressure of about 80GPa is needed to go
from the P63/m phase to the superhard P-phase and a little less at about 55 GPa to
go from the odd-phase p-C3N4 phase and only slightly more for other graphitic
phases. Such transition pressures are well above the value of 12 GPa suggested recently
by Teter and Helmley [35] to go from P63/m graphite to the superhard j3-phase,
although the way in which this value was estimated is not clear. Very likely this
value was based upon a upon common energy/pressure arguments that neglect the
importance of intermediate changes in volume of the structures.
8.4 Silicon Carbon Nitride

As discussed earlier the p-form of Si3N4was the prototype material leading to the
postulate that C3N4may be harder than diamond [5,40]. Recently tertiary ceramics
in the form Si-C-N are being investigated [4447] for potentially superhard proper-
ties. Amorphous forms of Si-C-N materials are well documented [48] but often such
materials segregate into Sic or Si3N4when heated; a stable crystalline form of Si-C-
N that does not behave in such a manner is being sought. Crystalline S i x - N com-
pounds that have so far been produced have had their structure contested depending
upon the nature of the locality about Si. Using extended X-ray-absorption fine
structure studies (EXAFS) either a local tetrahedral [49] or random [50] structure
about Si has been suggested. In both these cases Si atoms are linked by a C-N net-
work. More recent EXAFS measurements [47] coupled with computer modeling
favored the local tetrahedral Si-N-C arrangement. The crystalline structure of
this Si-C-N material was found to have a structure very near to that of
3 ~ ca) unit cell of dimension a = 0.6904nm and c = 0.5260nm.
~ t S i ~ N ~ ( Pwith
Figure 8. Beta phase of SiC2N4.
The importance of the local tetrahedral structure has also been noted in the
analysis of a lower density form of SiC2N4 compound recently produced by
Riedel and coworkers [45]. Here X-ray diffraction showed the possibility of an
almost cubic (Pn3m) phase with a unit cell of measured dimension a = 0.61885 nm.
In such a material the local Si related tetrahedral structure is realized as SiN4 units
that are connected through N=C=N chains. The flexibility of the N=C=N bond
has been considered an essential feature for more extended SiC2N4structures [45,51].
Following earlier theoretical results on fJC3N4,one of the hardest forms of
SiC2N4is expected to be fJSiC2N4.This is obtained by replacing two C atoms in
the 14-atom unit cell of fJC3N4with Si atoms maintaining lowest C-N bond lengths
[52]. Possibly this form can be attained through a synthesis starting from one of
structures discussed above.
8.4.1 PSiC2N4
In Fig. 8 we show the structure of fJSiC2N4as viewed along the c-axis of the
hexagonal unit cell with details of the structure being presented in Table 3 .
We see that the bulk modulus of SiC2N4is about $ that of C3N4.
In Fig. 9 we show the calculated charge density in one of the hexagonal planes.
There are two important features to be noted on comparing this with a similar
charge distribution for the fJC3N4shown in Fig. 1. First very little charge is located
about the Si atom, suggesting that Si bonding be insignificant compared with
Table 3. Calculated structure and Bulk Modulus of some phases of SiC2N4.
Phase Unit cell (nm) B (GPa) B' Relative energy (eVjatom)
PSiC2N4( P 2 / m) a = 0.678, b = 0.692; c = 0.258 330 4.01 0.000

a (Pn3m) a = 0.664 106 3.90 -0.048
b a = 0.665 95 2.68 -0.029
c (~4n2) a = 0.632 32 4.89 -0.052
Figure 9. Charge density of pSiC2N4 plotted perpendicular to the c-axis.
the C=N bond. Second is the direction of the C=N bonding; if we look at the
N atom we see that the main charge is not directed along the C=N bond itself
but with N(2p) orbitals being directed perpendicular to the bond. Surprisingly the
C-N bond is very similar in both silicon-carbon-nitride and carbon nitride again
suggesting the importance of the C=N bond.
8.4.2 Near-cubic Forms of SiC2N4

Recently a low density phase of SiC2N4has been investigated [45]with a possible
cubic structure (Pn3m) shown in Fig. 10.
In such a structure, Si surrounded tetrahedrally by four N atoms is the
Si N C N Si
a
N
b Si C Si
N
N N
C Si C Si
Figure 10. Cubic phase of SiC2N4.Large open circles are Si, small black circles C and small open
circles N. The nature of the Si-N=C=N-Si bond is shown for each case.
essential building block and a linear Si-N=C=N-Si bridging unit. The measured
unit cell constant is 0.618nm, and Riedel et al. noted that this size of cell would
have a short interatomic spacing between the C and N atoms of 0.11 9 nm. They
therefore suggested that the measured cell constant may correspond to an effec-
tive value relating to a random distribution of N atoms about the Si-C-Si unit.
The linearity of the N=C=N bond can also be maintained in a similar symmetry
in the manner suggested in Fig. 10b and more recently Kroll et af. [51] have sug-
gested another structure for the cubic phase involving N=C=N bond with overall
P4n2 symmetry shown in Fig. 1Oc. The nonlinear C=N=C bond is expected to
lead to a softening of the structure [53]. As yet the several cubic structures
have not been related to the potentially superhard BSiC2N4 phase. We have
calculated the structure of the three new possible phases and the results are
given in Table 3.
60 120
volurne(a.u.3)
Figure 11. Energy-volume behavior of different phases of SiC2N4:(a), (b), and (c) identify the
different phases shown in Figure 10. The pressure shown is obtained from the intersection point
of the c-phase and the P-phase.
8.4.3 Relative Stability

In Fig. 11 we show the energy/volume relations for each model of the cubic phase
and for the PSiC2N4phase.
The energy of each of the cubic phases lies below that of the P-phase. This is not
unlike the situation for graphite and diamond and so a cubic to P-phase synthesis
may be possible. Estimating the transition pressure needed to change from one
phase to another is from the slope at the energy intersection of the common
phases; this is the pressure necessary to overcome the energy barrier between the
two phases. In the softer of the three cubic phases the pressure is quite large,
about 62 GPa.
8.5 Conclusions
There is understandably some pessimism regarding a commercial synthesis of
superhard carbon nitride especially in view of the intense effort that has been put
into its synthesis by many laboratories in the last few years with no firm result to
date. Yet the wisdom of the computational arguments upon which the various
structures of the material have been predicted seems firm and they have been
shown to be capable of giving us valuable insight in many other areas of materials
science. Experiences gained regarding the commercial synthesis of diamond and
cubic BN could probably lead ways that carbon nitride can be synthesized, yet
other more novel routes also cannot be excluded.
Acknowledgements
The National Research Foundation (South Africa) is thanked for its support.
References 269
References
1. 0. Borgen and H. M. Seip, Actu. Chem. Scand. 1961, 15, 1789.
3. K. M. Yu, M. L. Cohen, E. E. Haller, W. L. Hansen, A. Y. Liu, and I. C. Wu, Phys. Rev. B,
1994,49, 5034.
4. M. L. Cohen and J. R. Chebkowsky, Electronic Structure and Optical Properties of Semiconduc-
tors, Springer Verlag, Germany, 1988.
5. A. Y. Liu and M. L. Cohen, Phys. Rev. B, 1990,41, 10727.
6. Y. Miyamoto, M. L. Cohen, and S. G. Louie, Solid State Commun. 1997, 102, 605.
7. Y. Guo and W. A. Goddard 111, Chem. Phys. Lett. 1995, 237, 72.
8. R. K. Kalia, A. Nakano, A. Omeltchenko, K. Tsurta, and P. Vashishta, Phys. Rev. Lett. 1997,
78, 2144.
9. J. Tersoff, Phys. Rev. Lett. 1988, 61, 632.
10. W. Sckkal, B. Bouhafs, H. Aourag, and M. Certier, J . Phys.: Condens. Mutter, 1988,10,4975.
11. F. de Brito Mota, J. F. Justo, and A. Fazzio, Phys. Rev. B, 1998, 58, 8323.
12. P. M. Kroll, PhD Thesis, University of Paderborn, 1996.
13. D. Porezag, Th. Frauenheim, Th. Kohler, G. Seifert, and R. Kaschner, Phys. Rev. B, 1995,51,
12947.
14. F. Weich, J. Widany, and Th. Frauenheim, Phys. Rev. Lett. 1997, 78, 3326.
15. R. M. Dreizler and E. K. U. Gross, Density Functional Theory, Springer Verlag, Berlin, 1990.
16. M. C. Payne, M. P. Teter, D. C. Allan, T. A. Arias, and J. D. Joanopoulos, Rev. Mod. Phys.
1992, 64, 1045.
17. D. M. Ceperley and B. J. Adler, Phys. Rev. B, 1980, 45, 566.
18. J. P. Perdew, Phys. Rev. B, 1986, 33, 8822.
19. M. Bockstede, A. Kley, J. Neugebauer, and M. Schemer, Comp. Phys. Comm. 1997,107, 187.
20. N. Troullier and J. L. Martins, Phys. Rev. B, 1991, 43, 1993.
21. F. D. Murnaghan, Proc. Nat. Acud. Sci. USA, 1994,3, 24.
22. F. Birch, J . Geophys. Res. 1952, 57, 227.
23. A. R. Merchant, D. G. McCulloch, D. R. McKenzie, Y. Yin, L. Hall, and E. G. Gerstner,
J. Appl. Phys. 1996, 79, 6914
24. M. Y. Ching, D. Li, X. Lin, V. P. Dravid, Y. Chung, M. Wong, and W. D. Sproul, J. Vuc.Sci.
Technol. 1993, 11, 521.
25. D. M. Bhusari, C. K. Chen, K. H. Chen, T. J. Chuang, L. C. Chen, and M. C. Lin, J. Muter.
Res. 1997, 12, 322.
26. J. Martin-Gill, F. Martin-Gill, M. Sarikaya, M. Qian, M. Jose-Yacaman, and A. Rubio, J . Appl.
Phys. 1997, 81,2555.
27. K. Yamamoto, Y. Koga, K. Yase, S. Fijiara, and M. Kubota, Jpn. J . Appl. Phys. 1997, 36,
L230.
28. Z. J. Zhang, S. Fan, and C. M. Lieber, Appl. Rev. Lett. 1995,66, 3582.
29. M. Tabbal, P. Merel, S. Moisa, M. Chaker, A. Ricard, and M. Moisan, Appl. Rev. Lett. 1996,
69, 1698.
30. Y. Chen, L. P. Guo, and E. G. Wang, Philos. Mag. Lett. 1997, 75, 155.
31. Kehui Wu, E. G. Wang, Jian Qing, and Guichang, Xu, J . Appl. Phys. 1998, 83, 1702.
32. C. Ronning, H. Feldermann, R. Merk, H. Hofsass, P. Reinke, and J. U. Thiele, Phys. Rev. B,
1998,58, 2207.
33. H. W. Lu, X. R. Zou, and J. Y. Feng, J . Phys. D: Appl. Phys. 1998,31,363.
34. J. Ortega and 0. F. Sankey, Phys. Rev. B, 1995, 51, 2624.
35. D. M. Teter and R. J. Hemley, Science, 1996, 271, 53.
36. J. Furthmuller, J. Hafner, and G. Kresse, Phys. Rev. B, 1994, 50, 15606.
37. A. M Liu and R. M. Wentzcovitch, Phys. Rev. B, 1994, 50, 10362.
38. J. E. Lowther, Phys. Rev. B, 1998, 57, 5724.
39. C. Cote and M. L. Cohen, Phys. Rev. B, 1997, 55, 5684.
41. J. V. Badding, Adv. Muter. 1997, 11, 877.
42. J. E. Lowther, Phys. Rev. B, 1999, 59,11683 .

43. Sjostrom, S. Stafstrom, M. Boman, and J. E. Sundren, Phys. Rev. Lett. 1995, 75, 1336.
44. R. Riedel, H. J. Kleebe, H. Schonfelder, and F. Aldinger, Nature, 1995, 374, 526.
45. R. Riedel, A. Greiner, G. Miche, W. Dressler, H. Fuess, J. Bill, and F. Aldinger, Angew. Chem.
Znternat. Edn, 1997 36, 603.
46. A. Bendeddouche, R. Berjoan, E. Beche, T. Merie-Mejean, S. Schamm, V. Serin, G. Taillades,
A. Pradel, and R. Hillel, J . Appl. Phys. 1997, 81, 6147.
47. Y. K. Chang, H. H. Hseich, W. F. Pong, M. H. Tsdi, T. E. Dunn, F. Z. Chien, P. K. Tseng, K. L.
Tsang, W. K. Su, L. C. Chen, S. L. Wei, K. H. Chen, D. M. Bhusari, andY. F. Chen, Phys. Rev.
B, 1998, 58,9018.
48. M. Driss-Khodja, A. Georghiu, G. Dufour, H. Roulet, C. Senemaud, and M. Cauchetier, Phys.
Rev. Lett. 1985, 54, 1985.
49. A. Gheorghiu, S. Senemaud, H. Roulet, G. Dufour, T. Moreno, S. Bodeur, C. Reynaud,
M. Cauchetier, and M. Luce, J . Appl. Phys. 1992,71,4118.
50. F. Tenegal, A. M. Flank, and N. Herlin, Phys. Rev. B, 1996, 54, 12029.
51. P. Kroll, R. Riedel, and R. Hoffman, Phys. Rev. B, 1999,60, 3126.
52. Cheng-Zhang Wang, En-Ge Wang, and Qingyun Dai, J. Appl. Phys. 1998, 83, 1975.
53. E. Kroke, personal communication, 1998.
9 Effective Doping in Novel sp2 Bonded Carbon

Allotropes
G. Jungnickel, P. K. Sitch, T. Frauenheim, C. R. Cousins,
C. D. Latham, B. R. Eggen, and M. I. Heggie
9.1 Introduction
It has been speculated that there exists a multitude of metastable forms of carbon.
This arises from the preparation of macroscopic quantities of fullerenes [l, 21, the
observation of carbon nanotubes [3] and onions [4]. and the successes in carbon
network chemistry [5,6]. This is based on the well-known fact that carbon atoms
can frequently occur in different local environments.
In particular, carbon can form hybrid orbitals and is found to be either sp3
bonded and four-fold coordinated as in diamond, sp2 bonded and, hence, three-
fold coordinated as in graphite, or sp bonded and two-fold coordinated as in
acetylene. Whereas in diamond one finds o bonds as in silicon, in the latter two
forms the carbon atoms can lower their energy by creating additional 7c bonds to
their neighbors, that is they form double and triple bonds. In contrast to diamond,
materials based on these under-coordinated types of carbon tend to form layered or
chain-like structures which are usually less dense than graphite. Diamond forms
very strong and short bonds and is, therefore, a promising material for applications
in harsh environments. The intraplane bonding in graphite is very strong but the
weaker interplanar forces prevent graphite from being well protected against
mechanical or chemical attack. Carbon atoms can develop intermediate hybridiza-
tion states dependent on their local neighborhood as in the various amorphous
phases that have been prepared during the past decades.
Semiconductor devices based upon carbon would be very attractive when taking
advantage of the short and stiff carbon-carbon bonds in combination with the light-
ness of the carbon atoms since this would result in a material with a thermal conduc-
tivity almost as good as in diamond, which is stable against thermal degradation. A
large band gap and a high saturation velocity for electrons as in diamond [7,8]
would make it a superior material for superfast, high-power, high-temperature,
and wide band gap applications.
Unfortunately, diamond itself, though intensively studied, has been found to be
less suitable for device applications. Although it can be doped by boron to make it
a p-type semiconductor it is almost impossible to obtain the n-type counterpart
for a p-n junction. Diamond cannot be doped by the most likely candidate, nitrogen,
as a substitutional impurity. Instead nitrogen moves off its lattice site causing a dan-
gling bond on one of its carbon neighbors which in turn results in a deep level [9-121.
Other impurities such as phosphorus [16-181 or lithium [19] have been tried for n-
type doping in diamond but their size reduces the solubility in diamond to near
272 9 Effective Doping in Novel sp2 Bonded Carbon Allotropes
zero [9]. The most promising alternative, P, results in an electrical activity much less
than would be expected from its concentration so that the doping efficiency is rather
low [17]. Moreover, it has been found that the phosphorus-vacancy complex [20]
is more stable than substitutional P. This complex traps injected electrons and
deactivates the impurities.
In amorphous carbon (a-C) substitutional impurities can relax off their sites as in
diamond or cause substantial rearrangements in the carbon host. Injected holes or
electrons migrate in a-C to occupied high energy n: or low energy unoccupied n*
levels, respectively. Associated n-bonded clusters relax to screen the additional
charge [12-141 and the conductivity can only occur through variable range hopping
between such clusters.
In the light of these findings, in a recent paper [ 151 the doping problem in carbon
has been attacked from a different point of view. Instead of searching for other atom
types or complexes that effectively would act as shallow donors metastable carbon
modifications which accept both boron and nitrogen as substitutional impurities
have been proposed to exist.
Among the hypothetical forms of carbon predicted prior to this work are stacked
layers made of double and acetylene triple bonds [21] yielding structures like
‘graphyne’ and related polymeric networks. These planar configurations were
predicted to have quite large electronic gaps but also found to have a lower
number of atoms per unit layer area than graphite. Equivalent building blocks
were used to create polyines in quasi-one-dimensional frameworks [22] or ‘super-
diamond’ and related structures [5] in three-dimensionally connected networks.
Another class of networks has been proposed by combining sp2 atoms in a non-
layered arrangement. The idea was to find a material based upon the thermo-
dynamically most stable short-range order but with properties close to diamond.
In order to make use of the short bonding between two graphitic carbon atoms to
obtain the highest bulk moduli, Hoffman et al. [23] predicted the particular structure
which they named BCT4 to be metastable. Tamor and Hass proposed that a very
similar structure called H6 possibly exists [24]. Though the latter is topologically
directly related to diamond it has been found to be rather unstable unlike the metal-
lic phase BCT4 [25,26] and converts without a pronounced barrier back to
diamond [27]. The most remarkable feature of these structures, as well as another
energetically unfavorable structure [28] (BCC4 in the following), is that they do
not contain any six-membered rings as is typical for graphite.
Other purely three-fold coordinated networks have been proposed inspired by
tesselations of infinite triply periodic minimal surfaces [28-321. All of these
structures are composed of five-, six-, and seven-membered rings which are the
major components in amorphous structures too. They establish the whole family
of negatively curved graphitic sheets. The density of these structures is usually
lower than that of graphite but they have been predicted to be more stable than
fullerene c 6 0 [32,33]. Electronically the proposed structures behave like either
metals or insulators with gaps smaller than that of f.c.c. c 6 0 [34]. An exception
with respect to this is polybenzene [28] which has been found to be a low
energy form of carbon that opens an indirect band gap of 2.96 eV [34]. This excep-
tionally large gap has been attributed to the very special topology which consists of
benzene-like six-membered rings interconnected by a three-dimensional framework

of single bonds [28].
By combining three- and four-connected atoms, Bucknum and Hoffmann gener-
ated a carbon network based upon the 1,4-cyclohexadiene motif. This structure has
a density between that of graphite and diamond. Owing to the geometric constraints
that force unconjugated double bonds face-to-face onto each other, the system
shows a significant indirect band overlap across the Fermi level and is predicted
to be thermodynamically quite unstable [35]. A whole family of similar networks
which mix three-fold and four-fold coordinated carbon atoms has been proposed
by Merz et al. [36].
Yin [37], Biswas et al. [38], Fahy and Louie [39], Mailhiot and McMahan [40],
Boercker [41], Crain et al. [42], Furthmuller et al. [43], and Teter [44] all investigated
a dense carbon phase containing sp3 atoms only as in diamond, which is known
from silicon and is called BC8. This and other interesting metastable structures
with four-fold coordination such as ST12 [45], R8 [42], lonsdaleite, and H3 [44]
have been investigated to establish theoretically the stability limit for diamond.
Another class of four-fold coordinated low-energy forms of carbon has been
predicted by Nesper et al. [46], Adams et al. [47], and Benedek et al. [48], and is
due to a carbon decoration of zeolite structures. In particular, structures known
as clathrates provide ideal hosts for four-fold coordinated carbon atoms. In con-
trast, the compact icosahedral clusters with tetrahedral bonding proposed by
Zeger and Kaxiras [49,50], though nice geometric arrangements, are quite high in
energy and thermodynamically rather unstable if the outer dangling bonds are
not saturated with hydrogen.
Finally, metallic allotropes with higher coordination numbers have been
studied [39,51] and found to be well removed in enthalphy from four-fold co-
5
ordinated structures below a pressure of 10 Mbar. Here, we report on the eometric
optimization and doping properties of the recently proposed purely sp bonded
carbon phases [15] and a propotype zeolite structure (sp3 bonded) called melanoph-
logite using the density-functional based tight-binding method (DF-TB) [52,53].
The sp2 bonded structures are based upon the concept of connecting ‘ideal’ con-
formations of eight-membered rings. Hence, the structures fill the gap between
carbon phases based upon benzene rings (graphite, polybenzene, some polyines)
and the models based on an interconnection of tenfold rings (BCT4, H6, BCC4).
The carbon crystals using eight-membered rings as the main structural motif have
a similar stability to BCT4 and fullerene C60,and belong to local minima in the
energy landscape for carbon [ 151. They are expected to be thermodynamically
very stable, since a transition to diamond or graphite would require a large
number of bonds to be broken and rearranged.
The sp3-bonded zeolite type structures are obtained from natural silica frame-
works by substituting the silicon atoms with carbon and removal of oxygen. We
demonstrate that considering n-type doping such structures [4643] behave very
much like diamond. A mid-gap state occurs after a substantial relaxation of the
impurity which is localized on a nearby carbon atom.
In contrast, the topologies of the purely sp2-bonded structures favor the incor-
poration of boron or nitrogen on substitutional lattice sites and do not permit a
reorganization of the carbon neighborhood. The relaxation is weak and ensures

that injected holes or electrons remain associated with the substituent. They provide
half-filled electronic levels close to either the valence or the conduction band. We
compare the situation with doping effects in diamond, amorphous carbon, poly-
benzene, and the zeolites.
The recent synthesis of crystals made of the DGhC36 fullerenes [54] and of a
material called Quasam [55] (which has been declared to be the lightest hard sub-
stance yet discovered) supports the belief that there is a real chance of preparing
novel forms of carbon such as those described in this paper. Quasam is believed
to be made up of graphite-like planes joined together by four-fold coordinated
carbon atoms and has been produced by fragmenting an organic compound.
9.2 Lattice Description

The hypothetical carbon phases that we predict to be metastable are presented in
their relaxed state in Fig. l(a, b). Figures l(c, d) show polybenzene and the melano-
phlogite structure.
By exploiting the special feature of boat and chair conformations of eight-
membered rings which contain sp2-sp2 bonds, potentially double bonded extremely
hard three-dimensional networks can be constructed. Similar to ethylene these
networks give rise to large n-n* splittings of the electronic levels at the Fermi
energy. As a result a wide bandgap material with superior properties for electronic
applications can be formed. For this, the connection of the octahedrons must be
such as to minimize the x-orbital overlap of adjacent double bonds.
A structure made from chair conformations of eight-membered rings is shown in
Fig. 1(a). There exist two very similar uniform three-connected networks, the space
groups of which have been identified by Wells [56] to be hexagonal P6222 and
rhombohedral RTm. Carbon models can be made for these nets with equal bond
length and angles as in graphite. The two topologies differ only slightly in the form
of the screw axis incorporated. Hence, the properties of crystalline structures
obtained by decoration with carbon atoms are expected to be rather similar.
The screw axis in both cases is such that there is little n-orbital overlap between
sp2 neighbors along the axis, whereas the n-orbital overlap parallel to bonds inter-
connecting the screw axis is maximized. Since both structures are topologically
almost identical we will concentrate on discussing the rhombohedral form only.
During relaxation an ideal crystal with graphite-like bond properties undergoes
a transition to a less symmetric monoclinic form of space group P112/m. Follow-
ing the naming conventions for BCT4 and H6, we will nevertheless call the
structure R6.
Structure l(b) which we will call BCT8 in its ideal configuration has body-
(a,
centered tetragonal symmetry (I4lmd) with the generating atom at the position
i, 0). For the idealized structure with equal bond lengths and angles the lattice
constants are given by a = b = 4d and c = J8d, where d is the common bond
9.2 Lattice Description 275
Figure 1. Perspective views of the wide band gap models showing the three-dimensionally connected
networks.
length. This structure results from connecting boat-shaped eight-membered rings

while avoiding the construction of smaller rings. Dropping the latter requirement
and allowing six-membered rings to occur gives rise to a simple cubic lattice which
is known as the polybenzene structure [28] and shown for completeness in
Fig. l(c). The properties of this simple cubic structure denoted here as SC24
are expected to be different since the n-bonded subunits are isolated six-membered
rings within the rigid single bond environment. Isolated double bonds do not exist
in this structure. Topologically, both models belong to the class of Archimedian
three-connected nets [56]. Each lattice point belongs to two eight-membered
rings and to one of another kind. In case (b) the other ring type is a ten-membered
ring whereas in model (c) it is the six-membered ring as mentioned above. It is
worth noting that structure BCT8 is topologically related to R6 in the sense
that it contains screw axes that are interconnected by bonds potentially double
276 9 Eflective Doping in Novel sp2 Bonded Carbon Allotropes
bonded. Here, however, the projection of the axis onto a plane perpendicular to it
is a square, whereas in the former case it is a triangle. SC24 does not contain such
a feature.
BCT8 and R6 both contain a rigid multichannel structure which may be promis-
ing for applications where anisotropic transport at a microscopic level is of interest.
Such properties have been predicted for hypothetical crystals made of carbon
nanotubes [57]. However, synthesizing such crystals requires one to prepare,
purify, and align carbon tubes of equal size [58].
Finally, we investigate the doping behaviour of a prototype zeolite structure
named melanophlogite, which has been detected to be one of the lowest energy
forms of this kind [46]. The space group of this structure is P m n . The simple
cubic system contains 46 atoms in its primitive cell and we will refer to it as
SC46. The model used throughout these studies is shown in Fig. l(d).
9.3 Computational Methods

To prove that the topologies described above form metastable configurations if
decorated with carbon we explored the energy landscape with respect to variations
of the lattice shape. For this, we started from the idealized structures with graphite
bond length and angles and let them relax to find the ground state with respect to
the energy functional as defined in the density-functional based tight-binding
method (DF-TB) [52,53]. Within this technique the total energy of a crystal is
approximated by a sum over all occupied electron levels obtained by solving the
many-body Schrodinger equation and a classical interaction that describes the
ion-ion core repulsion and the double counting corrections. To solve the
electronic problem efficiently we expand the wave function in a minimal basis of
localized contracted orbitals and transform the Schrodinger equation into a gen-
eral eigenvalue problem. For this we make use of the two-center approximation
to evaluate the Hamiltonian and overlap matrix elements. The eigenvalues are
then obtained by diagonalization and the Brillouin zone integration is done
employing a Gamma-point sampling through the large conventional unit cell
approach. The DF-TB method has been shown to coincide with more sophisti-
cated self-consistent field methods considering the energetics and ground state geo-
metries of small carbon clusters [52], fullerenes and fullerene oligomers [59,60],
diamond surfaces [61,62] and amorphous carbon phases [63]. B and N impurities
in amorphous material [ 121 and in sublayers of diamond surfaces [64,65] have been
successfully described by DF-TB.
The true local minima in the energy landscape were found by updating the atomic
coordinates for a sufficiently large number of lattice vector combinations using an
efficient conjugate gradient algorithm. It should be noted that the lattice vectors
have been varied independently in order to find energetic minima without applying
symmetry constraints. The lattice shape could change during the optimization
process. Finally, the vibrational density of states within the quasi-harmonic
9.3 Computational Methods 277
Table 1. DF-TB ground state properties of hypothetical model structures compared with the results
for natural carbon allotropes and recent calculations with more sophisticated methods.
Structurea Lattice Density Atomic Bulk Total energy Bond Bond

parameters volume modulus E-EDlamond lengthsb anglesb
[nml [ g ~ m - ~[nm’/atom]
] [GPa] [eV atom-’] [nm] [degl
DF-TB results
Diamond a = 0.3569 3.51 0.00568 49 1 0.00 0.1545 ( l / l ) 109.5 ( l / l )
{(66)-a}
Graphite a = 0.2458 2.27 0.00877 - -0.01 0.1419 ( l / l ) 120.0 ( l / l )
c = 0.6700
c60 0.41 0.1397 (1/3) 108.0 (1/3)
0.1448 (2/3) 120.0 (2/3)
BCT8 a = 0.5873 2.35 0.00849 297 0.52 0.1342 (1/3) 115.6 (1/3)
{ (82.10)-b} c = 0.3937 0.1496 (2/3) 122.2 (2/3)
R6 a = 0.4318 2.26 0.00882 316 0.54 0.1343 (1/3) 117.2 (1/3)
{(8,3)-b) b = 0.3589 0.1498 (2/3) 121.4 (2/3)
c = 0.7101
SC24 a = 0.6070 2.14 0.00932 324 0.22 0.1399 (2/3) 120.0 ( l / l )
{(6.8*)} 0.1494 (1/3)
SC46 a = 0.6671 3.09 0.00645 395 0.12 0.1550f 109.3 f 5.3
0.0028
BCC4 a = 0.41 16 2.29 0.00872 338 1.44 0.1455 (I/]) 120.0 (I/])
{(10,3)-a}
BCT4 a = 0.2502 2.90 0.0068 1 449 0.49 0.1425 (2/3) 118.6 (2/3)
{(10,3)-b) c = 0.8698 0.1492 (1/3) 122.8 (1/3)
H6 a = 0.2607 3.18 0.00627 519 0.75 0.1462 (2/3) 117.1 (2/3)
{(10,3)-~} c = 0.6396 0.1465 (1/3) 125.8 (1/3)
Reference section
SC24 [28] a = 0.6033 2.19 0.0091 1 - 0.21 0.1396 (2/3) 120.0 ( l / l )
0.1474 (1/3)
SC46 [47] a = 0.6680 2.95 0.00675 - 0.14 - -
BCC4 [28] a = 0.4065 2.37 0.00842 - 1.26 0.1437 ( l / l ) 120.0 ( l / l )

BCT4 [26Ic a = 0.2512 2.94 0.00679 362 0.52 0.1423 (2/3) 118.1 (2/3)
c = 0.8599 0.1480 (1/3) 123.8 (1/3)
H6 [25jC a = 0.2618 3.18 0.00626 372 0.82 0.1445 (2/3) 116.9 (2/3)
c = 0.6327 0.1468 (1/3) 126.2 (1/3)
a Wells’s [57] symbols for 3D-nets given in curly brackets.

Ratios in brackets denote percentage of bonds or angles.
Except for bulk moduli extended basis set results ([75] differences from [25,26] are small).
~
approximation has been calculated by diagonalization of the dynamical matrix for

the ground state lattice to guarantee that there are no negative nor zero phonon
modes.
The static properties of the relaxed ground state models of BCT8 and R6 are
summarized in Table 1 together with structural details of the lattices SC24,
SC46, BCC4, BCT4, and H6 for comparison, all handled as described above.
The results for the latter models agree very well with the recent more sophisticated
278 9 EfSeective Doping in Novel sp2 Bonded Carbon Allotropes
calculations, see Table 1, and demonstrate the accuracy of the tight-binding

method used here.
9.4 Static Properties

As has been expected from the local geometries, the lattices with screw axes inside
respond to the carbon interaction by a strong n-orbital overlap along bonds con-
necting these helical subunits. Hence, the associated bonds are strengthened and
shortened. Double bonds with the typical length of about 0.134nm are established.
The bonds within the helices, however, have a pronounced single-bond character
since the dihedral angles between the atoms prevent a strong n-orbital overlap.
Therefore, these bonds are longer and have typical o-bond lengths close to
0.150 nm. For comparison, the ideal bond length for the diamond lattice is calcu-
lated to be 0.155nm with the DF-TB method. These quite different bond lengths
introduce stress into the idealized crystals. Accompanied bond angles can no
longer be equal and differ significantly from 120". However, on average the graphite
value is still preserved which has been found for H6 and BCT4, too. The strain
energy is balanced by the energy gain due to the n-bonded atom pairs. As a
result, R6 and BCT8 are about 0.5eVlatom less stable than graphite. Due to the
strong overlap between the n-orbitals along the shorter bonds the new crystals are
expected to become insulators with a clean n-n* band gap.
The energy of the metallic phase BCT4 is found to be about the same. Here, the n-
bonded system which lowers the energy consists of extended zig-zag chains within
the crystal. Isolated double bonds do not exist in this system. Instead, graphite-
like bonds parallel to the chains are established with a length of about 0.142nm.
Note also that the unique bond length in graphite is calculated within DF-TB to
be 0.142 nm.
The polybenzene structure SC24 is more stable than the R6 and BCT8 by about
0.2eV/atom because all the bond angles in this structure remain at there ideal
values during relaxation and the n-interaction further stabilizes the six-membered
ring structures. While the former is also true for BCC4 the overall n-orbital overlap
weakens the whole structure and makes it less stable than graphite by almost 1.5 eV/
atom. For comparison, the total energy of the lowest energy forms of amorphous
carbon have been determined by the DF-TB method to be about 0.7-0.8 eV/atom
above graphite. C60 is less stable than graphte by about 0.42eV/atom with the
DF-TB technique which compares well to the experimental value of 0.44 eV/
atom [66]. Diamond is slightly less stable than graphite by 0.01 eV/atom which is
close to the experimental value of 0.025 eV/atom [67]. Therefore, the structures
BCT8, R6, SC24, and BCT4 may well exist, since their total energies are well
below amorphous carbon material and comparable to the natural allotrope fullerene
C60.In contrast, BCC4 is unlikely to be synthesized by the same arguments.
The bulk moduli of BCT8, R6, SC24, and BCC4 are found to be about the same,
i.e. almost 2/3 that of diamond. Hence, the structures are expected to be very hard.
9.5 Electronic Properties 279
H6 and BCT4 both have a bulk modulus which, within the tight-binding model is
comparable in magnitude to that of diamond. In accord with an empirical TB
scheme [24] H6 seems to be even harder than diamond while SCF plane wave
calculations [25,26] predict that the almost equal bulk moduli of BCT4 and
H6 are 17% below the diamond value. The bulk moduli of the models have been
determined by calculating the elastic compliances after applying suitable strains
to the crystals and inversion of the volume compressibility [68].
A four-fold coordinated structure obtained by decoration of melanophlogite with
carbon has been determined to be one of the two lowest energy forms of zeolite type
hypothetical carbon allotropes [46]. Within a converged SCF plane wave calculation
the total energy of this structure was determined to exceed that of diamond by only
0.09 eV/atom. The band gap was found to be almost 4 eV and only 6% lower than
that of diamond utilizing the local density approximation. We find the structure to
be 0.12 eV/atom less stable than diamond with a band gap which is about 4.4 eV and
68% that of diamond.
9.5 Electronic Properties

As mentioned above, BCT8 and R6 contain localized n-bonds within a o-bonded
frame. These n bonds are well relaxed so that there is maximum overlap between
the parallel n-orbitals. Hence, the structures develop pronounced gaps due to the
n-n* splitting of states as is shown in Table 2 and the densities of states (DOS), see
Fig. 2. In order to estimate the gap width and the activation energies more
accurately than is possible within a minimal basis (sp-basis) approach used to
relax the structures, we have recalculated the electronic properties of the models
using an extended basis set that includes d-type polarization functions (spd-
basis). This set was particularly constructed to improve the diamond and graphite
conduction band structure above the Fermi level. The electronic DOS obtained for
SC24, SC46, diamond, and graphite with the spd-basis is shown for comparison in
Fig. 2.
Table 2. Electronic properties of the hypothetical structures compared with diamond. sp indicates
minimal, spd extended basis set, see text. B-C and N - C mark the HOMO participation of the atomic
orbitals at the impurity and the nearest carbon neighbors.
Structure Gap width [eV] B acceptor level [ev] N donor level [eV] A 0 contribution to HOMO
SP SPd SP SPd SP spd B-C N-C
Diamond 7.75 6.49 0.09 0.09 4.61 3.28 0.03 0.50

BCT8 2.99 2.72 0.17 0.20 0.32 0.09 0.07 0.25
R6 3.37 2.79 0.37 0.24 0. I9 0.03 0.48 0.17
SC24 3.65 3.45 0.03 0.01 0.04 0.02 0.01 0.04
SC46 8.22 4.43 0.20 0.34 4.10 2.44 0.46 0.64
1.0
0.8
0.6
0.4
0.2
0.0
0.8
0.6
8n
0.4
0.2
"1" 0.0
0.8
0.6
0.4
0.2
0.0
-25 -20 -15 -10 -5 0 5 -20 -15 -10 -5 0
Figure 2. The (broadened) electronic densities of states for BCTB, R6, SC24, and SC46 show the
wide band gap character of these materials. Diamond and graphite DOS are plotted for comparison.
The most remarkable feature of the structures presented is that they can take up
boron and nitrogen at substitutional lattice sites without a strong relaxation. In
recent DF-TB studies of the doping effect in diamond [12] we could reproduce
the experimental findings that boron yields a shallow acceptor level and nitrogen
a deep donor level which is impractical for electronic applications. Whereas
boron relaxes only a little, nitrogen moves substantially off the substitutional lattice
site. A lone pair electron and a dangling state at one of the carbon neighbors is
created. The carbon defect lies deep in the gap at 3 eV below the conduction band
9.5 Electronic Properties 28 1
and pins the Fermi level. A Boron atom in contrast stays close to its lattice site. The
associated acceptor level is about 0.1 eV above the valence band and pins the Fermi
level. Both values are close to the experimental activation energies of about 2.00 eV
for the single nitrogen donor and 0.37 eV for the boron acceptor [8].
Testing zeolite structures as possible hosts for boron and nitrogen is more
complicated because the impurity can be placed on different lattice sites. We have
checked the three unique places in the melanophlogite prototype structure and
present the results for the best bonding situation found for either boron or nitrogen.
We only observed shallow acceptor (activation energy = 0.32 eV) and donor levels
(activation energy = 0.17eV) in the pure sp3 environment for SC46 when the local
symmetries forced all four carbon neighbors of boron or nitrogen to relax away by
exactly the same distance such that the B-N or C-N bonds had the same length.
This caused degenerate levels at the Fermi energy to appear. Jahn-Teller distortions
removed these degeneracies and yielded lower energy structures.
Generally, all the ground states found were characterized by a off lattice site
relaxation of the impurity. A bond to one of the carbon neighbors was stretched
and became weaker than the others. In the nitrogen case this bond finally was
broken due to the electrostatic interaction between the nitrogen lone pair orbital
and the carbon dangling bond leaving an occupied deep carbon state. Therefore,
it seems to be fairly general that n-type doping by nitrogen is inhibited in a
carbon sp3 environment. Significant lattice relaxations to screen the injected
charge plus the creation of too deep a carbon level seem to be always favored
over establishing a real donor state.
This is in sharp contrast to the hypothetical carbon crystals BCT8 and R6. In
these structures the nitrogen atom is naturally incorporated at a lattice site in a
favorable local geometry. It must take part in the two single and the isolated
double bond associated with each site. The lattice relaxation is extremely weak
and N stays at the lattice site as does a substitutional boron atom. The highest
occupied molecular orbital (HOMO) is occupied by just one electron and acts as
a donor and acceptor, respectively. The activation energies determined within
DF-TB are summarized in Table 2. The HOMO is primarily located at the
substitutional atom and spread over its neighbors as can be deduced from the
participation analysis, see Table 2. Note, that the numbers given are the largest
contributions found. The remaining parts are almost equally distributed over all
the other atomic orbitals in the system. In the case of diamond and the zeolites
the donor level participation ratio is very large indicating that the highest molecular
orbital is strongly localized in the neighborhood of the impurity.
Amorphous carbon materials consist of a mixture of sp3- and sp2-bonded carbon
atoms and the latter form more or less extended n-bonded subclusters in a rigid
single-bond environment [63]. Considering this structure a similar doping effect as
discussed above might be expected. However, the size distribution of x-
clusters [69] and the associated strain in the systems yield broad bond length distri-
bution functions [70]. As a result, the n- and rc*-bandsare broadened and the tails of
the bands overlap with potential acceptor and donor levels. This causes electrons to
be either trapped from the highest energy x-levels or injected into lowest energy x*-
bands, repectively. It is accompanied with relaxation processes of the carbon
network in order to saturate carbon defect states which are always present in
amorphous carbon. Hence, doping is very inefficient in full contrast to the situation
in the proposed crystalline allotropes where the injected hole or electron cannot
migrate to distant n-states.
Finally, we carefully checked the doping results obtained within the DF-TB
scheme against a parameter-free self-consistent-field method employing norm-
conserving pseudopotentials to model atom cores and a localized real-space basis
set of Gaussian orbitals to expand the molecular wave functions and the charge
density [7 11. This computational scheme invokes the local spin density approxima-
tion for exchange and correlation terms within the density functional theory to solve
the many-body Schrodinger equation. The technique has been successfully applied
to study diamond growth [72] reactions, graphitization tendencies at diamond
surfaces [73] and the auto-catalysis of Stone-Wales transformations in fullerene
C60[74]. However, the computational efforts within this code forced us to represent
the new crystals by relatively small clusters. We cut out representative pieces of the
structures including two screw axes and three double bonds in a row interconnecting
them. The cut was done to ensure that only (J dangling bonds occured on the outside
of the clusters which were saturated by hydrogen. The hydrogen atoms then were
kept fixed in order to simulate the crystal boundary. The carbon atoms and the
substitutional impurity could fully relax within this hydrogen cage by a conjugate
gradient algorithm.
The cluster representing BCT8 was found to have an electronic gap of 3.9 eV and
the one representing R6 had a gap of 2.7 eV. The energy differences of the acceptor
and donor levels to the respective band edges calculated within this technique are
0.63eV and 0.67eV in the case of BCT8 and 1.37eV and 0.38eV in the case of
R6. The activation energies tend to be a little higher than determined within DF-
TB. However, this is not surprising taking into consideration that a hydrocarbon
molecule with an artificially strained boundary is used to model the extended crystal.
In the molecules, the central double bond is elongated compared to the outer
two already without any impurity. Nevertheless, the calculations prove that the
relaxation of the impurities is very small so that they do not remarkably move off
the substitutional lattice site. Except for the boron acceptor in R6 we find shallow
levels close to states representing the valence or conduction band edges, respectively.
The differences compared to the periodic DF-TB calculation are enhanced owing to
charge transfer effects to the surface of the molecule in particular into constraint
bonds.
9.6 Conclusions
We could show that it is possible to get n-type doping effects in carbon based
materials using nitrogen as a dopant. The structures where this works, however,
are purely sp2 bonded low energy forms of carbon. The semiconducting allotropes
are characterized by unique screw axes of single bonded sp2 carbon atoms held
References 283
together by isolated well-relaxed double bonds. This particular local geometry

opens wide band gaps and provides an ideal host situation for nitrogen. In contrast
to usual sp3 bonded environments, nitrogen atoms stay at their lattice sites during
relaxation and take part in the isolated double bonds, one of which is associated
with each lattice point. The carbon crystals are metastable forms with a density
close to graphite and large bulk moduli. These mechanical properties together
with pronounced electronic gaps make the BCTS and R6 crystals attractive for
electronic purposes, as protective coatings, and for optical applications. The
purely sp2 bonded phases have a distinct multichannel architecture which may
cause unusual anisotropic transport effects at a microscopic level.
Acknowledgments
We gratefully acknowledge support from the Deutsche Forschungsgemeinschaft
and the EPSRC.
References
1. H. W. Kroto, J . R. Heath, S. C. O’Brien, R. F. Curl, and R. E. Smalley, Nature, 1985,318, 162.
2. W. Kratschmer, L. D. Lamb, K. Fostiropoulos, and D. R. Huffman, Nature, 1990, 347, 6291.
3. S. Iijima, Nature, 1991, 354, 56.
4. D. Ugarte, Europhys. Lett. 1993, 22, 45.
5. F. Diederich and Y. Rubin, Angew. Chem.: Internat. Edn. 1992, 104, 1123.
6. Z. Xu and J. S. Moore, Angew. Chem.: Internat. Edn. 1993, 105, 1394.
7. A. T. Collins, Muter. Sci. Eng. B, 1992, 11, 257.
8. 0.Madelung (Ed.), Semiconductors: Group IV Elements and III- V Compounds, Springer, Berlin,
1991.
9. S. A. Kajihara, A. Antonelli, J. Bernholc, and R. Car, Phys. Rev. Lett. 1991, 66, 2010.
10. C. Cunha, S. Canuto, and A. Fazzio, Phys. Rev. B, 1993,48, 17806.
11. S. J. Breuer and P. R. Briddon, Phys. Rev. B, 1996, 53, 7819.
12. P. K. Sitch, Th. Kohler, G. Jungnickel, D. Porezag, and Th. Frauenheim, Solid State Commun.
1996, 100, 549.
13. P. Stumm, D. A. Drabold, and P. A. Fedders, J . Appl. Phys. 1997, 81, 773.
14. S. Poykko, M. Kaukonen, M. J. Puska, and R. M. Nieminen, Comp. Muter. Sci. 1998,10,351.
15. G. Jungnickel, P. K. Sitch, Th. Frauenheim, B. R. Eggen, M. I. Heggie, C . D. Latham, and
C. S. G. Cousins, Phys. Rev. B, 1998,57, R661.
16. J. F. Prins, Diamond Rel. Muter. 1995, 4, 580.
17. H. Spicka, M. Griesser, H. Hutter, M. Grasserbauer, S. Bohr, R. Haubner, and B. Lux,
Diamond Rel. Muter. 1996, 5, 383.
18. S. Koizumi, M. Kamo, Y. Sato, S. Mitd, A. Sawabe, A. Reznik, C. Uzan-Saguy, and R. Kalish,
19. S. Prawer, Diamond Re/. Muter. 1995, 4, 684.
20. R. Jones, J. E. Lowther, and J. Goss, Appl. Phys. Lett. 1996, 69, 2489.
21. R. H. Baughman, H. Eckhardt, and M. Kertesz, J . Chem. Phys. 1987, 87, 6687.
22. U. H. F. Bunz, Angew. Chem.: Internat. Edn. 1994, 106, 1127.
284 9 Eflective Doping in Novel sp2 Bonded Carbon Allotropes
23. R. Hoffmann, T. Hughbanks, M. Kertesz, and P. H. Bird, J . Am. Chem. SOC.1983,105,4831.

24. M. A. Tamor, and K. C. Hass, J . Muter. Res. 1990, 5 , 2273.
25. A. Y. Liu, M. L. Cohen, K. C. Hass, and M. A. Tamor, Phys. Rev. B, 1991,43, 6742.
26. A. Y. Liu, and M. L. Cohen, Phys. Rev. B, 1992,45, 4579.
27. M. 1. Heggie, and C. D. Latham, Euromat 91: Proc. Second Europ. Con$ Adv. Materials and
Processes, T. W. Clyne, P. J. Withers (Eds), 1992, 3, 239.
28. M. O’Keeffe, G. B. Adams, and 0. F. Sankey, Phys. Rev. Lett. 1992,68,2325.
29. H. Terrones, and A. L. Mackay, Carbon, 1992, 30, 1251.
30. H. Terrones, J. Fayos, and J. L. Aragon, Acta Metall. Muter. 1994, 42, 2687.
31. D. Vanderbilt, and J. Tersoff, Phys. Rev. Lett. 1992, 68, 511.
32. S. J. Townsend, T. J. Lenosky, D. A. Muller, C. S. Nichols, and V. Elser, Phys. Rev. Lett. 1992,
69, 921.
33. R. Phillips, D. A. Drabold, T. Lenosky, G. B. Adams, and 0.F. Sankey, Phys. Rev. B, 1992,46,
1941.
34. M.-Z. Huang, W. Y. Ching, and T. Lenosky, Phys. Rev. B, 1993, 47, 1593.
35. M. J. Bucknum, and R. Hoffmann, J . Am. Chem. Soc. 1994, 116, 11456.
36. K. M. Merz, Jr., R. Hoffmann, and A. T. Balaban, J . Am. Chem. Soc. 1987, 109, 6742.
37. M. T. Yin, Phys. Rev. B, 1984,30, 1773.
38. R. Biswas, R. M. Martin, R. J. Needs, and 0. H. Nielsen, Phys. Rev. B, 1984, 30, 3210.
39. S. Fahy, and S. G. Louie, Phys. Rev. B, 1987, 36, 3373.
40. C. Mailhiot, and A. K. McMahan, Phys. Rev. B, 1991, 44, 11578.
41. D. B. Boercker, Phys. Rev. B, 1991,44, 11592.
42. J. Crain, S. J. Clark, G. J. Ackland, M. C. Payne, V. Milman, P. D. Hatton, and B. J. Reid,
Phys. Rev. B, 1994,49, 5329.
43. J. Furthmiiller, J. Hafner, and G. Kresse, Phys. Rev. B, 1994, 50, 15606.
44. D. M. Teter, private communication.
45. R. Biswas, R. M. Martin, R. J. Need, and 0. H. Nielsen, Phys. Rev. B, 1987, 35, 9559.
46. R. Nesper, K. Vogel, and P. E. Blochl, Angew. Chem.: Znternat. Edn. 1993, 105, 786.
47. G. B. Adams, M. O’Keeffe, A. A. Demkov, 0. F. Sankey, and Y.-M. Huang, Phys. Rev. B,
1994, 49, 8048.
48. G. Benedek, E. Galvani, S. Sanguinetti, and S. Serra, Chem. Phys. Lett. 1995, 244, 339.
49. L. Zeger and E. Kaxiras, Comp. Muter. Sci. 1993, 1, 403.
50. L. Zeger and E. Kaxiras, Phys. Rev. Lett. 1993,70, 2920.
51. M. T. Yin and M. L. Cohen, Phys. Rev. Lett. 1983,50, 2006.
52. D. Porezag, Th. Frauenheim, Th. Kohler, G. Seifert, and R. Kaschner, Phys. Rev. B, 1995,51,
12947.
53. G. Seifert, D. Porezag, and Th. Frauenheim, Znt. J . Quantum Chem. 1996,98, 185.
54. C. Piskoti, J. Yarger, and A. Zettl, Nature, 1998, 393, 771.
55. S. Hill, New Scientist, 1997, 153, 14.
56. A. F. Wells, Three-dimensional Net, and Polyhedra, Wiley & Sons, New York, 1977.
57. J. Tersoff and R. S. Ruoff, Phys. Rev. Lett. 1994, 73, 676.
58. P. M. Ajayan, Cond. Muter. News, 1995, 4, 9.
59. G. Jungnickel, D. Porezag, Th. Kohler, Th. Frauenheim, and M. R. Pederson, Fullerene,
and Fullerene Nanostructures, H. Kuzmany, J. Fink, M. Mehring, and S. Roth (Eds), World
Scientific, Singapore, 1996.
60. D. Porezag, G. Jungnickel, Th. Frauenheim, G. Seifert, A. Ayuela, and M. R. Pederson, Appl.
Phys. A - Mat. Sci. & Processing, 1997, 64, 321.
61. Th. Kohler, M. Sternberg, D. Porezag, and Th. Frauenheim, Phys. Stat. Sol. A , 1996, 154, 69.
62. G. Jungnickel, D. Porezag, Th. Frauenheim, M. I. Heggie, W. R. L. Lambrecht, B. Segall, and
J. C. Angus, Phys. Stat. Sol. A , 1996, 154, 109.
63. Th. Frauenheim, G. Jungnickel, Th. Kohler, and U. Stephan, J . Non-Cryst. Sol. 1995,182, 186,
and refs. therein.
64. P. K. Sitch, G. Jungnickel, M. Kaukonen, D. Porezag, Th. Frauenheim, M. R. Pederson, and
K. A. Jackson, J. Appl. Phys. 1998, 83,4642.
65. M. Kaukonen, P. K. Sitch, G. Jungnickel, R. M. Nieminen, S. Poykko, D. Porezag, and
Th. Frauenheim, Phys. Rev. B, 1998,57, 9965.
References 285
66. H.-D. Beckhaus, S. Verevkin, C. Ruchardt, F. Diederich, C. Thilgen, H.-U. ter Meer, H. Mohn,
and W. Miiller, Angew. Chem.: Internat. Edn. 1994, 33, 996.
67. M. T. Yin and M. L. Cohen, Phys. Rev. B, 1984, 29, 6996.
68. J. F. Nye, Physical Properties ojCrystu/s, Clarendon, Oxford, 1985.
69. U. Stephan, Th. Frauenheim, P. Blaudeck, and G. Jungnickel, Phys. Rev. B, 1994, 50, 1489.
70. G. Jungnickel, Th. Kohler, M. Haase, S. Deutschmann, U. Stephan, P. Blaudeck, and
Th. Frauenheim, J . Non-Cryst. Sol. 1995, 183, 161.
71. R. Jone, and A. Sayyash, J. Phys. C: Solid State Phys. 1986, 19, L653.
72. C. D. Latham, M. I. Heggie, R. Jone, and P. R. Briddon, Diamond Rel. Muter. 1994, 3, 1370.
73. C. D. Latham and M. I. Hegaie. Diamond Re/. Muter. 1995. 4, 528.
74. B. R. Eggen, M. I. Heggie, EyJungnickel, C. D. Latham, R. Jone, and P. R. Briddon, Science,
1996. 272. 87.
75. A. Y. Liu; private communication.
Synthesis and Processing
Part I1
1 Directed Metal Oxidation

V. Jayaram and D. Brandon
1.1 Historical Background

‘Directed metal oxidation’ (DMO) is a term coined by Newkirk and his coworkers
[ 1,2] to describe the formation of a ceramic product by oxidation of a liquid metal in
a growth direction normal to the original free surface of the liquid metal. In the
processes which have been patented by these workers, the liquid metal is commonly
an aluminum alloy and the oxidant air, while growth is usually into a porous
ceramic preform. The end product is a polyphase ceramic composite which may
be tailored to have rather unusual properties, most notably good thermal conduc-
tivity, high wear resistance, and good fracture toughness. These properties are a
direct consequence of the microstructure of the product, which usually contains
some residual metal (enhancing both the thermal conductivity and the fracture
toughness).
Although the commercial development and exploitation of DMO is relatively
recent, the phenomenon has been known for over 70 years, and can be traced
back to the search for suitable refractories which accompanied the dramatic
growth of the aluminum industry following the end of the First World War. The
refractory linings of crucibles and furnaces used in the new industry were sometimes
observed to be attacked by the molten metal, the products of the attack penetrating
into the refractory brick to form a hard dense mass which cracked under the thermal
and growth stresses, allowing the molten metal to escape. The removal and replace-
ment of the damaged refractory could only be accomplished using road-drills and
jack-hammers, so that, even at that time, the unusual mechanical properties of
the DMO material were noted, if not appreciated [3].
Residual sodium in the refractory was quickly identified as the primary element
catalyzing the attack, but with the development of magnesium-containing alumi-
num alloys it was recognized that sodium-free refractory linings might also be
attacked in the presence of magnesium, originating either in the refractory or in
the composition of the alloy melt [4,5]. Work on the development of refractories
for the aluminum industry continues to this day, with specific instances of refractory
penetration recurring as new alloys are developed, most recently for the Al-Li
alloys. In all cases it has been noted that the metallic forms of the elements cata-
lyzing the attack have unusually low vapor pressures: Na, Mg, Li, and to a lesser
extent Zn.
The process patented by Newkirk and his associates (more than 60 patents were
filed in the period 1985-90) covered both the composition of the aluminum alloy
(now to be used as a raw material for the manufacture of the DMO product), the
use of an ‘initiator’ (to promote nucleation of the growth process), control of the
growth rate and microstructure (primarily through the composition, structure,
290 1 Directed Metal Oxidation
and geometry of the preform), and the termination of the reaction (by a calcium
containing ‘poisoning’ agent, used to coat the external surface of the preform), as
well as the actual DMO process conditions (temperature and oxygen partial
pressure).
Subsequent work extended the DMO concept to the gaseous nitridation of
liquid aluminum to form AlN, as well as to a number of boride and carbide products
(especially titanium and zirconium) formed by preform infiltration [2,6]. At the time
of writing (1998) it is not at all clear what the commercial future of these processes
will be. It should be noted that there is a wide range of ‘ceramic’processing technol-
ogies under development, all of which are based on chemical reactions occurring in
combinations of solid, liquid, and gaseous phases. Concrete is an unconventional
example, not usually classed as a ceramic. More conventionally, reaction bonded
silicon nitride (RBSN - gaseous nitridation of a silicon preform), reaction sintered
silicon carbide (the REFEL process - molten silicon infiltrated into a carbon-
containing silicon carbide preform), reaction bonded aluminum oxide (RBAO -
gaseous oxidation of an aluminum-containing alumina preform), and boron carbide
reaction bonded by infiltrating a molten aluminum alloy, all rely on a chemical
reaction to bond the ceramic product into a shaped component.
The DMO process produces a near net-shape product without the need for any
externally applied pressure, and is (in principle) capable of being used for the
economic manufacture of very large ceramic components. It is also capable of yield-
ing a ceramic composite product with some very desirable engineering properties.
Despite the high expectations attached to the process and the extensive investment,
especially by the American government but also by private industry, it is still not
clear which areas of application are going to prove economically viable in the
long term.
1.2 Oxidation and Oxide Formation

The classical approach to the oxidation of metals has been as a one-dimensional
mass transport problem, in which the transport is across a growing, planar film
of the product and is driven by the free energy difference between the anodic and
cathodic processes occurring at the metal-oxide and oxide-atmosphere interfaces,
respectively.
The anodic reaction at the metal-oxide interface, in its simplest form, is just:
+
M = Ma+ ne, and represents the transfer of metal atoms across the interface to
form cations with the accompanying release of free electrons. The corresponding
cathodic reaction may take many forms, of which the simplest (in the presence of
dry oxygen and no other reactants) is: i02+ 2e = 02-,in which a partial pressure
of oxygen in the atmosphere supplies oxygen molecules which are ionized at the
oxide-interface to form anions.
Mass transport may limit the one-dimensional process in a number of ways.
Transport of electrons (or holes) across the growing film is generally rate limiting
1.2 Oxidation and Oxide Formation 29 1
in the passivation reaction occurring during the formation of stoichiometric,

nonconducting oxides (aluminum or titanium, for example). Electron and hole
transport often depends on quantum tunneling, leading to a logarithmic growth
law and the formation of the uniform, passivating film. In semiconducting, non-
stoichiometric oxides it is the diffusion of cations or anions through the growing
oxide layer that limits the growth. Ionic diffusion generally depends on the presence
of point defects: cation or anion vacancies or interstitials, and is a thermally
activated solid-state diffusion process. At constant temperature, diffusion limited
growth generally follows a parabolic growth law, and the oxide layer offers only
limited protection to continued oxidation of the metal.
In two well-recognized cases the oxidation reaction proceeds at a constant rate,
unimpeded by the reaction product. Thus, molybdenum at high temperatures
forms a volatile oxide which leaves the metal unprotected against continued
attack, while copper at elevated temperatures forms an oxide layer in residual ten-
sion which cracks and peels to expose fresh metal to attack. In such cases the rate of
reaction is only limited by the supply of reactants to the two interfaces, the sites of
the anodic and cathodic reactions. In general, the supply of metal atoms for the
anodic reaction is not a problem, and it is the supply of oxygen (the partial pressure
of oxygen in the gas phase) which is the rate-limiting step. However, neither the
anodic nor the cathodic reactions need be quite as simple as represented here, and
the rate-limiting reaction step in the formation of a steadily growing oxide may not
be obvious.
In the DMO process, as we shall see, the metal is supplied to the reaction front by
capillary flow through a network of interconnected channels and growth occurs
entirely at the oxide-atmosphere interface, outwards from the oxide product.
DMO is therefore one oxidation process in which the supply of metal for the
anodic reaction may indeed be rate controlling. However, the processes occurring
at the oxide-atmosphere interface have been found to be rather complex, and
involve both the formation of a metastable oxide film and a wetting reaction
between the oxide product and the molten metal emerging from the capillary
channels. Clearly, neither the multiple cracking and fragmentation of an oxide
product, to allow free access of oxidant to the metal surface, nor the percolation
of liquid metal through a network of capillary channels in the product, as occurs
in DMO, are particularly suited to process modeling based on one-dimensional
mass transport.
1.2.1 Initial Oxidation

Some of the voluminous literature on the oxidation and corrosion of aluminum and
its alloys has a direct bearing on DMO. Pure aluminum is normally covered by an
amorphous native oxide film which is partially converted to y-alumina at the inter-
face between the parent metal and the amorphous oxide when heated to ~ ~ 4 5 0 in°C
dry air [7-91. This unusual behavior is explained by growth of the amorphous phase,
through outward cation migration, while thickening of the y-alumina is by epitaxial
growth on the parent metal, controlled by inward oxygen diffusion. Termination
292 I Directed Metal Oxidation
of the oxidation process at elevated temperatures has been attributed [9] to the
build-up of biaxial residual stresses which inhibit further diffusion. The presence
of moisture initially increases the rate of oxidation due to the development of hydro-
gen blisters. Only when the metal is in the liquid state (700-1 100°C) does a-alumina
form directly [lo-121. The influence of magnesium additions has been summarized
by Lea and Ball [13] and Field [14]. The first authors [13] showed that y-alumina for-
mation is accelerated at the Al-amorphous alumina interface, as expected since
magnesium cations are incorporated in the spinel structure.
Subsequently, magnesium diffuses rapidly to the surface along grain boundaries
to form MgO, with concurrent crystallization of the amorphous phase to y-alumina.
Finally, the MgO layer becomes continuous so that the rapidly diffusing magnesium
no longer has direct access to external oxygen and reacts instead with y-alumina to
form MgA1204, releasing aluminum metal particles which are embedded in the
spinel. By contrast, Field [14] detected no spinel but rather direct oxidation of
magnesium to primary MgO below a cracked, amorphous surface oxide layer.
This was followed by secondary precipitation of MgO within the amorphous
alumina, which was ascribed to reduction of alumina by magnesium.
The formation of MgO on alloys containing as little as 0.5% Mg [15] at =40O0C
has been confirmed, while spinel and reduced, aluminum metal-rich particles are
frequently observed in the oxide film [16]. Additions of copper and silicon inhibit
crystallization [ 171, possibly by keeping magnesium in solution. Strong segregation
of magnesium at grain boundaries [18-201 and surfaces [20] was cited as the
probable cause for formation of MgO in dilute alloys [18,19,21]. In situ microscopy
[22] has revealed MgO nucleating at grain boundaries in Al-Mg-Zn, suggesting that
segregation could stabilize MgO and rapid grain boundary diffusion could maintain
a flux of magnesium for further growth.
Experiments on liquid alloys of A1-Mg [23] in the range 650-900°C have
demonstrated that a duplex layer of MgO and spinel is formed initially, the
former by direct oxidation at the free surface and the latter by reaction of MgO
with the liquid alloy. The proportion of MgO in the duplex layer peaked (after
2000 s at 800°C) and was gradually replaced by MgA1204,with reducing magnesium
content in the alloy. Measurements of the strength of the oxide film have implied [24]
that additions of Ca, Li, and Na initially increase the oxidation rate, while Be, in
quantities as small as 0.1 YO,considerably reduces the rate of oxidation. Zinc
additions to A1-Mg alloys [25] were found to greatly increase the oxidation rate
at very low concentrations (0.2%). In one elegant study of diffusion couples between
Al-Mg and alumina/spinel/magnesia [26] it was found that reduction of spinel to
alumina by aluminum was slow, while an A1-3.6Mg alloy readily reacted with
MgO and alumina to form spinel. In particular, silicon contamination was
found to promote alloy wetting of the oxide and to render the interfacial oxide
less protective.
The relevance of these results for the DMO process can be summarized by three
points: The importance of MgO as an ubiquitous surface layer that promotes rapid
oxidations; the importance of limiting the amount of magnesium in the alloy so that
the oxide layer, through which the melt migrates, is not too thick; and the effects of
Si, Zn, and Li in promoting the DMO process.
1.3 Related Ceramic Processing Routes 293
1.3 Related Ceramic Processing Routes

The production of ceramic components has traditionally been by powder-
processing, in which the conversion of the porous ‘green’ body into the final,
sintered product has been accompanied by a large volume change, typically of the
order of 40%. DMO is just one possible production route for ceramic products
which promises substantial savings by eliminating this sintering shrinkage and
ensuring that the initial product approximates the final, specified geometry. That
is, DMO is a ‘near-net-shape’ process. The casting of concrete and the blowing of
glass into a mold are also ‘near-net-shape’ processes, but in the case of structural
ceramics the mechanical properties of engineering interest are well beyond the
values achievable in a cement reaction or in a glass.
In RBSN, a silicon metal powder preform reacts with nitrogen gas to form silicon
nitride. The reaction occurs by percolation of the gas phase into the open porosity of
the silicon preform, while the large specific volume of the silicon nitride product
substantially reduces the volume fraction of residual porosity (down to a lower
limit of approximately YO). The mechanical properties of RBSN are inferior to
those of sintered silicon nitride powder products, but the reduced costs of machining
the near-net-shape product is a very attractive advantage.
RBAO involves a similar production route, but in this case the aluminum metal
powder is a minor component in a powder mix that contains a substantial fraction
of alumina. Oxidation of the aluminum metal by gas percolation yields a metastable
oxide phase, which is replaced at higher temperatures by the stable corundum phase.
Again, the large specific volume of the oxide product reduces the volume fraction of
porosity, but in this case the thermal cycle actually ‘sinters’ the product, and for the
optimized powder formulation, expansion during oxidation of the metal at a mod-
erate temperature is followed by the sintering contraction at a higher temperature to
yield a dense alumina product whose final shape and volume is still very close to that
of the ‘green’ powder preform.
A similar reaction route has been developed for reaction bonded mullite (RBM,
aluminum silicate). In this case the powder formulation also includes a source of
silicon (silicon metal, silicon carbide, or silica), in addition to aluminum metal
and alumina, and the final product contains mullite as the major phase. The
RBM process is frequently modified by additions of zircon (ZrSi04), which provide
an additional source of silicon and result in a dispersion of fine zirconia particles in
the product which restricts grain growth to ensure a fine uniform microstructure. It
is again possible to reduce residual porosity while retaining the near-net-shape char-
acteristics, but at the time of writing this process is still under development and some
way from commercial exploitation.
A rather different approach has been developed for reaction sintered silicon
carbide (RSSC), first developed in the former Soviet Union. In this process a
powder preform of mixed graphite and silicon carbide is immersed in a liquid
bath of molten silicon. The silicon wets and infiltrates the preform, reacting with
the finely-divided graphitic component (carbon black). In the best case, all graphite
is reacted and the residual silicon content is no more than a few percent. The product
of the reaction, silicon carbide, firmly bonds the silicon carbide preform powder
with effectively zero residual porosity. This process (patented as the REFEL process
in the western world) has been somewhat neglected in recent years, and the heat-
treatments developed to tailor high performance, high toughness microstructures
in Sic have yet to be applied to RSSC.
Infiltration of alumina preforms by aluminum alloy melts has also been achieved,
both by applying external pressure (in the absence of adequate wetting and capillary
absorption), and by ensuring the presence of wetting aids in the melt (most notably
magnesium, as in the DMO process). This infiltration processing route has not yet
been exploited commercially. Reactive wetting by an aluminum alloy melt is also the
basis of the process developed for boron carbide and marketed by the Dow
Corporation. In this case a B4C powder preform is reacted with an aluminum
melt at elevated temperatures. The molten aluminum is sucked into the pore
channels in the preform by capillary forces and reacts to form a range of mixed
aluminum borocarbides. The final ceramic product is essentially pore-free, but
contains a network of residual aluminum alloy. Since the ceramic phases are contin-
uous, the product is hard, but the residual aluminum phase is also interconnected
(while the volume fraction of residual alloy is to some extent controllable) so that
crack propagation is significantly inhibited. As we shall see below, both the micro-
structure and the mechanical behavior of DMO products are similar to those
obtained by reactive infiltration.
The combination of infiltration and reaction that characterizes DMO has been
exploited to make a number of composites. As long ago as 1953, it was shown
that silica containing refractories were reduced by molten aluminum to form
alumina and silicon [4]. Subsequently [27], the displacement reaction was extended
to the formation of composites of alumina with residual Al-Si. More recently, the
A1-Si02 displacement reaction has been used in the infiltration of dense preforms of
silica [28] and mullite [29,30] by molten aluminum. Extension of the reactive
infiltration process to porous silica-containing preforms [31,321 has resulted in
the fabrication of metal-matrix composites in which the silica was replaced by a
mixture of about 65% alumina and 35% metal, while the pores were infiltrated
by molten alloy. In contrast to DMO, the displacement reaction appears to proceed
at a critical temperature of X I 100-1200°C and without the need for a volatile solute
element or oxygen. Borosilicate glass has also been used as an initiator to enable the
infiltration of Al-Si alloys into alumina preforms [33].
Closer in concept to the DMO process is the infiltration of aluminum alloys in
nitrogen to yield A1-A1N composites. Low temperatures (< 1000°C) and high
magnesium or strontium content promote the spontaneous infiltration of liquid
metal with a small concurrent nitridation to yield dispersions of A1N [34]. Similar
alloys may be infiltrated at higher temperatures, resulting in nitride contents that
increase with temperature to yield A1N-matrix composites [35-381. This process
can result in particulate loadings of up to 75%. The mechanical properties of
these aluminum nitride composites have been extensively characterized [39].
A further extension of the DMO process has been in the fabrication of silicon
nitride and titanium nitride matrix composites by nitridation of the appropriate
molten alloys [6,40], but it is unclear whether these processes will have commercial
applications. Other examples of reactive infiltration include aluminum into Si3N4to
1.4 Directed Metal Oxidation Incubation 295
yield an A1N-Si3N4-Al-Si composite [41] and zirconium into B4C to yield ZrB2-
ZrC-Zr [4244].
1.4 Directed Metal Oxidation Incubation

Incubation is the first stage in the directed metal oxidation of a liquid metal,
preceding the actual growth process, and corresponds to a well-defined initiation
process. In the case of Al2O3-A1 composites, initiation is frequently delayed for a
considerable period, up to several hours, and depends on the alloy composition,
the heating rate, and the final temperature [45-511. Gross variations in this incuba-
tion period lead to lack of reproducibility in the thickness of composite obtained in a
given time. It has sometimes been concluded that melt oxidation was absent when
the duration of the experiment was shorter than required for the transition from
incubation to growth (e.g. [52]). Incubation can be avoided by seeding the melt
surface [45], but the development of surface oxide during the incubation period pro-
vides an understanding of the transition from low temperature behavior to the high
temperature growth regime and sheds some light on the phenomenon of oscillatory
growth rates, which commonly accompany formation of the aluminum/alumina
composite.
While incubation is found in nearly all DMO alloy systems, the most detailed
work has been on A1-Mg and Al-Mg-Si alloys. A typical isothermal ternary
section and PO2-composition diagram are shown for A1-Mg-0 in Figs 1 and 2.
+
The alloy composition corners for three-phase equilibria (alloy alumina spinel +
+ +
and alloy spinel magnesia) shift to higher magnesium content with increasing
temperature, while at a fixed temperature, magnesium depletion of the melt reduces
the stability of MgO to form first MgA1204 and then A1203.The PO2 dependence
shows that decreasing magnesium levels in the melt increase the equilibrium oxygen
activity of the melt in equilibrium with spinel. When heated in air, the oxide layer
formed on the melt at low temperatures is MgO. As the temperature increases,
the stable oxide in equilibrium with the melt changes from magnesia to spinel
[51]. For a typical magnesium content of 3%, the critical temperature for this
transition is about 750°C. Thermogravimetric analysis (TGA) at a constant heating
rate reveals a characteristic abrupt rise in the oxidation rate when spinel is formed at
the interface between the alloy and the surface film of MgO. The temperature of the
peak in oxidation rate increases with heating rate, confirming the importance of
kinetic factors.
Two mechanisms have been suggested for the formation of spinel: Direct
oxidation of the alloy melt through cracks in the MgO film [48,49]; or an interfacial
displacement reaction between aluminum and MgO, followed by solid state
diffusion of magnesium through the oxide to repeat the oxidation to MgO at the
surface [47,5 I].
Evidence for the latter mechanism includes the smaller molar volume of oxygen
ions in spinel when compared to that in magnesia, so that for a given quantity of
296 1 Directed Metal Oxidution
T=l100"C
A'2°3
\
\
\ iquid)
4 4 6 (0.40wt.%)
~ ~ ~ 0 . at.'IoMg
Figure 1. Ternary isothermal section for the A1-Mg-0 system at 1100°C [47].
T = 1100°C
.23
I I I I I I I I I 1-36
0.2 0.4 0.6 0.8
Al Mg
Figure 2. Stability fields for the alloy and corresponding oxides as a function of oxygen partial
pressure at 1100°C in the system A1-Mg-0 [47].
I .4 Directed Metal Oxidation Incubation 297
I
I
‘I
Composite Growth (Oscillatory)
Time
Figure 3. Schematic of weight gain (TG) and corresponding rate of weight gain (DTG) curves for
oxidation of a binary A1-Mg alloy in the temperature range 1050-1250°C[51].
oxygen, the volume of spinel formed should be less than the equivalent volume of
periclase, suggesting that the remaining volume could persist as liquid metal-filled
channels in the spinel which maintain contact with the MgO. It has also been
shown that mechanical disruption of the surface [48] can induce preferential for-
mation of spinel, although this does not appear to be a prerequisite for the DMO
process [5 11. A classical incubation period, characterized by extremely low rates
of weight gain, is then observed as the temperature increases (Fig. 3). The rate of
spinel formation drops as the spinel zone develops a duplex structure characterized
by a dense, metal-free layer at the interface with MgO but containing microchannels
of alloy near the melt [48,51]. It has been established that the true external surface is
composed of MgO, by using low voltage X-ray spectroscopy in the scanning elec-
tron microscope [51], and this has been attributed to de-mixing of cations in
spinel in the presence of an oxygen potential gradient [53]. These microstructural
changes have been attributed to the diffusion-limited supply of magnesium from
the melt reservoir which prevents the continued formation of MgO. Direct oxidation
of the alloy melt then leads to sealing of the metal channels by spinel, preventing
contact between the melt and the surface MgO.
It is worth emphasizing the change in the oxidation maxima, and the subsequent
drop in oxidation rate between the ‘low’ and ‘high’ temperature regimes. In the low
temperature regime, below 950°C for A1-3 Mg, magnesium is completely depleted
from the alloy reservoir and the oxidation product shifts from MgO to spinel,
though never to alumina [23,51]. In the high temperature regime, the incubation
period is characterized by bulk magnesium levels that are only marginally reduced,
together with a duplex surface oxide as described above. The temperature for the
onset of the spinel ‘peak’ increases with both magnesium content and heating rate
[46,51], as would be expected from both the phase equilibria and the activation
energy for spinel formation.
The end of the incubation period [49,5 11 is marked by a reversal of the phenom-
ena that led to the low incubation rate: Alloy channels that were isolated from the
surface, reconnect and advance through the spinel layer. The mechanism for regen-
eration of the liquid metal channels seems to be associated with an increase in melt
oxygen activity in equilibrium with spinel when the magnesium concentration
decreases. Hence magnesium depletion at the surface, due to the rapid formation
of MgO, should promote spinel dissolution and promote capillary wicking of
molten alloy through the spinel to the free surface. This hypothesis is supported
by the reduced incubation time for ternary alloys that contain silicon, since silicon
is known to reduce the activity of magnesium in the melt and should promote spinel
dissolution [54]. Cracking of the surface oxide film could also lead to direct exposure
of the melt to the gas phase [48], but this does not appear to be a prerequisite for
terminating the incubation stage. A simple thermodynamic criterion for the
molten alloy to repenetrate the channels is that the liquid-spinel interfacial energy
should be less than one half the grain-boundary energy, and this criterion may only
be satisfied above some minimum temperature [ 1,491. This could explain why alloy
repenetration does not begin at a temperature of 900°C for an A1-3Mg alloy,
although this temperature is past the peak rate of spinel formation and well into
the incubation regime.
When the alloy in the channels penetrates to the spinel-magnesia interface near
the free surface, their composition should be in equilibrium with spinel, leading to
the formation of spinel nodules [49,51,55]. Spinel-alloy composite growth should
then continue until limited by further magnesium solute depletion to below the
+
concentration defined by the alloy + spinel alumina ternary equilibrium. At this
juncture, the spinel nodules should be replaced by alumina nodules, and the alloy
composition in the advancing composite should be given by that determined by
the spinel-alumina buffer layer at the original melt surface. The extent of spinel
growth which occurs before the A1203-A1 composite begins to form increases
with the magnesium content of the alloy and, for a fixed magnesium composition,
decreases as the temperature increases, as predicted from the ternary Al-Mg-0
equilibrium. For example, an A1-8 Mg alloy has been observed to grow a spinel-
based composite to a thickness of hundreds of microns at 1100°C until the alloy
magnesium concentration reaches =0.5%, at which point alumina is the stable
oxide. In contrast, an A1-3 Mg alloy was depleted to a similar critical magnesium
concentration after ~ 1 0 pm0 of spinel growth [51].
The oxide at the external surface of the composite is predominantly a thin
(0.1 pm) layer of MgO with an intermittent, underlying spinel layer: The microstruc-
ture is discussed below. It could be argued that the entire process of incubation and
reinitiation is caused by excess magnesium, and that a starting composition near the
three-phase equilibrium should trigger immediate growth of alumina. However,
controlled growth of a DMO composite has never been reported from a bulk
1.4 Directed Metal Oxidation Incubation 299
binary Al-Mg alloy of the predicted composition, possibly due to additional

depletion of solute in the temperature gradient of the furnace during the initial heat-
ing period. In binary Al-Zn alloys, alumina composite growth has been initiated,
without the formation of a surface spinel layer, by immersing a magnesium spinel
seed in the melt [56]. Partial dissolution of the spinel provides the required concen-
tration of magnesium in the alloy. Similar growth could not be achieved in the
absence of zinc. Even in the case of Al-Zn it appears that incubation is not entirely
avoided, although the incubation time may now reflect the time taken for the seed
spinel to dissolve. Binary Al-Zn alloys containing no magnesium also have an
incubation period for DMO [57], even though thermodynamic criteria do not
predict any oxide other than alumina. In this case the delay in the onset of
growth could be associated with delayed rupture of the surface oxide by zinc
vapor. The duration of incubation is determined partly by the heating procedure.
Anomalously long incubation times are encountered at intermediate temperatures
(>70 000 s at 1 100°C) when the alloy is heated in argon [46]. Subsequent introduc-
tion of oxygen at the final processing temperature leads to reaction with magnesium
vapor to form a thick layer of MgO, which then has to be reduced and penetrated by
the underlying molten alloy. Because the magnesium vapor pressure increases
rapidly with temperature in the range 950-1200°C, the advantage of rapid reduction
could be outweighed by the greater thickness of MgO formed [46].
Consistent elimination of the incubation stage can be achieved by seeding the melt
surface with silica [45]. While bulk silicon appears to reduce the incubation time
marginally, the effect of a surface layer of S i 0 2 is much more pronounced, probably
serving as an additional source of oxygen during the spinel growth peak. The large
molar volume contraction which accompanies the reduction of silica and the for-
mation of spinel helps to form alloy channels that are enriched in silicon and are
associated with residual seed particles [51]. If the seed particle is large enough, the
magnesium concentration will be depleted to the point at which alumina is
stabilized, even before the particle is penetrated by melt channels. If the seed particle
is too small, spinel nodules will grow from the silica. In either case, the size of the
silica grains determines the time for the persistence of a silicon-rich alloy in the chan-
nels, which in turn determines the transient growth rate of spinel and the subsequent
reduction of magnesium activity. The silica seed particles are converted initially to
the binary spinel-alloy, and finally to the alumina-alloy composite, at a rate which
is determined by the A1-SiO2 displacement reaction. The passivating effect of spinel
and the ensuing solid state growth of MgA1204-Mg0 are then bypassed.
Other seed materials have been shown to eliminate the incubation stage, speci-
fically MgO [50] and NaOH [58]. In the case of seeding by MgO, an MgO-coated
Al-alloy gave immediate DMO alumina growth, although the extent of composite
formation was limited by the quantity of seed material. NaOH seeding gave the
same immediate onset of DMO growth on pure A1 with limited composite
growth. In both cases, it would appear that the seed particles act as sources of
oxygen and facilitate the formation of a surface aluminate by a displacement
reaction which is interspersed with alloy channels. Both the MgO and the NaOH
seed particles also act as a source of the volatile constituent that is essential for
growth of the DMO composite.
1.5 Directed Metal Oxidation Growth

1.5.1 Introduction
The formation and subsequent breakdown of the spinel layer which occurs during
the incubation stage is followed by the growth of A1203-Al composite. This is a
complex phenomenon whose details are still not fully understood. Incremental
advance of the growing composite involves at least three steps.
(i) Transfer of oxygen from the atmosphere to the condensed phases.
(ii) Reaction between A1 and 0 to form A1203.
(iii) Flow of molten alloy from the reservoir toward the surface.
In addition to these basic processes, the infiltration of preforms by the oxidation of a
complex alloy in an oxidizing atmosphere (air) must involve diffusion of the non-
oxidizing solute elements (and nitrogen) away from the interface, as well as
wetting of the preform by the molten alloy and adaptation of the growth process
to the tortuosity of the preform pore space. Early attempts to model DMO
[l, 45,591 assumed a steady state process controlled by ionic transport across a
surface oxide layer. Subsequent evidence demonstrated that the surface chemistry
was more complicated and that growth was cyclic, far from equilibrium, and self-
limiting. Most of the systems that have been studied were based on A1-Mg, but
DMO composites have also been grown from Al-Zn, Al-Na, and Al-Li alloys,
implying that the microstructures produced and the underlying growth mechanisms
should be similar.
The growth of an A1203-A1 composite from a binary Al-Mg alloy into air will
serve as a model system that avoids the complications of multiple solute elements
but retains the essential features of oxidative growth.
1.5.2 Directed Metal Oxidation Composites from

AI-Mg Alloys
An Al,03-A1 composite growing outward from the original molten Al-Mg alloy
surface into the atmosphere contains two fully interconnected, continuous phases,
the ceramic oxide and the molten metal. The ceramic volume fraction is normally
in the range 7&85% and the temperature of formation is in the range 1000-
1300°C [ 1,601. Higher growth temperatures lead to porosity and premature ter-
mination of the growth process. The contiguity of the metallic phase is a condition
for the supply of molten metal from the alloy reservoir to the advancing growth
interface, while the presence of an interconnected ceramic skeleton is inferred
from the crystalline texture in the alumina, which consists of columnar ‘grains’,
several tens of microns wide, percolated throughout the thickness of the composite
and with occasional low angle boundaries [47,60,61]. Elastic moduli measurements,
discussed later [60], confirm the high degree of connectivity of the stiffer oxide phase.
The mean intercept lengths of the ceramic and alloy channels, measured perpen-
dicular to the growth direction, are about 5 pm and 2 pm, respectively. These
intercept lengths are not related to the grain sizes, and both phases are mono-
crystalline over distances hundreds of times larger than the measured intercept
lengths. In effect, the liquid metal in the composite develops a grain size upon
solidification which is similar to that seen in a slowly cooled, cast aluminum alloy.
A typical growth history is shown schematically in Fig. 3. Following the
incubation period (when present) the growth rate rises to a maximum. During
this period, the oxide nodules nucleated during incubation (initially spinel and
subsequently alumina) grow radially from their point of nucleation, coalesce and
cover the surface of the melt. Within each nodule the orientation of the alumina
crystal lattice is approximately constant, with the c-axis aligned along the
growth direction [61]. The alumina grain size perpendicular to the growth direction
is therefore determined by the nodule spacing. As the nodules coalesce to form a
planar growth front the growth rate begins to drop monotonically with time
and (equivalently) composite thickness [46,62]. Occasionally, at high temperatures
or with small samples, growth acceleration has been observed and correlated,
through differential thermal analysis, with adiabatic heating due to the exothermic
nature of the oxidation reaction [46]. Frequently, oscillations in the growth rate
are superimposed upon the average, decreasing rate [47,63] and have been
associated with the passivation of old nodules and nucleation of new growth
centers.
Microstructural variations that accompany growth are of two kinds. Firstly, the
scale of the microstructure is finer near the oxidation front and coarser at the
original initiating surface [64,65], corresponding to the nucleation of new nodules
during the growth process. Secondly, oscillations in the growth rate are often
reflected in microstructural banding in which the residual metal content of the
layers differs slightly. At higher growth temperatures, above 1300°C for A1-3 Mg,
growth may cease abruptly after 1-2 mm of composite has formed.
The incubation period is terminated by re-penetration of molten alloy through the
passivating spinel layer and the nucleation of spinel nodules. Selective oxidation
of the volatile magnesium leads to depletion of the solute at the surface in order
+ +
to reach the three-phase equilibrium: (Al, Mg) alumina spinel. Alumina then
becomes the preferred oxide product. The top (1-2pm) of the surface at the
growth front has a complex chemistry whose details are still a matter for debate.
In general terms, the surface consists of a layer of MgO separated from the growing
composite by a thin wetting layer of molten alloy. This model borrows schematically
(Fig. 4) from the microstructure found in liquid-solid reactive diffusion couples,
such as Al-MgO or A1-Si02, in which the alloy wets its own oxidation product
and penetrates the solid-solid interface, which in the present case is A1203-Mg0.
Such microstructural observations were the basis for the steady state growth
hypothesis [45,48,59],which assumed anionic oxygen transport through the MgO
layer, controlled by either electronic or cationic vacancy defects, followed by
dissolution of MgO into the molten alloy, diffusion of dissolved oxygen through
the thin intervening alloy layer and, finally, an oxidation reaction to form alumina
at the interface with the preformed, columnar alumina crystals. As indicated in
Figure 4. Schematic of surface microstructure during DMO growth. A thin alloy film is formed
between a surface MgO layer and the growing alumina skeleton and is continuous with underlying
alloy channels. The reactions at A, B, and C are described in the text.
Fig. 4, the following reactions have been proposed.

$02(gas'-+ 3 0 z + 3 V k g + 6 h ) (1)
3 Vh, + 6 h' + 3 0; -+3 0 (liquid) (2)
3 0 (liquid) + 2 A1 (liquid) +. A1203 (3)
where Vhg indicates a doubly charged cation vacancy, h is a positive hole and 0;
refers to an oxygen ion occupying its own site.
The above mechanism was consistent with several experimental observations.
MgO was known to be preferentially formed on the surface of Al-Mg alloys
under many circumstances, as described earlier, either because of cation de-
mixing in spinel, to form periclase, or due to rapid diffusion of Mg in the solid
state and subsequent surface segregation. The 'single crystal' oxidation product
clearly precluded any mechanism that involved repeated nucleation of corundum
(alumina) grains, which in any event is known to be difficult at temperatures
below 1200°C. Finally, the presence of a film of molten alloy at a distance of
about 1 pm from the surface accounted for the relative absence of a growth rate
dependence on composite thickness (a rather slow decrease is actually observed).
The steady-state microstructure described above is a simplification. The MgO
layer is far from uniform, both with respect to position and time [50,56,66], and
thicknesses: <0.1 pm has occasionally been observed [66]. In some samples the
interface between aluminum and MgO was found to contain spinel [47], while the
solidified metallic phase frequently contained nanocrystals of MgO [66]. Even less
consistent with the steady state hypothesis was the apparent 'equilibrium' between
aluminum and MgO in the presence of a magnesium concentration that was
necessarily, and experimentally, close to the three-phase equilibrium with spinel
and alumina [50]. Cation de-mixing of spinel, as a source of MgO (periclase),
seems unlikely when spinel is the minor phase or apparently absent. An alternative
hypothesis assumes that oxidation of magnesium in the vapor phase is responsible
for the presence of MgO, an assumption that is supported by the demonstration of
DMO composite growth from aluminum alloys that contained no Mg but other,
alternative volatile elements (Na, Li, Zn). In each case DMO growth occurs at
temperatures for which the vapor pressure of the volatile solute is an appreciable
fraction of the partial pressure of oxygen in the gas phase.
The mechanism of oxygen transfer across the surface oxide layer is also problem-
atic. Observed growth rates require that diffusion occur through the surface oxide
at rates far greater than those measured for the fastest diffusing species (cation
vacancies) in polycrystalline MgO [66].In addition, the decrease in growth rate
with increasing thickness implies that viscous flow of liquid through the capillary
channels of the composite might also be rate determining. The results of interrupted
experiments [63] suggest that microstructural banding might be related to the
formation of alumina by alternate dissolution of both spinel and magnesia into
the molten alloy. The volume fraction of the alumina phase in these experiments
also corresponded approximately to that expected from a comparison of the
molar volume of oxygen in the alumina-spinel and alumina-magnesia phase
pairs. This hypothesis also predicts correctly the volume fraction of spinel formed
from MgO during the spinel growth peak in the incubation stage, as well as that
of alumina formed from a silica nucleant.
Alternating bands of magnesia and spinel at the growth interface with the molten
alloy could be associated with periodic fluctuations in the magnesium concentration
of the melt. Calculation of the partial pressure of oxygen (PO,) in equilibrium with
different alloy-oxide combinations suggests that a low magnesium content in the
melt maximizes the driving force for the formation of alumina by dissolution of
MgO, while the resulting enrichment of magnesium then favors epitaxial spinel
formation on the MgO crystals [63].Subsequent growth of alumina then occurs
by dissolution of spinel, leading to a decrease in the magnesium concentration by
diffusion into the bulk, and the cycle is then repeated. Isolation of molten metal
from the MgO decreases the rate of magnesium oxidation and hence the rate of
weight gain during this period of spinel dissolution. Subsequent re-penetration of
spinel by the molten alloy once more initiates direct dissolution of MgO. The
repeated formation of MgO required by the above sequence is most likely due to
the oxidation of magnesium in the vapor phase, although cation de-mixing may
also be possible if the spinel layer is sufficiently thick.
In summary, composite formation by DMO of Al-Mg alloys is considered to be
made possible by the following sequence of events.
0) A buffer layer of spinel at the original melt surface ensures a supply of molten
alloy, containing =0.2-0.5% of magnesium, to the free surface, close to the
three-phase equilibrium composition of the alloy with spinel and alumina.
(ii) The high vapor pressure of magnesium leads to the formation of MgO at the
surface of the composite.
(iii) A displacement reaction with aluminum leads to the formation of alumina, not
by direct nucleation but rather by transport of dissolved oxygen to the Al-
A1203 interface. This reaction is periodically mediated by spinel, due to local
variations in the magnesium activity that are not immediately compensated
by bulk diffusion from the melt reservoir.
The high-temperature limit for oxidative growth has been attributed to excessive
loss of magnesium vaporizing from the surface [50,62,65]. Unless the magnesium
can be replenished by diffusion through the liquid, a passivating layer of alumina
or spinel will form. In principle, providing the bulk reservoir contains sufficient
magnesium, oxidative growth is expected to resume eventually, the times involved
being comparable to the incubation period. Related experiments [65] have shown
that sufficient magnesium may be available, either in the alloy channels of a
previously grown composite or in the spinel layer at the interface with the melt,
to continue DMO growth when placed in contact with a pure aluminum melt.
This demonstrates the very low levels of magnesium actually necessary for com-
posite formation, but it is likely that, in the absence of a magnesium reservoir,
growth will eventually cease due to magnesium losses by spinel precipitation or
magnesium evaporation.
1.5.3 Directed Metal Oxidation Growth from other

Aluminum Alloys
Several other aluminum alloy systems which have added to an understanding of the
DMO growth process are based on additions of magnesium, silicon, zinc, lithium,
and sodium. Magnesium need not be an actual alloy constituent but may be
added by doping the surface of a pure aluminum melt with MgO powder [50].
This combination eliminated the incubation period and permitted composite
growth to a thickness determined by the amount of the original dopant. The elemen-
tal magnesium released by reduction of MgO remained in the vicinity of the surface
alloy channels long enough to allow oxidative growth to occur well beyond the
thickness of the oxide dopant layer. The MgO particles were therefore assumed to
migrate with the oxidation front. A suitable magnesium dopant concentration can
also be generated by immersing spinel in the alloy reservoir [56] or by doping the
preform with MgO-Mg [67]. Additional binary aluminum alloys which have been
oxidized by the DMO process include Al-Zn [57], aluminum doped with NaOH
[58], aluminum doped with sodium from p-alumina [68], A1-Li and Al-Na [66],
and aluminum doped with lithium derived from lithium carbonate [69]. In all
cases the growth of the composite was promoted by the presence of an oxide
which contained a potentially volatile cation (ZnO, LiAlsOs, NaA102 etc.) reducable
to aluminum through a displacement reaction. In the case of sodium, the cessation
of growth after only limited composite formation was attributed to evaporative loss
of the solute. A similar observation with zinc as the dopant was probably due to
localized surface depletion, reducing the activity of zinc below that needed to
retain a surface layer of ZnO. Al-Li alloys seem to be capable of sustained compo-
site growth but have not been studied extensively, presumably due to the cost and
difficulty of lithium doping. Al-Zn alloys with minor amounts of magnesium [57]
invariably lead to more sustained growth compared to the binary alloys, and it
has been argued that trace levels of magnesium ensure that, in the event of tem-
porary zinc depletion, the surface remains covered with MgO-MgA1204 rather
1 .SDirected Metal Oxidation Growth 305
than alumina. During growth in this system, however, the principal surface oxide
appears to be ZnO [56,57], although it should be noted that the two cations
magnesium and zinc readily substitute for each other in the periclase lattice. When
magnesium-rich alloys are used, the surface oxide has been found to revert to
MgO, as found in binary Al-Mg [65]. Additions of silicon improve the uniformity
of growth in Al-Mg [48] and Al-Zn-Mg alloys [56], even though binary Al-Si
alloys cannot initiate composite formation. For example, binary Al-Mg alloys may
nucleate only a few nodules per square centimeter, which grow irregularly outwards
from the isolated nucleation sites and never develop a planar growth front. Silicon
additions greatly increase the surface density of nodules and lead to a very much
more uniform and smooth growth front. Improved wetting by reduction of the
alloy-oxide interfacial energy has been suggested as the primary reason for more
regular growth, although prolonged incubation and microstructural banding, due
to periodic magnesium depletion, are also observed in Al-Mg-Si alloys.
It has already been suggested that oscillations in the growth rate are probably
linked to the periodic surface depletion of magnesium. The observation that
DMO composites grow initially with a uniform microstructure and a smoothly
varying growth rate, before the onset of the growth instability [62,63], is consistent
with the hypothesis that the instability occurs when the distance from the growth
surface to the magnesium reservoir in the alloy melt is sufficient for solute diffusion
through the liquid channels to become rate limiting. Furthermore, uniform micro-
structures are commonly found in ternary alloys which contain additions such as
silicon, palladium, or indium [70], and for these solutes thermodynamic calculations
have shown that the stability of alumina is extended to higher magnesium con-
centrations. This improved alumina stability should make spinel formation less
likely and so ensure a single alumina growth mechanism via MgO dissolution
into molten aluminum. However, an important feature of oscillatory growth that
still has no satisfactory explanation is the lateral uniformity of alloy composition
across the growth surface that should be necessary to achieve coordinated cyclic
behavior over the whole of the growth front.
1.5.4 Microstructural Scale

The volume fractions of the alumina and alloy phases present in a DMO composite
may be related to the relative molar volumes of oxygen in alumina, spinel, and
periclase, but the factors that determine the spacing of the metal alloy channels
and the ceramic ‘grain’ size (the nodule spacing) are much less apparent. Additions
of transition metals (Ni, Cu, Ti, or Zr) and some other elements, such as zinc, refine
the microstructure of composites grown from an Al-Mg-Si alloy [71-731. Iso-
thermal aging of composite samples coarsens the scale of the microstructure, as
would be expected for a classical precipitation or a spinodal system. The coarsening
mechanism has been attributed to capillarity-driven dissolution and reprecipitation
of alumina, with the liquid alloy providing the medium for oxygen transport [64]. A
similar phenomenon could also account for some coarsening of the microstructure
created early in the growth process, when compared to the microstructure formed in
the later stages. Nevertheless, there is still no clear explanation relating the micro-
structural length scale to the alloy composition and the DMO growth temperature.
It is possible to produce composites with an alumina content in excess of 85 vol-% at
high temperatures. However, this appears to be achieved either at the expense of a
reduced growth rate or with premature termination of the growth process before
complete conversion of the alloy to composite [60,74]. Excessive evaporative loss
of magnesium and sealing of alloy channels by alumina are the probable combined
causes of this high temperature behavior, which prevails above about 1300°C
(typically 1300°C for Al-Cu-Mg and 1200°C for Al-Ni-Mg [74], 1300°C for Al-
3 Mg [62], and 1400°C for Al-Zn-Mg [65] and Al-Si-Mg [46]). A further compli-
cation in complex alloys is the enrichment of nonoxidizing elements, such as silicon,
nickel, or copper, which fail to diffuse away from the interface and instead pre-
cipitate refractory intermetallic compounds [74].
A controversial area in the growth of DMO A1203-A1 are the temperature and
oxygen partial pressure dependencies of the oxidation rate. One difficulty in
reconciling experimental results obtained by different authors and assigning an
unambiguous growth mechanism has been the absence of a clear regime of constant
growth rate [65]. The available evidence suggests that the rate of weight gain starts to
fall after little more than 1-3 mm of composite growth. This decrease in growth rate
approximates a parabolic time dependence in some instances [46,75] and has been
attributed variously to enrichment of nonoxidizing elements near the surface,
depletion of magnesium in the melt, or percolation restrictions on the supply of
metal through the capillary channels [62,71].
Absence of data on melt viscosity, alloy channel tortuosity, alloy-alumina
interfacial energies, and melt-oxide contact angles precludes a meaningful analysis
of viscous flow of the melt through the percolation channels and its possible control
of the growth rate. Some thermogravimetric studies have been conducted over a
sufficient range of temperature and time to permit the determination of average
growth rates and an analysis of activation energies which are valid for a limited
range of temperature and composite thickness. Variations in alumina content
across the composite and uneven growth across the surface make it difficult to
define a linear interface mobility from gravimetric data. Measured activation
energies range from 90 kJ mol-' for a complex Al-Si-Zn-Mg-Cu-Fe alloy [45],
to 213 kJ mol-' for Al-1 Mg [76], 360 kJ mol-' for A1-5 Mg [77], and 400 kJ mol-'
for Al-Mg-Si [71], with other values typically around 270-300 kJ mol-' [46].
Attempts to relate these values to specific mechanisms, such as ionic diffusion in
MgO [59], have largely been abandoned because of the wide energy spread and,
in some cases, the absence of well-defined Arrhenius behavior [50].Similar inconsis-
tencies have been observed in the oxygen partial pressure dependence of DMO
growth. Power law dependencies of 1/4 and 1/6 for the oxygen partial pressure
were reported for Al-Mg-Si [71], but growth from Al-Si-Zn-Mg [45] was found
to be insensitive to PO2. In contrast, an exponent of unity was proposed for Al-
5Mg with 2 and 5% Si, while growth from an A1-5Mg-10% alloy was again
found to be insensitive to PO2 [78].
It would appear that the growth of the composite is inherently cyclic in the short
term, involving the formation and dissolution of oxides such as MgO and ZnO, and
is self-limiting in the long term, due to surface depletion of the volatile element,
enrichment of the melt in the channels by nonoxidizing elements, or to an
insufficient flux of molten alloy through the capillary channels.
1.5.5 Growth into Particulate Preforms

Commercial applications of melt oxidation generally require infiltration of a DMO
Al2O3-A1 composite into a shaped preform, either a particulate aggregate or a fiber
lay-up. The role of a particulate aggregate in modifying the DMO process is com-
plex. The aggregate poses a physical hindrance to the growth, forcing changes in the
growth direction and restricting the reaction between oxygen and the vaporizing
solute. When the particle size decreases to approach the alloy channel width the
growth constraints become pronounced, since the natural microstructural scale
characteristic of DMO growth in free space can no longer be maintained. Chemical
effects are also unavoidable, since the interfacial energy and contact angle between
the molten alloy and a preform particle replace those with the growing alumina
body to dominate the wetting by the melt of the percolation channels. In addition,
the depassivating MgO/ZnO phase may now form on the preform particles, while
heterogeneous nucleation of new alumina grains becomes possible. More sub-
stantial effects are to be expected if the particles react with the liquid alloy (partial
dissolution in the melt or the formation of additional phases).
Of the two most commonly used preform materials, one, alumina, is chemically
inert during DMO while the other, silicon carbide, is reactive. We first examine
the growth of a DMO composite into an alumina particulate, followed by that
into S i c preforms, and finally the DMO infiltration of a suitably coated fibrous
preform. Coatings of calcium sulfate with calcium silicate may be used to ensure
that oxidative growth ceases at the external boundary of the preform, thereby
ensuring near-net-shape fabrication. The mechanism by which these barriers
‘poison’ the oxidation without impeding the flow of oxygen is not clear at this time.
Oxidative growth of Al-Mg-Si into alumina preforms was first reported by
Nagelberg et al. [71,72]. It was subsequently observed that the crystallographic
texture of the matrix alumina was reduced in the infiltrated preform when compared
to that found after growth in free space [72]. In addition, wetting of the alumina
preform by the alloy was found several particle diameters ahead of the fully
infiltrated thickness [71]. The presence of an MgO-rich surface together with the
underlying alloy film appeared similar to that found during growth into free
space, and it was concluded that the growth mechanism remained substantially
unchanged. However, in high partial pressures of oxygen, the growth rate during
infiltration (normalized by the true surface area) was 2-3 times higher than that
into free space, but also dropped significantly as the composite thickened. This
behavior was attributed both to the wetting of alumina preform particles ahead
of the growth interface, and to the difficulty of transporting unoxidized silicon by
diffusion from the growth surface back into the melt through the tortuous alloy
channels. At low partial pressures of oxygen the growth rate was found to increase
as the composite growth interface approached the external surface of the preform.
Here, it was argued that the oxygen supply through the pores of the alumina
preform was the rate-limiting step, and that growth became progressively easier
as the alloy percolation channels approached the free surface. The oxidation rate
was also found to decrease with increase in the silicon content of the melt.
In the above work, the presence of alumina preform particles appeared to favor
infiltration but no residual metal was found between the preform alumina particles
and the matrix alumina and, surprisingly, neither was any epitaxy identified [72]. If
the particles aid infiltration, then the rate of infiltration would be expected to
increase with the particle surface area, and this was indeed the case for coarse
alumina particles DMO infiltrated using Al-Zn, Al-Zn-Mg, and Al-Mg alloys
[54,75,79,80]. A maximum in the growth rate was observed for particle sizes of
about 20-50pm [57] and 5-8pm [80]. The decrease in growth rate at smaller
particle sizes was attributed either to the difficulty of developing the surface
oxide layer (MgO/ZnO) or to a diminished supply of liquid metal owing to the
restriction in viscous flow due to the reduced size of the pore channels. Interest-
ingly, in contrast to earlier work on Al-Mg based alloys, composites grown
from Al-Zn-Mg alloys, in which copious nucleation of ZnO was observed on alu-
mina particles ahead of the main infiltration front, also exhibited rapid wetting of
the Zn0-A1203 interface, as well as epitaxy between the matrix alumina and the
particulate alumina preform. In one paper the microstructure of the preform-
infiltrated matrix has been reported to be refined when compared to that obtained
in the absence of an alumina preform, but the magnitude of the effect was not
specified [72]. The importance of wetting of the alumina preform by the liquid
alloy and of the preform-ZnO interface was demonstrated by the absence of
infiltration of Al-Zn-Mg into Zr02 [57].
Quantitative assessment of the role of pore size and particle packing fraction (the
‘green’ density of the preform) in the growth process remains elusive, largely owing
to the absence of steady-state growth over a sufficiently wide range of preform
parameters. Watari et al. [75] have reported a growth rate that was proportional
to the square root of the pore size. However, the rate of infiltration into the alumina
preform observed in this study was nonlinear and the data only appear to obey the
given pore size relationship for composites of 1-2 mm in thickness and at a single
temperature.
Silicon carbide has been a popular choice of particulate for preforms in the
DMO growth of A1203-A1 composites, promising both hardness and chemical
stability in an oxidizing atmosphere. A distinctive feature of DMO growth into
S i c preforms is the reactivity of Sic with both the alloy melt and with oxygen.
Sufficient silicon must be present in the alloy to prevent formation of aluminum
carbide during growth, since aluminum carbide is hygroscopic and leads to time-
dependent disintegration of the composite under ambient conditions [8 11. Super-
ficial oxidation of S i c leads to reactive infiltration of liquid aluminum into the
silica-coated Sic. Initial published work on the A1203-SiC-A1 formulations
showed that growth rates increased with decreasing particle size 1821, and attribu-
ted the increase to wetting of the silica layer by the melt. It was subsequently
pointed out [83] that oxidation of Sic, either during pre-sintering of the preform
or during infiltration, would be expected to increase the amount of silica available
1.5 Directed Metul Oxidation Growth 309
with decreasing Sic particle size in the preform. In effect, the volume fraction of
Sic in the composite is reduced and the silica product acts as an additional
source of oxygen that promotes reactive infiltration. This effect was found to
increase the residual alloy content of the DMO composite up to 30 vol-%, using
a 5pm particle size S i c particulate preform, as compared with 10-15 vol-% of
residual alloy with 50-300 pm S i c particles.
The reactive infiltration of a porous silica preform to yield a metal-rich A1203-A1
DMO composite appears to be similar [31]. In this case, the individual silica particles
are completely reacted and replaced by Al2O3-Al, while the pore spaces are
filled with the resulting silicon-rich alloy. Similarly, it was shown that additions of
1-20% Si02 and Sn02 to Sic preforms could increase the final alloy content of the
resulting composite [84]. In summary, optimizing the S i c preform particle size can
promote the production of DMO composites with a higher alloy content. However,
this is often at the expense of significant residual porosity in the final product,
particularly at higher temperatures (> 1000°C). Residual porosity may be avoided
by coating the Sic particles with reactive alumina (by the precipitation of aluminum
hydroxide followed by calcination). Composites made from preforms with these
alumina-coated S i c particles contain less porosity and residual metal, and have
higher hardness [85].
When excessive reactive wetting of the preform leads to metal migration over
particle surfaces ahead of the main reaction front, residual porosity becomes a
ubiquitous feature of DMO infiltration. The surface of a partially-infiltrated
region then becomes sealed before the underlying pores are filled with the advancing
composite. In the case of oxidation in air, a similar phenomenon can lead to the
entrapment of nitrogen-containing pores.
The crystallinity of alumina formed by DMO infiltration into S i c particulate pre-
forms follows the expected behavior of heterogeneous nucleation on S i c particles
and large monocrystalline regions with [00011 parallel to the growth direction
within the larger voids [83,86]. Manor et al. [82] found that the infiltration rate
increased as the Sic particle size decreased. However, no systematic trend was
found in another work [83], most probably due to widely different silicon concentra-
tions in the percolation channels resulting from reduction of the silica layer during
infiltration.
1.5.6 Growth into Fibrous Preforms

Silicon carbide fiber preforms have been successfully DMO infiltrated after being
coated in order to prevent attack by the melt and produce weak interfaces with
the DMO matrix in order to maximize pull-out during fracture. To achieve the
desired properties, a double layer coating was used in which the outer layer of
S i c provided an oxidation resistant surface on which the alumina could nucleate,
while the inner coating, probably BN, ensured a low shear strength for the interface
between the composite matrix and the reinforcing fiber [87]. Alumina fibers have
also been successfully infiltrated [88,89]. Double layers of BNjSiC and triple
layers of Zr02/A1203/Si02have been used to control the reactivity and shear
310 I Directed Metal Oxidution
strength of the fiber interface with the matrix. In the latter coating system, the role of
alumina is to reduce the thermal mismatch with the fiber, while the middle zirconia
layer both promotes fiber pull-out and protects the fiber against attack by the melt.
The outer silica layer promotes wetting by the molten metal. The alloy used in the
last example was A1-5 Mg, but a quaternary alloy, with silicon and zinc additions
and less magnesium, would probably have minimized degradation of the fiber and
reduced the incubation time, possibly obviating the need for the silica coating. An
attempt to infiltrate aluminosilicate (mullite-based) fibers resulted in extensive
attack associated with the reduction of silicon from the oxide and its dissolution
in the melt [90].
1.6 Mechanical Properties

Three distinctive features of the microstructure of DMO composites determine their
mechanical properties: the particulate or fiber loading in the preform, the presence
of a continuous ductile (and low-melting point) alloy phase, and the geometry of the
anisotropic ceramic skeleton. Not surprisingly, the transverse rupture strength (TRS
or MOR = modulus of rupture), which is sensitive to the initial flaw distribution,
varies widely with the dimensions of the reinforcement and the residual porosity.
The tribological properties, such as the resistance to adhesive wear, are more for-
giving. The thermal shock resistance can also vary widely, since this depends on
the thermal expansion coefficient, the thermal conductivity and the toughness, all
of which can be tailored by controlling the residual metal content, the nature and
volume fraction of the particulate preform, and the porosity. High temperature
creep of DMO composites is to some extent dependent on the content of the residual
alloy, which melts above about 600°C. A composite with most of the alloy phase
replaced by pores is creep resistant, similar to the behavior of chemical vapor
infiltrated (CVI) Sic-Sic composites, in which 1@15% of fine, uniformly
distributed porosity can be tolerated (at least, in the absence of a low-viscosity
grain-boundary phase). Published data on the mechanical behavior of DMO
composites is somewhat limited. The following section summarizes the different
mechanical properties and, where possible, compares the properties of the base
composite with those of a particulate-reinforced material. The fiber reinforced
composites are considered separately.

The stiffness of a DMO composite of A1203-Al with 22% alloy and 4% porosity
(231 GPa) was modeled successfully [60] by assuming the metal and the ceramic
skeleton to deform equally in series and in parallel, that is, by taking a Reuss-
Voigt average. The additional effect of isolated pores could be included by using
empirical expressions derived from data on porous alumina. The elastic modulus
1.6 Mechanical Properties 3 11
of a composite fabricated at a high temperature, yielding a higher ceramic content

with significant amounts of isolated alloy pockets (13% alloy, 5% porosity), was
also successfully modeled by using Eshelby's treatment for elastic spheres
dispersed in an elastic matrix. This latter composite had a Young's modulus of
304 GPa, illustrating the large increase in stiffness possible through a reduction in
volume fraction and connectivity of the metal phase. In both the above DMO
materials the modulus was measured by standing wave resonance, with the test
bar aligned perpendicular to the growth direction, approximating to a [uwtO]-alu-
mina orientation. The modeling of the results assumed isotropic, bulk polycrystal-
line behavior for all the phases present, ignoring textural anisotropy. Since elastic
anisotropy of the elastic constants for alumina is not very great, these assumptions
are reasonable. The major difficulty in measuring elastic moduli parallel to the
growth direction stems from the substantial infiltration times required to produce
the necessary length for the through-thickness resonating bar. Pickard et al. [91]
measured a modulus of 220 GPa for a DMO composite grown from A1-3 Mg that
contained 20% residual alloy and 10% porosity. They also reported values up to
260 GPa for composites grown from complex Al-Si-Mg-Fe-Cu alloys, and, since
there was no observable change in the proportions of alloy and ceramic, the increase
in modulus probably reflects the presence of a significant volume fraction of elasti-
cally stiff, intermetallic phases. The above values of the Young's modulus lead to
values of the specific modulus (modulus per unit weight) of the DMO composite
in the range 65-84MPacm3 g-', which may be compared with 100 (polycrystalline
alumina), 140 (dense Sic), and 25 (metals, silica).
DMO composites with particulate reinforcements of alumina and silicon carbide
were found to possess moduli of 302 and 324 GPa, respectively [92]. In the former
case, the total alumina content, including the reinforcement, was 8 1YOwhile in
the latter, the S i c volume fraction was 46% and the matrix alumina 37%. The
alloy content and porosity were in the vicinity of 15 and 2 vol-YO,respectively. A
similar value of 313 GPa was reported for a S i c reinforced DMO composite
whose composition was not disclosed [93].
1.6.2 Strength and Toughness

Most of the available data on strength and toughness have been reported for trans-
verse rupture strengths measured in three- or four-point bending, both at room and
elevated temperature. It is difficult to interpret strength data without a prior under-
standing of 'toughness'. Accordingly, we first examine the limited measurements of
the fracture toughness of these materials. Double torsion tests of DMO A1203-Al
grown from A1-3Mg have shown that K I C drops from a maximum of
9.5 MPam'/2 (74% alumina, 22% alloy) to a low of 3 MPam'/2 (73% alumina,
1% alloy) [60], with an intermediate value of 6MParn'I2 for a composite with
72% alumina and 13% alloy. Thus, the toughness, as might be expected, appears
to increase monotonically with alloy content. Flinn et al. [94], working with Lanxide
DMO composites, determined the contribution to the fracture energy from the plas-
ticity of ductile alloy ligaments. Their result, derived from the measured yield
strength of the alloy and fractographic analysis of the composite, gave an estimated
fracture energy contribution of 150-200 J mP2.This value corresponds to a tough-
ness increment of about 7MPam'/2 (assuming a Young's modulus of 250GPa)
over that for polycrystalline alumina, which is consistent with the maximum value
reported by Aghajanian and coworkers. The alloy in this case, though not specified,
was reported to contain Cu, Fe, and Si, implying a higher fraction of intermetallics
than in the Al-Mg alloy studied by Aghajanian and coworkers. Pickard et al. [91]
found R-curve behavior in four-point bending of A1203-Al composite samples
made from Al-Mg and Al-M Si Zn, with an initial toughness of 3 MPam'I2
P-
rising to a plateau of 5 MPa m 12. However,
-
a complex commercial alloy (A-380)
yielded DMO composites which contained intermetallic phases and also gave a
constant toughness of 3MPam'I2. The toughness appears to be increased by
additions of Sic [91]; in particular, the increment in the plateau of the crack
resistance curve increased with S i c particle size. It was suggested that particle
bridging and frictional sliding at the interface could contribute significantly to
energy dissipation. A model developed to explain the increase in KIc accounted
satisfactorily for the toughness of composites containing 67 and 165 pm Sic
particles, but underestimated the value for 35 pm Sic. It was concluded that
particulate contributions to toughening exceeded those due to plastic dissipation,
at least for the composites derived by DMO from this alloy (A-380), which con-
tained a high fraction of brittle intermetallics and displayed a flat R-curve. In situ
measurements of crack opening displacement in a A1203-SiC-A1 DMO composite
[95] confirmed that the intrinsic (crack tip) toughness was about 2 MPa similar
to many other alumina matrix composites) and that the observed increases of up to
7.5 MPa ml/' could be attributed entirely to crack wake effects, including frictional
sliding, S i c particle bridging and plastic deformation of alloy bridges, rather than to
changes in the crack-tip profile. Anderson [96] measured KI in four-point bending
for both a base composite and a Sic-reinforced composite and obtained 7.8 and
5.6 MPa m1/2,respectively. The decrease observed after growth into the particulate
preform is inconsistent with the results of Pickard et al. but it is unlikely that the
alloys used for the two composites were the same.
In additional results [97] for the fracture toughness, KIc, determined from a single
edge-notched beam test, values of 3.6MParn'l2 (for a 100pm Sic particulate
preform) up to 6.2MPam'12 (13 pm Sic) were reported. This trend is the opposite
of that reported by Pickard et al. [91]. Values of the elastic modulus reported in this
later work ranged from 261 to 328 GPa.
Most mechanical properties data report only rupture strength, and, since such
data are sensitive to the size and distribution of preexisting bulk defects and surface
finish, they are not easily interpreted without some help from microstructural
analysis. Flexural strengths in the range 250-300MPa were reported for two
alumina-aluminum composites, the higher values for composites that contained
more metal and less porosity, and which also had a higher fracture toughness
[60]. Values around 150-300 MPa have been published for alumina and silicon
carbide-reinforced compositions, but the particle sizes were not disclosed [92].
Pickard et al. [91] studied Sic reinforced materials and found that the finest
(35 pm) Sic particle size gave the highest strength (400 MPa) but the minimum
toughness. The coarsest preform particles, 165 pm Sic, yielded strengths only of the
order of 150 MPa. In this work microcracking at the Sic particles due to thermal
expansion mismatch with the alumina matrix was held responsible for the strength
degradation. Raman peak shift measurements have shown [98] that residual stresses
in the silicon carbide particulate preform are a minimum for large particles, so it is
possible that unfilled pore space in the preform (whose dimensions should scale with
the particle size) could be responsible for the strength loss.
1.6.3 Thermal Shock

The residual strength behavior of Al2O3-A1 DMO composites after quenching from
elevated temperatures is similar to that of other ceramic systems, with an abrupt and
monotonically increasing loss of strength above a critical temperature drop, A T ,
which increases in the range 200-300°C as the volume fraction of residual metal
increases [60]. In contrast, a composite in which most of the metal had been replaced
by open porosity, showed a low, monotonically decreasing strength with increasing
A T ,but with no abrupt transition above a critical A T . Such behavior is typical of
porous ceramics. Experiments on Sic reinforced DMO composites [97] showed
similar effects, with the thermal shock strength degradation increasing as the particle
size decreased, so much so that, although fine particulate material was initially
stronger, the coarser microstructures retained a higher residual strength after
thermal shock. These results were attributed to the beneficial role of an array of pre-
existing flaws at large particles, which could extend and prevent the growth of a
single large crack. The critical value of A T was found to be independent of particle
size in the range 13-100 pm.
1.6.4 High Temperature Strength

A Sic particulate reinforced (48%) alumina-aluminum DMO composite [99] has
been found to give lower creep rates at 1200-1300°C, when compared to silicon
carbide whisker-reinforced (10 and 50%) alumina. Creep cavitation damage was
attributed to alumina grain-boundary sliding, and stress exponents for the stress
dependence of the minimum creep rate were between 2 and 2.6. Molten metal
pockets were associated with cavitation at higher temperatures, but could also
blunt local stress concentrations, similar to the effect of viscous, glassy grain-
boundary phases at triple junctions in a conventional ceramic. Clearly, the presence
of the coarse-grained alumina skeleton in the DMO composite microstructure more
than compensates for any lack of load bearing capacity associated with the 10-15%
of liquid alloy which is present. Hot hardness values [92] and elevated temperature
transverse rupture strengths [60] show the expected strength reduction in the vicinity
of the melting point of the alloy phase (500-700°C). Nevertheless, significant high
temperature hardness remains ( 3 4 GPa at 900"C), at least in the particulate-
reinforced DMO composites.
1.6.5 Wear Properties
In terms of their potential engineering applications, the wear performance of DMO
composites is probably the most important class of mechanical behavior. Product
brochures published by the manufacturers have highlighted the superior performance
of silicon carbide reinforced composites when compared to monolithic alumina,
cemented carbides, cast irons or stellites, particularly in abrasive slurry wear tests
[loo]. Sliding block tests and slurry pump tests [93] have also demonstrated lower
wear rates for silicon carbide reinforced DMO composites in comparison with
ceramics such as silicon nitride-bonded silicon carbide and alumina. In block-on-
ring tests [loll it was found that the total wear of a A1203-SiC-AI DMO composite
was comparable to that recorded for Si3N4,S i c and SiAlON at high loads and tested
against steel, and superior to both alumina and cast iron. In addition, the wear of the
composite itself proved to be the lowest of all the materials studied, while the
measured friction coefficients were similar to, or lower than, those determined for
the other ceramics. Pin-on-disc wear tests of DMO composites against steel
[102,103] established that low wear rates and friction coefficients under high speed
dry sliding conditions were due to the formation of a lubricating tribofilm which
was rich in magnetite and which reduced both the contact stresses (by spreading
the load) and the friction coefficient. It was postulated [102-1041 that the
aluminum-rich metal phase in the composite could be crucial in forming a lubricating
surface layer by softening or melting at the high surface temperatures generated
adiabatically. At higher speeds and applied loads, cyclically varying wear rates and
friction were encountered, associated with the formation and removal of the tribo-
film. While wear rates were comparable to (or lower than) those of silicon nitride
and zirconia toughened alumina on steel, it was found that a higher metal content
in a fine particulate DMO composite reduced the wear rate, despite the lower hardness
and increased porosity [102]. A similar blunting role of aluminum has been suggested
during sliding of composites against abrasive S i c grinding discs [ 1051.
1.6.6 Mechanical Properties of Fiber-reinforced DMO Composites

Alumina-aluminum DMO compositions have been grown into both S i c and
alumina fibers. The fibers have generally been coated, either to maximize fiber
pull-out, or to protect the fiber from the melt and the atmosphere, or both. D M O
composites uniaxially reinforced with 15% Nicalon fibers [lo61 have been reported
to have mean bend strengths of 720 MPa with a fracture toughness of 27 MPam'/2
and a measured tensile strength of 460MPa at room temperature. Fareed et al.
[107,108] have published two data sets for the flexural strength of D M O composites
reinforced with 35% Nicalon cloth (satin weave), from which the residual aluminum
had been leached. In one case [107], the flexural strength dropped from 480MPa at
room temperature to 400 MPa at 1000°C and 350 MPa at 1200°C. This impressive
retention of strength in the presence of 20% open porosity in the capillary channels
was accompanied by a fracture toughness that increased from 21 MPa rn'/*at room
temperature to 23 MPam'12 at 1000°C, before dropping to 18 MPam112at 1200°C.

In their second report [108], the flexural strength increased slightly, from 460 MPa at
room temperature to 488 MPa at 1200°C before dropping to 400 MPa at 1300°Cand
340 MPa at 1400°C, while the fracture toughness dropped monotonically from
27.8 MPa m'12 at room temperature to 15.6 MPa m112at 1400°C. The fiber coatings
used in this work were not disclosed, but other reports on processing of DMO fiber
reinforced composites, described earlier, arc consistent with a duplex BN-SIC
system, the BN to introduce a low shear strength interface and the S i c to protect
the coating and fiber from oxidation.
Cyclic fatigue tensile and shear tests at room temperature, which were accompa-
nied by microstructural examination [ 1091, have revealed that the general behavior
of Nicalon-reinforced DMO composites is similar to other fiber-reinforced brittle
ceramic, glass, and glass-ceramic systems, such as Nicalon reinforced lithium
aluminosilicate (LAS) glass-ceramic. Residual stresses, introduced during fabrica-
tion, commonly generate extensive microcracking in the matrix of these brittle
materials, so that the matrix contributes little to the stiffness of the composite. In
the case of the DMO fiber-reinforced composites, the fatigue behavior in uniaxial
stress-strain tests was dominated by progressive matrix cracking (at a stress
amplitude of about 70 MPa), which was followed by interface debonding and
fiber sliding. The bend strength of notched samples remain high (70-80% of the
unnotched values) due to the well-distributed damage and global load sharing.
Tensile strengths under uniaxial loading for these biaxially reinforced composites
were found to be about 250MPa [108,109]. Substantial fiber pullout was observed,
with fiber fracture strengths, as deduced from the fracture mirrors, slightly reduced
compared to that of the virgin fiber before infiltration [109]. Strength retention of
70-100% was observed at temperatures up to 1100°C after thermal cycling to
1200°C in an inert atmosphere [108]. Tests in air at a constant stress level, just
above that required for matrix microcracking, gave a stress-rupture life of over
500h at 1100°C and over lOOOh at 1200°C. No creep was detected at 12OO0C,
even though unreinforced Nicalon fibers are known to degrade and creep under
such conditions. The apparent improvement in the high temperature properties of
the fiber was ascribed to the absence of oxidative degradation due to the protection
of the Al2O3-A1 matrix.
Some limited strength measurements have been made on alumina fiber (Nextel
610)-reinforced DMO Al2O3-Al, using duplex BN-Sic fiber coatings [88]. The
four-point load flexural strength was found to drop from 400 MPa at room tempera-
ture to 300 MPa at 700°C and 160 MPa at 1000°C. An exposure to air at 1000°C for
100 h decreased the strength by 20%. The drop in high temperature strength was
associated with a change from transgranular to intergranular fracture of the
fibers, possibly due to the precipitation of impurity phases at the grain boundaries.
The toughness of this oxide fiber-reinforced DMO composite fell from 17 MPa m'I2
at room temperature to 9MPam'/2 at 1000°C. In general, the high temperature
mechanical properties of the oxide fiber-reinforced composites appear to be inferior
to those of the Nicalon (Sic)-reinforced material. Some other grades of fiber have
also been used to reinforce DMO composites, including Sic-SCS-6, but data are
very limited.
1.7 Corrosion of Directed Metal Oxidation Composites

Published data on the resistance of DMO composites to oxidation, corrosion, and
other environmental factors is scanty. Standard Sic particulate-reinforced com-
positions have been rated for use at room temperature under conditions ranging
from slightly acidic to basic [ 1001. Concentrated hydrochloric acid is known to
cause progressive dissolution of the residual aluminum alloy. Resistance to acid
attack can be achieved [loll by active diffusion-infiltration of molten nickel to
replace residual aluminum and silicon with nickel-rich intermetallics. The corrosion
of Sic particulate-reinforced composites in chloride solutions [ 1 101 suggests that the
alloy phase is more susceptible to pitting than monolithic A1-6061, and that
corrosion is accompanied by hydrogen evolution under acidic as well as basic con-
ditions. The possibility of using SIC-DMO composites in heat exchangers for the
glass industry prompted a study of high temperature corrosion in the presence of
sodium silicate at 1000-1300°C [l 111. The composite DMO material appears poten-
tially superior to monolithic Sic and alumina, on the basis of corrosion resistance,
in the case of Sic, and thermal shock resistance, in the case of alumina. Under static
conditions, the principal mechanism of degradation was the continued oxidation of
residual alloy in the capillary channels to form additional composite at the surface.
By contrast, when a silicate was sprayed on to the surface under dynamic conditions
at high temperature, the weight gain due to oxidation of the alloy was accompanied
by dissolution of the protective silica film formed on the silicon carbide filler parti-
cles into the molten silicate (as observed in the hot corrosion of monolithic Sic). A
proprietary alumina-rich surface layer appeared to reduce the attack. Similar hot
corrosion tests with coal slags [ 1 121 demonstrated that a Sic particulate-reinforced
DMO could possess a lifetime of 4000-1 5 000 h at 1260°C, as estimated from the
time taken to degrade a 1 mm thick alumina-rich surface layer of the composite.
1.8 Other Properties

Some limited measurements have been made of the electrical conductivity [113] and
dielectric constant [114] of unreinforced DMO composites. These studies confirm the
much higher degree of connectivity of the alloy phase along the infiltration direction
as compared to the transverse directions. Anisotropy in the dielectric constant of
acid-leached composites may be ascribed to the crystallographictexture of the alumina.
1.9 Applications
The engineering advantages of DMO composite ceramics determine the possible
areas of application, although actual penetration into these markets depends on
I .9 Applications 3 17
economic factors, in particular the cost of replacing existing or competing materials.

It is helpful to list the primary engineering features of these materials and the
advantages they convey.
(i) Since no external pressure is applied and directed metal oxidation is a near-net-
shape process, very large components can be produced which approach
theoretical density and are essentially free of residual stress.
(ii) The hardness of the dense ceramic and the resistance to crack propagation
imparted by the network of metal-filled capillary channels give these materials
excellent wear resistance.
(iii) The presence of the continuous metal phase enhances the thermal conductivity,
allowing components to act as a heat sink and avoid excessive temperature
gradients.
(iv) The combination of high stiffness and high hardness, together with the relatively
low specific gravity and reasonable production cost, make the DMO material an
attractive candidate for ballistic protection.
No data are available on the sales of DMO products and the following assessment
is based almost entirely on unverified information taken from the commercial
literature and the manufacturer’s published reports.
1.9.1 Wear Resistant Components

Sic particulate-reinforced DMO materials are being used for the manufacture of large
protective chute liners in high-wear applications for the mining industry and are being
evaluated for applications as wear parts for pumps, valves, nozzles, ducts, plates, seals,
and piston engines [ 1 151. The standard DMO composites are suitable for neutral and
basic operating conditions, while modified compositions, possibly with nickel infiltra-
tion replacing the residual aluminum alloy in the capillary channels, are available for
use under acid conditions (pH as low as 1) [loo]. Results of tests in slurry pumps
indicate that a Sic particulate-reinforced DMO composite is superior to traditional
steels, cermets, and alumina in this application [93]. Extensive performance tests of
Sic and A1,03 particulate-reinforced composites for valve seat inserts in piston
engines have given promising results which were ascribed to the higher hardness and
thermal conductivity of the DMO material when compared to metallic inserts [92]. An
outer layer of a particulate-reinforced DMO composite has also been used to improve
the wear resistance of a metal-matrix composite camshaft [116]. Other applications
which have been explored include slide gates to control the flow of molten steel
(replacing magnesia or magnesia/carbon composites), and tooling for the plastics
injection molding industry (replacing various grades of tool steel) [117].
1.9.2 Ceramic Composite Armor

The published information on ceramic composite armor is neither extensive nor very
informative, although it is clear that incorporating a hard, stiff and light-weight
ceramic plate into a composite armor system can significantly reduce the weight of
armor which is required to defeat a given threat. Two examples are the boron
carbide armor used to protect the pilot’s seat in combat helicopters and the
alumina plate inserts used to improve the personal protection provided by a polymer
reinforced armored vest. The accepted criterion for judging the performance of an
armor system is the V5,test, which is the minimum weight per unit area of armor
required to defeat a given projectile under standard conditions (a 50% chance of
penetrating the armor module at a specified projectile velocity V ) .
Sic particulate-reinforced DMO tiles are known to have attractive ballistic
properties against light and medium threats (up to and including medium caliber,
high velocity rounds and fragments). The DMO materials are candidates for
vehicular add-on armor [ 1171 and are reported to have been used to provide
added protection to armored personnel carriers by the US Army in the Persian
Gulf war against Iraq. There is no indication that DMO armor has ever been
considered seriously as a candidate ballistic material to defeat the heavy threat
faced by a main-line battle tank. In terms of cost, DMO armor should be competi-
tive with steel because of the reduced weight [118]; however, it is by no means clear
that the DMO product is superior to other grades of alumina-based ceramic armor
tile, either in terms of cost or performance.
1.9.3 Thermal Barriers and Heat Sinks

Porous aluminum titanate DMO has both a low thermal expansion coefficient
(nearly zero at least up to 600°C) and a low thermal conductivity, suggesting
possible applications in thermal barriers and abradable seals for turbine rotor
shrouds [I 161 and automobile exhaust port liners and gas desulfurization nozzles
[117]. Other possible thermal applications of silicon carbide reinforced DMO
include heat exchangers [117] and flaps and seals in the afterburners of jet engines
[119]. In addition, Sic particulate-reinforced DMO could serve as a lossy insert
for high powered microwave tubes, withstanding temperatures up to 1000°C in
ultrahigh vacuum.
It has even been suggested that very large DMO composite cylinders might be
suitable for applications in deep-sea submersibles [ 1201.
References
1. M. S. Newkirk, A. W. Urquhart, H. R. Zwicker, and E. Breval, J . Muter. Res. 1986,1,81-89.
2. M. S. Newkirk, H. D. Lesher, D. R. White, C. R. Kennedy, A. W. Urquhart, and T. D. Claar,
Ceram. Eng. Sci. Proc. 1987, 8, 879-885.
3. E. T. Thews, Met. Ind. (Lond.) 1927, 31, 597.
4. K. J. Brondyke, J . Am. Ceram. SOC.1953, 36, 171-174.
5. J. G. Lindsay, W. T. Bakker, and E. W. Dewing, J . Am. Cerum. Soc. 1964, 47, 90-94.
6. A. W. Urquhart, Muter. Sci. Eng. A , 1991, A144, 75-82.
References 3 19
7. A. F. Beck, M. A. Heine, E. J. Caule, and M. J. Pryor, Corros. Sci. 1967, 7, 1-22.

8. R. K. Hart and J. K. Maurin, Surf. Sci. 1970, 20, 285-303.
9. G. M. Scamans and E. P. Butler, Metall. Trans. A , 1975,6A, 2055-2063.
10. W. Thiele, Aluminium, 1962, 38, 707-715 and 780-786.
11. S. Balicki, Pruce. Inst. Hutnic. 1958, 10, 208-213.
12. M. V. Mal’tsev, Izv. Akad. Nuuk. SSSR Ser. Fiz. (English Translation) 1956, 20, 747-750.
13. C. Lea and J. Ball, Appl. Surf. Sci. 1984, 17, 344362.
14. D. J. Field, G. M. Scamans, and E. P. Butler, Metall. Trans. A , 1987, 18A, 463472.
15. G. D. Preston and L. L. Bircumshaw, Philos. Mug. 1935,20, 70G720.
16. W. W. Smeltzer, J . Electrochem. SOC.1958, 105, 67-71.
17. M. J. Dignam, J . Electrochem. SOC.1962, 109, 184191 and 192-198.
18. J. M. Chen. T. S. Sun, R. K. Viswanadham, and J. A. S. Green, Metall. Trans. A , 1977,8A,
1935-1940.
19. T. S. Sun, J. M. Chen, R. K. Viswanadham, and J. A. S. Green, Appl. Phys. Lett. 1977,31,
580-582.
20. C. Lea and C. Molinari, J. Mater. Sci. 1984, 19, 23362352.
21. T. Malis and M. C. Chaturvedi, J. Muter. Sci. 1982, 17, 1479-1486.
22. G. M. Scamans, R. Alani, and P. R. Swann, Corros. Sci. 1976, 16,443459.
23. I. Haginoya and T. Fukusaka, Trans. Jpn. Inst. Met. 1983,24, 613-619.
24. W. Kahl and E. Fromm, Metall. Trans. B, 1985, 16B, 47-51.
25. M. Drouzy and M. Richard, Fonderie, 1974, 29, 121-128.
26. D. A. Weirauch Jr., J. Muter. Res. 1988, 3, 729-739.
27. B. F. Devereaux, Trans. TMS-AIME,1968,242, 795.
28. M. C. Breslin, J. Ringnalda, L. Xu, M. Fuller, J. Seeger, G. S. Daehn, T. Otani, and H. Fraser,
Mater. Sci. Eng. 1995, A195 113.
29. R. E. Loehman, K. Ewsuk, and A. P. Tomsia, J . Am. Ceram Soc. 1996,79,27-32.
30. W. G. Fahrenholtz, K. G. Ewsuk, D. T. Ellerby, and R. E. Loehman, J. Am. Ceram. Soc. 1996,
79, 2497-2499.
31. M. Hanabe and P. B. Aswath, Actn Mater. 1997, 45, 40674076.
32. M. Hanabe and P. B. Aswath, J . Muter. Res. 1996, 11, 1562.
33. W-P. Tai, T. Watari, and T. Torikai, Am. Cerum. Soc. Bull. 1997, 76, 8 6 8 9 .
34. M. K. Aghajanian, M. A. Rocazella, J. T. Burke, and S. D. Keck, J . Muter. Sci. 1991,26,447-
454.
35. D. K. Creber, S. D. Poste, M. K. Aghajanian, and T. D. Claar, Cerum. Eng. Sci. Proc. 1988,9,
975-982.
36. M. K. Aghajanian, J. P. Biel, R. G. Smith, and C. R. Kennedy, J. Muter. Sci. Lett 1994, 13,
293-296.
37. H. Scholz, R. Gunther, J. Roedel, and P. Greil, J . Muter. Sci. 1993, 28, 939-942.
38. V. Jayaram, B. S. S. Daniel, N. Nagendra, and H. R. Muralidhar, Truns. Muter. Res. Soc. Jpn
1993, 14A, 621-624.
39. M. K. Aghajanian, R. A. Langensiepen, M . A. Rocazella, J. T. Leighton, and C. A. Anderson,
J. Muter. Sci. 1993, 28, 6683-6690.
40. W. B. Johnson, Cerum. Eng. Sci. Proc. 1992, 13, 573-580.
41. M. K. Aghajanian, J. P. Biel, and R. G. Smith, J . Am. Cerum. Soc. 1994,77, 1917-1920.
42. T. D. Claar, W. B. Johnson, C. A. Anderson, and G. H. Schiroky, Cerarn. Eng. Sci. Proc.
1989, 10, 599-609.
43. E. Breval and W. B. Johnson, J . Am. Ceram. Soc. 1992,75, 2139-2145.
44. W. B. Johnson, A. S. Nagelberg, and E. Breval, J . Am. Ceram. Soc. 1991, 74, 2093-2101.
45. A. S. Nagelberg, Solid State Ionics, 1988, 32/33, 783-788.
46. K. C. Vlach, 0. Salas, H. Ni, V. Jayaram, C. G. Levi, and R. Mehrabian, J . Muter. Res. 1991,
6, 1982-1995.
47. 0.Salas, H. Ni, V. Jayaram, K. C. Vlach, C. G. Levi, and R. Mehrabian, J. Muter. Res. 1991,
6, 19641981.
48. A. S. Nagelberg, J . Muter. Res. 1992, 7, 265-268.
49. S. Antolin, A. S. Nagelberg and D. K. Creber, J . Am. Cerum. SOC.1992, 75, 447-454.
50. P. Xiao and B. Derby, J . A m . Cerum. SOC.1994,77, 1761-1770.
3 20 I Directed Metal Oxidation
51. 0. Salas, V. Jayaram, K. C. Vlach, C. G. Levi, and R. Mehrablan, J . Am. Ceram. SOC.1995,
78, 609-62 I.
52. C. D. Qin, M. 0.Lai, B. Cotterell, and S. Shi, Muter. Res. SOC.Symp. Proc. 1995,365, 133-138.
53. H. Schmalzried and W. Laqua, Oxid. Metals, 1981, 15, 339-353.
54. K. C. Vlach, C. G. Levi, and R. Mehrabian, in Processing and Manufacturing of Advanced
Materials for High Temperature Applications, V. A. Ravi and T. S. Srivatsan (Eds), The
Minerals, Metals and Materials Society, Warrendalc, USA, 1992, pp. 157-171.
55. V. S. R. Murthy and B. S. Rao, J . Muter. Sci. 1995, 30, 3091-3097.
56. M. Sindel, N. A. Travitzky, and N. Claussen, J . Am. Ceram. SOC.1990, 73, 2615-2618.
57. M. Hanabe, V. Jayaram, and T. A. Bhaskaran, Acta Muter. 1996,44, 819-829.
58. P. Xiao and B. Derby, J. Am. Ceram. Soc. 1994,77, 1771-1776.
59. A. S. Nagelberg, S. Antolin, and A. W. Urquhart, J. Am. Ceram. Soc. 1992, 75, 455462.
60. M. K. Aghdjanian, N. H. Macmillan, C. R. Kennedy, S. J. Luszcz, and R. Roy, J . Muter. Sci.
1989,24, 658-670.
61. E. Breval, M. K. Aghajanian, and S. J. Luszcz, J . Am. Ceram. SOC.1990, 73, 2610-2614.
62. 0.Salas, A Fundamental Study of the Formation of A1,03/Al Composites from the Oxidation
of Liquid A1 Alloys, PhD thesis, University of California at Santa Barbara, USA, 1992.
63. 0. Salas, V. Jayaram, K. C. Vlach, C. G. Levi, and R. Mehrabian, in Processing and
Manufacturing of Advanced Materials for High Temperature Applications, V. A. Ravi and
T. S. Srivatsan (Eds), The Minerals, Metals and Materials Society, Warrendale, USA, 1992,
pp. 143-156.
64. P. Padmapriya, T. A. Abinandanan, and V. Jayaram, Scripta Muter. 1997,37, 1883-1888.
65. P. Xiao and B. Derby, J. Europ. Ceram. SOC.1997, 12, 185-195.
66. V. Jayaram, J . Muter. Sci. 1996,31,45914598.
67. F. J. A. H. Guillard, R. J. Hand, and W. E. Lee, Br. Cerum. Trans. 1994, 93, 129-136.
68. X. Gu and R. J. Hand, J . Muter. Sci. 1995,30,6005-6012.
69. X. Gu and R. J. Hand, J . Europ. Ceram. SOC.1996, 16, 929-935.
70. M. Pinkas, A. Venkert, and E. Manor, Scripta Muter. 1997,36, 377-383.
71. A. S. Nagelberg, Muter. Res. Soc. Symp. Proc. 1989, 155, 275-282.
72. E. Breval and A. S. Nagelberg, Mater. Res. Soc. Symp. Proc. 1989, 132,93-98.
73. A. Guha and V. S. R. Murthy, Bull. Muter. Sci. 1996, 19, 1117-1124.
74. S. C. Khatri, M. J. Koczak, T. Chou, and Y. Kagawa, Ceram. Eng. Sci. Proc. 1992,13,494-502.
75. T. Watari, K. Mori, T. Torikai, and 0. Matsuda, J . Am. Ceram. Soc. 1994,77,2599-2602.
76. S. C. Koh, D. K. Kim, and C. H. Kim, J . Muter. Synth. Proc. 1995,3, 105-110.
77. H. Venugopalan, K. Tankala, and T. Debroy, Muter. Sci. Eng. 1996, A210, 6 4 7 5 .
78. S. C. Khatri, M. J. Koczak, T. Chou, and Y. Kagawa, Ceram. Eng. Sci. Proc. 1992,13,485493.
79. D. D. Upadhyay, R. Bhat, S. Ramanathan, and S. K. Roy, J . Alloys Compd. 1994, 205,
274279.
80. H. Venugopalan and T. Debroy, Muter. Sci. Eng. 1997, A232, 39-46.
81. S. Kumar, V. Jayaram, T. V. Mani, and K. G. K. Warrier, J . Muter. Sci. 1997,32,47194722.
82. E. Manor, H. Ni, C. G. Levi, and R. Mehrabian, J . Am. Ceram. Soc. 1993,76, 1777-1787.
83. S. P. Dhandapani, V. Jayaram, and M. K. Surappa, Acta Metall. Muter. 1994, 42, 649-656.
84. S. Lee and D. K. Kim, Ceram. Eng. Sci. Proc. 1990,11,795-805.
85. V. Jayaram, S. Kumar, T. V. Mani, M. S. M. Saifullah, J. Sarkar, and K. G. Warrier, in
Inorganic Matrix Composites, M. K. Surappa (Ed.), The Minerals, Metals and Materials
Society, Warrendale, USA, 1996, pp. 193-205.
86. V. S. R. Murthy and A. Deepak, Br. Ceram. Trans. 1996, 95, 173-176.
87. A. S. Fareed, G. H. Schiroky, and C. R. Kennedy, Ceram. Eng. Sci. Proc. 1993,14,794-801.
88. A. S. Fareed and G. H. Schiroky, Ceram. Eng. Sci. Proc. 1994, 15, 344352.
89. X. Gu and E. G. Butler, Ceram. Eng. Sci. Proc. 1997, 18, 371-378.
90. I. S. Kim and H. W. Hennicke, Znterceram. 1992,41, 75-76.
91. S. M. Pickard, E. Manor, H. Ni, A. G. Evans, and R. Mehrabian, Acta Metall. Mater. 1992,
40, 177-184.
92. D. J. Landini, H. D. Lesher, and J. Gesing, in Proc. Third International Symp. Ceramic
Materials und Components for Engines, V. J. Tennery (Ed.), American Ceramic Society,
Westerville, USA, 1988, pp. 1536-1551.
Rejerences 32 1
93. J. Weinstein and B. Rossing, in High Performance Composites for the 1990s, S. K. Das, C. P.
Ballard, and F. Marikar (Eds.), The Minerals, Metals and Materials Society, Warrendale,
USA, 1991, pp. 339-357.
94. B. D. Flinn, M. Ruehle, and A. G. Evans, Acta Metall. 1989, 37, 3001-3006.
95. J. Roedel, M. Sindel, M. Dransmann, R. W. Steinbrech, and N. Claussen, J . Europ. Ceram.
SOC.1994, 14, 153-161.
96. C. A. Andersson and M. K. Aghajanian, Ceram. Eng. Sci. Proc. 1988, 9, 621-626.
97. I-S. Kim and I-G. Kim, J . Muter. Sci.Lett. 1997, 16, 772-775.
98. R. Arvind Singh, A. K. Sood, V. Jayaram, and S. K. Biswas, Scripta Muter. 1998,38,617-622.
99. H-T. Lin and K. Breder, J . Am. Ceram. Soc. 1996,79, 2218-2220.
100. Reported in Product Brochures, ALANX Products, 101 Lake Drive, Newark, Delaware
19702, USA.
101. R. K. Dwivedi, Cerum. Eng. Sci. Proc. 1991, 12, 2203-2221.
102. V. Jayaram, R. Manna, M. G. Kshetrapal, J. Sarkar, and S. K. Biswas, J. Am. Ceram. Soc.
1996,79, 770-772.
103. A. Ravikiran, V. Jayaram, and S. K. Biswas, J . Am. Ceram. SOC.1997, 80, 219-224.
104. J. Zhang, D. G. Zhu, and L. Yang, Wear, 1998,215, 34-39.
105. R. Arvind Singh, V. Jayaram, and S. K . Biswas, J . Muter. Res. 1999, 14, 6 6 6 7 .
106. C. A. Anderson, P. Barron-Antolin, A. S. Fareed, and G. H. Schiroky, in Proc. Int. Con$
Whisker and Fiber Toughened Ceramics, American Society for Metals, 1988, pp. 209-21 5.
107. A. S. Fareed, B. Sonurpalak, C. T. Lee, A. J. Fortini, and G. H. Schiroky, Ceram. Eng. Sci.
Proc. 1990, 11, 782-794.
108. A. Fareed, B. Sonurpalak, P. A. Craig, and J. E. Garnier, Ceram. Eng. Sci. Proc. 1992, 13,
804818.
109. F. Heredia, A. G. Evans, and C. A. Andersson, J . Am. Ceram. SOC.1995,78,2790-2800.
110. D. G. Kolman and D. P. Butt, J . Electrochem. SOC.1997, 144, 3785-3791.
111. W. A. Kem, M. J. McNallan, and R. E. Tressler, J . Am. Ceram. Soc. 1996, 79, 1881-1891.
112. T. M. Strobel, J. P. Hurley, K. Breder, and J. E. Holowczak, Ceram. Eng. Sci.Proc. 1994, 15,
579-586.
113. H. Venugopalan, K . Tankala, and T. Debroy, J . Am. Ceram. Soc. 1994,77, 3045-3047.
114. S. L. Swartz, D. R. White, and L. E. Cross, J . Muter. Sci. 1992, 27, 3932-3938.
115. Reported in Futuretech, Technical Insights Inc., Englewood, USA, 1990, 101, 1-19.
116. B. W. Sorenson, G. H. Schiroky, and A. W. Urquhart, Turbomachinery, 1990,31,20-26.
117. S. Ashley, Mech. Eng. 1991, 113, 44-49.
1 18. Reported in Structural CeramicslCerumic Matrix Composites, Techmonitoring, SRI Inter-
national, 1995, 44-45.
119. L. P. Zawada and S. S. Lee, Ceram. Eng. Sci. Proc. 1995, 16, 337-339.
120. Reported in: Society and Industrial News, Am. Ceram. Soc. Bull. 1995, 73, 28.
2 Self-Propagating High-Temperature Synthesis

of Hard Materials
Z. A. Munir and U. Anselmi-Tamburini
2.1 Introduction
Of the many advantages of self-propagating high-temperature synthesis (SHS),
most can be realized in the synthesis of hard (primarily refractory) materials.
Among the most important and generally applicable of these advantages are: the
energy efficiency provided by this process, its tendency to produce relatively pure
products, and its suitability for the simultaneous synthesis and densification of
monolithic and composite materials. The synthesis and densification of composites
can be accomplished in one step through the use of this method. In recent years, the
utilization of the SHS process for the preparation of hard and other types of
materials has been extensive with a variety of experimental approaches and modifi-
cations, including chemical, thermal, mechanical, and electromagnetic activations.
The basis of the SHS method is the propensity of highly exothermic reactions to
become self-sustaining after they have been initiated by a brief energy input. A
requirement for such a phenomenon is that a portion of the heat generated by the
chemical reaction is conducted forward to raise the temperature of the next
reactants layer to a temperature at which the reaction rate becomes significant, a
temperature commonly referred to as the 'ignition' temperature. Part of the heat
generated in this layer will be conducted forward and thus the process is repeated
and the reaction moves along the reactants within a usually narrow zone, typically
referred to as the combustion zone or combustion wave. The process is demon-
strated by the sequence of photographs in Fig 1 .
As will be seen in more detail below, the SHS process is governed by the chemical
and physical parameters of the reacting system. And under real experimental condi-
tions, the process is also controlled by the ambient conditions. To understand the
SHS process and its governing parameters, we consider the general Fourier heat
transfer relationship accounting for a heat source,
+
pC,(dT/dt) = fi(d2T/dx2) p Q ( d q / d t ) - h / w ( T - To) - u s / w ( T 4- T t ) , ( 1 )
where p is the density of the product (g ~ m - ~Cp
) , is its heat capacity (J g-' K-I), K. is
the thermal conductivity (JcmP2s-' K-*), Q is the heat of the reaction (Jg-I), q is
the fraction of the reaction completed, h is the convective heat transfer coefficient
(J cm-2 sC1 K-' , us is the Stefan-Boltzmann constant (J cmP2s-' KP4), T is
temperature (K), To is the ambient temperature, w is the sample thickness (cm), t
is time (s), and x (cm) is the coordinate along which the combustion wave is
advancing. The term on the left-hand side of Eq. (1) is the algebraic total rate of
heat accumulated in the combustion zone and is the sum of the net rate of conducted
Figure 1. Photographic sequence of self-propagating combustion synthesis: reaction between Ti

powder compact and gaseous nitrogen.
heat, the rate of chemical heat evolution, the rate of heat loss due to convection,
and the rate of heat loss due to radiation. The reaction rate is assumed to have an
Arrhenius dependence on temperature and has the form
dq/dt = ko(l - q)"exp(-E/RT), (2)
where E is the activation energy for the process, n is the order of the reaction, R is
the gas constant, and ko is a constant.
Under ideal (adiabatic) conditions, the last two terms on the right-hand side of Eq. (1)
are neglected and the calculated maximum temperature in the combustion zone is the
adiabatic combustion temperature, Tad.For most of the important refractory materials
(including many hard materials) Tad is high, exceeding 4000 K in some cases. The high
value of this parameter is a thermodynamic consequence of the high exothermic heat of
formation of these materials. Table 1 lists the values for Tad and the enthalpies of
formation for selected hard materials, including carbides, nitrides, oxides, and borides.
With few exceptions, the adiabatic temperatures of the listed materials exceed the
empirically established minimum limit for SHS, about 2000 K [I]. Generally the excep-
tions are compounds of tungsten (W2C,WC, WB, W2B),molybdenum (Mo2C,MoN,
MOB),and chromium (Cr3C2).It should be emphasized that the temperatures in Table
1 are based on adiabatic conditions and thus are upper limits which are seldom
achieved in practice. In some cases, notably in the case of nitrides, the adiabatic tem-
perature is so high as to exceed the dissociation temperature of the nitride phase.
Examining Table 1 shows a general trend between the enthalpy of formation of the
material and its adiabatic temperature. However, a more accurate relationship is
between the ratio of the enthalpy to the heat capacity (H/Cp)29X and the adiabatic
324 2 Self-propagating High-Temperature Synthesis of Hard Materials
Table 1. Adiabatic combustion temperatures and enthalpies of formation of selected materials.
Material Enthalpy, - H (kJ) Combustion temperature, Tad(K)
TIC 184.5 3 290

HfC 208.2 3 802
ZrC 196.7 3913
NbC 138.9 2821
vc 100.8 2 228
TaC 144.1 2 902
Cr3C2 83.3 957
Mo~C 53.1 1015
wc 40.2 1164
w2c 26.4 603
TIN 323.0 4 459
HfN 373.6 5 622
ZrN 371.5 4941
NbN 236.4 3 508
VN 217.2 3 758
TaN 252.3 3 436
MoN 40.8 1239
TiB, 323.8 3 193
HfB2 329.3 3 979
ZrB2 328.0 4 183
NbB2 25 1.O 2715
WB 66.1 1617
W2B 66.9 1158
VB 138.5 2 490
TaB2 209.2 2 729
A1203 1675.3 11 312
Zr02 1097.5 12 153
Ti02 944.7 9 519
temperature (Tad),as has been shown previously [2]. Such a relationship is shown in
Fig. 2 for the materials listed in Table 1 for which reliable thermodynamic data are
available. With two notable exceptions, the data fit a straight line and a more useful
limit for SHS can now be stated as a (H/Cp)298 RZ 3000 K.
In SHS reactions, the maximum temperature is ideally achieved within the narrow
combustion zone, indicating that the reaction is complete within the combustion
wave. However, for kinetic reasons, some reactions continue beyond the zone and
consequently the maximum temperature is reached after the passage of the wave.
Schematic representations of these two cases are shown in Figs 3(a) and (b), respec-
tively. The figures show profiles of temperature and the conversion parameter. In the
latter case, diffusion of one or both reactants through a product layer can provide a
kinetic barrier for the process and hence its lack of completion within the com-
bustion zone. Such a behavior can be seen in systems where the particle size of
the reactants is relatively large, as was reported earlier for the synthesis of titanium
carbide [3]. Since in this case the conversion is incomplete in the wave, the tem-
perature of the wave is relatively low and hence its rate of propagation is low. In
general the nature of the product is not affected by a change in velocity as long as
the propagation is in the steady-state mode [4-6]. When the propagation is in the
E
2 0
1 5
1 0
5
F TiCh
0
0 2 4 6 8 1 0 1 2 1 4
Tadx1 0- 3( K )
Figure 2. The relationship between (AH,,,/C,) and the adiabatic temperature for hard materials.
non-steady-state mode (e.g. spin or pulsating wave propagation), the reaction may
be incomplete or the products are not the same as those obtained under steady-state
propagation [6]. An example of spin propagation is shown in Fig. 4 for the synthesis
of nickel silicides [7].
Wave
Propagation I I&
XrdL
Wave propagation
t-------
~
k 4b
T,
Vf
k i . After-bum 4 *
Figure 3. Schematic representation of the temperature, T , degree of conversion, 7,and rate of heat
release, 4, during wave propagation for: (a) a reaction completed within the wave, (b) a reaction
which extends beyond the wave.
Figure 5. Temperature profiles during the combustion synthesis of: (a) Ti + C + 25 weight-% Ni,
Figure 4. Evidence of spin combustion during the synthesis of Ni silicides [7].
2 Self-propagating High-Temperuture Synthesis of Hard Materials
60
45 52.5
22.5 30 37.5
time (s)
(b) Ti N2 (gas) [12].

15
7.5
1
+
0
2500
2500
326
2.2 Mechanistic Characterization of the Process

Determinations of the mechanism of self-sustaining synthesis reactions have been
attempted through two experimental approaches. The first, and most widely used,
method examines the temperature dependence of the wave velocity [8] and the
second utilizes temperature profiles associated with wave propagation [9-111.
Although in principle both methods should provide the same value for the
activation energy, in reality results obtained by the two methods can be significantly
different [ 11. The difference is partly attributable to the spatial distribution of
the conversion parameter. Referring to Fig. 3(b), which depicts a case where the
reaction is not complete within the wave, the temperature in the wave is not the
maximum, but it is the value that controls the rate of wave propagation. Thus
activation energies obtained for such cases are not valid for the process. Examples
of synthesis processes in which the reactions are complete within the zone and those
in which the reactions continue beyond the wave are shown in Figs 5(a) and (b),
2500
2000
G 1500
0
G
I000
500
O L J I 1 I I I I 1
0 15 30 45 60
t 6)
2500 -
2000 -
-Y 1500 ~
G -
1000
500 -
O r . I I I
Figure 6. Temperature profile of a 52% dense Zr + 1.5B + 15 weight YOTiB2: (a) complete profile,
(b) enlarged portion of the initial segment of the profile [lo].
328 2 Self-Propagating High-Temperature Synthesis of Hard Materials
v
1000
500
- 0
3.00 3.04 3.08 3.12 3.16 3.20
- 150000
50000
s
b 0
-50000
3.00 3.04 3.08 3.12 3.16 3.20
-.
PI
-I.E+7
" ! -1.5E+7
3.00 3.04 3.08 3.12 3.16 3.20

t 6)
(C)
+ +
Figure 7. Thermal profile characterizations of the wave in the reaction of Zr 1.5B 15 weight-%
TiB2: (a) smoothed temperature profile, (b) first derivative of the profile, (c) second derivative of the
profile [lo].
respectively. The former is the temperature profile for the synthesis of Tic-Ni
composite and the latter is for the synthesis of TIN [12]. In the latter case, both
visual observations and thermogravimetric analyses confirmed that the reaction
between Ti and nitrogen continues well after the passage of the wave [8,13].
As indicated above, determination of the kinetic parameters of the SHS process
can be made through a mathematical analysis of temperature profiles [14]. Integra-
tion of an adiabatic form of Eq. (1) along with appropriate substitutions gives the
following expression for the conversion parameter [9],
V(X) = [Cpu(T- TO)- 61( d T / d x ) I / [ ( - (3) ) QWI,
~ ~ l ) ( d T / d x+
where u is the wave velocity, K~ and K~ are the thermal conductivity of the reac-
tants and products, respectively, To is the ambient temperature, and the other
symbols are as defined above. Equation (3), which gives the spatial distribution
UL 20
10
0
3.00 3.04 3.08 3.12 3.16 3.20
t (s)
":E
(a)
0.6
0.2
0
-0.2
3.00 3.05 3.10 3.15 3.20
+ +
Figure 8. Thermal profile characterization of the reaction Zr 1.5B 15 weight-% TiB,: (a) the
time dependence of the reaction rate, aq/dt, (b) the time dependence of the conversion factor, q [lo].
of the conversion parameter, can be differentiated with respect to time to give the
reaction rate, &/at, which is related to x by d v / d t = u ( a q / a x ) .Utilization of this
(the Boddington-Laye) method of kinetic analysis has been made in investigations
on the combustion synthesis of zirconium boride, molybdenum silicide, and tita-
nium silicide [lo, 111. Figure 6(a) shows a typical profile during the reaction of
+ +
Zr 1.5B 15weight-% TiB2 and Fig. 6(b) shows an enlarged section of the
initial portion of the profile. A smoothed version of the latter as well as its first and
second derivatives are shown in Figs 7(a) to (c), respectively. From Fig. 7(b), the maxi-
mum heating rate at the leading edge of the wave is calculated as 1.6 x lo5K s-'.
Using Eq. (3), the reaction rate and the temporal distribution of the conversion
parameter can be calculated, as shown in Figs 8(a) and (b), respectively. It can be
seen from Fig. 7(b) that the conversion is complete in about 8 x lO-*s and with
the knowledge of the wave velocity, the width of the combustion zone is calculated
as about 0.4 mm.
Using the calculated reaction rates and their dependence on temperature, activa-
tion energies for the formation of zirconium boride can be calculated, as shown in
Fig. 9. The lines represent data for different levels of conversion. Those representing
7 values of 0.6, 0.7, and 0.8 have similar slopes. The lack of conformity of the lines
for q of 0.5 and 0.9 is related to the uncertainties in the process of obtaining
derivatives at these ends of the conversion profile, as explained elsewhere [lo].
The average value for the activation energy for the combustion synthesis reaction
Zr + 1.5B + XTiB2 (with X between 0.15 and 0.22) was calculated as 144.5 x
15 kJ molF' . The activation energy was also calculated from measurements of the
330 2 Self Propagating High-Temperature Synthesis of Hard Materials
4 -
g3 -
5 * F \ ,
5 2 -
I -
t 0.9
0 1 I I I
3 4 5 6 I
1 0 ' (IK)
~
Figure 9. Arrhenius plot of the temperature dependence of the reaction rate for Zr + 1.SB+XTiB,
for 7 values from 0.5 to 0.9.
temperature dependence of the wave velocity, shown in Fig. 10. The resulting value,
148.6 x 12kJmol-', is in good agreement with that determined from the profile
analysis, indicating that the reaction was complete within the relatively narrow
reaction zone.
Similar analyses were made for the synthesis of MoSi2 and Ti2Si3 [l 11. In these
cases, particularly for MoSi2, the reaction is not complete within a narrow zone
as can be seen from Fig. 11. Here the width of the zone is calculated as
approximately 1.3mm. The wide reaction zone for molybdenum silicide is, in
part, the consequence of a relatively low enthalpy of formation as manifested by
an adiabatic temperature near the empirically established lower limit. This is the
reason why some composites of MoSi2 cannot be directly synthesized by SHS
without some form of activation.
-7.8
E = 134 W/mol (Qconstant) 1
-8.0
r.
F
-3
e
-8.2
d
C
-8.4
-8.6
0 E = 148 Wimol (Qvarying)
I I I I
-II
3.4 3.5 3.6 3.7 3.8 3.9
1 0 ' (~1 1 ~ )
Figure 10. Arrhenius plot of the temperature dependence of the wave velocity in the reaction
Zr = 1.5B+XTiB2 [lo]: 0, from velocity measurements; 0 , from temperature profiles.
2.3 Efect of Experimental Parameters 33 1
0.9
0.8
0.7
0.6
0.5
Er
0.4
0.3
0.2
0.1
0.0
-0.1
2.4 2.5 2.6 2.7 2.8 2.9
t 6)
Figure 11. The time dependence of the conversion profile, 7, for the synthesis of MoSi,
2.3 Effect of Experimental Parameters

Typical SHS experiments for materials synthesis involve the use of powders.
Powders are mixed in the desired stoichiometry, pressed into a compact, and then
ignited at one end to produce the self-sustaining combustion wave. In some cases,
a diluent is added. This is done either because its presence in the product is desired,
or it is added to control (i.e. to lower) the temperature of the reaction. In the latter
case, the product of synthesis is added as a diluent. The SHS procedure, therefore,
includes the materials and experimental parameters of powder particle size, degree
of dispersion, relative density of the compact, the stoichiometric ratio, the degree
and nature of dilution, and the initial temperature of the reactants. The initial
temperature, To, has a direct influence on the adiabatic temperature and hence
the feasibility of establishing self-sustaining combustion. A schematic representa-
tion of the general effect of To on Tad is depicted in Fig. 12(a) and the specific
effect in the case of T i c is shown in Fig. 12(b) [15]. In the latter case, Tad coincides
with the melting point of TIC (3210 K) with the fraction of melted product = 0.33
for To 7299 K. Figure 12(a) shows that increasing To from T , to T, increases the
adiabatic temperature from Tal to Ta2.
The relationship between Toand Tad can be derived for a general case in which
Tad < T,, where T,, is the melting point of the product as
H298 +
TO
1298
(a + bT)d T = / Tad
TO
(up + b p )dT, (4)
where H298is the enthalpy of formation of the product at 298 K, A u and Ab are the
algebraic sums of the Cp coefficients of the product and reactants (assuming a
Cp = a + bT form), ap and b, are the coefficients for the heat capacity equation
of the product. In the case of TIC, shown in Fig. 12(b), the adiabatic temperature
(for To = 298 K) is at the melting point of the product and increasing To initially
3340 -
-
3320
3300 -
8 3280 -
2 3260 -
3240 -
3220 -
A
3200 -
I
400 800 1200 1600 2000 2400 2800
To(K)
(b)
Figure 12. (a) Schematic representation of the influence of the initial temperature on the adiabatic
temperature: His the enthalpy and T is temperature. (b) The dependence of the adiabatic tempera-
ture on the initial temperature for TIC [15].
has no effect on Taduntil all the product is in the molten state. Further increase in To
will result in an increase in the adiabatic temperature. The preheating of reactants
before ignition is a method employed in the synthesis of materials whose adiabatic
temperatures are relatively low. This is to be distinguished from the practice of heat-
ing the reactants (e.g. in a furnace) until the entire sample reacts. This latter form of
combustion synthesis has been erroneously referred to as SHS, but, in fact, it has no
discernible combustion wave and it is more appropriately referred to as the simul-
taneous (or volume) combustion form of combustion synthesis. For this form of
combustion synthesis, the rate of heating to the ignition temperature plays a
major role in the thermal behavior of the process and the nature of the resulting
products [16]. Lower heating rates result in the formation of precombustion
phases through interfacial diffusion with the consequence of a reduced driving
force in the main (combustion) reaction [17,18].
---
2.3 Effect of Experimental Parameters 333
1 9 o o r I ' ' I I ' ' ' I ' ' '
h
Y
Y
1800
e3
-
c
E i
-
1700
E
I-
1600 -
.-
c
m
n
m
*e I500 -
U
-0.2 0 0.2 0.4 0.6 0.8 1 1.2

Stoichiometric increment (x)
Figure 13. The effect of the stoichiometry of the reactants Mo + (l+x)Si and Si + (1 + x)C on the
adiabatic temperature.
In principle, the roles of stoichiometry and dilution are similar for product phases
with a narrow compositional stability range. For example, in the synthesis of MoSi2
+ +
and Sic, changing x in Mo (1 x)Si and Si (1 x)C has the same effect as + +
adding a diluent, as seen in Fig. 13, which shows the effect of stoichiometry on
the adiabatic temperature. This, however, is not the case when the product phase
is thermodynamically stable over a range of composition, as is the case of many
hard compounds (e.g. TIC, ZrC, NbC, TIN, ZrN, and others). For such materials,
the enthalpy of formation is a function of composition and thus changing the
stoichiometry of the reactants has the effect of changing the heat of the reaction
as well as the stoichiometry of the product. Using a modeling approach in which
the product is presumed to have a fixed nominal stoichiometry [19], the influence
of the mixture ratio of the nonmetal to the metal on the combustion wave velocity
was calculated and compared to experimental results, as shown in Figs 14(a) and (b)
for synthesis in the Ti-B and Zr-B systems [20-251. In these cases, increasing the
value of p from about 0.3 to 1.0 has the effect of reducing the degree of dilution
by the excess metal in the starting mixture. An experimental investigation on the
effect of stoichiometry on the combustion synthesis of TIC was recently carried
out by Makino et al. [26]. Figure 15(a) shows the dependence of the existence and
nature of combustion waves on the C/Ti ratio (p) for samples (1Omm diameter
and 35mm length) with initial relative densities ranging from about 55 to 95%
and a carbon particle (diameter) size of 20pm. Steady-state combustion was
observed in the range 0.47 5 p 5 1.4 and a pulsating mode of combustion was
observed for 1.4 5 p 5 1.9. For p > 1.9 and p < 0.47, the samples either were
ignited but the wave became extinguished, or the samples could not be ignited at
all. As the figure shows, the relative density had little influence on these boundaries,
The observed lower limit for steady-state propagation, p M 0.47, coincides with the
lower limit of phase stability in the Ticl p.y or about 32 at% C. On the other hand,
Ti - B
1 Zr- B
0 0.5 1 .o
P
(b)
Figure 14. Modeling results of the influence of the mixture ratio of nonmetal to the metal, p, on the
combustion wave velocity in the: (a) Ti-B system, (b) Zr-B system [19]. Tois the initial temperature:
data from [20-251.
the corresponding upper limit, p = 1.4, falls beyond the stoichiometric limit of TIC
phase and thus the product must contain excess carbon. The dependence of the
velocity on the stoichiometric ratio is shown in Fig. 15(b). The rate of propagation
increases to a maximum corresponding to the limit of steady-state propagation, that
is, the highest velocity (reaction rate) is obtained under a condition of excess carbon.
It has been suggested that excess carbon enhances the kinetics of the reaction
between liquid titanium and solid carbon [27].
The effect of dilution on the synthesis reaction of hard (refractory) materials has
been examined experimentally and through modeling. Figure 16 shows the effect of
adding TIC as a diluent on the adiabatic temperature and the fraction melted of the
product [ 151. The adiabatic temperature remains constant with the addition of TIC
up to 20 mol%, at this point the product is entirely solid and further addition of the
diluent decreases the adiabatic temperature as shown in the figure. The decrease in
the combustion wave temperature is expected to result in a decrease in the wave
velocity. Experimental confirmation of this is provided in Fig. 17 for the case of
2.3 Effect of’E.xperimcnta1 Parameters 335
I Specimen diameter=lO mm
0 Steady
0 Pulsating
OS
I
t I
A Extinction
X No ignition
0 1 2 3
15
- P*I
A 0.55-0.65
00.65-0.75
-f
fl
3 0 0 0.75-0.85
10 - 0 OM-0.95
0
.-
-
0 0 .
m *
a&
.E 5 -
.*
mE Open: steady
Filled: pulsating
0 1 2 3
Mixture ratio, p
(b)
Figure 15. The effect of the mixture ratio C/Ti (= p) on: (a) the relative density and nature of the
combustion mode, (b) the velocity of the wave for various relative densities of the reactants [26].
3300 160
3200 50
k
40
s. 3100
h
h
3000
i-”
20
2700 0
0 10 20 30 40 50 60 70
TIC (mol%)
Figure 16. The influence of dilution (by TIC) on the adiabatic temperature and fraction of product
melted (v) in the synthesis of TIC [15].
p = 1.0 d,. W)
h
Borovinskaya
E Present
.- 0
&el= 0.56
0
.-
0
m v
0.0 0.5
Degree of dilution
Figure 17. The effect of dilution on the wave velocity in the synthesis of TIC [26]. Experimental
results from [27].
TIC 1261 and compared to results of a previous investigation [27]. It is interesting to

note that the use of smaller carbon particles (0.1 pm) in the earlier study 1271resulted
in steady-state wave propagation in the same range as for the case of larger carbon
particles (25 pm). However, the use of smaller particles extended the dilution range,
albeit with pulsating wave propagation. Non-steady-state (pulsating and spin) pro-
pagation results in laminated products that are macroscopically inhomogeneous [7].
As indicated above, the addition of a diluent reduces the combustion temperature
and the velocity of the wave. When a sufficient amount of diluent is added, the wave
becomes unstable and ultimately ceases to advance. The concept of a wave extinc-
tion limit was utilized to construct SHS diagrams [4,5]. These diagrams were based
on the Zeldovich criterion for wave extinction [28],
where vmaxis the maximum (adiabatic) velocity and wmin is the minimum velocity
before the wave becomes extinguished. Equation (5) was used with the relationship
between velocity and temperature,
w2 = ( K RT2/QE)exp(-E/RT), (6)
where Q is the heat of the reaction, E is the activation energy of the process, T is
temperature, R is the gas constant, and K is a constant. The resultant relationship,
which provides boundaries for the existence of SHS, is derived as
where T,, and Tminare the maximum (adiabatic) temperature and the minimum, that
is, the temperature of the wave just before it becomes extinguished, and d is the degree
of dilution [4,5]. Using this relationship, SHS diagrams for TiB,, H B 2 ,ZrB2, and TIC
have been constructed. An example is shown in Figs 18(a) and (b) for TiB2 and Tic,
respectively. Details of the experimental results and the theoretical derivations of this
model as well as others are given in the literature [4,5].
2.3 Ejfect of Experimental Parameters 331
1400
TiB,
0.
400
200
0 20 40 60 80 100
-
(E=398KJ/Mol) -/ / \
3
1200
g 1000 - This Work
(E=398KJ/MoI)
800 -
600 -
400 - (E=l38KJ/Mol)
200 '
0
I
0.2
I
0.4
I
0.6
I
0.8 1.0
d
(b)
Figure 18. Theoretical limits of SHS (SHS diagram) and experimental results for: (a) TiB2, (b) Tic
[4,51.
The effect of particle size of one or more of the reactants on the SHS process has
been examined in experimental and modeling studies. In general, an increase in
particle size results in an increase in the diffusion length and hence a decrease in
the reaction rate, with the rate expected to be inversely proportional to particle
size. Figure 19 shows experimental observations on the relationship between the
particle size of titanium and the wave velocity in the synthesis of TIC and TiB2
[21]. This trend is also predicted by modeling studies of Makino and Law, showing
the anticipated inverse dependence of the velocity on particle size [29]. Their
prediction of this dependence is compared to results of experimental observations
in Fig. 20 for the synthesis of TIC [26]. The wave velocity decreases by a factor of
about four as the carbon particle size increased from 1 to 25pm. However, the
particle size of Ti had no apparent influence on the velocity within the range of
10-75 pm. The fact that titanium is molten in the combustion zone is believed to
provide the explanation for the lack of a dependence. The experimental results of
the trend of velocity with carbon particle size provide an unexpected observation
at the small particle size end. When small (<5 pm) carbon particles were used, the
338 2 Self-Propagating High-Temperature Synthesis of Hard Muterials
24 1 I
i
.
h
:
v 12
>
n
”
0 0.1 0.5 1.0 1.5
r, (mm)
Figure 19. Dependence of the combustion rate of Ti + 2B and Ti i-C mixtures on the Ti particle size
for several initial temperatures (To):0 , To = 20°C; 0, To = 200°C; 0,
TO= 400°C [21].
wave propagated in a pulsating mode, albeit at a much higher average velocity. The
significance of this, as was pointed out earlier, is that the products of non-steady-
state combustion are inhomogeneous. In another investigation, the size of the Ti
particles was shown to have little effect on the density of the TIC product, but it
had an effect on the resulting microstructure [30].
Another important experimental parameter in combustion synthesis is the relative
density of the reactants. Figure 21 shows the dependence of the wave velocity on
density in the combustion synthesis of TiNi [31]. These and similar experimental
p = 1.0
30 d&mm) P..,
0 Deevi et al. 1.6 0.56
1.0 0.52-0.60
2”E
h
25
-
”
25 ”
20
B
-
0
d)
Open: steady
Filled pulsating
I I I I
0 10 20 30 40 50
Diameter of C particles, d, (pm)
Figure 20. The effect of the particle size of carbon on the combustion wave velocity in the synthesis
of TIC [26].
2.3 Effect of E.uperimentul Purameters 339
1.5
1.o
0.5
0 20 40 60
Porosity ( % )
Figure 21. The effect of porosity on the wave velocity in the synthesis of TiNi [31].
results show a relatively common pattern in which the wave velocity and com-
bustion temperature have maximum values at an optimum relative density [213.
The relative density has three roles in the SHS reactions. It influences: the energy
density (e.g. J ~ m - of~ )the reaction; the kinetics of the reaction through changes
in the particle-particle interfacial contact; and the wave propagation dynamics
through changes in the effective thermal conductivity of the reactants. At low rela-
tive densities, the velocity and temperature are expected to be low because of all the
three factors listed above. On the other hand, at high relative densities the effective
thermal conductivity is so high that heat is conducted at such a high rate to distant
regions ahead of the wave making it impossible to reach the ignition temperature in
the layer immediately ahead of the wave. Although the existence of a maximum is
not always observed experimentally, the concept of wave extinction at high densities
is always verified. Highly dense reactant powders cannot be ignited. The role of
particle size in this behavior is shown by study on the synthesis of TIC, Fig. 22
[26]. Here the particle size of carbon is changed while holding the size of the Ti
particles constant at 25 pm. With small carbon particles (1 pm), a maximum in
40
4(w)p , ( w
0 Shkiro and
3 f AA Borovinskaya 0. I 20
h
A A Present 1 25
30 8 0 10 25
v
v " 25 25
0
.-
0
I OA 0
20
>
M
'gm 10
"
n
0.4 0.5 0.6 0.7 0.8 0.9 1.0
Relative density
Figure 22. The effect of relative density of the reactants on the wave velocity in the synthesis of TIC
P61.
340 2 Self-Propagating High-Temperature Synthesis of' Hard Materials
the wave velocity is observed at a relative density of 55%. However, regardless of

density, wave propagation was in the pulsating mode when these powders were
used. In contrast, no maximum is observed when relatively large particles (25 pm)
were used and in this case wave propagation was in the steady-state mode over
the entire range up to the extinction limit. The results further show that this limit
decreases as the particle size increases. For example, the limit is about 85% relative
density for samples with carbon particles of 1 pm diameter and it is about 75% rela-
tive density for samples with carbon particles of 25 pm diameter.
The role of relative density in gas-solid synthesis is significant for a different
reason. This case relates to the synthesis of nitrides for the category of materials
of interest to this chapter. Porosity in this case is essential to maximize the interfacial
contact area between the metal and nitrogen gas and to provide a conduit for the
replenishment of the gas to maintain the local supply at the interface. In principle,
the latter requirement can be negated if the total pores of the solid reactant contain,
at the outset, the needed number of nitrogen gas molecules for complete conversion
to the nitride phase. Permeation of the gas through the pores would not play a
significant role in such a case. An analysis has been made to provide the following
relationship between the degree of conversion to the nitride and the pressure of
nitrogen within the pores [32],
v = {(1/S)f'n[P/(1 VrnlIRT, (8)
where 7 is the degree of conversion to the nitride, S is the stoichiometric ratio of
the nitride (i.e. number of moles of N2 per mole of metal in the reaction), Pn is the
pressure of the nitrogen gas, V, is the molar volume of the metal, p is the initial
porosity of the sample, R is the gas constant, and T is temperature. Applying
Eq. (8) to the synthesis of selected refractory nitrides results in the curves of
Fig. 23. The figure shows, for example, that complete conversion of Ti to TIN
with an initial porosity of 0.7 would require a pressure of nitrogen of about
2 x lo3 atm. The corresponding value for the case of BN is about 6 x lo3 atm.
This does not imply that complete conversion would be accomplished with such
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Figure 23. The dependence of the degree of completion of conversion to TiN on the pressure of
nitrogen gas in preexisting pores: solid lines, 1000 K; dashed lines, 1500 K [32].
2.3 Effectof Experimental Parameters 341
10' ' I
4.0
I
6.0
I
8.0
I
10.0
I
12.0
I
14.0
V,"(cm')
Figure 24. A generalized relationship between the pressure of nitrogen and the molar volume of
elements with useful nitrides at complete conversions. T = 1500 K, porosity = 0.5. Nitride
phases: A = M3N4,B = MN, C = M,N,, D = M,N [32].
a pressure, since Eq. (8) does not account for the kinetics of the process. It only
provides an indication of the significance of permeation in the synthesis of the
nitrides. This can be better demonstrated by Fig. 24 in which the pressure for com-
plete conversion is plotted as a function of the molar volume of the metal and
shown for homologous series of nitrides. It can be seen from this figure, for
example, that the nitriding of Hf and Zr (to form HfN and ZrN) is less affected
by permeation than B (to form BN).
Experimentally, the role of porosity (relative density) in nitride formation by
SHS is complex and depends on phase transformation (as it affects permeation)
and the diffusion of nitrogen atoms through the metal or the nitride. The influence
of the former relates to the melting of the metal. If the combustion temperature
exceeds the melting point of the metal, then permeation is affected by the
concomitant change in the continuous porosity. Such a behavior is shown in
Fig. 25(a) where the degree of conversion to TIN is plotted against the relative
density of the starting Ti powder for synthesis in one atmosphere nitrogen pressure
[33]. Maximum conversion occurs at a relative density of about 55% and the
decrease with subsequent increase in density is due to a decrease in the local avail-
ability of nitrogen resulting from the decrease in permeability. The decrease on the
other side, that is, as the density decreases from 55%, is due to the melting of Ti
and its influence on the open porosity in the Ti reactant. For this end of the curve,
the surface area of the Ti in contact with gaseous nitrogen is high, leading to a
higher reaction temperature and more extensive melting of the titanium [34]. In
systems where the combustion temperature does not exceed the melting point of
the metal, the dependence of conversion on density does not show a maximum
but, as predicted, decreases with an increase in the relative density, as shown in
Fig. 25(b) for the synthesis of NbN [35]. As can be seen from Figs 25(a) and
(b), complete conversion to the nitride phases is not accomplished under the
experimental conditions of one atmosphere of nitrogen. In the case of TiN,
342 2 Self-Propuguting High-Tetnpeniture Synthesis of Hurt1 Muterid\
+
8
a 40
I I I I I I
45 50 55 60 65 I0 15
Relative density of Ti pellets (%)
(a)
1
48
44
3
-
0
z
2
0
40
- 0 0
* 0
-
3 36
.-
v1
0
32 - 0
28 -
0
,,LLL
0
2056 57 58 59 60 61
62
Relative Density (%)
Figure 25. The effect of relative density of the metal compact on the combustion wave velocity: (a)
for the synthesis of TIN, (b) for the synthesis of NbN [33,35].
complete conversion can be achieved with the appropriate control of the melting of
Ti by the addition of a diluent.
2.3 Synthesis of Dense Materials

When the temperature of the SHS process exceeds the melting points of all product
components, the product can be highly dense without the use of external pressure.
Such products are referred to as ‘cast SHS materials’ and are primarily the result of
thermite reactions [36]. However, in other than these extremely exothermic reactions
the products tend to be highly porous. While this is probably the most important
disadvantage of the S H S process, it can actually be an advantage in some special
cases, as in the synthesis of electrode materials [37].A direct relationship between
the initial and the final densities is observed, see Fig. 26 for the case of T i c [30]. How-
ever, attempts to use highly dense reactants, taken as a step towards ameliorating
2.3 Synthesis of Dense Materials 343
45
5
h
,x 40
.-
A
.
+
5
*
35
rt;
30 - I I I
45 50 55 60 65
Uncombusted reactants density (YO)
Figure 26. The relationship between the relative densities of the reactants and product in the
synthesis of TIC [30].
this problem, cannot overcome the primary intrinsic source of porosity (see below).
Thus the need to densify the product through simultaneous or subsequent applica-
tion of pressure is of major concern in SHS synthesis.
The sources of porosity in SHS products have been analyzed and classified as
intrinsic and extrinsic 1381. The extrinsic sources of porosity are the initial porosity
of the reactants and the porosity generated by the expulsion of volatile impurities
during synthesis. The contribution of the latter is difficult to calculate exactly, but
an approximate relationship has been derived which shows the dependence of the
level of porosity on the amount of impurity as
up + CipmRTIMiPc, (9)
where wp is the volume of pores per unit volume of product, Ci is the concentration
of the impurity in the reactants, pm is the theoretical density of the product material,
Mi is the atomic or molecular mass of the impurity, P, is the pressure to overcome
the confining forces in the powder, R is the gas constant, and T is the absolute
temperature. Results of such an analysis for the synthesis of TIC are plotted in
Fig. 27 for selected (assumed) impurities. The calculations were made assuming a
combustion temperature of 3000 K (the adiabatic temperature of TIC is 3210 K)
and the confining pressure, P, (the pressure needed to push the compacted particles
apart), was assumed to be 200 atm. The figure shows that at an impurity level of, for
example, lOOOppm the volume percentage porosity can range from about 4.5% for
Ba as an impurity to about 26% for Mg.
Intrinsic sources of porosity are the molar volume change (due to the formation of
the product) and the porosity generated by the thermal migration. Figure 28 shows
the relationship between the molar volume of the reactants and the product for
selected refractory materials. The lines represent a value of x in the relationship
V , = xVR, where Vp and VR are the molar volumes of the product and reactants,
respectively. The figure shows that the formation of TaC, B4C, for example results
in a porosity of about lo%, while the formation of Sic, MoSi2, results in a porosity
of about 30%. The other intrinsic source porosity is thermal migration (the Soret
344 2 SelflPropagating High-Temperature Synthesis of' Hard Muteriuls
100
Impurity:
.-
VI
a
.s 1.o
Tic : T = 3000K
2. 0.1
1.o 10 1o2 10' 10'
Impurity level, ppm
Figure 27. The contribution of impurity volatilization to pore formation in the synthesis of TIC [38].
effect). As a consequence of a temperature gradient, a nonuniform vacancy concen-

tration can arise, even in pure materials [39]. The coalescence of the vacancies can
give rise to porosity. The extent and location (relative to the temperature gradient)
of the excess vacancies from this effect depends on the magnitude and sign of the
heat of transport [40]. Calculations have been made to determine the size of the
pore diameter, dp, as a function of the temperature gradient and the result is
expressed as
dp = { (6e2/n)W J R ) ( d ~ / d x ) P3
(WT&- D C / ~31
(10)
where 1 is the length of the sample across which the temperature gradient exists, t is
the time interval over which the gradient exists, Q, is the heat of transport, DH and
Dc are the diffusion coefficients at the high (TH)and low (Tc) temperature regions
of the sample, respectively, and dT/dx is the temperature gradient. Assuming that
the temperature gradient occurs across one powder particle, the ratio of pore
34
v, = xv, x= I .o
30 -
26 -
10 14 18 22 26 30 34 38
v, ( c d
Figure 28. The relationship between reactants and product molar volume for selected compounds
1381.
2.3 Synthesis of' Dense Materirrls 345
30 r-- 50 I
40
50 rl
i
10
4 6 8
log (dT/dX) (Wcm) log (dl'idX) (Wcm) log (dT/dX) (Wcm)
Figure 29. The effect of temperature gradient on the pore/grain size ratio for: (a) Ti + Ni; (b)
Zr + 2B; (c) Ta + C [40].
diameter to particle diameter was calculated and the results are shown in Fig. 29 for
the three selected examples of TIN, ZrBz, and TIC [40]. For each case, curves are
shown as a function of combustion wave velocity, Slower wave propagation and
higher temperature gradients give rise to higher porosity from thermal migration.
Fi ure 29 shows that under realistic conditions of temperature gradients (lo5-
10f Kcm - ') [lo] and wave velocities (l-lOcms-'), the ratio d p / l ranges from 2 to
10%. However, it should be noted that while thermal migration can give rise to
pore formation, experimentally it is difficult to isolate its contribution. All of the
sources discussed above, in some combination, contribute to the overall porosity
of SHS products, which can be as high as 50%.
With few exceptions, the general desire is to prepare dense materials by the SHS
process and thus several simultaneous or sequential steps are taken to densify these
materials. In the former, the synthesis and densification is achieved in one step, and
in the latter the combustion process is followed, within a short time, by a densi-
fication process. The idea is to take advantage of the plasticity of the hot product.
Examples of the sequential densification include the application of uniaxial pressure
[4 1-44], hot-isostatic pressing (HIP) [45,46], pseudo-HIPing [47], dynamic
(explosive) compaction [48-5 11, high speed forging [52-541, and pressureless and
microwave sintering [55-571. Although the relatively rapid nature of the SHS
process was expected to give rise to products with high defect concentrations,
experimental observations on the sintering of SHS produced materials have not
substantiated this expectation [58,59]. In a related investigation, it was found that
shock compressed Ti + C powders could be reaction-sintered to give a highly refined
microstructure (grain size of 6 pm) [60].
The simplest form of densification, uniaxial pressing, has been used to densify a
variety of materials such as TIC + NiAl, TIC + Ni3Al [41], T i c + Ni, TIC + Fe [42],
and others [43,44]. Two parameters play an important role in this type of post-
reaction densification: the magnitude of the applied pressure and the time of its
application relative to the completion of the combustion process. Figure 30 shows
the dependence of the residual porosity in Tic-Ni/Al cermets on the applied
pressure. As the figure shows, the relative density increases from about 50 to 99%
as the pressure is increased from ambient (0.1 MPa) to about 28 MPa [41]. The
timing of the application of the pressure relative to the synthesis reaction has also
346 2 SelJ Propagating High-Temperature Synthesis of Hard Materials
Pressure (MPa)
Figure 30. The effect of the applied pressure on the residual porosity in the synthesis of: 0,
TiC-
NiAI, and W TiC-Ni3A1[41].
been investigated [42,61]. Fu et al. examined the various parameters related to pres-
sure application [42]. Figure 31 shows a schematic diagram of the pressure-time
parameters. The time for wave propagation, tw, is followed by a delay time, td,
before the pressure is changed from the ambient, Po, to the desired (high) pressure,
Ph. The time duration through which the pressure is maintained at P h is designated
as t,, as shown in the figure. The effect of the delay time on the (relative) density of
the composites 75 weight-% TiB2-25 weight-% Fe and 75 weight-% TIC-25
weight-% Ni is shown in Fig. 32. The results show that for these two cases, the
highest densities are obtained with little or no delay time (td 5 1 s). In both cases,
the time at pressure, t,, was 10s but the values for P h were 95 and 80MPa for
the Fe and Ni-containing cermets, respectively. The effect of the time of application
of the pressure is shown in Fig. 33. For both cermets, the densification approaches
an asymptotic value after a hold of about one minute. Furthermore, and in quali-
tative agreement with the work of Dunmead et al. [41], the final density depended
on the level of the applied pressure, Fig. 34. The difference in behavior of the two
Time
Figure 31. Schematic representation of the sequence of pressure application during SHS synthesis
and densification [42].
2.3 Synthesis of' Dense Materials 341
1 00
75TiB1-25Fe
t,= I0 scc
98'
5
h
92
.c
Q
88
84 75TiC-25Ni
t,=l 0 sec
P = 80 MPa
80
2 4 6 8 10
td(set)
Figure 32. Effect of the delay time in pressure application on the relative density of TiB2-Fe and
Tic-Ni composites [42].
material systems is believed to be a consequence of wettability differences. For

example, the addition of Mo improves the wettability in the 75 weight-% TiB2-25
weight-% Fe system such that when small amounts of Mo were added, a density of
98.5% could be reached. Thus, in addition to the intrinsic mechanical behavior of
the materials, the parameters of temperature and wettability play major roles in
the densification process. Poor wettability of the ceramic by the molten metal can
result in a nonuniform microstructure, and in the case of the pseudo-HIPing, it
can give rise to the expulsion of the metal phase from the composite to the surround-
ing pressure medium (e.g. silica) [62]. The difficulty in assessing the roles of these
parameters independently makes it necessary to rely on specific experimental data
for any given systems, but with the above consideration taken as a general guideline.
100
99
5
h
98
6 0 -
1
Q 4 0
75TiBJ25Fe
i
97 t,= 1 sec
P = YSMPa
96
I
0 20 40 60 80 100 120
t, (set)
Figure 33. Effect of the time of pressure application on the relative density of Tic-Ni and TiB2-Fe
composites [42].
348 2 Self Propagating High-Temperature Synthesis of Hard Materials
I00
99 tp= 10 sec
98
g 97
," 96 75TiBJ25Fe
0.
t"= 1 sec
95 tp= 10 sec
93
94 I
U
40 60 80 100
L
120
Pressure (MPa)
Figure 34. The effect of the applied pressure on the final density of Tic-Ni and TiB,-Fe composites
~421.
Simultaneous synthesis and densification has been attempted using different

experimental approaches. In the case of nitride cermet synthesis, the application
of a nitrogen pressure of 4.0MPa resulted in TiN/Ti-Ni cermets with relative
densities in excess of 97% without the application of external pressure [63].
Another approach for the simultaneous synthesis and densification is through
shock wave application [64]. The pseudo-HIPing technique has also been used to
simultaneously synthesize and densify T i c tiles [65]. For such large samples
(about 10 x 10 x 2.5cm), the formation of cracks during cooling is of major
concern and whose minimization requires the use of a less conducting pressure
transmitting medium to effect a slower cooling rate. Recently, a process was
developed to simultaneously synthesize and densify materials using a modified
field-activated combustion synthesis [66].
2.4 Synthesis by Field-Activated Self-propagating

High-temperature Synthesis
The occurrence of SHS reactions is predicated upon the ability of a reaction front to
produce heat at a magnitude and a rate which can raise the temperature of the
adjacent layer to the 'ignition' point, and thus sustain the process in the form of a
moving reaction front or combustion wave. However, for thermodynamic and
kinetic reasons this process may not be possible, as is the case for the synthesis of
many important materials. Referring to Eq. (l), a thermodynamic limitation is
operative when the reaction enthalpy, Q , is low and this is the general basis for
the use of the empirical limit of the adiabatic temperature discussed in an earlier
2.4 Synthesis hj. Fielcl-Activuted Sdflpropaguting High-trmperature Sjnthe.si.r 349
part of this chapter. Kinetic limitations of SHS reactions are dominant when the
rate of the reaction, &1/8r, is too low or when the thermal conductivity, K , is not
optimum. Too low a K value leads to inadequate heat build up in the reactant
layer just ahead of the front and hence the absence of ignition conditions. Similar
end results are obtained when, for too high a K , the heat is diffused to regions far
ahead of the front. Because of a low reaction enthalpy (thermodynamic limitation),
materials such as B4C, Sic, WC, Cr3C2,Mo2C, and others cannot be synthesized by
a normal SHS process. Kinetic limitations have been given as reasons for the
inability to synthesize such materials as TaC. AlN-SIC, and others. Many com-
posites of hard materials where the formation of one of the involved phases is
energetically lethargic cannot be synthesized by SHS because of these limitations.
These include such materials as MoSi,-SiC, .xB4C-TiB2 (for x > 0.5), and others.
The common approach to activate these reactions is to preheat the reactants in
order to either raise the adiabatic temperature for a subsequent SHS reaction or
to cause a reaction in a simultaneous combustion mode, as discussed above. This
is done through furnace heating or by a ‘chemical oven’ [67]. The latter involves
the placement of the reactants within a powder mixture of other more exothermic
reactants. However, in both of these methods diffusional reactions can occur
before the main combustion reaction and can result in the formation of extraneous
(undesirable) phases in the product [68,69]. To overcome the limitation of the SHS
process and thus facilitate the synthesis of a large number of materials systems, a
new process was recently developed in which the activation is effected by the
imposition of an electric field [70]. A schematic representation of this method of
activation is shown in Fig. 35. Two electrodes are placed in contact with opposite
faces of the compact containing the reactants and a voltage is applied across
them. In most cases, the application of a field (within limits) does not result in the
initiation of a combustion wave. Such a wave, however, can be initiated and
propagated after the addition of energy from an ignition source, in the usual way.
In all cases investigated, the initiation of a combustion wave is not accomplished
below a minimum (threshold) value of voltage (or voltage per sample thickness,
Product Reactants
Figure 35. Schematic diagram of the field-activated combustion synthesis

350 2 Self Propagating High-Temperature Synthesis of' Hard Materials
0 5 10 15 20 25
E (V/cm)
Figure 36. The effect of the applied field on the combustion wave velocity in the synthesis of p-Sic.
field). Figure 36 shows that in the case of the synthesis of Sic, the threshold is about
6Vcm-' [71]. The figure further shows that there are three regions in the depen-
dence of wave velocity on field strength. As indicated, below the threshold no
wave propagates, and above the threshold the wave velocity increases nearly linearly
with field. At relatively high field values, above about 20Vcm-', the reaction
between Si and C does not require an ignition source but takes place in a simulta-
neous mode. In this case the Joule heating by the current is sufficient to ignite the
reaction. As will be discussed in more detail, the role of the applied field is believed
to be primarily thermal in nature, providing additional (Joule) energy. In most
cases, this energy is localized to the narrow reaction zone, as will be shown later.
Real-time measurements of the changes in voltage and current during wave
propagation provide the results shown in Fig. 37 for the synthesis of Sic [71].
Both the measured parameters of current and voltage and the calculated parameter
of resistance remain reasonably constant during wave propagation, as indicated by
the interval between the start (S) and end (E) of the process. In view of the fact that
the electrical conductivities of the reactants and product are significantly different,
these observations are taken as indicative of a current that is confined to the com-
bustion wave. Modeling and other experimental studies have provided support for
this conclusion, as will be discussed below.
The modeling study was based on a modification of the applicable Fourier rela-
tionship, Eq. (1). The modification includes the addition of another heat source
from the Joule heating by the current. The rate of heat generation from this
source is uE2, where u is the electrical conductivity and E is the applied field. The
modeling was initially done on the synthesis of Sic, using a two-dimensional
approach and accounting for heat loss by radiation, to bring the results closer to
’7.4 Syirtliesis h!, Field- Activuted Self+vpagating High-temperature Synthesis 35 1
20 , 1
15
d
‘c)
10
>
-
5
0
0 5 10 15
time (s)
Figure 37. Variations of real-time voltage and current and calculated resistance during wave
propagation in the synthesis of P-SiC [71]: S and E signify the start and end of wave propagation,
respectively.
real experimental conditions [72]. The modified relationship is

+ +
pC,(dT/dt) = K [ ( ~ ’ T / ~ . Y ’( a) 2 T / a y 2 ) ] Q(i3v/at) oE2 + + ( o s / W ) ( T-4 T:),
(11)
where o,is the Stefan-Boltzmann constant, W is the thickness of the sample, and To
is the ambient temperature. In this modeling, the parameters C,, 6,and o are taken
as functions of composition and temperature and the reaction is assumed to obey
second-order kinetics. The modeling analysis provided results which were qualita-
tively consistent with the experimental observations. Of these the propagation
mode and the localization of the current are the most important results. Figure 38
shows conversion (v) profiles calculated at constant time intervals during wave
propagation for the modeled synthesis of Sic. The equal spacing between the
profiles is indicative of a steady-state wave propagation, in agreement with the
experimental observations. The corresponding current density profiles are shown
on Fig. 39. Two important aspects of this figure are noteworthy: the current is
0
x (cm)
Figure 38. Calculated conversion profiles for the synthesis of S i c with applied voltage 20V [72].
(E=14.2 V/cm)
0.0 0.5 1.0 1.5 2.0

x (4
Figure 39. Current density profiles for the synthesis of S i c [72].
confined to a very narrow zone; and the confinement zone is located in the leading
edge of the combustion wave. The confinement of the current to the reaction zone is
an obvious advantage when the energy efficiency of this form of activation is consid-
ered. However, this situation is not universally true but is dependent on the electrical
conductivity of the product relative to the reactants. The implication of this is that in
systems where the product is more conducting, the current is not confined to the
reaction zone but is, at least partially, carried by the product. Both experimental
and modeling studies have provided support for this expectation [6].
The application of a field has been shown to have the following effects in combus-
tion synthesis: it makes possible the synthesis by SHS of materials which, previously,
could not be prepared by this method [7&74]; influences the nature and the relative
amounts of product phases [75,76]; makes possible the synthesis in relatively dense
reactants [77]; and influences the microstructure of the product [78]. A few examples
of these will be discussed here and more details on the others are provided in the refer-
enced papers. An example of the benefit of the field is demonstrated by the synthesis of
B4C-TiB2 composites. With a goal of producing hard and light materials for armor
application, the desire is to increase the relative content of the carbide phase in this
composite. With normal SHS, composites with a B4C/TiB2molar ratio, y , of 50.5
can be synthesized without preheating the reactants [79]. While composites with
higher values y can be synthesized by preheating the reactants, the product contained
other extraneous phases (e.g. TIC). With application of a field, composites with y
values up to 5 could be synthesized [80]. Figure 40 shows the effect of the field on
the wave velocity in the synthesis of B4C-TiB2 composites with 1 5 y 5 5. As in
other field-activated combustion synthesis, a minimum (threshold) field value was
necessary to propagate the combustion wave. The threshold value depended on the
value of y , as seen in Fig. 41. Higher threshold values were required for higher
molar ratio 01) values. The products in all cases contained the two desired phases
only. Since the imposition of the field is a requirement for the synthesis of composites
2.4 Synthesis by Field-Activated Selfpropagating High-temperature Synthesis 3 53
H B,C/TiB,=I
0 B,CmiB,=2
0 B,C/TiB,=3
0 B,C/TiB,=5
0 10 20 30 40 50 60
Voltage ( V )
Figure 40. Effect of field on the wave velocity for the synthesis of yB4C-TIC composites with
1 5 y 5 5 [SO].
with y > 0.5, the removal of the field during wave propagation affords the oppor-
tunity to investigate the mechanism of formation of the two phases from elemental
reactants. Figure 42 shows the X-ray diffraction (XRD) patterns obtained from a
B4C-TiB2 composite with y = 1 in which the wave was frozen by the removal of
the field [80]. The upper pattern was obtained from the region behind the combustion
zone, and as can be seen it contains only the peaks for B4C and TiB2, the product of
synthesis. The middle pattern is for the combustion zone itself and shows the presence
of TiB2 and carbon (no peak for boron is expected since it was used as an amorphous
powder). Finally, the lower pattern is for the area ahead of the wave and shows the
presence of the reactant elements only. These observations indicate that the
formation of TiB2 is the first step in the composite formation, an expected
conclusion in view of the large difference in the thermodynamic driving force for
the formation of the two compounds. It should be recalled that B4C cannot be
formed without field activation.
2 2
F
Y
2 0
1 8
m
-o
C
1 6
>
8
0 1 2 3 4 5 6
B,C/Ti B, Molar Ratio
Figure 41. Effect of the molar ratio B,C/TiC on the threshold voltage in the combustion synthesis of
B,C-Tic composites.
354 2 SelfiPrapagating High-Ternperuture Synthesis of Hurd Materials
reacted region
300 TiR,
100 1 'fi unreacted region
0
25 30 35 40 45 50 55 60 65
2 theta
Figure 42. X-ray diffraction patterns of frozen reaction zone with analyses for the reacted region,
combustion zone, and unreacted region [80].
Another example of the role of the field in synthesis is that related to the prepara-
tion of tungsten silicides [75]. The two stable silicides, WSi2 and W5Si3, have
relatively low adiabatic combustion temperatures (1459 and 954 K, respectively)
and thus cannot be synthesized by SHS. However, both can be synthesized by
field activation. For WSi2, a combustion wave can be initiated above a threshold
value of about 15Vcm-' and the product is the desired phase. On the other
hand, the case of W5Si3 is more complex. Although a combustion wave can be
initiated above a threshold field of about 14Vcm-', the product does not contain
the desired phase unless the field is above about 19 Vcm-I. Microstructural obser-
vations on the products for the case of W5Si3showed that under the higher field the
product contained a liquid phase and it was proposed that the formation of the
desired phase was aided by the higher diffusion rates in the liquid phase.
The formation of composites and solid solutions in the binary A1N-SIC offers
another important example of the role of the field in the synthesis of materials.
The two compounds form extensive solid solutions with the hexagonal 2H structure
above about 1960°C [81]. The spinodal decomposition of these, at lower tempera-
tures, gives rise to the formation of composites modulated at the nanoscale [82].
The common method to synthesize the solid solutions is to heat mixed powders
of AlN and Sic (aor p) at 2100°C for up to 16h. As will be seen here, using
+
field-activated synthesis for the reaction S3N4 4A1+ 3C, the formation of solid
solutions can be effected in less than a minute [76,83]. Although this reaction is
highly exothermic (with an adiabatic temperature of 2504 K), it does not sustain a
self-propagating combustion wave without the application of a field above a
threshold value of 8.0 V cm-' . At this field, however, the reaction is not complete,
with the product containing SiC-rich and A1N-rich 2H solid solutions and Si and
C . The reaction becomes complete at a field of 12.5Vcm-' but the product is a com-
posite of two solid solutions. When the applied field is 25.0 V cm-I, the product is a
single phase solid solution. Careful analysis of the product phases showed that their
compositions changed with the strength of the applied field, as shown in Fig. 43. As
the field increased, the compositions of the two phases moved closer to each other
2.4 Synthesis hy Field-Activated SelJlpropagating High-temperature Synthesis 3 55
0.8
0.6
zd 0.4
0 WDS-AIN
o WDS-Sic
0.2 A XRD-AIN
x XRD-Sic
n I I
12.5 20 25 30
Field (V/cm)
Figure 43. Effect of the applied field on the phase composition of combustion-synthesized AIN-SIC
composites and solid solutions [76].
Figure 44. X-ray dot maps of A1 and Si in AIN-SIC solid solutions prepared under fields of 25 (left)
and 30Vcm-' (right) [76].
356 2 SeIf-Propagating High-Temperature Synthesis of Hard Materials
and at 25 V cm-' , the compositions are the same, indicating the formation of the solid
solution. Increasing the field to 30 V cm-', had no effect on the XRD generated com-
position, but a more significant effect on the elemental distribution, as seen in Fig. 44.
Although the XRD results indicate that the product with 25 V cm-' is single phase,
the elements (A1 and Si) are not uniformly distributed in the sample. However, a
nearly complete uniform phase is obtained when the field was 30 V cm-' . These obser-
vations are further evidence to the role of the field in the synthesis of materials.
2.6 Selected Recent Examples of Synthesis of Hard

Materials
During the past few years, the use of the SHS method to obtain technologically
important hard materials has been primarily focused on the synthesis of composites,
mainly TiC- and TiB2-based materials. Only in a few cases has the synthesis of
monolithic hard compounds been considered. In these cases the emphasis is on
the development of new techniques (such as the field-assisted combustion synthesis
described in the previous section) or on the understanding of the reaction mecha-
nism [5,7,8,84,85]. While the primary goal is to synthesize dense materials, the
production of porous materials, to be used mainly as a source of powders, has
also been investigated. In this case the powders can be the final desired product
(e.g. for use as abrasives) or can be subsequently densified. The emphasis on the
synthesis of dense materials is motivated by the desire to develop new processes
to replace the traditional techniques for the production of, for example, cutting
tools. The traditional methods for the fabrication of these materials typically
involve powder metallurgical techniques which are both labor and energy-intensive.
The extremely high temperatures, typically attained in the SHS process, makes it
feasible, at least in principle, to synthesize and densify pure carbides, through the
simultaneous application of pressure to the combustion products when they are
still above the ductile-brittle transition temperature, which for T i c is around
1700°C. Early attempts to utilize this approach were only partially successful, as
they were plagued by the macrocracking of the final products, probably due to
the presence of residual thermal stresses [86]. An example of such an approach is
the work of Wang et al. [87], who used explosive compaction on combustion syn-
thesized TIC and TiB2. The dynamic compaction followed the SHS process with a
time-delay of 5-20 s, with the best results being obtained with a delay time of 5 s. The
compacts produced had relative densities higher than 99% for TiB2 and 96.8% for
TIC. The microhardness, elastic modulus, and compressive strength of TiB2 were
comparable to those reported for hot-pressed materials. For the case of Tic, the
microhardness and elastic modulus (NO0 and 46 700 kg mm-2, respectively) were
equivalent to those obtained from hot-pressed samples, but the compressive
strength (103 kg mm-') was lower, presumably because of extensive cracking in
the SHS-prepared samples. Samples of TiBz were less prone to cracking, probably
because the melting point of TiBz (2900°C) is exceeded during the synthesis while
that of TIC (3140°C) is not likely to be reached during real (i.e. nonadiabatic)
conditions. Moreover, it was observed that the grain boundary strength of the
TIC and TiB2 prepared by the SHS-dynamic compaction method were relatively
weak. This conclusion was based on observations of grain boundary failure
during hardness measurements and was explained by a formation of microcracks
along grain boundaries during the rapid cooling. Support for this conclusion was
provided by significant improvements in mechanical properties after annealing for
several hours at 1500 and 2000°C. As pointed out above, the existence of cracks
is related to the rate of cooling subsequent to synthesis and densification. The use
of a pressure medium in the 'pseudo-HIPing' approach is believed to ameliorate
this problem by decreasing the rate of cooling of the densified sample [65].
In order to improve the mechanical properties of such monolithic hard materials,
a second phase is usually added. In many cases the second phase is a metal and thus
the formation of a cermet is accomplished. The addition of a ductile phase acts to
reduce the residual stress and to increase the fracture toughness of the material.
The characteristics of the combustion processes aimed at the formation of metal-
ceramic composites based on T i c and TiB2 have been investigated in several
recent studies. Capaldi and Wood [88] investigated the SHS processes of syn-
thesizing TIC cermets with the metallic component being iron, nickel, copper, and
manganese. Notable differences in the process characteristics were observed,
depending on the nature of the metal additive (binder). Figure 45 shows the
dependence of wave propagation velocity on combustion temperature for all four
metals. These results imply differences in the activation energy of the synthesis
process, see Eq. (6). Contrary to what may be attributed to the role of the metal
as an inert component, these results suggest that the metal plays a more direct
role in the synthesis. Thus differences between the metals in terms of influencing
the kinetics of the process may arise from possible differences, for example, in the
formation of intermediate intermetallic phases, or differences in the kinetics of a
dissolution-precipitation process responsible for the formation of the ceramic
component [88]. The possibility of formation of low melting phases, represented
-8.0
-8.2
-8.4
5
9
-8.6
-2 -8.8
3 -9.0
-2
u
-9.2
-9.4
-9.6
-9.8
- 10.0
3.5~10' 4.0~10' 4 . 5 ~ 1 0 5.0~10'
~ 5.5~10.~
I/T, (I/K)
Figure 45. Apparent activation energy determination for Tic, Fe-Tic, Ni-Tic, and Cu-Tic [88].
by eutectics between Ti and the binding metal, can alter the kinetics of the dis-
solution-precipitation process involved in the combustion synthesis of carbides.
Another difference in the synthesis of these cermets is represented by the variation
in the ‘ignition temperature’. In some cases, for example in the system Ti-C-Fe, the
ignition temperature is well below the melting point of the pure Ti metal, an
occurrence which has been subject to much debate in the early SHS literature.
The prevalent point of view has been that ignition occurs after the melting on the
metal. In the work of Capaldi and Wood [88], however, when iron was used as
the binder, titanium and iron react in the solid-state producing FeTiz which is a
eutectic compound with a melting point of 1085°C. In the case of the use of Mn,
the T i c that formed was carbon-deficient relative to that formed with other
metals, probably because of the formation of manganese carbide.
The synthesis of cermets in dense form have been reported by several authors. La
Salvia et al. reported the results of high-speed forging of TiC-Ni [54] and TiC-Ni-
Mo [89] cermets. In both cases the high speed load was applied 5-10 s after the end
of the combustion process, when the samples’ temperature was still above the
ductile-to-brittle transition. In the first case cermets containing 5 , 25, and
30 mol% of the metal were prepared. The microstructure was generally character-
ized by spheroidal grains of T i c surrounded by a discontinuous Ni layer probably
due to poor wetting of melted Ni on Tic. A substantial fraction of TIC particles
exhibited bonding and necking, forming a continuous carbide skeleton. The samples
with the lower metal content exhibited a certain porosity due to partial vaporization
of Ni due to the high reaction temperature. The hardness and compressive strength
of these samples were in the range of values of conventionally processed cermets.
This indicates that the bonding between Ni and T i c is excellent. The addition up
to 10 weight-% of Mo [89] was based on the suggestion that its presence improves
the wettability of the metallic phase on Tic. In traditional processing this
improvement in wettability decreases drastically the tendency of T i c grain
growth, reducing the possibility of carbide-carbide grain contacts. In the
combustion-synthesized/dynamically-compacted materials this effect was not so
evident. Figure 46 shows that the influence of Mo content on the size of the TIC
particles is not significant. On the other hand, evidence for incomplete solubility
of Mo within the matrix is observed. However, a decrease in the number of micro-
structural flaws such as voids, interphase debonding, and binder microcracks,
attributed to the improved wetting of the Ni-alloy binder on the carbide phase, is
observed with an increase in Mo content. As a result, the compressive strength,
bend strength, fracture toughness, and Young’s moduli, were slightly improved,
while the Vickers microhardness seemed to be independent of the Mo content.
Another example of synthesis of Tic-based compact cermets is reported by Xing
et al. [90]. These authors describe the synthesis of dense NiA1-20 vol-% T i c com-
posites through a reaction of a mixture of Ni, Al, Ti, and C powders in a hot
press. The reaction was performed using the volume combustion mode: heating
the green mixture until a reaction is observed in the entire sample. The samples
were heated up to 1500°C under the influence of an applied pressure. The products
were 98.9% dense but the T i c particles (0.2-1 pm) were not uniformly dispersed in
the matrix. The combustion process had similar ignition characteristics to the
E
I4.5
I
,
al
.*v1
-.U
0
z I I
4.0 -
.g 3.5 -
3.0 i
0 2 4 6 8 10 12
Mo (wt%)
Figure 46. Influence of the amount of Mo on carbide particle size for combustion synthesized/
impacted forged Tic-Ni based materials [SS].
system Ni-A1, with an ignition temperature of 588°C. This temperature is below the
lowest eutectic temperature in the Ni-A1 system, suggesting that the combustion
process commences through a solid state reaction.
A different approach to the synthesis of dense hard materials is reported by
Degnan and Wood [91]. They describe a process in which Fe-(W,Ti)C cermet
powders, produced by the SHS method, are used as additives to molten iron for
the production of a particulate-reinforced material with improved wear properties.
The presence of the additive provided an enhancement of wetting between the metal
matrix and the added particles. The product was a fairly uniform material which
showed a considerably improved wear resistance over iron-based alloys. Sharivker
et al. [92] followed a somewhat similar route by using SHS-produced TIC or
Tio.97M00.03C as additives to produce metal-based hard materials.
An interesting exception to the synthesis of dense materials through external
pressure application is the work of Ohyanagi et al. [93] in which the formation of a
surface dense TIC by a pressureless process is reported. Such a material was obtained
using reactants with high green densities (6549%). The formation of a fully dense
surface layer appears to be the result of the solidification of a supercooled solid solu-
tion of carbon in molten Ti. The presence of a nonequilibrium solution is suggested
by the existence of an exothermal peak on the cooling curve which became evident in
samples with green densities greater than 76%. The temperature corresponding to
the maximum of this peak decreased when the cooling rate was increased. The sur-
face became completely dense with a texture typical of solids obtained from solidifi-
cation of supercooled liquid. The presence of the liquid only at the surface was
explained by a combination of capillary and gas expulsion effects [93].
In the case of borides, the production of pure dense materials is more difficult
because TiB2 does not deform plastically even at very high temperatures due to
its intrinsically high concentration of Peierls barriers to dislocation movement.
Recent investigations on borides have been totally devoted to the synthesis of
composites either metal-ceramic or ceramicseramic. Woodger et al. [94,95]
2000,
h
5 1900
:1800
*
$
g2 1700
c
.gs 1600
D
u
3 1500
1400- I I
20 30 40 50 60 70 80
Percent relative green density
Figure 47. Influence of green density on the combustion temperature for the synthesis of Al-HfB2
composites for samples with 60 vol-% Al [94].
investigated the formation of dense materials through liquid metal infiltration

obtained by adding large amounts of metal to the reaction mixture. The infiltration
was done with pure A1 or with Ni3Ti intermetallic compound. In the case of
HfB2+ Al, up to 70 vol-% A1 was added to the reacting mixture. An unusual
effect of the green density of the reactants on the combustion temperature and
propagation rate was observed, Figs 47 and 48. Initially, as the relative density
increased, both the reaction temperature and propagation rate decreased, but
with still higher densities, these parameters increased dramatically. Such a behavior
is attributed to an assumed enhancement of liquid spreading at the reaction front for
reactants with high green densities. However, the presence of a large amount of
liquid aluminum does not lead to the formation of highly dense products. The
relative density of the products is primarily related to the density of the green com-
pacts, and it decreases with the amount of the metal. Furthermore, sedimentation
h*
$
v 0.35
.-3
-0
0
.>”
0.30 -
s
0.25 --
20 30 40 50 60
70 80
Figure 48. Influence of green density on the wave velocity in the synthesis of Al-HfB2 composites
for samples with 60 vol-% Al [94].
2.6 Selected Recent Examples of’Synthesis of Hard Materials 361
1600 1 T
92 1550 -
f 1500 -
2 1450
2 -
C
2
.3
1400 -
3E
P
-
1350
I , I I I
35 40 45 50 5 5 60 65 70 75
Figure 49. The influence of green density on the combustion temperature for the synthesis of Ni,Ti-
TiB2 composites. Samples containing 70 vol-% Ni,Ti [95].
effects, due to the large density difference between A1 and HfB2, were observed
together with the presence of small amounts of Hf-A1 intermetallic phases.
In the case of TiB2,cermets with up to 90 vol-% Ni3Ti have been synthesized from
elemental reactant powders. An unusual dependence of the combustion temperature
on relative green density was also observed, Fig. 49. The bimodal nature of the
combustion temperature curve was explained by a two-mechanism process. The
first peak in combustion temperature is likely to be related to the highly exothermic
TiB2 reaction in the presence of a small amount of diluent metals. The second peak
can be associated with the improved reaction kinetics resulting from the enhanced
capillary spreading of the molten phase. The presence of the molten phase makes
possible the synthesis of samples with very high relative densities. As shown in
Fig. 50 a rapid increase in relative density with metal content is observed up to
85% relative density, then this is followed by a marked decrease, probably related
to the decrease in the reaction temperature as a consequence of increased dilution
of the reacting mixture. The synthesis of a similar cermet has been reported by
Yang and Petric [96] and Yi and Petric [97] who used elemental mixtures to syn-
thesize Ti3A1(Nb)-TiB cermets. In this case the combustion process was reported
to be initiated by the reaction between Ti and Al, which subsequently ignites the
combustion between Ti and B.
A further example of production of boride-based fully-dense materials is recently
reported by Olevsky et al. 199). The product, a TiB,-TiN-Ni cermet, a material
with a unique combination of high temperature hardness and stability and with a
reasonable ductility/fracture toughness.
Other composite materials based on TIC and TiB2 have been produced recently
through the SHS process. Attempting to develop a process for the production of
ceramic armor, Lis et al. [55] considered the combustion synthesis of TiB2-Tic
composites. Such composites have a lower hardness than either TiB2 and T i c
single phases, but a fracture toughness and wear resistance which are significantly
higher than the corresponding values of the single phase materials. In order to
362 2 SeEf-Propagating High-Temperature Synthesis of Hard Materials
40 ' '
50
I
60
I
70
I
80
I
90
Volume percent Ni,Ti
Figure 50. The effect of Ni3Ti content on final density of Ni,Ti-TiB, composite. All samples were
60% relative green density [95].
reduce the cost of the raw materials, these composites were obtained through the use
of Ti and B4Cpowders as reactants with the anticipated product containing a TiB2/
T i c molar ratio of 2. Under such conditions, the adiabatic temperature is high
enough to reach the lowest eutectic temperature (T, = 2520°C) making it possible
to attain high product density. The combustion process was found to be strongly
dependent on the reactants' particle size. A stable combustion process could be
obtained only for B4C and Ti powders with particle sizes below 10 and 49pm,
respectively. The synthesized composite was then densified by pressureless sintering
at temperatures between 2000 and 2100°C or by hot-pressing at temperatures
between 1800 and 2000°C using a ram pressure of 200 atm. The sintering time for
the combustion synthesized samples was considerably shorter than for mixtures
of commercial T i c and TiB2 powders, and the microhardness of the product
(Hv = 29.5 GPa) was higher than values reported in the literature.
The addition of alumina to TiB2 or TIC has become an attractive approach in
recent years. The addition of alumina to titanium carbide or boride improves the
oxidation resistance and results in composites which possess superior mechanical
strength and fracture toughness, without a significant drop in their overall micro-
hardness. Moreover, these composites can be obtained through SHS using, as
reactant, the relatively inexpensive titanium oxide instead of the metal. In a
recent example, a ceramic-ceramic composite to be used as a grinding material
has been prepared by Vershinnikov et al. [98] using the reaction 3TiO2 + 3Bi2O3+
+
lOAl = 3TiB2 5A1203.Kecskes et al. [50] showed that the dynamic consolidation
technique can be used to produce dense TiB2-A1203 composites through SHS
using the previous thermite reaction. Sample densities of 90% can be obtained by
varying the reaction ignition-consolidation delay time. The TiB2 grain size was
found to be related to that of the precursor Ti02. A detailed study on the influence
of several experimental parameters on product microstructure was reported by
Bowen and Derby [loo] for TiC-A1203 composite always obtained through a reaction
between TiOz and A1 plus C. The combustion kinetics and the product microstructure
O h
0 10 20 30 40 50
Aluminium particle size (pm)
Figure 51. Combustion wave velocity against A1 particle size for the combustion synthesis of TIC-
A1203 composites [100].
result is strongly dependent on the reactant particle size. The wave propagation rate,
for example, was strongly dependent on A1 particle size, but less significantly on Ti02
particle size. In both cases a decrease of propagation rate is observed when the particle
size is increased, as seen in Figs 5 1 and 52. In contrast, when the carbon particle size
was increased from 0.02 to 7.2 pm, the wave propagation increased but then decreased
for larger particle size, Fig. 53. This behavior was explained by an assumed barrier
role played by the extremely fine carbon, decreasing the titanium-aluminum particle
contact, a necessary first step in the reaction, Fig. 54. The effect of diluents on the
reaction products was also considered in the cited work [loo]. When A1203 was
added as a diluent it tended to induce cracking in the final product, probably because
the combustion temperature decreased below the melting point of alumina. On the
other hand, the addition of A1 as a diluent produced an unexpected increase in
wave velocity with increasing diluent content, an observation that can be explained
2 '
0 5 10 15 20
Titania particle size (Fm)
Figure 52. Combustion wave velocity against TiOz particle size for the combustion synthesis of
TiC-A1203 composites [ 1001.
0 10 20 30 40 50
Carbon particle size (pm)
Figure 53. Combustion wave velocity against C particle size for the combustion synthesis of TiCi-
A120, composites [loo].
by an enhanced diffusion process or heat transfer in the presence of a higher liquid

content. The synthesis of TiC-AI2O3-Al composites by SHS has also been investi-
gated by Yi et al. [loll and Kim et al. [102]. In both investigations the products
were densified at the end of the combustion reaction using a hot-press system in
one case [loll and a weight-drop apparatus in the other [102].
In a few cases the SHS process was used as source of heat in order to achieve rapid
densification. An interesting result in this regard is the work of Li et al. [lo31 and
Miyamoto et al. [46], in which fully-dense, multifunctionally-graded A1203-TiC-
Ni-TiC-A1203 materials were produced using a HIP compaction method. The
Figure 54. Schematic diagram showing how small carbon black particles act as a barrier to TiO2-A1
particles contact [loo].
2.6 Selected Recent Examples of Synthesis of’ Hard Materials 365
I 2 3 4 5 6 7
vent
N,
Figure 55. Experimental set-up for SHS Cr7C3coating preparation: 1 =pressure gauge, 2 = reactor,
3 = heater, 4 = container, 5 =igniter, 6 =reactant charge, 7 = substrate [106].
samples, represented by a layered compact, were surrounded by a large amount of Si

powder and were ignited in high pressure nitrogen. The silicon burns in a nitrogen
pressure of above 3 MPa generating a high reaction heat and a temperature of over
2300 IS.The green body was densified within a few minutes. Because of the graded
composition, the products had a stress distribution of compression/tension/com-
pression which resulted in a desired compressive stress in the outer ceramic layers.
Thus far we have considered only the synthesis of powders or bulk densified
materials. However, the SHS approach has been successfully applied to the
production of thick protective layers. A typical example of such a process is the
development of technologies for the application of ceramic linings on hollow
cylindrical bodies by means of thermite reactions under the influence of a centrifugal
force [104]. A related technology, usually referred to as SHS cast-coating, has been
developed in the former USSR [105]. Several types of coatings have been produced
in this way, including carbides, borides, and silicides. A recent example of this
approach is reported by Lin and Xue [106]. In this work a layer of Cr7C3-Cr
cermet was produced on a flat substrate of carbon steel via the SHS cast route. The
+
synthesis is based on the thermite reaction 7Cr03 14A1+ 3C = Cr7C3 7A1203. +
The experimental setup is shown in Fig. 55. A layer of the reactants mixture, a few
millimeters thick, is compressed over the substrate, and a layer of an igniter
powder is placed on top of it. The entire assembly is then heated under an atmosphere
of nitrogen gas until a volume combustion is obtained. The two products of the
thermite reaction are obtained in the molten phase. Since the density of the chromium
compound is higher than that of alumina, a thick layer of Cr-containing phases is
formed in contact with the substrate. After cooling, the loose alumina layer on top
is easily removed. The thickness and composition of the green layers, the gas pressure,
and reactants preheating temperature were found to strongly influence the character-
istics of the final layer. Figure 56 shows that a strongly adherent layer could be
obtained only when using either high preheating temperatures or high gas pressure,
with a complex structure of the green compact, represented by three layers with
different proportions of the reactants. The best results were obtained with a triple
reacting layer, a 300°C preheating temperature and a 5-10MPa gas pressure, or
with a 200°C preheating temperature and a 15 MPa pressure. The surface coating
was primarily composed of Cr7C3and Cr with traces of Cr23C6.Transverse sections
(A) (B)
EI13 Non of any product is formed on the substrate
U Only some droplets are formed on the substrate
IT3 Only some islands are formed on the substrate
About 60% area is covered
111 Covered completely, but not smootly
Covered completely and smootly
Figure 56. Effect of preheating temperature and ambient pressure on the formation of SHS Cr7C3
coating on Cr: (A) single layer of reactants; (B) triple layer of reactants [106].
show that the composition of the layer is not homogeneous: the Cr content increases
and the C content decreases with distance from the top of the coating. Concomitantly,
the mechanical properties change with distance, with the maximum hardness being
measured on the top of the surface. The adhesion of the coating to the substrate
was found to be exceptionally good, even after several thermal cycles.
Considerable interest has been generated during the past few years in the synthesis
of diamond containing materials through SHS. In this regard the advantage of
SHS in avoiding the drawback of the traditional methods of production of
diamond-containing tools has been highlighted. These drawbacks relate to the
necessity to use long reaction times at very high pressures. In the SHS process,
reaction times can be reduced (from several hours in the traditional methods) to a
few seconds. Furthermore, the suitability of the SHS process for the production
of functionally-graded materials is an additional attractive feature. However,
there is one possible drawback in the use of the SHS process: the extremely high
reaction temperatures. Diamond transforms into graphite above about 1400-
1500°C under an inert gas atmosphere or in vacuum. As a result, it is of concern
that the mechanical characteristics of diamond might degrade during a typical
SHS thermal cycle. Theoretical analysis of the combustion process in diamond-
containing mixtures have been proposed by Padyukov et al. [107]. The extent of
diamond grain preservation was shown to depend more on the time the diamonds
stayed in the combustion zone than on the absolute value of the combustion tem-
perature. Surface graphitization, on the other hand, increases with the combustion
temperature. Three different approaches have been proposed in order to reduce the
effect of the combustion temperature on the diamond degradation [108]. In the first
approach two-layer samples are used, as shown in Fig. 57. One contains the pure
reacting mixture, for example Ti + B or Ni + Al, while the second contains the
same mixture diluted with diamond powder. The first layer, without diamond
acting as an inert diluent, has a more rapid combustion rate and acts to increase
the rate of combustion in the second layer. This results in a reduction of the dwell
time of the diamond in the high-temperature region. Moreover, such a procedure
Figure 57. Schematic of the bilayer pellet used for the SHS synthesis of diamond containing
materials [108].
makes possible the incorporation of higher amounts of diamond in the mixture

while still maintaining a self-sustaining process. Another approach is based on the
feasibility of producing a protective reducing atmosphere during the combustion
synthesis process. This can be accomplished by the addition to the reactants of a
hydride with a dissociation temperature well below the reaction temperature.
Titanium hydride (TiH2) is an example of such an addition. For the system
+ +
Ti B diamond an optimum concentration of 5 weight-% of the hydride was
shown to preserve the grain strength of the diamond, Fig. 58 [108]. A third modifi-
cation aims at lowering the reaction temperature by the use of less exothermic
reactions. The use of an intermetallic formation process, particularly in the Ni-A1
system, is an example of this.
The materials used for the incorporation of diamond fall in two categories,
corresponding to two different methods for fixing the diamond to the matrix [47].
In one case the diamond is retained only mechanically, while in the second case a
sort of chemical fixing is obtained through formation of a covalent bond between
the diamond and some component of the matrix. Metal carbides are generally
used in order to form covalent bonds with diamonds. A metal is also sometimes
z
a"
15
14 I I I I I
0 10 20 30 40 50
Mass % ,TIH,
Figure 58. Dependence of diamond grain strength in combustion products of Ti + B + 10 weight-%

diamond system on concentration of TiH2 additive [108].
included in the matrix in order to improve the mechanical properties of the product.
Intermetallics (Ni-A1 or Ni-Al-Cu-Ti), TiB2, and TiB2/Si represent examples of a
matrix of the first kind, while Tic-Si, Tic-Ti-A1, and Tic-TiN-Ni-Mo-A1 repre-
sent a matrix of the second kind. The densification of the products has been
obtained using the typical approaches already used for other SHS processes such
as hot-pressing, hot-isostatic pressing, pseudo-HIP, explosive consolidation, and
high-velocity forging. Ohyanagi et al. [47] obtained good results with a diamond
content up to 70 mol% (34 weight-%) using dynamic pseudo-isostatic compaction.
Similar results are reported by Padyukov and Levashov [109].
SHS reactions have also been used in the metal brazing of diamonds on cutting
tools [110]. In this case an SHS process is used to supply the heat necessary for
brazing a single particle of diamond on a copper-tungsten plate through a silver,
copper, and titanium alloy. A pellet of Ti + C was placed under the tungsten-
copper plate and then ignited. The speed of the thermal cycle ensured that no
damage to the diamond occurs. It has been suggested that a supercooled liquid
forms as a result of the rapid rate of cooling under such experimental conditions.
The presence of the liquid brazing alloy even at relatively low temperatures ensures
the process of stress relief.
Acknowledgment
This work was supported by a grant from the (US) Army Research Office, ARO.
References
1. Z. A. Munir and U. Anselmi-Tamburini, Self-propagating exothermic reactions: The synthesis
of high temperature materials by combustion, Mater. Sci. Rep. 1989, 3, 277-365.
2. Z. A. Munir, Synthesis of high temperature materials by self-propagating combustion
methods, Ceramic Bull. 1988, 67, 342-349.
3. V. M. Shkiro, V. N. Doronin, and I. P. Borovinskaya, Study of the concentration structure of
the titanium-carbon combustion wave, Combust. Explos. Shock Waves, 1981, 16,270-376.
4. Z. A. Munir and N. Sata, SSH diagrams: theoretical analysis and experimental observations,
Int. J. SHS, 1992, 1, 355-365.
5. Z. A. Munir and W. Lai, The SHS diagram of Tic, Combust. Sci. Technol. 1992,88,201-209.
6. A. Feng and Z. A. Munir, The effect of product conductivity on field-activated combustion
synthesis, J . Am. Ceram. SOC.1997,80, 1222-1230.
7. S. Zhang and Z. A. Munir, Spin combustion in the system nickel-silicon, J. Mater. Sci. 1992,
27,4789-5794.
8. M. E. Grami and Z. A. Munir, The effect of nitrogen pressure and diluent content on the
combustion synthesis of TiN, J. Am. Ceram. SOC.1990,73, 2222-2227.
9. S. D. Dunmead, Z. A. Munir, and J. B. Holt, Temperature profile analysis in combustion
synthesis: I, theory and background, J . Am. Ceram. SOC.1992, 75, 175-179.
10. S. D. Dunmead, Z. A. Munir, and J. B. Holt, Temperature profile analysis in combustion
synthesis: 11, experimental observations, J. Am. Ceram. SOC.1992, 75, 180-188.
References 369
1 1. L. L. Wang and Z. A. Munir, Kinetic analysis of the combustion synthesis of molybdenum and
titanium silicides, Metall. Trans. 1995, 26B, 591-601.
12. S. D. Dunmead, Z. A. Munir, and J. B. Holt, Gas-solid reactions under a self-propagating
mode, Solid State Zonics, 1989, 32/33, 47&480.
13. S. Deevi and Z. A. Munir, The mechanism of synthesis of titanium nitride by self-sustaining
reactions, J. Mater. Res. 1990, 5, 2177-2183.
14. T. Boddington, P. G. Laye, J. Tipping, and D. Whalley, Kinetic analysis of temperature
profiles for pyrotechnic systems, Combust. Flame, 1986, 24, 359-368.
15. J. B. Holt and Z. A. Munir, Combustion synthesis of titanium carbide: theory and experiment,
J. Mater. Sci. 1986, 21, 251-259.
16. K. A. Philpot and Z . A. Munir, The influence of pre-ignition processes on the combustion
synthesis of nickel aluminides, in Materials Processing by Self-Propagating High- Temperature
Synthesis, K. A. Gabriel, S. G. Wax, and J. W. McCauley (Eds), 1987, p. 105.
17. L. L. Wang, Z. A. Munir, and J. B. Holt, Synthesis of Al20,-B4C composites via a thermite-
based combustion reaction, J. Mater. Synth. Proc. 1994, 2, 227-237.
18. L. L. Wang, Z. A. Munir, and J. B. Holt, Formation of MgO-B4C composites via a thermite-
based combustion reaction, J. Am. Ceram. SOC.1995, 78, 756-764.
19. A. Makino and C. K. Law, SHS combustion characteristics of several ceramics and inter-
metallic compounds, J. Am. Ceram. Soc. 1994, 77, 778-786.
20. A. P. Hardt and R. W. Holsinger, Propagation of gasless reactions in solids - 11. Experimental
study of exothermic intermetallic reaction rates, Combust. Flame, 1973, 21, 91-97.
21. A. I. Kirdyashkin, Y. M. Maksimov, and A. G. Merzhanov, Effects of capillary flow on
combustion in a gas-free system, Combust. Explos. Shock Waves, 1981, 17, 591-595.
22. A. A. Zenin, A. G. Merzhanov, and G. A. Nersisyan, Thermal wave structure in SHS
processes, Combust. Explos. Shock Waves, 1981, 17, 63-71.
23. I. P. Borovinskaya, A. G . Merzhanov, N. P. Novikov, and A. K. Filonenko, Gasless combus-
tion of mixtures of transition metals with boron, Combust. Explos. Shock Waves, 1974,10,2-10.
24. A. G. Merzhanov and I. P. Borovinskaya, A new class of combustion processes, Combust. Sci.
Technol. 1975, 10, 195-201.
25. N. P. Novikov, I. P. Borovinskaya, and A. G. Merzhanov, Dependence of the composition of
the products and the combustion rates in metal-boron systems on the ratio of the reagents,
Combust. Explos. Shock Waves, 1974, 10, 177-178.
26. A. Makino, N. Araki, and T. Kuwabara, Flammability limits, dilution limits, and effect of par-
ticle size on burning velocity in combustion synthesis of TIC, J S M E Znt. J. B: Fluids Therm.
Eng. 1994, 37, 576582.
27. V. M. Shkiro and I. P. Borovinskya, Study of characteristics of the combustion of mixtures of
titanium with carbon, in Combustion Processes in Chemical Technologyand Metallurgy, A. G.
Merzhanov (Ed.), Akad. Nauk SSSR, OIKhF, Chernogolovka, 1975, p. 253.
28. Y. B. Zeldovich, Theory for the limit of the quiet flame propagation, J. Exp. Theor. Phys.
1941, 11, 159-169.
29. A. Makino and C. K. Law, Heterogeneous flame propagation in the self-propagating
high-temperature synthesis (SHS) process: theory and experimental comparisons, Twenty-
fourth Symposium (International) on Combustion, The Combustion Institute, 1992,
pp. 1883-1 89 1.
30. D. C. Halverson, K. H. Ewald, and Z. A. Munir, Influence of reactant characteristics on the
microstructure of combustion-synthesized titanium carbide, J. Mater. Sci. 1993, 28, 4583-
4594.
31. A. D. Bratchikov, A. G. Merzhanov, V. I. Itin, V. N. Khachin, E. F. Dudarev, V. E. Gyunter,
V. M. Maslov, and D. B. Chernov, Self-propagating high-temperature synthesis of titanium
nickelide, Sov. Powder Metall. Met. Ceram. 1980, 19, 5-8.
32. Z. A. Munir and J. B. Holt, The combustion synthesis of refractory nitrides, Part I: theoretical
analysis, J. Mater. Sci. 1987, 22, 710-714.
33. M. E. Grami and Z. A. Munir, The effect of porosity on the combustion synthesis of titanium
nitride, J. A m . Ceram. Soc. 1990, 73, 1235-1239.
34. Z. A. Munir, The mechanism of synthesis of high temperature materials by combustion
processes, High Temp. Sci. 1990, 27, 279-293.
35. S. Zhang and Z. A. Munir, The combustion synthesis of refractory nitrides: Part 11, the
synthesis of niobium nitride, J. Muter. Sci. 1991, 26, 3380-3385.
36. V. A. Gorshkov, G. N. Komratov, and V. I. Yukhvid, Production of cast higher chromium
carbide using self-propagating high-temperature synthesis, Sov. Powder Metall. Met. Ceram.
1992.31.952-955.
37. U. Anselmi-Tamburini, F. Maglia, G. Spinolo, and Z. A. Munir, NijYSZ cermets from
combustion synthesis: effect of process parameter on product microstructure, J . Am. Ceram.
SOC.1998,81, 1765-1772.
38. Z. A. Munir, Analysis of the origin of porosity in combustion synthesized materials, J. Muter.
Synth. Proc. 1993, 1, 387-394.
39. J. H. Matlock and J. P. Stark, Thermal diffusion ofvacancies in aluminum, Acta Metall. 1971,
14,923-929.
40. Z. A. Munir and L. L. Wang, The contribution of thermal migration to pore formation during
SHS reactions, in Proceedings of the First US-Japan Workshop on Combustion Synthesis,
Y. Kaieda and J. B. Holt (Eds), National Research Institute for Metals, Tokyo, Japan, 11-
12 January 1990, pp. 123-137.
41. S. D. Dunmead, Z. A. Munir, J. B. Holt, and D. D. Kingman, Simultaneous synthesis and
densification of TiC/Ni4-A1 composites, J. Mater. Sci. 1990, 26, 241c-2416.
42. Z. Y. Fu, W. M. Wang, R. Z. Yuan, and Z. A. Munir, Fabrication of cermets by SHS-QP
method, Int. J. SHS, 1993, 2, 307-313.
43. S. Adachi, T. Wada, T. Mihara, Y. Miyamoto, M. Koizumi, and 0. Yamada, Fabrication of
titanium carbide cermets by high-pressure self-combustion sintering of titanium powder and
carbon fiber, J. Am. Ceram. SOC.1989,72, 805-809.
44. P. D. Zavitsanos, J. J. Gebhardt, and A. Gatti, The use of self-propagating high-temperature
synthesis of high density titanium diboride, in Combustion and Plasma Synthesis of High-
Temperature Materials, Z. A. Munir and J. B. Holt (Eds), VCH Publishers, New York,
1990, pp. 170-178.
45. Y. Miyamoto, SHS/HIP compaction using inorganic fuels, Int. J. SHS, 1993, 1,479489.
46. Y. Miyamoto, Z. Li, Y. S. Kang, and K. Tanihata, Development of ceramic/metal hetero-
geneous composites by SHS/HIP compaction, Int. J. SHS, 1996, 5, 69-75.
47. M. Ohyanagi, T. Yoshikawa, M. Koizumi, S. Hosomi, E. A. Levashov, and I. P. Borovin-
skaya, Fabrication of diamond-dispersed cermets by SHSidynamic pseudo isostatic com-
paction (DPIC), Znt. J. SHS, 1995, 4, 387-394.
48. H. A. Grebe, A. Advani, N. N. Thadhani, and T. Kottke, Combustion synthesis and
subsequent explosive densification of titanium carbide ceramics, Metall. Trans. 1992, 23A,
2365-2372.
49. L. J. Kecskes, T. Kottke, and A. Niiler, Microstructural properties of combustion-synthesized
and dynamically consolidated titanium boride and titanium carbide, J. Am. Ceram. SOC.1990,
73, 1274-1282.
50. L. J. Kecskes, A. Niiler, T. Kottke, VK. V. Logan, and G. R. Villalobos, Dynamic consolida-
tion of combustion-synthesized alumina-titanium diboride composita ceramics, J. Am. Ceram.
SOC.1996,79, 2687-2695.
51. B. H. Rabin, G. E. Kroth, and R. L. Williams, Fabrication of TiC-AI2O3 composites by
combustion synthesis and subsequent dynamic consolidification, Int. J. SHS, 1992, 1, 336341.
52. J. C. La Salvia, L. W. Meyer, and M. A. Meyers, Densification of reaction-synthesized
titanium carbide by high-velocity forging, J. Am. Ceram. SOC.1992, 75, 592-602.
53. D. A. Hoke, M. A. Meyers, L. W. Meyer, and G. T. Gray, Reaction synthesis/dynamic
compaction of titanium diboride, Metall. Trans. 1992, 23A, 77-78.
54. J. C. La Salvia, M. A. Meyer, and D. K. Kim, Combustion synthesis/dynamic consolidation of
Tic-Ni cermets, J. Muter. Synth. Proc. 1995, 2, 255-274.
55. J. Lis, S. Majorowski, V. Hlavacek, and J. A. Puszynski, Combustion synthesis and densifica-
tion of TiB2-TiC composite powders, Int. J. SHS, 1995, 4, 275-285.
56. H. Zhang, S. Yin, J. Zhong, and S. Guo, TiC-TiB2-AI2O3 ceramics made by combustion
synthesis and pressureless sintering, Int. J. SHS, 1996, 5, 185-190.
57. G. Xu, H. Zhuang, J. Cai, and S. Xu, Microwave sintering of SHS-produced AIN powders,
J. Muter. Synth. Proc. 1996, 4, 307-314.
References 371
58. B. Manley, J. B. Holt, and Z. A. Munir, Sintering of combustion-synthesized titanium carbide,

in Sintering and Heterogeneous Catalysis, G . C. Kuczynski, A. E. Miller, and G. A. Sargent
(Eds), Muter. Sci. Res. 1984, 16, 303-316.
59. M. Ouabdesselam and Z. A. Munir. The sintering of combustion-synthesized titanium boride,
J. Muter. Sci. 1987, 22, 1799-1807.
60. J. Lee, N. N. Thadhani, and H. A. Grebe, Reaction sintering of shock-compressed Ti + C
powder mixtures, Metall. Muter. Trans. 1996, 27A, 1749-1 759.
61. A. G. Merzhanov, A. S. Rogachev, A. S. Mukasyan, and B. M. Khusid, Macrokinetics of
structural transformation during the gasless combustion of a titanium and a carbon powder
mixture, Combust. Explos. Shock Waves, 1990, 26, 92.
62. J. Martinez and Z. A. Munir, Combustion synthesis of Zr0,-Cu composites during the
application of a pseudo-hot isostatic pressure, unpublished work.
63. M. Shibuya, 0. Odawara, M. Ohyanagi, and M. Koizumi, Simultaneous synthesis and
densification of TiN/Ti-Ni composites by SHS nitridation, Int. J . SHS, 1996, 5, 77-83.
64. N. N. Thadhani, R. A. Graham, T. Royal, E. Dunbar, M. U. Anderson, and G. T. Holman,
Shock-induced chemical reactions in titanium-silicon powder mixtures of different morphol-
ogies: time-resolved pressure measurements and materials analysis, J. Appl. Phys. 1997, 82,
1113-1128.
65. R. V. Raman, S. V. Rele, S. Poland, J. Lasalvia, M. A. Meyers, and A. R. Niiler, The one step
synthesis of dense titanium carbide tiles, J . Muter. 1995, 47, 23-55.
66. Z. A. Munir, I. J. Shon, and K. Yamazaki, Simultaneous synthesis and densification by field-
activated combustion, US Patent, approved, 6 January 1998.
67. U. Anselmi-Tamburini and Z. A. Munir, The propagation of solid-state combustion wave in
Ni-A1 foils, J . Appl. Phys. 1989, 66, 5039-5045.
68. S. C. Deevi, Diffusional reactions in the synthesis of MoSi2, Muter. Sci. Eng. 1992, A149,241-
251.
69. D. C. Halverson, B. Lum, and Z. A. Munir, The combustion synthesis of boride composites,
Proceedings of the Symposium on High-Temperature Materials Chemistry - IV , Z. A. Munir,
D. Cubicciotti, and H. Tagawa (Eds), PV 88-5, pp. 613-622, The Electrochemical Society,
Pennington, NJ, 1988.
70. Z. A. Munir, W. Lai, and K. H. Ewald, Field-assisted combustion synthesis, US Patent No.
5,380,409; 10 January 1995.
71. A. Feng and Z. A. Munir, The effect of an electric field on self-sustaining combustion
synthesis: Part 11. Field-activated synthesis of Sic, Metall. Muter. Trans. 1995, 26B, 587-
593.
72. A. Feng and Z. A. Munir, The effect of an electric field on self-sustaining combustion
synthesis: Part I. Modeling studies, Metall. Muter. Trans. 1995, 26B, 581-586.
73. Z. A. Munir, Electrically-stimulated SHS, I n t . J . SHS, 1997, 6, 165-185.
74. Z. A. Munir, Field effects in self-propagating solid-state synthesis reactions, Solid State Ionics,
1997, 101, 991-1001.
75. S. Gedevanishvili and Z. A. Munir, The influence of an electric field on the mechanism of
combustion synthesis of tungsten silicides, J . Muter. Res. 1995, 10, 2642-2647.
76. H. Xue and Z. A. Munir, Synthesis of AIN-SIC composites and solid solutions by field-
activated self-propagating combustion, J . Europ. Ceram. Soc. 1997, 17, 1787-1792.
77. H. Xue, K. Vandersall, E. Heian, N. N. Thadhani, and Z. A. Munir, Initiation of self-
propagating combustion waves in dense Mo + 2Si reactants through field activation, to be
published.
78. H. Xue and Z . A. Munir, The effect of field on the microstructure of MoSi, synthesized by
field-activated combustion, to be published.
79. D. C. Halverson, Z. A. Munir, and B. Lum, Combustion synthesis of ceramic preforms for
molten-metal infiltration, in Combustion and Plasma Synthesis ofHigh Temperature Materials,
Z. A. Munir and J. B. Holt (Eds), VCH Publishers, New York, NY,1990, p. 262.
80. H. Xue and Z. A. Munir, Extending the compositional limit of combustion-synthesized BIG-
TiB2 composites by field activation, Metall. Trans. 1995, 27B, 475480.
81. A. Zangvil and R. Ruh, Phase relationships in the silicon carbide-aluminum nitride system,
J . Am. Ceram. Soc. 1988,71, 884890.
82. M. Miura, T. Yoyo, and S. Hirano, Phase separation and toughening of SIC-AIN solid-
solution ceramics, J . Muter. Sci. 1993, 28, 4188-
83. H. Xue and Z. A. Munir, The synthesis of composites and solid solutions of SIC-AIN by field-
activated combustion, Scripta Muter. 1996, 35, 979-982.
84. C. He and G. C. Stangle, The mechanism and kinetics of the niobium-carbon reaction under
self-propagating synthesis-like conditions, J . Muter. Res. 1995, 10, 2829-2841.
85. H. Xue and Z. A. Munir, Field-activated combustion synthesis of TaC, Int. J . SHS, 1996, 5,
229-239.
86. J. H. Lee, N. N. Thadhani, and H. A. Grebe, Reaction sintering of shock-compressed Ti + C
powder mixtures, Metall. Muter. Trans. 1996, 21A, 1749-1 759.
87. L. Wang, M. R. Wixom, and L. T. Thompson, Structural and mechanical properties of TiB,
and TIC prepared by self-propagating high-temperature synthesis/dynamic compaction,
J. Muter. Sci. 1994, 29, 524-543.
88. M. J. Capaldi and J. V. Wood, Production and characterization of Tic-containing
materials by self-propagating high-temperature synthesis, J. Muter. Synth. Proc. 1996, 4,
245-253.
89. J. C. La Salvia and M. A. Meyers, Microstructure, properties, and mechanisms of Tic-Mo-Ni
cermets produced by SHS, Int. J . SHS, 1995, 4, 43-57.
90. Z. P. Xing, J. T. Guo, G. Y. An, and Z. Q. Hu, Hot pressing aided exothermic synthesis and
densification of NiAl and NiAI-TIC composites, Int. J . SHS, 1996, 5, 51-56.
91. C. C. Degnan and J. V. Wood, Processing and wear properties of a (W,TiC) particulate
reinforced ferrous-based metal matrix composite, A. Poursartip (Ed.), Proc. 10th Int. Conf.
Compos. Muter. Vancouver, Canada, 1995, pp. 77 1-778.
92. S. Y. Sharivker, I. P. Golovinskaya, I. N. Vorob’eva, and V. M. Bunin, Wear of tungsten-
less hard alloys based on SHS-titanium carbide, Powder Metall. Metal Ceram. 1996, 35,
55-57.
93. M. Ohyanagi, M. Kanno, and M. Koizumi, T i c combustion synthesis and fabrication of a
body with a densified surface layer by the pressureless method, J . Muter. Synth. Proc. 1993,
1, 31 1-322.
94. T. C. Woodger, J. J. Moore, H. C. Yi, and J. Y. Guigne, Combustion synthesis of A1-HfB,
metal matrix composites, T. S. Srivatsan and J. J. Moore (Eds), Proceedings of Processing
and Fabrication of Advanced Materials V ,TMS, 1996, pp. 223-233.
95. T. C. Woodger, J. J. Moore, H. C. Yi, and J. Y. Guigne, Combustion synthesis of Ni3Ti-TiBz
intermetallic matrix composites, T. S. Srivatsan and J. J. Moore (Eds), Proceedings of Pro-
cessing and Fabrication of Advanced Materials V ,TMS, 1996, pp. 399408.
96. W. Y. Yang and A. Petric, Microstructure of the Ti3A1(Nb)/TiB composite produced by
combustion synthesis, Metall. Muter. Trans. 1995, 26A, 3037-3043.
97. H. C. Yi and A. Petric, Combustion synthesis of Ti-Al-Nb matrix composites reinforced by
titanium borides, J. Muter. Synth. Proc. 1994, 2, 161-167.
98. Y. I. Vershinnikov, N. S. Pesotskaya, S. S. Mamyan, and I. P. Borovinskaya, SHS of com-
posite ceramic powder TiB,-AI,03 and its abrasive characteristics, Muter. Powder Technol.
1993, PTM93, 749-753.
99. F. Olevsky, P. Mogilevsky, E. Y. Gutmanas, and I. Gotman, Synthesis of in situ TiB2/TiN
ceramic matrix composites from dense BN-Ti and BN-Ti-Ni powder blends, Metall.
Muter. Trans. 1996,27A, 2071-2079.
100. C. R. Bowen and B. Derby, The formation of TiC/Alz03 microstructures by a self-
propagating high-temperature synthesis reaction, J. Muter. Sci. 1996, 31, 3791-3803.
101. H. C. Yi, H. J. Feng, J. J. Moore, A. Petric, and J. Y. Guigne, Combustion synthesis and
mechanical properties of TiC/Al2O3-AI composites, Int. J . SHS, 1996, 5, 39-50.
102. C. W. Kim, C. S. Kim, S. C. Koh, and D. K. Kim, Combustion synthesis and densification of
the TiC-AIzO3-AI composites, Int. J . SHS, 1996, 5, 159-168.
103. Z. Li, K. Tanihata, and Y. Miyamoto, Fabrication of multifunctional gradient materials in the
system of A1~03/TiC/Ni/TiC/Alz03 by SHS/HIP, Advanced Materials ’93, N. Mizutani (Ed.),
Elsevier Science, Amsterdam, 1994, pp. 677-680.
104. 0. Odawara, Long ceramic-lined pipes produced by a centrifugal thermal process, J. Am.
Ceram. SOC.1990,73, 629-633.
References 373
105. A. G. Merzhanov, Self-propagating high-temperature synthesis: twenty years of search and

findings, in Combustion and Plasma Synthesis of High Temperature Materials, Z . A. Munir
and J. B. Holt (Eds), VCH Publishers, New York, 1990, pp. 1-53.
106. L. Lin and H. Xue, Investigation of Cr,C,/Cr hard surfacing via SHS cast route, Znt. J . SHS,
1995, 4, 171-182.
107. K. L. Padyukov, E. A. Levashov, and I. P. Borovinskaya, Regularities of the behaviour of
synthetic diamond in the combustion wave of self-propagating high-temperature synthesis,
Eng. Phys. J . 1992,63, 577-583.
108. E. A. Levashov, I. P. Borovinskaya, A. Rogachov, M. Koizumi, M. Ohyanagi, and S. Hosomi,
SHS: A new method for production of diamond-containing ceramics, Znt. J . SHS, 1993, 2,
189-201.
109. K. L. Padyukov and E. A. Levashov, Self-propagating high-temperature synthesis: a new
method for the production of diamond-containing materials, Diamond Rel. Muter. 1993, 2,
207-21 0.
110. M. Ohyanagi, Y. Taketani, Y. Nakamura, E. Kamijou, and M. Koizumi, Instantaneous
bonding of diamond and metal by brazing, Am. Ceram. SOC.Bull. 1993, 72, 8 6 8 9 .
3 Hydrothermal Synthesis of Diamond

K. G. Nickel, T. Kraft, and Y. G. Gogotsi
3.1 Introduction
Cubic diamond and hexagonal graphite are the thermodynamically stable con-
densed phases in the P-Tdiagram of carbon with diamond and graphite dominating
at very high and low pressures respectively. Direct transformation between the
forms is kinetically impeded due to the necessity of a complete reconstruction of
strongly bonded covalent networks.
Natural diamonds are supposed to have grown in the depth of the mantle 121
under high-pressure high-temperature (HPHT) conditions above the Berman-
Simon line [ 11, shown in Fig. 1. Although possible mechanisms of natural diamond
formation are still debated, C-H-0 fluids seem to play an important role [3].
Reports on man-made diamond obtained by HPHT synthesis were first published
in 1955 by General Electric [4]. Usually, metals able to dissolve carbon under HPHT
conditions are used as catalysts and increase growth rates. Diamond crystals of
several millimeters in size can be obtained in this way, but usually small grains
for abrasives are produced. Direct conversion of graphite to diamond without
catalyst in HPHT apparatus is possible, but uneconomical for industrial applica-
tion. Direct transformation can be done by the detonation method and produces
nanosized powders of diamond and diamond-like carbon [5].
Diamond films and coatings can be obtained via various chemical vapor deposi-
tion (CVD) methods at and below atmospheric pressure [6]. This highly metastable
and kinetically driven diamond growth process requires the presence of atomic
hydrogen, which selectively etches graphitic and amorphous carbon while diamond
is unaffected.
While HPHT, shock wave, and CVD synthesis of diamond have been commer-
cialized, these methods have serious limitations. None of them allows economical
growth of large single crystals and the diamond quality does not satisfy the require-
ments for electronic or gem applications. Since the quality and size of the best
natural diamonds have not yet been reproduced in the lab, the search for a better
method of diamond synthesis continues. Recent work demonstrates that diamond
can be formed metastably under a variety of conditions.
Related to the CVD process is the low-pressure solid-state source (LPSSS) pro-
cess, where preformed mixtures of carbon and diamond seeds [7] or carbon and
metal [8] are treated in a H2 plasma, thus forming compact diamond ceramics.
Diamond particles have been found in combustion products [9], synthesized from
fullerenes by pressurization at room temperature [ 101 or sputtering at high tempera-
tures [ 111, obtained by decomposition of polymers with an appropriate structure
[ 121, by heating carbon ‘onions’ with electron beam in a TEM [ 131, by chlorination
of S i c [14], or by deposition from supercritical fluids [15-171.
0 6 0
Temperature [“C]
Figure 1. P-Tdiagram of the low-pressure, low-temperature labile equilibriums of carbon solution:
1 = graphite-diamond equilibrium line, 2 = glassy carbon-diamond transition line, 3 =range of
pneumatolytic hydrothermal processes, 4 = oxidative corrosion of diamond, 5 = anticipated area
of diamond hydrosynthesis, 6 and 7 =diamond synthesis from glassy carbon precursors, 8 = low-
pressure, low-temperature hydrothermal homoepitaxy of diamond. Reproduced from [ 151 with
permission from A. Szymanski.
The presence of small diamond particles has been confirmed independently in all
cases using several methods of analysis. These recent findings became primarily
possible due to the development of new analytical techniques that allow the
detection of diamond in small quantities and with particle sizes down to a few nan-
ometers. The formation of diamond particles in the range of the thermodynamic
stability of graphite can be explained either by kinetic factors, or by a higher stabi-
lity of small diamond particles compared to graphite, which has been controversially
discussed [ 181.
While some of the metastable routes to diamond mentioned above will probably
remain a scientific curiosity, hydrothermal synthesis has attracted more attention
due to its commercial potential. Synthesis from a supercritical water-based fluid is
an excellent method for preparing powders, coatings, and single crystals of many
different materials [ 191, which offers many advantages over other preparation meth-
ods because it is inexpensive, environmentally benign, reduces free energies for var-
ious equilibria, enables the synthesis of phases that may not be stable under other
conditions, and permits production of composites from incompatible phases that
can not be obtained by other means. It is the only commercial method of growing
large quartz single crystals and synthetic emeralds. Thus, if applied to diamond, it
may potentially yield diamond crystals, which might be competitive with the Cullinan
376 3 Hydrothermal Synthesis of Diamond
diamond. Rustum Roy calls the hydrothermal synthesis ‘The fourth revolution in
diamond synthesis’ [20]. Although we think that it is too early to consider it to be a
‘revolution’, it is certainly worthy of attention and further evaluation as a potential
method of diamond synthesis. The goal of the ongoing research is to make diamonds
from a liquid phase which is several orders of magnitude denser than vapor, in the
hope of increasing growth kinetics compared to CVD synthesis.
In this chapter, we explain why hydrothermal synthesis deserves special consid-
eration and present a review of the current state of the art in hydrothermal synthesis
of diamond as a potential future method of making diamond or hard diamond-like
sp3-carbon.
3.2 Evidence from Nature

DeVries was among the first to suggest the possibility of the hydrothermal synthesis
of diamond [3]. A hot fluid in the C-H-0 system, not quite saturated with carbon,
should precipitate carbon as it cools down [21]. Depending on pressure and tem-
perature, diamond, graphite [22], or poorly crystallized carbonaceous materials
like anthraxolite [3] can form. Any supply of carbon rich C-H-0 fluids at a
depth greater than about 150km in the lithosphere has a high probability of becom-
ing diamondiferous as indicated by some investigations on diamond inclusions and
diamond host rocks [23-251. C-H-0 fluids should be the base of the inorganic
solutions from which graphite and diamond precipitate and grow in nature [3].
Little evidence for metastable growth of diamond in earth systems was found.
Diamond should form only at temperatures and pressures above about 1100°C
and 4.5 GPa, although a theoretical minimum set for diamond formation was calcu-
lated to be 2GPa and 1000°C [2].
However, it has been suggested [26] that natural epitaxial coarsening of diamonds
could proceed in colloidal pneumatolytic-hydrothermal environments in near-
superficial earth zones, because equilibrium data for material trapped as inclusions
in diamond indicated closing temperatures of 600-800°C and pressures below
10 MPa [15] and fit to early theories of diamond formation [27]. Based on the petro-
logic circumstances mentioned in [26], an area of probable diamond hydrogenesis
and hydrosynthesis was sketched (Fig. 1).
Theoretical calculations and experimental studies [28] suggested that smaller,
colloidal diamond particles of about 10nm may be stable at lower pressures and
higher temperatures than macroscopic diamond. A so called ‘crystallite size effect’
[29] might account for metastable diamond nucleation and growth under hydro-
thermal conditions. Thus, the HPHT graphite-diamond equilibrium should only
be applied to crystallites >100nm in diameter [28].
Based on observations of turbid rims of fibrous diamond and layers of the lons-
daleite polymorph on some natural diamonds [30] as well as of embryocrystals and
phantoms in central parts of natural diamonds [311, a mechanism for homoepitactic
diamond coarsening and crystal growth in nature was presented [32,33]. Assuming
that diamond grows from a nucleus in the postmagmatic fluid, the growth rate
should be controlled by the diffusion process of sp3-structural elements to the
seed crystal surface resulting in thin colloidal cluster layers which may in turn
form lonsdaleite or fibrous diamond. Due to coalescence, slow transformation to
thick polycrystalline layers or monocrystals with cubic structure might occur.
Coexistence, intergrowth and thus presumable paragenesis of crystals of dia-
mond, quartz and a-and 0-Sic have been found in a diamondiferous kimberlite
[34] where crystallization of quartz instead of coesite indicates maximum pressures
of <2.0-2.8 GPa at temperatures of 1000-1200°C. This might be important in the
view of diamond hydrosynthesis from silicon carbides in the metastable range.
3.3 Hydrotherma1 Synthesis

Investigations of hydrothermal processes and development of hydrothermal
equipment have their origin in the attempts to understand and mimic nature
which produces thousands of well-crystallized minerals at very modest temperatures
[19]. The hydrothermal processes in this paper deal with water or aqueous
fluids in the supercritical range of water, above its critical point ( T , = 374"C,
P, = 22.14MPa). This is opposed to 'just boiling in water', and supercritical pro-
cesses dealing merely with hydrocarbons or halogenated hydrocarbons [35].
Most interesting applications of supercritical fluids occur at 1 < TITc < 1.1 and
1 < PIP, < 2 [36]. At these conditions, the fluid exists as a single phase with suffi-
cient density to give high dissolving power and high diffusivity, facilitating mass
transport. Because of the high compressibility of fluids near their critical point,
physical and chemical properties change drastically and density and dissolving
power can be tuned through changes in temperature and pressure. Supercritical
water can act as an acidic or basic catalyst, and its reactivity can often be enhanced
by autocatalysis from water-soluble reaction products [37].
The discussion of hydrothermal diamond synthesis is divided into two sections,
dealing with synthesis from C-H-0 liquids and synthesis based on decomposition
of silicon carbide, respectively. Both start with thermodynamic calculations in
order to demonstrate the theoretical possibility of carbon formation before the
experimental findings are summarized. Naturally, equilibrium calculations do not
consider kinetic limitations.
3.3.1 C-H-0 System

In many instances diamond synthesis in the C-H-0 system is related to diamond
formation in nature as discussed in part 2. It was shown that precipitation of
diamond can occur in the C02/CH4 system [38]. No reagent external to the
C-H-0 system should be required for carbon precipitation while high water
fugacities may exist at the site of diamond precipitation.
400 500 600 700 800 900 1.000

Temperature I 'C
Figure 2. Calculated temperature dependence of the major species in the carbon-water system for a
C : H 2 0 molar ratio of 1 : 1 at 100 MPa.
3.3.1.1 Thermodynamic Analysis

Since experiments related to carbonaceous liquids summarized and discussed below
were carried out in the presence of diamond seed, the calculations presented here
[39] have been restricted to the water-diamond system, excluding graphite from
the calculations.
While CO and H2 are the main equilibrium species at low pressures (P< 1 MPa),
at higher pressures (P> 2MPa), a 1 : 1 mixture of CO and H2 converts to a four-
species assemblage involving C02, CH4, H20, and diamond, suggesting two
independent reactions:
CO+H2 =iCH4+iCO;?, (1)
co + H2 = Cdiamond+ H20. (2)
Combined, the overall reaction governing the system to a first approximation is
4
2CO + 2H2 = CH4 + C02 + Cdiamond + H20. (3)
The individual contributions of reactions (1) and (2) in reaction (3) at higher
pressures (100MPa) are strongly temperature dependent (Fig. 2). It is clear that
+
the equilibrium amounts run in pairs CH4 + C02, Cdiamond H20, and CO H2. +
The deviation at higher temperatures stems from the influence of the Boudouard
equilibrium
+
2C0 = Cdiamond c 0 2 . (4)
Clearly, at higher temperatures diamond will be oxidized to yield CO.
Reaction ( 3 ) shows the dependence of the equilibrium composition on the C: H 2 0
ratio. In order to maintain the equilibrium, an increase in the amount of H 2 0yields
less CH4, C02, and diamond and more CO and H2. The amount of diamond
decreases with decreasing C :H 2 0 ratio. Accordingly, the temperatures for the
elimination of diamond from the system decrease with decreasing C : H 2 0 ratio,

from 950°C at a C: H 2 0 ratio of 1 : 1.2, to 700°C at a C: H 2 0 ratio of 1 : 1.4 [39].
Repetition of the calculations for graphitic carbon gives a similar result with
shifted boundaries (we need a lower C: H 2 0 ratio of approximately 1:2 to eliminate
C at 700°C). Thermodynamic equilibrium calculations carried out for very high
pressures and temperatures [40] are consistent with our calculations.
Summarizing, the hydrothermal behavior of diamond is predicted to depend
crucially on temperature and the C : H 2 0 ratio with a strong interaction of both
variables while it shows a weak pressure dependence in the range of supercritical
water.
3.3.1.2 Interaction between Diamond and Supercritical Water

Knowledge of the hydrothermal behavior of different allotropes and forms of
carbon is essential for optimization of growth parameters of diamond.
Noticeable interaction between nanocrystalline diamond powder and supercritical
water starts at about 700°C but even long-term high-temperature treatment (94 h,
850"C, 200MPa) did not cause complete dissolution [39]. The sharpness of the
diamond Raman band increases with increasing reaction temperature and time,
while the band position moves from the downshifted value of 1323cm-' for
untreated nanocrystalline diamond powder to the typical diamond value of
1332 cm-* . This was explained by ordering and healing of the diamond structure.
Electron paramagnetic resonance (EPR) measurements indicated that the broken
carbon bond concentration decreased by a factor of 2 after 100 h at 750°C and
300 MPa. Hydrothermal treatment of mixtures of graphite and nanocrystalline
diamond showed, that nanocrystalline diamond reacts with supercritical water
faster than graphite [39]. Microcrystalline diamond powders showed a significant dis-
solution with weight loss and gas formation after a week at 750°C and 500 MPa [39].
Healing of diamond as found in the case of nanocrystalline diamond can be
explained by redeposition of carbon from the hydrothermal fluid, which also
explains the increasing formation of graphitic carbon with increasing temperature
observed in these experiments. It was reported that hydrothermal treatment in
the presence of a catalyst at 900°C and 1 GPa enhanced the graphitization of
amorphous carbon [41].
Formation of etching pits has been observed on the surface of diamond single
crystals after 15h at 800°C and l00MPa [39]. Long-term treatment at 800°C
caused significant weight loss. Graphitization of the diamond surface was not
observed. No weight change of the diamond single crystals occurred in the presence
of nanocrystalline diamond powder or graphite. Deposition of carbon films on the
walls of the gold capsules indicated the saturation of the fluid with carbon, and thus
preservation of diamond.
The interaction of artificial diamond single crystals with a high pressure water
fluid at 900 and 1400°C under 5.2GPa yielded nondiamond carbon films [42].
Scanning electron microscopy (SEM) examinations revealed the formation of
deep and shallow flat-bottomed trigons on diamond-( 11 1) faces at reaction tempera-
tures above 1000°C.
Summarizing, the hydrothermal behavior of diamond depends strongly on the

grain size. Noticeable dissolution of monocrystalline, microcrystalline and nano-
crystalline diamond start at 800, 750, and 700°C, respectively. In agreement with
thermodynamic calculations, complete dissolution of diamond was never observed
at high C : H 2 0 ratios. Precipitation of graphitic carbon and healing of defective
diamond crystals have been indicated. While some data suggest the formation of
nanocrystalline diamond, no epitaxial diamond growth was observed.
3.3.1.3 Hydrothermal High-pressure High-temperature Synthesis

Epitaxial growth of diamond on diamond has been demonstrated in the carbon-
water system at 7.7GPa starting at 2000°C and yielding a considerable weight
increase at 2200°C [40]. The newly grown diamond coats were tightly bonded to
the seed crystals. The colorless and transparent crystals became black and
opaque, indicating inclusions of graphite. Rounded corners and edges were
sharpened, and rough and irregular growth steps and numerous pits were formed.
No evidence of direct conversion of graphite to diamond was found. According
to the thermodynamic calculations, it was concluded that graphite dissolved in
the water rich fluid, forming COz and CH4. Subsequent carbon precipitation
from the carbon-saturated fluid resulted predominantly in homoepitaxial diamond
growth.
3.3.1.4 Synthesis outside the Range of Thermodynamic Stability

The range of thermodynamic stability is 30 MPa < P < 1 GPa. Hydrothermal
treatment of glassy carbon and diamond seeds at 80&900"C and 140MPa in Au
capsules resulted in a substantial increase of the diamond (1 1 1) reflection in X-ray
diffraction (XRD) patterns [35,43]. Hexagonal crystal shapes were etched out of the
substantially reacted glassy carbon but possibly new-grown diamond crystals were
too small to be seen by SEM. Less diamond growth was found at higher tempera-
tures. No diamond growth was observed using Pt capsules suggesting that hydrogen
presence is essential for hydrothermal growth of diamond.
Aggregates of small diamond crystals were obtained after experiments, when mix-
tures of glassy carbon, quarter micron diamond seed and different metal powders
(Ni, Pt, Fe) were hydrothermally treated at 800°C and 140MPa in gold tubes
[17]. In typical runs, the original black powder charge turned gray and contained
some large aggregates with irregular morphology. The aggregates, which were up
to tens of microns in size, contained no metal and were built of diamond particles
of 0.1-1 pm size. Some particles several microns in size were found. No diamond
growth was observed in the absence of diamond seeds. Formation of new diamond
was deduced from strong differences in the luminescence spectra of diamond seed
and hydrothermally treated material, intensity increase of the diamond Raman
band of an order of magnitude after hydrothermal treatment, and different down-
shifts of the Raman band with increasing excitation power.
We found similar agglomerates after hydrothermal treatment of nanocrystalline
diamond at 740°C and 300 MPa in pure water (Fig. 3a). Hydrothermal treatment
3.3 Hydrothermal Synthesis 38 1
Figure 3. SEM micrographs of: (a) aggregate after 12 h hydrothermal treatment of nanocrystalline
diamond at 740°C and 300 MPa, (b) plate-like crystals on the surface of a diamond seed crystal after
150 h hydrothermal treatment of graphite and diamond single crystals at 800°C and 300 MPa, (c)
liquid-phase hydrothermal homoepitaxy of diamond on diamond seed at 170MPa and 400°C in
a specially prepared water solution. Reproduced from [I51 with permission from A. Szymanski.
of nanocrystalline diamond can result in changes in its Raman-spectra without

change in grain size [39] and it is not clear whether new diamond is formed.
Hydrothermal growth of diamond on diamond single crystal substrates was
reported when mixtures of glassy carbon with quarter micron diamond seed or
with Ni and a diamond single crystal were treated with highly concentrated
NaOH at 800°C and 140MPa [44]. Aggregates of about 10pm of transparent
crystals were found in the presence of fine-grained diamond, while hexagonal, octa-
hedral and cubic shaped transparent crystals of about 2 pm were observed in the
presence of Ni. We observed formation of very similar two-dimensional crystals
on diamond single crystals without addition of Ni or NaOH (Fig. 3b). The crystals
formed could be diamond or a new crystalline transparent material without a
Raman band, however, their identification was difficult.
Successful hydrothermal diamond synthesis was carried out in autoclaves filled
with a ‘specially prepared carbon enriched water solution’, the composition of
which was not disclosed [15,45,46]. The carbon precursor should be fine-grained
diamond, vitreous carbon or emulsion of crude oil and water [29]. The presence
of free radical catalysts was mentioned and paragenetic crystallization of quartz
needles and diamond indicate the presence of silicon [45]. The synthesis was
described as a sol/gel colloidal process working in the range 200-600°C and 100-
200 MPa. Healing and joining of diamond crystals was reported. After 21 days at
400°C and 170 MPa, thin colorless films of polycrystalline diamond were obtained
on (1 11) surfaces of seed crystals (Fig. 3c). With a reported size of 1 5 4 0 pm, these
are the largest diamond crystals from hydrothermal experiments.
Summarizing: Strong experimental evidence exists for diamond growth on dia-
mond seed outside the range of its thermodynamic stability. While nanocrystalline
diamond might grow under a variety of conditions, possibly due to the ‘crystallite
size effect’ [18,28], growth of microcrystalline or possibly larger diamond requires
special conditions which have only been attained very roughly so far.
3.3.2 Hydrothermal Treatment of Sic

The first observation of hydrothermal carbon formation on S i c [47,48] led to the
suggestion of the following basic decomposition reactions of S i c [49,50]:
Sic + 2H20 = Si02 CH4 + (5)
+
S i c + 3H20 = Si02 CO 3H2 + (6)
Sic + 4H20 = Si02 + C 0 2 + 4H2 (7)
S i c + 2H20 = Si02 + C + 2H2 (8)
3.3.2.1 Thermodynamic Calculations of SiC-H20 Interactions

The calculations presented here are in qualitative agreement with other published
calculations [5 1,521. We concentrate on the activity of the condensed phases
SO2, C, and Sic, presented as a function of the amount of water added (Fig. 4).
p-Sic -
I
01 I
0 1 2 3 4 5
amount of water I mol
Figure 4. Calculated equilibrium activity of Si02, S i c and C for compositions of 1 mol S i c plus
variable amounts of H 2 0 at 500°C and 200 MPa.
This computation is typical for the whole supercritical T-P range. It shows that Si02
is stable in the whole range. Sic becomes unstable at a H 2 0 :Sic molar ratio of 2 : 1,
which is a critical point, whereas carbon remains stable until a slightly larger
H 2 0 :Sic molar ratio than 2: 1 is reached. This indicates the possibility of direct
carbon formation in the course of hydrothermal leaching of S i c by selective
oxidation of Si according to reaction (8). Reaction (8) is restricted to low
H20:SiC molar ratios whereas reaction (6) and (7) are impeded at H20:SiC
molar ratio 5 2 : 1. The equilibrium amount of carbon increases with increasing
temperature and decreasing pressure.
3.3.2.2 Experimental Results for the SiC-H20 System

Hydrothermal formation of carbon on Sic was first observed in the course of corro-
sion studies of amorphous Si-Ti-C-0 (Tyranno) fibers performed at 300-800°C at
about 100 MPa [47,48,53]. Increasing downshift of the carbon D band in the
Raman spectra with increasing reaction temperature from 1355-1360 cm-' in
microcrystalline graphite to 1350cm-', 1336 cm-I, and 1330 f 3 cm-' after hydro-
thermal treatments at 300°C, 500"C, and 600-800°C respectively indicated the
possible formation of sp3-bonded carbon [50].
The investigations were extended to a great variety of different forms of
Sic. Carbon formation was observed on powders, platelets, and single crystals of
a-Sic, powders and whiskers of p-Sic, B-doped SSiC and polycrystalline CVD-
Sic samples [16,48-501. The formation of carbon depends crucially on the
H 2 0 :Sic ratio, where three cases can be distinguished [50,52].
(i) Low H 2 0 :Sic ratio: Carbon and Si02 precipitate on the Sic surface.
(ii) Very high H 2 0 :Sic ratio: Neither carbon nor Si02 precipitates, only in some
cases carbon films of nanometer thickness where found.
(iii) Medium H 2 0 :Sic ratio: Carbon film formation accompanied by dissolution of
SO2.
I I
1100 1200 1300 1400 1500 1600
Wavenumber / cm-I
Figure 5. Raman spectra of carbon on a-Sic crystals, X = 632.8 nm, after 4 h hydrothermal treat-
ment at 600°C and 200 MPa: (a) homogeneous carbon film, (b) diamond particle at seed crystal
edge, (c) example of an inhomogeneous carbon film after 4 h hydrothermal treatment at 700°C
and 200 MPa.
Depending on the Sic material used, carbon formation starts at temperatures

between 300°C and 600°C. In all cases IR-absorptions corresponding to hydro-
genated sp3 carbon were present at about 2850, 2920, 2960 (C-H stretch) [16].
TEM investigations and analysis of the Raman spectra consistently showed for-
mation of fine graphitic carbon with a grain size of a few nanometers. Carbon
yield and grain size increase with increasing temperature. The carbon yield reaches
a maximum at about 700°C and decrease with further temperature increase due to
subsequent oxidation of the carbon films.
Downshift of the Raman D-band of carbon and splitting in two or more bands
in the range 1100-1400 cm-' were often observed in the whole temperature range
under investigation and indicated disorder as well as the presence of sp3 carbon
(Fig. 5), although the respective contribution of each of the processes into
observed shift has not been established. Hydrothermal treatment of p-Sic
whiskers for 24h at 600°C yielded Raman bands at 1327 and 1352cm-' which
were assigned to nanosized diamond and disordered carbon, respectively [53].
Generally, downshift of the Raman D-band increases with increasing reaction
temperature.
Carbon layers grown on a-Sic crystal surfaces, visible in the optical microscope,
revealed only the formation of small bulges, which may rather be attributed to
domain borders than to delaminating carbon on SEM micrographs. Homogeneous
films of amorphous carbon with high sp3-contribution (Fig. 5a) formed at 600°C.
Very few transparent crystals of micrometer size, exhibiting the typical diamond
Raman band at about 1330cm-' (Fig. 5b), were found at the edges of S i c crystals.
Higher temperatures led to inhomogeneous carbon films with often high sp3 content
(Fig. 5c) on the Sic-crystals as well as on the gold capsules which indicates also
carbon precipitation from the fluid.
Carbon formation via hydrothermal treatment of P-Sic was intensively studied at
400-800°C and 100-500 MPa [54]. Carbon formed at all temperatures, but
reasonable amounts in reasonable time were found only at and above 550°C.
Above 600°C carbon deposits on the capsule walls indicated carbon precipitation
from the fluid. In effort to reveal the presence of diamond after hydrothermal
treatment, the residues were leached in 20% H F solution to remove Si02 and in
hot HC104 to remove graphitic carbon because large amounts of silica and non-
diamond carbon obscured the picture.
XRD investigations showed graphite reflections and in some cases a very weak
reflection at 0.206nm, attributed to diamond. In the Raman spectra, shifts of the
graphite D band, and additional bands and broad features in the range 1100-
1340 cm-', confirmed the presence of amorphous and nanocrystalline carbon
and indicate the possible presence of disordered nanocrystalline sp3-carbon.
Small particles with a size below 0.3pm were found, and sometimes intergrown
with several larger P-SIC particles after wet oxidation, and exhibited predomi-
nantly hexagonal (0001) and face-centered cubic (1 11) and (100) faces that were
attributed rather to diamond than to graphite. The morphology of the particles
was very similar to that of gas phase diamond powders. Reflections corresponding
to lattice spacing of 0.206 and 0.192 nm in small angle diffraction (SAD) patterns
suggested the presence of cubic diamond as well as of lonsdaleite or other diamond
polytypes.
Hydrothermal experiments with P-Sic powders and watery solutions of 10%
Na2C03, 10% NaHC03 and 50% NaHCOO yielded results which were very
close to each other, because sodium hydrogen carbonate and sodium formiate can
be converted to disodium carbonate at high temperatures according to
2NaHC03 = Na2C03 H 2 0 + (9)
and
2NaHCOO = Na2C03+ C H 2 0 = Na2C03 C + + H20. (10)
Carbon formation was observed at and above 400°C and became pronounced at
600°C. Raman spectra strongly suggest the formation of larger quantities of sp3-
bonded carbon than in the experiments with distilled water.
Small amounts of well crystallized diamond were found after hydrothermal
treatment of p-Sic powder at 700-750°C in the presence of diamond seed [55].
After removing silica and nondiamond carbon, small (<3 pm) carbon particles
of predominantly octahedral shape, thus being probably diamond, were found.
They were attached to the surfaces of the single crystal seed but could be removed
by intense ultrasonic treatment. Tetrahedral hillocks of <1 pm size as well as
tetrahedral etch pits appeared on the seed diamond surfaces. The hillocks always
showed a faceted structure and sometimes common orientation, as expected for
homoepitaxial diamond growth.
The microcrystalline diamond seed did not show any increase in size after hydro-
thermal treatment, but particles of <50 nm size, which typically show a regular
crystalline shape, were found. They were identified as new-grown nanocrystalline

diamond by Raman spectra, which show a downshift of the diamond band to
1323-1324 cm-' and an increased band halfwidth, full width at half maximum
(FWHM) of 10-12 cm-' [56] compared with 1332 cm-' and 5 cm-' respectively,
found for the microcrystalline diamond seed used.
In another attempt to obtain diamond from Sic, no evidence for diamond growth
was found by means of XRD and SEM after hydrothermal hydrolysis of p-Sic
powder at 140MPa and 800°C [57]. Formation of quartz, a second acicular
phase, graphite and maybe nanosized diamond crystals was reported. TEM analysis
of whiskers of the second acicular phase revealed lattice spacing of 0.9 nm, a d-value
of 0.56nm and a composition of Si01,6Co,2. A very sharp band at 1330cm-I was
found in a Raman spectrum and assigned to the silicon oxycarbide, Si01.6Co.2,
rather than to diamond.
However, we would like to comment, that the lattice spacing, the d-value, and the
acicularity of the so called 'oxycarbide' whiskers are close to the c-axis (0.861 nm),
the (101) reflection (0.564nm) and the space group (P41212)of the silica modifica-
tion keatite [58], which has no Raman band at 1330 cm-' . We often found keatite
after hydrothermal experiments with p-Sic powder and CVD-SIC fibers [Sl]. The
small amounts of carbon found can easily be introduced during hydrothermal treat-
ment or TEM investigation. Thus, there is no alternative assignment to the band at
1330 cm-' except of diamond.
Summarizing, strong evidence for the formation of nanosized diamond in the
course of hydrothermal treatment of Sic was found. Seeding might enhance the
size of newly formed diamond crystals. Above all, the hydrothermal method
seems to be a versatile way for carbon coatings on Sic with emphasis on formation
of amorphous sp3-bonded carbon. Depending on the substrate, reaction conditions
and composition of the fluid, it should be possible to vary the coating structure from
graphitic to diamond-like carbon.
3.3.2.3 Comparison of CVD- and Hydrothermal Diamond Synthesis from the

Sic-H20 System
Formation of diamond-like sp3-carbon in the course of hydrothermal decom-
position of Sic may be explained in the same way as diamond formation at low
pressures by CVD. Due to its crystal structure, cubic p-Sic is a good substrate
for heteroepitaxial diamond growth. Tetrahedral coordinated carbon in the Sic
structure may become a good precursor for the further growth of sp3-bonded
carbon (diamond) when Si is extracted. Hydrothermal treatment of S i c results
in the formation of hydrogen, which, dissociated into atoms, plays a crucial role
in CVD diamond synthesis by etching preferentially sp2-bonded carbon, thus
facilitating formation of sp3-carbon and nucleation of diamond. Even though
the equilibrium concentration of atomic hydrogen is very low, it can not be
ruled out that hydrogen initially forms as atomic or highly reactive species, thus
affecting the carbon in the neighborhood of its formation. These facts indicate
some similarities between hydrothermal diamond synthesis from carbides and
diamond synthesis via the vapor phase.
References 387
3.4 Outlook
Taking into account the fact that hydrothermal formation of carbon takes place at
relatively low temperatures (200-800°C) and pressures (<100 MPa), this method
seems to be very attractive. It has obvious advantages over the HPHT synthesis
due to less extreme temperature-pressure conditions and the possibility of much
larger pressure vessels, which would make the hydrothermal method attractive for
industrial use. However, the temperature of synthesis should be decreased prefer-
ably to 200-250°C to make commercialization and up-scaling easier.
A better understanding of the mechanisms of diamond growth under hydrother-
mal conditions should make it possible to optimize the conditions for the growth of
diamond coatings or large single crystals. High growth rates compared to CVD
synthesis may be achievable.
Current understanding of the diamond synthesis under hydrothermal conditions
and quantity of diamond produced are comparable with those of CVD synthesis of
diamond at the end of the 1950s or the beginning of the 1960s. Thus, we are at the
beginning of the road and further research is strongly recommended.
Acknowledgments
We thank Professor Dr. Andrzej M. Szymanski, Faculty of Chemistry of the
Warsaw University of Technology for supply of materials for publication, and
the German Research Society for financial support (grant DFG Ni 299/4).
References
I . R. Berman and S. F. Simon, Z . Elektrochem. 1955, 59, 333-338.
2. R. C. DeVries, R. Roy, S. Somiya, and S. Yamada, Trans. Mater. Res. Soc. Jpn. 1994, 19B,
64 1-663.
3. R. C. DeVries, in Advanced Ceramics IIZ, S. Somiya (Ed.), Elsevier, Amsterdam, 1990, pp. 181-
205.
4. F. P. Bundy, H. T. Hall, H. M. Strong, and R. H. Wentorf, Nature, 1955, 176, 51-55.
5. K. Yamada and A. B. Sawaoka, J . Am. Ceram. Soc. 1994, 77, 1104-1106.
6. P. K. Bachmann, D. Leers, and H. Lydtin, Diamond Rel. Muter. 1991, 1, 1-12.
7. R. Roy, H. S. Dewan and P. Ravindranathan, Mater. Res. Bull. 1993, 28, 861-866.
8. R. Roy, H. S. Dewan, and P. Ravindranathan, J. Mater. Chem. 1993,3,685-686.
9. B. V. Derjaguin and D. V. Fedoseev, in New Diamond Science and Technology,M R S Znt. Conf.
Proc. 1991, pp. 3-10.
10. M. N. Regueiro, P. Monceau, and J.-L. Hodeau, Nature, 1992,355, 237-239.
11. A. R. Krauss, D. M. Gruen, D. Zhou, T. G. McCauley, L. C. Qin, T. Corrigan, 0. Auciello,
S. Pimenov, V. Konov, and A. Karabutov, in M R S Fall Meeting Abstracts, Paper W10.5,
MRS, 1997, p. 499.
12. G. T. Visscher, D. C. Nesting, J. V. Badding, and P. A. Bianconi, Science, 1993,260, 1496-1499.
13. F. Banhart, Phys. Bl. 1997, 53, 33-35.

14. Y. G. Gogotsi, I.-D. Jeon, and M. J. McNallan, J . Muter. Chem. 1997, 7, 1841-1848.
15. A. M. Szymanski, E. Abgarowicz, A. Bakon, A. Niedbalska, R. Salacinski, and J. Sentek,
Diamond Rel. Muter. 1995, 4, 234-235.
16. Y. G. Gogotsi, P. Kofstad, M. Yoshimura, and K. G. Nickel, Diamond Rel. Muter. 1996, 5,
151-162.
17. X.-Z. Zhao, R. Roy, K. A. Cherian, and A. Badzian, Nature, 1997, 385, 513-515.
18. P. Badziag, W. S. Verwoerd, W. P. Ellis, and N. R. Greiner, Nature, 1990, 343, 244-245.
19. R. Roy, in Workshop on Solvothermal and Hydrothermal Reactions, Takamatsu, Japan, 1996,
pp. 11-120.
20. R. Roy, in M R S Fall Meeting Abstracts, Paper W10.4, MRS, 1997, p. 499.
21. B. R. Frost, Amer. J. Sci. 1979, 279, 1033-1059.
22. D. Rumble I11 and T. C. Hoering, Geochim. Cosmochim. Acta, 1986,50, 1239-1247.
23. 0. Navon, I. D. Hutcheon, G. R. Rossman, and G. J. Wasserburg, Nature, 1988,335, 784789.
24. 0. Navon, Nature, 1991, 353, 746-748.
25. G. D. J. Guthrie, D. R. Veblen, 0. Navon, and G. R. Rossman, Earth Planet. Sci. Lett. 1991,
105, 1-12.
26. A. M. Szymanski, Arch. Miner. 1994, 50, 13-30.
27. S. J. Thugutt, Rev. Sci. 1923, 61, 97-102.
28. A. Niedbalska, R. Salacinski, J. Sentek, and A. M. Szymanski, in Proceedings of the Fourth
International Conference on Advances in New Diamond Science and Technology, S. Saito,
N. Fujimori, 0. Fukunaga and M. Kamo (Eds), Scientific Publishing Division MYU,
Tokyo, Japan, 1994, pp. 415418.
29. A . M . Szymanski, J. Miner. SOC.Korea, 1997, 10, 18-32.
30. C. Frondel and U. B. Marvin, Nature, 1967, 214, 587-589.
31. V. Mironov and B. Antonyuk, Arch. Miner. 1994, 50, 3-12.
32. A. M. Szymanski, in Paper presented at the European Conference on Diamond Materials,
Barcelona, 1995.
33. A. M. Szymanski, Szklo Ceram. 1996,47, 26-28.
34. I. Leung, W. Guo, I. Friedman, and J. Gleason, Nature, 1990, 346, 352-354.
35. R. Roy, D. Ravichandran, P. Ravindranathan, and A. Badzian, J . Muter. Res. 1996,11, 1164-
1168.
36. C. A. Eckert, B. L. Knutson, and P. G. Debenedetti, Nature, 1996,383, 313-318.
37. M. Siskin and A. R. Katritzky, Science, 1991, 254, 231-237.
38. P. Deines and J. W. Harris, Geochim. Cosmochim. Acta 1995,59, 3173-3188.
39. Y. G. Gogotsi, T. Kraft, K. G. Nickel, and M. E. Zvanut, Diamond Rel. Mater. 1998,
submitted.
40. S. Yamaoka, M. Akaishi, H. Kanda, and T. Osawa, J. Cryst. Growth, 1992, 125, 375-377.
41. %-I. Hirano, K. Nakamura, and S. Somiya, in Hydrothermal Reactions for Materials Science
and Engineering, S. Somiya (Ed.), Elsevier, London, 1989, pp. 331-336.
42. N. A. Bendeliani, T. D. Varfolomeeva, A. N. Glushko, N. A. Nikolaev, V. N. Slesarev, and S. V.
Volkov, Kristallograjiya, 1995, 40, 345-346.
43. R. Roy, D. Ravichandran, and P. Ravindranathan, in Proceedings of the Applied Diamond
Conference 1995, Applications of Diamond Films and Related Materials, Third International
Conference, A. Feldman, Y. Tzeng, W. A. Yarbrough, M. Yoshikawa and M. Murakawa
(Eds), NIST, Gaithersburg, MD, 1995, pp. 355-359.
44. D. Ravichandran and R. Roy, Muter. Res. Bull. 1996,31, 1075-1082.
45. A. M. Szymanski, J. Chem. Vap. Dep. 1996,4, 278-285.
46. E. Abgarowicz, A. Bakon, S. Maciejowska, R. Salacinski, A. Niedbalska, and A. M. Szymanski,
in Proceedings of the Joint XV AIRAPT & XXXIII EHPRG International Conference Warsaw,
W. A. Trzeciakowski (Ed.), World Scientific Publishing Co., Singapore, 1995, pp 247-249.
47. Y. G. Gogotsi and M. Yoshimura, J. Am. Ceram. SOC.1995,78, 1439-1450.
48. Y. G. Gogotsi and M. Yoshimura, Nature, 1994,367, 628-630.
49. Y. G. Gogotsi and M. Yoshimura, M R S Bull. 1994, XIX, 3 9 4 5 .
50. Y. G. Gogotsi, in Korrosion und VerschleiJ von keramischen Werkstoffen, R. Telle and
P. Quirmbach (Eds), DKG, Aachen, 1994, pp. 114-122.
References 389
51. T. Kraft, K. G. Nickel, and Y. G. Gogotsi, J. Muter. Sci. 1998, submitted.

52. N. S. Jacobson, Y. G. Gogotsi, and M. Yoshimura, J. Muter. Chem. 1995, 5, 595-601.
53. Y. G. Gogotsi, M. Yoshimura, M. Kakihana, Y . Kanno, and M. Shibuya, Ceram. Trans. 1995,
51, 243-241.
54. Y. G. Gogotsi, K. G. Nickel, D. Bahloul-Hourlier, T. Merle-Mejean, G. E. Khomenko, and
K. P. Skjerlie, J . Muter. Chem. 1996, 6 , 595-604.
55. Y. G. Gogotsi, K. G. Nickel, and P. Kofstad, J. Muter. Chem. 1995, 5, 2313-2314.
56. V. D. Andreev, T. A. Nachalnaya, and E. V. Gabrusenok, Sverktverd. Muter. 1993, 2, 11-18.
51. R.Roy, D. Ravichandran, A. Badzian, and E. Breval, Diamond Rel. Muter. 1996,5, 913-916.
58. J. Shropshire, P. P. Keat and, P. A. Vaughan, Z. Kristallogr. 1959, 112, 409413.
4 Chemical Vapor Deposition of Diamond Films

C.-P. Klages
4.1 Introduction
The development of the scientific knowledge of diamond, of its properties and com-
position, and the various attempts to synthesize this much sought-after material
form an exciting chapter in the general history of science. Several review articles
covering historical aspects of diamond in general or of its synthesis under conditions
of metastability in particular are available. A recent short history of diamond
synthesis, in which older sources back to the year 1711 are quoted, was written by
Angus [l]. DeVries et al. [2], in an article containing numerous references to earlier
publications, laid their emphasis on the contributions of Russian researchers.
The modern era of diamond growth can be traced back to the days between 26
November 1952 and 7 January 1953, when Eversole at the laboratories of Union
Carbide Corporation in the USA achieved the first overgrowth of diamond from
low-pressure gas phases on the surface of diamond powder particles. Thus the
growth of diamond under pressure-temperature conditions at which graphite is
the thermodynamically stable phase of carbon, obviously predates the first success-
ful high-pressure, high-temperature syntheses of diamond at ASEA in Sweden in
February 1953 and at the General Electric Company in the USA in December
1954 [l]. Eversole was the first to synthesize diamond by any method, the procedure
used by him, however, was far from economic viability, owing to a very low yield
and the necessity of repeated removal of simultaneously deposited graphite by a
treatment in an atmosphere of pure hydrogen. Additionally, there was much
doubt in the scientific community that diamond could be grown at all under condi-
tions of thermodynamic metastability. Therefore the work of Eversole did not
attract enough attention to be a starting point of much more intensified efforts.
The results of Eversole were confirmed about 15 years later independently by the
group of Angus at Case Western Reserve University in the USA and by researchers
of the Physical Chemistry Institute of Moscow.
The important role of hydrogen in the low-pressure synthesis of diamond was
recognized in the work of Angus. Molecular hydrogen suppresses the spontaneous
nucleation and growth of graphite during the formation of diamond, but it does not
completely prevent the formation of the stable carbon allotrope, which has to be
removed from the deposit in a separate etching step. Therefore it was a very
important breakthrough in the efforts to grow diamond by a chemical vapor
deposition process, when in the seventies the Soviet group found out (stimulated
by results of Case Western Reserve University group on the action of atomic
hydrogen as an efficient remover of nondiamond carbon phases), that in the
presence of a superequilibrium concentration of atomic hydrogen diamond could
be deposited continuously, with much higher growth rates, largely free from
4.2 Preparation Methodsfor Diamond Films 391
graphite and not only epitaxially on diamond crystallites but also on nondiamond
substrates.
This Russian achievement can most probably be dated back to the early 1970s.
Nevertheless, it took another decade, until (thanks to a series of papers from the
Japanese NIRIM (National Institute for Research in Inorganic Materials), reveal-
ing detailed experimental information) the possibility and the potential of diamond
growth from the gas phase was recognized by the general scientific community. The
Japanese work, connected with the names of Kamo, Matsumoto, Sato, and Setaka,
had started in 1974, based on results in the Russian literature. Among various
diamond CVD processes, developed over about 15 years, the hot-filament CVD
process (HF-CVD) and the microwave-plasma CVD method (MWP-CVD), for
the first time described in detail by the Japanese group, have continued to be the
most used diamond deposition processes.
4.2 Preparation Methods for Diamond Films

At pressures below the diamond-graphite equilibrium line of the carbon phase
diagram, graphite will eventually be the only product of any gas-phase pyrolysis
process of a carbon-bearing atmosphere running very close to chemical equilibrium
and leading to the deposition of carbon. All continuously working diamond
deposition methods, on the other hand, have as a common feature the presence of
a pronounced gas-phase nonequilibrium in close vicinity to the substrate, main-
tained by some sort of gas phase activation. As an empirical fact, the gas phase
must contain, in addition to a carbon carrier, a powerful carbon etchant like
oxygen, fluorine or (as indicated above) atomic hydrogen. The nonequilibrium
situation close to the growing diamond film can be maintained either by thermal
means (external heating like in the hot-filament process or by a chemical reaction
as in the combustion (oxyacetylene) CVD method) or by the generation of a
plasma, sustained by direct current (d.c.), radio frequency (RF) or microwave
(MW) electrical fields.
A thorough comparison of diamond CVD processes, including a discussion of
their advantages and drawbacks, was given by Bachmann and Lydtin [3]. It was
pointed out by these authors that typical linear growth rates, which can be achieved
by these methods, correlate positively with the estimated gas phase temperatures,
reaching approximately 1 mm h-l in plasma jets running at pressures between
several 100mbar and 1 bar at gas temperatures exceeding 4000K, see Fig. 1.
Growth rates less than 1 pm h-' are obtained in thermal decomposition processes
as investigated mainly in the 1950s and 1960s and in nonisothermal plasmas at
pressures typically below 1 mbar, in which heavy-particle temperatures are at least
an order of magnitude smaller than the electron temperatures (ECR in Fig. 1 stands
for electron cyclotron resonance, a special form of a microwave plasma supported
by a magnetic field at pressures in the pbar region, see below). An intermediate
position is occupied by low power density MW plasmas and hot-filament activated
392 4 Chemical Vupor Deposition of Diamond Films
1000 -
DC arcjet
100 *
10.
10
V oxy-acetylene
flames
r I pmlh
1 low-power density MW plasmas
hot filaments
0.1
non-isothermal DC, RF or ECR plasmas
thermal decomposition
0.01
1000 2000 3000 4000 5000 6000 71 DO
gas phase temperature I K
Figure 1. Correlation of typical gas temperatures and growth rates r (in pm h-') of diamond CVD
processes (after [3]). See text for the explanation of abbreviations.
processes, running typically at pressures beyond 10 mbar and allowing growth rates
of several pm h-' .
In the following, the HF-CVD method will be presented in some detail because of
its simplicity and scientific as well as technical importance (thanks to its up-scaling
potential). Several results of studies on HF-CVD processes are of fundamental
importance for an understanding of diamond deposition in general. The presenta-
tion of plasma-based methods will focus on microwave-activated processes because
they are probably the most often used diamond CVD growth methods for virtually
impurity-free material.
4.2.1 Hot-filament Chemical Vapor Deposition

The essential feature in the HF-CVD process which turns the pyrolysis of a
hydrocarbon from a graphite-forming process to a diamond CVD method, is a
hot refractory metal wire as an atomic hydrogen source in the close vicinity of the
substrate. The elementary principle was described by Matsumoto et al. in 1982 [4,5],
see Fig. 2. The researchers at NIRIM used coiled tungsten filaments, mounted
roughly 1 cm from the substrate surface and heated to about 2000°C by passing
an electrical current. The temperature of the substrate was usually 700-1000"C, as
measured directly at the substrate holder. From a mixture of hydrogen and methane
(about 1 vol-%) as a feed gas (10-1 00 cm3min-I), diamond was formed at pressures
between roughly 10 and l00mbar. Dependent on the substrate conditions, the
deposits consisted of cubo-octahedral or multiply twinned isolated particles or
closed films of diamond. A substrate scratching with diamond powder is not expli-
4.2 Preparation Methods,for Diamond Films 393
To pump
Figure 2. Schematic view of the HF-CVD deposition chamber as used by Matsumoto et al.
(adapted from [4,5]).
citly mentioned in these papers; it is however probable that this or a similar kind of
pretreatment was used in order to achieve a dense nucleation and early coalescence
of the crystallites to closed films. Figure 3 shows the surface of a closed diamond film
with well-shaped facets grown by HF-CVD in the author’s laboratory. The typical
appearance of isolated, non-twinned particles, is shown schematically in Fig. 5.
There can be no reasonable doubt that the diamond phase of carbon was actually
Figure 3. Scanning electron micrograph of a diamond film grown by HF-CVD at Fraunhofer-IST

(by courtesy of L. Schafer).
394 4 Chemical Vapor Deposition of Diamond Films
formed by the HF-CVD method as run at NIRIM: Scanning electron micrographs

(SEM) clearly showed the cubo-octahedral morphology typical for a material with a
cubic crystal lattice, electron diffraction data were consistent with ASTM values and
the Raman peak at 1334.5cm-' was very close to the position reported for diamond
single crystals. SEM photographs, X-ray or electron diffraction, and especially
Raman spectroscopy are still the most conclusive methods for identifying the
diamond phase and these methods have therefore been used routinely in many
subsequent publications on diamond CVD.
Linear growth rates around 1 pm h-' are obtained under the conditions quoted
above. A growth velocity of this order of magnitude can still be considered as typical
for HF-CVD processes employing hydrocarbon-hydrogen mixtures forming good
quality diamond. Numerous investigations of the influence of deposition parameters
on diamond growth rates and film qualities have been reported in the literature.
Several important factors can be summarized from various investigations leading
to basically identical results.
(i) There is generally hardly any nucleation of HF-CVD (or MWP-CVD) diamond
on a smooth, thoroughly cleaned substrate. Therefore, a suitable substrate
pretreatment is essential in order to increase the nucleation density, leading to
a faster formation of a closed film and better utilization of growth species.
This can be done, for example, by ultrasonic irradiation in an aqueous diamond
powder suspension or by lapping with diamond paste. Diamond residues are
most probably responsible for this effect.
(ii) A maximum growth rate is achieved at substrate temperatures between 850 and
950°C: the exact figure depends on the pressure used.
(iii) The growth rate increases with the filament temperature and higher growth
rates are achieved at lower filament-substrate distances, thanks to an increased
flux of H atoms to the substrate 'driving' the formation of diamond.
(iv) Increasing diamond growth rates can also be achieved with methane concentra-
tions growing from 0.1 to 2 vol-% in hydrogen, however the film quality will
deteriorate due to increasing amounts of sp2-bonded carbon.
(v) As a function of gas pressure, the growth rate has a maximum at pressures
typically between 10 and 100 mbar, dependent on the methane concentration,
and the temperatures of filament and substrate.
The positive effect of increasing carbon concentration in the feed gas on the linear
growth rates, accompanied by a deterioration of the film quality under otherwise
unchanged conditions, is a very general effect in all diamond CVD processes. There-
fore, a compromise between growth rate and film quality will have to be made,
depending on economic considerations and the film properties required for a specific
application.
In the early HF-CVD studies tungsten was exclusively used as a filament material.
Among other refractory metals, only tantalum and rhenium have since then gained
greater importance. An advantage of tantalum over tungsten is its much smaller
carburization rate: W as well as Ta are thermodynamically unstable with respect
to carbides Me2C or MeC under typical diamond forming conditions; however
the carbide-forming reaction which severely deforms and embrittles the filaments,
4.2 Preparation Methods for Diamond Films 395
is generally significantly slower for a tantalum wire. Under comparable conditions,

Ta keeps its mechanical strength longer than a W filament of the same thickness,
thanks to a core of uncarburized material [6]. In contrast to Ta and W, Re does
not form a carbide, albeit it dissolves a certain amount of carbon. Therefore
rhenium filaments, if used under suitable conditions, can retain their ductility and
do not show significant dimensional changes [7]. However, the much higher price
of Re has to be taken into account, if deposition of diamond on a larger than labora-
tory scale is considered.
The role of the hot filament in diamond HF-CVD is clearly to dissociate
molecular hydrogen in order to meet the requirement for a continuous growth of
the diamond phase. The question, if there are any differences between filament
materials (for the same geometries and surface temperatures), as far as the hydrogen
dissociation capabilities and achievable growth rates are concerned, is closely
related to chemical surface reactions involving carbon and hydrogen. If the concen-
trations of carbon- and hydrogen-bearing species on and near to the filament surface
are close to equilibrium concentrations at the filament temperature, no significant
effects of the filament material on gas phase composition near the substrate will
be expected. From the experimental results reported in the literature, no clear con-
clusions can be drawn with respect to this question: In a comparison of tantalum
and tungsten filaments Singh et al. [8] did not find significant differences in diamond
growth rates, and the authors concluded that the dissociation of the gas mixture is
independent of the filament material. In contrast, distinct differences between
rhenium, tantalum, and tungsten filaments in HF-CVD of diamond were found
by Okoli et al. [6]. In order to obtain well faceted good quality diamond films
different methane concentrations were needed for the three materials at a filament
temperature of 2200°C. According to results of Sommer and Smith [7], material
and temperature dependent formation of carbon contamination surface layers,
reducing the filament activity, has to be taken into account in interpreting com-
parisons of this kind.
Gas phase diagnostic investigations also show differences in the hydrogen disso-
ciation capabilities of different filament materials: In methane-containing gas phases
the quantitative determination of H concentrations by two-photon laser-induced
fluorescence yielded values below the equilibrium concentrations calculated for
the filament temperatures. Among the materials investigated (W, Ta, Ir) the highest
concentrations of atomic hydrogen were generated by W filaments [9]. The deviation
of molecular hydrogen dissociation from thermal equilibrium, indicating material
specific surface effects, is also indicated by higher degrees of dissociation in pure
hydrogen gas phases, where no carburization or carbon impurities affect the
filament surface.
Aside from the filament stability and its hydrogen dissociation capability the
incorporation of evaporated filament material into the growing diamond films is
one aspect which has to be taken into account in HF-CVD: Quantitative determi-
nations of filament-related impurities in diamond films have been carried out for
rhenium, tantalum and tungsten filaments. For filament temperatures between
1900°C and 2350°C the amount of incorporated Re increases by three orders of
magnitude to a maximum of about 0.1 at-% [lo]. Using Rutherford backscattering
(RBS) a tungsten concentration around 0.1 at-% was measured by Gheerhaert et al.
in films grown at filament temperatures of about 2250°C [Ill. For tantalum
filaments at 2300°C maximum Ta concentrations of 0.03 at-% were detected
using secondary mass spectrometry (SIMS) [121. For carbide-forming filament
materials the degree of filament material incorporation strongly depends on the
carburization state of the filaments. This is caused by the fact that at the relevant
temperatures the carbides have lower vapor pressures than the corresponding
uncarburized materials [ 121. Therefore, in several studies relatively high amounts
of filament impurities were found at the diamond-substrate interface when the
deposition process was started with uncarburized filaments.
The mostly used gas atmospheres for hot-filament deposition of diamond gener-
ally consist of mixtures of 0.1 to a few vol-% methane and hydrogen. In contrast to
plasma-activated deposition methods, additions of oxygen-containing source gases
in HF-CVD of diamond are possible only to a limited extent, due to the tendency of
oxidation and volatilization of the filament material. Nevertheless, additions of
oxygen, water vapor, or oxygen-bearing organic compounds to the HF-CVD gas
feed can have several advantages if the quantitative gas phase composition is
carefully chosen. This was first demonstrated by Hirose and Terasawa, who used
a range of oxygen- and nitrogen-containing organic compounds [13]. Diamond
films with good crystallinity and high phase purity (according to Raman and
electron diffraction characterization) were obtained at growth rates about an
order of magnitude higher than typical for oxygen-free gas feeds. Compared with
methane-hydrogen mixtures, much higher gas pressures, up to nearly one atmos-
phere, and gas phase concentrations of the carbon-bearing species were possible
in these experiments. These results were later essentially confirmed by other authors,
achieving growth rates up to 10 pm h-' by using ethanol and acetone (1-8 vol-%) in
hydrogen at pressures up to 13 kPa [14].
Kawato and Kondo carried out a comparison of CH4-H2 and CH4-H2-02
mixtures and showed that an increased growth rate and extended pressure and con-
centration ranges (40 kPa, 4 vol-%) are also obtainable with additions of molecular
oxygen [15]. Numerous investigations have been performed in order to reveal the
reasons for the beneficial effects of oxygen-containing gases observed in diamond
HF-CVD (and MWP-CVD) [15-201. Results of these studies can only be sum-
marized here without further comments. The effects of oxygen have generally
been attributed to
- promotion of hydrogen dissociation and methane decomposition,
- a reduction of the acetylene concentration, leading to less nondiamond carbon
phases,
- formation of active growth sites by improved hydrogen abstraction, and
- an efficient removal of non-diamond carbon phases by hydroxy radicals.
As shown by Bachmann, there are more general limits to the C,H,O-concentrations
which can be used in diamond HF-CVD and similarly in most other diamond CVD
processes. A compilation of diamond-forming gas phase compositions from a large
number of publications shows that (within a ternary C,H,O-concentration diagram)
those concentrations leading to diamond growth are restricted to an essentially
4.2 Preparution Methods for Diamond Films 397
wedge-shaped area extending from the H-corner along the CO line of the triangle
[21]. Smaller carbon concentrations do not lead to the formation of deposits at
all, due to the dominance of etching processes over carbon deposition. At con-
centrations higher than those situated in the diamond wedge sp2-coordinated
carbon phases tend to dominate.
As a key to the growth of fiber-textured diamond films, additions of molecular
nitrogen or nitrogen-containing compounds have in the recent years gained
increased interest, see below. Homoepitaxial growth experiments of Cao et al. [22]
showed, that nitrogen strongly enhances the growth rate of { 100) facets. This effect
can be attributed to nitrogen compounds breaking the dimers of the (2x1)
reconstructed {loo} surface, thus favoring the (100) growth direction.
4.2.2 Microwave-plasma-based Methods

As indicated above already, diamond growth rates beyond a few 100 nm h-' gener-
ally require gas phase temperatures above 2000K. Aside from hot filaments and
combustion flames, various kinds of plasmas are in use today to heat the gas
phase. Depending on the ratio of average energies of electrons and heavy particles
(neutrals and ions), electrical gas discharges can be divided into nonisothermal
plasmas, characterized by a large difference between the average energies (or tem-
peratures) of electrons and neutrals (or ions), and isothermal plasmas, in which
electrons and heavier particles have virtually the same temperature. In the pressure
region below about 1 mbar, for example in sputtering, many low-pressure plasma-
CVD or glow-discharge polymerization processes, average electron energies of a
few electron volts, corresponding to several l0000K, are quite typical while the
gas phase temperatures usually do not exceed a few hundred degrees Celsius. The
success of plasma CVD methods in diamond growth is generally based on the use
of higher pressures, mostly beyond 10 mbar and up to 1 bar. Under these conditions,
a local thermal equilibrium prevails between the electrons, gaining their energy from
an electrical field, and heavier particles which are heated by frequent collisions with
the hot electrons. As an example, the use of microwave plasmas will be discussed in
the following sections.
4.2.2.1 Isothermal Microwave Plasma Chemical Vapor Deposition

Aside from hot-filament chemical vapor deposition, microwave plasma CVD
(MWP-CVD) of diamond has become the most popular method for growing
diamond films, since the pioneering paper of Kamo et al. appeared in 1982 [23],
in which the experimental setup and deposition conditions were described in
detail. The tubular apparatus, as used by the NIRIM researchers and constructed
in many other laboratories subsequently, typically consists of a silica reactor tube
running through two opposing bores in a rectangular microwave guide, connected
to a 2.45 GHz magnetron source. Using a stub tuner between the magnetron and the
silica tube and a plunger at the end of the waveguide, a position of maximum MW
electrical field strength can be generated within a section of the tube, just beyond a
substrate table. Owing to the geometric limitations of suitable waveguides at
2.45GHz, quartz tubes are limited in diameter to around 5cm. Therefore the
coatable area on a substrate is usually between 1 and 10cm2. The experimental
conditions leading to successful growth of diamond are very similar to those in
HF-CVD: Pressures are typically between 10 and loombar, the feed gas in the
first NIRIM publication consisted of 1 vol-YOmethane in hydrogen and the sub-
strate was heated to 800-1000°C in a plasma generated at several 100W to 1 kW.
Under these conditions, growth rates between 0.5 and 3 pm h-’ can be achieved,
the lower value being more realistic for well faceted, high phase purity films.
In early MWP-CVD experiments oxygen-free gas feeds, generally 0.1-1 vol-%
methane in hydrogen, were commonly in use. However, since it was shown by
Hirose and Terasawa and others, that higher growth rates and better film qualities
at larger pressures and carbon concentrations could be achieved using oxygen com-
pounds in HF-CVD [13], it became popular to use oxygen-containing compounds
like 02,H 2 0 , CO, C02, alcohols, ethers, and others in microwave plasma CVD
of diamond. Independent of the gas composition actually used, hydrogen, methane,
and acetylene are the prevailing components of the in situ and exhaust gas of typical
MWP-CVD processes in C,H gas phases. Oxygen-containing components are
predominantly converted to carbon monoxide. Oxygen additions suppress the
acetylene formation and lead eventually to a practically complete disappearance
of hydrocarbons in favor of CO. The spectrum of stable products is fairly similar
to that of HF-CVD processes, pointing to a basic similarity of both types of
processes [24]. In fact the main role of the plasma in MWP-CVD (analogous to
the filament in HF-CVD) seems to be the dissociation of molecular hydrogen into
atoms, which initiate a reaction chain leading to the methyl radical as the most
important diamond growth species, and help to gasify unwanted sp2-bonded
carbon deposits (see below).
While the tubular reactor design was used in most laboratories working on MWP-
CVD of diamond in the mid-80s, commercial reactors appeared on the market after
a few years. Major drawbacks of the NIRIM-type reactors are the limited substrate
size and difficulties in preventing the plasma from interactions with the tube
material, leading to etching and redeposition of silicon contaminations in the
films, localized heating, deformation, or even puncturing of the tube. Also, MW
power independent adjustment of the substrate temperature by external heating is
difficult. The development of MWP-CVD sources at ASTeX, Applied Science and
Technologies, at Woburn, MA, USA was outlined by P. K. Bachmann [25], who
also contributed to the design of the so called ‘bell jar’ reactor which was the first
microwave plasma CVD apparatus permitting growth of diamond on substrates
up to 7.5cm in diameter and became the first commercially available diamond
CVD reactor at all. This reactor contains a dome-shaped silica bell jar of lOcm
diameter in which a plasma is generated at a location of maximum electrical field
strength well apart from the wall material.
Today the ASTeX high-pressure microwave source reactor (HPMS) is in use in
many laboratories. A schematic cross-section of this apparatus is shown in Fig. 4
[25]. Microwave power is transmitted by a coupling device from a rectangular wave-
4.2 Preparution Metkods,for Diamond Films 399
MW
=i inlet
gas ,- substrate
t o pump
+----
Figure 4. Schematic cross-section through a modern 2.45 GHz ASTeX HPMS (‘High-pressure
Microwave Source’) MWP-CVD reactor. Microwave radiation is coupled through a silica
window separating the atmospheric pressure region within the waveguide and around the antenna
from the plasma region under reduced pressure. The reactor is equipped with a R F induction-
heatable graphite stage and can be used for substrates up to lOcm diameter. Modified by a
water-cooled graphite stage, reactors of this type can be used for up to 5 kW MW power. Inspection
windows are not shown here (adapted from [25]).
guide through a silica MW window into a cylindrical double-walled water-cooled

steel shell equipped with quartz windows for process inspection and diagnostic
investigations. The substrate is located on a graphite substrate stage independently
heatable from below by a R F induction coil. The system was introduced with a
1.5 kW MW magnetron and power supply and can now be obtained in a modified
version equipped with a 5 kW magnetron and a water-coolable substrate stage.
Growth rates up to 15 pm h-’ can be obtained at 5 kW using suitable process param-
eters. However, process conditions leading to very phase-pure material as usually
needed for optical, electrical, or thermal purposes, always lead to much smaller
deposition rates. Under suitably optimized deposition conditions, reactors of this
type are very stable and can be left running unattended for days or even several
weeks if very thick diamond films are needed.
In order to coat substrates with diameters in excess of 4 inches, reactors working
at lower frequency have been developed in recent years. Using a 75 kW source oper-
ating at 91 5 MHz, an apparatus was built by ASTeX permitting a coatable area of 8
inches diameter and a total mass deposition rate which is a factor of 10-100 larger
than for the 5 kW HPMS source [25].
400 4 Chemical Vapor Deposition o j Diamond Films
4.2.2.2 Non-isothermal Plasma Deposition

A different approach was pursued by Kawarada, Suzuki and others [26,27], trying
to increase the size of the microwave plasma by operating it at pressures well below
1 mbar. In the case of MW fields, it becomes increasingly difficult to sustain a plasma
under decreasing pressure owing to insufficient energy absorption by the electrons
from the oscillating electrical field in the absence of frequent collisions with heavy
particles. The number of ionizing collisions will be decreased and eventually the
loss rate of electrons by recombination and diffusion to the chamber walls will
become larger than the generation rate; the plasma will be extinguished. In order
to operate a plasma sustained by microwave radiation down to pressures of
lop4mbar, electron cyclotron resonance (ECR) sources have gained a certain
importance in the field of thin film deposition and etching. The Lorentz force
imposed by a magnetic field of 875 G field strength normal to the oscillating electri-
cal field of 2.45 GHz will force the electrons onto closed circular loops (cyclotron
motion), allowing a much higher average electron energy. However, in case of
diamond CVD deposition, use of an ECR source at pressures below 0.1 mbar,
which is about the maximum pressure for the ECR mechanism to come into play,
yields only fairly small growth rates and a poor crystallinity and phase purity of
the deposit. In a wide range of experiments conducted in the author’s laboratory
at pressures below mbar, no carbon deposition could be detected at all. This
is just one example of the failure of conventional nonisothermal plasmas, which
are frequently used for thin film deposition and etching, to grow diamond films
with acceptable rate and quality.
4.2.3 Preparation of Special Forms: Textured and

Heteroepitaxial Films
Throughout the first decade of the modern diamond CVD era, counted from the
first NIRIM papers in the early 80s, it was common practise to treat a substrate
to be coated by scratching with diamond paste in order to enhance film nucleation.
Diamond films of a few micrometers thickness, grown on such a substrate, consist of
randomly oriented crystallites because neither the nucleation nor the growth process
introduce any significant ordering of crystallite orientations in the film. The
development of methods allowing the growth of diamond films with an increased
orientational order (i.e. a reduced degree of orientational freedom) of crystallites
was not only of academic interest; for many application of diamond films it
would be of considerable importance to control the orientation of crystallites
during film growth and thereby improve the film morphology or physical character-
istics like the optical or (thermal and electrical) transport properties. The ultimate
goal would doubtlessly be the growth of large-area films approaching the quality
of single crystals.
Two methods will be presented here allowing the preparation of films in which the
crystallites have lost two and three degrees of freedom. By an appropriate control of
the growth process, diamond films with fiber texture can be fabricated, in which
4.2 Preparation Methods for Diamond Films 40 1
special crystallographic axes of individual crystals are restricted to make a given

angle with the film normal, which in a perfectly textured film is zero (loss of two
degrees of freedom). Three degrees of orientational freedom are lost in epitaxial
and heteroepitaxial films, in which the regularity of a substrate lattice imposes a
complete, or nearly complete, orientation on the (nuclei of the) growing film.
4.2.3.1 Textured Diamond Films

The formation of a texture, that is, a preferred orientation of crystallites, during
deposition of diamond is a nice example of a mechanism which can be observed
during mass crystallization of many materials. The solidification of a melt in a
container, for example, usually starts simultaneously at many nucleation sites on
the container walls. During the competitive struggle among the growing crystallites
only those with fastest growth directions close to the substrate normal survive. Crys-
tals with unfavorable orientations will die out during growth, leaving a continuously
decreasing number of still-growing crystals, accompanied by an increase of their
frontal area. This process of ‘geometric selection’ was analyzed theoretically by
Kolmogorov in 1945, leading to his law n(h) 0; l/h”* for the crystallite population
density n at an ingot thickness h [28]. In 1967 the principle was re-named by Van der
Drift as an ‘evolutionary selection’ and discussed quite generally for various situa-
tions playing a role in vapor-deposition of thin films (crystal morphology, angle of
incidence of growth species, surface diffusion) [29]. For an isotropic supply of
growth species, freely growing crystals and crystals in a film develop in the same
way: The direction of fastest growth in the isolated crystal is the direction from
the center to the remotest point of the crystal. In a film, the crystals which are the
‘fittest’ for survival during the evolutionary selection, are those which have this
direction most parallel to the average growth direction.
Texture formation in growing diamond films is closely related to the morphology
of individual diamond crystallites formed under identical conditions. In discussing
crystal morphologies, on must differentiate between the equilibrium shape of a
crystal, which is determined by the thermodynamics of equilibrium between, for
example, vapor and solid phase, taking into account the orientation dependence
r(n) of surface energy of the solid phase, and the growth shape, determined by
the kinematics of faces during growth, allowing for the orientation dependence of
growth rates V(n).
Sharp cusp-like minima in polar diagrams of r(n) or V(n) correspond to flat faces
on the equilibrium and growth shape. Dominant faces are those with the smallest
distances from the origin, that is, those with smallest surface energy and growth
rates. Frequently growth shapes are quite similar to equilibrium shapes, as lowest
growth rates are expected for faces parallel to a maximum number of atomic
chains connected by strong bonds, so-called F-faces (flat faces) containing at least
two different ‘periodic bond chains’ (PBCs [30]), characterized by the lowest surface
energies. In the absence of defects like screw dislocations, growth on these faces has
to overcome the barrier for the formation of two-dimensional nuclei of new layers.
Higher surface energies and growth rates are generally found for S-faces (Stepped)
and K-faces (kinked), characterized by the presence of only one type of parallel
PBCs or the lack of parallel PBCs. Growth on these faces involves the addition of
growth units to pre-existing steps or kinks, without an activation barrier. These
interrelations between surface energy and growth rates (in spite of often being
able to describe general features of crystal morphology) do not necessarily hold
always and exactly: Growth rates of different faces of crystals are often observed
to depend differently on growth conditions like temperature, supersaturation
and kind of growth species, impurities acting as inhibitors or catalysts and so on.
Diamond CVD is a nice example illustrating the influence of growth conditions
on crystal shape and film texture.
A PBC analysis of the diamond lattice leads one to expect ( 11 1) faces (containing
at least two different types of periodic bond chains) as the only well-defined flat
faces of diamond crystallites. (001) faces should be kinked (K-faces) and appear
(if at all) as curved surfaces owing to the lack of pronounced cusp-like growth
rate or surface energy minima in (001) directions. This expectation is, however, in
conflict with experimental evidence, showing that (001) faces of diamond and
other crystals with a diamond lattice are generally slow-growing well-defined flat
faces of similar or even larger morphological importance than ( 111) faces. Giling
and Van Enckevort explained this prominence of (001) faces in terms of a surface
reconstruction whereby surface atoms are able to lower their energy by dimerization
with neighboring atoms, thus reducing the number of dangling bonds at the surface.
The surface reconstruction process leads to F-face-like behavior of (001) faces, that
is, growth by two-dimensional nucleation of new layers [311.
A cubic crystal bounded by (001) and (111) faces generally forms a cubo-
octahedron which degenerates into a cube or an octahedron as the limiting shape
for vanishing portions of ( 111) and (001) faces. The relative share of these two
types of faces depends on the ratio of growth rates Vool/Vlll. In Fig. 5 crystal
shapes are shown for a number of ‘growth parameters’ a, defined by
31/2VooI/ V11 [32]. The arrows indicate the largest diameter, which is the direction
of fastest growth of an individual crystal. In order to realize a cubic diamond crys-
tallite, the growth rate ratio has to become Vool/V111I 3-‘/2 = 0.58; for a pure
octahedron a ratio 2 31/2= 1.73 is required.
As shown by Clawing et al. [33], all forms of cubo-octahedrons, including the
limiting forms of cubes and octahedrons, can be produced as single isolated crystal-
lites by a suitable selection of growth conditions, that is, growth rate ratios
0.58 V0ol/Vl11 2 1.73 are accessible. Thick films grown under the extreme
condition of a growth rate ratio near 0.58 (a = 1) exhibited a (1 11) texture with
1 1.5 2 2.5 3
Figure 5. Morphologies of cubo-octahedral crystals, exhibiting (001) and { 11I } facets, grown at
different growth parameters, a. Arrows indicate the directions of fastest growth (Reprinted from
Diamond and Related Materials, 3, C. Wild et al., Oriented CVD diamond films: twin formation,
structure and morphology, pp. 373-38 1, Copyright 1994, with permission from Elsevier Science).
4.2 Preparation Merhodsfor Diamond Films 403
Figure 6. SEM micrographs of a (001)-textured diamond film with {OOl) morphology (by courtesy
of M. Fryda).
(001) facets as to be expected from the considerations made above. Similarly, (001)-
textured films with { 11 1) facets were obtained in the opposite extreme (a! = 3),
where pure octahedra were grown as single crystallites. Upon a reduction of a! to
about 1.9, corresponding to single crystallites appearing as octahedra with truncated
corners, a material dominated by (001) facets, with (001) about 10-15" from the
substrate normal, was formed. Figure 6 shows the surface of a film of this kind,
grown at the author's institute.
Growth of diamond films is frequently accompanied, however, by the formation
of twins stabilizing a pronounced (011) texture. An analysis of the formation
mechanism of this type of texture was presented by Wild et al. According to the
principle of evolutionary selection, a (01 1) texture should only occur at cy = 1.5,
with protruding corners formed by {OOl} and { 11 1) faces. However, roof-shaped
structures bounded only by { 11 1) faces were observed in thick (01 1)-textured
films [34]. Taking into account the effect of twinning on (111) faces the authors
presented arguments for the destabilization of growth of (001) tips in favor of a
preferential (01 1) growth direction, forming the observed roof-shaped appearance.
As shown in [34] and [35], high values of a are favored by high methane concen-
trations and low deposition temperatures. These findings are in agreement with
results of direct growth rate measurements performed on single cubo-octahedral
crystallites and homoepitaxial films, showing a stronger methane-concentration
dependence of the [OOl] growth rate, however a larger activation energy (tem-
perature dependence) for [11 11 growth [36]. Knowledge of the growth parameter
dependence of a permits a large scope in tailoring film texture and morphology
independently. Thus, using a two step procedure, the growth of smooth, perfectly
(001)-textured films with (001) morphology is made possible [37].
The concentration of molecular nitrogen in the source gas was recognized as a
very important third parameter determining the growth morphology and texture:
Small amounts (10-100 p.p.m.) of nitrogen were found to improve the crystalline
perfection and favor the formation of smooth (001 }-faceted (001)-textured films
considerably [22]. An analysis of HF-CVD homoepitaxial film growth rates
showed, that the addition of 100 p.p.m. nitrogen accelerated the [ l l l ] growth by a
factor of 2.2, however, a 4.4-fold increase was found for the [OOl] growth rate.
Therefore, (Y is multiplied by a factor of 2. A probable explanation of this phenom-
enon was given in terms of the surface reconstruction by dimerization on (001)
faces, leading to the F-face type behavior (see above): The adsorption of nitro-
gen-containing species could partially break this surface reconstruction, leading to
the formation of kink sites and an enhancement of the growth rate on this face.
An alternative mechanism leading to the growth of textured diamond films is
related to the bombardment with low-energy ions [38]. By using a carefully
controlled substrate bias-voltage during film growth, films with (001) texture and
(001) morphology could be obtained on diamond (film) substrates of any orienta-
tion. Seemingly the ion flux leads to a renucleation and a preferential growth of only
those crystallites which are oriented with one [OOl] direction parallel to the ion flux
direction, possibly due to their higher stability with respect to a flow of etching H+
ions.
4.2.3.2 Heteroepitaxy of Diamond Films

4.2.3.2.1 Bias-induced Heteroepitaxy on Silicon Substrates In the early nineties it
was discovered independently by research groups in the USA and in Germany,
that heteroepitaxial nucleation of diamond can be achieved on 3C-Sic and silicon,
by applying a negative substrate bias voltage to the substrate for a limited period of
time preceding conventional microwave plasma-assisted diamond CVD (MW-
PCVD) 139,401. The work at the author’s institute was started in 1991, inspired
by literature on the ‘beneficial’ effects of ion beam assistance on heteroepitaxial
nucleation and growth. The use of an ion beam during MW-PCVD is precluded
due to the low mean free path length of ions at the pressures typically used. The
energy of positively charged species arriving at the substrate can, however, easily
be increased beyond the l-lOeV region, which is typical for self-bias effects, by
connecting the substrate holder to an external d.c. voltage source.
The bias-induction of diamond nucleation had already been reported by Yugo
et al. in 1991 [41], and thoroughly studied also by Stoner et al. [42]. However,
they found no evidence of the formation of heteroepitaxially oriented nuclei, due
to the selection of experimental parameters in their studies.
Using a commercial MW-PCVD reactor equipped with a substrate bias, nuclea-
tion and growth of epitaxially orientated diamond films can be achieved on silicon
substrates by applying a three-step procedure consisting typically of:
- plasma etching of the substrate in pure H2 at 850°C and an applied substrate bias
voltage, V,, of -150V;
- a nucleation step in CH4/H2 (partial pressure ratio 1:lOO) at 850°C and
VB = -150V, 5-10min; and
- film growth in CH4/H2(0.5: 100) at 800°C.
A scanning electron micrograph of a heteroepitaxial film grown on (001) silicon is
shown in Fig. 7. The epitaxial orientation of most of the grains relative to the
substrate is obvious: The grains exhibit nearly square (001) facets bounded by
4.2 Preparation Metho& for Diamond Films 405
Figure 7. SEM micrograph of a heteroepitaxial diamond film obtained on (001) silicon by bias-
induced nucleation, film thickness =6.5 pm. Vertical and horizontal edges of the photograph are
orientated parallel to the Si [I 101 and [LlO] directions, respectively (by courtesy of X. Jiang).
edges oriented along [110] and [I101 with

(OO1)diamondIl(OO1)silicon and [l 10]diamondll[l10]silicon.
The overall four-fold symmetry of a heteroepitaxial film on silicon is also evident
from polar plots of Raman intensities taken with parallel polarization of incident
and scattered light as a function of the azimuthal angle, 4, between the polarization
vector and the [110] substrate direction.
X-ray pole figures also reveal the symmetry of heteroepitaxial films on silicon and
confirm the orientation relations given above for lattice planes of silicon and diamond.
In addition, this technique is able to give a measure of the crystallographic quality of
the films: Due to a residual misalignment between individual diamond crystallites and
the substrate (a tilt misorientation between the [OOI] directions of individual grains and
the substrate as well as a twist misorientation around the [OOI] substrate direction) the
reflections corresponding, for example, to { 11l} lattice planes in [001]-oriented films
are broadened. The full width at half maximum (FWHM) of pole figures taken on
early samples was about 12" in both the x and 4 variable [43]. The tilt component
of the misalignment can also be measured by quantitative image analyses of atomic
force microscopy (AFM) data: Tilt angles of diamond (001) facets with respect to
the substrate plane exhibit a Gaussian distribution around o", independent of the
azimuthal direction of the measurement [MI.
The latter result indicates that the perfect orientation represents an energy mini-
mum in this epitaxial system (albeit possibly a rather shallow one): Interfacial tilting
and azimuthal rotational misorientations are not necessarily a consequence of the
large lattice constant misfit between diamond and silicon. Therefore, the crystallo-
graphic quality of heteroepitaxial diamond films on silicon can be considerably
improved by adjusting the process conditions (substrate cleaning, base vacuum,
nucleation parameters, especially voltage and time). In fact, considerable progress
406 4 Chemical Vapor Deposition of’Diamond Films
in this direction has already been made; tilt angle as well as pole figure FWHM
values of less than 5” were observed in films of 10-20pm thickness; in thicker
films (100 pm), about 2” have been reached.
4.2.3.2.2 The Mechanism of Bias-induced Nucleation The mechanism of bias-

induced nucleation is presently still a matter of many investigations. Owing to the
difficulty of surface analyses under the high-pressure conditions typical of diamond
MW-PCVD, experimental clues to the kind of decisive steps are rare. Only two fairly
different hypotheses out of many discussed in the literature shall therefore be
mentioned here.
A subplantation process was discussed by Gerber et al. [45,46], due to the
pronounced voltage dependence which they found for the bias-induced nucleation
of diamond on silicon, exhibiting a sharp maximum at -225V for 2% methane.
Measurements of the ion energy distribution (IED) by a retarding field probe in
the substrate table showed that these voltages lead to ion energies of 70-80eV,
according to the authors the optimum region for the formation of dense sp3-
bonded carbon by a subplantation mechanism. The subplantation model was
originally proposed by Lifshitz et al. [47] in order to explain what was claimed
(but seemingly not confirmed later) to be an epitaxial growth of diamond on
(111) Si at room temperature from a mass-selected 120eV C+ beam [48]. Sub-
plantation according to this model involves:
- shallow implantation of hyperthermal ions (energies of the order of 100 eV) onto
subsurface interstitial sites;
- internal growth of a dense, rigid phase and simultaneously substrate surface
sputtering and dilution; and
- possible oriented growth by a ‘mold’ effect, that is, endotaxy with the surrounding
crystalline environment.
On the other hand, a more classical reaction scheme was proposed by Jiang et al.,
based on AFM and SEM observations of the temporal development of diamond
crystallites. They considered the formation, destruction and capture of ‘active
sites’, ‘germs’, and ‘nuclei’ and attributed the role of ion-bombardment to an
enhancement of adatom diffusion [49].
4.2.3.2.3 Heteroepitaxy on Other Substrates Many other substrate materials have

in recent years been used to grow epitaxially oriented diamond crystallites or
continuous films. An exhaustive review with 50 references, covering growth on
the nonmetals c-BN, Si, Sic, BeO, graphite and metals like Cu, Ni, Pt, Ir appeared
recently [50]. In spite of a good lattice match (misfit 1.3%) and a ready nucleation of
diamond on its boron-terminated {lll}, face, c-BN has hardly any value as a
substrate for technical purposes because it can only be grown by high-pressure
high-temperature synthesis with sizes of, at most, 0.5 mm diameter. Cubic silicon
carbide is an important substrate as far as potential applications are concerned,
because 3C-Sic films of high quality can be grown on Si either by chemical conver-
sion in carbon-containing gas phases or by CVD processes. Using bias-induced
nucleation, very smooth heteroepitaxial diamond films have recently been grown
4.3 Thermochemistry and Mechanism of Chemical Vapor Deposition Diamond Growth 407
on 3C-Sic films on silicon, exceeding probably the quality of other films grown on Si
or S i c before [51]. The ultimate goal of diamond heteroepitaxy is doubtlessly the
achievement of deposition of very high quality films on silicon, owing to its cheap.-
ness and ready availability of large-diameter wafers.
4.3 Thermochemistry and Mechanism of Chemical

Vapor Deposition Diamond Growth
4.3.1 Transformation of Graphite to Diamond at Low Pressures
In the early era of metastable diamond growth, the decisive role of atomic hydrogen
for the achievement of continuous diamond CVD growth was usually explained by
its etching effect on graphite; it was implied that there was simultaneously a
precipitation of diamond from the gas phase and a gasification of graphite, if
there was any. In at least two publications, it was explicitly demonstrated that in
the presence of atomic hydrogen diamond can nucleate and grow, while graphite
is simultaneously etched under formation of hydrocarbons. At first glance, this
seems to be a thermodynamic paradox because in these reactions graphite is
virtually converted to diamond, which should be impossible at subatmospheric
pressures owing to an overall increase of the Gibbs free energy:
C, --+Cd, AG;98 = +2.9 kJmole-'. (1)

It can be shown, however, that the formation of diamond from graphite under the
conditions of typical diamond CVD processes is not in conflict with classical
thermodynamics if the CVD process is looked at in some more detail [52 and refs.
therein].
An inevitable first highly endothermic step in the CVD reaction is the dissociation
of molecular hydrogen by the application of energy, for example by means of a
heated filament, a flame or an electrical gas discharge, reaction (2). The gasification
of graphite, reaction (3) on the other hand, is strongly favored thermodynamically:
H2 +2H AG$,OO
= +75 kJ mole-', (2)
C, + 4H +CH4 AGy500 = -470 kJ mole-'. (3)
Methane reacts with H forming a series of CH, radicals (x = 3.0)
CH4
CH3 + H --
+ H +CH3 + H2
CH2 + H2
CH2 + H +CHI + H2
AGy5oo = -42 kJ mole- 1 ,
AGySoo= +17 kJ mole-',
AGySoo= -38 kJmole- 1 ,
(4)
(5)
(6)
CHI+H+C+H2 AGy5oo = -84 kJ mole-', (7)
408 4 Chernicd Vapor Deposition of Diurnond Films
and carbon atoms are finally added to diamond ( c d is a carbon atom in the diamond
lattice):
c + Cd --f 2cd AGy5oo = -480 kJ mole-'. (8)
Summarizing reactions (3) to (8), one obtains a thermodynamically strongly favored
net reaction (9) for the graphite-to-diamond transformation, driven by the input of
atomic hydrogen:
C, + 8H -+ Cd + 4H2 AGy5oo = - 1097 kJ mole-', (9)
4.3.2 Reactive Species in Diamond Chemical Vapor Deposition,

the Role of CH3
For the purely thermodynamic argument given above, it is not necessary that the
direct incorporation of atomic carbon into the diamond lattice is actually taking
place in any diamond CVD mechanism. It could have been formulated alternatively
using the incorporation of a methyl radical or an acetylene molecule into the
diamond lattice as a central step in the reaction sequence. Many theoretical and
experimental investigations have focused on the question: Which gas phase species
must actually be present at the diamond surface in order to keep the crystal lattice
growing? Several molecular models, involving mostly an attachment of methyl
radicals or acetylene molecules to the diamond surface, have been proposed in
order to evaluate possible growth mechanisms. A review of these theoretical
investigations, which are not completely conclusive as to which species are actually
most responsible for CVD diamond growth, was recently given by Goodwin and
Butler [53].
Several gas-phase diagnostic studies have been performed especially on the hot-
filament CVD process in an attempt to analyze the spatial distribution of stable
and reactive species in the region between filament and substrate and to obtain
information about the decisive steps in the diamond formation mechanism. One
of the first quantitative investigations of this kind was performed by Celii et al.
using tunable infrared-laser detection of methane, acetylene, ethylene and methyl
[54,55]1.Aside from a considerable conversion of methane (starting concentration
8 x 10 ~ m - ~to) acetylene (2 x 1014cmP3) near the hot filament (tungsten,
22OO0C), methyl radicals were found with a concentration of 2 x 10l2~ m - ~ .
An indirect access to the concentrations of methyl and hydrogen atoms was
used by Harris and Weiner [56]. The authors used a quartz probe 3-5mm
from the hot (2600 K) tungsten filament and determined the concentrations of
ethylene and ethane formed by 'self-scavenging' reactions (10, 1 1) of methyl,
forming the main loss channels of CH3 at relatively low pressure within the
probe (<10 mbar):
+ +
CH3 CH3 -+C2H5 H +C2H4 H2 + (or +C2H6), (10)
CH3 + CH3 --+ C2H6. (11)
atid Mecliunisrn of Clirmicul Vripor Deposition Diurnoticl Growth
4.3 Tl~ertiioi.hmii.~tr~ 409
The mole fraction of methyl in the growth chamber was then calculated by doubling
the sum of the ethane and ethylene mole fractions. From the methyl mole fractions
H atom concentrations in the chamber were estimated by assuming a partial
equilibrium for the reaction (12):
CH4+H*CH3+H2. (12)
The methyl mole fraction increased from 2 x lop4 to 7 x lop4, when the methane
concentration was raised from 0.5 to 3.2 vol-YO. The H fraction amounted to
about 0.01 at 2vol-% methane.
The most direct experimental approach to the species involved in diamond
growth, molecular beam mass spectrometry of stable and reactive species extracted
through a 300 pm orifice in the substrate plane, was used by Hsu to investigate the
HF-CVD as well as the MWP-CVD process [57,58]. The most interesting results of
these measurements are as follows.
(i) In HF-CVD, acetylene is the predominant carbonaceous species near the
substrate, if a multi-filament arrangement is used, while methane prevails
under a single filament.
(ii) Methyl and acetylene mole fractions saturate at lo-’ and 7 x lo-’, if the
methane concentration is increased beyond 2%, accompanied by a decrease
in the H atom fraction from 2 x lop’ at methane fractions below 1% to
about 1 Op4 at 7 % methane (attributable to filament poisoning).
(iii) Owing to a rapid removal of H atoms by surface recombination, reaction (12) is
not in partial equilibrium, as assumed [56], but shifted to the methane side.
Based on these relatively early investigations, no decision could be made as to which
of the various species detected was actually responsible for diamond growth,
although in many of these studies only CH3 and C2H2 were left as possible
candidates. Very sophisticated diagnostic techniques involving nonlinear laser
spectroscopy have, in recent years. been applied to study growth atmospheres in
diamond CVD processes. However, the most conclusive information about which
species are responsible for bringing carbon to the growing diamond film came
from a number of suitably designed growth experiments. Several of these experi-
ments point to CH3 as the principal species in commonly used CVD environments,
although minor contributions of other species like acetylene cannot completely be
excluded.
Isotopic labeling experiments using 13CH4/’2C2Hz feed gases in the hot-filament
process were performed by Chu et ul. and D’Evelyn et al., first for polycrystalline
film growth [59,60], later for homoepitaxial growth on diamond {OOl}, {Oll},
and { 111) surfaces [61], and reanalyzed in order to account for the nonlinearity
of the first-order Raman shift frequency versus I3C mole fraction relation, which
was used to determine the isotopic composition of the diamond films grown [62].
Irrespective of the crystallographic orientation, corrected ”C mole fractions in
the films were identical, within experimental errors, to the corresponding mole
fractions in methane sampled directly above the growth surface. Owing to the fast
reaction (12), the isotopic composition of methyl and methane can be assumed to be
virtually equal. Isotopic scrambling between methane and acetylene, on the other
410 4 Chiwiccrl Vripor. Deposition of Dicltnoncl Films
hand, was suppressed by injecting the gas sufficiently close to the diamond surface.
Similar experiments were performed by MWP-CVD, also demonstrating that the
methyl radical is about an order of magnitude more efficient in diamond formation
than acetylene [63].
Prevention of the scrambling between C, and C2 species is also the essential idea
behind the flow tube studies of Martin and Hill [64], experiments with forced
convection of Schiifer et ul. [65],and experiments with supersonic jets performed
by Lee et ul. [66],all demonstrating the importance of methyl radicals as diamond
precursors. Martin and Hill used a microwave plasma in order to generate a stream
of hydrogen atoms in a flow-tube system. After injection of a small amount of
methane into the flowing gas the formation of diamond could be achieved in a
furnace at 8SO’C a few centimeters downstream, corresponding to a reaction time
of the order of 1 ms. This time is sufficiently long for reaction (12) to come to
equilibrium, but too short for major Cl-species formation. Diamond formation
was also observed after acetylene injections, however the rates were about an
order of magnitude smaller than for methane.
The investigations of Schiifer et ul. were stimulated by observations made during
HF-CVD experiments using arrays of four parallel uncoiled Ta wires, 1 1 .S mm from
each other. Under conditions of free convection of the gas phase a strong decrease of
the diamond growth rate from the edges of a wafer towards its center was noted. The
authors concluded that the reason for this observation was a virtually complete
conversion of carbon species to acetylene, which does not act as a growth species.
In order to test these assumptions, they investigated the effects of vertical gas jets
(hydrogen with 0.5% methane and acetylene) blown through the filament array with
a velocity of about 1000 cm s-’ and striking the substrate vertically. With methane,
maximum growth rates of several pm h-’ were found at the stagiiation point of the
jet while in the case of acetylene a growth rate depression was observed at this point.
A practically complete prevention of the interconversion between CH, and C2H,
was achieved in the experiments of Lee rt al., using supersonic free jets of methyl
radicals, generated by the pyrolysis of azomethane, and atomic hydrogen. Gas
phase chemistry can thus be virtually eliminated. Good quality diamond crystallites
could be grown on submicron diamond seeds, exposed to these jets at 650-8SO’C.
Diamond deposition ceased, when either the azomethane pyrolysis or the glow
discharge generating atomic hydrogen was turned of. Graphitic carbon was the
main product, when an acetylene jet was used instead of a methyl jet.
4.4 Properties and Applications of Chemical Vapor

Deposited Diamond
The world-wide interest in CVD diamond is in a considerable part due to its
unsurpassed mechanical hardness and wear resistance, promising the ultimate
wear protection for tools and machine parts to be achieved by direct coating
from the gas phase. However, the exceptional position of diamond under natural
4.4 Properties and Applications of Chemical Vapor Deposited Diamond 411
and man-made materials is not only based on its mechanical properties but also its
optical transparency over a broad wavelength range, its unsurpassed thermal con-
ductivity at ambient temperatures, several outstanding electrical properties (band
gap, breakdown field strength, charge carrier mobilities, saturated carrier velocities)
and chemical inertness. Many prospective applications make use of one or more of
these physical properties. Several handbooks have meanwhile appeared which give
very exhaustive accounts of the properties and applications of natural or man-made
diamond single crystals [67] and diamond films [68,69]. The subsequent sections will
therefore be restricted to a concise introduction of only the most important applica-
tions (which often are still in a state of infancy or even only prospective) and the
properties which are relevant for these applications.
4.4.1 Diamond Coated Cutting Tools

Before the deposition by CVD methods became possible, diamond was in use as a
cutting material in three different forms: As a single crystalline natural material since
early times; as a synthetic single crystal and as compacted and sintered polycrystal-
line diamond chips (pcD) after the high-pressure, high-temperature synthesis was
invented in the mid-1950s. CVD allows two further forms of diamond to be used
for the machining of materials: a free standing sheet of several 100 pm thickness,
grown on a temporary substrate and used similarly to the pcD chip (and in many
cases outperforming the corresponding pcD tool); and a diamond film of up to a
few 10 pm thickness deposited directly from the gas phase onto a suitably shaped
cutting tool [69]. The technical potential of diamond CVD is fully exploited only
by the last method, because it allows the simultaneous conformal coating of
many, possibly different, cutting tools with complex three-dimensional shapes
(drills, mills, cutting tool inserts with multiple cutting edges equipped with chip
breakers) in a single run. In spite of a tremendous technological potential and (at
a first glance) lack of sophistication, diamond coating of cutting tools has not yet
found the widespread use which was expected 15 years ago.
Owing to the catalytic effect of iron on the conversion of diamond to graphite at
elevated temperatures [67], diamond tools cannot be used for cutting steels. Never-
theless, there is a growing application potential for diamond tools thanks to the
growing use of materials like hypereutectic AlSi alloys, A1 alloys containing alumina
or silicon carbide, and fiber reinforced plastics, which are difficult to machine. The
introduction of diamond-coated tools, especially those based on cemented carbide,
has mainly been slowed down by the difficulty of obtaining sufficient adhesion
between the hard metal substrate and diamond film reliably and reproducibly.
Therefore, a major part of recent work on diamond tool coatings consisted of the
development of suitable substrate pretreatment and process schemes aiming at a
tight mechanical anchoring of the diamond phase, removal of Co (which has a
very detrimental effect on the film adhesion because it tends to convert diamond
into graphite) from the upper substrate regions and/or formation of an interlayer
between WC/Co and diamond. Although a consensus as to an optimum procedure
cannot be expected, due to different tool materials, deposition conditions and work
piece materials, significant progress has been made in tool lifetime and reliability.
Therefore, diamond-coated tools are now just beginning to play a role in the
machining of nonferrous and nonmetallic materials [70].
4.4.2 Thermal Conductivity of Chemical Vapor Deposited

Diamond: Thermal Management Applications
Pure diamond is the material with the highest thermal conductivity, K., at room
temperature. A compilation made by Vandersande [68] compares the best results
of thermal conductivity measurements performed by several groups on the very
rare and pure (less than 100 p.p.m. N) type IIa of natural diamond. At 80-100 K,
K. has a maximum value of about 125 W cm-' K-'; at room temperature about
25 W cm-' K-', more than six times the value of copper (4 W cm-' K-I), is
obtained. Within the classical kinetic theory of heat conductivity, treating the pho-
nons, carriers of heat in insulating solids, as a gas of particles with mean free path A
average velocity v and specific heat C per unit volume, the expression (13) can be
deduced from a few simple arguments:
K. = f(CvA). (13)
Using v w 1O6cmC' (of the order of the velocity of sound in diamond) and the
literature value of C (1.82 J K-' cmP3), a mean free path length of 0.4 pm can be
calculated, roughly two orders of magnitude larger than for NaCl or silica single
crystals. Theoretically, the large value of A can be traced back to the strong
bonds between light atoms forming the diamond lattice. This consideration also
demonstrates that a diamond crystal has to be perfect over dimensions of at least
a few micrometers in order to prevent phonon scattering by point, line, or planar
defects like impurities, dislocations, or stacking faults, reducing K. below the best
single crystal values. The thermal conductivity is therefore critically dependent on
crystalline quality and purity of a diamond crystal or film.
Early thermal-conductivity measurements on CVD diamond were performed on
relatively thin films (less than 30 pm) with average lateral grain sizes of only a few
micrometers, grown by MWP-CVD [71] or HF-CVD [72]. For both growth
methods the in-plane thermal conductivity K~~ decreased from about 10 to less
than 2 Wcm-' K-', if the methane concentration in the source gas was increased
to several percent, accompanied by a deterioration of the film quality as to be
judged from Raman spectra and (in the case of the HF-CVD samples) an increase
of the H-content from 0.1 to 1.0%.
Significantly larger thermal conductivities were achieved, when films of several
l00pm thickness were grown by MWP-CVD or HF-CVD [73,74]. Owing to
grain dimensions attaining 50-100 ym in the top regions of films of 350-600 pm
thickness, grain boundary scattering was strongly reduced and room temperature
thermal conductivities of 17 Wcm-' K-'could be achieved. The average thermal
conductivity of a thick diamond film measured in-plane or vertically to the film
plane can be considered as an integral over increments from local contributions
4.4 Properties and Applications of' Chemical Vapor Deposited Diamond 4 13
of differential slices parallel to the film plane. Graebner and coworkers analyzed
thick films with respect to the local values of in-plane and vertical thermal con-
ductivities and discovered, that both quantities increased with the distance from
the substrate surface up to values of 21 and 23.5 W cm-' K-' in the top region
of a 350 pm thick film [75,77]. The latter value is already close to the best single
crystal data.
The high thermal conductivity of diamond forms the basis for the application of
CVD diamond in heat sinks which is expected to play an increasing role in the near
future: Semiconductor-based laser diodes or IMPATT microwave diodes generate a
large amount of heat which has to be removed efficiently in order to prevent an early
failure of the device due to excessive heating. The efficiency of water-cooled heat
sinks is limited, due to the relatively small area across which the heat flux is entering
into the sink. Therefore heat spreaders are used in order to enlarge the area, over
which the heat is conducted to the heat sink. These spreaders must not present a
large thermal resistance themselves. Single-crystalline diamonds synthesized by
the high-pressure, high-temperature method are mainly used for this purpose
today [67]; however, the price of these crystals increases very rapidly with size.
Therefore CVD films can gain importance where larger heat sinks with dimensions
reaching 1 cm or more are necessary [69].
Thermal management of computers could possibly open a much larger future
market for thick diamond films of maximum thermal conductivity. Modern multi-
chip modules involve power density levels which are difficult to cope with using
conventional substrate materials. The physical properties of diamond, aside from
thermal conductivity, the highly insulating nature, and a very low coefficient of
thermal expansion, make it an optimal choice for managing the thermal problems
encountered in present days two-dimensional packaging and in prospective three-
dimensional arrangements of integrated circuits [69].
4.4.3 Electrical Properties and Electronic Applications

The literature on the electrical properties of CVD diamond and on the performance
of first laboratory samples of electronic devices made from it, has grown to such an
extent that an appropriate review would need at least an order of magnitude more
space than available here. Important publications are referred to in Chapters 8 and
29 of [69], a good introduction was recently given by Prins, who had been the first to
demonstrate transistor action in natural type IIb semiconducting diamond in 1982
[78]. Table 1, adapted from his article, compares the most relevant properties of
diamond to those of other semiconductors. JFM and KFM are the Johnson and
Keyes figures of merit, in the table normalized to the corresponding figures for Si.
They are defined as
JFM = (27t-'ESus (14)
and
KFM = K.(cv,/~~cE)''~,
Table 1. Comparison of physical properties of diamond with those of other semiconductors (after
[781)
Si GaAs 3C-Sic Diamond
Band gap, E, (eV) 1.1 1.4 2.3 5.47

Thermal conductivity, K (W cm-' K-') 1.5 0.46 4 5 25
Carrier mobility, p (cm2V-' s-l)
Holes 600 400 70 1700
Electrons 1400 8500 800 2150
Saturated electron velocity, us (10' cms-:) 1.o 1-2 2.5 2.7
Breakdown field strength, E, ( lo5 V cm- ) 3.0 4.0 10-50 50-200
Dielectric constant, E 11.8 12.5 9.1 5.7
JFM (relative to Si) 1 2.6 34 90
KFM (relative to Si) 1 0.5 6.3 35
where c in KFM is the speed of light, the other symbols are defined in the table.
JFM was designed to measure the upper limits of a material's performance in a
transistor handling simultaneously high frequency and high power. The superior
position of diamond is mainly due to its very large breakdown field strength. In
the derivation of KFM the thermal limits on the speed of a semiconductor
device are considered; it is mainly the unsurpassed thermal conductivity of
diamond which gives it the highest KFM value. Apart from these advantages of
diamond as a semiconductor material for high-power, high-frequency applications,
its very large band gap, making undoped pure diamond very highly insulating at
room temperature and up to very high temperatures, is the prerequisite for
prospective applications in active devices working at elevated temperatures up to
500°C or more.
So far, only p-type doping with boron (incorporated from a suitable volatile
source during CVD growth) has become a standard practice. In homoepitaxial
films, mobilities reaching 1600 cm2V-' s-' have been reported, indicating a good
crystalline quality similar to natural IIb-type diamond single crystals. Doping of
the n-type by phosphorus during CVD growth is still in its infancy. Correspond-
ingly, the list of 24 publications on active diamond devices in [69] mainly refers to
unipolar field-effect transistors, mostly prepared from in situ p-doped homoepitaxial
films or B ion implanted layers on single crystals. In several cases transistor function
and operation at elevated temperatures up to 500°C were demonstrated. In order to
develop an economically viable diamond device technology, however, diamond
single crystals are hardly suitable substrates. Owing to grain-boundary effects in
randomly oriented polycrystalline films grown on cheap silicon substrates, device
performance in this material is inferior by orders of magnitude. Therefore, the
growth of heteroepitaxial films on Si or Sic substrates might gain importance for
future electronic applications. Even if the films are still polycrystalline, the detri-
mental effect of grain boundaries on physical properties is reduced owing to
much smaller intergrain angles. Electrical properties (in-plane hole mobility,
through-plane leakage currents) as well as possibly the thermal conductivity are
therefore superior to those of randomly oriented material [79]. For a metal-oxide
4.4 Properties and Applications of Chemical Vapor Deposited Diamond 4 15
E"AC
----xt- 5.5 eV
clean H-terminated
(1 11) diamond surface
Figure 8. Alignments of valence and conduction bands of clean and H-terminated (1 11) surfaces of
p-doped diamond, EF = Fermi level. The figures have been aligned at the vacuum level (adapted
from [SO]).
silicon field-effect transistor (MOSFET) made from heteroepitaxial material,

transistor operation significantly improved over that of a random material has
already been demonstrated up to 400°C [69].
Apart from transistors, several other solid state devices have been discussed [78],
like junctions, photon and electron beam switches and various kinds of sensors. One
property of diamond which has stimulated considerable interest in the recent years is
the negative electron affinity (NEA) of suitably prepared surfaces [78, SO]. The elec-
tron affinity, x,of a material is defined as the difference between the energy of a free
electron in vacuum and the bottom of the conduction band Eva,- E,. In Fig. 8 the
electronic bands of p-doped clean and H-terminated (1 1 1) diamond surfaces near
the surface are depicted, based on the results of UV-photoemission measurements.
For the H-terminated surface, the electron affinity becomes negative; once an elec-
tron is injected into the conduction band from a suitable contact or by UV excita-
tion, it will easily leave the crystal and be emitted into vacuum. This effect, which is
also observed on monohydride terminated (1 00) surfaces, is not unique to diamond
but was also observed in a few other semiconductors with high band gaps [80]. Apart
from a scientific interest, the NEA of diamond makes it an attractive candidate for
the replacement of thermionic emitters as electron beam sources and as a miniature
electron emitter for field emission displays.
4.4.4 Electrochemical Use of Chemical Vapor

Deposited Diamond
The chemical inertness of diamond, even in hot acidic or caustic oxidizing media
[8 I], and the possibility of making conductive CVD diamond films with resistivities
down to less than 0.01Rcm [82] on a wide variety of substrates and over large
areas, has in the recent years stimulated an increased interest in applications for
analytical and preparative electrochemistry. In a comparative study, Swain
4 16 4 Chemical Vapor Deposition of Diamond Films
investigated the response of electrodes consisting of heavily boron-doped diamond

films, highly ordered pyrolytic graphite (HOPG) and glassy carbon (gC), towards
repeated potential cycling (PC) for 2 h in a solution of 1 M HN03 0.1 M NaF+
at 50°C. The electrode properties were investigated before and after PC by
Raman spectroscopy, optical and scanning electron microscopy, and electro-
chemical techniques. The potential ranges extended to the onset potentials for the
evolution of hydrogen and oxygen, for each electrode. For diamond, as a result
of the potential cycling the anodic onset potential shifted from 1.2 to 1.7V (vs.
Ag/AgCl), at the cathodic potential of -0.6V, a current reduction was observed.
Simultaneously, the total charge in a cyclovoltammogram taken after PC was
reduced by a factor of two. These changes were discussed in terms of a surface
cleaning by removal of nondiamond phases. In contrast to diamond, a much
larger degree of microstructural damage and surface oxydation was observed for
HOPG and gC electrodes [83].
The width of the ‘potential window’ with very low background currents in
aqueous electrolytes depends critically on the diamond phase purity. Comparing
HF-CVD samples grown from atmospheres with 1 and 4% methane, and doped
with about lo2’ boron atoms per cm3, a potential range extending from -1.25V
to +2.3V (vs. SHE) without significant water decomposition could be detected
for the high quality sample grown from the carbon-poor atmosphere. Samples
grown from 4% methane, on the other hand, showed a range of only -0.6 to
1.7 V, owing to their content of nondiamond phases. The inert nature of the H-
terminated diamond surface can be made responsible for the high overpotentials
observed for hydrogen and oxygen [84] as well as halogen evolution [85].
The evolution of halogen is promoted by the absorption of halogen atoms on the
electrode surface (16); however, owing to a small affinity of the hydrogenated
diamond surface towards absorption of atomic species, the subsequent reaction
(17) is kinetically strongly hindered:
+
Xads X-
X- -Xads
+X2
+ e- ,
+ e-.
(16)
(17)
It is inferred that generally electrochemical reactions involving surface-absorbed
species will proceed comparably slowly on pure diamond surfaces. On the other
hand, outer sphere electron transfer reactions not requiring a strong surface
interaction of the species involved, will in principle not be affected by the inert
nature of the diamond surface, as to be demonstrated, for example, by the reversible
behavior of the Ru(III)/Ru(II) redox couple [85].
The actual technical potential of diamond electrodes for analytical and prepara-
tive electrochemical applications can so far not be fully assessed. From various
prospective applications mentioned in the literature, making use of the exceptional
stability, wide potential window and low background currents of diamond
electrodes, only one example shall be mentioned here, the reduction of nitrate to
ammonium according to the net Eq. (18)
NO, + 7H20 + 8e- +N H 4 0 H + 90H-, Eo = -0.15 V vs. SHE. (IS)
References 4 17
Thanks to the special properties of diamond electrode surfaces, this reaction can be
run with high efficiency and stability [86]. A potential application is the removal of
nitrates from the effluents of combustion processes or from drinking water.
4.5 Summary
Only a small selection of possible applications of CVD diamond could be touched
on here. Many of the expectations which were placed on the applications of CVD
diamond in the early years of the ‘new diamond era’ have so far not come true,
due to reasons like still too high deposition costs or insufficient film qualities for
certain applications, major advances in competitive materials and technologies, or
just the lack of a market. Nevertheless, thanks to the tremendous progress which
has been made in the recent years and will undoubtly be made in the future, CVD
diamond can be expected to play an ever increasing role as the ultimate material for
a variety of technical purposes.
References
1. J. C. Angus, in The Physics of Diamond, A. Paoletti and A. Tucciarone (Eds), 10s Press,
Amsterdam, Oxford, Tokyo, Washington DC, 1997, p. 1.
2. R. C . DeVries, A. Badzian, and R. Roy, M R S Bulletin, 1996, 65-75.
3. P. K. Bachmann and H. Lydtin, in Diumondand Diamond-like Films and Coatings, R. E. Clausing,
L. L. Horton, J. C. Angus, and P. Koidl (Eds), Plenum, New York, 1991, p. 829.
4. S. Matsumoto, Y. Sato, M. Kamo, and N. Setaka, Jpn. J. Appl. Phys. 1982, 21, L183-Ll85.
5. S. Matsumoto, Y. Sato, M. Tsutsumi, and N. Setaka, J. Muter. Sci. 1982, 17, 3106-31 12.
6. S. Okoli, R. Haubner, and B. Lux, J. PhJjsique ZV 1991, C2, 923-930.
7. M. Sommer and F. W. Smith, J . Muter. Res. 1990, 5, 2433-2440.
8. B. Singh, Y. Arie, A. W. Levine, and 0. R. Mesker, Appl. Phys. Lett. 1988, 52, 4511152.
9. L. Schafer, U. Bringmann, C.-P. Klages, U. Meier, and K. Kohse-Hoinghaus, in Diamond and
Diamond-Like Films and Coatings, R. E. Clausing et al. (Eds), Plenum Press, New York, 1991,
p. 643..
10. F. Jansen, M. A. Machonkin, and D. E. Kuhman, J. Vac. Sci. Technol. 1990, AS, 3785-3790.
11. E. Gheeraert, A. Deneuville, M. Brunel, and J. C. Oberlin, Diamond Rel. Muter. 1992,1,504-507.
12. M. Griesser, G. Stingeder, M. Grasserbauer, H. Baumann, F. Link, P. Wurzinger, H. Lux,
R. Haubner, and B. Lux, Diamond Rel. Muter. 1994, 3, 638-644.
13. Y. Hirose and Y. Terasawa, Jpn. J. Appl. Phys. 1986, 25, L519-L521.
14. M. Murakawa, S. Takeuchi, H. Miyazawa, andY. Hirose, Surf: Coat. Technol. 1988,36,303-310.
15. T. Kawato and K. Kondo, Jpn. J. Appl. Phys. 1987,26, 1429-1432.
16. Y. Saito, K. Sato, H. Tanaka, K. Fujita, and S. Matsuda, J. Muter. Sci. 1988, 23, 842-846.
17. C . Chen, Y. C. Huang, S. Hosomi, and I. Yoshida, Muter. Res. Bull. 1989, 24, 87-94.
18. J. A. Mucha, D. L. Flamm, and D. E. Ibbotson, J. Appl. Phys. 1989, 65, 3448-3452.
19. S. J. Harris and A. M. Weiner, Appl. Phys. Lett. 1989, 55, 2179-2181.
20. D. N. Belton and S. J. Schmieg, J . Appl. Phys. 1991,69, 3032-3036.
21. P. K. Bachmann, D. Leers, and H. Lydtin, Diamond Rel. Muter. 1991, 1, 1-12.
22. G. Z. Cao, J. J. Schermer, W. J. P. van Enckevort, W. A. L. M. Elst, and L. J. Giling, J. Appl.
Phys. 1996, 79, 1357-1364.
23. M. Kamo, Y. Sato, S. Matsumoto, and N. Setaka, J. Cryst. Growth, 1983, 62, 642-644.
24. C.-P. Klages, Appl. Phys. A 1993, 56, 513-526.
25. P. K. Bachmann, in Handbook of Industrial Diamond and Diamond Films, M. A. Prelas,
G. Popovici, and L. K. Bigelow (Eds), Marcel Dekker, New York, Basel, Hong Kong, 1998,
Chapter 21.
26. H. Kawarada, K. S. Mar, and A. Hiraki, Jpn. J . Appl. Phys. 1987, 26, L1032-LIO34.
27. J. Suzuki, Jpn. J . Appl. Phys. 1989, 28, L281LL283.
28. A. A. Chernov, Modern Crystallography ZIZ, Springer Verlag, Berlin, Heidelberg, New York,
1984, p. 283.
29. A. Van der Drift, Philips Res. Repts. 1967, 22, 267-288.
30. P. Hartman, in Crystal Growth; An Introduction, P. Hartmann (Ed.), North-Holland, Amsterdam,
1973, p. 367.
31. L. J. Giling and W. J. P. Van Enckevort, Surface Science, 1985, 161, 567-583.
32. C. Wild, P. Koidl, W. Miiller-Sebert, H. Walcher, R. Kohl, N. Herres, R. Locher, R. Samlenski,
and R. Brenn, Diamond Rel. Mater. 1993, 2, 158-168.
33. R. E. Clausing, L. Heatherley, L. L. Horton, E. D. Specht, G. M. Begun, and Z. L. Wang,
34. Ch. Wild, N. Herres, and P. Koidl, J. Appl. Phys. 1990, 68, 973-978.
35. Ch. Wild, R. Kohl, N. Herres, W. Muller-Sebert, and P. Koidl, Diamond Rel. Muter. 1994, 3,
373-381.
36. H. Maeda, K. Ohtsubo, M. Irie, N. Ohya, K. Kusakabe, and S. Morooka, J . Muter. Res. 1995,
10, 3115-3123.
37. R. Locher, C. Wild, N. Herres, D. Behr, and P. Koidl, Appl. Phys. Lett. 1994,65, 34-36.
38. X. Jiang, W. J. Zhang, M. Paul, and C.-P. Klages, Appl. Phys. Lett. 1996, 68, 1927-1929.
39. B. R. Stoner and J.T. Glass, Appl. Phys. Lett, 1992, 60, 698-700.
40. X. Jiang and C.-P. Klages, lecture at Diamond 92, Heidelberg 1992, published in: Diamond Rel.
Muter. 1993,2, 1112-1 113.
41. S. Yugo, T. Kanai, T. Kimura, and T. Muto, Appl. Phys. Lett. 1991,58, 10361038.
42. B. R. Stoner, G.-H. M. Ma, and S. D. Wolter, J.T. Glass, Phvs. Rev. B 1992.45. 11067-11084.
43. X. Jiang, C.-P. Klages, M. Rosler, R. Zachai, M. Hartweg, and H.-J. Fusser, Appl. Phys. 1993,
A51,483489.
44. K. Schiffmann and X. Jiang, Appl. Phys. 1994, A59, 17.
45. J. Gerber, S. Sattel, K. Jung, H. Ehrhardt, and J. Robertson, Diamond Rel. Muter. 1995, 4,
559-562.
46. S. Sattel, J. Gerber, and H. Ehrhardt, Phys. Stat. Sol. ( A ) 1996, 154, 141-153.
47. Y. Lifshitz, S. R. Kasi, and J. W. Rabalais, Phys. Rev. Lett. 1989, 62, 1290-1293.
48. J. L. Robertson, S. C. Moss, Y. Lifshitz, S. R. Kasi, J. W. Rabalais, G. D. Lempert, and
E. Rapoport, Science, 1989, 243, 1047-1050.
49. X. Jiang, K. Schiffmann, and C.-P. Klages, Phys. Rev. B 1994,50,8402-8410.
50. T. Suzuki and A. Argoitia, Phys. Stat. Sol. ( A ) 1996, 154, 239-254.
51. H. Kawarada, T. Suesada, and Nagasawa, Appl. Phys. Lett. 1995, 66, 583-585.
52. J. C. Angus, in The Physics of Diamond, A. Paoletti and A. Tucciarone (Eds), 10s Press,
53. D. G. Goodwin and J. E. Butler, in Handbook of Industrial Diamond and Diamond Films, M. A.
Prelas, G. Popovici, and L. K. Bigelow (Eds), Marcel Dekker, New York, Basel, Hong Kong,
1998, Chapter 11.
54. F. G. Celii, P. E. Pehrsson, H.-T. Wang, and J. E. Butler, Appl. Phys. Lett. 1988,52,2043-2045.
5 5 . J. E. Butler and F. G. Celii, in Proc. 1st Znt’l Symp. on Diamond and Diamondlike Films, J. P.
Dismukes et al. (Eds), The Electrochemical Society, Pennington, NJ, 1989, p. 317.
56. S. J. Harris and A. M. Weiner, J . Appl. Phys. 1990,67,6520-6526.
57. W. L. HSU,Appl. P h y ~Lett.
. 1991,59, 142771429,
58. W. L. Hsu, in Proc. 2ndZnt’l Symp. on DiamondMaterials, A. J. Purdes et al. (Eds), The Electro-
chemical Society, Pennington, NJ, USA, 1991, p. 217.
59. C. J. Chu, M. P. D’Evelyn, R. H. Hauge, and J. L. Margrave, Muter. Res. SOC.Symp. Proc.
1990, 162, 85-90.
60. C. J. Chu, M. P. D’Evelyn, R. H. Hauge, and J. L. Margrave, J . Muter. Res. 1990,5,2405-2413.
References 419
61. C. J. Chu, M . P. D’Evelyn, R. H. Hauge, and J. L. Margrave, J . Appl. Phys. 1991,70, 1695-1705.
62. M. P. D’Evelyn, C. J. Chu, R. H. Hauge, and J. L. Margrave, J . Appl. Phys. 1992,71, 1528-1530.
63. C. E. Johnson, W. A. Weimer, and F. M. Cerio, J. Muter. Res. 1992, 7, 1427--1431.
64. L. R. Martin and M. W. Hill, J. Muter. Sci. Lett. 1990, 9, 621-623.
65. L. Schafer, M. Sattler, and C.-P. Klages, in Applications of Diamond and Related Materials,
Y. Tzeng e f al. (Eds), Elsevier, Amsterdam, 1991, p. 453.
66. S. S. Lee, D. W. Minsek, D. J. Vestyck, and P. Chen, Science 1994, 263, 1596-1598.
67. J. E. Field (Ed.), The Properties qf Diamond, Academic Press, London, 1997.
68. G. Davis (Ed.), Properties and Growth qf Diamond, EMIS Datareview Series, INSPEC,
London, UK, 1994.
69. M. A. Prelas, G. Popovici, and L. K. Bigelow (Eds.) Handbook of Industrial Diamond and
Diamond Films,Marcel Dekker, New York, Basel, Hong Kong, 1998.
70. C.-P. Klages, M. Fryda, T. Matthee, L. Schafer, and H. Dimigen, in Proceedings ofthe 14th
International Plansee Seminar, G. Kneringer, P. Rodhammer und P. Wilhartitz (Eds), Plansee
AG, Reutte, Austria, 1997, Vol. 3, 1-15.
71. A. Ono, T. Baba. H. Funamoto, and A. Nishikawa, Jpn. J . Appl. Phys. 1986, 25, L808-LSlO.
72. D. T. Morelli, C. P. Beetz, and T. A. Perry, J . Appl. Phys. 1988, 64,3063-3066.
73. D. T. Morelli, T. M. Hartnett, and C. J. Robinson, Appl. Phys. Letf. 1991, 59, 2112-2114.
74. J. E. Graebner and J. A. Herb, Diamond Films Technol. 1992, 1, 155-163.
75. J. E. Graebner, S. Jin, G. W. Kammlott, J. A. Herb, and C. F. Gardinier, Appl. Phys. Lett. 1992,
30, 1576-1 578.
76. J. E. Graebner, S. Jin. G. W. Kammlott, B. Bacon, L. Seibles, and W. Banholzer, J. Appl. Phys.
1992,71, 5353-5356.
77. J. E. Graebner, S. Jin, G. W. Kammlott, J. A. Herb, and C. F. Gardinier, Appl. Phys. Lett. 1992,
359,401403.
78. J. F. Prins, in The Physics of Diamond, A. Paoletti, and A. Tucciarone (Eds), 10s Press,
Amsterdam, Oxford, Tokyo, Washington DC, 1997, p. 41 1.
79. C.-P. Klages in The Physics of Diamond, A. Paoletti, and A. Tucciarone (Eds), 10s Press,
80. R. J. Nemanich, P. K. Baumann, M. C. Benjamin, S. P. Bozeman, and B. L. Ward, in The
Physics of Diamond, A. Paoletti and A. Tucciarone (Eds), 10s Press, Amsterdam, Oxford,
Tokyo, Washington DC, 1997, p. 537.
81. P. M. Natishan and A. Morrish, Materials Letters, 1989, 8. 269-272.
82. M . Werner, R. Job, A. Zaitzev, W. R. Fahrner, W. Seifert, C. Johnston, and P. R. Chalker,
Phys. Stat. Sol. (‘A), 1996, 154, 385-393.
83. G. M . Swain, J. Electrochem. Soc. 1994, 141, 3382-3393.
84. H. B. Martin, A. Argoitma, U. Landau, A. B. Anderson, and J. C. Angus, J. Electrochem. Soc.
1996, 143, L133-LI36.
85. N. Vinokur, B. Miller, Y. Avyigal, and R. Kalish, J . Electrochem. Soc. 1996, 143, L238-L240.
86. R. Tenne, K. Patel. K. Hashimoto. and A. Fujishima, J . Elecfroanal. Chem. 1993,347,409415.
5 Vapor Phase Deposition of Cubic Boron

Nitride Films
K. Bewilogua and F. Richter
5.1 Introduction
Connected with the advancement of synthetic diamond as a technical material, cubic
boron nitride (cBN) has attracted increasing interest during the past decades.
Boron nitride exists in high density diamond-like and in low density graphite-like
modifications. High density cubic boron nitride has a zinc-blende structure with
eight atoms in the elementary cell. Each B atom and each N atom is surrounded
by four N or B atoms, respectively. Like in the diamond structure the neighboring
atoms are tetrahedrally arranged with fourfold coordination. Diamond and cBN
have similar lattice constants. However, diamond has a pure covalent bonding
and cBN a covalent-ionic mixed bonding. Hexagonal boron nitride (hBN) and
graphite have similar structures, both with a threefold coordination of nearest
neighbors. Beside hBN and cBN two other BN structures are known, the rhombo-
hedral (rBN) and the metastable high-density wurtzite-type BN (wBN). Apparently
the wBN phase only can be synthesized by shock compression. On the one hand
hBN and rBN and on the other hand cBN and wBN have similar structures and
the same short range order but different stacking sequences of atomic planes (see
[l]). In Table 1 the structural data as well as some properties of cBN, diamond,
hBN, and graphite are summarized.
Cubic BN is the second hardest material after diamond. It does not react with iron
in tribological contact [2] and can therefore (in contrast to diamond) be used to coat
tools for machining of ferrous materials.
Because of its wide band gap (Eg= 6.4eV [ 3 ] ) , cBN transmits electromagnetic
radiation in a broad range including the near ultraviolet as well as the visible and
infrared region. Moreover, cBN can be doped both n- and p-type [4] and has
good thermal conductivity. Therefore, additional potential applications are as pro-
tective coatings for optical devices or in electronics, either as a high temperature
semiconductor or as heat dissipating material.
Analogous to graphite in the carbon case, boron nitride exhibits a low-density
relatively soft hexagonal phase hBN. Formation of boron nitride by chemical
reaction at atmospheric pressure yields the hexagonal phase. However, by using
the high pressure, high temperature (HPHT) synthesis established by Wentorf [5]
it is possible to transform hBN into cBN. In this method, ( p , T ) values in the
range 1200-2000°C and 4.5-7.5 GPa, respectively [6], were applied corresponding
to conditions where cBN is thermodynamically stable [7] and formed comparatively
readily. Today, small crystallites of cBN having a typical size in the several 100 mm
range [S] are produced in considerable quantities with this method and are processed
5.2 Empirical Results 42 1
Table 1. Structure and properties of cBN, diamond, hBN, and graphite.
cBN Diamond hBN Graphite
Crystal structure zinc blende diamond hexagonal hexagonal

Lattice constants (nm) 0.362 0.356 a 0.250 a 0.246
c 0.666 c 0.671
Bond length (nm) 0.157 0.154 0.144 0.142
Interplanar spacing (nm) ~
0.333 0.335
Density (gcm-') 3.48 3.52 2.28 2.27
Hardness (GPa) 5&70 9&100 approx. 10 <10
Young's modulus (GPa) 700-900 1050 a 87 a 19
c 34 c 5
Band structure Indirect Indirect Direct Metallic
semiconductor semiconductor semiconductor
Optical band gap (eV) 6.1 5.5 5.2 -
Refractive index 2.12 2.42 1.72 -
(A = 589.3 nm)
Resistivity (0 cm) 10'O 1013 a 10'' a lo-'
c lo1* c 1
Dopability p and n P (n ?)
Thermal conductivity <750 2000 1 6.4-10.8
(W m-I K - I )
Oxidation at air ("C) >loo0 >500 >I000 -
Reactivity with iron at No Yes No Yes
elevated temperatures
mainly to sintered polycrystalline boron nitride (pcBN) pieces as cutting edges for
tools.
However, by growing thin BN films from the vapor phase it is possible to get cubic
BN as a thin film at moderately increased temperatures as well as subatmospheric
pressure. Starting in the SOs, mainly by Japanese researchers [9], great efforts have
been made in this field up to now.
The aim of this chapter is to present an overview of some important recent
developments in cBN thin film deposition. The physical mechanisms which are
causal for cBN formation in ion-assisted thin film deposition will be discussed in
some detail.
5.2 Empirical Results

5.2.1 Deposition Methods
A great variety of deposition methods has been successfully used up to now to
deposit cBN-containing thin films, among them both physical (PVD) and chemical
vapor deposition (CVD) techniques. All these methods have in common that the
growing film is exposed to an intense ion bombardment which plays a crucial role
in cBN thin film formation. Among the PVD methods initially such techniques
422 5 V q m Phuse Deposition of' Cubic Boron Nitride Films
did predominate which use electron beam evaporation of boron. The necessary ion
impact to the growing film was performed either by special activation and ionization
of the nitrogen containing gas atmosphere together with a suitable substrate bias to
accelerate the ions [10-12] or by means of ion beam sources [13,14]. The use of ion
beam sources to realize ion beam impact to the growing film has also been combined
with boron vaporization by ion beam sputtering [15,16] or laser ablation [17] of
either a boron or BN target. Mass-selected ion beam deposition (MSIB) [18] and
cathodic arc deposition (CAD) [19] have also been successfully used to deposit
cBN thin films. With the last mentioned two methods the film is formed essentially
by deposition of ions.
Because of its potential for application, R F sputtering [20,21] and R F magnetron
sputtering [22,23] has attracted increasing attention. Deposition by d.c. magnetron
sputtering (which is even more interesting from an application point of view) has
also been successfully applied to cBN thin films, either using a B4C target [24]
(section 5.4.1) or a boron target which is operated at elevated temperature where
boron gets conductive [25] (section 5.4.2).
Different variants of CVD techniques were applied, for example R F glow dis-
charge using the gases diborane B2H6, N2, Ar, He, and H2 [26] as well as electron
cyclotron resonance (ECR) plasma CVD processes with B2H6and N2 [27] or with
the vapor of N-trimethylborazine [28]. The authors reported that cBN could only be
generated if the negative substrate bias was sufficiently high. BN films deposited by
such CVD methods seem to have in general somewhat lower cBN contents than
those prepared by PVD techniques. Probably this is due to the incorporation of
hydrogen into the films [29].
Zhang et ul. [30,31] deposited films on nickel in a hot filament assisted R F glow
discharge with the reactant gases being B2H6, NH3, and H2. Obviously the sub-
strates were not biased. Scanning electron microscopic images revealed crystallites
with dimensions of about 2 pm. As an essential reason for the rather large crystal-
lites, a catalytic effect of Ni was assumed. Unfortunately, for the evidence of the
cubic phase only X-ray diffraction patterns, but no infrared spectra, were published.
Therefore considerable doubt that these crystallites consist of cBN remains.
Konyashin et al. [32] deposited BN films using plasma assisted CVD with gas mix-
tures of argon, hydrogen, and borane-ammonia (BH3-NH3) without substrate bias
at temperatures up 1000°C. Regrettably no data on hydrogen content and B and N
concentrations in the films were given. From IR spectra and X-ray diffraction
patterns of films deposited with high hydrogen concentrations in the process gas,
the existence of single phase cBN was concluded. The authors assumed a deposition
mechanism which includes a selective etching of sp2 bonded BN and the formation
of metastable BNH, species.
Reinke et al. [33] analyzed numerous deposition methods with respect to film
structure (cBN or hBN) on the one hand and ion bombardment on the other
hand. To describe the ion bombardment quantitatively they used the ion energy
and the ratio of fluxes of impinging ions (i) and deposited boron atoms (a).
Figure 1 shows the results for the ratio i/u against the ion energy. The total
number of deposited atoms can be estimated taking into account the deposition
rate and the mass densities of the BN phases. As can be seen from that presentation,
0
0
0
0 oo 0
0 0 .
0 0 0
10
I
100
0
.
1000
Ion energy [eV]
Figure 1. Data compilation according to Reinke et al. [33] showing the phase of BN thin films in
dependence on ion energy, Ei, and the ratio of bombarding ions and deposited atoms, i/a, for
numerous deposition methods. Filled and open symbols represent cBN and hBN films, respectively.
cBN has been grown in a broad range of ion energies Ei,beginning with quite low
energies connected with a large i/u (these are mostly CVD methods) up to higher E;
together with small i/u. The limiting case is represented by mass selected ion deposi-
tion where the film grows solely by deposition of Bf and Nf ions. In this case i/u is
one and cBN was formed at ion energies down to 125 eV [34].
In their successful attempts to grow cBN thin films, most authors have applied
moderately increased substrate temperatures, T,, mostly in the range between 200
and 500°C. However, the growth of cBN thin film is also possible at temperatures
as high as 1000°C [35]. The low temperature limit for cBN film deposition is at
about 150'C, as has been found by Mirkarimi [36] for ion-assisted pulsed laser
deposition and more recently by Hofsass et ul. [34] for MSIB deposition.
5.2.2 Morphology and Structure of cBN Films

Up to now, most work on cBN deposition has been done using single crystal silicon
substrates. Those films grow in a typical phase sequence which has been shown for
the first time by Kester et al. [37] in 1993 and since then confirmed by many groups:
At first an amorphous layer at the interface to the substrate is formed which is
usually referred to as aBN. This layer probably contains also some silicon from
the substrate and is due to ion-induced intermixing. After that a turbostratic
boron nitride (tBN) layer is formed. Turbostratic boron nitride is a BN form
consisting of nearly parallel hBN-like layers which, however, do not have defined
three-dimensional orientation. The distance of these layers is a few percent enlarged
in comparison to the hBN crystal (see e.g. [ 11). Under deposition conditions which
424 5 Vapor Phase Deposition of Cubic Boron Nitride Films
Figure 2. Cross-section TEM micrograph of a cBN film on single crystal silicon with the typical
interlayers of amorphous material (probably a mixture of B, N, and Si), turbostratic BN.
finally yield cBN growth (i.e. under proper substrate temperature and ion bombard-
ment as discussed above) the tBN exhibits a preferred orientation such that the c-
axis of the tBN clusters are predominantly perpendicular to the substrate surface
normal. After a certain amount of tBN deposition the film structure changes into
cBN. In Fig. 2 a cross-section transmission electron micrograph of this typical
phase sequence is shown. The thickness of the amorphous and tBN interlayers
depends on the deposition conditions. Especially the tBN thickness may be very
large, however, under optimum conditions the transition to cBN growth takes
place very soon and the thickness of the hexagonal BN layer is only about 1 nm
or even not detectable in cross-section TEM.
The cBN films are polycrystalline with columnar grains about 10-20nm in
diameter extended perpendicular to the surface.
The cBN crystallites have a preferential orientation. Some authors report on a
(1 10) texture [38,39], others [40] found a so-called double-fiber texture. Both
textures are characterized by an alignment of { 11l} planes perpendicular to the
surface. The difference is that in the case of the (110) texture the (111) planes
have a preferred azimuthal orientation (i.e. the (1 10) direction is parallel to the
substrate normal), whereas in the case of the double-fiber texture the orientation
about the in-plane [l 1 11 axis is random. However, both textures do support the
suggestion that the structural transition from hBN to cBN growth is characterized
by three (1 11) spaces of cBN (dl11 = 0.209 nm) matching two (0002) spaces of tBN
-Yr c-BN (1 1I), 2.09 A
Figure 3. High-resolution cross-section electron micrograph of the tBN/cBN 2 : 3-transition in a

magnetron sputtered BN film.
(do,,, M 0.34 nm), thus forming a semicoherent interface. In the meantime this
model has been confirmed using cross-section TEM by us and others (see Fig. 3).
A necessary condition for cBN phase formation in thin films is that boron and
nitrogen are incorporated in a nearly 1: 1-ratio. Based on highly accurate com-
position measurement using neutron depth profiling, Hackenberger et al. [41]
investigated the relationship between film stoichiometry and the phases present in
the film. From these measurements, together with an analysis of data from the
literature, they found evidence that film stoichiometry is one of the factors that
stabilize the cubic phase.
Boron nitride films containing both the hBN and cBN phase cause strong infrared
absorption bands at about 1050cm-' owing to the presence of cBN (transversal
optical mode) [42] as well as at about 800 and 1400cm-' corresponding to B-N
bending (out-of-plane vibration) and stretching (in-plane vibrations) modes, respec-
tively [43].
Actually, the IR absorption features depend on the chemical short-range order of
the lattice, and their appearance is only a necessary (not a sufficient) condition for
the existence of the respective phases. There exist some more BN phases with either
sp2 or sp3 short-range order which show essentially the same IR absorption as hBN
and cBN, respectively. However, if used in combination with more direct methods
such as electron diffraction, IR absorption is a fast and reliable method for determi-
nation of the phase content of BN thin films and has found widespread application.
In Fig. 4 the IR absorption spectrum of a BN film is given as an example which
shows a distinct absorption near 1050cm-' and only weak absorption at about
426 5 Vupor Phusr Depositiori of Cubic Boron Nitride Films
1 .o
0.9
2
T-
Y
t 8~1 8 - , -7- , '

7- 1 -
7---
600 800 1000 1200 1400 1600 1800 2000
Wavenumber [cm']
Figure 4. Infrared absorption spectrum of a 130 nm thick BN film on silicon consisting mainly of
nanocrystallline cBN
800 and 1400 cm-I, respectively. This film consists practically of phase pure cBN.
The slight sp2-like absorption is mainly due to the non-cubic (amorphous and turbo-
stratic) interlayer mentioned above. Probably the not perfectly sp3 bound grain
boundary regions of the nanocrystalline cBN film contribute to the sp2 signal, too.
There were some attempts to reveal quantitative data on phase contents in mixed
BN films from IR spectra. Jager et al. [44] determined the cBN content from the
absorption coefficients a h (at a wave number near v = 1400cm-' for hBN) and
a, (near v = 1050cm-' for cBN). The ratio of the volume fractions,f,, and,fvhof
both phases was calculated as fvc/fih = 0.6ac/ah. However, in practice it is much
more simple to consider the peak heights 1;(i = h, c) in the spectra than to determine
the absorption coefficients ai. Schiitze et al. [24] derived the rough approximation
Z; z Q; (i = h,c) which leads to,jvc/fvh= 0.61,/Zh.
Considering the mass fractions,f,, it follows that,fi,,,/f;,,h = ( p , B N / p h B N ) *jvc/fvh
and,fm,/fn,h z Z,/Z, and the cBN content simply can be calculated as Zc/(Zc + Zh).
Using this relation one gets for the film with the spectrum shown in Fig. 4 a cBN
content of SO-90% which is also in good agreement with the impression gained
from cross-section TEM images.
5.3Morlel.~of c B N Forrnution 427
5.2.3 Film Adhesion

An important and typical experimental result is the strong biaxial compressive stress
which is always found in cBN thin film. The stress usually has values of 4-10 GPa
which is high enough to cause serious adhesion problems as soon as the film thick-
ness exceed a few 100 nm. In fact, bad film adhesion is the most important obstacle
in making cBN ready for application.
Beside high stress. the ambient conditions also influence the adhesion of cBN
films. Several research groups observed that cBN films stored in vacuum, dry air,
or nitrogen did not delaminate while those kept under normal humid environmental
conditions delaminated much faster from the substrates. In detail this effect was
discussed by Cardinale rt al. [45]. They suggested a generation of oxide compounds
by an aqueous oxidation of BN. This oxidation is accompanied by a volume expan-
sion which destroys the film. Probably water vapor comes into the films via defects
which can be assumed to be located in the regions between the columnar grains.
Therefore, the way to overcome the adhesion problems should be to prepare
more homogeneous films with higher density.
Most cBN deposition experiments reported so far were carried with silicon sub-
strates. Only a few papers concerned ‘non-silicon’ substrates. Mirkarimi rt al. [46]
used different metallic substrates and found a decrease of the cBN content with
decreasing metal hardness. Schutze rt al. [47] found that it is not possible to deposit
cBN films on steel substrates under conditions which are suitable for silicon sub-
strates. On steel cBN could only be prepared after deposition of a high-density
hBN buffer layer. Concerning the effect of the substrate material on the cBN film
growth there are still a lot of open questions.
5.3 Models of cBN Formation

McKenzie et al. [48] considered the above mentioned strong biaxial stress in cBN
and mentioned that the respective stress tensor may be broken down into two
components: a tensor describing a hydrostatic pressure and a tensor describing a
pure shear stress. The hydrostatic pressure has a magnitude of two thirds of the
biaxial compressive stress. Because of these stress values in thin BN films, the hydro-
static pressure may have a magnitude of several GPa. This is approximately the
same magnitude as the pressure values applied to transform hBN into cBN using
the HPHT method. The authors argue that provided the pressure is high enough,
the formation of cBN is in principle possible, even at low temperature, due to the
fact that the layer is formed from individual atoms. As a possible mechanism,
they discuss the localized melting of small regions of the layer due to energy transfer
from the ions (‘thermal spikes’) followed by the solidification of the material as cBN
under the influence of the high stress.
The authors back up their model by experimental results. For instance they have
varied the conditions for boron nitride thin film deposition which yielded different
428 5 Vupor Plitrsr Deposition o f ’ Cubic BOIYJ~I
Nitride Filtm
100
a,
20 c m I
8
0 2 4 6 8 10 12
Compressive Stress [GPa]
Figure 5. Cubic phase content in RN thin film deposited by IBAD against the compressive film
stress (adapted from [48]).
stress level in the films. An analysis of the respective cBN content of the films (Fig. 5)
showed that a certain threshold of compressive stress exists above which cBN for-
mation in the film takes place. They also reported on a typical dependence of the
film stress on the deposition time (zero stress + increase up to maximum + decrease
to saturation value) which at least qualitatively agrees with the above mentioned
sequence aBN + hBN cBN. An important conclusion from this model is that
---f
the compressive stress not only exists in conjunction with the cBN phase in the
layer, but is the cause of the cBN formation. An analogous model has already
been applied by the same authors to explain the formation of tetrahedral amor-
phous carbon (taC) [49].
Several authors conducted experiments under well-controlled conditions and
varied both the energy and mass of the ions in order to discriminate between the
specific role of ion energy and momentum. Kester and Messier [ 141 found a distinct
increase of the cBN content of the films at a certain threshold value of the trans-
ferred momentum per arriving atom. For the momentum transferred from a
single impinging ion they used an expression given in [50] for a single, binary, elastic,
head-on collision betwcen the ion and an atom of the film. These results underline
the relevance of momentum transfer processes in cBN formation. However, because
of the quite restricting suppositions as well as an error existing in the formula used,
their quantitative accuracy is limited.
Mirkarimi rt al. [17] reported on similar BN deposition experiments performed by
pulsed laser evaporation of an hBN target together with an ion beam bombardment
of the growing film performed by a special ion source. They varied the ion energy
(Eion= 500-1200 eV) and used ions of different mass (mion = ???Ar, m K r , nzxe).The
authors found that the cBN content of the films scaled best with ( M Z ~ ~ ~ E ~ ~ ~ )
(i.e. with the total momentum of impinging ions per deposited atom, ptoi/ii).Despite
the quite different deposition parameters in the particular experimental series all
dependences of the cBN percentage on those parameters could be traced back to
practically one and the same curve when ptoi/a as parameter was used. As an
5.4 Spirfter-Deposirion of cBN Films 429
100 ,
80
c
--
E
6 60
s
2 4 0
41
20
0
0 100 200 300 400
P,,,/a (eV x amu)’
Figure 6. Dependence of the cBN percentage in BN thin films on p t o t / ufor different ion masses
(adapted from [ 171).
example, in Fig. 6 the cBN content dependence on ptot/u for different ion masses is
+
shown. To get practically phase-pure cBN films (i.e. Zc/(Zc Zh)2 0.8),p,,,/a has to
exceed a critical value of about 20(1-250 (eV amu)’I2 (amu: atomic mass units).
In addition, based on their ion impact approach, Mirkarimi et (11. [ 171 established
a quantitative model of defect generation during ion-assisted film growth. They
found that the maximum height of the defect production distribution exhibits
the same (w~lol,Eion)”2 dependence as cBN formation. Moreover, as the authors
emphasize, this is the very same dependence as that found by Windischmann [51]
for the formation of intrinsic compressive stress in thin films deposited under ion
bombardment.
Therefore, we conclude that the ion impact and stress models approaching the
subject from different directions using different ‘pictures’, however, have a great
deal in common. The ion impact during film growth causes subplantation [52] of
particles into the subsurface region of the film thus modifying its physical state:
The film gets more defect-rich, densified and compressively stressed. These are
all different characteristic of the same physical phenomenon. From that point
of view it is not surprising that critical values of the compressive stress or ptot/u
can be found which mark the onset of cBN growth. However. those critical
values represent ‘projections’ of the cBN formation mechanism onto the p l o , / a
and film stress scales. respectively. They d o not deliver information on the physical
mechanism of cBN formation itself.
5.4 Sputter Deposition of cBN Films

In this section cBN film preparation by sputter deposition will be discussed in detail.
Such techniques have a high potential for later scaling up of deposition devices and
for practical utilization. such as in the field of tool coatings.
430 5 Vapor Phase Deposition of' Cubic Boron Nitride Films
5.4.1 Sputter Deposition with Conducting Targets

Most of the sputter deposition experiments known so far were performed using
either hexagonal boron nitride (hBN) or boron (B) targets. Due to the low
electrical conductivity of hBN and B, the sputter deposition of cBN films requires
radio frequency (RF) power at the target electrodes, both in diode [20,21] or
magnetron [23] arrangements. With respect to industrial demands like high rate
and large area deposition processes, d.c. magnetron sputtering seems to be very
promising. However, for the application of this technique, conducting target
materials have to be used. Possible candidates are low-resistivity boron rich
compounds, like the boride ZrBI2.In the work reported here, as target materials
boron carbide (B4C, p < lo2 Rcm at room temperature) was chosen [24,53]. The
development of a d.c. magnetron sputter process for the B4C target was subdivided
in different phases. First, an R F diode sputter process was optimized for an hBN
target [21] and transferred to a B4C target [53], followed by d.c. magnetron
sputtering experiments [24,54]. Boron carbide was preferred because it could be
expected that the only 'impurity' carbon, at least partially, can be removed by a
reaction with nitrogen, such as towards stable gas molecules like C2N2 [55]. For
the d.c. magnetron a mixed material target consisting of boron nitride and
titanium diboride (BN/TiB2, both with 50 weight-%, p 5 R cm) was also
tested [56].
As mentioned above, Hackenberger et al. [41] concluded that stoichiometric or
nearly stoichiometric BN films contained the highest amount of the cubic phase.
In sputter deposited films both from the hBN target and the B4C target, atomic
ratios B : N z 1 in the films were achieved [21,24,54].
The process data for the sputter deposition of cBN films by R F (13.56MHz)
diode sputtering with hBN and B4C targets as well as by d.c. magnetron sputtering
with B4C targets have been reported in detail elsewhere [21,24,54]. In both con-
figurations the ionization was enhanced by additional magnetic coils positioned
around the substrate holder. The substrate electrodes were operated either with a
d.c. potential, with pulsed d.c. [57] or with R F power.
Some important parameters of the deposition experiments with conducting
targets will be summarized as follows.
R F diode sputtering:
- target diameter: 150 mm (hBN and B4C)
- target-substrate distance: 70 mm
~ target power: 750-1000 W

- substrate potential UBs)-50 to -500 V (d.c. or d.c. self bias/RF)
- substrate temperature: up to 350°C
- nitrogen flow (N2/Ar + N2): 0-100%

- deposition rate: 5 0.5 pm h-'
d.c. magnetron sputtering:

- target dimensions: 254 mm x 127mm (B4C and BN/TiB2)
- target-substrate distance: 80 mm
5.4 Sputter Deposition of'cBN Filnts 43 1
- target power: 750-1 500 W

- negative substrate potential ( UHS):up to -550 V (d.c. or d.c. self mode)
- substrate temperature: up to 500°C
- nitrogen flow: up to 50% for B4C, up to 100 YOfor BN/TiB2
- deposition rate: 5 0.5 pm h
In the d.c. magnetron mode at higher nitrogen flows for the boron carbide target the
danger of poisoning and arcing increased. On the other hand, the BN/TiB2 target
could be operated with a high stability even at high target power and high nitrogen
flows.
With the R F as well as the d.c. magnetron sputter technique a films consisting of
almost single phase cBN could be deposited. For both techniques the process
parameters (target power, substrate voltage, ion current density) necessary to
generate cBN (cBN parameters) were determined [24,54]. To achieve B : N ratios
near 1 and high ( ~ 6 0 % cBN
) contents, for the hBN target only a few percent nitro-
gen flow in the sputter gas and for the boron carbide target N2 flows > l o % were
needed [24,58].
Cross-section TEM investigations of these sputter deposited films revealed the
typical phase evolution well known for cBN films. Electron diffraction patterns
indicated that in our films the c-axis ([0002]) of turbostratic BN is parallel to
( 1 1 1 ) direction of cBN and both are perpendicular to the substrate normal [58].
The maximum thicknesses reached so far with the both sputter techniques were
about 500 nm on Si and about 200 nm on steel substrates.
5.4.2 Deposition by d.c. Magnetron Sputter with a Hot

Boron Target
In order to prevent any incorporation of undesired target constituents, we devel-
oped a special d.c. magnetron sputtering technique utilizing a pure boron target.
In this technique the target is operated at elevated temperature. Since boron has
semiconducting properties it becomes increasingly conductive if the temperature
is raised. Owing to the special design of the magnetron source the thermal contact
between the water-cooled copper support and the clamped target plane was
relatively bad. Therefore, by R F sputtering under appropriate conditions, the tem-
perature of the boron target could be increased quickly to about 800°C. This led to
an increase in conductivity of six orders of magnitude compared to that at room
temperature [59], which was high enough to ignite a stable d.c. magnetron sputter
discharge with the boron cathode. During d.c. discharge the steady state tempera-
ture of the target is about 600°C [60].
The experiments have been performed using a circular R F magnetron source
(diameter l00mm). The substrate holder (diameter 150mm) was mounted at a
distance of l00mm from the target surface and was separately R F powered yield-
ing a certain negative d.c. potential of the substrate. This d.c. self-bias will be
referred to in the following as the substrate bias voltage, UBS.The substrate tem-
perature and total pressure were held constant during deposition at about 350°C
1.0 1000
-*- intensity ratio
0.8 - - 800
n
4 -
+
3u 0.6 - 5?
v
1
3 0 - -1500 3r
0
3 0.4-
2
8
-400 2
0
.M
c
3
cn
8
U
0.2-
-
- 200
E
Y
Y
0.0 - *-*'
l ~ l ' l ' l ~ l ' l I ~' l0 ' ~ '
0 -50 -100 -150 -200 -250 -300 -350 -400 -450
Substrate bias voltage [V]
Figure 7. cBN content together with the deposition rate of BN films deposited at different substrate
bias voltages by d.c. magnetron sputtering of a boron target.
and 0.2Pa. The sputter gas was argon plus 10% nitrogen and polished silicon
wafers have been used as substrates. For d.c. sputtering of the boron target a
power of 200 W was utilized.
Varying the substrate bias voltage leads to the evolution of BN layers containing
various phase compositions (Fig. 7): From noncubic modifications obtained with-
out external substrate bias and at UBs = -lOOV to films with the highest cBN
content at UBs = -350 V. A further increase of the substrate bias to UBs = -400 V
causes growth conditions resulting in BN films with a lower cBN/hBN ratio. The
growth rate decreases if - UBs is increased mainly due to resputtering. In contrast
to that, for R F magnetron sputtering with either an hBN or a boron target
performed in the same vacuum chamber maximum cBN content was obtained at
UBs = -15OV [25]. However, it has been revealed by Langmuir probe measure-
ments [61] that this reduction of substrate bias voltage for optimum cBN formation
is connected with an about twice as high plasma density in front the substrate.
Therefore, qualitatively the decreased plasma density has to be compensated by a
higher ion energy.
A quantitative evaluation of the three magnetron sputter processes with respect to
the film-forming particle fluxes (for the analysis procedure see section 5.5.2) revealed
that within the error of measurement the d.c. process with boron target had the same
criticalptotlabetween 180 and 265 (eV amu)'I2 as the two R F processes [62]. This is
in good agreement with the values found for pulsed laser deposition by Mirkarimi
et al. [17].
5.5 Discrimination between Nucleation and

Growth Phase
In nearly all investigations published until now certain experimental conditions were
maintained during the whole growth process and then assessed with respect to the
cBN percentage in the films grown. This is also true for all experiments discussed
in section 5.2 of this chapter. To the knowledge of the authors there are only few
exceptions: McKenzie et al. [63] used a helicon activated reactive evaporation pro-
cess and found that for good cBN formation an R F substrate bias of 100 W was
necessary. However, after growing 40 nm of the film the cBN growth could be main-
tained even when the substrate bias was reduced to 50 W. They concluded that the
cBN growth, once nucleated, can be continued at a lower bias level. However, they
did not give quantitative data in terms of bias voltage, ion energy or ptot/a.Schutze
et al. [57] investigated the reactive sputtering from a B4C target. They monitored the
tBN/cBN transition by measuring the substrate current and changed the parameters
after cBN nucleation to ‘softer’ conditions. Hahn et al. [64,65] were the first who
systematically varied and quantitatively characterized the deposition conditions in
order to investigate nucleation and growth of cBN separately. In an R F magnetron
sputtering process they started with conditions for optimum cBN formation in order
to nucleate the cubic phase. After that, however, they continued the growth under
reduced ion impact. It was found that the cBN growth can be maintained even when
the ion impact is drastically reduced in comparison with the conditions which are
necessary to start cBN formation. Experiments concerning the discrimination
between nucleation and growth will be discussed in the following sections in more
detail.
5.5.1 Detection of hBN-cBN Transition

To the knowledge of the authors there are few experimental experiences concerning
the discrimination of nucleation and growth of cBN. Schiitze et al. [57] developed a
simple method to control the cBN nucleation in situ by monitoring the substrate
current. This is possible, for example, if in the R F sputter device the substrate
electrode is operated with d.c. voltage. Fig. 8a shows the principle.
Keeping all process parameters constant, an increase in the substrate current was
observed if the deposition had been performed in the cBN mode, but it was not
observed if the deposition was performed in an hBN mode. This can be explained
+ +
as follows. The measured substrate current Zsub = Zi ZsE = Zi( 1 r i ) is the sum
of the ion current Zistriking the substrate surface and the current of secondary elec-
trons, ZsE, generated at the substrate surface by the incoming ions (ri = secondary
electron yield). With the assumption that the 7icBN > YihBN, one can correlate the
increase of the substrate current with the change from hBN growth (range A) to
cBN growth (B) in the films. The idea of different secondary electron yields was
proved by measurements of the ion current component Zi using a Faraday cup
arrangement mounted into the substrate holder. While the substrate current
80
b
A1
20
0
Deposition time [a. u.]
0.o001 0.001 0.01 0.1 1
AUI
Figure 8. (a) Scheme of the observed substrate current increase during BN film deposition where f
means the current during hBN growth, (b) cBN content against relative current increase A f/f, the
parameters of the deposition experiments were: R F powered boron carbide target (750W), 10%
nitrogen in Ar/N2 and UBs = -200 V.
increased (typical values of AZ/Isub z 0.1) the measured ion current Z,stayed nearly
constant. Furthermore a clear correlation between AI and the cBN content in the
films was found (Fig. 8b). With increasing AZ the cBN content in the films increased,
too. An additional hint to higher secondary electron yields of cBN compared to
hBN was the brightness difference in scanning electron microscope images taken
in the secondary electron mode under identical conditions for both samples [55].
The method described allows control of the cBN deposition process. For example
in the cBN nucleation range (left part of interval B in Fig. 8a) the deposition
parameters can be modified to ‘softer’ conditions. One possibility is to change the
sputter gas composition from high Ar contents, which are related to high stress
and cause cBN nucleation [48], to pure nitrogen. Figure 9 shows a SIMS profile
of a film grown under changed conditions. The deposition was started under typical
‘cBN conditions’ with 10% N2 in the sputter gas and after nucleation the gas
composition was changed to 100% NZ.The diagram reveals that the higher nitrogen
’
r_l
60
50
s40
Y
+g
,g* 30
20
10
u 0
0.00 0.05 0.10 0.15 0.20
Depth [PI
Figure 9. SIMS depth profile of boron, nitrogen, and carbon in a cBN film sputter deposited under
different conditions on a silicon substrate.
5.5 Discriniination between Nucleation and Growth Phase 435
Table 2. Deposition conditions for separate investigations of nucleation and growth phase
Step R F power (W) N2 (YO) At (min) uBS (v) Aiil (nm)

~
1 1000 3 -15 - -
2 500 3 -15 -400 2 -100

3 1000 3 -1 -150 -10
4 1000 100 1&30 -200
-175
-150
-125
-100 -100-150
-80
-60
-40
+15
For all steps Ts = 350"C, p = 0.2 Pa.
concentration leads to a considerable decrease in the carbon content to about 5 at-%

This behavior is to be expected, taking into account element concentration
measurements of cBN films deposited with different nitrogen contents in the sputter
gas. An interpretation of the observed carbon reduction was given in section 5.4.1.
The total incorporated momentum per deposited atom, p t o t/ a , was estimated to be
about 20% lower in the pure nitrogen deposition compared to the Ar/N mixture.
There are still other methods which are suitable for controlling the phase evaluation
in cBN films. Barth et al. [66] measured in situ infrared reflectance spectra during the
deposition time and revealed different deposition processes on silicon and steel sub-
strates, respectively. McKenzie et al. [48] concluded from in situ stress measurements
that there is a stress threshold above which the cBN growth is beginning. However, in
both cases the process conditions were not modified during film growth.
5.5.2 RF Magnetron Sputtering

The deposition experiments using RF magnetron sputtering of an hBN target
[64,65]were carried out using the deposition chamber described in section 5.4.2
following the experimental design given in Table 2. The whole deposition process
consisted of four steps. During step 1 the hBN target was sputter-cleaned in an
Ar/N2 mixture. Then, the shutter between target and substrate was opened, and a
relatively high substrate bias voltage UBs = -4OOV was applied. Owing to the
corresponding high ion energy, strong resputtering prevents any net film growth.
Nevertheless, an intense intermixing of boron and nitrogen with the silicon substrate
takes place yielding an intermixed region and hence an improved film adhesion.
After that, step 3 was carried out with UBs reduced to - 150 V which had been
found before to be the optimum value for cBN growth for that process (see section
5.4.2). Under these conditions a thin cBN film (about 5nm thick) was already
formed after a deposition time of 1 min. In step 4 the deposition was continued in a
pure nitrogen atmosphere. The whole multistep process has been repeated several
times and the substrate bias voltage during step 4 was varied in each case within
the range from -2OOV and the floating potential of the substrate holder (about
+10 V). The total deposition time was chosen so that all films got comparable thick-
ness values of 100-150nm.
The film structure has been characterized by infrared absorption, electron diffrac-
tion, and cross-section transmission electron microscopy. It was found that by
changing the gas atmosphere to pure nitrogen during step 4, but maintaining the
substrate bias voltage at -150V, the IR peak intensity ratio Zc/(Zc + Zh) was
about 0.84.9, which is the same value as in the case of cBN growth under optimum
conditions in an argon-nitrogen mixture (97% Ar/3 YON2). By transmission elec-
tron microscopy and electron diffraction it was confirmed that the BN films were
composed of small (10-50nm) grains of the cubic phase. The film composition
was measured using elastic recoil detection analysis (ERDA) and it was found
that the films were (within the error of measurement) stoichiometric BN
(B/N = 0.95 f 0.05). Practically 100% cubic film structure was obtained within a
wide range of substrate bias voltage during step 4, ranging from -200 V to values
as low as -6OV (Fig. 10). The deposition rate was about 200nmh-' for
UBs = -200 V and increased gradually with decreasing bias voltage due to reduced
resputtering. The increase of the deposition rate between -40 and -60 V (Fig. lo),
however, is mainly due to the different mass density of the cubic and hexagonal BN
phase. From that it was concluded that for voltage values below - 100 V, resputter-
ing could be neglected and the rate of mass deposition was practically constant.
0.8 - - 0.8
n
cf
+ 0
u* 0.6 - -0.6 3
--.
W
U"
3.
v
.-0 3
E 0.4 - -0.4 8
."0
8 0.2 - I
: E-0.2
U
-*- intensity ratio
0.0 - a 0 0 - 0.0
I I I I I I I I
0 -40 -80 -120 -160 -200
Substrate bias voltage [V]
Figure 10. Cubic BN phase content together with the deposition rate of BN films deposited at
different substrate bias voltages during step 4.
5.5 Discrimination between Nucleation m d Growth Phase 437
k' I ' I ' I ' I ' I " '

600 800 1000 1200 1400 1600 1800 200
Wavenumber [cm']
Figure 11. FT-IR transmission spectra demonstrating the phase inhomogeneity of BN films depos-
ited at U,, = -60 V. The IR beam with a diameter of about 1 mm was moved in 1 mm steps radially
from the outer region towards the center of the substrate (spectra a-g). Within the lateral resolution
limit of the method a sharp transition from noncubic (outer part) to cubic BN (inner part) is found.
For UBs= -60V and -8OV, only a circular central part of the film (having a
diameter of about 3cm) was cubic BN. Outside this central part, the film was
purely sp2 bonded (hBN). Considering the lateral resolution of the method, the
boundary between the cBN and hBN region was very sharp. This can be seen
from Fig. 11: the infrared spectra shown in this figure were measured in distances
of 1 mm using a beam diameter of 1 mm. Within a distance of only about 3 mm
the absorption features change significantly between the typical pictures of the
cubic and hexagonal BN phase.
The charge carrier density in the magnetron plasma was measured by Langmuir
probe measurements [61]. As an example, in Fig. 12 the distribution of the charge
carrier density is presented which had been measured lOmm in front of the sub-
strate. Because the charge carrier density at the plasma sheath is directly related
to the ion current towards the substrate (see below), the inhomogeneity visible in
Fig. 12 will be reproduced in the ion flux towards the substrate, that is, the ion
438 5 Vapor Phase Deposition qf Cubic Boron Nitride Filnis
1 ' 1 ' 1 ' ' 1 . 1 '

-60 -40 -20 20 40 0
Radial probe position [mm]
Figure 12. Radial distribution of the charge carrier density lOmm from the substrate surface (hBN
target, R F target power IOOOW, substrate bias voltage UBS= -15OV.p = 0.2Pa, 97% Ar. 3% N?).
flux will be about 15% higher in the center of the substrate holder than at r = 2cm.
This is obviously a relevant fact for the BN phase formation and will be discussed
later.
In order to describe the growth conditions in a quantitative manner the particle
fluxes towards the substrate have been estimated. The total flux of incorporated
atoms (both boron and nitrogen), Fa, was estimated from film stoichiometry,
growth rate, and mass density. The flux of ions cannot be directly measured for
several reasons, above all because of the high resistivity of the film which necessitates
an RF substrate bias. Therefore an estimation based on the Bohm sheath criterion
was made [25]. The Bohm criterion [67] yields an estimation of the ion flux, F,,
towards the substrate when ion density, ni, electron temperature, T,, and ion
mass, mi, are known. These quantities, however, can be determined by Langmuir
probe measurements (ni, T,) or estimated from gas composition (mi). Finally, the
kinetic ion energy, Ei, was roughly estimated (neglecting the plasma potential) to
be Ei = elUBsl.
Using the values of Fa, Fi, and Ei, the total incorporated momentum per
deposited atom, ptot/a, for the experimental series was calculated (Fig. 13). Under
optimum conditions for cBN growth with the given method used during step 3
(97% Ar/3% N2, Uss = -150V) ptot/a is about 200(eVamu)'/2, which is in good
P '
agreement with Mirkarimis results. Changing to ure nitrogen as the sputtering
gas, ptot/a reduces to approximately 150 (eV amu)' *. With an additional reduction
of UBs it was finally found that the continuation of cBN growth withptot/avalues as
low as about 50 (eV amu)'/2 was possible. These results lead to the conclusion that
the continuation of cBN growth after a certain initial cBN layer has been formed is
800
H
h-BN C-BN
N
-
-
n
600 -
s
*2
2cd
v
u
---
400-
. .
ArM2=9713
a” 200 -
0
. I
.
I
. a
1
H
I I I
governed by an epitaxial mechanism. It has been found that the upper few mono-
layers of a growing cBN film are sp2 bonded [68,69]. The film forming particles
penetrate this layer and reach somewhat deeper regions of the film where they finally
become incorporated into the already existing cBN. The certain minimum energy
and/or momentum of the ions which was found to be still necessary to maintain
the cBN growth is obviously due to the fact that the film-forming particles have
to penetrate the sp2 bonded layer.
In the meantime, Litvinov and Clarke [70] have confirmed the results of Hahn
et al. [64,65]. They deposited boron nitride films in an UHV chamber by ion
beam sputtering of a boron target together with a nitrogen ion bombardment of
the growing film. The substrate temperature was 1000°C. The authors found that
after cBN nucleation at UBs = -96 V the substrate bias voltage can be significantly
reduced without interruption of the cBN growth. At UBs = -56V the crystalline
quality of the layers is even better that at -96 V. The reduction of the bias voltage
below approximately -50 V, however, leads to the formation of predominantly
hexagonal phase material.
Interestingly, a similar nucleation effect as found for the ion impact seems to exist
in the case of the substrate temperature. As has been found by several authors
[34,36] a minimum substrate temperature of 150°C is necessary in order to nucleate
cubic BN in ion-assisted deposition. However, very recently Feldermann et al. [71]
have found for mass selected ion beam deposition that switching off the substrate
heating after cBN nucleation and allowing the substrate to cool down to room
temperature, does not interrupt cBN growth.
5.6 Properties of cBN Films

The aim of this section is to give an impression on properties of thin cBN films
reported so far. It was found that the film properties are very similar to those
known for bulk material cBN.
5.6.1 Mechanical and Tribological Properties

Only few reports concerning mechanical and tribological data are available so
far. Therefore these properties are a main topic of this chapter. Because of the
relatively small thicknesses (<1 pm) of cBN films which can be reached with
good adhesion, it is difficult to determine reliable data of hardness, Young’s
modulus, and wear.
Knoop hardness values of about 50 GPa for cBN and 25 GPa for amorphous BN
(aBN) films were reported by Murakawa et al. [72]. From nanometer indentation
measurements hardness ( H ) and Young’s modulus ( E ) were derived for different
sputter deposited cBN films (thickness some 100 nm) [54,55,73]. The hardness
values ranged between 60 and 80GPa which is close to the bulk hardness of cBN
(up to 90GPa [l]). On the other hand the Young’s moduli, E, in the range 300-
600GPa, are lower than the cBN bulk values (800-900GPa) [l]. Mirkarimi et al.
[73] explain this discrepancy by substrate effects which decrease with increasing
film thickness.
Microwear tests performed using atomic force microscopy (AFM) were reported
by Miyake et al. [74] and by Schutze [55]. The result was that the wear caused by a
diamond tip is much greater for hBN than for cBN. The microwear rates of sputter
deposited cBN films were found to be nearly the same as that of hard diamond like
carbon films [55].
Friction coefficients, p, of cBN films were measured under different conditions
(counterpart, humidity, number of cycles, sliding speeds) at normal loads up to
10N. Data available from the literature are summarized in Table 3. In all cases
considered hBN films had clearly higher friction coefficients than cBN.
Firsts attempts to apply cBN coatings for wear protection on drills were reported
already in 1990 by Murakawa and Watanabe [77]. Owing to the still unsolved
problem of depositing films of several micrometers thickness, no essential progress
in tool applications of cBN films has been made so far.
5.6.2 Optical Properties

Several data on refractive index, n, and extinction coefficient, k, of BN films for the
visible, UV, and near IR spectral ranges were published. In all cases the refractive
index of cBN films varies between about 2.0 and 2.1, slightly decreasing with
increasing wavelength (A) [78,55]. These n values agree well with data known for
bulk cBN [79,42]. In the range X > 400 nm the extinction seems to be negligible.
5.6 Properties of cBN Films 441
Table 3. Data on friction behavior of cBN films.
Counterpart Relative humidity Friction coefficient, Remarks Reference

("/.I P
Steel 45 0.2 [751

Diamond 45 <o. 1 Decreasing with [751
increasing load
Steel 60 0.35 Extreme decrease [76]
under vacuum
(<0.01)
Steel 40 0.1-0.4 Strong fluctuations, [55]
p decreases under
dry conditions
Alumina 40 0.0 5-0.5 Strong fluctuations [55]
For films sputter deposited from a boron carbide target, k markedly increased at
X < 400nm [55]. This indicates an effect of the incorporated carbon (5-10 at-%,
see section 5.5.1) in the films. The cathodoluminescence behavior of hBN and
cBN films was investigated by Taylor et al. [80] with respect to possible applications
of the large band gap material BN in the deep blue and UV range.
5.6.3 Electrical Properties

Undoped BN films exhibit low electrical conductivities. Ronning et al. [81] deter-
mined for ion beam deposited films resistivities of 109-10'0Rcm and the Z-U
curves were best approximated assuming a Frenkel-Poole conduction mechanism.
These resistivities are near to the expected bulk value (up to 10" R cm [82]). Further-
more the dielectric constant was estimated as e = 8-10.
The resistivities of films sputter deposited with a boron carbide target were lower
(107-109 R cm). Comparing the results of lateral and sandwich measuring arrange-
ments, no essential differences were found [55].
Obviously the conductivity of BN films can be decreased by doping, for example
with magnesium to prepare a p-type semiconductor [83].
5.6.4 Other Properties

Boron nitride is a wide band gap semiconductor. For BN a negative electron affinity
(NEA) can be expected [84]. Materials with this property can be used, for example,
in cold cathode emitters. NEA occurs if the vacuum energy level lies below the mini-
mum of the conduction band. Beside the electronic structure of the bulk material the
surface conditions of the considered material are quite important [85].
For sputter deposited cBN films no hints of NEA were found, but UPS spectra of
the same cBN films, additionally coated with a monolayer of Cs, provided indica-
tions of NEA [55,86].
442 5 Vapor Phuse Deposition of Cubic Boron Nitride Films
With the Faraday cup arrangement in the substrate holder used for the experiments
on cBN nucleation which were discussed in section 5.5.1 additionally secondary elec-
tron yields, yi, both of hBN and cBN films could be determined. Here yi is the yield for
ion-induced electron emission. For both phases in the ion energy range between 100
and 600 eV for an argon-nitrogen gas mixture, "ii near to 1 were estimated. According
to the observed substrate current increase (section 5.5.1) the values of cBN were
calculated to be about 20% higher than those of hBN [86].
The plasma etching behavior of BN films is interesting, for example with respect
to deposition process modifications. However, there are still many open questions
concerning the effect of additional gas components in the process gases of cBN
deposition processes. For ion etching in pure Ar the ratio of the sputter yields of
hBN and cBN was revealed to be YhBN/YcBN5 1 [55,64]. On the other hand,
Harris et al. [87] observed, for an etching gas consisting of 1% CH4 and 99% HZ,
a considerably higher etch rate for the hBN than for the cBN phase.
The surface tension of sputter deposited cBN films was determined by contact
angle measurements with different well known liquids [88]. The value of
40 mN m-l is nearly the same as that of DLC (diamond like carbon) [89].
5.7 Summary and Outlook

There are much experimental data on deposition of cBN films. Both CVD and PVD
methods are suitable for preparing films with high contents of the cubic phase
material. Essential parameters determining the film growth are ion impact onto the
growing film as well as the substrate temperature. The cBN content in the films
obviously scales best with the total momentum of impinging ions per deposited atom.
Sputter deposition techniques in the R F as well as in the d.c. mode were success-
fully used to prepare nearly single phase cBN films.
At present the insufficient adhesion, caused by high compressive stress and sensi-
tivity to humidity, is the most essential problem to be solved. A promising way to
overcome this difficulty could be discrimination between the nucleation and
growth phases. Furthermore a better understanding of the correlation between
growth conditions, film structure, and stress is needed.
If a sufficient adhesion of cBN films on different substrates can be achieved for a
thickness range of 1-2pm or more, cBN coatings will have a high potential in
several fields of application, especially in the field of tools for ferrous materials.
References
1. J. H. Edgar, in Properties ojGroup III Nitrides, J. H. Edgar (Ed.), INSPEC, the Institution of
Electrical Engineers, London, 1994, pp. 3-21,
2. A. Inspektor, C. E. Bauer, and E. J. Oles, Surf. Coat. Technol. 1994,68/69, 359-368.
References 443
3. W. R. L. Lambrecti and B. Segall, in Propertics qf Group III Nifrides, J. H. Edgar (Ed.),

INSPEC, the Institution of Electrical Engineers, London, 1994, pp. 129-1 34.
4. R. H. WentorfJr., J . Chem. Phys. 1962,36, 1990-1991.
5. R. H. Wentorf Jr., J . Chem. Phys. 1957, 26, 956.
6. R. H. Wentorf Jr., J . Chem. Phys. 1961, 34, 809-812.
7. V. L. Soloshenko, Diamond Rel. Muter. 1994,4, 1 4 , has delivered strong evidence that cBN is
the thermodynamically stable phase at ambient conditions, too. The formation of cBN at those
conditions, however, is obviously kinetically inhibited.
8. R. J. Caveney, Muter. Sci. Eng. 1992, B11, 197-205.
9. For a review see: J. J. Pouch and S. A. Alterowitz (Eds), Synthesis and Properties of Boron
Nitride, TransTech Publications, Aedermannsdorf, Switzerland, 1990.
10. K. Inagawa, K. Watanabe, K. Saitoh, and Y. Yuchi, Surf: Coat. Technol. 1989, 39, 253-264.
11. M. Murakawa and S. Watanabe, Surf: Coat. Technol. 1990,43/44, 128-136.
12. T. Ikeda, Y. Kawate, and Y. Hirai, J . Vac. Sci. Technol. A , 1990, 8, 3168-3174.
13. T. Wada and N. Yamashida, J . Vuc. Sci. Technol. A , 1992, 10, 515-520.
14. D. J. Kester and R. Messier, J . Appl. Phys. 1992, 72, 504513.
15. N. Tdnabe, T. Hayashi, and M. Iwaki, Diamond Rel. Muter. 1992, 1, 883-890.
16. N. Tanabe, M. Iwaki, Diamond Rel. Muter. 1993, 2, 512-516.
17. P. B. Mirkarimi, K. F. McCarty, D. L. Medlin, W. G. Wolfer, T. A. Friedmann, E. J. Klaus,
G. F. Cardinale, and D. G. Howitt, J . Muter. Res. 1994, 9, 2925-2938.
18. H. Hofsass and C. Ronning, in Proc. Int. Conf. on Beam Processing of Advanced Materials,
ASM International, Materials Park, OH, 1996, pp. 29-56.
19. G. Krannich, F. Richter, J. Hahn, R. Pintaske, V. B. Fillipov, and Y. Pddero, Diamond Rel.
Muter. 1997, 6, 1005-1009.
20. M. Mieno and T. Yoshida, Jup. J . Appl. Phys. 1990,29, L1175-Ll177.
21. K. Bewilogua, J. Buth, H. Hiibsch, and M. Grischke, DiamondRel. Muter. 1993,2, 1206-1210.
22. V. Yu. Kulikowsky. L. R. Shaginydn, V. M. Vereschaka, and N. G. Hatynenko, Diamond Re/.
Muter. 1995, 4, 113-1 19.
23. D. G. Rickerby, P. N. Gibson, W. Gissler, and J. Haupt, Thin Solid Films, 1992, 209, 155-160.
24. A. Schiitze, K. Bewilogua, H. Liithje, S. Kouptsidis, and S. Jager, Surr Coat. Technol. 1995,74-
75, 711-122.
25. J. Hahn, M. Friedrich, R. Pintaske, M. Schaller, N. Kahl, D. R. T. Zahn, and F. Richter,
Diamond Rel. Muter. 1996, 5, 1103-1 112.
26. W. Dworschak, K. Jung, and H. Ehrhardt, Diamond Rel. Muter. 1994,3, 337-340.
27. M. Okamoto, Y. Utsumi, and Y. Osaka, Jpn. J . Appl. Phys. 1992,31, 3455-3460.
28. A. Weber, U. Bringmann, R. Nikulski, and C.-P. Klages, S u r t Coat. Technol. 1993,60,493497.
29. R. Freudenstein, S. Reinke, and W. Kulisch, Diamond Rel. Muter. 1997, 6, 584588.
30. F. Zhang, Y. Guo, Z. Song, and G. Chen, Appl. Phys. Lett. 1994,65,971-913.
31. Z. Song, F. Zhang, Y. Guo, and G. Chen, Appl. Phys. Lett. 1994,65, 2669-2671.
32. I. Konyashin, J. Loeffler, J. Bill, and F. Aldinger, Thin Solid Films, 1997, 308-309, 101-106.
33. S. Reinke, M. Kuhr, W. Kulisch, and R. Kassing, Diamond Rel. Muter. 1995, 4, 272-283.
34. H. Hofsass, H. Feldermann, M. Sebastian, and C. Ronning, Phys. Rev. B, 1997, 55, 13230-
13233.
35. P. B. Mirkarimi, D. L. Medlin, K. F. McCarty, D. C. Dibble, W. M. Clift, J. A. Knapp, and
J. C. Barbour, J . Appl. Phys. 1997,82, 1617-1625.
36. P. B. Mirkarimi, D. L. Medlin, K. F. McCarty, and J. C. Barbour, Appl. Phys. Lett. 1995, 66,
28 13-28 15.
37. D. J. Kester, K. S. Ailey, R. F. Davis, and K. L. More, J . Muter. Res. 1993, 8, 1213-1216.
38. D. R. McKenzie, W. D. McFall, S. Reisch, B. W. James, I. S. Falconer, R. W. Boswell, H. Persing,
A. J. Perry, and A. Durandet, Surf: Coat. Technol. 1996,78,255-262.
39. A. K. Ballal, L. Salamanca-Riba, C. A. Taylor 11, and G. L. Doll, Thin Solid Films, 1993,224,
46-5 1.
40. D. L. Medlin, T. A. Friedmann, P. B. Mirkarimi, and K. F. McCarty, J . Appl. Phys. 1996,79,
3567-351 1.
41. L. B. Hackenberger, L. J. Pilione, R. Messier, and G. P. Lamaze, J . Vac. Sci. Tecl?nol.A , 1994,
12, 1569-1575.
42. P. J. Gielisse, S. S. Mitra, J. N. Plendl, R. D. Griffis, L. C. Mansur, R. Marshall, and E. A. Pascoe,

Phys. Rev. 1967,155, 1039-1046.
43. R. Geick, C. H. Perry, and G. Rupprecht, Phys. Rev. 1966, 146, 543-547.
44. S. Jager, K. Bewilogua, and C.-P. Klages, Thin Solid Films, 1994, 245, 50-54.
45. G. F. Cardinale, P. B. Mirkarimi, K. F. McCarty, E. J. Klaus, D. L. Medlin, W. M. Clift, and
D. G. Howitt, Thin Solid Films, 1994, 253, 130-135.
46. P. B. Mirkarimi, K. F. McCarty, G. F. Cardinale, D. L. Medlin, D. K. Ottesen, and H. A. Johnsen,
J. Vac. Sci. TechnolA, 1996,14,251-255.
47. A. Schiitze, K. Bewilogua, H. Liithje, S. Kouptsidis, and M. Gaertner, Surf. Coat. Technol.
1997, 97, 33-38.
48. D. R. McKenzie, W. D. McFall, W. G. Sainty, C. A. Davis, and R. E. Collins, Diamond Rel.
Muter. 1993, 2, 970-976.
49. D. R. McKenzie, D. Muller and B. A. Pailthorpe, Phys. Rev. Lett. 1991, 67, 773-776.
50. J. D. Targove and H. A. McLeod, Appl. Opt. 1988,27,3779-3781.
51. H. Windischmann, J . Appl. Phys. 1987, 62, 1800-1807.
52. J. Robertson, Diamond Rel. Muter. 1996, 5, 519-524.
53. H. Liithje, K. Bewilogua, S. Daaud, M. Johansson, and L. Hultman, Thin Solid Films, 1995,
257,4045.
54. S. Kouptsidis, H. Liithje, K. Bewilogua, A. Schutze, P. Zhang, Diamond Rel. Muter. 1998, 7,
2 6 3 1.
55. A. Schutze, Dr. rer. nat. Thesis, TU Braunschweig/Germany, published in: Fraunhofer-Institut
fur Schicht- und Oberflachentechnik - Berichte aus Forschung und Entwicklung Nr. 2, Fraunhofer
IRB Verlag, 1997.
56. S. Kouptsidis, Dr. rer. nat. Thesis, TU Braunschweig/Germany/Fraunhofer-Institutfur
Schicht- und Oberflachentechnik-Berichte aus Forschung und Entwicklung Nr. 4, Fraunhofer
IRB Verlag, 1998.
57. A. Schiitze,K. Bewilogua, H. Liithje, and S. Kouptsidis, DiamondRel. Muter. 1996,5, 1130-1135.
58. K. Bewilogua, A. Schiitze, S. Kouptsidis, and H. Liithje, Proc. Third Int. Conf. on Applications of
Diamond Films and RelatedMaterials, A. Feldman, Y. Tzeng, W. A. Yarbrough, M. Yoshikawa,
and M. Murakawa (Eds), NIST Special Publications, Washington DC, 1995,831-838.
59. L. Gmelin, Handbuch der anorganischen Chemie, Suppl. Vol. 2, Springer Verlag, Berlin,
Heidelberg, New York, 1981, p. 82.
60. V. LinD, Diploma Thesis, Institut fur Physik, Technische Universitat Chemnitz, Germany,
1998.
61. R. Pintaske, Th. Welzel, N. Kahl, M. Schaller, J. Hahn, and F. Richter, Surf. Coat. Technol.
1997, 90, 275-284.
62. F. Richter, R. Pintaske, J. Hahn, and Th. Welzel, in Hard Coatings Based on Borides, Carbides and
Nitrides: Synthesis Characterization and Applications, A. Kumar, Y.-W. Chang, and R. W. J. Chia
(Eds), The Minerals, Metals and Materials Society, Warrendale, PA, 1998, pp. 153-168.
63. D. R. McKenzie, W. D. McFall, H. Smith, B. Higgins, R. W. Boswell, A. Durandet, B. W.
James, and 1. S. Falconer, Nucl. Instr. Meth. B, 1995, 106, 90-95.
64. J. Hahn, Dr. rer. nat. Thesis, Technische Universitat Chemnitz, Germany, 1997.
65. J. Hahn, F. Richter, R. Pintaske, M. Roder, E. Schneider, and Th. Welzel, Surf. Coat. Technol.
1997,92, 129-134.
66. K.-L. Barth, W. Sigle, D. Stockle, J. Ulmer, and A. Lunk, Thin Solid Films, 1997, 301, 65-70.
67. B. Chapman, Glow Discharge Processes, Wiley, New York, 1980, Chapter 3.
68. G. Sent, D. Bouchier, S. Ilias, M. A. Djouadi, V. Stambuli, P. Moller, G. Hug, and S. Reinke, in
Proc. 4th Intern. Symp. on Diamond Materials, K. V. Ravi, J. L. Davidson, R. H. Hauge, J. P.
Dismukes, K. E. Spear, and B. V. Spitsyn (Eds), Reno, NV, 1995, The Electrochemical Society,
Pennington, NJ, 1995, pp. 377-382.
69. T. A. Friedmann, P. B. Mirkarimi, D. L. Medlin, K. F. McCarty, E. J. Klaus, D. Boehme, H. A.
Johnsen, M. J. Mills, and D. K. Ottesen, J . Appl. Phys. 1994,76, 3088-3101.
70. D. Litvinov and R. Clarke. Aool. Phvs. Lett. 1997. 71. 1969-1971.
71. H. Feldermann, R. Merk, 'H: Hofsass, C. Ronning, T. Zheleva, Appl. Phys. Lett., 1999, 74,
1552-1554.
72. M. Murakawa, S. Watanabe, and S. Miyake, Diamond Films Technol. 1991, 1, 55-66.
References 445
73. P. B. Mirkarimi, D. L. Medlin, K. F. McCarty, D. C. Dibble, W. M. Clift, J. A. Knapp, and J. C.

Barbour, J. Appl. Phys. 1997,82, 1617-1625.
74. S. Miyake, S. Watanabe, M. Murakawa, R. Kaneko, and T. Miyamoto, Thin SolidFilms, 1992,
212, 262-266.
75. S. Watanabe, S. Miyake, and M. Murakawa, Surf: Coat. Technol. 1995, 76-77, 600-603.
76. Z. Yu, K. Inagawa and Z. Jin, Surf. Coat. Technol. 1994,70, 147-150.
77. M. Murakawa, S. Watanabe, Surf. Coat. Technol. 1990,43144, 145-153.
78. W. D. McFall, D. R. McKenzie, and R. P. Netterfield, Surf. Coat. Technol. 1996, 81, 72-78.
79. G. L. Doll, in Properties of Group III Nitrides, J. H. Edgar (Ed.), INSPEC, the Institution of
Electrical Engineers, London, 1994, pp. 169-1 74.
80. C. A. Taylor 11, S. W. Brown, V. Subramaniam, S. Kidner, S. C. Rand, and R. Clarke, Appl.
Phys. Lett. 1994,65, 1251-1253.
81. C. Ronning, E. Dreher, H. Feldermann, M. Gross, M. Sebastian, and H. Hofsass, Diamond Re/.
Muter. 1997,6, 1129-1 134.
82. V. V. Lopatin, in Properties of Group III Nitrides, J. H. Edgar (Ed.), INSPEC, the Institution of
Electrical Engineers, London, 1994, pp. 89-100.
83. M. Lu, A. Bousetta, A. Bensaoula, K. Waters, and J. A. Schultz, Appl. Phys. Lett. 1996, 68,
622-624.
84. M. J. Powers, M. C. Benjamin, L. M. Porter, R. J. Nemanich, R. F. Davis, J. J. Cuomo, G. L.
Doll and S. J. Harris, Appl. Phys. Lett. 1995, 67, 3912-3914.
85. R. J. Nemanich, P. K. Baumann, M. C. Benjamin, S. W. King, J. van der Weide, and R. F.
Davis, Diamond Rel. Muter. 1996, 5, 790-796.
86. K. Bewilogua, A. Schiitze, H. Walter, S Kouptsidis, Proceedings of the Second Symposium on
III-V Nitride Materials and Processes (Eds. C. R. Abernathy, W. D. Brown, D. N. Buckley,
J. P. Dismukes, M. Kamp, T. D. Moustakas, S. J. Pearton, F. Ren), The Electrochemical
Society, Inc., Pennington, NJ, 1998, Proceedings volume 97-34, 134141.
87. S. J. Harris, A. Weiner, G. L. Doll, and W.-J. Meng, J . Muter. Res. 1997, 12, 412415.
88. K. Trojan, M. Grischke, and H. Dimigen, Phys. Stat. Sol. ( a ) , 1994, 145, 575-585.
89. M. Grischke, K. Bewilogua, K. Trojan, and H. Dimigen, Surf. Coat. Technol. 1995,7475,739-
745.
6 Polymer to Ceramic Transformation: Processing

of Ceramic Bodies and Thin Films
G. D. Soraru and P. Colombo
6.1 Introduction
Polymer pyrolysis to form advanced ceramics allows the production of highly
covalent refractory components (fibers, films, membranes, foams, joints, monolithic
bodies, ceramic matrix composites) that are difficult to fabricate via the traditional
powder processing route [14]. Yajima was the first to demonstrate the feasibility of
producing high-strength SiC-based fibers from pyrolysis of polycarbosilane [5]. In
this process, a thermoplastic pre-ceramic polymer is first shaped into the desired
form, cross-linked into a pre-ceramic network and finally converted into a ceramic
material by a pyrolysis process in a controlled atmosphere (Fig. 1). A common
feature of the polymer route is the formation of intermediates called amorphous
covalent ceramics (ACC) [6]. These are formed after removal of the organic
components and before crystallization that occurs at higher temperatures.
Many experimental techniques have been applied to characterize the polymer-to-
ceramic transformations. They range from differential thermal analysis/thermo-
gravimetrical analysis/mass spectometry (DTA/TGA/MS) [7,8] to spectroscopies
which are especially helpful for the low temperature range when the ceramic residue
is still amorphous such as: magic angle spinning nuclear magnetic resonance
(MAS NMR) [9,10], XPS [11, 121, and X-ray absorption near edge structure-
extended X-ray absorption fine structure (XANES-EXAFS) [I 3,141, to X-ray
diffraction (XRD) and transmission electron microscopy (TEM) to characterize
the crystallization process [15,16]. Among the various systems already synthesized,
S i c obtained from polycarbosilane is certainly the most widely studied and the
proposed structure of the amorphous silicon carbide phase obtained after pyrolysis
in Ar at 800°C is shown in Fig. 2 [17].
One of the most attractive and challenging features of this process is the possibi-
lity of designing the composition, the microstructure, and the properties of the final
ceramic material by controlling each processing step. The processing steps are:
- the polymer composition and/or its architecture,
~ the cross-linking step,
- the pyrolytic conversion process, and
- the final high temperature annealing to induce densification and/or crystal-
lization.
Indeed, many Si-containing ceramic systems have been already prepared from poly-
mers of different composition such as: polysiloxanes (-R, R2Si-O-), to give SiOC
[ 18,191, polysilanes (-R1 R2-Si-), [20,2 I], and polycarbosilanes (-R2Si-CH2-),
f Pb'O'ecular
recursor I
1
Inert or
Reactive
Atmosphere
Fibers
A m o r p hou s
Films Covalent
Foams C e ra m ic
Joints
lnfiltralion
for CMC
N a n o - c' r y s t a I Iin e
Bulks Ceramic
Figure 1. Flow chart for the preparation of ceramic components viu polymer pyrolysis.
[22-241 to give S i c and polysilazane (-R2Si-NH-), to give SiCN [25] and SiBCN
ceramics [26].
Moreover, the various polymeric precursors can be easily modified by chemical
reactions to obtain new systems. A good example is the preparation of polytitanocar-
bosilane (PTC) from polycarbosilane and titanium alkoxide [27]. This precursor
transforms into SiC/TiC during pyrolysis. The fibers produced from this modified
precursors seem to exhibit better mechanical properties than those of S i c [28]. The
same reaction used to produce PTC gives the opportunity of modifying the polycar-
bosilane with different elements just by changing the nature of the metal alkoxide and
accordingly a large variety of new metal organic polymers, precursors for S-M-C4
(M = Al, Zr, Ta) systems have been obtained [29-311. Another approach to control
the composition and the microstructure of the final product is to use ceramic or metal-
lic powders as filling agents or reagents in the starting polymers [32,33]. The selectivity
toward defined ceramic phases can also be controlled by varying the polymer archi-
tecture. In this case the pyrolysis in inert atmosphere of-[MeHSiNH],- leads to SiCN
ceramics at temperatures >8OO0C whereas pyrolysis of isostructural -[H2SiNMeIx-
gives Si3N4and carbon [34] . The precursor molecular structure can also lead to dif-
448 6 Polymer to Ceramic Transformation: Processing of Ceramic Bodies and Thin Films
, Si, I -
\
. I -Si-C, I ’S .i .
CH+~, / “i:
si-Cqc/
’ \ I
Figure 2. Proposed structure of the amorphous silicon carbide phase obtained from polycarbo-
silane.
ferent microstructures of the cross-linked pre-ceramic networks that in turn have a sig-
nificant effect on both the pyrolysis chemistry and the composition and microstructure
of the ceramic products. Indeed, a study performed on polycarbosilane/siloxane pre-
cursors having similar chemical composition but different molecular structures
pointed out the formation of different preceramic microstructures that survive up
to high temperatures (1000°C) leading to SiOC glasses of different microporosity
and thermal stability [35].
The curing step can play a vital role in defining the structure and mechanical
properties of polymer derived ceramics. Indeed, the tensile strength of Sic-based
Nicalon fibers drops quickly at high temperatures due to the oxygen introduced
during the thermal cross-linking stage. It is believed that the oxygen gives rise,
upon pyrolysis, to a SiOC phase that decomposes above 1300°C with evolution of
SiO and CO [36]. This problem has been overcome by using a radiation (electron
beam or y-ray irradiation) curing process in helium gas which reduces the oxygen
content down to 0.4weight % and improves the fiber resistance up to 1750°C [37].
A different approach to controlling the composition and structure of the
ceramic products is through the control of the transformation process itself. The
heating rate and the maximum pyrolysis temperature can have a strong influence
on the ceramic yield and the ceramic composition [38]. In Fig. 3 thermogravimetry
curves obtained on a commercial polycarbosilane at different heating rates are
reported. It can be noticed that the ceramic yield is strongly related to the
pyrolysis rate, and it increases by using low heating rates. The pyrolysis atmosphere
can be varied from inert (Ar, He, N2) to reactive (NH,, 0,)and accordingly silicon
nitride and carbonitride-based materials, such as Si-N, Si-C-N, Si-0-N, and Si-
M-0-N (M = Al, Zr, Ti) have been produced from various polysiloxanes, poly-
carbosilane and polysilanes precursors [3942]. More recently synthetic silicates
have been fabricated by thermooxidation of polysiloxanes filled with different
oxide or metallic powders [43].
As far as the final heat treatment step is concerned, it is well known that the
maximum pyrolysis temperature governs the densification of the amorphous
ceramic phase and its crystallization process [44]. The densification of polycarbosi-
lane-derived amorphous silicon carbide was studied [ 17,451 in the temperature
-1 0
-$
u1
- -20
Em
.-
0)
-30
200 400 600 800 1000 1200

Temperature ("C)
Figure 3. Thermogravimetry curves recorded on a commercial polycarbosilane (Dow Corning X9-

6348) in Ar flow (100cm3 min-') at different heating rates.
range 900-1000°C (Fig. 4). An activation energy of 82 kcal mole-' was obtained
from the initial slope of the densification curves (Fig. 5). The observed increase of
density during the isothermal firing treatments has been primarily ascribed to reduc-
tion of the free volume of the amorphous phase. Two main processes may account
for this effect: the progress of the condensation reactions between residual CH
groups in the structure with the elimination of H2 and CH4 and the formation of
new Si-C-Si bridges with a consequent increase of the cross-linking of the network;
and a rearrangement of the open amorphous covalent structure toward a more com-
pact configuration with no change in chemical composition. For both of these
0 2 4 6 8 10 12
Time (h)
Figure 4. Evolution of the density, p, of polycarbosilane-derived amorphous silicon carbide with the
firing temperature. 900°C; +
1000°C; 950°C.
450 6 Polymer to Ceramic Trunsftirrnution: Processing of' Ceramic Bodies and Thin Films
3.9 4.0 4.1 4.2 4.3
IO~IRT
Figure 5. Arrhenius plot for the densification rate of the polycarbosilane-derived amorphous silicon
carbide.
mechanisms, a key step should be the cleavage of chemical bonds: C-H

(99 kcalmole-') and Si-C (76 kcal mole-'). The obtained value of activation
energy suggests that, at least in the initial stages, the densification process occurs
via the cleavage of chemical bonds present in the material.
Up to now, polymer pyrolysis has been investigated especially to develop ceramic
fibers [46,47] and ceramic matrix for ceramic matrix composites [48-501. More
recently studies have been undertaken to exploit this method to develop ceramic
thin films [51-531, foams [54], joints [55], and bulk materials [56]. Moreover, noncon-
ventional heating systems such as laser [57],microwave heating [53], or even athermal
conversion processes such as ion bombardment are just now starting to be applied to
the polymer route and the preliminary results are very promising [58-601. In this chap-
ter we focus on the polymer processing of bulk ceramics obtained by pyrolysis of par-
tially cross-linked preceramic bodies and of thin ceramic films (obtained either by
traditional pyrolysis or by the innovative ion irradiation process).
6.2 Processing of Monolithic Components

The use of polymer pyrolysis for the fabrication of bulk ceramics has been hindered
mainly by the significant volume shrinkage (250%) accompanying the polymer-to-
ceramic conversion, often resulting in the formation of cracks which destroys the
structural integrity of the ceramic component.
In order to overcome these difficulties, many hybrid processes have been studied
for the fabrication of bulk ceramic components via polymer pyrolysis. Greil and
coworkers proposed to reduce the shrinkage by the use of active fillers that can
react during pyrolysis leading to a volume expansion [33], pre-ceramic polymers
6.2 Processing of Monolithic Components 45 1
have been used as active binders in the traditional powders processing [61-631; and
polymer-derived amorphous ceramic powders have been sintered to give dense,
monolithic ceramics [64-671. However, in all these hybrid processes, the potential
advantages of the polymer route are lost, at least to some extent.
Recently, Riedel and coworkers [56,68-721, demonstrated the possibility of
producing a dense, crack-free, monolithic ceramic in the Si-C-N system by direct
pyrolysis of a polymeric green compact. In this method, a pre-ceramic polymer is
first converted into a pre-ceramic network with a controlled cross-linking process.
The powders obtained are then sieved and pressed to get a polymeric pre-ceramic
component, that can be pyrolyzed without cracking, into the final dense ceramic
body. These innovative ceramic components display negligible creep rates up to
very high temperatures [73,74]. The key point of this process is the partial cross-
linking of the polymer precursor that leads to an infusible network preventing the
component from melting and bloating during the first stages of the pyrolysis pro-
cess. Moreover, the open porosity of the green body guarantees the efficient removal
of the gaseous products during pyrolysis preventing the development of cracks.
Thus, the significant volume shrinkage usually associated with the polymer-to-
ceramic transformation no longer prevents the fabrication of bulk, crack-free
ceramic components. However, in order to optimize the microstructure of the cera-
mic product, the influence of many polymer properties and processing parameters
still must be assessed. As an example, the cross-linking degree and the weight loss
during pyrolysis of the polymeric precursor, the powder size and powder size
distribution, the amount of porosity and the pore size distribution of the green
compact, the pyrolysis rate and atmosphere are just some of the parameters that
could strongly influence the microstructure and the related physical, chemical,
and mechanical properties of the final ceramic product.
Here we report on the fabrication and characterization of dense ceramic materials in
the SiAlOC system from pre-ceramic green bodies obtained by pressing partially cross-
linked polyaluminocarbosilane (PAlC) powders with different aluminum loads. The
synthesis of PAlC is carried out reacting a commercial polycarbosilane (PCS) with
an aluminum alkoxide, following a published procedure [27]. PAlC has been already
investigated in our laboratory both with the aim of getting insights into the chemical
reaction between the polycarbosilane chains and the aluminum alkoxide and to clarify
the structure of the resulting amorphous Sic-based ceramic [29,40,47].
The objective of the present work is to investigate the role of some polymer prop-
erties (cross-linking degree, weight loss during pyrolysis, etc.) on the microstructure
and mechanical properties of the ceramic materials obtained by pyrolysis in inert
atmosphere at 1000°C. Indeed, the forementioned parameters can be easily modified
by varying the amount of aluminum alkoxide reacted with the polycarbosilane.
Accordingly, several PAlC compositions have been prepared with a nominal Al/
Si ratio between 0 and 0.25. The corresponding ceramic materials have been
characterized mainly by measuring their density, porosity and mechanical proper-
ties through four point bending tests both at room temperature and at high tempera-
ture up to 14OO0C,in air. The results will be discussed with the aim of disclosing the
key parameters that allow the processing of high strength and high modulus ceramic
components.
Finally, we will show that this innovative processing route can be extended to the
production of bulk ceramic components by pyrolysis in reactive atmosphere, such as
for example, production of SiAlON ceramics by nitridation of PAlC in ammonia
flow [75].
6.3 Preparation and Characterization of SiAlOC

Ceramic Bodies by Pyrolysis in Inert Atmosphere
6.3.1 Experimental Procedure
Polyaluminocarbosilanes (PAlCs), with a nominal AljSi ratio of 0.05, 0.1, 0.15, and
0.25 were prepared reacting a commercial polycarbosilane (Dow Corning X9-6348)
with an aluminum sec-butoxide modified with ethylacetoacetate, A1(Obus)*etac
(Alfa product, 89349). Details of the synthesis are reported elsewhere [47]. A PAlC
composition with Al/Si= 1 was also prepared and used only for the MAS NMR
study. PAlCs were subsequently heat-treated for 1 h at 300°C in Ar flow in a silica
tube furnace in order to ensure that all the starting pre-ceramic polymers were treated
at the same maximum temperature. In some cases, more than one batch of PAlC with
the same AljSi ratio was prepared; however, every batch was studied individually. For
comparison purposes, partially cross-linked, pure polycarbosilane was also prepared.
In this case, however, the PCS was cross-linked at 400°C for 1h since polymers pro-
cessed at lower temperatures showed melting and bloating during pyrolysis.
The pre-ceramic polymers were ball-milled in an agate mortar and the powders
with a size lower than 63 pm were used to prepare green compacts by cold uniaxial
pressing into a steel mold 60 mm in length and 8 mm wide. Green samples (approxi-
mately 60 x 8 x 5 mm3) were accordingly obtained with a pressure of 143 MPa
applied for 30min (see Fig. 6). After complete release of the molding pressure
and extraction of the sample from the mold, the length of the sample was measured
and compared with the length of the mold in order to evaluate the relative size
increase, usually designated as ‘springback’ and defined as:
IC - Id x 100,
S=-----
ld
where Id is the mold length and I, is the corresponding compact length. This param-
eter is a function of many factors, such as the powder properties, the compacting
pressure and time, and the die rigidity. However, in our experiments the forming
process was performed always with the same conditions. Thus, the springback is
directly related to the ability of the pre-ceramic polymers to be permanently
deformed under load and gives a measurement of the cross-linking degree of the
pre-ceramic network. Indeed, highly cross-linked polymers behave as purely elastic
materials and display a higher value of springback compared to polymers with a
lower cross-linking degree that show a permanent deformation under pressure,
ultimately leading to a lower S value. The green components were also characterized
6.4 Results 453
Figure 6. Green and corresponding ceramic components obtained from PAlC with AljSi = 0.05
ratio.
measuring their open porosity (P)by mercury infiltration (Carlo Erba equipment)
and their bulk density, Pb. The structure of the partially cross-linked polymers
was investigated by recording 29Si CP MAS NMR spectra on a MSL 400 spec-
trometer operating at 79.6MHz. Details of the experimental set up are reported
elsewhere [47].
Monolithic, dense and crack-free SiAlOC ceramic components were obtained (see
Fig. 6) after a pyrolysis process (1 "C min-' , Ar flow, 100 cm3min-') at 1000°C for
1 h. The weight and the length of the samples were recorded before and after the heat
treatment in order to measure the associated weight loss and the linear shrinkage.
The microstructure of the ceramic samples was investigated measuring the open
porosity, P,and the bulk density, &,.
Modulus of rupture, MOR. was obtained on polished samples by four point bend
tests (4 x 20 mm) with 1 mm min-' deformation rate using Instron equipment. The
elastic modulus, E , was measured at room temperature, by the acoustic resonance
method using a self-assembled apparatus and, at high temperature, from the
stress-strain curve obtained in the four point bend tests.
6.4 Results
6.4.1 Characterization of the Pre-ceramic Precursors
The cross-linking mechanism of polycarbosilane with various metallic alkoxides has
been already published [45]. The Si-H bonds of the PCS react with the M-OR
groups (M=Ti, Al, Zr, Ta) of the metallic alkoxide to give Si-0 bonds that
probably form siloxane bridges between the polycarbosilane chains. The 29Si CP
MAS NMR spectra recorded on the starting PCS and on two PAlCs with an AljSi
ratio of 0.25 and 1 are presented in Fig. 7. The spectra show two peaks at 0p.p.m.
(Sic4 sites) and at - 17.6p.p.m. (SiC3H sites), characteristic of the polycarbosilane
backbone and a third peak, present only in the Al-containing compositions, at
10 p.p.m. that is assigned to SiC30 sites. By increasing the A1 content, a decrease of
AVSi = 0.25
50 0 -50 -100
Chemical Shift (ppm)
Figure 7. 29Si CP MAS NMR spectra of the partially cross-linked pre-ceramic polymers.
the component at -17.6p.p.m. can be noticed and an increase of the peak at

10p.p.m., in agreement with the proposed mechanism. Thus, these results suggest
an increase in the number of the siloxane bridges, that is, of the cross-linking
degree, with the amount of aluminum alkoxide.
6.4.2 Characterization of the Pre-ceramic Components

Bulk density of all the pre-ceramic bodies was ~1 g ~ m - ~Figure . 8 shows the
evolution of open porosity and elastic recovery as a function of the Al/Si ratio.
For all the PAlC precursors P falls in a narrow range between 9% and 12%. A
higher value ( P = 17%) has been observed for the green bodies obtained from
pure PCS. These low values of open porosity are due to the plastic deformation
during pressing of the partially cross-linked precursors. Moreover, mercury infiltra-
tion experiments always resulted in a bimodal distribution of the open porosity, in
agreement with the published literature [56]. Accordingly, the finer pore fraction
below 10 nm can be assigned to flat pores between two particles whereas the coarser
one, with a maximum in the range 8&600nm, is due to pores at triple point
junctions. The size of the pores at the triple point junctions is ~ 6 0 0 n mfor the
pure PCS precursor and increases with the A1 load, that is with the cross-linking
degree, from a minimum value of ~ 8 nm, 0 observed for the AljSi = 0.05 composi-
tion to 300 nm for the Al/Si = 0.25 samples.
The elastic recovery (Fig. 9) shows an evident relation with the aluminum content:
the lowest S values being observed for the samples with AljSi = 0.05 ( S M 0.4%) and
6.4 Results 455
18
A
8
W
s
14
c,
.d
v1
g 10
0
pc
6 t. 1 .. I . . I .
0.3
# . . . I . . I
0 0.1 0.2
Al/Si (Atomic ratio)
Figure 8. Evolution of the porosity of the pre-ceramic samples as a function of the AljSi ratio.
the highest ones for the samples with Al/Si=0.25 (1% 5 S 5 1.5%). The S value
measured for the green compact derived from pure PCS is 1.3%, close to the
values measured on PAlCs with the maximum aluminum load. These results suggest
that, by increasing the amount of aluminum alkoxide used for the preparation of the
modified polycarbosilanes, the cross-linking degree advances, in agreement with the
NMR results, and correspondingly increases the elastic recovery of the pre-ceramic
components. The high S value measured for the PCS green bodies indicates that the
thermal cross-linking process used in this case leads to a polymer with a high cross-
linking decree which exhibits a highly elastic behavior during pressing.
6.4.3 Characterization of the Ceramic Components

All the pyrolyzed samples were XRD amorphous. Weight loss ( ~ 2 weight-%)
8 and
linear shinkage ( ~ 2 9 % )associated to the pyrolysis treatment were found to be
0 0.1
0.2
i
1
2
3
4
T
0.3
Figure 9. Evolution of the elastic recovery of the pre-ceramic samples as a function of the AljSi
ratio.
.cd
200
5
W
d 100
0
E
0 0.1 0.2 0.3
Figure 10. Evolution of the flexural strength, modulus of rupture (MOR), of the ceramic samples
pyrolized at 1000"C, 1h, with the AljSi ratio.
independent of the A1 content. Values of density, porosity, flexural strength, and elastic
modulus measured on the various ceramic components are reported in Table 1.
The open porosity of the PAlCs-derived ceramic samples shows a weak depen-
dence from the amount of aluminum and falls in the range ~ 8 - 1 3 % .It increases
up to ~ 2 0 %for the PCS-derived component. Such low values of open porosity
have been already reported in the literature for similar polymer-derived ceramics
[56] and indicate that this processing route is suitable for the fabrication of dense
(up to 92% of theoretical density) covalent ceramics at low temperature. The
porosity values measured for the ceramic products are comparable with the starting
pre-ceramic samples. Indeed, results of the mercury infiltration experiments indicate
that the pyrolysis process leads to the removal of the finer pore fraction (below
10 nm) leaving almost unchanged the amount and the size of the coarser pores.
The evolution of the MOR and E with the Al/Si ratio is reported in Figs 10 and
11. Surprisingly, contrary to the behavior of all the other studied parameters, these
Table 1. Results of the microstructural and room temperature mechanical characterization per-
formed on the ceramic samples pyrolyzed at 1000°C for 1 h.
AljSi Batch Bulk Porosity Flexural strength Elastic

(Atom number density P (Yo) (MOR) modulus
ratio) Pb (gcm-3) (MPa) (GPa)
PCS I 1.82 f 0.01 *

19.0 0.3 40 f 1 55 f 3
0.05 I 2.15 f 0.03 7.8 f 0.5 181 f 15 117 f 1
0.05 I1 2.19 f 0.01 8.0 f 0.5 214 f 10 136 f 1
0.10 I 2.14 f 0.04 10.3 f 0.5 190 f 38 123 f 9
0.10 I1 2.24 f 0.01 7.5 f 0.4 177 f 16 177 f 7
0.10 111 2.12 z t 0.06 12.4 f 0.5 - 112 f 5
0.10 IV 2.13 f 0.05 - - 122 f 3
0.15 I 2.15 f 0.05 10.6 f 1.3 101 f 7 78 f 8
0.25 I 2.11 f 0.05 11.9 f 1.5 63 f 2 56f 1
0.25 I1 2.12 f 0.05 ~ - 90 f 1
6.4 Results 457
160
9 batch 2
2 120 A batch 3
k
W
u
W 80
40
0 0.1 0.2 0.3
Figure 11. Evolution of the elastic modulus, E , of the ceramic samples pyrolized at lOOO"C, 1 h, with
the Al/Si ratio.
properties show a strong dependence from the composition of the starting polymers
with up to 100% of variation. The highest values of MOR (up to ~ 2 5 MPa)
0 and of
E (up to ~ 1 4 0 G P ahave
) been recorded for the samples with AljSi ratio of 0.05,
then the increase of the A1 load leads to a decrease of both mechanical parameters.
PCS-derived samples display very low values of MOR and E similar to the samples
with the highest AljSi ratio.
6.4.4 Mechanical Characterization at High Temperature

High temperature mechanical characterization was performed on the PAlC com-
positions with AljSi = 0.05 and 0.1. The elastic modulus has been measured in air
at various temperatures between 800 and 1400°C by four point bend tests
(40 x 20 mm) with a 0.2 mm min-' deformation rate. E has been calculated, through
the standard equation valid for rectangular bars, by measuring the displacement
with a LVDT. All the ceramic samples obtained by pyrolysis at 1000°C have been
pre-annealed at 1400°C for 1 h in argon atmosphere, before the high temperature
tests. This treatment lead to the crystallization of microcrystalline PSiC with a
minor amount of clSiC. Aluminum atoms are present both as a solid solution of
A120C in clSiC and in the residual amorphous phase.
The results of the elastic modulus measurements are reported in Table 2. The
evolution of the elastic modulus with the temperature for the Al/Si = 0.05 composi-
tion is reported in Fig. 12. It can be noticed that the elastic modulus is constant up to
Table 2. Results of the elastic modulus, E , measurements performed at high temperature, in the
range 800-1400°C.
~~~~~~ ~
T ("C) 20 800 900 1000 1100 1200 1300 1400
AljSi = 0.05 129 f 6 118 f 8 - 118f7 115f6 104f6 9 4 f 4 81 f 2

AljSi =0.1 126 f 5 107 f 1 108 f 8 104f 7 96 f 12 85 f 13 72 f 7 -
200 1 ' ' ' ' 1 ' ' ' ' ' " 1 ' 1 ' ' ' ' ' ' ' ' ' ' ~ ' '
1 5 0-
-
n
Q i I T
0
100-
T +
* r
-
m -
W
50- -
0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1100°C. Above this temperature the slight decrease of E is probably related to the
viscous deformation of the amorphous silicon oxycarbide phase that shows a glass
transition temperature of ~ ~ 1 3 0 0 [76].
°C
6.5 Discussion
In this work, crack-free, dense (up to 92% theoretical density), amorphous SiAlOC
ceramic components with good mechanical properties have been obtained at low
temperature ( T = 1000°C) by direct pyrolysis of pre-ceramic green bodies.
Elastic modulus and flexural strength show a clear dependence from the A1
content and the maximum values were recorded for the composition low in Al.
These behaviors cannot be ascribed to a different weight loss or linear shrinkage
during pyrolysis due to the constancy of these parameter with the polymer com-
position. Thus, in order to explain this experimental data we tried to correlate the
elastic modulus with the amount of porosity measured by mercury infiltration.
The dependence of E on the porosity shows a high level of scatter (see Fig. 13).
As an example, ceramic components with a porosity level of about 10% display
elastic modulus values from z60GPa to =120GPa, with a 100% variation. These
findings suggest that other microstructure parameters, apart from the amount of
open porosity must play a crucial role in governing the Young modulus of the
studied polymer-derived ceramics. Recently some studies have been published
that take into account the role of the contact area on which mechanical stresses
are really acting between the particles [77]. According to these models, samples
with the same P but with different E values must have a different amount of contact
area between the ceramic grains.
In our samples, the interface between the grains (derived by the decomposition of
the initial pre-ceramic particles) in the ceramic component can be high only if the
6.5 Discussion 459
t
5 10 15 20 25
Porosity, P (YO)
Figure 13. Elastic modulus of the samples pyrolyzed at 1000°Cfor 1 h plotted against the porosity of
the ceramic components.
contact area between the pre-ceramic powders in the pressed green body is also high.
In fact, during the pyrolysis, the development of the ceramic bond between the indi-
vidual particles takes place with the formation of primary chemical bonds via con-
densation reactions such as:
-Si-CH3 + H-Si- +S - C H 2 - S i s + H2
ESi-CH, + H3C-Si- +=Si-CH2-Si- + SH4.
These reactions can occur only if the reacting groups are close enough for the
formation of -Si-CH2-SiE bridges, that is, if they belong to the interface between
two pre-ceramic grains. The extension of the contact surface between the partially
cross-linked polymeric grains should be high for the compositions that can be
easily plastically deformed under pressure, with a low cross-linking degree. In
order to test this hypothesis, the elastic modulus has been plotted against the elastic
recovery, that is a parameter directly related to the cross-linking degree of the pre-
ceramic polymers (see Fig. 14). In this case a very good relation between E and S can
be observed suggesting that, actually, the cross-linking degree is a vital parameter
for tailoring the mechanical properties of polymer derived ceramics. In particular,
the cross-linking degree must be high enough to prevent bloating but also low
0 0.5 1 1.5 2
Elastic Recovery, S (YO)
Figure 14. Elastic modulus of the samples pyrolyzed at 1000°C for 1 h plotted against the elastic
recovery of the green components.
enough to allow viscous deformation during pressing with the formation of

extended contact area between the polymeric powders, which in turns gives rise to
ceramic components with high mechanical properties.
6.6 Preparation and Characterization of SiAlON

Ceramics by Pyrolysis in Reactive Atmosphere
6.6.1 Experimental
The green pre-ceramic samples were nitrided in NH3 flow (100 cm3min-') in a silica
tube furnace at different temperatures up to 1000°C for 1 h with a heating rate of
1"Cmin-I. The ceramics obtained at 1000°C were subsequently fired in N2 flow
in a graphite furnace at various temperatures up to 1500°C. At the end of the pyr-
olysis treatments the weight loss and the specific surface area (BET method) was
recorded. The skeleton density was measured with a helium picnometer while the
porosity was obtained by mercury infiltration tests. In order to follow the sintering
process, dilatometric tests and differential thermal analysis experiments were per-
formed in flowing nitrogen at 5°C min-' up to 1550°C on samples already pyrolized
under ammonia at 1000°C. Structural characterization of the final PSiAlON ceramic
was accomplished by X-ray diffraction analysis on powdered samples. Micro-
structural investigation of the ceramic phase was performed by SEM on polished
or fracture surfaces. Mechanical characterization, namely hardness (Hv), elastic
moduls (E), and fracture toughness (&) was carried out by means of Vickers
and Knoop indentation tests performed at 40-60 and 10 N, respectively.
6.7 Results and Discussion

The nitridation under ammonia of the yellowish PAlC compacts leads to the for-
mation at 1000°C of white, crack-free amorphous SiAlON pellets with a linear
shrinkage of 20%. After an additional pyrolysis step under flowing nitrogen at
1500°C the amorphous samples convert into microcrystalline PSiAlON monolithic
components with a further linear shrinkage of 20%.
A weight loss of 25% has been observed from room temperature up to 400°C
(Fig. 15). It is due to the decomposition of the organic component of the modified
aluminum alkoxide, since PCS chains are stable up to this temperature. This process
leads to a corresponding increase of the specific surface area up to 350 m2 g-'
(see Fig. 16), without any appreciable variation in the apparent density. In the
temperature range 400-600°C the nitriding of PCS occurs with a further weight
loss of up to ~535%and an additional increase of surface area to 500 m2 g-I. In
this case, however, the formation of a three-dimensional silicon nitride network
6.7 Results and Discussion 46 1
/I
0
- I \ I
-10 *
2.
z
(P
-20 -
-.'
Adensity
0
-E
fn
fn
/.
-30
-
-40
0 400 800 1200 1600
T ("C)
Figure 15. Processing of SiAlON ceramics by nitridation of PAlC precursors: evolution of the
density and weight loss with the firing temperature.
leads to an increase in density to 1.4 gcmP3,indicating that a sintering mechanism is

active in this temperature range. Above 600°C the weight of the samples does not
change up to the maximum temperature reached in this study (1 500°C). However,
its apparent density remains constant (1.4 g cmp3)up to 12OO0C,then increases dra-
matically above 13OO0C,approaching a value of 2.5 g cmp3at 1500°C.The evolution
of the surface area with the pyrolysis temperature is similar to the density, that is, it
remains constant (500m2g-') up to 1200°C and then decreases rapidly above
1300°C, being close to zero at 1500°C. The dilatometric and DTA tests performed
I -> N2
600
NH3 <-
N
.ol
E
Y
m
E!
m
.-
0
c0
al
n
a
0 400 800 1200 1600
T ("C)
Figure 16. Processing of SiAlON ceramics by nitridation of PAlC precursors: evolution of the
specific surface area with the firing temperature.
0.25 4
.
0
-I
-I
-0.05
LIL" 1
-0.15 1 I
T ("C)
Figure 17. Processing of SiAlON ceramics by nitridation of PAlC precursors: differential thermal
analysis (DTA) and dilatometry of a monolithic amorphous sample obtained after pyrolysis in
ammonia at 1000°C.
on samples previously nitridated at 1000°Care reported in Fig. 17. The dilatometric

curve clearly shows that the shrinkage of the amorphous AlON compacts starts
around 1300°C and it is complete before 1500°C. DTA analysis indicates that, in
the same temperature interval, a strong exothermic effect, related to the crystalliza-
tion of PSiAlON phase, is active. Indeed, the crystallization of the sample has been
verified by XRD analysis and the phase was identified as PSi2.5A10,500,5N3.5, with a
mean particle size of 40 nm. Thus the observed sintering process has to be related
with the amorphous to crystalline transformation that occurs via the formation of
a liquid phase.
Monolithic PSiAlON ceramics obtained at 1500°C are 85% dense, according to
the porosity measurements. The results of the mechanical characterization per-
formed on samples pyrolyzed at 1300°C and 1500°C are reported in Table 3.
This work his clearly shown, for the first time, that the direct pyrolysis of partially
cross-linked pre-ceramic bodies can be applied for the processing of monolithic
ceramics even when a reactive pyrolysis process is used.
Table 3. Measured values of hardness (H,), elastic modulus ( E ) and fracture toughness (KIc) for
amorphous (1300°C) and microcrystalline (1500°C) DSiAlON pellets.
Temperature ("C) H, (GPa) E (GPa) KIc (MPa m112)
1300 2.0 f 0.3 28 !i 12 1.2 f 0.5

1500 11.0 f 1.4 198 f 28 4.7 f 0.8
6.9 Experimental 463
6.8 Processing of Thin Ceramic Films

Processing is always a key step in the development of a material or a device. Param-
eters such as the heat treatment temperature, the heating atmosphere, and the
dwelling time greatly influence the microstructure and composition of the material.
Moreover, thin films are more likely to be affected by the processing conditions than
a monolithic body, so more care should be taken to avoid the presence of undesired
impurities, such as oxygen.
This work reports the development of a polymeric/sol-gel route for the deposition
of silicon carbide and silicon oxycarbide thin films for applications such as heat-,
corrosion-, and wear-resistant coatings, coatings on fibers for controlling the inter-
action with the matrix in ceramic matrix composites, or films in electronic and
optoelectronic devices. This method, in which the pre-ceramic films are converted
to a ceramic coating either by a conventional high temperature annealing or by
ion irradiation, is alternative to conventional methods such as chemical or physical
vapor deposition (CVD, PVD), molecular beam epitaxy, sputtering, plasma spray,
or laser ablation, which are not always practical or cost efficient.
Using a conventional high temperature annealing (500-12OO0C), important
characteristics like the degree of crystallization, the crystallite dimension, the
presence of hydrogen and the CjSi ratio can all be varied in a fairly large range of
values. However, low-temperature melting substrates cannot be employed, and if
the precursors contain carbon, it is difficult to obtain pure ceramics due to the reten-
tion of a noticeable amount of hydrogen (because of the strength of the single C-H
bonds) and of free carbon. The use of ion implantation allows for the conversion to
ceramics through the breaking or three-dimensional cross-linking of bonds in the
precursor films, accompanied by the radiolysis of hydrogen atoms and alkyl
groups. Several advantages might result from the use of this room temperature
processing technique with respect to conventional annealing. In particular ceramic
coatings can be obtained on non-refractory substrates or as components of
integrated circuits for electronics without degrading metallic overlayers. Other
interesting potentialities are the introduction of dopants and the patterning of the
coating by using masks or by focusing the ion beam.
6.9 Experimental
S i c and SiOC films were produced by spinning a solution containing the precursor
on (100) silicon waters or silica slides. Detailed experimental and analysis procedure
is reported elsewhere [59,60]. The pre-ceramic polymers used were: for S i c films,
polycarbosilane (PCS, Dow Corning), or allylhydridopolycarbosilane (HPCS,
Starfire Systems, Watervliet, NY). For SiOC films, methylhydroxylsiloxane or
phenylmethylhydroxylsiloxane (SR350 and SR355, respectively, General Electric
Silicone Products, Waterford, NY). In addition, SiOC films were produced also
from different alkoxides: methyltriethoxysilane or phenyltriethoxysilane (MTES
and PTES, respectively, Aldrich, Milan), or from a mixture of methyldiethoxysilane

and triethoxysilane (DH and TH, respectively, ABCR, Karlsruhe), with a DH/TH
molar ratio of 1/2. The precursors were chosen in order to achieve different CjSi
ratios in the final ceramic product.
The conversion from precursor to ceramic was obtained either by conventional
heating in a tube furnace (heating at temperatures in the range 200-1200°C in
different inert atmospheres: regular grade Ar (99.99%), ultra high purity Ar
(99.999%), vacuum (1 x 10-6mbar)), or by irradiating the films with different
doses of ions (C, Au, He). Part of the irradiated specimens were subsequently
heat treated at 1000°C for 1 h in vacuum. After conversion, the average thickness
for the films was in the range 200-500 nm.
The composition and microstructure of the films was investigated using several
techniques, such as Fourier transformed infrared spectroscopy (FTIR), secondary
ion mass spectroscopy (SIMS), Rutherford back-scattering spectrometry (RBS), elas-
tic recoil detection analysis (ERDA), nuclear reactions analysis (NRA), glancing angle
X-ray diffraction (GA-XRD), Raman spectroscopy, transmission electron micros-
copy (TEM), scanning electron microscopy (SEM), and atomic force microscopy
(AFM). Film thickness was measured by ellipsometry or using a tactile profilometer.
Some optical and mechanical properties of the films were also evaluated, using a UV-
vis spectrophotometer and a nano-indenter [51, 52, 59, 60, 78-80].
In this paper, only the results concerning SiOC films obtained from SR350 and
S i c films obtained from PCS will be reported.
6.10 Results and Discussion

6.10.1 Conventional Conversion Process: Annealing in
Controlled Atmosphere
When considering the thermal conversion, it was found that the processing gas, the
heating temperature, and the dwelling time have a strong influence on the structure
and the composition of the produced ceramic thin films.
In Fig. 18 are reported the FTIR spectra collected for SR350 and PCS films heat
treated in argon and vacuum. The peaks at about 1090cmpl (with the shoulder at
1190cm-') and at 460 cm-' are assigned to the Si-0 stretching and rocking modes,
respectively, while the peak at about 790cm-' corresponds to the S i x stretching
vibration. Pyrolysis in regular grade Ar results in coatings whose structure resembles
that of silica, and compositional analysis confirmed that little carbon was retained in
the SR350 and PCS films. It has to be emphasized that, in the case of PCS, oxygen
was not present in the as deposited coatings, so it can be concluded that oxygen was
introduced during the heat treatment. When UHP Ar gas was used, the com-
positional analysis showed that carbon was retained in both the SR350 and PCS
coatings (with a very high CjSi ratio in PCS films), and that oxygen contamination
was almost completely suppressed in the latter. X-ray diffraction analysis shows that
I " ' 1 ' ' -
SR350
I , . . l . . .
1600 1200 800 400 1600 1200 800 400
Wavenumbers / cm-' Wavenumbers / cm-l

Figure 18. Fourier transform transmission spectra for SR350 and PCS films heat treated in: (a) reg-
ular grade Ar, 1000°C; (b) ultra high purity Ar, 1000°C;(c) vacuum, 1000°C; (d) vacuum, 1200°C.
the structure of the PCS films remained amorphous, possibly due to the excess
carbon inhibiting crystallization or growth of Sic nanocrystals.
Performing the pyrolysis of PCS films at 1000°Cin vacuum, a structure character-
istic of nanocrystalline silicon carbide was obtained, but the compositional data
show that some oxygen contamination was present. Increasing the processing
temperature to 1200°C results in the increase of crystallization of PSiC (see later).
The SR350 films heat treated at 1200°C in vacuum exhibits a composition and
structure that can best be described as an oxycarbide glass, that is an amorphous
silicate material in which some of the Si-0 bonds are replaced by Si-C bonds. In
fact, the FTIR spectrum shows the persistence of Si-C bonds after heating, even
if this finding is less evident when heating at temperatures 11000°C.
It is thus clear that the processing atmosphere strongly affects the resulting
composition and microstructure of both SiOC and S i c coatings, and that vacuum
is the only environment which allows the synthesis of both silicon carbide and silicon
oxycarbide thin films [51].
In Fig. 19 is shown the influence of the processing temperature on the com-
positional and microstructural development of PCS films, heat treated in vacuum
for 1h. The peak at about 2100cm-' (see Fig. 19a) is attributable to the Si-H stretch-
ing; the peaks at 1400cmp' are due to the CH3 asymmetric in Si-CH3 whereas the
peak at 1250cm-' is due to the Si-CH, symmetric deformation. The peaks at 1355
and 1020cm-' are respectively assigned to the CH2 deformation and wagging of
Si-CH,-Si, and the peaks at 820 and 790cm-' are respectively assigned to the Si-
CH3 rocking and S i x stretching. The spectra show that, for heat treatments above
600°C, the absorption bands that are associated with organic species are eliminated,
indicating the conversion to a ceramic material of the pre-ceramic polymer.
2000 1600 1200 800 10 20 30 40 50 60 70 80

Wavenumbers / cm" 29 / Degrees
Figure 19. (a) Fourier transform transmission spectra and (b) glancing angle X-ray diffraction
(XRD) patterns for PCS films heat treated for 1 h in vacuum at different temperatures.
The narrowing and intensification of the Si-C absorption band located at

800 cm-' indicates an increase of the ordering of the structure, which corresponds
to a crystallization of the films. Indeed, glancing angle X-ray diffraction analysis
(see Fig. 19b), reveals that above 1000°Cthe PCS films develop into nanocrystalline
ceramic coatings. The compositional analysis showed that, with increasing pro-
cessing temperature, a strong decrease in the hydrogen concentration occurs [8 11.
In the intermediate pyrolysis temperature, just after the polymer-to-ceramic con-
version, there is a rather large oxygen contamination, which is 19.7 (at 800°C)
and 15.6at-% (at 9OOOC). This is believed to be due to the presence of reactive
sites in the films at this stage of the transformation, which are generated by the
cross-linking of the PCS polymer and the creation of Si radicals. With increasing
processing temperature, the oxygen contamination strongly decreases, and films
heated at 1100°C have a C:Si ratio of about 1, while the ones heated at 1200°C
appear to be silicon-rich. Raman analysis allowed to follow the evolution of the
free carbon (turbostratic C) contained in the Sic films: at elevated heating tempera-
tures (> lOOO"C), the free C phase is drastically reduced, and disappears completely
in films pyrolyzed at 1200"C, presumably being eliminated as CO and/or CH4. The
effect of heat treatment at 1200°Cvs. 1000°Cis primarily to enhance the evolution of
hydrogen and the more effective crystallization of PSiC at the higher temperature.
Combining compositional and microstructural data we can thus infer that the
presence of free carbon has a hindering effect on the development of PSiC crystal-
lites in the silicon carbide films [52]. With increasing processing temperature the
PCS-derived films approach the composition and microstructure of stoichiometric
polycrystalline 0 silicon carbide. Concurrent with the compositional and micro-
structural development, a shrinkage of the films occurs, which is of about 75%
1600 1200 800 400 26 28 30 32 34 36 38 40

Wavenumbers / cm-' 28 / Degrees
Figure 20. (a) Fourier transform transmission spectra and (b) glancing angle X-ray diffraction
(XRD) patterns for PCS films heat treated for different times in vacuum at 1000°C.
at 1200°C yielding a Sic coating 200 nm thick. No major defects (cracks, pits) are
visible in the films.
The effect of heating time on PCS films, pyrolyzed at 1000°C in vacuum, is
reported in Fig. 20. We can see that the crystallinity of the S i c coatings clearly
increases with increasing the isothermal treatment time. Accordingly, the composi-
tional analysis revealed that for heat treatments shorter than 6 h (at 1000°C), the
films are carbon-rich, whereas for longer dwelling times they become silicon-rich,
with a significant reduction in hydrogen and a lesser reduction in carbon and
oxygen. At the same time, Raman spectroscopy shows a strong reduction of the
free turbostratic carbon phase [52]. In Fig. 21 is reported the variation of the
average dimension of the PSiC crystallites in S i c films from polycarbosilane, as
a function of temperature and dwelling time (at 1000°C). The grain size seems
to be enhanced with respect to PCS-derived Nicalon fibers and PCS pyrolysis
residues that have been treated at similar temperatures. This is most probably
due to differences in oxygen contamination among the different PCS-derived
materials, and/or to their different processing histories (especially the kind of
annealing atmosphere). Again, concurrent with the transformations happening
in the material during the isothermal treatment, the film thickness decreases as a
function of the dwelling time.
6.10.2 Nonconventional Conversion Process: Ion Irradiation

In the experimental conditions adopted in these studies, the irradiating ions
penetrate through the entire thickness of the film, and stop into the substrate.
6 Polymer to Ceramic Transformation: Processing of Ceramic Bodies and Thin Films
-800 900 1000 1100 1200

Pyrolysis Temperature /"C Dwelling Time at 1000°C / h
Figure 21. Average PSiC crystallite dimension for PCS-derlved thin films as a function of: (a) pro-
cessing temperature and (b) dwelling time (heating in vacuum at lO0OT).
During this travel, the ions transfer energy to the atoms, which constitute the
coating material, by nuclear collisions and electronic excitations. As a result, a
change in composition, structure and density occurs, which depends on the product
of the fluence (which is proportional to the number of irradiating ions) by the linear
density of energy deposited by each impinging ion. In particular, besides merely
promoting densification, ion irradiation has a much higher efficiency than UV or
electrons for breaking all bonds and creating three-dimensional cross-links in
polymers (organic or inorganic), and especially for radiolyzing hydrogen and
alkyl groups. One has to bear in mind that the number of irradiated ions amounts
to just a few monolayers, and thus the properties of the substrates are not strongly
affected by this processing method (Si is amorphized within the ion range).
In Fig. 22 are reported the FTIR spectra for SR350 and PCS films, submitted to
irradiation with different doses of 3 MeV Au ions, or annealed, or irradiated and
then annealed (1 h at 1000°C in vacuum). We can easily see that with increasing
ion dose the bands relative to the organic moieties (see previous discussion) show
a progressive smearing out, and are replaced by broad bands of conjugated bonds
when the deposited energy reaches values of ~ 4 eV.0 Similar behavior and threshold
of collapse into a three-dimensional network are generally found for organic poly-
mers, and in our case indicate the actual occurrence of the polymer-to-ceramic
transformation. With respect to spectra of annealed films, the bands of conjugated
bonds are significantly broader and shifted toward lower wavenumbers (of
=2Ocm-' for the main Si-C and Si-0-Si modes) in spectra of films converted by
irradiation. An oxygen sub-stoichiometry in silica films or a compressive state of
stress can induce such a shift. In our case, this is more probably due to a higher
degree of bond distortion because of the combination of free radicals at random,
Irradiated
+ Annealed
Annealed
=
5x10''
5x10''
As deposited
2000 1600 1200 800

4 As deposited
2000 1600 1200 800
Wavenumbers I cm-' Wavenumbers I crn"
Figure 22. Fourier transform spectra of SR350 and PCS films irradiated with different doses of
3 MeV Au ions, or annealed (1 h lOOO"C), or irradiated with 5 x lOI5 Au' at cm-2 and then annealed
(1 h lO0OT).
and to some atomic disorder in Au-irradiated films. Associated with the structural
transformation in the specimens, there is a compositional change, which depends on
the dose and the type of irradiating ion (see Table 4). At high doses, the irradiation
process allows to obtain ceramic films whose composition is very similar to that of
the starting precursor, apart from a large decrease in hydrogen content. No major
contamination (oxygen) is introduced, and carbon is retained in the structure,
whereas conventional annealing is shown to strongly alter the Si:C ratio (and to
introduce some oxygen in the PCS films) [58,59,60]. Under irradiation, the films
also are subjected, without cracking, to a compaction that can be as high as 65%.
After irradiation, all the films are X-ray amorphous, while PSiC nanocrystallites
develop in PCS films, even after they have been irradiated, during the annealing
treatment.
SiOC and Sic films obtained by ion irradiation contain some carbon clusters (free
C). Ion irradiation, besides provoking structural and compositional chances in the
films, affects the hybridization state of the carbon clusters, as can be seen in Fig. 23,
which reports some Raman spectra collected on irradiated, annealed, irradiated and
then annealed SR350 and PCS samples. The presence of a single Raman peak in the
case of both SR350 and PCS irradiated films, indicates that the carbon clusters
possess a high degree of sp3 hybridization, which is typical of diamond-like films.
The hybridization state of the carbon clusters becomes a little more graphite-like
in films irradiated and then annealed (according to the degree of splitting in two
components of the Raman carbon peak), and is absolutely graphitic (sp2 hybridiza-
tion, turbostratic carbon) in precipitates formed during annealing of as-deposited
films [58].
Table 4. Composition, density, and hardness of the studied films.

Film Treatment Composition (Si,C,O,H,) Density Hardness
(g ~ m - ~ ) (GPa)
X J
PCS As-deposited 1.90 f 0.1 0.08 f 0.02 4.50 f 0.2 1.10 0.28
He 1 x 10l6 1.90 0.92 1.oo 2.14 17.1
c 1 x 10I6 1.85 0.06 0.60 2.15 29.0
Au 1 x 10" 1.75 0.06 0.62 2.01 19.4
Annealed 0.80 0.20 0.25 2.45 19.0
Irradiated Au + annealed 1.65 0.05 0.04 2.41 34.6
SR350 As-deposited 1.00 4~0.05 1.55 f 0.05 2.9 f 0.1 1.25 0.23
He 5 x lot6 1.oo 1.40 0.29 2.05 21.0
c 1 x 10'6 1.oo ISO 0.37 2.16 18.5
Au 1 x 10'' 1.oo 1S O 0.55 2.06 15.5
Annealed 0.60 1.28 0.52 2.08 7.0
Irradiated Au + annealed 0.98 1.45 0.02 2.40 18.5
The carbon clusters afford some interesting optical properties to the SiOC and
Sic films [79], but are also responsible for the high hardness measured on irradiated
films (see Table 4). This is at least two times larger than that of conventionally
annealed specimens, mainly because of the previously mentioned diamond-like
nature of the C precipitates in the amorphous SiOC or Sic ceramic matrix of
irradiated specimens. The effect depends on the type of irradiating ion, but also
on the nature of the side groups of the polymeric chain (CH3 vs. C6H5) [59,60].
The annealing of irradiated films does not seem to affect much their hardness,
1"'""""""l I * ' ' I ' " I ' ' '
Irradiated Irradiated PCS
I . . , I . . . I . . .
1000 1200 1400 1600 18001000 1200 1400 1600 1800
Wavenumbers/ cm-l WavenumbersI cm-'

Figure 23. Raman spectra of SR350 and PCS films irradiated with 5 x 10153 MeVAu' atcm-', or
annealed (1 h 1000°C), or irradiated with 5 x 10"Au+ atcm-2 and then annealed (1 h lOOWC).
6.10 Results and Discussion 47 1
Figure 24. Profilometer trace and AFM image of a -250 pm-wide strip realized by ion irradiation
(1 x loL5500 keV C+ at cm-’) in a PCS film.
despite the partial change in hybridization state of the C clusters, probably because
this effect is compensated by the further hydrogen removal and compaction.
Another interesting consequence of the ion irradiation of polycarbosilanes and
polysiloxanes is that the resulting films are thermochemically very stable towards
high temperature annealing, as shown by compositional analysis (see Table 4).
Indeed, carbon atoms having already lost their hydrogen neighbors can no longer
evolve (as CO or C,H,) and leave Si atoms with dangling bonds able to react
with O2 or H 2 0 of the atmosphere, as instead happens during annealing of non-
irradiated films [79]. Thus, the postirradiation annealing treatment results mainly
in the nearly complete elimination of hydrogen from the ceramic films. Finally,
ion irradiation promotes a good adherence of the coatings to the substrate, due
to some interfacial mixing.
The capability of patterning the films using ion irradiation, which can be quite
important when fabricating actual devices, is illustrated in Figs 24 and 25. Figure 24
shows the profilometer trace and the AFM image of a strip realized in a PCS film
by ion irradiation (1 x lOI5 500 keVC+ at cm-*) using a 250 pm mask, while Fig. 25
Figure 25. Profilometer trace and AFM image of a -250 pm-wide strip realized by ion irradiation
(1 x 500 keV C+ at cm-2) in a PCS film, after dissolving the non-irradiated part of the film.
shows the same specimen after the nonirradiated part of the film was removed by dis-
solving it in monochlorobenzene. It is evident the ~ 4 8 % compaction of the film due to
the irradiation, as well as the rather good lateral resolution of the strip profile. The high
edges present in the strip after dissolving the non-irradiated part of the film (Fig. 25) are
probably due to an irradiation-induced cross-linking of the PCS polymer at the inter-
face between the irradiated and non-irradiated areas, which makes it insoluble, but is
not accompanied by any compaction.
6.11 Conclusions
Polymer pyrolysis appears to be a very promising processing route to advanced
covalent ceramics. This process allows the formation of new amorphous covalent
ceramic phases in the general system Si-M-C-N-0 (with M = B, Ti, Al, Zr . . .)
which show exceptional oxidation and creep resistance at high temperature up to
1500°C.
Another advantage of this process is its high flexibility. Indeed, the composition,
structure (amorphous or crystalline), microstructure, and properties of the ceramic
material can be controlled and adjusted by varying many different parameters such
as: the composition and architecture of the pre-ceramic polymer, the amount and
nature of the filler, the cross-linking step, and the pyrolysis parameters (atmosphere,
heating rate, final temperature). Also, nonconventional heating systems such as
laser or microwave heating or even athermal conversion processes like ion bombard-
ment can be efficiently used.
Polymer pyrolysis is also very flexible regarding the type of component that can be
obtained: fibers, films, membranes, foams, ceramic/ceramic joints, ceramic matrix
composites, and monolithic bodies can all be fabricated. The present study has
been focused on the formation of monolithic component and thin ceramic films.
Dense Sic-based monolithic components with MOR values up to ~ 2 5 MPa 0
and E values up to ~ 1 4 0 G P ahave been fabricated from partially cross-linked
polyaluminocarbosilanes, PAlC, by pyrolysis at 1000°Cin a controlled atmosphere.
The cross-linking degree of the pre-ceramic polymers was controlled by varying the
amount of aluminum alkoxide reacted with the PCS. Among the many investigated
parameters (weight loss and linear shrinkage during pyrolysis, porosity of the pre-
ceramic and ceramic components) only the cross-linking degree was shown to really
play a key role in the development of ceramic materials with elevated mechanical
properties. This result has been related to the ability of the pre-ceramic powders
to be permanently deformed under pressure leading to an extended contact surface
both between the green particles and between the ceramic grains.
Defect-free Sic and SiOC ceramic thin films were obtained by heating in inert
atmosphere or by ion irradiation. The composition and microstructure of the
coatings can be varied by changing processing conditions (heating atmosphere,
maximum heating temperature, dwelling time, type and dose of irradiating ions).
The segregation of C clusters during ion irradiation affords interesting optical
(luminescence) and mechanical (high hardness) properties to the films.
References 473
Acknowledgments
Gian Domenico Soraru would like to thank Prof. J. D. Mackenzie, who introduced
him to the field of polymer-derived ceramics, Dr. Florence Babonneau for the MAS
NMR study performed in the framework of a very effective, 10-year long collabora-
tion, and the contribution of his students and colleagues: Dr. Alberto Ravagni, Ing.
Massimo Mercadini, Ing. Fabiano Vulcan, Ing. Alberto Chiari, Ing. Marcello
Beber, Ing. Paolo Bosetti, Dr. Vincenzo Sglavo, Dr. Antonella Glisenti. MURST
is also acknowledged for the 40% financial support.
Paolo Colombo would like to thank Prof. C. G. Pantano for the constant support
and encouragement in pursuing research in the field of polymer-derived ceramics.
He also gratefully acknowledges the fundamental contribution of Dr. J. C. Pivin,
who performed the ion irradiation experiments as well as many characterizations
on the specimens.
References
1. J. Lipowitz, Cerum. Bull., 1991, 70, 1888-1894.
2. Y. D. Blum, R. M. Platz, and E. J. Crawford, J . Am. Cerurn. Soc. 1990,73, 170-172.
3. R. Riedel, Advanced ceramics from inorganic polymers, in Materials Science and Technology,
Vol. 176, Processing of Ceramics, R. J. Brook (Ed.), VCH, Weinheim, 1996, pp. 1-50.
4. J. Bill and F. Aldinger, Precursor-derived covalent ceramics, in Precursor-Derived Ceramics,
J. Bill, F. Wakai, and F. Aldinger (Eds.), Wiley-VCH, Weinheim, 1999, pp. 33-51.
5. S. Yajima, J. Hayashi, and M. Omori, Chem. Lett. 1975, 931-934.
6. G. D. Sorard, F. Babonneau, and J. D. Mackenzie, Structural concepts on new amorphous
covalent solids, J . Non-Cryst. Solids, 1988, 106, 25G261.
7. V. Belot, R. J. P. Corriu, D. Leclercq, P. H. Mutin, and A. Vioux, Thermal redistribution
reactions in crosslinked polysiloxanes, J . Polym. Sci.: Part A: Polym. Chem. 1992,30,613-623.
8. R. Campostrini, G. D’Andrea, G. Carturan, R. Ceccato, and G. D. Sorard, Pyrolysis study of
methyl-substituted Si-H containing gels as precursors for oxycarbide glasses, by combined ther-
mogravimetric, gas chromatographic and mass spectrometric analysis, J . Muter. Chem. 1996,6,
585-594.
9. F. Babonneau, J. Livage, G. D. Sorard, G. Carturan, and J. D. Mackenzie, SiC/TiC ceramics
via polymer route: a structural investigation. New J . Chem. 1990, 14, 539-544.
10. F. Babonneau, G. D. Sorard, and J. D. Mackenzie, 29Si MAS-NMR investigation of the
conversion process of a polytitanocarbosilane into SiC/TiC ceramics, J. Muter. Sci. 1990, 25,
36643670,
11. G. D. Sorard, A. Glisenti, G. Granozzi, F. Babonneau, and J. D. Mackenzie, The pyrolysis
process of a polytitanocarbosilane into SiC/TiC ceramics: an XPS study, J . Muter. Res.
1990, 5, 1958-1962.
12. G. Granozzi, A. Glisenti, and G. D. Sorard, XPS characterization of mixed carbides obtained
from polymer precursors, Muter. Rex Soc. Symp. Proc., 1990, 180, 81 1-814.
13. C. Laffon, A. M. Flank, P. Lagarde, M. Laridjani, R. Hagege, P. Orly, J. Cotteret, J. Dixmier,
J. L. Miquel, H. Hommel, and A. P. Legrand, Study of Nicalon-based ceramic fibres and
powders by EXAFS spectrometry, X-ray diffractometry and some additional methods,
J . Muter. Sci. 1989, 24, 1503-1512.
14. F. Babonneau and G. D. Sorard, Spectroscopic characterization of the nitridation process of
polymeric precursors to Si-M-N-0 systems (M = Ti, Zr, Al), Muter. Res. SOC.Symp. Proc.
1993, 287, 239-244.
15. B. A. Bender, R. W. Rice, and J. R. Spann, Transmission electron microscopic characterization

of ceramics formed by pyrolysis of organometallic polymer precursors, J . Am. Ceram. SOC.
1987,70, C58-C60.
16. H.-J. Kleebe, D. Suttor, H. Miiller, and G. Ziegler, Decomposition-crystallization of polymer-
derived Si-C-N ceramics, J. Am. Ceram. Soc. 1998,81, 2971-2977.
17. G. D. Sorarh, F. Babonneau, and J. D. Mackenzie, Structural evolution from polycarbosilane
to S i c ceramics, J. Muter. Sci. 1990, 25, 38863893.
18. G. M. Renlund, S. Prochazka, and R. H. Doremus, Silicon oxycarbide glasses: Part I. Prepara-
tion and chemistry, J . Muter. Res. 1991, 6, 2716-2722.
19. G. M. Renlund, S. Prochazka, and R. H. Doremus, Silicon oxycarbide glasses: Part 11. Struc-
ture and properties, J. Muter. Res. 1991, 6, 2723-34.
20. R. West in Ultrastructure Processing of Ceramics, Glasses and Composites, L. L. Hench and
D. R. Ulrich (Eds), Wiley, New York, 1984, pp. 235-244.
21. R. H. Baney, J. H. Gaul, and T. K. Hilty, Methylchloropolysilanes and Derivatives Prepared
from the Redistribution of Methylchlorodisilanes, Organometallics, 1983, 2, 859-863.
22. S. Yajima, J. Hayashi, M. Omori, and K. Okamura, Development of Silicon Carbide Fiber with
High Tensile Strength, Nature, 1976, 261, 683.
23. C. L. Schilling, J. P. Wesson, and T. C. Williams, Polycarbosilane Precursors for Silicon
Carbide, Am. Ceram. Soc. Bull. 1983, 62, 912-915.
24. L. V. Interrante, C. W. Whitmarsh, W. Sherwood, H.-J. Wu, R. Lewis, and G. E. Maciel, High
yield polycarbosilane precursors to stoichiometric Sic. Synthesis, pyrolysis and application,
Mater. Res. SOC. Symp. Proc. 1994,346, 593-603.
25. D. Seiferth and G. Wiseman, High-Yield Synthesis of Si,N,/SiC Ceramic Materials by
Pyrolysis of a Novel Polyorganosilazane, J. Am. Ceram. SOC.1984, 67, C-132.
26. R. Riedel, A. Kienzle, W. DreBler, L. Ruwisch, J. Bill, and F. Aldinger, A silicon carbonitride
ceramic stable to 2,0Oo”C, Nature, 1996, 382, 796-798.
27. S. Yajima., T. Iwai, T. Yamamura, K. Okamura, and Y. Hasegawa, Synthesis of Polytitanocar-
bosilane and Its Conversion into Inorganic Compounds, J . Muter. Sci. 1981, 16, 1349-1355.
28. K. Okamura, M. Sato, T. Matsuzawa, and Y. Hasegawa, Formation of S i c Fibers and Related
Ceramic Fibers from Polycarbosilane, in Ultrastructure Processing of Advanced Ceramics, J. D.
Mackenzie and D. R. Ulrich (Eds), Wiley-Interscience, New York, 1988, pp. 501-518.
29. F. Babonneau, G. D. Soraru, K. J. Thorne, and J. D. Mackenzie, Chemical characterization of
Si-AI-C-0 precursor and it pyrolysis”, J . Am. Ceram. Soc. 1991, 74, 1725-1728.
30. F. Babonneau and G. D. Sorarh, Synthesis and Characterization of Si-Zr-C-0 ceramics from
polymer precursor, J. Eur. Ceram. Soc. 1991, 8, 29-34.
31. K. Thorne, E. Liimatta, and J. D. Mackenzie, Synthesis of SiC/TaC ceramics from tantalum
alkoxide modified polycarbosilane, J. Muter. Res. 1991, 6, 2199-2207.
32. D. Seyferth, N. Bryson, D. P. Workman, and C. Sobon, Preceramic polymers as “reagents” in
the preparation of ceramics, J. Am. Ceram. Soc., 1991,74,2687-2689.
33. P. Greil, Active-Filler-Controlled Pyrolysis of Preceramic Polymers, J . Am. Ceram. Soc. 1995,
78, 835-848.
34. R. M. Laine, F. Babonneau, K. Y. Blohowiak, R. A. Kennish, J. A. Rahn, and G. J. Exharos,
The evolutionary process during pyrolytic transformation of poly(n-methylsilazane) from a
preceramic polymer into an amorphous silicon nitride/carbon composite, J . Am. Ceram. Soc.
1995, 78, 137-145.
35. G. D. Sorarh, Q. Liu, L. V. Interrante, and T. Apple, The role of precursor molecular structure
on the microstructure and high temperature stability of silicon oxycarbide glasses derived from
methylen-bridged polycarbosilanes, Chem. Muter. 1998, 10, 4047-4054.
36. S. Yajima, K. Okamura, T. Matsuzawa, Y. Hasegawa, and T. Shishido, Anomalous Character-
istics of the Microcrystalline State of S i c Fibers, Nature, 1979, 271, 706.
37. K. Okamura, M. Sato, T. Seguchi, and S. Kawanishi, Preparation of high-temperature
strength S i c fiber, in Controlled Interphases in Composite Materials, H. Ishida (Ed.), Elsevier,
Amsterdam, 1990, pp. 209-218.
38. D. Bahloul, M. Pereira, P. Goursat, N. S. Choong Kwet, and R. J. P. Corriu, Preparation of
silicon carbonitrides from an organosilicon polymer; I, Thermal decomposition of the cross-
linked polysilazane, J. Am. Ceram. Soc. 1993, 76, 1156-1 162.
References 475
39. K. Okamura, M. Sato, and Y. Hasegawa, Silicon nitride fibers and silicon oxynitride fibers by
the nitridation of polycarbosilane, Cerum. Int. 1987, 13, 55-61.
40. G . D. Soraru, A. Ravagni, R. Campostrini, and F. Babonneau, Synthesis and characterization
of p-sialon ceramics from organosilicon polymers, J . Am. Ceram. Soc. 1991, 74, 2220-2223.
41. G. D. Soraru, A. Ravagni, R. Dal Maschio, G. Carturan, and F. Babonneau, Polymer-derived
Si3N4-Zr02 nanocomposite powders, J. Muter. Res. 1992, 7, 1266-1270.
42. G. Granozzi, A. Glisenti, R. Bertoncello, and G. D. Soraru, XPS study of the nitridation
process of a polytitanocarbosilane into Si-Ti-N-0 ceramics, Muter. Res. Soc. Symp. Proc.
1992,271, 899-904.
43. D. Suttor, H.-J. Kleebe, and G. Ziegler, Formation of mullite from filled siloxanes, J . Am.
Ceram. Soc. 1997,80,2541-2548.
44. M. Monthioux and 0. Delverdier, Thermal behavior of (organosilicon) polymer-derived
ceramics. V: main facts and trends, J . Eur. Ceram. Soc. 1996, 16, 721-737.
45. G. D. Soraru, F. Babonneau, and J. D. Mackenzie, Mixed carbides via polymer route, Muter.
Res. Soc. Symp. Proc. 1990, 180, 815-818.
46. M. Narisawa and K. Okamura, Polymer pyrolysis, in Chemical Processing ofcerumics, B. I.
Lee and E. J. A. Pope (Eds), Marcel Dekker, New York, 1994, pp. 375-394.
47. G . D. Soraru, M. Mercadini, R. Dal Maschio, F. Taulelle, and F. Babonneau, Si-AI-0-N
fibers from polymer precursor: Synthesis, structural and mechanical characterization, J . Am.
Ceram. Soc. 1993, 76, 2595-2600.
48. L. V. Interrante, C. W. Whitmarsh, C-Y. Yang, W. Sherwood, W. Schmidt, P. S. Marchetti,
and G. E. Maciel, Processing of Si-based ceramics and ceramic composites using hydridopoly-
carbosilane (HPCS), in Silicon-BusedStructural Ceramics, Ceramic Transactions, Vol. 42, 1994,
pp. 57-69.
49. S. T. Gonczy, E. P. Butler, N. R. Khasgiwale, L. Tsakalakos, W. R. Cannon, and S. C.
Danforth, Blackglas-Nicalon composites with CVD boron nitride fiber interface coatings,
Ceram. Eng. Sci. Proc. 1995, 16, 433441.
50. P. Carri and G. D. Soraru, Sol-gel processing of continuous fiber reinforced composites by
the liquid infiltration and pyrolysis (LIP) method, in Innovative Processing and Synthesis of
Ceramics, Glasses and Composites, N. P. Bansal, K. V. Logan, and J. P. Singh (Eds), Ceramic
Transactions, Vol. 85, 1997, pp. 405416. The American Ceramic Society, Westerville, Ohio,
USA.
51. P. Colombo, T. E. Paulson, and C. G. Pantano, Atmosphere effects in the processing of silicon
carbide and silicon oxycarbide thin films and coatings, J . Sol-Gel. Sci. Technol. 1994,2,601-604.
52. P. Colombo, T. E. Paulson, and C. G. Pantano, Synthesis of S i c thin films with polycarbosilane
(PCS), J . Am. Ceram. Soc. 1997, 80, 2333-2340.
53. P. Colombo, G. A. Danko, and R. Silberglitt, Production of SIC and SiOC from preceramic
polymers: comparison of microwave and conventional heating, in Ceramics ~- Getting into the
2000s, Part C (Section F), Techna Srl, Faenza, 1998. pp. 353-360.
54. P. Colombo and M. Modesti, Silicon oxycarbide ceramic foams from a preceramic polymer,
J . Am. Ceram. Soc. 1999,82, 573-578.
55. P. Colombo, V. Sglavo, E. Pippel, and J. Woltersdorf, Joining of Reaction-bonded silicon
carbide using a preceramic polymer, J. Muter. Sci. 1998, 33, 2409-2416.
56. R. Riedel, G. Passing, H. Schonfelder, and R. J. Brook, Synthesis of Dense Silicon-based Cera-
mic at Low Temperatures, Nature, 1992, 355, 714717.
57. K. Jakubenas and H. L. Marcus, Silicon carbide laser pyrolysis of polycarbosilane, J . Am.
Ceram. Soc. 1995. 78, 2263-2266.
58. J. C. Pivin and P. Colombo, Conversion of inorganic-organic polymers to ceramics by ion
implantation, Nucl. Instr. Meth. Phys. Res. B, 1996, 120, 262-265.
59. J. C. Pivin and P. Colombo, Ceramic coatings by ion irradiation of polycarbosilanes and
polysiloxanes. Part I: Conversion mechanism, J . Muter. Sci. 1997, 32, 61634173.
60. J. C. Pivin and P. Colombo, Ceramic coatings by ion irradiation of polycarbosilanes and
polysiloxanes. Part 11: Hardness and thermochemical stability, J . Muter. Sci. 1997, 32, 6175-
6182.
61. S. Yajima, T. Shishido, and K. Okamura, SIC bodies sintered with three-dimensional cross-
linked polycarbosilane, Ceram. Bull. 1977, 56, 1060-1063.
476 6 Polymer tn Ceramic Transformation: Processing of Ceramic Bodies and Thin Film&
62. K. B. Schwdrtz, D. J. Rowcliffe, and Y. Blum, Microstructural development in Si3N4/Polysila-

zane bodies during heating, Adv. Ceram. Muter. 1988, 3, 320-323.
63. R. Riedel, K. Strecker, and G. Petzow, In Situ Polysilane-Derived Silicon Carbide Particulates,
J . Am. Ceram. SOC.,1989 72, 2071-2077.
64. H. Kodama and T. Miyoshi, Preparation of Fine-Grained Sic Ceramics from Pyrolyzed Poly-
carbosilane, Adv. Ceram. Muter. 1988, 3, 177-179.
65. H. Tanaka and Y. Kurachi, Sunthesis of P-SiC Powder from Organic Precursor and is Sinter-
ability, Ceram. Int. 1988, 14, 109-1 15.
66. R. Riedel, M. Seher, and G. Becker, Sintering of Amorphous Polymer-Derived Si, N and C
Containing Composite Powders, J . Europ. Ceram. SOC.,1989, 5, 113-122.
67. V. M. Sglavo, R. Dal Maschio, G. D. Sorard, and A. Bellosi, Fabrication and characterization
of polymer-derived Si2N20-Zr02 ceramics, J . Muter. Sci. 1993, 28, 6437-6441.
68. G. Passing, R. Riedel, H. Schonfelder, and R. J. Brook, Polymer-Derived Ceramic Compo-
nents, in Euro-Ceramics II, Vol. 2, G. Ziegler and H. Hausner (Eds), Deutsche Keramische
Gesellschaft, Koln, 1992, pp. 601-606
69. R. Riedel, H. J. Kleebe, H. Schonfelder, and F. Aldinger, A Covalent Micro/Nano-Composite
Resistant to High-Temperature Oxidation, Nature, 1995, 374, 526528.
70. R. Riedel, M. Seher, J. Mayer, and D. V. Szabo, Polymer-Derived Si-Based Bulk Ceramics,
Part I: Preparation, Processing and Properties, J . Europ. Ceram. Soc., 1995, 15, 703-715.
71. M. Mayer, D. V. Szabo, M. Ruhle, M. Seher, and R. Riedel, Polymer-Derived Si-Based Bulk
Ceramics, Part I: Microstructural Characterization by Electron Spectroscopy Imaging,
J . Europ. Ceram. SOC.1995, 15, 717-727.
72. J. Bill, R. Riedel, and F. Aldinger, Hybrid Processing of Silicon-Based Polymers to Composite
Ceramics, in Euro-Ceramics IV, Vol. 4, A. Bellosi (Ed.), Gruppo Editoriale Faenza Editrice,
1995, pp. 125-132.
73. L. An, R. Riedel, C. Konetschny, H.-J. Kleebe, and R. Raj, Newtonian viscosity of amorphous
silicon carbonitride at high temperatures, J . Am. Ceram. SOC.,1998, 81, 1349-1352.
74. R. Riedel, L. M. Ruswisch, L. An, and R. Raj, Amorphous siliconboron carbonitrlde ceramic
with very high viscosity at temperatures above 1500"C,J . Am. Ceram. Snc. 1998,81,3341-3344.
75. G. D. Sorard, V. M. Sglavo. F. Vulcan, and F. Babonneau. Fabrication and characterization of
p-SiAlON components from polymeric precursors, Muter. Res. SOC.Symp. Proc. 1993, 287,
245-2 50.
76. T. Rouxel, G. Massouras, and G. D. Sorard, High temperature behavior of an SiOC oxycarbide
glass: Elasticity and viscosity, J. Sol-Gel Sci. Technol. 1999, 14, 83-94.
77. R. Cytermann, A New Way to Investigate the Dependence of Elastic Moduli on the Micro-
structure of Porous Materials, Pow. Mettall. Int. 1987, 19, 27-30.
78. J. C. Pivin, P. Colombo, and M. Tonidandel, Ion irradiation of preceramic polymers thin films,
J . Am. Ceram. SOC.1996,79, 1967-1970.
79. J. C. Pivin, P. Colombo, M. Sendova-Vassileva, J. Salomon, G. Sagon, and A. Quaranta, Ion-
induced conversion of polysiloxanes and polycarbosilanes into ceramics: Mechanisms and
properties, Nucl. Instr. Meth. Phys. Res. B, 1998, 141, 652-662.
80. J. C. Pivin and M. Sendova-Vassileva, Visible photoluminescence of ion irradiated polysiloxane
films, Solid State Commun. 1998, 106, 133-136.
81. T. E. Paulson, Polymer processing of silicon (0xy)carbide thin films and microcomposite
coatings, M.Sc. Thesis, The Pennsylvania State University, University Park, PA, 1994.
Materials and Applications
Part I11
Diamond Materials and their Applications

Edited by I. Sigalas and R. J. Caveney
1.1 Superabrasive tools: A Brief Introduction

M. W. Bailey
1.1.1 Introduction
Tooling systems containing diamond and cubic boron nitride materials are used
today across the breadth of modern industry, which could not function effectively
without them. Their range of application covers petroleum exploration, mining,
stone and concrete sawing, the grinding of tungsten carbides, glass, and ceramics,
the machining of a wide range of engineering materials including aluminum
alloys, modern composite materials, hard ferrous materials, and also in such
unlikely fields as the cutting of frozen foods. Most applications take advantage of
the extremely high hardness and wear resistance of these so-called ‘ultrahard’ or
‘superabrasive’ materials. However, more recently, the outstanding thermal and
optical properties of diamond are being more widely exploited, giving rise to further
(and new) areas of application.
1.1.2 Early History

Diamond, as a material, has been recognized as having unique properties for several
thousands of years and references of it being used as an industrial tool can be traced
back to at least 300 BC. At this time, whole stones appeared to have been used for
engraving and, by 150 BC, India, where the early diamond deposits were discovered,
had established an export trade with China for such diamond-tipped engraving tools
for use in cutting very hard jade stone (Fig. 1).
No major change in the use of diamond as an industrial tool occurred for some
considerable time. Reference was made by Pliny in his ‘The History of the
World,’ published in AD 77, to splinters of diamond being used as engraving
tools. Other industrial applications developed very slowly until the 18th and 19th
centuries. In AD 1751, Diderot describes in his ‘Encyclopaedia’ a method of drilling
rock using fragments of diamond held in the lower end of a diamond rod. This rod, a
precursor to modern rock drills, was raised and dropped by an operator who also
rotated it with his hand (Fig. 2). By the mid 18OOs, such rock drills were machine
powered and a report in 1852 claimed that a hole, 5 cm in diameter, had been drilled
in granite to a depth of 37 cm in 87 min. With modern diamond technology, this can
now be achieved in a matter of seconds.
480 1 Diamond Materials and their Applications
Figure 1. Diamond trade routes from the 1st century BC to the 3rd century AD.
Other developments were taking place during this period. A diamond tipped
lathe tool was used by J. Ramsden in 1773, the beginning of precision turning.
The use of diamond as a wire drawing die was patented in England in 1819. By
the end of the 19th century, large saw blades with diamond set around their
Figure 2. The first diamond-tipped ‘percussion’ drill (Diderot).

1. I Superabrasive tools: A Brief Introduction 48 1
periphery were in use in France for cutting stone. The cutting of marble and lime-
stone used during the construction of large buildings in Paris in the early 1900s
utilized diamond-containing saw blades up to 2-3 m in diameter.
In 1842, the first mention was made of a diamond tool utilizing relatively small
pieces of diamond when Pritchard reported grinding and polishing microscope
lenses using a tool consisting of diamond grit hammered into the surface of a cast
iron tool. The first patent for a diamond wheel was granted to A. L. Caverdon in
France, in 1878, when he produced a wheel containing fine diamond particles
held onto a metal wheel form possibly by jamming them into indentations made
in the surface of the wheel. This method was described by Carl Zeiss in 1906.
Most modern diamond tools, including the majority of saw blades and grinding
wheels, consist of small particles of diamond held in a matrix or ‘bond’ which
wears down during use to expose new diamond particles, which carry out the
cutting. This is in contrast to the very early tools which used relatively large
single stones or slivers from single stones. A patent for such a bond system was
granted to the Western Electric Company in 1927. During the early 1930s, in
order to grind the then-new material, cemented tungsten carbide, which had
been invented by Krupp in Germany in 1928, grinding wheels containing diamond
held in a resin bond were developed and several patents were granted between 1932
and 1933.
1.1.3 Synthetic Diamond

The commercial availability of synthetic diamond has been a major contributor to
the dramatic advances which have occurred during the past 50 years in both
diamond tool technology and also in the range of applications in which they are
used. In the period between 1953 and 1958, three independent teams of scientists
from ASEA in Sweden, General Electric in the United States and De Beers in
South Africa were successful in producing synthetic diamond by converting graphite
to diamond using high-pressure, high-temperature technology. Prior to this, the
limited supply of natural diamond impeded development of new diamond tools.
The commercial availability of synthetic diamond offered two advantages. First,
there was potentially unlimited availability of industrial diamond compared to the
limited volumes of suitable natural material and, second, it offered the opportunity
of engineering material to have specific properties suited to particular industrial
applications.
Of the diamond used for industrial purposes today, some 90% is synthetic, and
10% natural. The major use of diamond grit is in the machining (sawing and dril-
ling) of stone and concrete (Fig. 3). Finer sizes of diamond are used widely for the
grinding, sawing, and polishing of glass, ceramics, tungsten carbides, and a very
wide range of other mainly nonferrous engineering and industrial materials.
Specialist applications, for example fine high precision turning (Fig. 4) of nonfer-
rous and precious materials, drawing of fine wire, and the dressing of conventional
abrasive wheels have, for several decades, used natural diamond. During the 198Os,
the technology to produce, economically, large synthetic single diamond crystals
Figure 3. Diamond drilling a 100 mm hole for a new sea wall drainage system
was developed. Whole synthetic crystals or cut pieces of the order of several milli-
meters in size are used for these high precision industrial applications.
1.1.4 Cubic Boron Nitride

Following the successful commercial synthesis of diamond in the 1950s, the second
hardest material known, cubic boron nitride, cBN, was introduced to the market in
the 1960s and is complementary to diamond. The iron, and its alloying elements, in
ferrous materials has a tendency to react chemically with diamond under machining
conditions and this can reduce the efficiency of the tool. cBN, however, although not
as hard as diamond, does not react chemically with iron and is therefore particularly
well suited to machining hard ferrous materials.
1.1.5 Polycrystalline Diamond and Cubic Boron Nitride

The next notable material development was the introduction to the market in the
early 1970s of polycrystalline diamond (pcD). This made available to industry
1.1 Superabrasive tools: A Brief Introduction 483
Figure 4. Natural diamond cutting tool machining a contact lens.
relatively large pieces of diamond, albeit polycrystalline rather than mono-

crystalline, at an economical cost. Such materials can be produced with varying
mechanical properties and are used in a wide variety of cutting applications and
also as a wear resistant surface. As a cutting tool material, it is used predominantly
to machine nonferrous abrasive materials including drilling of rock in oil and
gas exploration, and in the engineering and manufacturing industries for the
machining of aluminum alloys, reinforced plastics, new wear resistant lightweight
composite materials, wood, and wood composites (Figs 5 and 6). pcD is also used
in noncutting applications and one of its original and important applications is as
a die for drawing wire. More recently, the use of pcD as a wear resistant compo-
nent in thrust bearings, work rests, and high precision gauging equipment has
become more widespread.
The development of polycrystalline cBN (pcBN) in the mid-1970s made it
economically possible to machine fully hardened ferrous workpieces on a lathe or
milling machine when previously the only method of machining such workpieces
was by grinding, normally with silicon carbide or aluminum oxide (conventional)
abrasives. Huge increases in productivity can be achieved by changing to turning
(or milling) with pcBN. In addition, environmental improvements can also be
achieved, since in many cases pcBN does not need coolant.
Figure 5. A pcD-tipped cutter machines oak panels for kitchen furniture.
1.1.6 Chemical Vapor Deposited Diamond

The development in the late 1980s of an entirely new method of diamond synthesis
using chemical vapor deposition (CVD), as an alternative to ultra-high pressure and
temperature systems, has caused a considerable stir in the industry.
Figure 6. pcD machining a cast metal-matrix composite automotive brake motor in A1 20% Sic,.
1.2 The Crystullizution qf Diamond 485
With this technique, it is possible to deposit layers of diamond onto a substrate

from a hot carbon-containing plasma, enabling relatively large areas of diamond
to be produced (at the time of writing, the order of 10000mm2). Traditional dia-
mond tools utilize the extreme hardness of diamond (and cBN) but, with this new
technology, which is still developing, the exceptional optical and thermal properties
can be applied. Diamond has a very high thermal conductivity (up to six times that
of copper at room temperature) and is transparent to visible and infrared radiation,
opening the door to a new varied range of industrial applications where these
properties can be exploited in areas such as thermal management of electronics
and windows for high technology infrared equipment, in addition to its use as an
alternative material for use in the more traditional diamond tool areas.
1.1.7 Outline of Chapter

Section 1.2 of this Chapter reviews the crystallization of diamond and cubic boron
nitride using high pressure, high temperature techniques. Chapter 4 of Part I1 deals
with the subject of chemical vapor deposition of diamond at low pressure.
The high pressure sintering of diamond and cubic boron nitride to form super-
hard composites is described in Section 1.3, a review of attempts to produce other
hard materials is given in Section 1.4.
Section 1.5 summarizes the many and varied applications of diamond and cBN
materials. The applications of diamond grown using the CVD technique are dis-
cussed in Chapter 2 of Part 111.
1.2 The Crystallization of Diamond

G. J. Davies
1.2.1 The Carbon Phase Diagram

It was established that diamond is an allotrope of carbon by Tennant in 1797 [1].
This led to many attempts to crystailize diamond using various carbonaceous start-
ing materials, but it was not until about a century and a half later that successful
synthesis was proven, as referred to in Section 1.1.3. The first clear success was by
a Swedish group at the ASEA Company in February 1953 [2]. This was followed by
the General Electric Company in December 1954 [3]. The early attempts and the
subsequent successes are well reviewed [4-61.
Natural diamond has been used by man since at least biblical times, not only as a
gem but also, due to its extreme properties relative to other materials, as a n abrasive
and even as a medicine in crushed form, ‘a panacea for all ills’. It is not intended here
to enter the debate concerning the crystallization of natural diamond. Some general
h
(3
W
cc
I3
v)
v)
W
cc
a
0 1000 2000 3000 4000 5000 6000

TEMPERATURE (K)
Figure 7. The phase diagram of carbon adapted from [9].

A: region of solvent/catalyst-based synthesis of diamond from graphite using:
A, : commercially used transition metal solvent/catalyst [ 161
A2: nonconventional metal solvents: Cu, Zn, Ge [17-201
A,: nonmetallic solvents, mainly metal salts, silicates [22-281
B: the synthesis of diamond from phenolic resin and cobalt [30]
C: region of direct conversion of hexagonal graphite to either hexagonal or cubic diamond, where
the dotted line C1-2,represents the solid-solid transformation line of graphite or hexagonal
diamond to cubic diamond through fast pressure/temperature cycles [9]
C,: conversion of hydrocarbons to cubic diamond [33]
C4: transformation of C60fullerenes into polycrystalline diamond [35]
D: low pressure metastable growth of diamond [12, 13, 36-38]
N: regions of natural diamond genesis [7, 8, 21, 291.
natural diamond genesis information is to be found elsewhere [7, 81. The conditions
under which natural diamonds are formed will, however, be touched upon when
the phase transformation diagram for carbon is discussed later in this section on
crystallization.
To gain a balanced general view of the many ways in which man has crystallized
diamond, the thermodynamically stable and metastable phases of elemental carbon
and reaction dynamics between them, over obtainable pressures and temperatures,
should be considered. The most up to date phase and transformation diagrams are
to be found in a review by Bundy et al. [9]. Figure 7 is an adaptation of the pressure-
temperature phase transition diagram taken from this review article, together with
further information gleaned from the extensive diamond- and carbon-related scien-
tific literature.
The dominant thermodynamically stable forms of carbon are graphite, diamond,
liquid, and vapor. The phase diagram presented by Bundy [9] and adapted in Fig. 7
1.2 The Crystallization of‘ Diamond 487
has some new notable features over many of the earlier diagrams. The slope of the
diamond melting line is positive, and molten carbon is said to be metallic with little
evidence for transformations from electrically conducting to nonconducting forms.
For consideration of diamond crystallization, one of the most important features is
the diamond-graphite equilibrium line which stretches from 1.7GPa at OK to
12 GPa at 5000 K, the diamond-graphite-liquid carbon triple point. The activation
energies required to initiate transformations from one form of carbon to another are
very largely due to the high cohesive energy of carbon in its crystalline structures.
Davies and Evans [lo] showed that the activation energy for the graphitization of
the { 1 lo} surface of pure diamond in vacuo is 728 f 50 kJ mol-’, which is about
twice the bond energy of the C-C bond in aliphatic hydrocarbons (3.7eV per CJ
C-C bond). This is also about the same energy required for the vaporization of gra-
phite. These very high activation energies result in very large areas of pressure and
temperature over which a metastable form of carbon can subsist. Probably the
most important example of this is the indefinite existence of diamond at room tem-
perature and pressure even though, at these conditions, it is very ‘deep’within the gra-
phite stability field. To overcome the activation energy for conversion to graphite in
vacuo, diamond must be heated to a temperature-of about 2000 K before graphitiza-
tion becomes rapid [ 101. In addition to the allotropes of carbon, there are polytypes of
carbon, that is structural variations based upon lattice plane stacking difference.
Important examples are hexagonal diamond, which occurs in methods of very
rapid and hot direct transformation into diamond, and rhombohedra1 graphite, pro-
duced by cold working normal hexagonal graphite crystals. When the nanoscale
groupings of pure carbon atoms are considered, other solid forms with very distinctive
properties are known and are probably best described as ‘metastable’ over the whole
range of pressures and temperatures in Fig. 7. These are amorphous forms such as
carbon black and glassy carbon, condensate of very long linear carbon molecules col-
lectively called carbynes and, of course, the now very extensive variations of fullerenes
(C60, C70, etc.) and the derived nanotubes [ 1 I].
Importantly for diamond crystallization, all these forms of carbon, including gra-
phite, very stubbornly persist at pressures far into the diamond stable region. The
practical synthesis of diamond could thus be considered as the search for means
of finding practical convenient kinetic pathways to avoid the high activation barriers
between the multifarious nondiamond carbon structures and diamond. The two
very general approaches are to use solvents for carbon and to precipitate diamond
primarily in the diamond stable part of the phase diagram or to ‘quench’ carbonac-
eous species from very high energy states such as vapors, liquid, or plasmas. Some of
the more recently reported means could well be considered to be combinations of
these [12, 131. Some of the most important diamond crystallization techniques in
the recent literature will now be itemized using Fig. 7 as a guide.
1.2.2 Diamond Crystallization at High Pressure

The shaded region in Fig. 7 labeled with pressure from about 4.5 GPa to 9 GPa
and temperature from about 1475K to 2475 K, gives the conditions under which the
bulk of the reported solution grown, diamond stable region, crystallization of

diamond has taken place.
The graphite-diamond equilibrium expression in this region, following Berman
and Simon [14] is
P
AG:=AH~-7.AS~,o[ AVdP=O
It has stood the test of time and experimental verification [ 151.

Al is where most of the commercial diamond abrasive products and high pressure
polycrystalline diamond products are manufactured using alloys of transition
metals, mainly iron-nickel, cobalt-iron and to a lesser extent nickel-manganese
for the grits and larger individual crystals, and mainly cobalt for the polycrystalline
products [16]. In this reference, the general transition metal solvent work up to the
end of the 1980s is covered. In principle many of the other transition metals and
alloys can be used but a large number of them require very difficult to attain and
control temperatures and pressures dependent upon melting points, carbide stabili-
ties, and carbon solubilities.
The position A2, somewhat higher in pressure and temperature than A , , covers
the region where the higher melting and or lower carbon solubility metals and
alloys can be used. Diamond synthesis from graphite has been demonstrated
using copper, zinc, and germanium [17, 181. Magnesium was shown to be a workable
solvent/catalyst [19] at about 7.7 GPa and 2275 K. Later, by the addition of 5&60%
copper to magnesium, diamond synthesis conditions were reduced to 6 GPa and
1825 K [20].
The A3 position in Fig. 7 is the general condition under which inorganic com-
pounds have been used to aid in the conversion of graphite to diamond. The
more recent work in this area has mainly been carried out at the National Institute
for Research in Inorganic Materials, Tsukuba, Japan. This work was probably sti-
mulated by the possibility that volatiles such as C 0 2 , H 2 0 , CH4, and O2 may play
an important role in natural diamond formation [21]. Carbonates [22], hydroxides
[23] and sulphates [23, 241 with alkali and alkali earth cations have proven to
work at 7.7GPa and temperatures in the range 1825-2475K. The diamonds
formed were aggregates with particles up to 20 pm in size with normal faceting.
In order to clarify the possible role of C 0 2 , silver carbonate has been investigated
compared to silver metal [25]. This compound aided diamond formation above
about 2075K and 7.7GPa, whereas pure silver did not act as a solvent/catalyst
under these conditions. Further, volatile-rich silicate melts of kimberlite compo-
sition (kimberlite is the igneous rock usually associated with natural diamond in
diamond bearing pipes) and Si02-H20 combinations aided graphite conversion
to fine diamond at 7.7 GPa and 20750-2475 K [26, 271. The need for the presence
of volatiles when using silicates to aid diamond formation has been recently con-
firmed [28]. In this report, anhydrous magnesium silicate melts corresponding to
the composition of the minerals forsterite (Mg2Si04)and enstatite (MgSiO,) were
compared with hydrated versions of these compositions. Fine diamonds of euhedral
shape were formed only in the hydrated magnesium silicate graphite systems above
about 2075 K at 7.7 GPa. No diamond was detected in the anhydrous systems up to
2175 K at this pressure even after treatments lasting 8 h or more. These findings are
consistent with the current theories of natural diamond genesis.
The two regions both marked N in Fig. 7 are intended to illustrate the conditions
under which natural diamond formed in the earth’s crust, probably over geological
time scales (about 3.3 billion years). There are two environments linked to natural
diamond formation, the peridotitic (depicted by the lower pressure N region on
Fig. 7) and the eclogitic (at higher pressures and about 200 K hotter, also marked N).
The pressure given by the N positions are estimated from the accepted
correspondence of 0. 1 GPa pressure to 3 km depth into the earth crust. The lower
pressure limit of N is given by the boundary zone between the upper region litho-
sphere and the mobile asthenosphere at about 150 km in the earth’s mantle.
It is believed that natural diamonds, crystallized from C 0 2 or CH4 volatiles below
the lithosphere and after existing for great periods of time at these depths, are
brought to the surface by subsequent eruptions of magma. For more detailed discus-
sions of the geology, mineral inclusions in natural diamond and general matters
related to diamond genesis, the reader is referred to Harris [7], Kesson [8], Haggerty
[29], and Meyer [21].
The position marked B at 4 GPa and temperatures between 1775 and 2075 K,
gives the condition under which well faceted diamonds of about 200 pm diameter
have been crystallized using specially treated phenolic resins as the source of
carbon and molten cobalt as solvent [30]. The notable point of this particular crys-
tallization is that it occurred well into the graphite stable region of Fig. 7 and is thus
an example of the ‘metastable growth of diamond’. An explanation for this may be
found in consideration of the relative solubilities of phenolic resins and diamond in
molten cobalt allowing dissolution of one metastable form and precipitation of
another, namely diamond. This is an example where rules governing the transitions
between metastable states, such as the Ostwald and Ostwald-Volmer rules, can be
applied [3I].
The curved dashed line CI-Cz in Fig. 7 depicts the proposed boundary between
the formation of the hexagonal (below the line) or cubic (above the line) polytypes of
diamond, after explosive shock compression or fast heating, mainly using laser
heated diamond anvil cells, of graphite structures in the diamond stable region,
described by Bundy et al. [9]. This reference is a detailed review article and is a
major source of reference for work concerned with the direct conversion of graphite
and other carbon structures to diamond. The diamonds that can be retrieved from
these various methods, either of the hexagonal or cubic form, are always small and
crystallographically defective. This is because of the very short times at high tem-
perature and pressure, and the need to satisfy fast thermal quench requirements.
Despite this, diamond powders have been commercially produced by the Du Pont
Company of the USA and others using explosive shock of metal containing small
crystals of graphite as a carbon source [32].
The direct conversion of hydrocarbons to diamond has also been demonstrated.
C 3 is the region where organic compounds such as anthracene, camphene, fluorene,
pyrene, sucrose, polyethylene, adamantane, and paraffin wax have been converted
to very fine diamonds under high pressure (about lpm in size) (Wentorf [33]).
Aliphatic compounds seem to work but compounds with aromatic rings and/or
large amounts of nitrogen only formed graphite of excellent crystallinity under high
pressure pyrolysis and did not form diamond.
There is a growing body of literature describing the crystallization of diamond
from fullerenes as a starting material. One of the latest [34] claims that c 6 0 can be
transformed to diamond at 5GPa and 1675K using a solvent/catalyst without
forming graphite first. The generation of ‘bulk’ polycrystalline diamond at room
temperature has been reported by the crushing of c 6 0 at about 20 GPa [35], (position
C4 in Fig. 7). The conditions used in the polycrystalline diamond (cobalt-matrix)
anvil cell were deliberately nonhydrostatic with respect to pressure. It was suggested,
from the high efficiency of conversion and fast kinetics involved, that, if cheap bulk
C60became available at some time in the future, a feasible method for the commer-
cial production of polycrystalline diamond might result.
The metastable crystallization of diamond in the appropriate pressure range of
0-0.5 GPa and temperature range 675-1300K has now been accomplished using a
‘galaxy’ of techniques in region D in Fig. 7. These include chemical vapor deposi-
tion from carbonaceous gases, mainly methane, hydrogen mixtures using many
varied means of generating plasma [36, 371 (see also Chapter 4 of Part 11), the
use of a combination of lasers and carbon dioxide as source (the so called QQC
deposition process) [13], hydrothermal synthesis in the C-H-0 and C-H-O-
halogen systems [38], low pressure solid source processes (the so called LPSSS
methods) [12], and many more. The literature in this general area of metastable
diamond growth has now become extensive, as discussed in Section 1.2.6.
From a commercial point of view, the production of diamond abrasives is, at this
time, dominated by the high-pressure, high-temperature, transition metal solvent/cat-
alyst techniques where the crystallization takes place in the diamond stable region.
1.2.3 High Pressure Apparatus

Early attempts at generating high pressure used opposed anvils, frequently referred
as ‘Bridgman anviis’ after the father of high pressure, Professor P.W. Bridgman [39].
A constraint of this system was the limited volume of material to which the high
pressure could be applied. This problem was addressed by forming recesses in the
two opposing anvils, to form the so-called) ‘toroidal’ device used extensively in
Russia [40]. However, most commercial growth of diamond using metallic sol-
vent/catalysts has been carried out in the belt/girdle and multi-anvil devices
described below.
1.2.3.1 Belt/Girdle Devices

The pressure limit of the piston and cylinder apparatus [39] was extended to allow
commercial production of synthetic diamonds by designing the cylinder to accom-
modate pistons shaped as truncated cones.
If the die is reinforced by prestressing with concentric rings with interference fits,
the device can sustain a pressure of about 6GPa through use of an appropriate
gasket (talc, pyrophyllite, and mixtures thereof) at the die/anvil interface.
I .2 The Crystallization of Diamond 49 1
The gasket allows an even distribution of stress in the die taper and along the
tapered anvil cone. In this way, both the die and anvil are supported by the resulting
compressive stresses.
The essential difference between the belt and girdle devices is the shape of the die
bore/taper and the matching anvils. For the belt apparatus, the die bore is curved
continuously from center to outer surface whereas, for the girdle, the die bore is
straight and proceeds out on both sides in a linear progression.
1.2.3.2 Multi-anvil Devices

The first multi-anvil press was invented by Von Platen [41]. It comprised six anvils in
a cubic arrangement. The benefit of a muiti-anvil device is the possibility of multi-
staging, which allows the yield strength of a compressed component to be increased.
Russian workers [42] have made extensive use of a multi-staged split sphere device.
Pressure is applied to two sets of anvils. The outer arrangement contains eight anvils
which form an octahedral-shaped cavity, and an inner set of six anvils is placed to
form a cubic-shaped central cavity that contains the high pressure capsule in which
the diamonds are grown.
1.2.4 The Synthesis of Particulate Diamond Abrasives

Crystal growers in most of the world’s research and commercial organizations [ 16,
43,441 have efficiently harnessed the solvent/catalyst growth of diamond to produce
both abrasive grits and large single crystals. Scientists have developed mechanisms
to explain the formation of diamond. These have been described by many workers
and could involve the formation of a C+ ion or the formation of an intermediate
carbide which decomposes at synthesis conditions. However, the more accepted
explanation of growth is based on supersaturation. For example, if the well
known Ni-C equilibrium phase diagram, Fig. 8 [45], at 5.7 GPa is studied, an expla-
nation of the formation of diamond under isothermal (thin film growth) or tempera-
ture gradient conditions can be obtained. Both diamond growth techniques are
possible since the solubility of diamond in the molten solvent/catalyst, nickel, is
less than the solubility of graphite.
This section will focus upon the crystallization principles used to produce com-
mercially synthetic diamond ‘grit’ abrasives. The size of particles discussed here
are within an envelope of about 850-200 pm for grits used mainly in stone and con-
crete sawing (saw sizes) and from about 200-50 pm used for grinding applications
(wheel sizes).
The solvent/catalyst crystallization method generally uses graphitic carbons as
the source material at constant high temperature and high pressure. The general
pressure and temperature conditions used are all within the diamond stable
region, at position A, in Fig. 7. The lowest conditions are about 5 G P a and
1500 K, and the highest about 5.8 GPa and 1775 K. These conditions are determined
by the transition metal alloy systems together with the crystal nucleation and growth
conditions all necessary to crystallize the diamond structures required for specific
1700 , I I
1600
1300
1240
0 1.0 2.0 3.0 4.0 Wt.% C 99 100
4.7 9.0 13.15 16.7 At.% C
Figure 8. Ni-C phase diagram at 5.7 GPa [45].
product types. The diamond industry has spent many years and significant amounts
of money developing and refining the large scale high pressure equipment appro-
priate to generate and maintain economically the required crystallization condi-
tions. The high pressure equipment used by the two largest manufacturing
companies, namely De Beers Industrial Diamond Division and the General Electric
Corporation are proprietary, highly evolved designs based upon the belt and/or
girdle apparatus described in Section 1.2.3.1. The Chinese producers mainly use
cubic multi-anvil high pressure designs, and the Russians and Ukrainians use
various modifications of the toroidal equipment.
Possibly the first question a crystal grower must ask when investigating or devel-
oping a crystal growth method is, ‘how is an appropriate driving force for nucleation
and growth generated and controlled’?
The central part of Fig. 8 is considered as representative of a general transition
metal/carbon phase diagram at a pertinent pressure (5.7 GPa) and temperature
range. The driving force for crystallization of diamond using ‘ideal’ graphite as
source carbon would be given by the difference in solubility between graphite and
diamond (the supersaturation) at the chosen conditions. For example, in Fig. 8 at
1450°C and 5.7 GPa this would be the solubility difference AB, (A’B’). Note that
this supersaturation for diamond crystallization (let it be Ac) would decrease as tem-
perature is increased and would essentially be zero at the diamond/graphite equili-
brium temperature, E(eq.)in Fig. 8. The slopes of the diamond/metal liquidus line
and graphite/metal liquidus line at constant temperature will change with pressure.
The difference between these solubility lines, Ac, will thus also change with pressure
and similarly become zero at the diamond/graphite equilibrium condition. Thus the
driving force for crystallization, Ac, is a function of the ‘distance’ from the diamond
graphite equilibrium line in pressure/temperature space, Fig. 9. The super-
saturation, Ac, can of course be related to the nucleation rate and growth rate of
the diamond in the classical way as shown in Fig. 10.
Note the extremely strong dependence of nucleation rate upon supersaturation
and that there is generally a critical level of supersaturation, Accrit,which must be
DIAMOND STABLE
a
GRAPHITE STABLE
TEMPERATURE
Figure 9. Pressure-temperature diagram for diamond synthesis showing typical conditions ( P I ,T I ) .

The supersaturation Ac is a function of P , T I giving rise to different crystal morphology at different
conditions.
Figure 10. The classical relationships of the rates of nucleation and growth to supersaturation, Ac,
for crystallization from a solution.
exceeded before the probability for nucleation becomes large enough for an obser-
vable nucleation rate. The growth rate is an increasing function of supersaturation.
The relative growth rates of different crystallographic growth zones is also a func-
tion of Ac, hence the morphology shift with pressure, temperature condition as
depicted in Fig. 9.
The relationships between solubility, supersaturation, pressure, and temperature
depend upon the details of the microscopic mechanisms operative in the context of
the detailed active impurity chemistry and the specific geometric relationship set up
between the source carbon, molten metal solution, and nucleation site, and or grow-
ing diamond. For an introduction into these extensive subjects the reader is referred
to Burns and Davies [16] where the issues such as nitrogen and boron active impur-
ity chemistries, diamond surface reconstruction, morphology and solvent/catalyst
are discussed.
An important conclusion to be drawn from the above fundamental explanation is
that the structure of the source carbon as it pertains to solubility in the chosen
molten metal is very important. Will and Graf [46] compare six graphite types
using an iron-nickel solvent and show a clear dependence of rate of diamond forma-
tion and diamond size upon the starting crystallinity of the graphites. Moreover
these authors include consideration of the influence of gaseous impurities in the gra-
phites and the cluster theories of Sunagawa [47] concerning the carbon species in
solution in the metal. As the source carbon became more crystalline, the pressure
needed for diamond nucleation increased by up to 0.5GPa. This could be inter-
preted as the need to increase the pressure in order to exceed the critical supersatura-
tion for spontaneous nucleation, Accrit,owing to the much lower solubility of the
recrystallized and thus more crystalline graphite.
Another striking example of the importance of the structure of source carbon
leading to crystallization consequences via the supersaturation driving force is
described in Wang et al. [48]. Here, the pressure needed for spontaneous nucleation
was shown to increase with time as the graphitic source material recrystallized in the
high pressure reaction chamber.
In addition to the explained (at least in general terms) dependence of crystalliza-
tion on the structure of the source carbon, the character of the transport of solute
from dissolving source to growing diamond crystals needs to be considered.
When a diamond synthesis reaction chamber from a process for making grits is
broken open after completion of the crystallization cycle, it is observed that each
grown crystal is separated by a metal alloy film from the undissolved carbon
source material [16]. The metal film, which of course was molten during the
growth of the diamonds, is usually about 50-200 pm in thickness. Figure 11 is a
schematic drawing of this growth geometry. The carbon source material is dissolved
at the molten metal interface, D,and the resultant carbon species in solution trans-
ported by diffusion to the growing diamond surface. The dimensions of these molten
metal films are such that it is unlikely that convection plays any significant role in
carbon transport for this type of crystallization. The growth of diamond grits of
good morphology occurs through lateral propagation of steps over the surface.
The detailed kinetics of surface adsorption, surface diffusion and incorporation
of carbon species into the step structures needs to be studied to understand the
1.2 The Crystallizatioiz of Diamond 495
Figure 11. Schematic growth environment of a synthetic diamond particle.
microscopic mechanisms of growth. The different chemical and reconstructed struc-

tures of the lowest energy crystal facets of diamond give rise to different growth rate
responses to the flux of carbon species resulting in a growth zone history and parti-
cular final morphology for each crystal. The morphology of diamond grits may be
related to the pressure and temperature conditions of growth through the relative
growth rate response of the different crystallographic surfaces to the flux set up
by the supersaturation, Ac, produced by these pressure and temperature conditions,
as in Fig. 10 [ 161. Thus a morphology map may be drawn on such diagrams showing
the regions of pressure and temperature of common morphology [ 161. The difficulty
of investigating the detailed growth mechanisms is extreme due to the ‘inaccessibil-
ity’ of the growing crystal in practical high pressure, high temperature equipment.
The crystal defect types and distributions which characterize any real crystal,
diamond abrasive grits being no exception, may also be discussed in terms of the
driving force for crystallization, and the impurities in solution and/or suspension
in the solvent and the geometry/transport considerations of the growth environ-
ment. Examples of this are the oriented metal solvent inclusions described by
Wakatsuki et al. [49]. These inclusions occur as an array, along certain growth
zone boundaries leading to certain edges of the crystals. The most important crystal
defects for diamond abrasives are those which may affect their behavior in applica-
tions and also those which give rise to their appearance, such as color and clarity. A
discussion of the characterization of synthetic diamond abrasives in terms of strain,
impurity content, both as inclusions and lattice defects, and crush strengths is given
by McCormick et al. [50].These authors consider the macroscopic structure such as
shape, external morphology, cracks, and large inclusions together with the micro-
scopic structure such as small inclusions, dislocations, and some nitrogen related
lattice defects. Using strain sensitive techniques such as Raman spectroscopy,
photoluminescence, and birefringence, together with information from extensive
literature, they arrive at hierarchies of the macroscopic and microscopic structures
496 I Diamond Materials and their Applications
and properties that affect the strengths of the diamond particles. Not surprisingly,
the most important macroscopic structures of the diamond particles turn out to be
size, shape, crack, and inclusion content. It was also demonstrated that particular
nitrogen related lattice defects such as the H3 and N-V center affect the strength
of the crystals. For a comprehensive discussion of the observed impurity lattice
defect centres such as H3 and N-V, the reader is referred to Davies [51]. An impor-
tant observation in [50] was that some of the microscopic defects may well
strengthen the diamond by impeding dislocation motion and possibly acting as
crack arresters.
The very complex nature of the macroscopic and microscopic structures as they
affect strength and the behavior of diamond abrasive/particles still requires
extensive work to elucidate fully. However, from a crystallization point of view,
gaining control over the crystallization behavior is the key to the production'of
optimal diamond abrasives. This, of course, may be achieved by choice and
manipulation of the pressure/temperature conditions, source carbon structure
and solvent/catalyst metal type, leading to control over nucleation and growth
rates.
1.2.5 Growth of Large Synthetic Diamonds

R. C. Burns and M. Sibanda
Large diamonds are best grown using the temperature gradient technique, where
diamond is used as the source of carbon. In this technique, as described by Wentorf
[52], the driving force for reconstitution, under diamond stable conditions, is pro-
vided by the higher solubility of diamond in a hot zone of the solvent/catalyst
and the consequent crystallization of diamond in a cooler zone at constant pressure
(Fig. 9).
The design of the high pressure cell also influences the rate at which this carbon
transport mechanism proceeds. A schematic diagram of a high pressure cell used to
grow reconstituted diamond is shown in Fig. 12. A useful axial temperature gradient
is established by manipulating the distance between the diamond source and seed
pad, which (at constant cell pressure) determines the growth rate and hence, quality
and size of the diamonds grown on the seeds in a fixed time. The small, well formed
diamond seeds, which act as a template for new growth, are mechanically attached
to the seed pad. For a reaction cell pressure of about 5.5 GPa, the temperature at the
diamond source is about 1450°C. The solvent/ catalyst bath is agitated by convec-
tion and soon saturates with carbon. The solvent/catalyst in the vicinity of the seeds,
at a temperature of 1420°C, is therefore supersaturated with respect to diamond,
providing the driving force for diamond growth on the seeds, which are selected
to be about 0.5 mm in size [53, 541.
The nucleated diamond is restrained by the seed pad, resulting in truncated cubo-
octahedral growth, whereas the orientation of the grown diamonds depends on the
orientation of the seed. The habit of the reconstituted diamond depends largely on
the solvent/catalyst, internal pressure and temperature of the reaction cell.
t
DIAMOND POWDE
CARBON HEATER
SOLVENT-CATALYS AT
GROWING CRYSTAL
SEED CRYSTAL T2 f
Figure 12. Schematic diagram of a high pressure cell used in the reconstitution method.
Diamonds of a cubic habit tend to be formed at a lower growth temperature than

those of an octahedral habit [47] (refer to Fig. 9).
The chosen spacing between diamond seeds is determined by the intended size of
the reconstituted diamond. This spacing also affects the growth rate. In practice it is
found, for a fixed temperature gradient, that the smaller the spacing between seeds,
the slower the growth rate. Figure 13 shows a selection of large synthetic diamonds
(0.5 to 1.8ct in weight) grown by De Beers Diamond Research Laboratory from
seeds using the temperature gradient technique.
Crystal perfection is determined by a reasonably low growth rate which is
achieved by optimizing, in combination, the space between seeds, the axial tempera-
ture gradient and the solvent/catalyst under diamond synthesis conditions.
Figure 13. Diamonds grown by De Beers Diamond Research Laboratory from seeds using the tem-
perature gradient technique.
1.2.5.1 Solvent/Catalysts for the Growth of Large Synthetics

1.2.5.1.1 Type Ib Diamonds Effective solvent/catalysts for high pressure, high
temperature diamond synthesis (thin-film and temperature gradient techniques)
are derived from metals and alloys of Group VIII of the periodic table and
metals such as chromium, manganese, tantalum and niobium [55-571. The effective-
ness of transition metals and their alloys depends on a number of factors which
include: their melting points, the degree to which they influence the supersaturation
of the molten solution and also the stability of possible metal carbides at synthesis
conditions [58].
Most commercial solvent/catalyst alloy systems make use of a combination of
metals [16, 59-61] (e.g. iron-nickel-chromium, iron-nickel, iron<obalt, nickel-
manganese-cobalt, or manganese-nickekopper), which enable lower temperatures
and pressures to be used. Large synthetic diamonds of different colors can be grown
using such combinations with or without introducing additives or getter elements,
the most common color being yellow (type Ib). The amount of nitrogen in synthetic
diamonds depends on the composition of solvent/catalyst used, synthesis tempera-
ture and pressure and also the growth rate [16]. In this regard, the solubility of nitro-
gen in the solvent/catalyst and how it is influenced by other elements is also an
important consideration [62]. The higher the solubility, the more readily nitrogen
will be retained by the solvent/catalyst rather than be incorporated into the growing
crystal. This is the case for iron relative to cobalt [63]. Furthermore, the solubility
also increases with temperature. This is consistent with observations of very pale
crystals being grown with pure iron at relatively high temperatures [43, 641.
The choice of solvent/catalyst alloy for synthetic diamond growth not only has an
influence on nitrogen content, but also on other diamond characteristics such as
morphology, color, and inclusion content. The dependence of some of these dia-
mond characteristics on solvent/catalyst typically used for research and commercial
production by institutes and companies such as De Beers, Sumitomo, General
Electric, and NIRIM, amongst others, are summarized in Table 1.
1.2.5.1.2 Type IIa and IIb Diamonds Aluminum, titanium, niobium, and zirconium
are used to remove nitrogen, in the growth of near colorless synthetic diamonds
(type IIa) [62, 66, 671. These nitrogen-getter elements have a high affinity for nitro-
gen and contribute to nitrogen removal by forming stable nitrides. As mentioned
earlier in the chapter, there is a relationship between nitrogen in the grown diamond
and nitrogen solubility in the solvent/catalyst.
The dependence of nitrogen content on the alloy composition is shown in Fig. 14,
where it can be seen that an increase in cobalt content (which is accompanied by
decreased nitrogen solubility in the iron-cobalt solvent/catalyst alloy) results in
increased nitrogen concentration in the grown diamond [67].
Nitrogen concentration in the crystals decrease with increasing amounts of alumi-
num or titanium added to the solvent/catalyst (Fig. 15). Even when large amounts of
aluminum are added, it is difficult to achieve purity levels of less than 0.1 p.p.m.
nitrogen. This is to be expected as the aluminum nitrides are believed to readily
decompose, whilst titanium nitrides are not readily decomposed, making titanium
Table 1. Summary of large synthetic diamonds grown for research purposes by some of the leading producers of synthetic diamonds
Max. size Growth Growth Nitrogen
Solvent ( 4 time (h) rate Color content Morphology
(mg h-') (p.p.m.1
~~ ~
Co-Ti [62] 3.4 300 2.3 near colorless <0.01-2 ( l l l ) , (100) > {113), {115), {llO}
Fe-A1 [65] 4.6 500 1.8 near colorless 0.4-1 1 { l l l ) > (113) > (100) > (110)
Fe-Co [62] 25 1000 5 yellow 100-300 { l l l } , {loo} >> {IlO), j113)
Fe-AI-B (0.02 wt YOB) [65] 5.1 160 1.3 blue - { l l l ) , {110), (100) {113), {115} x
- - - yellow 50-100 { I l l } , (110) >> {110), (113) 2
Ni [43]
Ni -Fe 3.5 - - yellow 50--400 (1111, {loo) 0
Fe-Co-Ti [65] 4 - - near colorless <O.Ol {Ill}, (100) > {113}, {115}, {llO) $CI
-
=:
b
8'
;
R
P
W
W
Co content in Fe-Co solvent (&A)
Figure 14. Nitrogen concentration in diamond versus the cobalt content of iron + cobalt solvent.
The diamonds were grown with 1.5 weight-% titanium addition using (a) high-purity graphite
and (b) synthetic diamond powder [67].
a more efficient nitrogen-getter under conventional synthesis conditions. At higher

temperatures, however, aluminum becomes a more efficient nitrogen-getter than
titanium as shown by diamond synthesis experiments at 1550°C where the use of
titanium yielded yellowish and heavily included crystals [62].
Boron is often added in the form of a stable metal boride to gettered reaction
volumes, to generate blue semiconducting diamonds (type IIb) [62].
1.2.5.2 Purity of Large Synthetic Diamonds Synthetic diamonds invariably contain
,
\
,
', Al
\
Ti
--_
Added amount of Al or Ti to the solvent-catalyst(wt%)
Figure 15. Nitrogen concentration of diamond grown with A1 or Ti added as a nitrogen getter [67].
1.2 The Crystullization of' Diamond 501
impurities. These impurities fall into two classes. There are impurities on an atomic
scale (point defects) and these include nitrogen, boron, and metallic elements such as
nickel. The second class of inclusions are present in the form of metallic or nonme-
tallic particles, both microscopic and macroscopic. The occurrence of these inclu-
sions is controlled by crystal growth kinetics.
1.2.5.2.1 Inclusions Solvent/catalyst metal inclusions can occur in large synthetic

diamonds. The number, size, and shape of these inclusions depends to a large
extent on the growth rate and the synthesis temperatures and pressures. Extensive
research in this area shows that the uptake of metallic inclusions occurs predomi-
nantly during the early stages of growth when the carbon flux density is very
large, particularly when a small seed crystal is used [68]. One suggestion is the use
of a two-stage growth method [63, 691 in which a small recess is made on the seed
pad and the seed is located at the bottom, such that it does not protrude above
the recess. The net result is that carbon supply to the seed is limited as the seed
grows within the recess during the initial stages of growth. Once the seed has
grown out of the recess, it has an enlarged surface which can accept more carbon
without compromising quality. The uptake of inclusions during the early stages
of a growth can, therefore, be minimized by paying special attention to the depth
and diameter of the seed recess.
The use of larger seed crystals tends to increase the growth rate. The coupled effect
of growth rate and seed crystal size [70] on the uptake of inclusions is summarized in
Fig. 16 where it is evident that the size of the seed crystal is an important parameter in
the growth of high purity diamonds. In general, the population and size of inclusions
increases with an increase in growth rate. There are many different types of inclusions
that occur and most of these have been comprehensively dealt with elsewhere [44].
One of these types of inclusions results from hoppered growth when there is a high
supersaturation [71], as shown in Fig. 17. Low growth rates and stable growth condi-
tions are therefore important for the growth of high purity large synthetic diamonds.
The choice of solvent/catalyst alloy which incorporates a getter is important
in controlling the quality of grown diamond. In the growth of near-colorless
diamonds, the tendency for the getter elements to form stable carbides can be a
problem. When titanium is used, a very stable carbide is formed whilst, with alumi-
num, the A14C3carbide formed is readily decomposed at diamond synthesis condi-
tions. Inclusions therefore occur more readily with titanium than with aluminum
gettered systems. At high levels of titanium or aluminum, inclusion content can
be a problem. When titanium is used, many fine particles of TIC are formed
[72, 731 and included in the growing diamond crystal in addition to the solvent/cat-
alyst metals (Fig. 18). It has been suggested [73] that the addition of copper to the
reaction volume has the effect of suppressing the formation of titanium carbide.
Copper is thought to form a cation in the molten solvent/catalyst bath which can
decompose the carbide.
1.2.5.2.2 Point Defects Nitrogen and boron are ubiquitous atomic-scale impurities
in synthetic diamonds. The free energy of formation of nitrides and borides in the
solvent/catalyst melt will determine the extent to which these impurities are
' Crystals containing inclusions
Seed crystal (100) face area (mmz)
Figure 16. Dependence of critical growth rate of diamond crystals on { 100) faces of seed crystals, on
varying seed size [70].
available for incorporation into ie growing crystal.

As mentioned in Section 1.2.5.1.1, unless nitrogen-getters are used during synth-
esis, nitrogen is responsible for the highest concentration of impurity-related point
defects in synthetic diamond. This impurity is incorporated principally as isolated
substitutional atoms which give rise to a strong optical absorption below 500 nm,
leading to a yellow or yellow-brown color. Such diamonds are classified as
Figure 17. Cross section of a cobalt-titanium grown crystal with inclusions resulting from hoppered
growth.
1.2 The Crystullizution of Diamond 503
Figure 18. Micrograph of cross-section of a carbon, solvent/catalyst mixture Co-Fe-Ti showing

large Tic crystallites and some Tic particles that have been trapped at the diamond-solvent/catalyst
interface.
type Ib. Synthetic diamonds tend to take up impurities at different concentrations

for each of the types of growth sector (regions of the crystal with a common crystal-
lographic growth plane). Under standard synthesis conditions, nitrogen concentra-
tions are highest in { 11 1) growth sectors, typically 100-300 p.p.m. for iron-cobalt
grown crystals [62]. Generally, the average nitrogen concentration in {loo} sectors
is about half that for { 11 l} sectors [73, 741. However, at low growth temperatures,
cube growth predominates and the nitrogen concentration in {loo} sectors exceeds
that in { 11 l} sectors [75]. The reduced solubility of the solvent/catalyst for nitrogen
at low temperatures has led to nitrogen concentrations up to about 1000p.p.m. for
the ironxobalt system. Nitrogen concentrations are substantially less in the minor
growth sectors ({I lo}, (1 13) and occasionally { 115)) compared with { 11 l} and
{ IOO}, with the { 110) sectors containing about 1 p.p.m. If nitrogen getters are
used to reduce the amount of nitrogen available to the growing diamond, as
described in Section 1.2.5.I .2, the residual nitrogen concentration tends to be
approximately proportionally similar to the ungettered case.
Blue, semiconducting type IIb diamonds are synthesized by adding both boron,
and a nitrogen getter to the reaction volume. The total boron acceptor concentra-
tion is highest for { 11 l} sectors, somewhat less for { 1 lo} sectors and substantially
less for the remaining types of growth sector [74-761. The presence of uncompen-
sated boron acceptors gives rise to a blue coloration. The strength of the blue
color, or indeed any residual yellow color, depends upon the difference between
the boron acceptor and nitrogen donor concentrations for any growth sector.
When nickel or cobalt is used as the solvent/catalyst, these elements are incorpo-
rated in the diamond lattice as optically active point-defects, but only in (111)
growth sectors [77, 781.
Frequently, when large, high-quality synthetic diamonds are discussed, the ques-
tion of their potential use as gemstones is raised. At the time of writing, none of the
major manufacturers of synthetic diamonds markets material for gem use. How-
ever, small numbers of cut and polished synthetic diamonds, mostly produced in
the former Soviet Union using the so-called BARS equipment, have appeared in
the gem trade in recent years. The majority have been yellow in color but a few
near-colorless synthetic diamonds have also appeared. As uncut crystals, it is
easy to distinguish synthetic diamonds from natural diamonds by their distinctive
morphology and the presence of a seed. But cut and polished synthetic diamonds
are also readily identifiable as such by standard gemmological techniques such as
microscopy of inclusions and of features on polished faces (due to sector-
dependent differential hardness) and zonation of color or fluorescence (due to
sector-dependent differential uptake of optically active impurities) [79-841. Near-
colorless and boron-doped, blue in color, synthetic diamonds tend to exhibit
strong, long-lived phosphorescence. To enable large numbers of diamonds to be
screened rapidly for the presence of synthetic diamonds, instruments have been
developed. The first of these instruments detects an optical absorption line at
415 nm present in the majority of natural diamonds. Those diamonds not exhibit-
ing this feature may be examined using a second instrument which, by exciting
near-surface fluorescence with very short wavelength ultraviolet light, produces a
fluorescence pattern from which the growth structure of the diamond may be
inferred. These patterns are very different for synthetic as compared to natural dia-
mond [85].
1.2.6 Novel Diamond Synthesis Routes

J. 0. Hansen
1.2.6.1 Introduction: Thermodynamically Stable and Metasfable Processes

The two major innovations in diamond technology in this century must be the
reproducible high pressure process with metal solvent/catalyst, and the low pressure
sustained growth from methane-hydrogen plasma reviewed by Kamo et al. [86]
which we call CVD. There may be some argument that the work of Angus et al.
[87] or the patents of Eversole [88] predate and anticipate metastable diamond
synthesis by CVD, but the authors did not achieve sustained growth. They did
demonstrate that the decomposition of methane or carbon monoxide can produce
active carbon atoms or hydrocarbon fragments, which in a suitable temperature
range and in the presence of diamond surfaces to act as nuclei, will permit over-
growth of a new diamond phase. The appearance of graphite nuclei leads to a cessa-
tion in diamond growth which suggests that the success of the ‘metastable’ growth,
without plasma methods, depends upon an energy barrier to nucleation of graphite.
In the now mature CVD technology, graphite nucleation is inhibited by the contin-
uous bombardment of the growing surface by atomic hydrogen, which has a greater
tendency to etch graphite than diamond [89]. Even so, it appears that certain ‘tricks
of the trade’ are required to initiate nucleation of diamond on a nondiamond sub-
strate because of a substantial energetic barrier.
The graphitediamond equilibrium line in Fig. 7 is based on a thermodynamic ana-
lysis by Berman and Simon [ 141, which has been reviewed recently [90] in the light of
new data, particularly the direct electrochemical determination of the free energy dif-
ference for the diamond to graphite transformation [91]. From the equilibrium expres-
sion of Berman and Simon, Eq. (1) in Section 1.2.2, it may be possible to include
additional terms which offer an explanation for some cases of novel non-CVD synth-
esis, or to expand the existing terms such as AH', the enthalpy change at zero pressure.
1.2.6.2 Increase in Enthalpy: Synthesis from Nongraphitic Carbons

Onodera et ul. [30] and Brannon and McCollum [92] have published evidence of dia-
mond growth from nongraphitic carbons at pressures below the phase-equilibrium
line at 2 4 GPa and 1300-1900°C. They suggest as explanation an increased chemi-
cal potential (free energy) of the nongraphitic phase with respect to diamond, or the
catalytic action of the hydrogen associated with the cokes, and not present in gra-
phite, by analogy with diamond CVD. The second explanation seems unlikely,
since hydrogen has been shown to be a deleterious impurity in metal solvent/catalyst
synthesis from graphites [93, 941.
The enthalpy of nongraphitic carbon (furan coke) has been measured at low tem-
perature [95,96] and found to be higher than both graphite and diamond, by between
0.4 and 2.8 kJ mol-* . This would have the effect of eliminating or reversing the sign of
the first term in Eq. (1) and so decreasing the pressure requirement for diamond synth-
esis, perhaps as far as point B in Fig. 7. As Vereschagin et al. [96] point out, there is an
even greater driving force for graphite crystallization. Even in synthesis from synthetic
graphites of high graphitization index (dOo2= 0.337 nm), there is a competitive process
of recrystallization of graphite as coarse single crystal flakes [56]. See also Fig. 19.
The increased enthalpy of the nongraphitic carbon follows from a higher internal
energy because of lower total bond energy or equivalently, a higher concentration of
defects by comparison with single-crystal graphite. A similar argument has been
used by Bar-Yam and Moustakas [97] for the case of CVD synthesis. High vacancy
concentrations in graphite and diamond can cause diamond to become the stable
phase because the formation energy of vacancies in diamond is lower than that in
graphite.
The possible success of a synthesis route using nongraphitic carbon would seem to
depend upon the suppression of graphite nucleation and purity factors, since growth
of high quality diamond requires low levels of impurities. This is more difficult to
achieve in a carbon which has not been graphitized.
1.2.6.3 Interfacial Energy: Growth in the Colloid Size Range

Ideal diamond has an exceptionally high surface energy because of the high density
of strong bonds which must be destroyed in order to generate unit area of surface. It
has been estimated at 5000-9000 mJ mP2 depending upon the crystal face [98].
Graphite has a lower average surface energy but there is extreme anisotropy, with
the basal plane having an energy of 150mJm-* and the prism planes over
4000mJm-* [99]. This explains the tendency for graphite to crystallise as large
thin flakes, so reducing the contribution of the edge surfaces to the total surface
506 I Diamond Muteriuls and their Applications
Figure 19. Micrograph of a section of a diamond synthesis capsule in polarized light at 200x mag-
nification. Recrystallization of coarse graphite flakes of high crystallinity is a process which
competes with diamond growth even in the diamond stable region. The tabular shape suggests
that surface energy is a significant driving force for recrystallisation even in this size range.
(interfacial) energy.
In practice, these solid-vacuum interfacial energies are not likely to be seen, since
the diamond surface and the graphite prism-face surfaces will reconstruct or react
with adatoms such as hydrogen and oxygen to reduce their surface energy. Under
atmospheric conditions, diamond is terminated by a range of hydride and oxide
groups and is a low energy surface [loo].
During high pressure solvent/catalyst synthesis, the interfacial energies of metal
melt-diamond and metal melt-graphite have values lying between the metal surface
tension, 1800mJ mP2,and the solid-vacuum surface energy. Some values have been
estimated from contact angle data [loll although the complexities of contact angle
work in the diamond stable region are formidable.
By including a surface energy (interfacial energy) term in the equilibrium expres-
sion it is possible to make some interesting predictions.
1
AG', = A H ! - T L I S ; , ~ A VTd P + A(yA) = 0
where y is the interfacial (surface) energy of any phase and A is its molar area, which
is not an intensive variable like pressure or temperature but varies with both the size
of the particles and the density of the phase, diamond having a lower molar area for
the same particle size than graphite.
For instance, using this approach, Nuth [ 1021 has predicted that diamonds (in
vacuo) are more stable than graphite in the size range 1-5nm, a result which was
supported by Badziag et al. [lo31 using a different approach.
There is experimental support for this in the finding that nanometre diamonds are
abundant in primitive meteorites and possibly in the interstellar dust clouds [ 1041.
Similarly, growth of graphite in the diamond-stable region during high pressure,
high temperature synthesis, and the corollary, an overpressure for nucleation of
diamond, are explainable using the additional interfacial energy term without the
need for kinetic arguments [105, 1061. Once the growing particles are in the size
range of microns, the molar surface area becomes too small for this term to have
any effect and the classical phase equilibrium line is applicable.
There is another way of looking at the same effect, since a surface energy per unit
area is equivalent to a line force per unit length, or surface tension, hence there is an
excess pressure in a spherical droplet given by the Laplace equation:
A P = -2Y (3)
Y
where y is the surface energy of any phase and r is the radius of curvature. For diamond
with sigma at 5 J mP2and Y at 1 nm, the excess pressure in a nanometre sized sphere is
5 GPa, which would be sufficient to stabilize the diamond phase to at least 1600K.
This could explain processes such as laser pyrolysis of hydrocarbons [ 1071, where
small quantities of nanometre sized diamond have been detected in the carbon-
aceous products. A recent result suggests that it is even possible to generate and
stabilize a diamond core in the centre of a graphitic ‘onion’ a giant fullerene, by
electron irradiation and annealing of the outer graphite layers. In this process,
‘the sputtering-induced loss of atoms and the closure of the shells around vacancies
lead to extreme surface tension, which is finally responsible for the spherical shape of
the onions’ [108].
This publication is a good example of the development of analytical techniques
such as high resolution electron microscopy lattice imaging [1081 and micrometre-
area Raman spectroscopy (microRaman) [ 1091 which permit the researcher to
determine that diamond is present in crystallites of submicrometre, perhaps even
nanometre size. Early scientists like Moissan [I 101 and Hannay [I 113 performed
experiments in the pressure and temperature ranges now being cited as feasible
for hydrothermal synthesis of diamond [38, 112, 1 131. Had they been equipped to
use the analytical techniques now available, they may have been proven correct in
their claims, but in a particle size regime much finer than they were able to inspect.
1.2.6.4 Non-equilibrium Thermodynamics: Superequilibrium Atomic Hydrogen

The process of CVD of diamond may be explained by kinetic arguments [89], or
alternatively by the nonequilibrium thermodynamic coupling theorem [ 1 141.
In the latter case, it is the presence of a superequilibrium concentration of atomic
hydrogen and the coupling of the process of recombination of hydrogen atoms with
the processes of diamond growth which allow the diamond deposition reaction to
occur in a direction contrary to that prescribed by conventional thermodynamics.
Diamond synthesis processes which are characterized by continuous generation of
atomic hydrogen or halogens would then fall into the class of nonequilibrium
processes. The generation of atomic hydrogen could be by means of microwave
or discharge plasma, flame or even a hot filament. This topic is covered more
fully in Chapter 4 of Part I1 of this book (C.-P. Klages), but some special cases
will be listed here.
Table 2. Classification of low-pressure synthesis methods
Nominally metastable Truly metastable: explanations in Undetermined: perhaps

thermodynamic terms of nonequilibrium controversial
explanation is possible thermodynamics, atomic H
HPHT: nongraphitic LPSSS (low pressure solid state Triple laser process using
carbon [30], [92] source) as by Roy et al. [ 1 16, 1 171 CO, (QQC process) [ 131
Hydrothermal synthesis - Hydrothermal flow synthesis with PVD using ion beams -
nonflow [113, 381 flame [116, 1171 reviewed by Prins [123]
Laser in absence of Laser in presence of hydrogen [ 1181 Precipitation from
hydrogen [105, 1151 nickel-sodium hydroxide
melts [ 1241
Diamonds in interstellar Fluidized bed deposition on
space [102-1041 particles from acetylene flame [ 1191
Electron irradiation of Discharge electrolysis from
carbon ‘onions’; nanometre water:organics mixtures [120-1221
scale HPHT [ 1081
1.2.6.5 Classification of Low-pressure Methods

Using this criterion, it is possible to separate those methods which are nominally
metastable, but for which there is an explanation in terms of the additional free
energy components discussed in Sections 1.2.6.1 to 1.2.6.3, from those which are
truly metastable. There is a third category, undetermined, where the authors have
insufficient knowledge of the process to make a judgement. Perhaps these are
truly different in their mechanism.
1.2.6.6 Extension of High-pressure Methods

In Section 1.2.3 we reviewed the design of high pressure systems for diamond synth-
esis. The major development in recent decades has been an almost thousand-fold
increase in working volumes. The pressures achieved using cemented carbide
technology have, however, not increased significantly. The advent of the dia-
mond-anvil-cell DAC [1251has allowed researchers to attain pressures of thousands
of kilobars, whilst simultaneously collecting spectroscopic data on phase changes.
This is possible only because of the transparency of the diamond anvils. Most
research on ultrahigh pressure has been done with DACs. There might be two
reasons for industrial interest in pressures of 10GPa. The first is the synthesis of
diamond from nonmetallic solvents [24, 1261 such as carbonates at pressures of
7.7GPa and 2000°C. This might offer different and novel types of diamond to
those presently commercially made.
The second is the possibility of direct conversion, solvent free, of graphite to
diamond. Table 3 lists reported conditions for the direct conversion.
The direct conversion is dependent upon the crystallinity of the starting carbon
[127], and the product, when analysed by XRD, appears to be hexagonal diamond.
Figure 20 shows the martensitic transformation, by compression in the c-direction
and in-plane buckling, of two crystalline forms of graphite. Hexagonal graphite,
Table 3. Direct conversion of graphite to diamond
Authors Apparatus Date Pressure and Temp Comments
Bundy and Kaspers [I271 Modified belt 1967 13 GPa and1000"C Hexagonal diamond
Boehler [I281 DAC 1994 10 GPa and 1700°C
Xu and Huang [ 1291 DAC 1994 6GPa and 1700°C Raman at 1333 cm-'
Figure 20. Computer-generated perspectives of the martensitic transformation processes, hexagonal

graphite to hexagonal diamond, and rhombohedra1 graphite to cubic diamond. The structure on the
right is generated by compacting the structure on the left in the vertical direction and puckering
those atoms that find a neighbor in an adjacent plane so as to encourage the formation of the
new sp3 bonds in the vertical direction [132]. Software designed by Pawel Wzietek, Massy, France.
5 10 I Diamond Materials and their Applications
which has layers repeating in the order ABAB . . ., transforms to hexagonal

diamond, or lonsdaleite. This is a metastable form of diamond and is never seen
in diamond synthesized from a solvent/catalyst. Higher pressures and temperatures
are required for the conversion of hexagonal graphite to cubic diamond, above line
C, in Fig. 7, which suggests that the mechanism requires melting of carbon [130].
Rhombohedral graphite has a layer sequence ABCABC.. . (Fig. 20) and can
convert to cubic diamond by a martensitic process of compression and buckling.
Rhombohedral graphite is metastable towards hexagonal graphite and rare except
in some natural crystals. Shear forces in high pressure or shock synthesis processes
could cause rapid conversion of small domains of hexagonal graphite to the
rhombohedra1 form, after which the direct conversion to cubic diamond may
proceed. It is unlikely that this could happen over domains much larger than the
100 nm crystallite size seen currently in shock-synthesized diamond [13I].
1.2.6.7 Summary and Conclusions

There have been many innovations leading to growth of nanometre sized crystals of
diamond. A few processes, like the LPSSS and Me-C-H [117] appear to offer the
possibility of combining the advantages of low-pressures with the commercial
need for a volume-based rather than a surface-based process, and of growing
crystals of micron rather than nanometre size. Direct conversion of graphite to
diamond at macroscopic sizes will depend upon the synthesis and stabilisation of
suitable graphite or mixed sp2/sp3precursors.
1.2.7 Cubic Boron Nitride Crystallization

G. J. Davies
Cubic boron nitride (cBN) is presently the second hardest material known to man
after diamond. The compound boron nitride is unknown in nature. The combina-
tion of atoms of boron and nitrogen in stoichiometric arrangements is such that
B-N is isoelectronic with C-C. It is not surprising that, by analogy with carbon
structures, similar crystalline structures of boron nitride occur. By analogy with gra-
phite, hexagonal boron nitride is a soft, slippery material with an anisotropic, hex-
agonal structure made up of sheets containing interlinked six-membered rings
(3 x B-N). There are some differences however, notably, the n-orbital overlap of
graphite is absent, so hBN is an electrical insulator and is white in color. For kinetic
reasons, all conventional chemical synthetic routes for making boron nitride result
in the hexagonal structure. Again, by analogy with carbon, boron nitride can take
up the diamond structure with alternating atoms being boron and nitrogen. This is
the common zinc blende or sphalerite structure. This phase of boron nitride has a
hardness about half that of diamond; and a bulk modulus of about 367 GPa com-
pared to the 435 GPa of diamond.
The value of cBN as an abrasive lies in its much higher oxidation stability com-
pared to that of diamond (1200°C compared to 600°C) and in its reduced chemical
1.2 The Crystallization of Diamond 5 11
Figure 21. Phase diagram of boron nitride showing the latest stability fields suggested by Soloz-
henko [ 1341 compared to the earlier Bundy-Wentorf diagram [ 1381.
interaction with ferrous materials when compared to diamond. It is, therefore, the
abrasive of first choice for the grinding of hard, ferrous materials.
When the formation of cBN was first reported [133], the pressure and temperature
phase diagram was believed to be very similar in form and stability field positions to
that of carbon, with the equilibrium line between the hexagonal and cubic phases
intercepting the pressure axis when temperature is at absolute zero [134]. The
most recent phase diagram for boron nitride is that of Solozhenko, shown in
Fig. 21 [135]. This diagram was initially calculated and subsequently partially
experimentally substantiated. The hexagonal-cubic-liquid triple point is at about
7 GPa and 3800 K. The hexagonal-cubic equilibrium line intercepts the temperature
axis at just above 1500K. However, the exact temperature of the intercept is not
well-established [136], although it is now generally agreed that cBN is the stable
form of boron nitride at room temperature and pressure. This is where boron nitride
differs significantly from diamond. It does not transform to the hexagonal structure
until temperatures of at least 1200°C are exceeded.
cBN is synthesized using the same general technology as that used for commercial
diamond synthesis. A mixture of hBN as source and various catalyst/solvents are
subjected to pressures in the range 4.5-5.5 GPa and temperatures between about
1500 K and 2000 K. The elements initially investigated and demonstrated to
function as solvent/catalysts were the alkali and alkali earth metals, as well as
antimony, tin, and lead [137]. Subsequently, the nitrides of lithium, calcium. and
magnesium were shown to be more effective. Single crystals of cBN, in the size
ranges appropriate to grinding applications, both black and amber in color, are
now readily synthesized by several companies on a commercial basis.
The recent developments in the boron nitride phase diagram, particularly with
respect to the room temperature and pressure thermodynamic stability of cBN,
have stimulated work aimed at producing cBN over a wider range of conditions
[136]. In conjunction with this, it is now known that a wide range of nitrogenous
5 12 1 Diamond Materials and their Applications
compounds are capable of aiding the transformation of the hexagonal to the cubic
form [138].
1.3 Polycrystalline Diamond and Cubic Boron Nitride

S . Ozbayraktar
1.3.1 Natural Polycrystalline Diamond

Although the first natural polycrystalline diamond was discovered in the 19th cen-
tury, it was only investigated on a scientific basis during the 1970s, and was classified
into two broad types: Carbonado and Ballas [139].
Carbonado is ‘a porous, randomly polycrystalline diamond aggregate’ [ 1401.
Ballas on the other hand is defined as ‘polycrystalline diamond of oriented globular
growth’. Its crystallites have (110) directions oriented radially.
Experiments so far have clearly indicated that these natural forms of sintered
diamond are truly polycrystalline ceramics, exhibiting transgranular fracture as a
result of diamond to diamond bonding.
1.3.2 Synthetic Polycrystalline Diamond

In 1958, Hall [141] discussed the desirability of preparing a ‘cemented diamond com-
position analogous to WC’ and hinted that experiments to produce polycrystalline
diamonds were underway. But it was not until 1970, when he reported details of his
procedures [1421, that he established experimentally practical pressure and tempera-
ture fields where pure diamond powder can be sintered within times ranging from
several days down to about one second. He mentions hard refractory materials
like borides, carbides, nitrides and oxides as suitable binders.
In December of the same year, Stromberg and Stephens [143] published a paper
titled ‘Sintering of Diamond at 1800-1900”C and 60-65 kBar’. Their manuscript
was originally dated 17 December 1969. In their work, natural diamond powders
of sizes 0.1-10pm were carefully cleaned in several solvents and then heated at
500°C in a vacuum of torr in order to remove adsorbed gases. The cleaned
particles were then loaded, under argon, into a similarly cleaned and outgassed
tantalum crucible which was sealed by electron beam welding. The capsule was
heated under pressure and temperature for 1 h. They used indium as a ‘cushion’
to protect the diamond specimen on unloading. Small amounts (less than lwt%)
of boron, silicon, or beryllium were found to aid the sintering process. Most of
the sintered specimens gave hardness values of over 7000 kg mm-*, with a maximum
of 8800 kg mm-’.
I .3 Polycrystalline Diamond and Cubic Boron Nitride 5 13
In 1971, Katzman and Libby, from the University of California at Los Angeles
[144], described the formation of sintered diamond compacts in the presence of 20
vol% of cobalt binder. The sintering was carried out for about 20 min at 6.2 GPa in
a piston and cylinder type high pressure-high temperature apparatus, and at a
temperature (159OOC) which was above the cobalt-diamond eutectic, 1570°C at
62 kBar. They claimed their results suggested that the cobalt cleans the diamond
surfaces of any adsorbed gases that might prohibit or retard grain growth or
fusion, thereby making unnecessary the surface cleaning and degassing procedures
described by Stromberg and Stephens. They also claim that the optimum cobalt
content was 20% by volume. An attempt to substitute cobalt with nickel under iden-
tical sintering conditions also led to a somewhat softer material. Compacts made
with diamond grit sizes of 1-5 pm and 0-2 pm were harder than those made with
grit sizes of 10-20 pm.
Wentorf and Rocco in 1973 [145] described in situ diamond sintering on top of a
layer of cemented tungsten carbide (WC). The idea here was that any solvent/catalyst
metal (cobalt) which was present in the metal carbide would act as a source for liquid
phase sintering. The sintering method involves placing in a cylindrical mold, a 0.5-
1 mm layer of diamond powder on top of a thicker layer (lomm) of WC-lOwt%
cobalt powder. The mold is then subjected to high pressures (>5 GPa) and tempera-
tures (1400-1600°C) for at least 1Omin to 1 h. Under these conditions, the cobalt
melts and liquid phase sintering of both the carbide and the diamond occur.
Veraschagin et al. [146], in 1975 showed that, if the metallic binder was initially in
the form of a disc (rather than a powder), better packing densities were achieved and
the molten binder would thus be found mainly in the voids rather than in between
diamond grains, thus resulting in strong and uniform diamond-to-diamond bonding
throughout the compact.
1.3.3 Mechanisms involved in Polycrystalline Diamond

Manufacturing Process
The pcD manufacturing process can be roughly divided into three stages: cold com-
paction, hot compaction, and liquid phase sintering of the diamond compact.
1.3.3.1 Cold Compaction

Kolomiitsev [147] and Uehara [I481 studied the compaction behavior of diamond
compacts under high pressures. Kolomiitsev showed that there are basically three
processes taking place during cold compaction, see Fig. 22: particle rearrangement,
crushing of diamond particles, and filling of voids by crushed particles.
At a given pressure, coarser particles crush more than finer particles. Their aver-
age particle size changes drastically and the particle size distribution becomes truly
bimodal. At a given pressure, porosity of coarser powders is lower than that of finer
powders. The change in porosity with pressure is bigger at lower pressures than at
higher pressures for a given starting particle size. Pore size of coarser powders is
Figure 22. Schematic description of processes taking place during cold compaction.
larger at a given pressure. But the difference between coarse and fine grains gets
smaller as pressure is increased.
Crushing of diamond particles during this stage dramatically shifts the particle
size distribution.
The powder, which has a lower number of sharp edges per particle, is more stable
to compression forces than the one with many spallings, sharp edges, etc. Since their
weakest planes are the planes where they contain inclusions; the density, size and
shape of the inclusions may be a critical factor in the crushing of diamond particles
during the cold compaction stage.
1.3.3.2 Hot Compaction

We describe hot compaction as the stage where temperature is applied together with
pressure but the temperatures are not high enough to start infiltration/liquid phase
sintering processes.
During hot compaction the following processes are believed to be taking place, as
shown in Fig. 23: graphitization of the diamond surfaces facing the voids, plastic
deformation of diamond grains, and densification of the compact (shrinkage of
pores).
4
pressure
temper fure
Figure 23. Deformation of diamond grains under high pressure and temperature
Once the heat is turned on at high pressures, the densification of the diamond
compact proceeds mainly by further crushing and rearrangement of the crushed par-
ticles up to 700°C [149]. After 700°C, densification proceeds with plastic deforma-
tion [150-152]. The extent of plastic deformation is a function of temperature and
pressure. At higher temperatures, the diamond grains become more round, their
sharp edges disappear, they start to deform in the zones of contact with each
other. Pores get smaller, even diminish at a rate initially determined by their size
at the end of the cold compaction stage. Pore size reduction rate is faster for coarser
pores than for smaller pores [149]. But the time required for a pore to disappear is
constant at a given pressure and temperature to a first approximation, regardless of
size [153, 1541.
Also at higher temperatures, graphitization of the diamond surfaces, which are
not in contact with another diamond surface, takes place. The amount of graphiti-
zation is a function of the temperature at a given pressure [150]. The degree of
graphitization in diamond compacts can be related to the specific resistivity of the
compacts since graphite is a relatively conductive material compared to diamond
which is an insulator. That is, the more graphitization takes place, the more conduc-
tive the material becomes.
At even higher temperatures, bulk graphitization of diamond grains starts to take
place and the hardness and wear properties of pcD produced (binderless) start to
degrade [ 1501. At a given temperature, if the pressure is increased, graphitization

seems to decrease and the properties improve. One can conclude from these obser-
vations that, at a given plastic deformation, the degree of graphitization determines
the change in hardness and wear resistance. But when both plastic deformation and
graphitization are taking place, properties are more sensitive to plastic deformation
until bulk graphitization starts taking place. Finer particles tend to graphitize more
than coarser particles at a given pressure and temperature. This is most probably
due to their larger surface area in contact with the pores.
1.3.3.3. Surface Chemistry Effects

Surface chemistry of the starting diamond particles has an effect on the graphitiza-
tion kinetics of the free surfaces of diamond particles. Surface adsorbed oxygen gas
reacts with diamond and speeds up the process of graphitization on these surfaces
[149]. This may be due to the fact that the activation energy for the process of oxida-
tion is lower than the activation energy for the process of graphitization without
participation of the gas phase. Moreover, such graphitization may begin at lower
temperatures.
1.3.3.4 Liquid Phase Sintering

There are a number of binders used for pcD sintering. The most commonly used
ones are iron group materials which are known to be good solvent/catalysts for dia-
mond synthesis, namely cobalt, iron, nickel, and manganese or various combina-
tions of these.
Utilization of binder materials for the sintering of diamond compacts has three-
fold benefits: decreasing of sintering temperatures and pressure, cleaning diamond
particle surfaces of graphite, and electron discharge cuttability for tool making.
There are various methods of adding binder material into diamond compact:
mixing it with the diamond powder, coating diamond particles with it, infiltrating
from a disc of binder metal, and infiltrating from a substrate containing binder
metal.
We will concentrate here on the mechanism of infiltration from a substrate
because it is the most commonly employed method in the industry.
There are basically four stages in the infiltration of diamond compacts [ 155, 1561.
During first stage, temperature is just sufficient to cause melting of Co-W-C at
the interface due to excessive amounts of carbon which cause the eutectic tempera-
ture to drop. The amount that is melted during this stage immediately infiltrates into
diamond due to the very high pressure gradient between the porous body and the
solid interface. However, this stage is very short and it is rapidly followed by the
second stage.
The second stage corresponds to melting of the binder in the bulk of the substrate.
As the temperature increases, it reaches the eutectic temperature of binder in the
substrate and binder becomes liquid and starts infiltrating into diamond compact
again under the same pressure gradient which is roughly equal to the capsule inter-
nal pressure.
As the pores in the diamond layer become filled, the magnitude of the pressure
gradient is decreased. Accordingly, the rate of mass transfer is decreased,
although its direction is maintained. As this pressure gradient decreases, the
effect of the pressure gradient in the reaction volume of the capsule, which was
found to be about 1 GPa in the axial direction by Russian researchers in their
own capsule [ 1571, starts dominating the infiltration gradient. Whatever the direc-
tion of the pressure gradient is, this stage of mass transfer is completed when
there is no pressure gradient at the interface between diamond layer and the
substrate and when the pores are completely filled. Thus, the rate of mass transfer
approximates zero and the binder content in the diamond layer becomes a
maximum.
The graphitization of the diamond surface, described in Section 1.3.3.2, is impor-
tant in terms of infiltration of the solvent/catalyst binder. Metals such as cobalt dis-
solve graphite more readily than diamond at the sintering pressures (Fig. 9b).
Hence, the more graphitization that has taken place, the more cobalt is expected
to be in the final material.
A characteristic feature of the fourth mass transfer stage is a change in its direc-
tion: it has been observed [ 155-1 571 that the quantity of liquid phase in the diamond
layer is decreased and increased in the substrate. The beginning of mass transfer
indicates that a pressure gradient arises in the system, directed from the diamond
layer to the substrate. The origin of this phenomenon was explained by Shulzhenko
et al. [ 155, 1561 in terms of temperature gradients and their changes resulting from
the surface tension of the molten binder-diamond interface.
Also, this effect can be explained by the fact that, as the sintering of diamond
compact proceeds, the area of the contacts between the diamond grains increases.
This accelerates shrinkage and the densification process. This in turn according to
Shulzhenko et al. squeezes the excess cobalt, which cannot be accommodated in
the ever shrinking pores and cavities in between diamond grains, out of the diamond
layer. One destination of this stage of mass transfer is the diamond/substrate
interface.
As a result of the change in direction of binder mass transfer in the fourth
stage, the binder content of the diamond layer is decreased by an amount of the
order of couple of percent [155, 1581. The binder content of the diamond layer
can be controlled by lowering the sintering temperature and extending the sintering
time.
Almost complete exclusion of cobalt from large areas where two grains have
grown together is observed by TEM studies on polycrystalline diamonds sintered
with cobalt binder [159]. No thin, residual layer of cobalt was detected at the
grain boundaries. The thickness of regrown diamonds is usually much lower than
the original diamond grains it is precipitated onto, and it joins with the regrown
diamond precipitated onto a neighboring diamond grain forming a high angle
grain boundary. So it can be said that only the surfaces of the diamonds in contact
with a pore just before infiltration takes place are potential candidates for
reprecipitation of new dislocation free diamond. The areas next to the diamond
to diamond contact points or surfaces are the preferred locations for diamond
reprecipitation due to lesser surface energy requirements.
1.3.3.5 Conclusions
During cold compaction of diamond powders, coarser powders densify more and
have larger size pores than finer powders at a given pressure.
During hot compaction, the densification of the diamond compact proceeds
mainly by crushing and rearrangement of the crushed particles up to 700°C.
Above that temperature, the densification proceeds with plastic deformation. The
extent of plastic deformation is a function of temperature and pressure. Pore size
reduction rate is faster for coarser pores than for smaller pores.
The degree of plastic deformation and of graphitization taking place during hot
compaction seems to influence the wear resistance and hardness of the sintered dia-
mond compacts. When both plastic deformation and graphitization are taking
place, properties are more sensitive to plastic deformation until bulk graphitization
of diamond grains starts.
lnfiltration rate of the metallic binder from the substrate is influenced by pressure
gradients created by the applied external load, pressure gradients inside the capsule
and the pore size of the diamond compact at the onset of infiltration, temperature,
and the substrate binder mean free path and chemistry.
1.3.4 Polycrystalline Cubic Boron Nitride

As referred to in Section 1.2.7, cubic boron nitride (cBN) was first synthesized in
1957 [133].
Polycrystalline cubic boron nitride is preferable over single crystal cBN because it
can be manufactured in bigger sizes and has higher fracture toughness. However,
like its counterpart pcD, it can only be sintered at very high pressures and tempera-
tures with the aid of binders because of its strong covalent bonds.
pcBN, having a hardness second only to pcD in terms of polycrystalline materials,
and being relatively less reactive with ferrous metals, is the cutting tool material of
choice for a wide range of applications such as finish and interrupted cutting of
hardened steel, gray, and ductile cast iron machining and finds a wide range of
applications in the production of automotive parts. These applications will be
discussed further in Section 1S.4.
Commercially, pcBN tool materials are produced using sintering additives and/or
binding materials at pressures of 4-6 GPa and at temperatures of 1200-1500°C. The
sintered pcBN has a higher thermal stability than pcD.
The first polycrystalline cBN material was sintered by Wentorf and Rocco [I 601 in
1971 on a cemented carbide substrate using alloys of nickel, cobalt and iron with
aluminum as binders. cBN to cBN bonding is believed to be achieved by a liquid
phase sintering process involving these binder alloys.
In February, 1980, Sumitomo from Japan filed the patent ‘Sintered compact for a
machining tool and a method of producing the compact’ [161]. This patent basically
covers any compact with 1 W O vol% cBN and a balance of binder material that can
comprise any carbides, nitrides, borides, or silicides of metals of groups IVa, Va, or
VIa. Specifically mentioned are titanium, zirconium, hafnium, vanadium, niobium,
I .3 Polycrystalline Diamond and Cubic Boron Nitride 5 19
Table 4. Properties of the pcBN products and other cutting tool materials.
DBC Amborite* SYNDITE WC A1203 Sialon

50* 010 + TIC
Density [g ~ m - ~ ] 4.297 3.41 4.127 14.7 4.28 3.20

Compressive strength [GPa] 3.55 2.73 4.74 4.50 4.50 3.5
Fracture toughness [MPa.m'/2] 3.64 6.36 8.39 10.80 3.31 5
Knoop hardness [GPa] 27.5 31.5 50 13.0 17.0 13
Young modulus [GPa] 607 653 993 620 370 300
Modulus of rigidity [GPa] 258 288 453 258 160 117
Bulk modulus [GPa] 315 297 412 375 232 227
Poisson's ratio 0.178 0.13 0.102 0.22 0.22 0.28
Thermal expansion 4.7 4.9 3.8 5 7.8 3.2
coefficient [10-6K-1]
Thermal conductivity [W.m-IK-'] 44 100 120 100 16.7 2&25
tantalum, chromium, molybdenum, and tungsten (which are, in fact, the elements of
groups IVb, Vb and Vlb).
In Sumitomo's patent, high heat conductivity together with high hardness are
mentioned as paramount properties when selecting a suitable binder material. A
rapid rise and fall in temperature in the tool during interrupted cutting often
leads to the formation of cracks. Therefore, a binder material that couples high ther-
mal stability and hardness with high thermal conductivity at high temperatures is
suitable for this kind of application.
The manufacturing process (especially the cold and hot compaction stages) of
pcBN is very similar to that of pcD, summarized in Section 1.3.3. However, the
types of binders used and liquid phase sintering mechanisms can be quite different.
In pcBN production, 'reactive sintering' plays a major role in terms of driving the
densification process, whereas in pcD, dissolution and precipitation and/or
adhesion/coalescence are the main driving mechanisms. A typical example for
pcBN would be the following reaction [162].
3A1+ 2BN 4 A1B2 + 2A1N (4)
where the reaction products A1B2 and AlN act as binders [ 1631. However, the chemi-
cal reaction to achieve these products is the driving force for sintering this particular
pcBN product.
The properties of the pcBN products are not only determined by the final micro-
structure, but also by the phases formed during sintering. The pcBN properties are
listed in Table 4 and the properties of the main phases are given in Table 5.
A direct conversion process can produce translucent pcBN which has almost
theoretical density and very high thermal conductivities. In 1972 and later in
1974, Wakatsuki et al. [164, 1651 sintered a polycrystalline cBN directly from
hBN at pressures as low as 55kBar and temperatures of 110&1400°C. Corrigan
[1661 observed that thermal conductivity of directly converted compacts increases
with increasing grain size. Fukunaga and Akaishi [167] used a small amount of
Mg3BN3 to promote conversion from hBN to pcBN at pressures of 5-6GPa and
at 1500°C.
ul
N
0
b
6'
$
a
6'
c
Table 5. Properties of the main phases found in pcBN products after sintering. D
%
a
cBN AIN TiB2 TIN TIC A1203 Diamond
(for reference) $
L
L
Crystal sructure cubic hexagonal hexagonal cubic cubic hexagonal cubic %2
Density [g ~ m - ~ ] 3.48 3.26 4.52 5.22-5.44 4.92 3.96 3.51
~
Melting point ["C] 2700 2300 3197 2930 3065 2050 s.
Knoop hardness [GPa] 41 12 26.5 19 28-35 23 57-104 2
Young modulus [GPa] 700-800 318 434540 370 450 421 1141
Thermal expansion coefficient [10-6K-'] 3.2 3.9 8.1 8 8.6 8.1 1.54.8
Thermal conductivity [W.m-'K-'] 150-700 200 80 25 3343 14 500-2000
1.4 New Ultrahard Materials

I. Sigalas
1.4.1 Introduction
Diamond’s combination of properties make it a unique material. Although hardness
is its primary characteristic, thermal conductivity, compressive strength, refractive
index, spectral transmittance, and chemical stability are either the highest or among
the highest found in nature.
This combination of properties can be traced to the same structural characteris-
tics of diamond that give rise to its high hardness. It is therefore reasonable to expect
that other ultrahard materials would also exhibit such a suite of properties. This
would make them also desirable for a number of industrial applications.
New ultrahard materials might possess new attributes other than a higher hard-
ness that would make them more attractive than diamond in some cases. A different
chemical composition would give rise to different interaction of cutting tool and
workpiece as is the case for cBN, as would a different crystal structure and ultrahard
particle morphology.
The search for new ultrahard materials is motivated largely by these considera-
tions, but also from purely economic ones. It might be possible to synthesize such
materials through routes cheaper than those involved in the case of diamond,
thus accessing new applications and new markets. In this paragraph we briefly
review the basic ideas behind the search for new ultrahard materials, as well as
the latest developments in the search for some of the ‘identified’ cases.
1.4.2 Hardness
Hardness is a measure of a material’s ability to resist elastic and plastic deformation.
The hardness of non-ideal material is determined by the intrinsic stiffness of the
material, as well as by the nature of its defects, be they point defects, dislocations,
or macroscopic defects such as microcracks etc. For ideal systems, the hardness of a
material will scale with its bulk modulus.
Figure 24 shows the Knoop hardness as a function of the bulk modulus for a
number of representative materials [168]. Table 6 shows the hardness as well as
the bulk modulus of a number of representative materials.
The bulk modulus of a solid can be calculated by means of ab initio calculations.
A review of this field can be found elsewhere [169]. Such calculations require exten-
sive computer time. Liu and Cohen in 1989 [ 1701 proposed an empirical formula for
the bulk modulus of solids with the zinc blende structure:
B = (19.71 - 2 . 2 0 h ) p (5)
where B is the bulk modulus in GPa, d is the bond length in Angstroms, and h is a
measure of the ionicity of the compound. For purely covalent compounds (group
80oo
diamond
7000
6ooo
n
5ooo
N cubic BN
E
E
m 4000
?5
z
I 3Mx) ZrC
Sic
Y
0
2ooo oAL03
rn-ZrO,
Si
1000
a 0"
OMgO
NaCl %a0
a 0
I I I I I I I I
0 50 loo 150 200 250 300 350 400 450
6,( G W
Figure 24. Knoop hardness (KHN) as a function of bulk modulus (B,) for representative materials.
IV), h = 0 while, for compounds of groups 111-V and 11-VI, h is equal to 1 and 2
respectively. This expression can be modified to allow for nontetrahedrally coordi-
nated compounds as follows [ 171, 1721:
B = (Nc/4)(1971 - 220qd-3.5 (6)
Where Nc is the average coordination number.
From Eq. (6) it follows that in order for the bulk modulus to be high, the crystal
must have: a large coordination number; a high degree of covalency; and short
interatomic distances, meaning small atoms. Although a high coordination
Table 6. Bulk moduli and microhardness for light covalent ceramics
Material B (GPa) H (GPa) H actual GPa [228]
C (diamond) 443 77 75-100

Si3N4 220 35 17
Sic 21 1 33 26
number may be attainable in metallic materials, where it can be as high as 12, metal-
lic bonds are much weaker than covalent bonds which can allow a maximum coor-
dination number of 4 in naturally occurring materials. Because of these
considerations the search for new materials is confined to tetrahedrally coordinated
compounds with high degree of covalency and with small atoms.
On the basis of this model, it was suggested that, for h = 0.5 and d = 1.47-1.49, a
hypothetical tetrahedral compound between carbon and nitrogen would have a bulk
modulus of 461-483 GPa, which would exceed that of diamond (443 GPa).
However, not all potentially hard or ultrahard materials can be found within the
two top periods of the periodic table, nor are they confined to tetrahedral coordi-
nation. A number of sp2 coordinated structures have been proposed which,
mainly due to their high density, would posses a high bulk modulus and would
therefore potentially exhibit high hardness [ 173-1 751. Superdiamond structures
have been proposed by Diedrich et al. [176] and Alberts et al. [177], based on poly-
merized derivatives of tetraethylmethane C(C2H)4and the carbon skeleton of allene
CH2=C=CH2 respectively. Leger [ 1681 investigated the feasibility of generating
ultrahard materials by producing the high pressure phases of a number of oxides.
Such phases would have a higher degree of coordination and bonds with a higher
degree of covalency than their ambient pressure allotropes.
1.4.3 C3N4
In 1989, Liu and Cohen reported local density approximation pseudopotential cal-
culations on a hypothetical carbon nitride phase, PC3N4 which suggested that it
could be metastable under ambient conditions [170, 1781 and would have a short
c-N bond length (0.147 nm), low ionicity ( ~ 7 % and ) a bulk modulus higher than
that of diamond. Figure 25 shows the crystal structure of hexagonal PC3N4derived
from pSi3N4 [ 1791.
Alternative structures have been proposed by Liu and Wentzcovitch [ 1SO], Wentz-
covitch and Martins [181] and Wentzcovitch [182], proposing a cubic zinc blende
structure with one carbon vacancy per unit cell, and a structure resembling graphitic
CN with one carbon vacancy per four nitrogen sites. This study was based on an ab
initio molecular dynamics scheme with a variable cell shape algorithm.
More recently Teter and Henley [ 1831, using first principles pseudopotential total
energy techniques, predicted a cubic form of C3N4with a zero pressure bulk mod-
ulus of 495GPa, higher than that of diamond. The same authors predicted that
aC3N4and graphite-structure C3N4are energetically more favorable than PC3N4.
Figures 26 and 27 show the structures of the proposed cubic C3N4 and aC3N4
compounds respectively.
The above theoretical predictions have generated a great deal of experimental
activity aimed at synthesising a new ultrahard compound.
Several attempts have been made to synthesize thin film C3N4,be it cubic or hex-
agonal. These include reactive magnetron sputtering [184187], laser ablation [188]
ion beam assisted deposition (IBAD) [189], plasma [191, 1921 and plasma-enhanced
chemical vapour deposition [190].
Figure 25. Hexagonal crystal structure of PC3N4 derived from pSi3N4. Two types of sp2’ bonded
N atoms are present. The first type is in a trigonal planar configuration with the trigonal plane
perpendicular to the c-axis. The second type forms, together with the carbon atoms, the tunnel at
the center [179].
In most efforts, the films produced were either amorphous, or their nitrogen con-
tent was less than the target of 57%. D. Li et al. [184] using a magnetron sputtering
system produced crystalline carbon nitride/titanium nitride composite coatings with
a reported hardness of 55 GPa.
Figure 26. Two adjacent unit cells of cubic C3N4carbon atoms are shown schematically as the black
atoms [192].
I .4 New Ultrahard Materials 525
P’C3N4
Figure 27. Predicted crystal structure of aC,N4 [192], compared with that of PC3N4.
Zhang et al. [ 1941 at Harvard were able to synthesize an amorphous CN thin film
with composition C2N. This material had exceptional hardness [195-1971. In some
cases, tiny crystallises of covalent C-N were observed in the amorphous films [19&
2021. Bhusari et af. in February 1997 [193] were able to synthesise large crystals
(750 pm) of Si-containing carbon nitride consisting of a predominantly C-N
network, by microwave CVD.
More recent efforts involving nitrogen ion beam assisted deposition [203, 206,
2111, hot filament CVD with or without rf plasma [204, 205, 2091, or microwave
plasma enhanced CVD [213, 2141, with bias assistance [207, 208, 2121 were able to
produce thin films containing C3N4,some with the BC3N4structure and some with
the cubic or a-structure.
Yan et al. [205] using hot filament CVD were able to synthesize fully crystalline
films combining both aC3N4and PC3N4.
Xu et al. [210] using a reactive magnetron plasma source claim to have deposited
polycrystalline BC3N4films with crystallites as large as 20 pm.
Xiao-Ming et al. [211] using argon arc nitrogen-assisted bombardment claim to
have deposited completely polycrystalline BC3N4 films. They quote a measured
hardness of 52.6 GPa.
Similar results were obtained by Wu et al. [212] who used RF plasma enhanced
CVD.
Finally Chen et al. [213], using microwave plasma enhanced CVD, claim to have
grown crystals larger than 10 pm, and propose that the incorporation of silicon from
the substrate into the C-N structure promotes BC3N4crystal growth.
Efforts to synthesize C3N4at high pressures have been much less extensive than
those made to synthesize this material in the thin film form. A number of trials [214]
indicate that it may be possible, through the use of pressure to incorporate nitrogen
into sp’-bonded carbon nitride.
1.4.4 Boron Rich Nitride

B3N, B5N and B4N have been synthesized by CVD methods [215-2171. The B4N
structure, with a crystal structure analogous to that of rhombohedra1 B4C, may
exhibit high hardness. No reported hardness values are available at present.
1.4.5 Boron Carbonitrides

In addition to binary compounds made with elements from the 2d period, ternary
compounds may also exhibit higher thermal stability that diamond [218]. Such com-
pounds can be considered as solid solutions of carbon in boron nitride. Efforts to
synthesize such compounds have been made both at low and at high pressures.
Ternary BC,N, films were first produced in 1972 by Badzian et al., using a CVD
process [219]. Montasser et al. [220] in 1984 produced films of hardness in the range
4-33GPa. Loeffler et al. [221] studied the influence of substrate temperature on the
crystallinity of BC,N, films deposited by PA CVD. Hegermann et al. [222] investigated
the influence of carrier gases nitro en, argon and helium and the influence on the applied
power density (about 14Wcm-') of the film deposited by the PA CVD process.
1.4.6 Boron Suboxides

Boron suboxides with hardness comparable to that of diamond were first reported
by A. R. Badzian in 1988 [223]. These suboxides were produced by reacting B2O3
with boron at 1600-2000°C and 7 kbar. Subsequent melting under argon gave a sin-
tered compound with an oxygen content of 4 5 % and a microhardness of 60 GPa.
The introduction of oxygen reduces the electron deficiency of the P-rhombohedra1
structure of elemental boron. Filling these voids with atoms results in a cross linking
of the boron icosahedra and can contribute to a significant increase of the rigidity of
the structure.
Recently McMillan et al. [224] reported the synthesis of B60 icosahedra at pres-
sures of 5-6 GPa. ltoh et al. [225] reported the synthesis of B 6 0 powder by reacting
B203withboron at 1350-1400°C and the subsequent sintering of that powder with
B4C or cBN at 3-7GPa and 1500-1800°C for 10-30min.
Although the hardness of the resulting compacts exceeded 40 GPa, the fracture
toughness did not exceed 1.5 MPa m1/2.
The synthesis of B 6 0 was patented initially in 1992 [226, 2271, but no boron sub-
oxide products have appeared in the market as yet.
1.4.7 Stishovite
The search for new ultrahard materials has also been extended into nonboron based
oxides. The expectation was that high pressure allotropes would exhibit a high
coordination of the metallic ion, as well as more ionic bonding, thus hopefully pos-
sessing higher stiffness and, therefore, higher hardness.
Leger et al. [166] in 1994 measured the bulk modulus of the high pressure phases
of HfOz and RuOz, which they obtained at pressures above 42 and 12GPa. The
resulting bulk moduli place the two compounds above diamond in the case of
Hf02 and between cBN and diamond for Ru 0 2 .
In 1996 Leger et al. [228] reported the synthesis of stishovite from c1 quartz at
20 GPa and 1100°C and measured a hardness of 33 GPa for the polycrystalline com-
pact produced. As the samples obtained were very small, a maximum load of 1.9 N
during hardness testing was possible. Synthesis of this material in larger quantities
has not yet been possible.
1.5 Industrial Applications of Diamond and cBN

M. W. Bailey
1.5.1 Introduction
As discussed earlier, the modern industrial diamond business began to expand sig-
nificantly following the commercial availability of synthetic diamond in the late
1950s. Before then, only natural diamond produced as a by-product of gem dia-
mond mining was available and was, by its nature, limited in volume. Demand
was increasing to the point where new application development was a largely self-
defeating exercise since, the more successful they were in terms of using large
volumes of diamond, the less feasible they became in terms of total world supply.
With the introduction of synthetic diamond, this situation changed dramatically
on two important fronts. Firstly, the only constraint remaining on the development
of new applications was the cost and technology of producing a suitable product;
and secondly the constraints on product characteristics imposed by the nature of
a natural raw material were removed: diamond could now be specifically engineered
for its intended use.
Natural and synthetic diamond are complementary, not competing products.
While today a minor player, natural industrial diamond still plays an important
role in a variety of special applications where one or other of its particular
characteristics will make it the preferred material (refer to Section 1.5.5 for some
examples).
Industrial applications of diamond have developed over the years as a result of
developments by tool makers, machinery manufacturers and the advert of new
materials. In addition, new diamond (and cBN) products have evolved, either as
a result of technological advances in synthesis or in response to the requirements
of a new application, and it is this multi-partnership relationship within the
industry which has resulted in the dramatic growth and diversification since the
early 1960s.
1.5.2 Abrasive Application

1.5.2.1 Bonded Tools
One remarkable property of diamond is its hardness and associated resistance to
abrasion and it is this which has led to its use as an industrial abrasive. Cutting
and polishing stone, glass, and other hard materials were among its early areas of
application. Diamond has one limitation in respect of its application, this being
its reaction with iron at high temperatures (causing a reversion to graphite and
hence high rates of wear), which in general can make it uneconomic in the machining
of the ferrous materials by comparison with other conventional abrasives such as
aluminum oxide and silicon carbide. Cubic boron nitride (cBN) does not react in
this way and hence, although only having 50% of the hardness of diamond, it is
still substantially harder than the conventional abrasives and this makes it suitable
as a high performance abrasive for use on ferrous workpieces. Figure 28 summarizes
the major areas of application for diamond and cBN abrasives.
The advantages of using diamond and cBN abrasives are due mainly to their high
hardness and abrasion resistance which, when applied correctly, result in longer tool
lives, higher levels of productivity, closer tolerances, and many other advantages.
The majority of diamond and cBN abrasive is used in so-called ‘bonded tools’,
although much polishing and fine finishing is carried out using the abrasive as a
loose powder or in a slurry when mixed with an oil or other carrier medium. Dia-
mond and cBN abrasives are normally used in particle sizes of approximately
1 mm down to less than 0.1 pm, and their sizing is subject to international standards,
the most widely used being the FEPA standard [229].
In a bonded tool, the abrasive particles are held in a bond or matrix material,
usually distributed randomly but in a controlled concentration. The main functions
of the bond are:
- to hold the abrasive particles

- to form a molded profile of the required shape which is usually attached to a hub
or carrier
Diamond Cubic Boron Nitride (cBN)
Non-ferrous materials: Ferrous materials:

Glass -flat, decorative, optical Hard ferrous components
Engineering ceramics
Cemented tungsten carbides
Reinforced plastics
Natural stone
Semiconductor materials
(Diamond accounts for 75-80%, cBN accounts for 20-25%)

Figure 28. Main application areas for diamond and cBN abrasives.
Layers of diamond particles in a

resin, metal or vitrified bonded
tool
Figure 29. Basic principle of bonded tools.
- to release worn abrasive particles as they reach the end of their useful lives, thus
exposing new ones.
It is therefore a general requirement of a bonded tool that the abrasive and the bond
material wear at similar rates.
The most common types of bond used are as follows:
- resin, normally thermosetting phenolic or polyimide resins

- vitrified, based on glass frits
~ metal, normally sintered alloys
- electroplated, single layer or a few layers of abrasive attached to a hub by an elec-
tro-deposited metal layer, normally nickel.
Figure 29 shows the concept diagramatically. Figure 30 shows a selection of bonded

tools including saw blades and grinding wheels.
In the case of the first three bond types, since the rate of bond erosion and rate of
wear of the abrasive have to be matched if optimum performance is to be obtained,
the selection of the abrasive which has the correct characteristics of size, impact
strength, and fracture mode matching the characteristics of the bond is very impor-
tant. Workpiece material plays a major role in this selection process but also the type
of machining operation, the machining conditions and the primary requirement of
the operation are important. Maximum tool life, high rates of material removal or
high quality of surface finish are common requirements but normally a compromise
between one or more is required. Figure 31 shows some of the issues which affect the
final machining performance. In the case of the fourth bond type, electroplated (or
EP), the abrasive particles are embedded in a metal layer to something less than 50%
of their total dimension as shown schematically in Fig. 29.
The abrasive particles wear down to the level of the supporting EP layer, at which
point the tool is at the end of its life. With this type of tool, the choice of abrasive
Figure 30. A selection of bonded tools.
depends primarily upon the workpiece material, the machining conditions and the
primary role of the machining operation (e.g. to produce a high surface finish or
high material removal rates). The role of the EP bond layer is to support the abrasive
and not wear significantly during the life of the tool. When new, EP tools have a high
degree of protrusion of the abrasive particles above the bond layer (much higher
than the other types of bond) and hence can machine at high material removal
rates while generating low cutting forces. The term often used to describe such a
Task Wheel type

Machine Wheel bond
Workpiece 4''" Wheel grit
Coolants Preparation
Figure 31. Main parameters affecting tool performance.

1.5 Industrial Applications of Diamond and cBN 53 1
General purpose bond

Simple to manufacture
Easier to use than vitrified

or metal (more forgiving)
Profile holding not as good

as vitrified or metal
General ouroose for all tvoes of material

Resin bond diamondgrinding the
Sintered‘catbide, Non-oxide ceramics
teeth of a cemented carbide cutter Cermets, Natural stone (polishing)
Figure 32. Main attributes of a resin bond.
characteristic is ‘free-cutting’. As the abrasive particles become worn, so this free

cutting character is gradually reduced and the cutting forces increase until the end
of the useful life of the tool is reached.
1.5.2.2 Grinding with Diamond and Cubic Boron Nitride Abrasives
1.5.2.2.1 Introduction Grinding with diamond and cBN abrasives is employed by a

very wide range of industries and application areas. In dentistry, small EP diamond
drills are widely used, spectacle lenses and crystal glass items are machined using
diamond, large diamond wheels are used to produce high volumes of cemented
tungsten carbide cutting tool inserts to precise tolerances and cBN wheels are
widely used in the automotive and bearing industries to produce high volumes of
precision ground components.
In grinding operations, all four of the main bond types are used. As a general
guide they may be considered to have specific advantages and the choice is often
dependent upon which characteristic is the more important. Figures 32-35 show a
Very high wear resistance

+Good profile holding
+Difficult to dress
+ Manufacturing technology
more advanced than resin
Decorative, flat, optical glass

Aluminium oxide
Metal bond diamond grinding

a high alumina ceramic seal
Figure 33. Main attributes of a metal bond

Single layer of Ni-electroplated

abrasive
Cheap to manufacture
Excellent for profile grinding
Limited life
High stock removal possible
Green carbide, Aluminium oxide

Ferrites, Natural stone (profiling)
Precious stone, Dental
Reinforced plastics,
Semiconductors
- Profile grinding hard ferrous (cBN)
Electroplated diamond saw blade
slicing a block of optical glass
Figure 34. Main attributes of an electroplated bond
typical application for each of the four bond types and some of their particular
advantages. The characteristics of the abrasive used can be selected from a wide
range available; size, resistance to impact and fracture characteristics being
among the more important and Fig. 36 summarizes the main differences and their
effects.
Grinding operations can take place with the application of a coolant or lubricant
(wet grinding) or with no such application (dry grinding).
1.5.2.2.2 Grinding with Diamond Abrasives In grinding applications, the main

types of bond used are metal and resin bonds.
In resin bond wheels, low strength friable diamond types are normally used, a
typical diamond being shown in Fig. 37. Since the resins normally used for the man-
ufacture of diamond tools have relatively low strength and are also sensitive to high
Hard and abrasion resistant

Up to 30% porosity improves chip
clearance independent of grit size
Holds its edgelprofile well
Longer life than resin
Easily dressed
More difficult to manufacture
than resin
Vitrified bond diamond grinding

d a tungsten carbide roll Non-oxide ceramics
Hardened steels (cBN)
Figure 35. Main attributes of a vitrified bond.

~ ~ ~ oblockv ~ -r ,
n ~orits
withstand higher forces - withstand lower forces -
greater exposure - less exposure -
higher rates of stock removal - lower rates of stock removal -
lower degree of surface finish - higher degree of surface finish -
suited to more ductile, suited to hard, brittle,
long chipping workpiece materials short chipping workpiece materials
Figure 36. Basic effects of abrasive characteristics.
temperatures, a thick metal cladding of about 10 pm of nickel or sometimes copper

is applied to the diamond particle such that typically it will comprise 50-60% of the
total particle weight (Fig. 37). This cladding functions in two main ways (Figs 38
and 39), firstly by increasing the surface area of the particle and hence aiding its
retention by the bond, and second by acting as a heat sink to protect the bond
from thermal damage due to the high interfacial temperatures generated between
the abrasive particle and the workpiece, particularly under heavy grinding condi-
tions. These temperatures, although of short duration, can reach in excess of
800°C as the particle passes through the arc of cut.
Figure 37. A typical friable diamond abrasive PDA321 used (normally in metal clad form) in resin
bonded grinding wheels.
Figure 38. The role of metal cladding.
Resin bond tools are used to machine a wide variety of non-ferrous, usually
abrasive, workpiece materials. A very small number of ferrous materials can be
ground economically with diamond, for example, some carbon rich cast irons
which, providing machining temperatures are maintained relatively low, do not
chemically react with the diamond, and other difficult-to-machine materials such as
some grades of stainless steel which have a tendency to work harden during machin-
ing. The major application area for resin bond diamond wheels is the grinding of
cemented tungsten carbide (which was also its initial application area in the 1940s).
The list below gives the main areas of application in descending order of relative size:
- grinding cemented tungsten carbide
- grinding ceramics and cermets
- grinding semiconductors
- grinding polycrystalline diamond (pcD)
- machining stone/glass
- others
Although the machining of cemented tungsten carbide represents approximately
half the resin bond tools used, other areas are growing, notably semiconductors,
ceramics and cermets, pcD and the machining of stone.
The bond of an abrasive tool performs various tasks:

* Holds the abrasive particles
* Forms a moulded part or segment
(attached to a carrier -blank or hub- by brazing/welding/
adhesive fixinglsintering or electroplating methods)
+ Release worn diamond at a controlled rate

- wear rates of abrasive particles and bond must match
for a given workpiece and machining para~eters
Figure 39. The role of the bond.

Figure 40. A typical high strength diamond abrasive PDA999 used in grinding wheels.
Metal bond diamond wheels are predominantly used in the machining of glass
including windows for the automotive and construction industries, decorative
glass such as lead crystal and also optical and electronic components. The diamond
types most commonly used in metal bond tools are relatively high in impact strength
with regular crystal shapes. Figure 40 shows a typical high strength diamond abra-
sive. Sintered metal bonds commonly based on bronze are much more abrasion
resistant than resin bonds, and therefore better suited to machining workpiece
material which produce highly abrasive swarf, glass being one example. The list
below summarizes the main application areas for metal bond diamond wheels.
- grinding glass
~ machining refractory materials
- machining semi-precious stones
- grinding pcD
- grinding ceramics
- honing (mostly ferrous materials)
Electroplated diamond wheels are used in applications where free-cutting character-

istics are required. Fibreglass and other composite materials are in general highly
abrasive although not particularly hard. The high abrasion resistance of diamond
makes it the ideal tool material for machining these types of material and the
very open texture and free-cutting nature of EP tools mean that these materials
can be machined quickly. Figure 41 shows an example.
Vitrified bond diamond wheels are most commonly used to machine polycrystal-
line diamond. pcD is difficult to machine and often has to be machined to close tol-
erances when being used as a cutting tool material as shown in Fig. 42. Both metal
and resin bond grinding wheels are also used, but vitrified bond wheels are becoming
the most popular, since they provide a good compromise between relatively good
wheel life, and hence ease of maintaining dimensional tolerance on the pcD work-
piece, and the rate at which the pcD can be ground. Another advantage of vitrified
Flank grinding a PCD-tipped saw

Figure 42. A circular saw blade tipped with pcD.

Figure 41. An EP diamond wheel cutting glass.
CTCOO2-tipped saw blade for wood

520mm diameter SYNDITE
536
bond systems is that a degree of porosity can be built into the bond. Resin and metal
bonds are effectively 100% dense and clearance for the swarf produced during
grinding is provided by the protrusion of the abrasive particle above the surface
of the bond. The pores, or voids, in porous vitrified bonds provide extra clearance
volume for the removal of swarf out of the grinding zone and also the transport of
coolant into it.
In general, all diamond grinding wheels operate at average peripheral speeds in
the range 2&30ms-'. If a material is being ground dry, without coolant, the
wheel speeds tend to be lower, and also the difficult materials such as cermets and
pcD tend to be ground at lower speeds to avoid the excess generation of heat. A
small number of specialized applications use much higher wheel speeds, one example
being the sawing of glass tubes which, providing the machine is specially designed
for the purpose, with particular attention to the coolant application, can be cut
much more quickly with a wheel rotating at speeds up to 60-80 ms-' .
1.5.2.2.3 Grinding with Cubic Boron Nitride Abrasives Cubic boron nitride abra-
sives are used to machine hard ferrous materials. As with diamond, all four of the
main bond types are used and, during the early days of its use, in the 1970s and early
1980s, resin bond tools were the majority. Today, vitrified bond tools are the most
commonly used type, particularly in high volume production operations, with resin,
electroplated, and metal bonds failing in second, third and fourth place respectively.
A reason for the growth in the use of vitrified bond cBN tools is that this type of
bond system offers a good compromise between high material removal rates, ease
of use, and low wear rate resulting in the ability to hold tight tolerances.
The growth in the use of cBN since the early 1970s has been due to its penetration
into volume production operations, for example in the automotive and aerospace
industries. Prior to that time, cBN wheels were used primarily in the toolroom for
sharpening and re-sharpening high speed steel drills, milling cutters and other tool-
ing and also in small wheels for precision jig grinding (Fig. 43).
The economics of cBN grinding have been improved by the adoption of creep feed
grinding techniques and also the use of higher wheel peripheral speeds. Both of these
developments required the development of suitable grinding machines.
Creep feed (or deep) grinding techniques were originally developed using con-
ventional wheels and the principle of the technique is shown in Fig. 44. In 'normal'
or reciprocating grinding mode, a shallow depth of cut is used together with a relatively
high table speed to achieve the required material removal rate. Under these conditions,
the arc of cut, and hence the area of wheel in contact with the workpiece at any one
time, is very small. In creep feed grinding, a significantly deeper cut is made at a
slow table speed. Under these conditions, the arc of cut is much longer and hence
the number of particles instantaneously in contact with the workpiece is much greater.
The specific rate of material removal is a function of the depth of cut and the feed rate
of the table. In general, when comparing the two modes of grinding, and assuming
equal material removal rates, the main differences for creep feed grinding are that
- the number of particles in contact is higher
- the load on each particle is lower
538 I Diuniontl Matrrials mu' their. App1iuition.s
Figure 43. Dry grinding with cBN
~ the total normal load in the machine is greater (due to the greater number of
particles)
- the life of the grinding wheel is longer
~ surface finish produced is normally improved.
Reciprocating Grinding Creep Feed Grinding
DOC = S - SO Micron DOC = 5 - 25 millimetres

Table speed = 5 - 15 m/min Table speed = 0.1 - 1 mlmin
Figure 44. Principle of creep feed and reciprocating grinding.

1 .SIndustrial Applications of Diamond and cBN 539
Wheel Speed
Figure 45. Effect of wheel speed on grinding wheel performance.
A second development in cBN grinding is the adoption of higher wheel peripheral

speeds. In the early days of grinding with cBN, wheel speeds similar to those used
for diamond wheels were used. It was found however that increasing the wheel periph-
eral speed improved the life of the grinding wheel providing coolant was used (Fig. 45).
Initially speeds up to 60 ms-' were thought to be optimum but today, wheel speeds in
the range 6&150ms-' are common in industry. This has been made possible by
improvements in the mechanical design of both the grinding wheels and the machines
on which they are used since, at high speeds, safety becomes a major issue which has
to be addressed by suitable speed certification for the wheels and safety guarding of
the machine. At various academic institutions, much higher wheel speeds have been
used, and wheel life has continued to improve. In one such series of tests in Germany,
speeds of 500 ms-' were run in one series of experiments.
As a general rule in any grinding operation, increasing the rate of material
removal results in a decrease in wheel life. Hence, there is usually a balance in the
cost of the time taken to grind and the cost of the wear on the wheel (Fig. 46). By
adopting either, or both, the technologies of creep feed grinding techniques and high
wheel speeds, higher material removal rates (and hence rates of production) can be
achieved whilst retaining, or even improving, the wheel life and hence direct machin-
ing cost due to wheel wear.
The field of application for cBN grinding is almost entirely that of hard ferrous
metal machining, although some superalloys are also ground using cBN. The list
T
cost
Material Removal Rate-

Figure 46. Trend of cost versus removal rate.
600mm dia.resin bond wheel containing

ABN360
Grit size 8126 (120/140)
Material removed 5-20 micron
Figure 47. Grinding automotive piston rings with cBN.
below gives the main application areas:

- automotive and aerospace material machining
- tool and cutter grinding
- gear grinding
- machining of bearings
- general grinding of hardened steel components
- honing of hardened steel
The largest area of application is in the automotive and aerospace industries, grind-
ing of camshafts, crankshafts and other precision engine parts such as turbine blades
and injector nozzles being examples (Fig. 47). Precision grinding of gear tooth pro-
files is another area where cBN, either in vitrified or EP tools, is used (Fig. 48). The
original application area for cBN, tool and cutter grinding, is still a cBN grinding
application, the grinding of high speed steel drills and milling cutters, both during
manufacture and re-sharpening during use being examples (Fig. 49).
In addition to faster metal removal rates and better tolerances compared to grind-
ing methods using conventional abrasives, cBN wheels have the ability to produce
lower workpiece temperatures during grinding, which reduces the potential for ther-
mal damage. This can be of significant importance when critical components are
being machined.
1.5.3 Machining of Stone and Concrete

The largest application area for diamond abrasives in terms of volume consumed is
the sawing, drilling and surfacing of natural stone and concrete. This application
area grew dramatically during the 1970s and 1980s, due to a number of influences.
Industrial diamond became more widely available and also became available in an
Figure 48. Electroplated gear grinding wheel
ever increasing range of size and also other characteristics such as strength. Devel-
opments by tool manufacturers and machine builders improved the economics of
machining hard natural stone types such as the granites, making their processing
more economical. There was also a movement towards the use of granite from
marble in the construction industry. One of the reasons for this was that it was
recognized that granite was far more durable than marble in respect of both wear
if used on floors, and also more resistant to chemical attack and discoloration in
urban environments if used as a cladding on the exterior of prestige buildings. Its
use, therefore, became more widespread and the quantity of diamond abrasives
used to process it has continued to increase.
Figure 49. Flute grinding high speed steel.

Figure 50. Section of a diamond wire saw.
The types of diamond abrasive used for cutting stone and concrcte are normally
those with high impact strength and of a relatively coarse size.
The normal production route for stone products includes:
- extraction from the quarry in the form of large blocks
- squaring and stabbing the rough blocks in the stone yard
- cutting the blocks into either cladding or tiles
- surfacing and polishing the finished items.
In all these operations, diamond tools are widely used. For quarrying, diamond wire
sawing is established as the standard method of extraction for soft rock (e.g. marble)
and the technology for hard rock is being established. Development of diamond
wire began in the early 1970s but it came into common use only in the late 1980s.
It consists of a high tensile steel wire fitted with beads containing the diamond abra-
sive, typically 40 beads per metre. The overall diameter of the beads is 6-12mm
(Fig. 50). Two intersecting holes drilled in the quarry face and the wire, which
can be up to 500m long, is fed through them, joined to form an endless loop and
then driven at speeds of 20-50msC' while tension is applied, thus cutting a slot
through the rock and enabling a block to be cut (Fig. 51). The resulting block suffers
much less damage compared with using explosive techniques and this, in turn, leads
to less waste in subsequent processing. A major problem encountered with early
designs was erosion of the wire by the fine particles of stone produced by the
sawing, leading to breakage. This is now alleviated by protecting the wire with
rubber or plastic sleeving bonded to the wire between the beads.
When quarrying granite, a relatively low wire speed is typically used, 20-30 ms-'
and, for a medium grade of material, a block measuring 30m long x 10m
wide x 5 m deep can be separated from the surrounding material in less than 12 h.
Figure 51. Wire sawing in a stone quarry.
However, because the block produced is of a regular size, up to a 40% increase in

usable material is possible with commensurate similar improvements in the econom-
ics.
Following extraction from the quarry, the large blocks are either sawn into thick
slabs for use in the monument industry or they are squared for subsequent produc-
tion into tiles or stone cladding. Diamond wire saws are used for these operations
together with large circular sawing machines utilizing diamond saw blades of up
to 4m diameter (Fig. 52). The advantages of using diamond tools compared to
machines utilising silicon carbide abrasives are increased speed of production,
improved accuracy of cut and much reduced pollution.
Tile production is a major application for diamond saw blades. The sawn blocks
are machined on multi-blade machines using up to 32 blades, 1.6 m in diameter on
one spindle (Fig. 53). These machines typically cut the granite blocks into a series of
slabs 12mm thick, with a thickness tolerance of f0.5mm normally required. A
horizontal blade then traverses the block to separate the slabs produced and these
Figure 52. Block sawing of granite.

Figure 53. Multi-blade sawing of granite to produce tiles.
slabs in turn are ground to final thickness and sawn to final size, both operations
using diamond tools. The final stage in the production of tiles is polishing which
is normally achieved using automatic multistage machines (Fig. 54). In this final
process, many of the stages are now completed using diamond tools, the early
Figure 54. Polishing natural stone tiles.

1.5 Industrial Applications of Diamond und cBN 545
Figure 55. Resin bond polishing head.
stages in polishing using metal bond tools with high strength diamond to achieve the
high material removal rate required and the later stages using resin bond tools incor-
porating lower strength, more friable, grades of diamond particle. Figure 55 shows a
selection of polishing tools.
Another major application area for diamond abrasive tools is in the construction,
refurbishment and demolition industries. In many instances, concrete, masonry and
also roadways need to be cut or drilled and diamond tools provide an economic
solution, their advantages being speed, accuracy and also reduced damage to the
surrounding structure compared to other methods using percussive hammers or
thermal lances. Circular saw blades are used in a wide variety of operations and
metal bond tools are most commonly used. On roadways, both asphalt and con-
crete, diamond saws are used to cut trenches during repairs and also to provide
slots to embed sensor wires for traffic management systems (e.g. traffic signals)
(Fig. 56). Concrete roadways, runways and pedestrian walkways can, under wet
conditions, become slippery and diamond blades, up to 100 spaced along a spindle,
are used to cut a series of parallel grooves in the concrete to aid water drainage and
improve skid resistance and hence safety in critical areas (Fig. 57).
Saw blades are used to cut openings in walls and floors during both building and
refurbishment and can also be used during controlled demolition. The advantage of
a diamond saw, compared to a saw of conventional abrasive, is its ability to cut both
concrete, masonry and steel reinforcement (Fig. 58).
The majority of all diamond sawing of concrete is carried out with use of water as
a coolant. This has the combined effect of controlling the temperatures at the cutting
edge and also removing the swarf produced by the saw. Some modern blades are
designed to run dry (Fig. 59). These are now widely used on construction sites for
cutting tiles and the minor correction work which needs to be carried out.
Wire saw machines, similar in principle to those used for cutting stone, are also
used widely for large demolition operations. When using a circular saw, there is a
limit on the depth of cut which can be made due to the presence of the drive spindle
Figure 57. Roadway grooving to improve skid resistance.

Figure 56. Roadway sawing.

546
547
1 .SIndustrial Applications of Diamond and cBN
Figure 58. Sawing in the construction industry.
Figure 59. Dry sawing of concrete.

Figure 60. Wire sawing of heavily reinforced concrete.
in the centre of the blade. This is a problem if deep sections have to be cut. The wire
saw does not have any such limitation and concrete, heavily reinforced with steel as
used in large structures such as road bridges, can be cut with modern wire saw
machines. Figure 60 shows such an operation.
Drilling stone with diamond bits is now widely used and concrete, both plain and
steel reinforced, brickwork and other masonry have been drilled for many years
using diamond core bits (Fig. 61). These tools can range in size from approximately
20mm to in excess of a meter. They are all metal bonded tools and were until
recently designed to operate with water as a coolant. In some areas, the use of
water can be inconvenient and a relatively recent development has been the
design of core bits which do not need to be cooled with water. These dry drill bits
are widely used on construction sites to produce accurate holes in walls for electrical,
plumbing and other services (Fig. 62).
1.5.4 Applications of Polycrystalline Ultra-hard Materials

1.5.4.1 Introduction
The polycrystalline derivatives of diamond and cBN, normally referred to as pcD
and pcBN respectively, provide engineers with materials which have many of the
properties of diamond and cBN, notably hardness and abrasion resistance, but in
the form of relatively large isotropic pieces, usually in the form of flat discs.
These materials have a wide variety of uses, both as a defined edge cutting tool
Figure 61. Core drills ~iseclin the coiistriiclioii inclustrq
element and also as a wear resistant material. As a cutting tool material, the general
rules which apply to abrasives also apply to pcD and pcBN.
~ pcD is for use in non-ferrous applications.
~ pcBN is for use on ferrous workpieces.
Figure 62. Dry drilling of t i u w n r q

550 1 Diamond Muteriuls and their Applications
cutting 10000 - CVD

speed diamond
[mlmin]
-
1000
Cemented
-
100
-
10
Carbon tool
- steel
0.1
I I I I I I
1.5.4.2 Cutting Tools of Polycrystalline Diamond

In most successful applications of pcD cutting tools, the advantages over more con-
ventional tool materials, including sintered tungsten carbide and ceramics, are due
to their high hardness and abrasion resistance. This in turn results in longer tool life
giving rise to closer tolerances and the ability to work efficiently at high cutting
speeds and hence give higher levels of productivity. In addition, workpiece materials
which are difficult to machine can be machined efficiently with pcD. Continuous
material’s development over the years has played a very significant role in the
advances made in machining technology, resulting in higher levels of productivity
and, in particular, closer tolerances over long machine runs which are now required
by modern manufacturing industries. Figure 63 shows the achievable cutting speeds
which have been made possible by advances in cutting tool material technology.
Figure 64 shows schematically how current materials relate to each other in terms
of two important characteristics, toughness and abrasion resistance/hot hardness.
The ideal cutting tool material would feature very high toughness with equally
high wear resistance. Although, today, such a material does not exist, it is the objec-
tive of many development programmes (see Section 1.4).
The main application areas for pcD are the machining of
- non-ferrous metals, e.g. aluminum alloys, magnesium alloys, brass, copper
- wood and wood composites
- fibre-glass and carbon fibre composites
- plastics and rubber
- mineral materials
The automotive and aerospace industries provide a large number of applications for
pcD, particularly with the movement to more widespread use of advanced alloys
I .5 Industrial Applications of Diamond and cBN 55 1
I PCD
Ideal cutting
tool material
Toughness, transverse rupture strength
Figure 64. Properties of cutting tool materials

-
and reinforced plastics. The main application areas in the automotive industry are
summarized in Fig. 65.
An example of an engine component commonly machined with pcD is an
aluminum alloy engine block (Fig. 66) manufactured from an silicon-silicon alloy
Engine components
Gearbox components
Transmission components
Braking systems
Interior fittings
Body panels
Figure 65. pcD applications in the automotive industry
Figure 66. Rough milling A1-9% Si engine blocks with pcD.

Figure 67. Face milling A1-12% Si gearboxes with pcD.
which, dependent upon the alloy, can be very abrasive. Similar classes of material
are used in the manufacture of gearbox casings and these can also be machined
more efficiently with pcD than with any other cutting tool materials (Fig. 67).
The cylinder bores of aluminum-silicon alloy engine blocks are also typically
machined with pcD. Figure 68 shows such a component from a V-12 engine
being finish bored with pcD.
Figure 68. Finish boring of cylinder blocks with pcD

Figure 69. SYNDITE 025 machining 20% S i c reinforced Al M M C brake discs.
A material originally used in racing cars and motorcycles is a metal matrix com-
posite (MMC) made of aluminum reinforced with silicon carbide particles. Its light
weight and high strength made it ideal for use as a disc brake rotor and also other
components. The presence of the silicon carbide makes it a difficult material to
machine, pcD being the best cutting tool material. Figure 69 shows a typical
MMC machining operation on a brake disc rotor. Machining of reinforced plastics
is also a major application for pcD tooling. Figure 70 shows part of a glass
reinforced plastic (GFRP) truck body section which is machined with a pcD router.
A relatively recent application area for pcD has been in the woodworking
industry. Wood composites such as medium density fibreboard (MDF), chipboard,
laminated boards with plastic coatings, for use in furniture and on floorings, and
also composites, used in wall claddings and in ceilings, all tend to be highly abrasive
Figure 70. pcD milling cutter machining a GFRP truck bumper.

Figure 71. Schematic cross-section through typical abrasive particle board.
when being machined. Figure 71 shows a typical cross-section through a wood

laminate. Saw blades, routers and other profiling tools now use pcD tips for high
volume production. Figure 72 shows a profiled pcD tipped cutter being used to
produce furniture panels in MDF, a commonly used material for this application
and Fig. 73 shows typical wear rate results obtained when machining such materials
and compares high speed steel, cemented tungsten carbide and pcD in terms of tool
wear rate as a function of distance machined.
1.5.4.3 Drilling with Polycrystalline Diamond

The major advantage of using pcD in a rock drilling application is that pcD drill bits
are designed to remove rock by shearing rather than by crushing or grinding, as is
the case with most other types of drill. Rock removal by shearing is very efficient,
requiring only 15-20% of the energy required by crushing and grinding. This, in
Figure 72. pcD is typically used on medium density fibreboard (MDF), chipboard, and abrasive
hardwoods.
1.5 Industrial Applications of Diamond und cBN 555
Flank wear Vb [mm]
Speed = 1000 d m i n
Depth = 10mm
Feed ~ 0 1 mdtooth
0.15 Infeed = 2 rnmlpass
f
c_tc;;3p 3';
Distance machmed [m]
Figure 73. Wear rates of HSS (high speed steel), WC (tungsten carbide) and pcD routers machining
chipboard.
turn, means that pcD drill bits have the potential to drill much faster than conven-
tional bits and, in addition, the high abrasion resistance of pcD results in the cutters
remaining sharp and hence extending considerably the life of the drill bit.
Drilling for oil and gas is a highly specialized subject: pcD cutters were introduced
to this application area in the early 1980s and, since that time, the design of the drill
bits and the pcD cutting elements themselves have resulted in ever-increasing perfor-
mance.
A factor which limited the early pcD oil and gas exploration bits was premature
failure, either of the tungsten carbide supporting the pcD or of the braze attaching
the cutter to the bit body. Today, most of these difficulties have been overcome, and
pcD bits offer high penetration rates and long bit lives when drilling medium hard
abrasive rock formations.
pcD drill bits are also used in mining, although the volume is considerably less
than that in the oil and gas exploration industry. One example is in methane drai-
nage, where holes are drilled into the rock surrounding the coal seam and the
methane pumped out to prevent it seeping into the mining area and hence causing
a fire hazard. Also, in underground mining, holes have to be drilled in large quan-
tities for roof bolts to be inserted to provide roof support, and blast holes have to be
drilled into which explosives are packed. Figure 74 shows a selection of typical pcD
mining bits.
1.5.4.4 Non-cutting Application of Polycrystalline Diamond

pcD, being hard and abrasion resistant, is potentially ideal for use as a wear resistant
material. One traditional non-cutting application is in wire drawing, where pcD is
Figure 74. Selection of mining bits.
used as a die to reduce fine wire (Fig. 75). pcD wire drawing dies are most widely
used to draw non-ferrous wires such as copper where the life, compared to cemented
tungsten carbide, is up to 1000 times longer.
A more recent application field has been the use of pcD as a wear part, such as
a bearing, a workpiece support rest in a machining operation, a nozzle for
abrasive fluids and in high precision gauges. Figure 76 shows a pcD used as a
workpiece support rest for an automotive engine component and Fig. 77 shows
precision automatic gauging equipment fitted with pcD tipped measuring fingers.
In such cases, the high wear resistance of pcD leads to improved accuracy being
achieved.
Figure 75. Drawing wire with diamond dies.

1.5 Industrial Applications of’Diamond und cBN 557
Figure 76. SYNDITE pcD wear parts support a crankshaft during machining
1.5.4.5 Machining Ferrous Materials with Polycrystalline Boron Nitride

pcBN is used almost exclusively for machining hard ferrous materials. As described
earlier, cBN and pcBN do not react with iron at high temperatures in the same way
as diamond, and hence can be used to machine hard ferrous metals. To machine a
hard ferrous material effectively, the temperature of the workpiece in the zone imme-
diately ahead of the contact between it and the cutting tool (the shear zone) has to be
increased such that the workpiece becomes softer and hence easier to machine.
Figure 78 shows the hardness/temperature curve for a typical hard ferrous material.
By selecting the correct machining conditions of cutting speed and tool geometry,
these relatively high temperatures can be achieved, and because pcBN retains its
Figure 77. Measuring diamond honing heads with a pcD-tipped micrometer.

558 1 Diamond Materials und their Applicutions
Figure 78. Workpiece material hardness as a function of temperature.
hardness at these high temperatures, efficient machining can be conducted. The high
temperatures generated are limited to the cutting zone only, providing the machin-
ing conditions are selected correctly. The bulk temperature of the workpiece remains
low and coolant is only required in a minority of cases. pcBN tool materials fall lar-
gely into two classes:
- high cBN content (>go% by volume cBN) in a ceramic binder
- low cBN content ( ~ 6 0 %by volume cBN) in a ceramic binder
High cBN content materials are used in rough machining applications, whereas
lower cBN content tools tend to perform better in finishing applications where shal-
low depths of cut are being taken and a good surface finish is required.
Abrasion resistant cast-irons, for example Ni-hard, are used to manufacture
pumps for pumping slurry and gravel in the quarrying, mining and associated
heavy industries. These materials are difficult to machine with conventional tooling
materials but can be machined easily with a high cBN content tool material such an
Amborite. Steel rolling mill-rolls, which are normally ground to final size, can also
be machined with pcBN, the advantage being higher production rates and easier
swarf disposal since most rough machining operations using pcBN are conducted
dry, without coolant (Fig. 79). pcBN can also be used in milling, and Fig. 80
shows an example of a milling operation on a machine tool slideway.
In the automotive industry, high content pcBN is used to machine a wide range of
cast iron components such as brake discs (Fig. 81) and cast iron cylinder blocks
(Fig. 82).
Using low-content pcBN tools for finishing previous operations often replaces a
grinding operation. In many cases, a hardened steel component was finish ground
because its hardness meant that it was not possible to machine it in any other
way. Many such components may now be turned to final tolerance and surface
Ni-HARD Pump Bodies (left)

High Chromium Iron Rolls (right)
Figure 79. Typical applications of high cBN-content pcBN tools
finish using suitable pcBN tooling. Figure 83 shows a gear component being finished
in this way.
Low content pcBN can also be used in fine milling. Figure 84 shows a n automo-
tive gearbox component being finish milled with a low content pcBN (DBCSO).
1.5.5 Applications of Single Crystal Diamond

Natural diamond crystals were the original diamond tools used hundreds of years
ago for cutting and engraving purposes, because the outstanding properties of
Figure 80. AMBORITE milling hardened meehanite slideways.

5 60 I Diamond Materials und their Applicutions
Figure 81. Machining cast iron brake discs with AMBORITE.
hardness and abrasion resistance were recognized, but it was only possible to use
relatively large pieces of diamond, several millimeters in size. More recently, other
outstanding properties have been recognized, in particular the high degree of optical
transparency of type I1 diamonds and the very high thermal conductivity of (the
relatively rare) type HA diamond. In recent years, the technology to synthesize
Figure 82. AMBORITE machining a six-cylinder cast iron engine block.

1.5 Induxtrial Applications of Diamond and cBN 561
Figure 83. Finish machining a case-hardened automotive gear with DBCSO.
economically synthetic single crystal diamond of several millimeters in size has been
developed and this has contributed to the availability of relatively large diamonds
for industrial use.
Single diamond crystals are currently used in a wide variety of industrial appli-
cations, albeit some highly specialized, and these range from rock drilling to
sophisticated applications in micro-surgery and electronics.
One advantage of synthetic crystals is that they can be synthesized and processed
to specific required sizes and also crystallographic orientations chosen according to
the application (Fig. 85).
Figure 84. DBCSO finish milling a transmission component (60HRc).

i/--g
MONODITE \-Facet
MONODITE Triangle
Cube
Whole
Stone
4
-
c
MONODRESS
MONODITE Triangle
Octaiedron MONODIE 111

Figure 85. MONOCRYSTAL orientation.
1.5.5.1 Diamond Truing and Dressing Tools

Diamond tools are employed for truing and dressing conventions- abrasive w ,.eels,
like aluminum oxide and silicon carbide wheels. Truing ensures that newly installed
wheels run concentrically on the machine spindle, whilst dressing restores the desired
surface topography and texture to grinding wheels that have become dull, glazed, or
loaded with grinding debris. The standard types of truing and dressing tools available
include single-point, multi-point, chisel, rotary, and indexing tools (Fig. 86). This wide
variety caters for all types of conventional abrasive wheel and all types of truing and
dressing operations, such as those required to produce both straight and profiled
wheel faces, those that have steps, radii and shoulders, and for plunge forming etc.
Rotary dressers are used for truing and dressing vitrified-bond cubic boron nitride
wheels. Custom-made tools are also produced for specialized applications.
Figure 86. Selection of diamond dressers.

1.5 Industrial Applications of' Diamond and cBN 563
1.5.5.2 Diamond Cutting Tools

Diamond cutting-tool applications are many and varied, ranging from rock drilling
to the finish machining of nonferrous metals and plastics. In rock drilling with sur-
face-set diamond drill bits, the structural strength, fracture resistance, hardness and
wear resistance of diamond are most in evidence because rock is removed by crush-
ing rather than shearing.
Its crystal structure and extreme hardness also make it possible for diamond to
accept and retain cutting edges which are flawless and unbroken over their entire
length when viewed at high magnification.
Diamond tools are used for turning, milling and engraving non-ferrous metals,
including precious metals, non-metallic materials, abrasive composites etc.
(Fig. 87). Typical of applications for which diamond has been successfully used
are the turning of aluminum alloy photocopier cylinders and computer memory
discs, copper printing rolls, non-ferrous metal optical components and plastic
contact lenses, the fly-cutting of multi-facetted scanner mirrors, the production
of precious metal jewellery and the restoration of crazed plastic aircraft windows
to their original transparency. In such operations, diamond tools give reduced
downtime and improved quality, thus making it possible to achieve greater
economy in comparison with other types of cutting tool. In many cases, the
superior finish obtained with diamond machining means that no subsequent
polishing or buffing operations are required, thus improving the economics
even further.
Figure 87. Typical diamond cutting tools.

Figure 88. Surgical scalpel for micro-surgery.
In the medical field, diamond blades are used in surgical scalpels (Fig. 88) and in
microtome knives (Fig. 89) to prepare tissue samples for microscopic examination.
Diamond knives are also used for cleaving fibre optics.
1.5.5.3 Non-Cutting Applications of Diamond Tools

I n comparison with the use of diamond as a cutting tool material, the consump-
tion of diamond for non-cutting applications is small. Nevertheless, high-quality-
single crystal industrial diamond is often the only suitable material for specific
tasks, due to its outstanding mechanical, optical, electrical, chemical or thermal
properties.
Figure 89. Diamond microtome knife.

Figure 90. Ultra-high pressure diamond anvils.
The types of equipment in which diamond crystals are used include anvils for
high-pressure research (in excess of a million atmospheres or Megabar) (Fig. 90),
heatsinks, bearings for precision chronometers and other ultra-sensitive electronic
meters, styli for audio equipment and surface measuring instruments, hardness-
testing indenters (Fig. 91) and distance stops on machine tools.
In addition, diamond is used for metal-forming applications, such as burnishing,
engine turning, and the ruling of diffraction gratings, where little or no material is
actually removed. The most important of such forming operations is wire drawing.
In top-quality wire-drawing dies, the ultimate wear resistance and surface polish rely
on the particular qualities of diamond crystals. Such dies enable large quantities of
Figure 91. Diamond hardness indentor.

Figure 92. Diamond wire drawing dies.
very fine wire (down to 10pm in diameter) to be produced with the required
precision (Fig. 92).
Acknowledgments
The authors are indebted to Dr I. Sigalas and Dr R. J. Caveney for the overall
editing of the article, and to Mrs Chancellor Teffo for her very competent secretarial
help.
References
1. S. Tennant, Phil Trans. R. Soc. 1797,97, 591.
2. H . Liander and E. Lundblad, Arkiv Kemi, 1969, 16, 139-149.
3. F. P. Bundy, H. T. Hall, H. M. Strong, and R. H. Wentorf, Nature, 1955,176 (No. 4471), 51-55.
4. F. P. Bundy, Nature (London), 1973, 241, 116-1 18.
5. G. Davies, Diamond, Adam Hilger, Bristol, 1984, Chapter 4.
6. H. M. Strong, Am. J. Phys. 1989, 57, 794-802.
7. J. W. Harris, Diamond geology, in The Properties of Natural and Synthetic Diamond, J. E.
Field (Ed.), Academic Press, London, 1992, pp. 342-393.
8. S. E. Kesson and A. E. Ringwood, Chemical Geology, 1989, 78, 97-1 18.
References 567
9. F. P. Bundy, W. A. Bassett, M. S. Weather, R. J. Hemley, H. K. Mao, and A. F. Goncharov,

Carbon, 1996, 34, 141-153.
10. G. J. Davies and T. Evans, Proc. Ro. Soc. Lond., 1972, A328, 413427.
11. M. L. H. Green, M. J. Rosseinsky, and E. S. C. Tsang, 2nd International Interdisciplinary Col-
loquium on the Science and Technology of the Fullerenes, Keble College and University of
Oxford Museum, Oxford, Abstracts, Elsevier Science, Oxford, 1996, p. 186.
12. R. Roy, in Advances in Nav Diamond Science and Technology, S. Saito (eds), MYU, Tokyo,
1994, pp. 17-22.
13. P. Mistry, M. C. Turchan, S. Liu, G. 0. Granse, T. Baurmann, and M. G. Shara, Innovations
in Materials Research, 1996, 1(2), 193-207.
14. R. Berman and F. Simon, Zeitschrijt,fur Elektrochemie, 1955, 59, 333-337.
15. C. S. Kennedy and G. C. Kennedy, J . Geophys. Res., 1976,81(14), 2467-2469.
16. R. C. Burns and G. J. Davies, Chapter 10 in The Properties ofNaturalandSynthetic Diamond,
J. E. Field (Ed.), Academic Press, London, 1992.
17. H. Kanda, M. Akaishi, and S. Yamaoka, Appl. Phys. Lett. 1994, 65, 784.
18. S. K. Singhal and H. Kanda, J . Cryst. Growth, 1995, 154, 297.
19. N. V. Novikov and A. A. Shul’zhenko, in Science and Technology of”ew Diamond, S. Saito,
0. Kukunaga, and M. Yoshikawa (Eds), KTK Sci, Tokyo, 1990, p. 339.
20. A. V. Andreyev and H. Kanda, Diamond Relat. Mater., 1997, 6, 28-32.
21. H. 0.A. Meyer, Inclusions in Diamond in Mantle Xenoliths, P. H. Nixon (Ed.), John Wiley and
Sons, 1987, pp. 501-522.
22. M. Akaishi, H. Kanda, and S. Yamaoka, J . Cryst. Growth, 1990, 104, 578.
23. M. Akaishi, H. Kanda, and S. Yamaoka, Jap. Appl. Phys., 1990, 29, L1172.
24. M. Akaishi, Diamond Relat. Muter., 1993, 2, 183-189.
25. L. Sun, M. Akaishi, S. Yamaoka, H. Yamaoka, and S. Nakano, Special Issue of The Review
of High Pressure Science and Technology, AIRAPT-16 1997, 6, 36.
26. M. Arima, K. Makayama, M. Akaishi, S. Yamaoka, and H. Kanda, Geology, 1993, 21, 968.
27. M. Akaishi, Proceedings of 3rd NIRIM International Symposium on Advanced Materials,
Tusukuba, Japan, Mar 4-8, 1996,75.
28. H. Yamada, M. Akaishi, and S. Yamaoka, Special Issue of The Review of High Pressure
Science and Technology, AI RAPT-16, 1997, 6, 35.
29. S. E. Haggerty, Nature, 1986, 320, 34-37.
30. A. Onodera, K. Terashima, T. Urushihara, K. Suito, H. Sumiya, and S. Satoh, J . Mater. Sci.,
1997,32, 430943 18.
31. A. Barti, S. Bohr, R. Haubner, and B. Lux, Int. J . Refract. Metal HardMater., 1996,14,145-147.
32. G. R. Cowan, B. W. Dunnington, and A. H. Holtzman, U. S. Patent, USP34010119, Du Pont
de Nemours USA, 1965.
33. R. H. Wentorf Jr., J . Phys. Chem., 1965, 69, 3063-3069.
34. Y. Ma et a/., Appl. Ph)x Lett., 1994, 65, 822.
35. M. N. Rugueiro, P. Monceau, and J. L. Hodeau, Nature, 1992, 355, 237-239.
36. Y. Muranaka, H. Yamashita, and H. Miyadera, Diamond Re/. Mater., 1994, 3, 313-318.
37. H. E. Spear and J. P. Dismukes (eds), Parts I1 and 111 in, Synthetic Diamond, Emerging CVD
Science and Technology, John Wiley and Sons, 1994, pp. 21-304.
38. R. Roy, D. Ravichandran, P. Ravindranathan, and A. Badzian, J . Mater. Res., 1996, 11(5),
116441168.
39. P. W. Bridgman, Phys. Rev. 1935,48, 825-832.
40. L. F. Vereschagin, A. A. Semerchan, N. N. Kuzin, and Y. A. Sadkov, Dokl. Akad. Nauk S S R ,
1968, 183, 565-567.
41. B. von Platen, in Modem Very High Pressure Techniques, R. H. Wentorf (Ed.), Butterworths,
Washington, DC, 1962, pp. 234-239.
42. Yu. N. Pal’yonov, Yu. I. Malinvosky, Yu. M. Borzdov, A. F. Khokhryakov, A. I. Chepurov,
A,, Godovikov, and N. V. Sobolev, Doklady Akademii Nuuk S S S R , Earth Science Section
1990,315, 1221-1224.
43. H. Kanda and 0. Fukunaga, in Advances in Earth and Planetary Science, High Pressure
Research in Geophysics, V o l I 2 ,S. Akimoto and M. H. Manghnani (Eds), Riedel, Dordrecht,
1982, pp. 525-535.
44. Yu. N Pal’yonov, A. F. Khokhryakov, Yu. M. Borzdov, A. G. Sokol, V. A. Gusev, G. M.

Rydov, and N. V. Sobolev, Geol. Geojiz, 1997,38, 882-906.
45. H. M. Strong and R. E. Hanneman, J. Chenz. Phys., 1967,46, 3668-3676.
46. G. Will and G. Graf, High Press. Res., 1994, 12, 17-27.
47. I. Sunagawa, Morphology of natural and synthetic diamond crystals, in Material Science of
the Earth’s Interior, I. Sunagawa (Ed.), Tokyo Terra Scientific Publishing Co, 1984, pp.
303-330.
48. Y. Wang, R. Tanakabe, and M. Wakatsuki, Proc. 15th AIRAPT Int Conf. on High Pressure
Science and Technology, Warsaw, Sept 11-15, 1995, pp. 235-237.
49. M. Wakatsuki and K. Takaro, K. Inone, Inclusions in synthesised crystals of diamond, in 8th
AIRAPT Covf. Proc., High Pressure in Research and Industry, Uppsala, pp. 369-372.
50. T. L. McCormick, W. E. Jackson, and R. J. Nemanich, J. Muter. Res., 1997, 12(1), 253-263.
51. G. Davies, Chemistry and Physics ofcarbon, Vol. 11, P. L. Walker (Ed.), Marcel Dekker, New
York, 1982, pp. 70-71.
52. R. H. Wentorf, J. Phys. Chem. 1971, 75, 1833-1837.
53. H. M. Strong and R. E. Tuft, US Pat. 4073380, 1978.
54. Y. Suji, et al., Jpn. Pat 2631/1986, 165-167.
55. F. P. Bundy, H. M. Strong, and R. H. Wentorf Jr, Chemistry of Physics of Carbon, Vol. 10,
1973, 213-272.
56. M. Wakatsuki, in Materials Science of the Earth’s Interior I. Sunagawa (Ed.), Terra Scientific
Publishing Company, Tokyo, 1984, pp. 35 1-374.
57. Yu. A Kocherzhinskii and 0. G. Kulik, Powder Metal1 Metal Ceramics, 1996, 35(7-8), 470-
483.
58. H. Kanda, in Advances in New Diamond Science and Technology, S. Saito, N. Fujimori, 0.
Fukunaga, M. Kamo, K. Kobashi, and M. Yoshikawa (Eds), MYU, Tokyo, 1994, pp. 507-
512.
59. M. Wakatsuki, Jpn J. Appl. Phys. 1966, 5, 337.
60. T. Sugano, N. Ohashi, T. Tsurumi, and 0. Fukunaga, Diamond Relat. Muter. 1996,5,29-33.
61. H. M. Strong and R. M. Chrenko, J. Phys. Chem. 1971,75(12), 1838-1843.
62. R. C. Burns, S. Kessier, M. Sibanda, C. M. Welbourn, and D. L. Welch, Proceedings of the 3rd
NIRIM International Symposium on Advanced Materials, 1996, pp. 105-1 11.
63. H. Kanda, T. Ohsawa, 0. Fukunaga, and I. Sunagawa, J. Cryst. Outgrowth, 1989,94, 115-124.
64. P. Cannonand and E. T. Conlin, J. Am. Chem. SOC. 1964,86,45404544.
65. R. C. Burns, J. 0. Hansen, M. Sibanda, R. A. Spits, C. M . Welbourn, and D. L. Welch, Dia-
mond Re/. Muter., 1999, 8, 1433-1437.
66. H. Kanda and S. C. Lawson, Indust. Diamond Rev. 1995,2, 56-61.
67. H. Sumiya and S. Satoh, Diamond Re/. Muter., 1996, 5 , 1359-1365.
68. W. Li, H. Kagi, and M. Wakatsuki, Trans. Muter. Res. SOC. Jpn., 1994, 14B, 1451-1454.
69. M. Wakatsuki, W. Li, Y. Gohda, and L. Y. Ding, Diamond Rel. Muter., 1996, 5, 56-64.
70. S. Yazu, J. SOC. Muter. Sci. Jpn., 1993, 42(476), 588-592.
71. H. Kanda, New Diamond, 1990, 58-62.
72. R. A. Chapman, Private Communication, 1996, De Beers Diamond Research Laboratory,
South Africa.
73. S. Satoh and H. Sumiya, Jpn. Muter. Res., 1995, 15, 183-195.
74. R. C. Burns, V. Cvetkovic, C. N. Dodge, D. J. F. Evans, M. L. T. Rooney, and C. M. Wel-
bourn, J. Cryst. Growth, 1990, 104, 257-279.
75. S. Satoh, H. Sumiya, K. Tsuji, and S. Yazu, in Science and Technology of New Diamond, S.
Saito, 0. Fukunaga, and M. Yoshikawa (Eds), KTK Scientific Publishers/Terra Scientific
Publishing Co., Tokyo, 1990, pp. 351-355.
76. H. Kanda, T. Ohsawa, and 0. Fukunaga, in Abstracts 2nd Meeting of Diamond, Tokyo, 1987
pp. 23-24.
77. M. L. T. Rooney, J. Cryst. Growth, 1992, 116, 15-21.
78. A. T. Collins, H. Kanda, and R. C. Burns, Phil. Mag. B, 1990, 61(5), 797-810.
79. S. L. Lawson, H. Kanda, K. Watanabe, I. Kiflawi, and Y. Sato, J. Appl. Phys., 1996,79(8), 1-
10.
80. J. I. Koivula and C. W. Fryer, Gems and Gemology, 1984, 20, 146-158.
References 569
81. J. E. Shigley, E. Fritsch, C. M. Stockton, J. I. Koivula, C. W. Fryer, and R. E. Kane, Gems and
Gemology, 1986,22, 192-208.
82. J. E. Shigley, E. Fritsch, C. M. Stockton, J. I. Koivula, C. W. Fryer, R. E. Kane, D. R.
Hargett, and C. W. Welch, Gems and Gemology, 1987, 23, 187-206.
83. M. L. T. Rooney, C. M . Welbourn, J. E. Shigley, E. Fritsch, and I. Reinitz, Gems and
Gemology, 1993, 29. 3 8 4 5 .
84a. J. E. Shigley and E. Fritsch, J . Cryst. Growth, 1993, 128, 425428.
84b. J. E. Shigley, E. Fritsch, I. Reinitz, and T. M. Moses, Gems andGemology, 1995,31, 256-264.
85. C. M. Welbourn, M. Cooper, and P. M. Spear, Gems and Gemology, 1996, 32(3), 3 156-169.
86. M. Kamo, Y. Sato, S. Matsumoto, and N. Setaka, J . Cryst. Growth, 1983, 62(3), 642-644.
87. J. C. Angus, H. A. Will, and H. S. Stanko, J . Appl. Phys., 1968, 39, 2915.
88. W. G. Eversole, US Patent 3, 030, 187 and US Patent 3, 030, 188 1962.
89. K. V. Ravi, Mater. Sci. Eng., 1993 B19. 203-227.
90. R. Bennan, Solid State Commun., 1996, 99(1), 35-31.
91. K. T. Jacob, Solid State Commun., 1995, 94(9), 763-765.
92. C. J. Brannon and S. L. McCollum, in New Diamond Science and Technology, R. Messier et al.
(Eds), Materials Research Society, Pittsburgh, PA, USA, 1991, p. 117.
93. A. Tsuzuki, S.-I. Hirano, and S. Naka, J . Muter. Sci., 1985, 20, 2260-2264.
94. H. Uchikawa, H. Hagiwara, and K Nakamura, in Science and Technology of New Diamond, S.
Saito et al. (Eds), Terra Scientific Publishing Co., 1990. 227.
95. Hawtin, J. B, Lewis, N. Moul, and R. H. Philips, Roy Soc. Phil. Trans. Ser A , , 1966,261(116),
67-95.
96. L. F. Vereschagin, E. N. Yakoviev, L. M. Buchnev, and B. K. Dymov, Teplofizika Vysokikh
Temperafur, 1977, 1S(2), 316-321.
97. Y. Bar-Yam and T. D. Moustakas, Nature, 1989, 32, 786787.
98. A. A. Abramzon and A. A. Novozhenets, Russ. J . Phys. Chem., 1992, 66(8), 122c1222.
99. J. Abrahamson, Carbon, 1973, 11, 337-362.
100. J. 0. Hansen, R. G. Copperthwaite, T. E. Derry, and J. M. Pratt, J. Colloid Interface Sci.,
1989, 130(2), 347-358.
101. R. A. Munson, Carbon, 1967, 5, 471474.
102. J. A. Nuth, Astrophys. Space Sci., 1987, 139(1), 103-109.
103. P. Badziag, W. S. Verwoerd, W. P. Ellis, and N. R. Greiner, Nature, 1990,343(6255), 244245.
104. E. Anders and E. Zimmer, M eteoritics, 1993, 28, 49c514.
105. A. N. Nesterov, Y. I. Merezhko, and V. V. Chernikov, Run. J . Phys. Chem., 1983,57(2), 307-308.
106. D. V. Fedoseev, ColloidJ. USSR, 1978,40(2), 341-342.
107. D. V. Fedoseev, B. V. Deryagin I. G. Varshavskaya, and A. V. Laverentev, Prog. Surf. Sci.,
1994,4S, 84-87.
108. F. Banhart and P. M. Ajayan, Nature, 1996. 382,433435.
109. B. Wei, J. Zhang, J. Liang W. Liu Z Gao, and D. Wu, J . Muter. Sci. Lett., 1997, 16,402403.
110. H. Moissan, C. R. Acad. Sci. Paris, 1894, 118, 32G32.5.
111. J. B. Hannay, Nature, 1880, 22, 255-256.
112. Y. G . Gogotsi K. G . Nickel, and P. Kofstad, J . Muter. Chern. 1995, S(12), 2313-2314.
113. M. A. Cappeli, US Patent 5,417,953, 1995.
114. J-Y Wang, Y-Z Wan, D-W Zhang, and Z-J Liu, J . Muter. Res., 1997, 12(12), 3250-3253.
115. F. Davanloo, E. M. Juengerman, D. R. Jander, T. J. Lee, and C. B Collins, J . Muter. Res.,
1990, S(l l), 2398-2404.
116. R. Roy, H. S. Dewan, and P. Ravindranathan, Muter. Res. Bull., . 1993, 28, 861-866.
117. R. Roy, K. A. Cherian: J. P Cheng, A. Badzian, C. Langlade, H. Dewan, and W. Drawl, lnnov.
Muter. Res., 1996, 1(1), 165-87.
118. P. A. Molian and A. Waschek, J . Muter. Sci.,1993, 28, 1773.
119. M . Horio, A. Saito, K. Unou, H. Nakazono, N . Shibuya, S. Shima, and A. Kosaka, Chem.
Eng. Sci., 1996, Sl(1 l), 3033-~3038.
120. Y. Namba, J . Vac. Sci. Technol. A , , 1992, 10(5), 3368-3370.
121. M. Takaya, Japanese Patent Hei 6-25896, 1994.
122. T. Suzuki, T. Ishihara, T. Yamazaki, and S. Wada, Jpn J . Appl. Phys., 1997,36 part 2, No. 4B,
L50kL506.
123. J. F. Prins, Diamond Rel. Muter., 1993, 2, 646-655.

124. K. U. Cherian, M. Komath, S. K. Kulkarni, and A. Ray, Diamond Rel. Mater., 1994,4,20-25.
125. A. Jayaraman, Sci. Am., 1984,250,42-54.
126. Y. A. Litvin and L. T. Chudinovskikh, Trans. Russian Acad. Earth Sci., 1997,35SA(6),908-91 1.
127. F. P. Bundy and J. S. Kaspers, J . Chem. Phys., 1967,46(9), 3437-3446.
128. R. Boehier, Private communication, Jun 1994, Max Planck Inst, Mainz Germany.
129. J-A Xu and E. Huang, Rev. Sci. Instrum., 1994, 65(1), 204-207.
130. J. Kleiman, R. B. Heimann, D. Hawken, and N. M. Salansky, J . Appl. Phys. ( U S A ) , 1984,
56(5), 5 1440-1454.
131. 0. Bergmann and N. F. Bailey, in High Pressure Explosive Processing of Ceramics, R. A.
Graham and A. B. Sawaoka (Eds), Trans Tech duPont Wilmington, USA, 1987.
132. S. Fahy, S. G. Louie, and M. L. Cohen, Phys. Rev. B, 1986, 34(2), 191-199.
133. R. H. Wentorf, Jr., J . Chem. Phys., 1957, 4, 956.
134. R. H. Wentorf, Jr., New Diamond Sci. Technol., 1990, 1029-1037.
135. V. L. Solozhenko, J . Hard Materials, 1995, 6, 51-65.
136. N. V. Novikov and V. L. Solozhenko, J . Chem. Vap. Dep., 1996,4, 240-252.
137. R. H. Wentorf, Jr., J . Chem. Phys., 1961, 34(3), 809-812.
138. V. L. Solozheuko, Adv. Muter. ‘96: Proceedings of the Third NIRM International Symposium,
Tsukuba, Japan, 1996, pp. 119-124.
139. F. P. Bundy and R. H. Wentorf, J. Chem. Phys., 1963,38, 11441149.
140. Y. Moriyoshi, M. Kamo, N. Setaka, and Y. Sato, J. Muter. Sci., 1983, 18(1), 217-224.
141. H. T. Hall, Rev. Sci. Instrum., 1958, 29(4), 267.
142. H. T. Hall, Science, 1970, 169, 868-869.
143. H. D. Stromberg and D. R. Stephens, J . Am. Ceram. Soc., 1970,49(12), 1030-1032.
144. H. Katzman and W. F. Libby, Science, 1971, 172, June 1971, 1132-34.
145. R. H. Wentorf, W. A. Rocco, Patent SAP 7315038, 1973.
146. L. F. Veraschagin, A. A. Semerchan, Patent UKP 1382080, 1975.
147. A. I. Kolomiitsev and V. E. Smirnov, Almazyi Sverkhtverdye Materialy, 1980, 7, 4 5 .
148. K. Uehara and S. Yamaya, Science and Technology of New Diamond, S. Saito (Ed.), Tokyo,
KTK Sci Publ., 1990, pp. 203-209.
149. A. A Shulzhenko, V. G. Gargin, V. A. Shishkin, and A. A. Bochechka, Ukrainian Academy of
Sciences, Institute for Superhard Materials, printed in Kiev Naukova Dunko, 1989, Chapter 4,
93-173.
150. Yu A. Kocherzhinskii, A. A. Shulzhenko, and V. A. Shishkin, Vliyanie Vysok Davieniya Na
Structuru Z Svoistva Materialov, 1983, 34-40.
151. 0. A. Voronov and A. A. Kaurov, Soviet J . of Superhard Materials, 1994, 16(1), 6 1 0 .
152. S. H Robertson, PhD Thesis, University of Reading, 1984.
153. D. V. Fedoseev, G. A. Sokolina, and E. N. Yakoviev, Sov. Phys. J., 1985, 30(5), 425427.
154. D. V. Fedoseev, G. A. Sokolina, E. N. Yakoviev, and A. V. Lavrentev, Sov. Powder Metall.
Ceram. Muter., 1984, 23(9), 683-686.
155. A. A. Shulzhenko, S. A. Bozhko, I. A. lgnatusha, and A. N. Vashchenko, Sov. J . Superhard
Muter., 1990, 12(5), 14-19.
156. A. A. Shulzhenko, S. A. Bozhko, A. I. lgnatusha, and A. N. Vashchenko, Sov. J . Superhard
Muter., 1988, 10(5), 13-17.
157. S. A. Bozhko, A. I. Ignatusha, V. G. Delevi, and A. A. Budyak, Sov. J. Superhard Muter.,
1989, 11(5), 32-35.
158. M. Akaishi, S. Yamaoka, J. Tanaka, T. Ohsawa, and 0.Fukunaga, J . Am. Ceram. Soc., 1987,
70(10), C237-C239.
159. J. C. Walmsley and A. R. Lang, J . Muter. Sci., 1988, 23(5), 1829-1834.
160. R. H. Wentorf, Jr. and W. A. Rocco, U. S. Patent 3 767 371, 1973.
161. Sumitomo Electric Industries, Ltd., US Patent 4 334 063, 1982.
162. A. A. Shulzhenko, S. A. Bozhko, A. N. Sokolov, A. Petrusha, N. P. Bezhenar, A. I. Ignatusha,
Cubic Boron Nitride Synthesis, Sintering and Properties, The Ukrainian Academy of Sciences,
Institute for Superhard Materials, Kiev Naukova Dumka, 1993, 254.
163. P. N. Tomlinson, R. J. Wedlake, Znd. Diamantenrundschau, 1983, 7(4), 234241.
164. M. Wakatsuki, K. Ichinose, and T. Aoki, Muter. Res. Bull., 1972,7, 999-1004.
References 57 1
165. M. Wakatsuki and K . Ichinose, 4th Internutionul Conference on High Pressure, Kyoto, 1974,
441445.
166. F. R. Corrigan, 6th AIRAPT Conference on High Pressure Science and Technology,' 1979, 1,
994-999.
167. 0. Fukunaga and M. Akaishi, High Press. Res., 1990, 5, 911-913.
168. J. M. Leger, J . Haines, and B. Blanzat, J . Muter. Sci. Lett., 1994, 13, 1688-1690.
169. M. C. Payne, M. T. Teter, D. C. Alan, T. A. Arias, and J. D. Ioannopoulos, Rev. Mod. Phys.,
1992,64, 1045.
171. M. L. Cohen, J . Hard Mater., 1991, 2, 13.
172. P. K. Lou, M. L. Cohen, and G. Martinez, A m . Phys. Soc., 1987, 35, 919&9194.
173. R. Hoffmann, T. Hughbanks, M. Kerttesz, andP. H. Bird,J. A m . Chem. Soc., 1983,105,4831.
174. M. A. Tamor and K . C. Hass, J . Muter. Res., 1990, 5, 2273.
175. A. Y. Liu, M. L. Cohen, K. C. Hass and M. A. Tamor, Phys. Rev. B, 1991,43, 6742.
176. F. Diedrich and Y. Rubin, Angew. Chem. Int. Ed. Engl., 1992, 31, 1101.
177. H. Alberts. T. Sekine, H. Kanda, Y. Bando, and K. Hojou, J . Mater. Sci.Lett., 1990,9, 1376.
178. A . Y. Liu and M. L. Cohen, Phys. Rev. B., 1990,41, 10727-10734.
179. J. V. Badding, Adv. Mater., 1997, 9, 877-886.
180. A. Y. Liu and R. M. Wentzcovitch, Phys. Rev., 1994, 1350, 10362.
181. R. M. Wentzcovitch and J. L. Martins, SolidState Commun., 1991, 78, 831.
182. R. M. Wentzcovitch, Phys. Rev., 1991, B44, 2358.
183. D. M. Teter and R. J. Henley, Science, 1996, 271, 53-55.
184. D. Li, S. Lopez, Y. W. Chung, M. S. Wong. and W. D. Sproul, J . Vuc.Sci. TecAnol., 1995,13,
1063.
185. T. A. Yeh, C. L. Lin, J. M. Sivertsen, and J. H. Judy, IEEE Trans. Laga., 1991, 27, 5163.
186. J. J. Cuomo, P. A. Leary, D. Yu, W. Reuter, and M. Frisch, J . Vac. Sci. Technol., 1979, 16,
299.
187. H. Sjostrom, W. Lanford, B. Hjovarso, K. Xing, and J. E. Sundgren, J . Muter. Res., 1996, 11,
981.
188. C. Niu, Y. Z. Lu. and C. M. Lieber, Science, 1993, 261, 334.
189. K. Ogata, J. F. D. Chubaci, and F. Fujimoto, J . Appl. Phys., 1994, 76, 3791.
190. J. Schwan, W. Dworschak, K . Jung, and H. Erhardt, Diamond Rel. Muter.. 1994,3, 1034.
191. T. Y. Yen and C. P. Chon, Appl. Phys. Lett., 1995, 67, 2801.
192. Y. Guo and W. Goddard 111, Chem. Phys. Lett., 1995,237,12-76; J. E. Lowther, Phys. Rev. B ,
1998,57, 57245727.
193. D. M. Bhusari, C. K. Chen, K. H. Chen, T. J. Chuang, L. C. Chen, and M. C. Liu, J . Muter.
Res., 1997, 12, 322-325.
194. Z. J. Zhang, S. Fan, J. Huang, and C. M. Lieber, Appl. Phys. Lett., 1996, 68(19), 2639-2641.
195. F. Fujimoto and K. Ogata, Jpn. J . Appl. Phys., 1993, 32, L. 420.
196. D. Liu, Y. W. Chung, M. S. Wong, and W. D. Sproul, J . Appl. Phys., 1993 74, 219.
197. Z. M. Ran, Y. C. Du, Z. F. Ying, Y. X. Qiu, X. X. Xiong, J. D. Wu, and F. M. Li, Appl. Phys.
Lett., 1994, 65, 1361.
199. S. Kumai and T. L. Tansley, Solid State Commun., 1993, 88, 803.
200. D. Norton, K. J. Boyd, A/H. Al-Bayari, S. S. Todorov, and J. W. Rabalais, Phys. Rev. Lett.,
1994, 73, 118.
201. K. M. Yu, M. L. Cohen, E. E. Hailer, W. L. Hausen, A. Y. Liu, andI. C. Wu, Phys. Rev., 1994,
B49, 5034.
202. J. P. Riviera, D. Texier, J. Delafond, M. Jaouen, E. L. Mathe, and J. Chanmond, Muter. Lett.,
1995, 22, 115.
203. X. W. Su, H. W. Song, F. Z. Cui, W. Z. Li, and H. D. Li, Surj: Coat. Technol., 1996,84,388-391.
204. Y. Zhang, H. Li, and Q. Xue, Muter. Res. Soc., 1996, 317-322.
205. C. Yan, Liping Guo, and E. G. Wang, Phil. Mag. Lett., 1997, 75, 155-162.
206. K. Yanamoto, Y. Koga, K. Yase, S. Fuziwara, and M. Kubota, Jpn. J . Appl. Phys., 1997,36,
L23CL233.
207. E. G. Wang, Chen Yan, and Guo Liping, Physica Scripta, 1997, T69, 108-114
208. M. J. Yacaman, J. J. M. Gil, and M. Sarikaya, Muter. Chem. Phys., 1997,47, 109-117.
209. D. W. Wu, W. Fan, H. X. Guo, M. B. He, X. 0. Meng, and X. J. Fan, Solid State Commun.,
1997, 103, 193-196.
210. S. Xu, Han-Shi Ki, S. Lee, and Yin-Au Li, IEEE, 1997,213-214.
21 1. Xiao-Ming He, Li Shu, Wen-Zhi Li, and Heng-De-Li, J . Muter. Res., 1997, 12, 1595-1602.
212. Wu Dawei, Fu Dejan, Guo Huaixi, Zhang Zhihong, Meng Xianquan, and Fan Xiangiun,
Phys. Rev. B, 1997,56,49494954.
213. Chen, L. L., D. M. Bhusari, C. Y. Yang, K. H. Chen, T. J. Chuang, M. C. Lin, C. K. Chen,
and Y. F. Huang, Thin Solid Films,1997, 303, 66-75.
214. J. Badding, Adv. Mater., 1997, 9, 877.
215. K. Ploog, P. Rauh, W. Stoeger, and H. Schmidt, J . Cryst. Growth, 1972, 13/14. 350.
216. K. Ploog, H. Schmidt, E. Amberger, G. Will, and K. H. Kossobutzki, J . Less-Common Met.,
1972, 29, 161.
217. H. Saitoh, K. Yoshida, and W. A. Yarbrough, J . Muter. Res., 1993, 8, 8.
218. L. Vel, G. Demazeau and J. Etournea, Mat. Sci. Eng., 1991, BID, 149.
219. A. R. Badzian, T. Niemyski, and E. Ockusmik, Proc. 3rd lnt. Con$ on CVD, F. A. Glaski
(Ed.), The Electrochemical Society, Penington, NJ, 1972, p. 747.
220. K. Montasser, S. Hattori, and S. Morita, Thin Solid Films,1984, 117, 311.
221. J. Loeffier, F. Steinbach, J. Bill, J. Mayer, and F. Aidinger, Z. Metailkol, 1996, 87, 170.
222. D. Hegermann R. Riedel, W. Dresier, C. Oehr, B. Schindler, and H. Brunnen, Chem. Vap.
Depos., 1997, 3, 257
223. A. Badzian, Appl. Phys. Lett., 1988, 53, 2494.
224. P. F. McMillan, W. T. Petuskey, H. Hubert, K. J. Kingma, L. A. Garvie, A. Grzechnik, and A.
Chizmeshya, Adriutico Research Conference, Miramare, Trieste, Italy, July 1997.
225. H. Itoh, I. Maekawa, R. Yanamoto, and H. Iwahara, Rev. High Press. Sci. Technol., 1998,7,
986-988.
226. C. Ellison-Hayashi, M. Zandi, F. J. Csillag, and Shih-Yee Kuo, US Patent 5, 135, 892, 19912.
227. C. Ellison-Hayashi, M. Zandi, D. K. Shetty, P. Kuo, R. Yeckley, and F. Csillag, US Patent
5330,937, 1994.
228. J. M. Leger, J. Haines, M. Schmidt, J. P. Petitet, A. S. Pereira, and J. A. H. da Jomada, Nature,
1996,383,401.
229. Federation Europkene des Fabricants de Produits Abrasifs (FEPA), Standard for Superabra-
sives Grain Sizes, 2nd Edn. September, 1997.
2 Applications of Diamond Synthesized by

Chemical Vapor Deposition
R. S. Sussmann
2.1 Introduction
The ability of Chemical Vapor Deposition (CVD) technology to synthesize diamond
as large plates of high and controlled purity [l-51 is enabling a host of new
applications in science and technology that were hitherto hampered by the practical
difficulties inherent in the use of natural diamond. The basic principles of CVD for
the synthesis of polycrystalline diamond have been discussed in the chapter by
Klages et al. [6]. The aim of this chapter is to discuss a selected range of applications
to illustrate how CVD diamond is emerging as a key (and in some cases unique)
engineering material.
The impact that CVD diamond is starting to have on some areas of modern
technology is in some way similar to the impact that the invention of the high
pressure and high temperature (HPHT) techniques for the synthesis of diamond
had on the abrasive industry in the early sixties [7]. At that time the industrial
implementation of HPHT technology gave a renewed impetus to an existing but
relatively small diamond abrasive industry. HPHT synthesised diamonds have
been the foundation of what is now a relatively mature industry in which the
extreme hardness and wear resistance of diamond are exploited for numerous
abrasive and cutting tool applications [S].
The potential importance of diamond in technology follows from the wide range
of technically desirable thermomechanical, optical, and electronic properties
exhibited by diamond [9-131. In addition to its extreme hardness and exceptional
wear resistance, diamond is the stiffest known material, has a broad transmission
spectral range (from the ultraviolet to the far infrared and extending to microwave
frequencies), has the highest room-temperature thermal conductivity, one of the
lowest thermal expansion coefficients and is radiation hard and chemically inert
to all acid and base reagents.
What makes diamond so attractive, and in many cases unique, is not only the
extreme value of some of the above properties such as hardness, thermal
conductivity or the Young modulus, but the combination of two or more of
these properties in a high performance product, as will be discussed later in this
chapter.
The enormous interest that CVD diamond has generated since the mid eighties
has been based on the perception that this new diamond synthesis technology
could overcome all the limitations previously encountered with the availability of
large size natural diamond specimens [ 14,151.As was the case in the abrasives indus-
try, the use of natural diamond in other fields of technology has been relatively small
514 2 Applications of Diamond Synthesized by Chemical Vapor Deposition
and restricted to niche applications in optics [ 161, the thermal management of small
electronic devices (such as laser diodes or impatt oscillator diodes) [17,18], and
others [19,20]. For these types of applications only the relatively rare high-purity
Type IIa form of natural diamond could be used. The most abundant form of
natural diamond, Type Ia, contains relatively large concentrations of nitrogen
which degrade properties such as optical transmission [21],thermal conductivity
[22],and electronic transport [23]. Another major difficulty in the use of natural
diamond is the stringent limitation to the size of specimens that are practically avail-
able. Most natural Type IIa windows sold commercially are below 5 mm in diameter
and typical heat sinks are smaller than 1 x 1 mm [16, 181. The cost of natural single
crystal windows of sizes in excess of 19 mm, if available, would be outside the budget
range of most practical applications [16]. The largest and the most spectacular nat-
ural diamond product ever made was a window, 18.2mm in diameter and 2.8 mm
thick, used for the main pressure cell of the pioneer Venus probe launched on
August 1978 [24]. Purity and size are also serious limitations in the use of HPHT
synthetic diamond in these types of applications because of the high intrinsic
single substitutional nitrogen concentration and the practical maximum size of typi-
cal HPHT diamond crystals [7].
It is within this perspective that efforts in recent years have been directed to the
development of a CVD diamond technology that allows the routine manufacture
of large plates of diamond of high and controlled purity. Some of this effort has
met with considerable success [l-5,251. In the following sections of this chapter
we will briefly review the properties of various types of CVD diamond materials
that have been developed to address specific applications. We will illustrate through
a few selected examples how the current ability to manufacture CVD diamond
(including large area plates and three-dimensional shapes) as a reliable and robust
engineering material is enabling the use of diamond in an increasing range of tech-
nically demanding applications.
2.2 Properties of Chemical Vapor Deposited Diamond

2.2.1 Material Grades
For most practical applications CVD diamond is synthesized as a polycrystalline
material. Many of its properties can be expected to be strongly influenced by its
grain structure, including factors such as grain size, preferred orientation, intergra-
nular purity, microstress, etc. All of these factors depend strongly on the synthesis
technique and process conditions [7] and although there are generic features which
are common to all CVD diamond materials, substantial variations in properties such
as tensile strength, optical transparency, thermal conductivity and others can be
found in material from different origins. It is not yet possible, for instance, to
define a generic optical grade of CVD diamond because optically transparent
CVD diamond specimens prepared in different laboratories are known to exhibit
2.2 Properties of' Chemical Vapor Deposited Diamond 575
Figure 1. Optical grade CVD diamond window, 100mm diameter, 0.7mm thick
different values of fracture strength, intrinsic bulk absorption, and total forward
scatter [26,27].
The majority of the properties described in this section refer to the type of CVD
diamond that is currently available as a commercial product sold under the trade
name of DIAFILM [28]. With sufficient control of synthesis conditions, the extra
degree of freedom conferred by the polycrystalline structure has been used to
formulate different 'grades' of CVD diamond in which properties are optimized
for specific applications. Optical grades, for instance, exhibit the highest optical
transparency and can be made routinely in discs up to 100 mm in diameter. In con-
trast mechanical grades are dark in color but have a higher mechanical strength and
can be made in plates up to 160mm in diameter. Various thermal grades have been
developed for thermal management applications which offer a compromise between
good thermal conductivity and cost. Most of the applications currently pursued
with CVD diamond are relatively new and in certain cases the quality of the
corresponding grade is subject to continuous development in which factors such
as cost and performance are gradually optimized.
Figure 1 shows an example of a l00mm diameter, 0.7mm thick, optical grade
CVD diamond window which has been polished on both sides to an optical
finish. The as-grown grain morphologies of optical (2 mm thick) and mechanical
(0.5 mm thick) grade samples are shown in Fig. 2. Figure 3 shows the Raman spec-
trum of an optical grade CVD diamond sample measured by a Renishaw instrument
with an excitation wavelength of 633nm. The main Raman line is centered at
1332cm-' and the full-width-half-maximum (FWHM) line width is approximately
3.0cmP', which compares well with typical values of 2.8 cm-' to 3.1 cm-' observed
in high-quality single-crystal Type IIa natural diamond specimens measured with
the same instrument.
576 2 Applications of' Diamond Synthesized by Chemical Vapor Deposition
Figure 2. (a) SEM of the as-grown morphology of an optical grade DIAFILM sample 2mm thick.
(b) SEM of the as-grown morphology of a mechanical grade DIAFILM layer 0.5 mm thick.
2.2.2 Optical Properties

For a sample of optical grade CVD diamond 1 mm thick in which both faces have
been polished to an optical finish the optical transmission from the fundamental cut-
off at 220 nm to the far infrared (IR) part of the spectrum is shown in Figs 4 and 5.
For comparison, the transmission of a high-quality single-crystal natural Type IIa
sample is also shown. For wavelengths beyond the fundamental cut-off at 220nm
the transmission rises to reach values close to the theoretical reflection limited
value of approximately 71 'YO assuming a refractive index of approximately 2.4
[29,30]. At shorter wavelengths there is a reduction in transmission relative to the
Type IIa sample, which can be partly attributed to scattering and, close to the
fundamental edge, to a bulk absorption mechanism [311.
2.2 Properties of Chenzical Vapor Deposited Diamond 577
10,000
.-yE
n
8,000
4Q 6,000
a
Y
.w 4,000
C
3
0
0 2,000
n
(200 1,300 1,400 1,500 1,600 1,700
Raman shift from 633 nm (cm-1)
Figure 3. Raman spectrum of an optical grade CVD diamond specimen measured with a Renishaw
instrument with an excitation wavelength of 633 nm.
Beyond the intrinsic, multi-phonon absorption bands (between 4000 cm-' and
1500 cm-') [30], the transmission of the CVD diamond specimen reaches the maxi-
mum reflectivity limited value (Fig. 5) of 71.4% (using a refractive index value of
2.375 [32,3]) and is identical, within experimental accuracy, to that of high-quality
single crystal Type IIa samples measured with the same instrument.
80
Single crystal Type Ha, 0.5 mm thick
I
60
.-0
C
.-
v)
u)
40
E
s
20
n
u
500 1,000 1,500 2,000 2,500
Wavelength (nm)
Figure 4. Transmission spectrum for an optical grade CVD diamond window 1.0mm thick in the
UV-visible-near IR spectral range compared to that of a high quality natural Type IIa window
0.5 mm thick.
---------- 1
.-
v) Optical grade DlAFlLM
v) 1.0 mm thick sample
71.4% transmission
L - 1 ~
0
3,000 2,000 1,000 500
Wavenumber cm-I
Figure 5. Transmission spectrum for an optical grade CVD diamond window 1.Omm thick in the
infrared spectral range.
Transmission measurements have been recently reported [33] extending the range
of the spectrum of Fig. 5 to much longer IR wavelengths (to 20 cm-' or 500 pm)
showing that there are no measurable absorption features within that range.
The intrinsic absorption at 10.6 pm wavelength of optical grade samples measured
by calorimetry is in the range 0.10-.029 cm-' [27,34]. It is thought that some of these
values may be affected by surface absorption. This aspect is currently being investi-
gated. As will be discussed below, values of intrinsic absorption below 0.1 cm-' are
acceptable for most practical applications, due to the very high thermal conductivity
of diamond, coupled with the low change in refractive index with temperature.
The low angle scattering of optical grade CVD diamond at ~ 1 pm 0 wavelength
has been assessed by measurements of modulation transfer function (MTF) using
samples of optical grade CVD diamond with accurately polished surfaces to evalu-
ate one aspect of their performance as imaging windows [4,35]. The MTF is the best
parameter available for defining image quality. A modulation transfer function
curve effectively shows the contrast in the image as the spatial frequency of that
image changes (so that the curve will always drop to zero at a high enough spatial
frequency). The MTF of the components was determined from the width of the
image of a slit source. Any component which is placed in the light path which has
a poor optical performance (whether because of form shape errors or scatter) will
increase the broadening of the image of the slit source (the line spread function,
LSF).
An infrared interferometer was used to determine the LSF due to the form shape
error and this can be subtracted from the total broadening as measured in the infra-
red MTF instrument. What remains is the broadening due to the scatter alone.
Figure 6 shows a comparison of the LSF for the system with and without a
sample showing that the CVD diamond optical window introduces only a very
small broadening. The computer converts these data into an MTF trace such as
2.2 Properties of' Chemical Vapor Deposited Diamond 579
2,500
2,000
e
$-= 1,500
b
'i1,000
32
- FWHM = 115 pm
500
n
-200 -300 -200 -100 0 100 200 300 400
Position (microns)
Figure 6. Comparison of the Line Spread Function measured with and without a sample in the
system.
that in Fig. 7 (MTF vs RSF, reduced spatial frequency) which also shows the MTF
degradation caused by form shape errors and the M T F of the system with no
window in place (the diffraction limited curve). The effect of scatter is therefore
reflected in the difference between the lower and middle curves. The conclusion is
that optical grade CVD diamond shows minimal scatter and degradation in MTF
diffraction limited 1.s.f. measurement interferometerresults

+ .. .A- - - . ... .Q ..
Figure 7. Modulation Transfer Function (MTF) measured with a CVD diamond window (lower
curve), the M T F degradation caused by form shape errors (middle curve) and the MTF of the
system with no window in place showing the diffraction limited case (upper curve).
in the far infrared and is therefore a good candidate for windows required in ima-
ging systems. Such an application will be discussed in 2.3.1.
2.2.3 Strength of Chemical Vapor Deposited Diamond

C. S. J. Pickles
The fracture stress of CVD diamond has been evaluated using the three-point bend
geometry [15]. The details of these measurements and its relative merits have been
discussed elsewhere [4,35].
The fracture stress is plotted against thickness in Figs 8 and 9 for the growth
(coarse grain) and nucleation (fine grain) surfaces respectively, for the optical and
mechanical grades of material. All of the samples in this study were in the as-
grown state. These strength data suggest several conclusions.
First, the measured strength depends strongly on thickness. There is an increase in
grain size on the growth side with thickness and it is speculated that the size of the
critical flaw is limited by the grain size close to the tested face, as discussed
previously [4,25,35]. The variation of the stress level through the thickness is
another factor which may influence the thickness dependence of the strength. A
more precise analysis has been performed in which the experimental data seem to
be in good agreement with the model, as reported elsewhere [35].
Second, the reproducibility of the strength values is high. If the data from samples
of thickness between 0.4 and 1.4 mm are assessed, they give the Weibull moduli [ 151
shown in Table 1 (a higher value indicates a more reproducible strength, c.f.
sapphire = 2.1, zinc sulphide = 5.4, silicon carbide = lo). These can be corrected
to allow for the thickness effect [25,35]. The corrected values for the growth surface
are also given in the table.
The observations suggest that CVD diamond has bulk (rather than surface)
IGrowth surface strength 1

h
m
Q
600
8
e
c
u) 8Optical grade
200
0 0.5 1 15 2 2.5 3
Thickness (mm)
Figure 8. Fracture stress as a function of thickness measured with the growth face in tension.
2.2 Properties of Chemical Vapor Deposited Diamond 58 1
1400
T 1200 -
2i
* loo0 -
I
v)
P 600
8oo
I I I I I
400
strength controlling flaws [26] which are related to the size of the crystal grains.
2.2.4 The Young Modulus

Early measurements of the Young modulus on relatively thin specimens (up to
300 pm) gave values in the range 986-1079 GPa [2,36]. More recent results have
been obtained on optical and mechanical grade specimens as a function of tempera-
ture in the range 20-800°C [37]. For temperature values below 700°C (which is the
onset of oxidation in diamond) the Young modulus is relatively insensitive to tem-
perature as shown in Fig. 10. The average value for the Young modulus in optical
grade material at room temperature is 1 133 GPa and that of mechanical grades is
1166 GPa. Results published previously using samples of a different origin [38]
show that values between 242 and 539 GPa may be exhibited if the CVD diamond
contains a high concentration of defects such as micron-size voids or microcracks.
This illustrates the great variability in material properties that can be found in CVD
diamond of different origins and the need for CVD diamond suppliers to provide
precise material specifications to end-users.
Table 1. The Weibull moduli calculated from the data in Figs 8 and 9
CVD Growth Growth surface (corrected Nucleation

diamond type surface for thickness variation) surface
Mechanical grade 9.8 11.6 6.5

Optical grade 11.6 23.1 11
1160
-2 1140
1120
9
-g
1
1100
1080
-& 1060
c
ga 1040
1020
1000
0 100 200 300 400 500 600 700 800 900
Temperature [“C]
Figure 10. Young modulus of an Optical Grade CVD diamond specimen in the 20°C to 800°C tem-
perature range [37].
2.2.5 Thermal Conductivity

The thermal diffusivity of CVD diamond has been measured over the temperature
range 200425 K using the thermal flash technique which measures the thermal
diffusivity perpendicular to the plane of the layer (through-the-plane) [l, 31.
Figure 11 shows the temperature dependence of the thermal conductivity for
different grades of CVD diamond: optical grade, and three grades of thermal
I I
A
Y
E.
4000
P
3 3000
u
S
U
s 2000
0
I I
150 200 250 300 350 400 450
Temperature (K)
Optical Thermal 1 Thermal 2 Thermal 3 Type 118 natural
0 4 * -
Figure 11. Temperature dependence of the thermal conductivity for different grades of DIAFILM
CVD diamond: optical grade and three thermal grades.
DIAFILM specimens. These results are compared with literature values of natural
Type IIa single crystal diamond [39]. The difference between the thermal conductiv-
ity of optical and thermal grade specimens becomes more appreciable at lower tem-
peratures. The optical grade specimen seems to follow the Type IIa trend reaching a
value of 4000 W mK-’ at 200 K, probably the highest reported so far for polycrys-
talline CVD diamond at that temperature. For the ‘thermal-3’ grade, which shows
the lowest conductivity and is of dark color, the temperature dependence is not very
pronounced, in agreement with previous publications [40,41]. The ‘thermal-1’ grade
has a thermal conductivity better than 1800W mK-’ at room temperature which is
adequate for most demanding applications, as discussed in 2.5.2. The ability to
define different grades of thermal material is an example of the cost-performance
optimisation that is possible with CVD diamond.
2.2.6 Dielectric Properties

At microwave frequencies in the range 72-145 GHz, the critical parameters for high-
power transmission are the dielectric characteristics of the window material: the
dielectric loss factor tan 6 and the permittivity E: (or the refractive index n = &:I/*)
because they affect power absorption and reflection [42]. The dielectric loss factor
tan 6 in low loss samples is usually measured as the decrease in the Q factor of a reso-
nant cavity [43]. Low dielectric loss materials find application as the output windows
of high-power microwave tubes. A specific case is that of windows for Gyrotron
tubes operating in the 70-170 GHz frequency region with output powers in excess
of 1 MW, as will be discussed later.
Values of the dielectric loss of CVD diamond have been measured over the past 3
years as a suitable material grade for dielectric window applications was being devel-
oped [5]. For open resonant cavity measurements, samples are usually required to be
of at least 30 mm in diameter and of thickness in excess of 0.87 mm depending on the
measurement frequency and the accuracy required. For recent CVD diamond, values
of tan 6 below lop5 have been achieved. A specific example is a window l00mm in
diameter and 1.6mm thick which exhibited a tan 6 value of 0.6 (& 0.2) This
is the lowest value so far reported for CVD diamond and would enable the material
to be used as output windows in Gyrotron tubes of powers in excess of 2 M W as
discussed in 2.4.
2.3 Optical Applications

In this section, applications will be discussed which illustrate the versatility and
advantages of CVD diamond as an infrared and multi-spectral window material.
We describe the use of CVD diamond optical elements including CVD diamond
domes and flat plates as windows for IR seekers or imaging systems in high-speed
flight or other mechanically aggressive environments. Then we describe the use of
CVD diamond windows for the transmission of high-power IR laser beams.
2.3.1 Chemical Vapor Deposited Diamond for Passive Infrared

Windows in Aggressive Environments
C. J. H. Wort
The extreme hardness and abrasion resistance of diamond makes this material an
ideal choice for applications in which the optical components are exposed to aggres-
sive environments. The use of diamond for optical elements for infrared seeking mis-
siles and other military uses constitute probably the best examples of this type of
application as discussed in more detail in this section. There are, however, a host
of other applications in industry outside the military field in which the attributes
of diamond optics are of great advantage. These include windows used in the mon-
itoring of chemical reactions, or in the analysis of fluids which contain abrasive com-
ponents.
Originally conceived during the 1940s to allow armed forces to operate effectively
under the cover of darkness, infrared thermal imaging fulfilled a need that could not
be met with radar, acoustic, or visible sensors [44]. Thermal imaging relies on sen-
sing the heat emitted by a body, by virtue of its temperature alone; thus thermal ima-
ging is purely passive in its operation. The wavelength of the emitted light is
temperature dependent and the IR detector is optimized for maximum sensitivity
over a limited wavelength (and hence temperature) range.
Infrared detector materials are delicate and require protection from the environ-
ment by a transparent window. In many military applications this environment can
be extremely harsh, such as the environment encountered by a high-speed, heat-
seeking missile or a tank window exposed to a sandstorm. The damage caused by
a high velocity rain drop impact onto a zinc sulphide window [45] is clearly illu-
strated in Fig. 12. The erosion/impact damage seriously affects the component
strength and also the ability to form an acceptable image behind the window.
Owing to the absence of strong absorption by water or carbon dioxide molecules
in the atmosphere, two wavebands are basically transparent and are generally used
by IR imaging equipment. The mid waveband is between 3 and 5 pm wavelength and
the long waveband is between 8 and 14 pm wavelength. For a body at 300 K there is
considerably more emission per unit area (termed excitance) in the 8-14 pm wave-
band than the 3-5 pm waveband (the excitance is about thirty times more). For a
body at 600K, the levels of excitance in each waveband are similar but at 2000K
the 3-5 pm excitance is a factor of about nine times higher [15]. This is one of the
main reasons why the 3-5 pm operating waveband has been favored for heat-
seeking missiles designed to follow the hot air trail left by a jet engine. The other
main reason why the mid waveband seekers are currently preferred is that there
are several durable window or dome materials that are transparent in that spectral
range, such as sapphire or magnesium oxide [ 151 and that are currently available on
a commercial basis. Figure 13 shows typical heat-seeking missiles fitted with 3-5 pm
wavelength transparent domes.
In contrast, the longer wavelength infrared (LWIR) band is desirable for imaging
cooler bodies (such as the aircraft itself rather than just the hot engines, which can be
‘masked’) and although LWIR detector materials and missile seekers have been
585
Figure 12. Example of the damage caused by a high velocity water drop onto a zinc sulphide
Figure 13. Examples of air-to-air heat-seeking missiles fitted with 3 to 5 pm wavelength transparent
2.3 Optical Applications
window (from [45]).
domes.
Figure 14. Example of a CVD diamond hemispherical dome, 70 mm in diameter and over 1.O mm
thick that has been semi-processed on both surfaces.
available for some time, they have not been exploited due to the lack of erosion resis-
tant windows and domes that transmit in the 8-14pm waveband [14]. Materials
such as germanium and zinc sulphide, for instance, are transparent in the longer
wavelength IR band but even with the best available protective coatings they are
not able to withstand the rain erosion encountered during air carriage, let alone a
missile flight scenario.
Diamond has been recognized to be an ideally suited material for this application
[15,14,25] because of its good IR transmission and its extreme mechanical proper-
ties. It was only with the introduction of CVD diamond technology and its potential
for synthesizing diamond films over large areas that the IR seeker community
started to consider the practical possibility of using diamond for this application.
One approach has been the use of CVD diamond as a protective coating for
other softer IR materials such as ZnS or Ge [46]. However the successful develop-
ment of thick-film or bulk CVD diamond [2f, 131 motivated active development
projects on the synthesis of free-standing CVD diamond domes. This work has
met with considerable success [25,47] resulting, for instance, in the synthesis of
CVD diamond hemispherical dome geometries 70 mm diameter, 2 1 mm thick
that have been semi-processed on both surfaces as shown in Fig. 14. For imaging
applications the form shape needs to be very precise and the radii of the two surfaces
have been consistently machined to better than fl pm, which is technically difficult
to achieve in such a hard material.
Figure 15 shows the thermal image of a commercial aircraft through a CVD
diamond dome fitted (such as that shown in Fig. 14) to an 8-14 pm thermal imaging
camera. There is a small loss of image resolution due to imperfections in both the
surface finish (causing scattering) and the form shape of the dome being tested; how-
ever, the engines are clearly visible as being hotter than the bulk of the plane, and the
plane shape is clearly identified.
Figure 15. Thermal image of a commercial aircraft through a CVD diamond dome such as that
shown in Fig. 14. The engines are clearly visible as being hotter than the main body and the
shape of the aircraft is clearly identified.
Table 2 compares IR window properties of CVD diamond with other, 8-14 pm

wavelength transparent materials showing the exceptional optical and thermo-
mechanical properties available with CVD diamond. As in other applications, it
is the combination of several of these properties within one window material
that makes diamond so special. One example is the remarkable sand erosion resis-
tance of CVD diamond coupled to the high transparency in several spectral
regions. The latter property may be important for future military imaging systems
since there are considerable advantages in being able to ‘see’ simultaneously in
several different spectral regions through a common window. When compared
with multi-spectral ZnS (which is the only other truly, multi-spectral material
for visible, LWIR, and radar frequencies) it can be appreciated that diamond is
not only more transparent but also more resistant to sand erosion by close to a
factor of a million times! The extreme sand erosion resistant properties of diamond
are graphically illustrated in Fig. 16 which compares the reduction in transmission,
as a function of sand erosion time, for CVD diamond and sapphire (which is the
most erosion resistant of the multi-spectral, 3-5 pm waveband materials). A poten-
tial future application for a multi-spectral window would be on a tank, where both
the IR sensing and laser range finding could be performed through the same
window as used by the tank driver.
Thermal shock of the window (or dome) is an important consideration for the
design engineers of future missile systems because of the high thermal gradients gen-
erated during rapid in-flight acceleration, and is currently limiting the performance
of existing heat-seeking missiles. The combination of a high thermal conductivity
coupled with a low thermal expansion coefficient render diamond virtually
immune to thermal shock [15]. Using a thermal shock figure of merit as defined in
note 3 of Table 2 diamond is approximately 200 times better than sapphire, which is
Table 2 Property comparison between high quality CVD diamond and alternative window
materials that transmit in the long wave (8-14 pm) thermal imaging region
Property CVD Diamond ZnS ZnS (Multi- Germanium

(high optical (FLIR spectral)
quality) grade)
Vicker's hardness (GPa)

Fracture toughness (MPa.m'/')
81 18 * 2.3
0.81
1.5
1
8.5
>6 0.57
Young modulus (GPa) 1000-1 100 74.5 74.5 103
Poisson ratio 0.1 (assumed) 0.29 0.28 0.28
Tensile strength (MPa) 35&800 103 60 90
Rain impact DTV (m s-I) 525 (>1.25mm 170 165 175
2 mm drop size thick)
Sand erosion (mg kg-I) 0.18 19,000 (2) 156,750 (2) 12,382 (2)
@ 80 m s-' C25/52 sand
Sand erosion (mg kg-I) Negligible 950 7,840 1,730
@ 34 m/s C25/52 sand
Thermal conductivity 1800-2200 19 27 59
@ 300K (W mK)
Thermal conductivity 1,200 -
@ 750 K (W mK)
Thermal expansion coeff. (p.p.m./K) 0.9 6.6 6.3 6.1
at 273 K
Thermal expansion coeff. (p.p.m./K) 2 7.7 7.5
-
-
at 473 k
Thermal shock (mild) FOM 1000 2.6 2.1 6.1
( x l o ' w m - ') (3)
Refractive index at 10.6 pm 2,376 2.19 2.19 4
dn/dT (K-') (at 300 K)
*
9.6 x 41 x 10 36 x 400 x
Dielectric const. D (35 GHz) 5.68 0.15 8.35 8.39 -17
YOincrease in D at 773 K (Yo) 4.3 -8 -8 -
Loss tangent at 35 Hz <0.0002 - - -
10.6 pm abs coeff. (cm-') 0.03-0.1 0.2 0.2 0.02

Max. window temperature (K) for 1023** 573* 573% 343***
extended operation in air at 10.6 pm 873** 873**
wavelength ( I )
Useful transmission range (pm) 0.3 to 3.0 and 7.7 to 11.2 0.4 to 11.5 1.8 to 23.0
7.0 to >SO0
(1) *** Band cdge moves towards shorter wavelength,introducing significant absorption.
Mass loss due to oxidation.
***
Free-carrier absorption makes material opaque.
(2) Calculated sand erosion mass loss values at 80 m/s for ZnS and Ge (to allow a comparison
with CVD diamond) are based upon up-scaling theories (using erosion rate velocity scaling
factors) from work undertaken at the Cavendish Laboratory [45].
(3) FOM = (Material strength x (1 - Poisson's Ratio) x thermal conductivity/thermal expan-
sion coeff. x Young modulus) [IS].
generally considered to be the 3-5 pm material which is most resilient to failure by

thermal shock.
The above considerations illustrate the versatility of CVD diamond technology in
the ability to synthesize three-dimensional shapes of high quality diamond and
% of rnax. Transmission retained
90
\
CVD diamond, growth surface
70
,
40
30
I’ I I
sapphire
I I
0 500 1,000 1,500 2,000 2,500

Exposure time (s)
Figure 16. Reduction in transmission as a function of sand erosion time for a CVD diamond
window compared to a sapphire window.
explains why CVD diamond is emerging as one of the preferred material options for
applications requiring multi-spectral transmission in harsh environments.
2.3.2 Windows for High-power Infrared Lasers

Considerable progress has been achieved in recent years on improving the optical
properties of CVD diamond and in making this material available in large
sizes and with reproducible and consistent properties [4,5,35] so that its use in
actual laser systems has become a practical reality [48]. This section reviews the
technical issues relevant to the performance of windows for high power lasers
comparing the specific case of CVD diamond and ZnSe for the transmission of
C 0 2 laser beams.
A major requirement for windows used in the transmission of high power beams
is the reduction of wave-front distortion which may result in defocusing of the beam
and loss of power density. There are two major sources of wave-front distortion:
thermal effects and geometrical or form-shape aberrations [49].
2.3.2.1 Thermal Effects

S. E. Coe
A window can absorb heat from the laser beam by intrinsic bulk absorption or at the
surfaces (through intrinsic surface absorption mechanisms or at the anti-reflection
(AR) coatings). The absorbed heat induces radial and axial temperature gradients
which in turn result in two distinct forms of optical distortion:
Table 3. Comparison of properties between CVD diamond and ZnSe
Property at 300 K Diamond (optical-grade DIAFILM) ZnSe
Thermal conductivity (W mK-') 1800-2100 16

Thermal expansion (p.p.m. K-') 0.9 1.6
Thermo-refraction (dn/dT) ( T - ' ) 1x 5.8 x 1 0 - ~
Fracture strength (MPa) 400-600 55
Weibull modulus 11-23 5
Young modulus (GPa) > 1050 70
Typical thickness (mm) used for a 0.7-1.0 6
25 mm diameter window
Bulk absorption coefficient (cm-') 0.1-0.03 -0.0005
Total absorption (AR/AR coated, 0. 9 0.23
0.1 % absorption per surface) (YO)
(i) The thermo-refractive effect is the result of changes in refractive index with
temperature and is described by the thermo-refractive coefficient dnjdT,
where n is the refractive index of the window material.
(ii) Thermo-elastic effects which are a caused by bulging (or other form-shape dis-
tortions) or by changes in the optical constants by thermal stress gradients
through the photo-elastic effect.
Materials which have relatively small thermal gradients and do not change geometry
with temperature will show the lowest values of thermal distortion. Critical param-
eters of relevance in these effects are thermal conductivity, thermal expansion
coefficient, thermo-refractive constant, bulk absorption coefficient, absorption at
the AR coatings and the photo-elastic constants. The total wave-front distortion
can be described by the variation in phase (or optical-path-length) of a transmitted
beam with radial position. For the thermo-refractive effect, for instance, the phase
change will be (dn/dT) x t x A T (where t is the thickness of the window and AT is
the temperature change) and is thus smaller for a thinner window. The minimum
acceptable thickness of a window will be determined largely by the ability to support
vacuum or pressure without breaking or distorting [50]. Critical parameters in these
cases are the fracture strength and the Young modulus respectively.
The current standard material for optical elements in C 0 2 lasers is ZnSe [27]
because of its very low intrinsic absorption at 10.6 pm [14, 15,341. Table 3 compares
the critical parameters of CVD diamond and ZnSe showing that CVD diamond, for
the reasons discussed above, has the potential to handle much greater beam powers.
This has been long recognized [51] and the thermal effects in diamond laser windows
have been theoretically modelled in previous work [27,51-531. Some of the earlier
results [5 1-53] were derived before reliable data were available on the properties
of polycrystalline diamond and were therefore very speculative. The following is
an up-date of the predicted thermal effects in optical grade CVD diamond and
ZnSe windows where some of the earlier calculations arc revisited.
The analysis presented here assumes that bulging and thermo-elastic effects in
either diamond and ZnSe can be neglected [27,52-541. The assumed configuration
for each case is that of a circular window, 25mm in diameter, which is edge
2.3 Opticul Applications 59 1
Figure 17. Schematic of the axially symmetrical temperature gradients generated in an edge-cooled
window fitted with AR coatings when traversed by a laser beam.
cooled at a fixed temperature resulting in an axially symmetrical temperature profile

as shown in Fig. 17. An AR coating with a total absorption at 10.6 pm of 0.1 % per
surface is assumed for both windows. In the case of ZnSe, the absorption at the AR
coating dominates, leading to the highest temperature being at the surface.
Because of the difference in mechanical strength and Young modulus (see
Table 3), it is expected that the thickness of CVD diamond windows could be
made thinner than those of ZnSe. Using the criterion that the thickness should be
sufficient to prevent fracture, the thickness-to-diameter ratio must be [35,50,55]:
t/D > 0,5(kjP/S)'/* (1)

where:
k constant determined by the boundary conditions and dependent on
=
the Poisson's ratio
t = sample thickness
D = sample diameter
P = pressure differential
f = safety factor (depends on the value of the Weibull modulus)
S = fracture strength.
Using the criterion of acceptable deformation imposes the condition [50,52]:
t/D > l.OI{(n - 1)(P/E)2(D/h)}'/2 (2)

where E is the Young modulus.
Using Eq (l), the ratio of the CVD diamond-to-ZnSe thickness (for the same
diameter) is:
/
tCVDD tZnSe = { (kCVDafCVDDSZnSe) / (kZnSefZnSeSCVDD) 1112 (3)
Using the parameters listed in Table 3, assuming that the k parameter is similar in
both materials and using the same safety factor for ZnSe and CVD diamond results
in tCVDD/tZnSe z 1/3. Equations (1) and (2) predict a minimum t / D ratio of 0.017 for
CVD diamond (using a safety factor of 4 and a fracture strength value of 400 MPa)
or a thickness above 0.4mm for a 25 mm diameter window. In practice, it is found
that a better flatness and overall shape stability can be achieved if the thickness is
close to 1.Omm. For these calculations it will be assumed that the CVD diamond
window is 1.O mm thick and the ZnSe 6.0 mm, which is the typical thickness used
for ZnSe windows in the industry [27]. The higher thickness used for ZnSe may
be partly justified by the need to use a higher safety factor due to the lower value
of its Weibull modulus [35].
The absolute bulk absorption of optical grade CVD diamond windows has been
found to be in the range 0.1--0.03cm-' [27].A value of 0.05 cm-' has been assumed
for this calculation.
The analysis has been done for a continuous wave, 5 kW incident beam that is
assumed to have a Gaussian profile with a 14mm diameter at 14% (lie2) of maxi-
mum intensity. For this level of power the temperature increases are relatively low
and the temperature dependence of parameters such as dn/dT [54] have been
ignored, but the temperature dependence of the thermal conductivity for optical
grade CVD diamond (Fig. ll), although not very pronounced, has been taken
into account.
The analysis solves the fundamental heat propagation equations using a two-
dimensional computational model in which values of the radial and axial variations
of the temperature and the refractive index are calculated. The calculated tem-
perature gradients for the two cases are shown in Fig. 18. Figure 19 shows the
radial dependence of the average (across the thickness) temperature. The tempera-
ture in CVD diamond shows a maximum variation (center to edge) of 3.4"C
compared to 23.4"C for the ZnSe case. The temperature variations over the effective
diameter of the beam are 1.7"C and 12.4"C for the CVD diamond and ZnSe
windows respectively.
The phase or optical path length variation A@ = (dn x t ) is a function of radial
position in the ZnSe and CVD diamond windows and is shown in Fig. 20. The
variation for CVD diamond over the effective beam width is over a factor of 250
smaller. Three factors combine to yield this large difference: the smaller temperature
rise in CVD diamond due to the higher thermal conductivity ( A T a factor of more
than 7 smaller in CVD diamond), the smaller value of dn/dt for diamond (a factor of
5.8), and the thickness of the window, which for CVD diamond is only 1/6th of that
in ZnSe. For an example similar to that shown in Fig. 18 the thermal lensing effect in
the ZnSe window was found to be equivalent to a lens of 3-5 m focal length, whereas
for the CVD diamond window the effect was negligible [27,34].
As a general rule it can be assumed that the lensing effect can be neglected if the
difference in A@is smaller than hjl0 [51],or 1 pm for a C 0 2 laser beam. As shown in
Figure 18. Calculated temperature gradients in a CVD diamond and a ZnSe windows, edge cooled
and with AR coatings.
Fig. 20, A@is approximately 8 pm for ZnSe but under 0.04 pm for CVD diamond.
Recent results have been reported using electronic speckle pattern interferometry
(ESPI) [56] using a C 0 2 laser of power up to 1 kW on a CVD diamond and a
ZnSe windows of 25mm aperture and AR coated on both sides. The ZnSe
window starts to deform at moderately low power levels (50W). In the CVD
diamond window discernible distortions start at powers over 700 W. These
0 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3
Radial Position (mm)
Figure 19. Radial dependence of the average (across the thickness) temperature for the CVD
diamond and the ZnSe windows derived from the calculations shown in Fig. 18.
10
8 --...._
-...--.
.-
a
--......
h
fn *%.
C
$ 6 *.' ..%.
.-
E
Y
*%.
CI
x 4 -*...
rn
C
2 -*...+.-
5.
-...
-*..
....*-.
I .%
0
measurements predict a deformation in the ZnSe window for a beam power of 5 kW

of approximately 5pm, in rough agreement with our previous calculation. The
equivalent deformation measured in that experiment for a 1 kW beam was at least
one order of magnitude lower in the CVD diamond window than in ZnSe.
Using parameters similar to the ones used in the above modeling, Wild [54] esti-
mated that a CVD diamond window will introduce an optical distortion of 1 pm for
a laser beam of at least 160 kW and at that level of incident power, the generated
thermal stress in the window should not exceed 100 MPa, a factor of at least four
below the fracture strength of optical grade CVD diamond < l m m thick
[25,35,54]. From this perspective it could be concluded that the power handling
capability of an edge cooled CVD diamond window l.Omm thick is of at least
160 kW before serious thermal degradation is observed.
However, this may be an oversimplification because it ignores possible laser
damage effects. There is evidence [57] that for CW beams:
(i) the laser induced damage threshold (LIDT) is a decreasing function of total
power, in fact the power damage threshold scales with diameter rather than
with area, and
(ii) LIDT can be considerably reduced due to surface imperfections such as pits or
digs.
The first is a heating effect and it can be expected that, because of its high thermal
conductivity, diamond will outperform most other materials. For ZnSe, LIDT
values of 3 kW mm-' have been measured [54]. Since temperature excursions in
diamond are at least a factor of seven lower than in ZnSe (Fig. 19) a thermal
LIDT of at least 20 kW mm-' can be expected for CVD diamond (probably much
larger because of the higher mechanical and chemical integrity of diamond compared
Figure 21. lnterferometric surface profile of a 25mm CVD diamond window showing a flatness
better than one fringe (633 nm illumination).
with ZnSe). For a 160 kW beam, this implies a minimum beam diameter of 8 mm,
considerably smaller than actual beam sizes in lasers of this power level. What has
not yet been investigated is the potential decrease in LIDT that could be introduced
by surface and bulk features typical of polycrystalline CVD diamond [2,35]. For
pulsed lasers, considerably higher LIDT values have been reported [58]. Clearly,
more experimental work is required before the true power handling capability of
CVD diamond is truly understood.
2.3.2.2 Geometrical Distortions: Window Flatness

C. S. J. Pickles
Beam distortion can be generated when the form shape of the surfaces of a window
deviate from perfect flatness. The flatness of optical elements as measured by a inter-
ferometer is expressed as the number of fringes (twice the wavelength of the light
beam probe) shown in the interference pattern of the optical element placed against
a perfect flat surface [59]. Depending on the application, an acceptable number of
fringes is specified. Applications in laser optics often require a flatness of < 4 fringes
and sometimes < 1 fringe using visible illumination (typically 633 nm wavelength).
Figure 21 shows the interferometer pattern of the surface of a CVD diamond
Window which has been polished to a flatness better than 1 fringe (633 nm illumina-
tion). This is typical of what can be achieved presently for CVD diamond showing
that current technology is sufficient to address applications such as exit windows,
where a flatness of 1 to 3 fringes is considered adequate. Better results have been
achieved using techniques currently under development as shown in Fig. 22 where
a flatness of 0.5 fringe has been achieved in a window 48mm in diameter. This
Figure 22. lnterferometric surfxe profile of a CVD diamond window showing that by the use of an
improved process, a flatness of 0.5 fringe can be obtained over a 48 mm diameter.
suggests that CVD diamond will be able to address a wider range of applications as
an optical element in laser systems.
The ability to process CVD diamond to controlled form shapes has also allowed
the production of samples with wedge angles up to 1 degree, as required to eliminate
interference fringes for broad band IR transmission, as is the case in synchrotron
beam lines [4,33].
Figure 23. The Rofin Sinar ‘Diffusion cooled’ DC-series COz slab laser with output power up to
3500 W [60].
2.4 Windowsf o r High Power Gyrotron Tubes 597
,Bending mirror
Figure 24. Schematic diagram showing the location of the CVD diamond window in the ‘Diffusion
cooled’ Rofin Sinar DC-series COz slab laser [60].
2.3.2.3 Scattering
Scattering in the infrared has been discussed in 2.2.2 and by Union et al. [34] con-
cluding that, at ~ 1 pm,
0 a good quality optical grade CVD diamond window does
not show a level of scattering that is likely to affect its infrared imaging properties
and is capable of transmitting a diffraction limited image. For systems operating at
wavelengths in the UV-visible spectral range, scattering effects may be more pro-
nounced, although recent measurements indicate that a good quality CVD diamond
window could exhibit values of scattering at 633 nm similar to that of a single crystal
Type IIa specimen [54].
CVD diamond windows are now being used in commercial high power COz
lasers. An example is the Rofin Sinar Diffusion-cooled C 0 2 slab laser with an
output power of 3500 W [48] shown in Fig. 23. In this laser, the CVD diamond
window is used to seal the slab cavity, as shown schematically in Fig. 24.
2.4 Windows for High Power Gyrotron Tubes

Measurements of dielectric loss in high quality CVD diamond at millimetre wave
microwave frequencies, reported in 1993 [2], furnished for the first time clear
evidence that the loss tangent of CVD diamond could be comparable or lower than
conventional dielectrics such as sapphire or boron nitride. This initial data and sub-
sequent first dedicated dielectric property studies [61] intensified the considerable
amount of interest that had already existed in the nuclear fusion community for
CVD diamond as a high power window material [62] especially for the development
598 2 Applications of’Diamond Synthesized by Chemicul Vupor Deposition
of Gyrotron tubes [63] for the first ‘International Thermonuclear Experimental

Reactor’ (ITER). This section will discuss why, once other advantages of diamond
are taken into account (high thermal conductivity, radiation hardness, the relative
insensitivity of the dielectric properties with temperature), CVD diamond is being
considered not only as an alternative option but probably as the only practical
solution for high power Gyrotron tube and Torus windows in ITER.
2.4.1 Window Requirements

It is believed that electron cyclotron heating (ECH) at frequencies in the range 7&
170 GHz is one of the major candidates for heating, current drive, and start-up of
plasmas in thermonuclear reactors such as the ITER Tokamak or the Stellarator
W7-X [64]. The preferred option to supply this source of microwave power is Gyro-
tron tubes [63] which are currently being developed with an output CW power of at
least 1 MW [65]. Figure 25 shows an example of a 1.3 MW, 140 GHz Gyrotron tube
being developed at the Forschungzentrum Karisruhe (FZK) [66]. A key component
in these tubes is the exit window which should be able to carry high output powers.
Figure 25. Example of a 1.3 MW, 140 GHz Gyrotron tube built at the Forschungzentrum Karlsruhe
WI.
Additionally, the barrier windows in the Plasma Vacuum Torus should be able to
withstand high doses of radiation without degradation.
The critical parameters for high-power windows are the dielectric characteristics
of the windon, material;;he dielectric loss-factor tan 6 and the permittivity E:. (or the
refractive index 11 = E:.' ' - ) because they affect power absorption and reflection. The
power absorption coefficient 'x is related to tan 6 by [43]
'x = 2x11 tan ijj h
where h is the wavelength of the microwave radiation in free space.
To minimize power reflection at a given frequency the thickness t of the window
needs to be a11 integer number of half dielectric wavelengths [64]
t = Nhj21l
where N is an integer. This requirsment can be relaxed if a Brewster angle window is
used. but the penalty is that a much larger diameter window is necessary [67].
The requirement of CW operation puts extremely high demands on the material
properties of the dielectric vacuum barrier window that serves as both the primary
tritium containment boundary and as the output window of the Gyrotron. The
window options available to Gyrotron manufacturers have been discussed in
detail by Th~iiiiiiiet ril. [64]. From this analysis it is appreciated that all other options
apart from diamond require sophisticated and, in some cases, impractical cooling
system involb ing, for instance. cryogenically edge-cooled windows, liquid-surface-
cooled double-disc windows or distributed windows i n which segments of dielectric
are metal bonded to microchannel cooled metal ribs. Another shortcoming of some
of the materials considered. such as sapphire or gold doped silicon, is the
pronounced temperature dependence of the loss factor and refractive index,
which may lead to thermal runaway and lor detuning of the window resulting in
high levels of reflected power.
This background explains the keen interest in diamond as an alternative material.
Theoretical thcrmal modeling results such as that illustrated in Fig. 26 [66] show that
a simple edge cooled diamond window approxiinately 100 mm in diameter and with
a value of loss tangent below 1 O-'. should be able to withstand powers in excess of
1 M W with a temperature rise at the centre of the window below 240 C, which is far
below the temperature (600-C)at which diamond starts to degrade by oxidation,
although thermal stress may be a limiting factor at these temperatures. This
relatively low temperature excursion is a direct consequence of the high thermal
conductivity of diamond. Table 4 compares properties of diamond relevant to
this application with those of other competitive materials.
2.4.2 The Development of Chemical Vapor Deposited Diamond

Gyrotron Windows
J . R. Brandon
The detailed requirements for the use of diamond as a window material in high
power Gyrotron tubes can be listed as follows.
600 2 Applications of’ Diamond Synthesized by Chemical Vapor Deposition
Frequsncy= 140 GHz

Powar=I,ZMW
l h i i s s = 1.748 mm
Waveguide radius = 40 mm
C m l i = 293 K: 12000 W,/ Boo
580
5M)
510
520
sw
480
460
440
420
4lm
380
360
340
320
3w
0.00
Figure 26. Results from a theoretical thermal model showing the predicted temperature rise at the
center of an edge cooled diamond window for a 1 MW incident microwave beam. Results are shown
for different values of dielectric loss and for two different cooling diameters assuming an exposed
aperture of 100mm diameter [66].
A reproducible value of loss tangent below lop4 (from modeling results as in

Fig. 26) achieved in discs of at least l00mm in diameter and up to 2.2mm
thick with reasonable radial and axial uniformity. The thickness requirement
is dictated by the need to minimize power reflection as discussed above and
by the need to support vacuum with a reasonable safety factor.
The variation with temperature of the dielectric constants (tan 6 and n ) needs to
be very small in the temperature range of interest (room temperature to
approximately 200°C) to minimize the risk of thermal runaway or the de-
tuning of the window for minimum power reflection.
The diamond windows need to be bonded to vacuum-tight and high-tempera-
ture bakeable metal flanges for mounting to the Gyrotron tube and reactor
ports.
The dielectric properties need to be relatively insensitive to radiation damage.
This requirement is critical for the tritium barrier window that will be exposed
to a radiation field of both neutrons and gamma radiation [68].
following summarizes some of the Drotzress achieved to date partly as part of

a collaboration between De Beers IndistrTal Diamonds and t h e FZK-ITP, that
show convincingly that CVD diamond is indeed a very attractive option for this
application.
2.4 Window~sjorHigh Power Gyrotron Tubes 60 1
Table 4. Properties of candidate Gyrotron window materials at room temperature.
Property Diamond Sapphire Silicon BN CVD

CVD P.C. S.C.
- Au-doped S.C. - p.c.
Thermal conductivity (WmK-’) > 1800 40 150 50

Fracture strength (MPa) 400-600 410 3000* (170) 80
Density (gcm-3 3.5 4 2.3 2.3
Specific heat capacity (J gK-’) 0.5 0.8 0.7 0.8
Young modulus (GPa) >I050 385 190 70
Thermal expansion coefficient K-’) 1.1 5.5 2.5 3
Permittivity (at 145GHz) 5.67 9.4 11.7 4.7
Dielectric loss (at 145GHz) tan 6 0.8-10 20 0.35 1 I5
Temperature dependence of dielectric small steep steep
properties
P.c.: polycrystalline, s.c.: single crystal

* The fracture strength of conventional single crystal silicon is 170 MPa [36]. The value of 3000 MPa
has been reported after a special surface treatment [76].
2.4.2.1 Dielectric Loss

Table 5 shows the dielectric loss values that have been achieved for different types of
CVD diamond windows. The four samples DB 1 to DB 4, of diameters between 30
and 40mm, were the result of a development effort to achieve the lowest value of
dielectric loss in specimens of the required thickness. The results achieved demon-
strate that values of tan F below lop5 are possible in specimens of thickness in
excess of 1.7mm. These values of loss are close to the sensitivity of the measuring
technique for specimens in this thickness range. When the four samples were stacked
together, the more accurate loss value measured was tan 6 = l ( f 0 . 3 ) x lop5 [69].
Samples DB 5 , DB 6, DB 7 and DB 8 are full size (96120mm diameter) windows,
synthesized under different conditions.
Sample DB 5 (named ‘Super FZK’ elsewhere [70]) is l00mm diameter and
demonstrates that the low values of dielectric loss measured in small diameter
Table 5. Dielectric loss of CVD diamond window samples.

~
Sample Diameter Thickness Dielectric loss

(mm) (mm) tan 6 ( m 5 )
(at 145GHz)
DB- I 50 1.8 0.8

DB-2 30 1.79 0.8
DB-3 30 1.74 0.8
DB-4 30 1.72 0.8
DB - 1 to 4 (stacked) 7.06 1.o
DB-5 I00 1.6 0.6 (10.2)
‘Super FZK’ [70] (70 to 370 K)
DB-6 120 2.2 2
‘Star of FZK’ [70] (70 to 370K)
DB-7 100 1.85 2
DB-8 96 2.23 13-15
Figure 27. Spatial distribution of the dielectric loss in a 100mm diameter CVD diamond window
1.6mm thick measured with a spatial resolution of 6mm. The isolated points of higher loss at the
edge may be caused by scattering losses. Over most of the area the measured dielectric loss is equal
or below tan 6 = which is close to the sensitivity of the measuring technique. This may account
for some of the measured isolated spurious higher values of loss.
specimens (DB 1 to 4) can also be achieved in full size windows. The spatial unifor-
mity of the dielectric loss in this window is shown in Fig. 27 which shows a map of
tan 6 measured with a beam diameter of 6 mm [69]. It can be seen that over most of
the active aperture of the window (a central area of approximately 80mm diameter)
tan 6 is below lop5 and close to the sensitivity of the measuring technique, which
may account for spurious isolated higher values of loss. The temperature depen-
dence of this window over the 70 to 370K temperature range indicates that the
loss is largely independent of temperature over this range and that the average
tan 6 exhibits a value of 0.6 (* 0.2) [70], which is the lowest loss tangent so
far reported for a CVD diamond window.
Sample DB 6 is probably the largest Gyrotron window ever made (named 'Star
of FZK') and exhibits a loss of 2 x over the same temperature range as
DB 5 [70].
The window DB 8, which has a relatively high loss tangent of over 13 x lop5,was
the first CVD diamond window to be mounted in a high power Gyrotron tube
(shown in Fig. 28) at the Japan Atomic Energy Research Institute (JAERI) [71-
731. This window was tested in 1998 using the 170 GHz JAERI/Toshiba Gyrotron
at output powers of 0.5 MW for 8 s with a water cooled edge showing a temperature
increase a t the center of not more than 130°C,consistent with theoretical modelling
f o r High Power G y o t r o n Tubes
2.4 Windo~vs 603
Figure 28. The first high-power Gyrotron (JAERI/Toshiba, 170 GHz) fitted with a CVD diamond
window. Measurements using beam powers up to 1 MW for 10 seconds have already been
performed.
[73]. The second window for the JAERI Gyrotron (DB7) has a lower loss tangent of
approximately 2 x lo-’. This window was tested at the end of 1999 using also the
170GHz Gyrotron for powers up to 1 MW for 10s. The temperature increase at
the center was less than 3 0 T , also in very good agreement with theoretical model-
ling [73]. Because diamond reaches almost steady state for times in excess of 3 s the
above experiments show convincingly that CVD diamond windows are able to
transmit millimeter wave powers of at least 1 MW continuous wave. At the
moment of going to press, in excess of 20 full size CVD diamond windows have
been delivered to members of the international nuclear fusion community. These
exhibit routinely values of loss tangent of 2-3 x lo-’.
It is not clear yet if the low values of dielectric loss of samples DB 1 to DB 7
represent a fundamental limit of a loss mechanism in diamond or if they are still
affected by impurities or defects in the polycrystalline CVD diamond structure.
From measurements in previous samples [74], there is clear evidence that, in
samples of relatively low quality (tan 6 greater than 70 x darker regions of
the sample have higher values of loss. Even in the highest quality polycrystalline
CVD diamond, such as that of the sample shown in Fig. 27, it is possible to observe
localized features related to the grain structure of the layer. It can be speculated that
these features may be a contributing factor to the dielectric loss of polycrystalline
CVD diamond. However, from a practical point of view, this is probably not of
great concern in the foreseeable future since it has been estimated that diamond
windows with values of tan 6 below 2 x lo-’ (a factor of two above that of the
window shown in Fig. 27) should have a power handling capability of at least
2.5 MW [64].
604 2 Applications of' Diamond Synthesized by Chemical Vapor Deposition
10
5
7
<
0
s 2
*
E
$ 1
C
m
I- 0.5
v)
v)
0
2
A 0.2
Q,
ii 0.1
0.05
1 100 150 200 250 300 350 1 0
TtKl
Figure 29. Temperature dependence of tan 6 for a CVD diamond sample measured at 145GHz
compared with sapphire [75].
2.4.2.2 Temperature Dependence of Dielectric Constants

Preliminary measurements of the temperature dependence of the dielectric proper-
ties of polycrystalline CVD diamond are available. Figures 29 and 30 show the tem-
perature dependence of tan 6 and EL measured at 145GHz for a CVD diamond
sample compared to sapphire [75].In this temperature range (100-370 K), the dielec-
2.4 Windowsf o r High Powcu Gyrotron Tubes 605
Figure 31. Examples of CVD diamond windows mounted onto vacuum flanges. The two at either
end show 100 mm diameter windows mounted to conventional CONFLAT vacuum flanges. The one
in the middle illustrates a double flange configuration designed specifically for Gyrotron tube assem-
blies. All these vacuum assemblies have been tested to be vacuum tight to better than 10-9 mbar I s C '
after thermal cycling up to 450°C.
tric constants of diamond are largely insensitive to temperature, in contrast to sap-

phire. In 1999 the FZK team extended the measurements of dielectric loss to tem-
peratures up to 700K [75] showing that in a high quality CVD diamond window
the loss tangent at 700K is not more than 5 x lop5 from a value at 300K of
3 x lop5. The results described in 2.4.2.1 indicate that even with transmitted
powers of up to 1 MW the temperature increase is likely to be considerably less
than 100°C and therefore there is no major risk of thermal runaway.
2.4.2.3 Effects of Radiation

Measurements of the effect of radiation on the dielectric properties of CVD
diamond have been reported usin neutron irradiation experiments with fast
neutron fluences up to at least 10" n m p 2 (energy >0.1 MeV) [65]. Differences
before and after irradiation were found to be more pronounced at lower frequencies.
At 145 GHz specimens that started with values of tan 6 of 2 x lop5maintained these
levels (or even showed a decrease in loss). Further experiments to extend radiation
fluencies to lo2*n mp2 are in progress [68].
2.4.2.4 Mounting Techniques

A metal-to-diamond bonding technique has been developed to attach CVD dia-
mond windows to vacuum flanges or other assemblies. This is required to mount
the windows to the Gyrotron tubes or reactor ports via suitable waveguide sections.
Figure 31 shows examples of three l00mm diameter CVD diamond windows
mounted to two different types of flanges. The two at either end of the picture are
Figure 32. Window disc with slots to reduce hoop stresses and to facilitate bonding alignment
test mechanical grade specimens mounted to conventional CONFLAT vacuum

flanges. The one in the middle illustrates a double flange configuration designed
specifically for mounting on to a Gyrotron tube allowing for water cooling of an
exposed section of the edge of the window. This bonding technique is based on
an aluminum alloy and has been demonstrated to withstand thermal cycling at
temperatures up to 450°C without degradation of the vacuum integrity (measured
at better than lop9 mbar 1 spl [64]. Figure 32 shows a new development in which
slots are cut at the edge of the window to reduce hoop stresses and to facilitate
bonding alignment.
The results described above indicate that CVD diamond windows are very
close to meeting all the requirements listed at the beginning of this section, and
promise to be the preferred option for high power Gyrotron windows for ITER
and other experiments.
2.5 Thermal Management of Laser Diode Arrays

Natural Type IIa single crystal diamond heatsinks have been used for the
thermal management of microwave and laser diode devices for well over a
decade [17, 181. The restriction in the size of available natural diamond has, how-
ever, limited the use of diamond to the heat management of small, discrete devices
such as single junction laser diodes. The commercial availability of large CVD
diamond plates has opened a host of new possible applications in which diamond
can be used in the heat management of larger electronic and optoelectronic
devices [ 1,77,78].
2.5.1 Laser Diode Arrays: General Issues

A specific example is the use of CVD diamond to improve the heat dissipation in
laser diode arrays (LDA) [78,79]. In these devices a set of laser diode junctions
are monolithically integrated on a single chip. A typical LDA bar is 0.1 mm in
height, the resonator length of each junction between 0.3 and 2mm, the junction
width between 0.1 to over 0.2 mm and the total length of the array between 5 and
15mm to accommodate 5-75 single emitters side by side along the length of the
bar [80]. During operation of a LDA, more than half of the electrical input power
is dissipated in heat which needs to be efficiently removed in order to reduce the
junction temperature and thus prolong the life of the device. Typical operating junc-
tion temperatures are 50-55°C and raising this temperature by only 10°C results in a
reduction by a factor of two in lifetime [79].
The thermal resistance of a diode laser bar depends on the area size of the active
junctions. Mounting the diode laser bar on a high thermal conductivity material
serves to increase the effective area of the junction before contacting the water
cooled, copper heat-sink. These submounts are therefore heat-spreaders, rather
than heat-sinks [l]. A typical LDA requires a heat-spreader submount over
10 mm in length, which is well within the current capability of CVD diamond tech-
nology but would have been almost impossible to implement by the use of natural
diamond. The other advantage of using electrically insulating materials, such as
CVD diamond, is the possibility to address electrically single emitters of an array
individually for applications such as marking and printing [80]. The efficiency of a
heat-spreader will be a function of its intrinsic thermal conductivity and of its actual
dimensions and is measured by the decrease in the total thermal resistance of the
package, equivalent to the total reduction in temperature of the device junction
for a given dissipated power.
There are a number of issues that need to be addressed before CVD diamond can
be used effectively as a heat-spreader submount.
(i) Optimization of submount geometry and thermal conductivity.
(ii) The flatness of the submount, to ensure uniform thermal contact.
(iii) Metallization issues such as thickness and thermal stability.
(iv) Thermal stress and expansion mismatch.
2.5.2 Modelling of Submount Heat Resistance

S. E. Coe
Numerical simulations of the junction temperature increase of a laser diode array

have been performed for a package configuration as illustrated in Fig. 33. Calcu-
lations have been performed with and without a CVD diamond heat-spreader.
The decrease in thermal resistance has been calculated for different values of the
heat-spreader thickness, depth and thermal conductivity. The simulations have
been performed for an array of laser diodes which are individually 200pm wide
Figure 33. Package configuration analysed in the work consisting of the laser diode array mounted
on a CVD diamond spreader which is mounted on a copper heat-sink block.
and separated by a spacing of 200pm. The length of the heat-spreader has been
assumed to be constant at 11mm. This is a three-dimensional calculation which
simultaneously takes into account heat dissipation parallel and perpendicular
to the direction of the LDA bar, in contrast with previous two-dimensional
calculations reported in the literature [79].
Figures 34 and 35 show that the thermal resistance decreases as a function of heat-
spreader thickness and depth. Heat dissipation through the rear (or depth) of the heat
spreader is important, because in an array of diodes there is thermal cross-talk
between junctions in which the temperature of one junction is affected by that of
adjacent junctions. Figure 34 shows that heat-spreader depths between 2 and 3 mm
are required for a thickness of 300pm. Also, because of thermal cross-talk, the
0.2'
0
I
100
I
200
1
300
I
400
I
500
I
600 700
Thickness of CVDD layer (microns)
Figure 34. Thermal resistance decrease as a function of the thickness of the CVD diamond heat
spreader for the configuration shown in Fig. 33.
8c 0.32 ~
.-4
3 0.28
-m
'
p?
0.24
c
I-
0.2
0
' 0.5 1 1.5 2 2.5 3 3.5 4
Length of CVDD (mm)
Figure 35. Thermal resistance decrease as a function of the depth of the CVD diamond heat sprea-
der for the configuration shown in Fig. 33.
influence of heat spreader thickness is not as pronounced as would have been expected
in a single diode junction and very little gain is achieved for a thickness above 300 pm,
as shown in Fig. 35.
Figure 36 shows the decrease in thermal resistance as a function of the thermal
conductivity of the spreader, showing that, for a total dissipated power of 90 W
(or 3.6 W per junction), a reduction close to 5°C is obtained by increasing the
thermal conductivity from 1000 W m-' K-' to values close to 2000 W m-' K-I.
This reduction in temperature is significant in view of the increase in life of a
factor of two expected for a decrease on lO"C, as mentioned above. As discussed
in 2.2.5, values of thermal conductivity of 1000W m-' K-' are typical of the
lower thermal grade whereas the higher grades exhibit a thermal conductivity
close to 2000 W m-' K-l (Fig. 33). The result of Fig. 36 shows, therefore, that for
this application it is advantageous to use a higher thermal conductivity CVD dia-
mond grade.
18'
500 1000 1500 2( 10
Themal Conductivity (WlmK)
Figure 36. Decrease of the thermal resistance as a function of the thermal conductivity of the CVD
diamond heat spreader.
These results indicate that an ideal CVD diamond heat-spreader would reduce the
thermal resistance of a 25 element array by approximately 36% from 0.36 K W-' to
0.23 K W-I. Experimental results [79] show that, in a similar array, the measured
reduction in thermal resistance is only 22%. This may be a consequence of the
extra thermal resistance introduced by the metallization. If a metallization thermal
resistance value of 0.12 K W-' is added, the results from the above calculations
agree well with experimental results. Using a novel microchannel heat-spreader
made of CVD diamond reductions in thermal resistance up to 75% have been
predicted [81].
2.5.3 Flatness of Submount

The analysis discussed above is an ideal case in which the effects of intermetalic
layers is ignored. If the heat spreader is not very flat there will be differences in
metal layer thickness between different sections of the array, leading to a non-
uniform temperature distribution. This effect is thought to account for differences
in thermal resistance observed between different heat-spreaders [79]. It is estimated
[80] that the required flatness should be better than 1 pm cm-'.
As discussed in 2.3.2.2, it is possible to process CVD diamond plates to accuracy
close to one visible fringe over 25 mm. This is equivalent to a deviation from flatness
of less than 0.13 pm cm-', well within the requirements for an efficient LDA heat
spreader.
2.5.4 Thermal Stress

Due to the substantial difference between the thermal expansion coefficient of CVD
diamond (1 p.p.m. K-') and laser diode material (6p.p.m. K-' for GaAs), thermal
Figure 37. Example (courtesy of Jenoptik [83]) of a laser diode array bar which has been pre-cracked
to reduce effects of thermal stress.
2.6 Cutting Tools. Dressers and Wear Parts 61 1
stress can seriously affect the reliability and life of these devices [79,80,82]. A novel
approach has been proposed called stress induced dicing (SID) where the tensile
stress in the laser diode bar after soldering and during cooling down is relaxed by
cracking the bar at defined lines between emitters [83]. Figure 37 shows an example
(courtesy of Jenoptik) of such a laser diode bar mounted on a pre-metallized CVD
diamond spreader using a Au:Sn (80:20) solder alloy.
2.6 Cutting Tools, Dressers and Wear Parts

J. L. Collins, M. W. Cook and P. K. Sen
2.6.1 Cutting Tools Trends

The use of nonferrous materials and composites in modern industry, particularly
automotive, is substantial and becoming increasingly important as the demand
for stronger, lighter, more wear resistant components increases. Highly abrasive,
high performance materials, such as fibre reinforced plastics, high SiAl alloys,
metal matrix composites (MMCs) and even abrasive wood composites are steadily
increasing in demand
Also, there is significant development within the machining industry to increase
machining speed and eliminate or reduce the amounts of cutting fluid used for
environmental and cost reasons. Components are being manufactured to continu-
ally higher levels of precision and consistency.
These trends in workpiece material, manufacturing philosophy, and, product
quality have led to a strong demand for polycrystalline diamond (pcD) when
machining nonferrous materials, and polycrystalline cubic boron nitride (pcBN)
when machining ferrous materials. These materials are increasingly providing the
best economic solution to high volume, precision manufacturing problems. More
recently, CVD diamond has become available and is currently being evaluated
and developed in cutting tool type applications. CVD diamond cutting tool
materials fall into two categories a thin film conformal coating less than 30 pm
~
in thickness or a thick film solid diamond usually around 0.5mm in thickness.

The application of thin film CVD diamond is highly substrate sensitive. The adhe-
sion to tungsten carbide, particularly three dimensional tools like taps and drills,
remains a persistent technical obstacle to overcome. CVD thick film can be used
free-standing, or has to be brazed directly to a tool body or a substrate that is in
turn brazed again to a tool body.
The strong acceptance of pcD in machining non-ferrous materials is based on its
properties being highly suited to the requirements of modern day manufacturing.
CVD diamond has similar properties and it is anticipated that it will gradually
become integrated into diamond based cutting tool material selection and
application.
6 12 2 Applications of Diamond Synthesized by Chemical Vapor Deposition
Figure 38. Schematic diagram showing the relative merits of various cutting tool materials for the
machining of different materials. It can be seen that CVD diamond occupies a specific niche in the
machining of materials requiring exceptional wear resistance.
2.6.1.1 Polycrystalline and Chemical Vapor Deposited Diamond

An indication of the relative position of various cutting tool materials with respect
to their wear resistance and fracture toughness is shown in Fig. 38. Diamond is by
far the most wear resistant material, particularly in pure single crystal form, but
because single crystal diamond is a brittle material, it does not have the average
toughness of the polycrystalline forms both pcD (cobalt-containing) and CVD
~
diamond (which contains no metal phases).

CVD thick film diamond complements pcD throughout a wide range of general
machining applications. Its main advantage over pcD is its greater thermochemical
stability. Potential disadvantages have been its relative brittleness, and its lack of
electrical conductivity. But recent advances have resulted in the development of
an electrically conductive grade of CVD diamond material, demonstrating that
the use of spark erosion cutting and processing techniques (as are used extensively
in the diamond tool making industry, and in particular for woodworking tools)
could be also adopted for CVD diamond tools.
2.6.2 Cutting Tool Application of Chemical Vapor Deposited

Diamond
The following case studies are typical examples of where CVD diamond’s perfor-
mance is increasing the range of diamond cutting tool materials available to industry.
Figure 39. Section through typical HPL wood flooring panel.
2.6.2.1 High Speed Finish Machining of High Pressure Laminate Wood Flooring
During the 1990s, the wood working industry has seen a phenomenal growth of the
use of PCD cutting tools in sawing, milling, and profiling of high and medium
density fibre board, chipboard, particle board - all known to be abrasive materials.
In the last five years, high pressure laminate (HPL) - a generic wood product, with
its decorative aluminum oxide coated surface for high wear resistance, has emerged
as a flooring material. Figure 39 shows the makeup of a HPL wood flooring
material. In this application, CVD thick film diamond has recently been tested
alongside pcD and single crystal diamond tools for the finish machining of the
highly abrasive coated layer of HPL wood flooring material, with very encouraging
results [84] as described below.
2.6.2.1.1 Machining Parameters
Material machined : HPL boards Al-oxide layer

Cutter diameter : 200 mm
No. of teeth : 8
Spindle speed : 6000 r.p.m.
Cutting speed : 63 m s-'
Feed rate : 53 m min-'
Feed per tooth : 1.1 mm
For CVD diamond materials, the tool edges were prepared by mechanical grinding
using a standard cutter grinder and diamond wheels. The cutting edge of synthetic
single crystal diamond was prepared by scaife polishing, and the pcD tools' edges
were prepared by the standard Electric Discharge Grinding (EDG) method.
6 14 2 Applications of’Diamond Synthesized by Chemical Vupor Deposition
The criterion by which tool life is measured is chipping of the board edge. When
chips of an ‘unacceptable’ size are produced, then the tool is changed. ‘Unaccepta-
ble’ chip size is determined visually by the machine operator.
2.6.2.1.2 Test Results Figure 40 shows, for each test, the average linear length
machined by the diamond tools, before the tools were taken out on the basis of visi-
ble edge chipping of the floor panels.
The edge wear of CVD tools appears to have been more influenced by grain pull
out and erosion at the grain boundaries than wear.
All the materials evaluated in these tests are capable of machining HPL flooring,
however, it is quite clear that single crystal diamond gives the longest tool life and
CVD diamond gives a marginal advantage compared with PCD. Both synthetic
single crystal diamond and CVD diamond materials, while providing high tool
life, need to be evaluated in the greater cost benefit equation for specific workpiece
materials, the requirements of individual HPL flooring products and the manufac-
turing environment pertaining.
Figure 40. Average linear length machined by different tool materials.
2.6.2.2 Dry Machining of SiCp A1 Metal Matrix Composite

MMCs are typically alloys of aluminum reinforced with ceramic particles
(usually silicon carbide). This composite material combination has the lightness
of aluminum but the strength and temperature resistance of cast iron. Con-
sequently, MMCs have tremendous potential in the automotive and aerospace
industries for the replacement of cast iron to considerably reduce weight.
MMCs, although readily machinable, are highly abrasive and will dull the cutting
edge of conventional tools in a matter of seconds. The performance of CVD dia-
mond thick and thin film and the extremities of pcD grain sizes of 2 and 25pm
have been compared when machining an MMC. The machining conditions and
parameters are detailed below.
i
0.35
PCD 025
3 0.3
E PCD 002
0.25
m
g 0.2 CVD thick film
5 0.15
Y
= /&.5:: CVD coating

.-8 0.1
2 0.05
0
2.6.2.2.1 Machining Conditions/Parameters
Material Machined : 40% Sicp A1 (A356) - MMC

Insert Style : TN MN 160408
Cutting Speed : 400 m min-l
Feed Rate : 0.05 mm rev.-’
Depth of Cut : 0.5 mm
No Coolant
2.6.2.2.2 Test Results Figure 41 shows the flank wear on the respective tool relative
to the machining time. In the case of the CVD thin coated tool, initially the rate of
tool wear is low but as soon as the ‘thin’ coating is worn through the flank wear
increases rapidly. The lowest rate of tool wear is observed on the CVD thick film
(0.5 mm) insert. Although testing is limited, the CVD thick film flank wear indicates
that the edge has ‘bedded in’ and steady state wear has been established. This sug-
gests that the ultimate tool life could be quite considerable.
The machining of MMCs, particularly high volume fraction, > 35%, generates
high temperature and involve having to efficiently shear hard silicon carbide parti-
cles in the cut zone. The high thermal conductivity/resistance to thermal degrada-
tion and abrasion resistance of CVD are particularly suitable for machining high
volume fraction MMCs.
Table 6 summarizes the relative merits of the three major diamond cutting tool
materials - pcD, CVD diamond, and synthetic single crystal as current technology
~
stands, from the perspective of tool performance.

It is clear that the various diamond cutting tool formats are largely complemen-
tary and, with the development of electrically conductive CVD diamond, the poten-
tial application areas of the machining of highly abrasive workpiece materials
becomes a reality.
Table 6. The relative merits of different diamond cutting tool materials: tool performance.
PCD CVD diamond Single crystal

(thick film)
Toughness JJJ JJ J
Abrasion resistance JJ JJ JJJ
Thermo-chemical resistance J JJJ JJJ
2.6.3 Chemical Vapor Deposited Diamond Dressers

Diamond is extensively used for the dressing of precision grinding wheels. The
grinding process using conventional, diamond and cBN abrasives is extensively
used in the finishing of ferrous, tungsten carbide and ceramic components. Here,
the geometrical tolerance, the surface finish of the wheel and its condition is critical
and the dresser's function is to achieve and maintain these characteristics.
CVD diamond has recently been available in a range of dresser log formats and
has shown promising results. CVD diamond potentially has a number of advantages
compared to other diamond dresser material formats. The polycrystalline nature of
CVD makes it insensitive to orientation. It also has extremely uniform wear charac-
teristics. This means that dresser manufacture is more straight forward and the wear
of the dresser is more consistent. The possibility of extracting long sections of CVD
from a large disc means that products like blade dressers can be made with contin-
uous diamond pieces compared to numerous discrete stones or pieces that are
required to make up the working length of the dresser. Also CVD diamond has
increased toughness, meaning that chip-free grinding of the dresser point is easier
to achieve.
CVD diamond has been compared to single crystal synthetic diamond in
single point dressing tests. The conditions and results of these tests are described
below.
2.6.3.1 Dresser Test Conditions

Wheel Type: Aluminium oxide WA46 KV (300 mm O/D)
Wheel Speed: 2300 r.p.m.
In-Feed: 0.02 mm wheel diameter-'
Feed: 4 mm s f l
Flood Coolant
2.6.3.2 Test Results

From Fig. 42 it can be seen that CVD diamond exhibits almost identical rates of
wear compared to single crystal synthetic diamond. Effectively this means that
CVD diamond offers a range of geometric and orientation advantages in dresser
design without having to compromise on the wear resistance of the dresser.
’
Dresser Wear (mm)
0. 5 Dnrarw paramatem
Meel AI-0xde WA46KV (300mm dla x 50 rnm d e )
Speed 2300rpm
In-Feed 0 020 mm I wheel dla
Feed 4 rnrnlrec
Caolant Floodmolant
0.05 -
0 ,
0 200 400 600 800
Volume of Wheel Removed (cm3)
CVD Diamond
-m- MONZESS
Figure 42. CVD diamond and Monodress single point dresser wear relative to volume of grinding
wheel removed.
2.6.4 Chemical Vapor Deposited Diamond Wear Parts

The extreme abrasion and erosion resistance of polycrystalline CVD diamond
combined with chemical inertness, make it a perfect candidate for use in appli-
cations such as wear parts, solid-state sensors, actuators and tools [85,86]. The
combination of diamond’s hardness, high Young modulus, very low coefficient
of friction (less than 0.1 in air) and low wear rates (less than m3 N-’ m-’)
under extreme loading conditions [87,88] potentially give diamond great tribo-
logical advantage over most other materials: The advent of polycrystalline CVD
diamond deposition technology means that these tribological advantages can be
exploited in applications such as large bearings or seals. These can be fabricated
from either CVD diamond coated parts or from thick layers of free standing
CVD diamond. Examples of these applications areas are pumps, valves and pipe-
lines used in the extremely harsh tribological conditions seen in oil fields, often in
remote locations [89].
Limited data is available on the wear rate of CVD diamond [90,91]. A recent
study has been conducted in order to investigate the basic tribological properties
of bulk and thin-film CVD diamond. The sliding wear characteristics of a number
CVD diamond materials have been evaluated using a modified Denison T62
abrasion [92] pin-on-disc test rig, operated in dry and lubricated sliding wear
conditions, and compared with a typical hardened steel which is often used for
wear part components.
Discs and pins of each material were made for testing; the pins all had chamfered
edges to prevent edge-galling and were made square to minimise any misalignment
thus reducing the contact area. Each material was tested first in dry sliding wear and
then in lubricated sliding wear and only tested against components made of the same
material.
The test parameters were identical in each case, with a contact stress of 15 MPa
and sliding speed of 150 m min-' . The tests were performed for 48 h where possible
(to steady state or until sliding speed diminished through increased friction).
The tests all began at nominal room temperature of 20°C and, where required, the
lubricant used was a typical heavy white mineral oil; excess lubricant was removed by
a repeated solvent wash prior to weighing the pin to determine the specific wear rate.
The temperature rise associated with both the dry and lubricated wear tests did
not exceed 90°C. Such a temperature rise can be associated with the low coefficients
of friction (and carrier mechanisms during the lubricated tests). Such increases are
not sufficient to noticeably alter the properties of the CVD diamond or to break-
down the mineral oil lubricant.
The coefficient of friction determinations have been made at specific times into the
sliding wear tests. These have an associated uncertainty of approximately flo%,
predominantly from the (unsteady) change in friction coefficient with time as the
sample surface is modified by wear and vibration/noise associated with the friction
monitoring transducer.
It is worthwhile noting that there are several sources of error in the determination
of the specific wear rate which give approximately an order of magnitude uncer-
tainty in the quoted values, in particular, the mass loss through wear of the pins
is minuscule - a consequence of the excellent abrasion resistance of CVD diamond
~ and such a low loss is easily affected by any residual lubricant.
As can be seen from the data presented in Table 7 the surface finish of the CVD
diamond has a considerable effect on both the wear rate and coefficient of friction.
Table 7. Wear test results of CVD diamond and hardened steel
Material Surface Dry or Specific Friction Friction

finish - lubricated wear rate coefficient coefficient
Ra ( w ) ( x lo-'* m3 (initially) (at 24 hours)
N-l ,-I
1
Hardened steel 0.07 Dry NIA Galled at -15 s N/A
Hardened steel 0.07 Lubricated 1 0.08 0.08
CVDD (bulk) 0.06 Dry 986 0.115 <24 hour test
(Nucleation face)
CVDD (bulk) 0.1 Lubricated 0.4 0.038 0.049
(Nucleation face)
CVDD (bulk) 0.06 Dry 835 0.15 <24 hour test
(Growth face)
CVDD (bulk) 0.1 Lubricated 0.7 0.018 0.018
(Growth face)
CVDD (thin film) 1.52 Dry 61 1 0.124 0.246
(As-grown face)
CVDD (thin film) 1.52 Lubricated 27 0.11 0.08
(As-grown face)
CVDD (thin film) 0.23 Dry 758 0.116 0.196
(Lapped face)
CVDD (thin film) 0.23 Lubricated 84 0.11 0.05
(Lapped face)
References 619
However, whereas the hardened steel could not be run without a lubricating layer,
the CVD diamond does not require lubrication or might be self-lubricating.
It is clear that there is a great potential for very low wear rate and low coefficient
of friction CVD diamond materials prepared with the required surface finish for
wear part applications, where component lifetime is determined by sliding wear in
low lubricant or lubricant-free applications.
References
1. R. S. Sussmann, DIAFILM, A New Material for Optics and Electronics, Indust. Diamond Rev.,
1993, 53,555, pp. 63-72.
2. R. S. Sussmann, J. R. Brandon, G .A. Scarsbrook, C . G. Sweeney, T. J.-Valentine, A. J. White-
head, and C . J. H. Wort, Diamond Re/. Muter., 1994, 3, 303-312.
3. C. J. H. Wort, C. G. Sweeney, M. A. Cooper, G. A. Scarsbrook, and R. S. Sussmann, Diamond
Re/. Muter., 1994, 3, 1158-1 167.
4. R. S. Sussmann, C. S. J. Pickles, J. R. Brandon, C . J. H. Wort, S. E. Coe, A. Wasenczuk, C. N.
Dodge, A. C. Beale, A. J. Krehan, P. Dore, A. Nueara, and P. Calvani, Nuovo Cimento, Section
D , 1998, 12 (April), 503-526.
5. R. S. Sussmann, J. R. Brandon, S. E. Coe, C. S. J. Pickles, C . G. Sweeney, A. Wasenczuk, C. J.
H. Wort, and C . N. Dodge, CVD Diamond, A. New Engineering Material for Thermal Dielec-
tric and Optical Applications, Proceedings of the Industrial Diamond Association of America
meeting on Ultrahurd Materials, Windsor, Ontario 28-30 May 1998, and Indust. Diamond
Rev., 1998, 58(578), 69-77.
6. C. P. Klages, Chemical Vapor Deposition of Diamond Films, in Handbook qf Ceramic Hard
Materials, R. Riedel (Ed.). Wiley-VCH, Weinheim, 2000, pp. 390-419.
7. G. J. Davies, Section 2 of Diamond Materials and their Applications, in Hundbook of Ceramic
Hard Materials, R. Riedel (Ed.), Wiley-VCH, Weinheim, 2000, pp. 485-504.
8. M. W. Bailey, Section 1 of Diamond Materials and their Applications, in Handbook ofceramic
Hard Materials, R. Riedel (Ed.), Wiley-VCH, Weinheim, 2000, pp. 479-485.
9. R. Berman Chapter I , J. E. Field Chapter 9, D. Tabor Chapter 10, A. T. Collins Chapter 3, C . D.
Clark, E. W. Mitchell and D. J. Parson Chapter 2, and C. A. Brooks Chapter 12, in The
Properlies qf Diamond, J. E. Field (Ed.), Academic Press, London, New York, San Francisco,
1979.
10. C. D. Clark, A. T. Collins, and G. S. Woods Chapter 2, R. Berman Chapter 7. J. E. Field
Chapter 12. C . A. Brookes Chapter 13. D. Tabor and J. E. Field Chapter 14, J. Wilks and
E. M. Wilks Chapter 15, and J. E. Field Chapter 18, in The Properties of Natural andSynthetic
Diamond, J. E. Field (Ed.), Academic Press London, San Diego, New York, Boston, Sydney,
Tokyo and Toronto, 1992.
11. J. E. Field in Proceedings of the N A T O Advanced Study Institute on Diamond and Diamond-like
Films and Coatings 22 July 1990, R. E. Clausing, L. L. Hardon, J. C. Angus, and P. Koidl (Eds),
Plenum Press New York, London, 1991, pp. 17.
12. Chapters I , 3, 5 , 7 and 8. in Handbook uf’lndustrial Diamond~sand Diamond Films M. A. Prelas,
G. Popovici. and L. K. Bigelow (Eds), Marcel Dekker Inc. New York, Basel, Hong Kong,
1998.
13. Chapters I , 2, and 5. in Properties andGrovvth ofDiamond, G. Davies (Ed.), EMIS Data Review
Services, INSPEC IEE, London, 1994.
14. J. Save, in Infrared Optical Materials und Their Antirefection Coatings, Adam Hilger, Bristol,
Boston, 1985, p. 119.
15. D. C. Harris in Itfrared Windows and Dome Materials, Tutorial Texts in Optical Engineering,
Vol. TTlO SPIE Optical Engineering Press, Bellingham, WA (USA), 1992, Chapter 5.
-
16. M Seal and W. J . P. van Enckevort, Proceedings of SPZE- Diamond Optics, 1988,969, 144152.
17. M Seal in Industrial Diamond Review, P. A. Daniel (Ed.), 1971, 31 (Nov), 464469.
18. J. Doting and J. Molenadr, Proceedings ofthe Fourth Annual ZEEE Semiconductor Thermal and
Temperature Measurement Symposium, Sun Diego California, February. IEEE CH2530-
418810000-0113, 1988, pp. 113-117.
19. R. J. Caveney in Chapter 20 of Ref. [9].
20. M. Seal in Chapter 16 of Ref. [lo].
21. G. S. Woods in Chapter 3 of Ref. [13].
22. J. W. Vandersande in Chapter 3. of Ref. [13].
23. A. T. Collins in Chapter 9. of Ref. [ 131; B. A. Fox, L. Harstell, D. M. Malta, H. A. Wynands, G.
Tessmer, and D. L. Preifus in Proceedings of the Materials Research Society Symposium, 1996,
416, 319-329.
24. M. Seal, Zndust. Diamond Rev., 1979, 39(461), 115-118.
25. J. A. Savage, C. J. H. Wort, C. S. J. Pickles, R. S. Sussmann, C. G. Sweeney, M. R. McClymont,
J. R. Brandon, C. N. Dodge and A. C. Beale. SPZE Proceedings of the Windowand Dome Tech-
nologies and Materials Symposium, Orlando, Florida, April, 1997, 3060, 1 4 4 159.
26. D. C. Harris, Development of Chemical Vapour Deposited Diamondjor Infrared Optical Applica-
tions, Status Report and Summary of Properties, Report No N AWCWPNS TP 8210, Naval Air
Warfare Center Weapons Division, China Lake, CA (USA), 1994.
27. M. Massari, P. Union, G. A. Scarsbrook, R. S. Sussmann, and P. Muys, Proceedings ofthe SPZE
Laser-Znduced Damage in Optical Mateilals, Boulder, Colorado, November, 1995, SPZE, 2714,
177-184.
28. DIAFILM is a. trade mark of De Beers Industrial Diamond Division.
29. R F. Potter in Handbook of Optical Constants of Solids, E. D. Palik (Ed.), Academic Press Inc.
Orlando. San Diego, New York, London, Toronto, Montreal, Sydney, Tokyo, 1985, Chapter 2.
30. M. E. Thomas and W. J. Tropf, SPZE, 1994,2286, 114&151.
31. V. Vorlicek, J. Rosa, M. Vanacek, M. Nesiadek, and L. M. Stals, Proceedings of the 7th Euro-
pean Conference on Diamond, Diamond-like and Related Materials, Tours France, September
1996, Diamond Rel. Muter., 1997, 6 , 704707.
32. D. F. Edwards and H. R. Philipp, in Handbook of Optical Constants of Solids, E. D. Palik (Ed.),
Academic Press Inc. Orlando, San Diego, New York, London, Toronto, Montreal, Sydney,
Tokyo, 1985, subpart 3, pp. 668.
33. P. Dore, A. Nucara, D. Cannavo, G. De Marzi, P. Calvani, A. Marcelli, R. S. Sussmann, A. J.
Whitehead, C. N. Dodge, A. J. Krehan, and H. J. Peters, J. Appl. Optics, 1998,37 (July), no. 21.
34. P. Union, P. Muys, D. Vyncke, B. Depuydt, and P. Boone, SPIE, 1995,2870, 521-527.
35. C. S. J. Pickles, J. R. Brandon, S. E. Coe, and R. S. Sussmann, Proceedings of Symposium ZV,
Diamond Films, 9th Cimtec - World Forum on New Materials, Florence, Italy, June 1998,
P Vincenzini (Ed.), Faenza, Italy, 1999, pp. 435454.
36. T. J. Valentine, A. J. Whitehead, R. S. Sussmann, C. J. H. Wort, and G. A. Scarsbrook, Dia-
mond Rel. Muter., 1994, 3, 1168-1 172.
37. F. Sziics, University of Berlin and California Institute of Technology, Pasadena, California.
Private communication, 1998, see also: F. Sziics, M. Werner, R. S. Sussmann, C. S. J. Pickles,
and H. J. Fecht, J. App. Phys., 1999,86 (December) no. 11, 6010-6017.
38. F. Sziics, C. H. Moelle, S. Klose, H. J. Fecht, S. Fassbender, N. Meyendorf, and M. Werner, in
Proceedings of the International Conference on Micro Materials, April I997 (MICRO-MA ’97)
B. Michel and T. Winkler (Eds), 1997, pp. 569-572.
39. J. W. V. Vandersande in Chapter 1 of Ref. [13].
40. J. E. Graebner, M. E. Reiss, L. Seibles, T. M. Hartnett, R. P. Miller, and C. J. Robinson, Phys.
Rev. B, 1994, 50(6), 3702.
41. E. Wormer, C. Wild, W. Muller-Sebert, R. Cocher, and P. Koidl, Diamond Rel. Materi., 1996,5,
688-692.
42. See for instance P. A. Rizzi, Chapter 2.6 in Microwave Engineering, Prentice-Hall International,
Englewood Cliffs, NJ (USA), 1988, p. 41.
43. A. C. Lynch and R. N. Clarke, IEE Proceedings A , 1992, 139, 221.
44. A. P. O’Leary in Electro-Optics, Special Report on Thermal Imaging Systems, 2nd edn, Jane’s
Information Group, Antony Rowe Ltd, Chippenham, Wiltshire, UK, 1994.
References 62 1
45. E. J. Coad, The Response of C V D Diamond and other Brittle Maierials to Multiple Liquid
Impacts, PhD Thesis, University of Cambridge, 1996.
46. M. D. Hudson, C. J. Brierley, A. J. Miller, and A. E. Wilson in the Proceedings of the Windou.
and Dome Technologies and Materials Synzposium, Orlando, Florida, April 1997, SPIE, 3060,
196202.
47. C. J. H. Wort, J. R. Brandon, B. S. C. Dorn, J. A. Savage, R. S. Sussmann, and A . J. Whitehead
in Applications o f Diamond Films and Related Materials: Third International Conference, A.
Feldman, Y. Tzeng, W. Yarbrough, M. Yoshikawa and M. Murokawo (Eds), NIST Special
Publications, 1995, 885, 569-572.
48. Rofin Sinar Technologies, Annual Report, 1997, pp 6. 7.
49. C. A . Klein, Optical Engineering, 1990, 29(4), 343-350.
50. M . Sparks and M. Cottis, J. App. Phys., 1973, 44, 787-794.
51. S. Singer, Diamond optics, A. Feldman and S. Holly (Eds), SPIE, 1988, 969, 168-177.
52. C. A. Klein in Laser Induced Damage in Optical Materials, 1991, SPIE, 1624, 475492.
53. C. A. Klein, Dianiond and Relaied Materials, 1993, 2, 102&1032.
54. C. Wild, Chapter 10.3 in Lou-Pressure Synthetic Diamond, B. Dischier and C. Wild (Eds),
Springer Veriag, Heidelberg, 1998.
55. S. P. Timoshenko and J. N. Goodier, Chapter 12 in Theory of Elasticity, 3rd edn, 1970,
McGraw-Hill, NY, pp. 385.
56. B. Depuydt, P. Boone, P. Union, and P. Muys in Proceedings y f t h e 4th International Worshop on
Laser Beam and Optics Characterisation, Munich, June 1997, A. Giesen and M. Morin (Ed.),
1997, 657-663.
57. N. Ellis and D. E. Greening Diamond and Related Materials, 1993, 2, 572-581.
58. R. S. Sussmann, C. A. Scarsbrook, C. J. H. Wort, and R. M. Wood. Diamond Rel. Muter., 1994,
3, 1 I73 1177.
-
59. H.-K. Karow, Fabrication Methods fbr Precision Optics, John Wiley and Sons, 1993, p. 620.
60. Tim Holt, Rofin Sinar 1998, Private communication.
61. R. Heidinger. Digest 19th International Confirence on Injiared and Millimeter Waves,Lausanne
( C H I , SPIR, 1991, 1576,441.
62. L. Redcliffe, Digest 16th International Conference on Injiared and Millinieter Waves.Sindon ( J ) ,
JSAP Cat No AP 941228, p. 277.
63. M. Thumm, State of the art of high power Gyro-devices and,free-electron Masers, Wissenscliaf~
tliche Berichte F Z K A 5564, Forschungszentrum Karlsruhe GmbH, Karlsruhe 1995, (ISSN
0947-8620).
64. M. Thumm, J. of Infrared and Millimeter Wave, 199X, 19, No. I , pp. 3-14.
65. M. Thumm, Fusion and Engineering Design, 1995, 30, 139-170.
66. M . Thumm, FZK Institute, 1997, Private communication.
67. 0. Braz, G. Dammertz M. Kujntze, and M. Thumm, International Journal of Infrared and
Milliineter Waves, 1997, 18(8), 14651477.
68. R. Heidinger, A . Ibarra, and J. Molia, J. Nuclear Materials, 1998, 258-263B, part B, pp. 1882-
1826.
69. R. Sporl, R. Schwab, R. Heidinger, and V. Parshin, C V D Diamondfor High Power Gyrotrons:
Characterisutian of Dielectic Properties, 1TG-Pachbericht, Berlin, April 1998.
70. R. Heidinger, R. Sporl, M. Thumm, J. R. Brandon, R. S. Sussmann, and C. N. Dodge,
Progranime Commitire, 23rd International Conference on Infrared and Millimeter Waves,
Colchester, Essex. UK. ‘CVD Diamond Windows for High Power Gyrotrons’, September
1998, pp. 223-225.
71. 0. Braz, A. Kasugai, K. Sakamoto, K. Takahashi, M. Tsunekoa, T. Imai, and M. Thumm, Int.
J . Infrared Millimeter Waves, 1997, 18(8), 1495-1 503.
72. A. Kasugai, K. Sakamoto, K. Takahashi. M. Tsuneoka, T. Kariya, T. Imai, 0. Braz, M.
Thumm, J. R. Brandon, R. S. Sussmann, A. Beal. and D. C. Ballington, Rev. Sci. Instrum,
1998, 68(5).
73. K. Sakamoto, Japan Atomic Energy Research Institute, Private communication.
74. R. Heidinger. R. Schwab, R. Sporl, and M. Thumm, 22nd International Conference on Infrared
and MiNimeter Waves, Wintergreen. Virginia, July 1997, P. Freund (Ed.).
75. R. Heidinger. FZK Institute, Private communication.
622 2 Applications of Diamond Synthesized bqi Chemical Vapor Deposition
76. S. M. Hu, Critical stress in silicon brittle fracture, and effect of ion implantation and other
surface treatments, J. Appl. Phys., May 1992, 53(5), 35763580.
77. R. C. Eden, in Handbook of Industrial Diamonds and Diamond Films, M. A. Prelas, G. Popovici,
and L. K. Bigelow (Eds), Marcel Dekker Inc, New York, Basel, Hong Kong, 1998, Chapter 30,
1073-1102.
78. J. Bonhaus, D. Lorenzen, E. Kaulfersch, A. Denisenko, A. Zaitsev, S. Kwong, K. Man,
H. Reichi, and W. R. Fahrner, Proceedings of the International Conference on Micro Materials,
Micro Mat '97, Berlin, April 1997, B. Michel and T. Winkler (Eds), 1997, 505-508.
79. D Lorenzen, S. Heineman, J. Bonhaus, Proceedings of the International Conjerence on Micro
Materials, Micro M a t '97, Berlin, April 1997 (Ed.: B. Michel and T. Winkler), 1997, 513-516.
80. D. Lorenzen, Jenoptik Laserdiode, 1998, Private communication.
81. K. E. Goodson, K. Kurabayashi. and R. F. W. Pease, IEEE Transactions on Components
Packaging and Manujacturing Technology, 1997, Part B. - Advanced Packaging 20-1, 104109.
82. S. Weiss, E. Zakel, and H. Reichi, IEEE Transactions on Components Packaging and Manu-
facturing Technology, 1996, Part A. Vol. 19-1, 46-53.
83. Patent assigned to Jenoptik Laserdiode, DE 196 44 941 CI (D. Lorenzen, 1998. Private
communication).
84. P. K. Sen, M. W. Cook, and R. Achilles, To be published: Proceedings of the Industrial Diamond
Association of America Meeting on Ultrahard Materials, Windsor, Ontario 28-30 May, 1998.
85. S. J. Bull and A. Matthews, Diamondjor Wear and Erosion Applications, Diamond and Related
Materials, 1992, 1, 1049-1064.
86. B. Lux and R. Haubner, Phil. Trans. R . SOC.Lond. A, 1993, 342, 297-311.
87. T. Le Huu, H. Zaidi, and D. Paulmier, Wear, 1977, 203-204,442446.
88. M. N. Gardos in Protective Coatings and Thin Films,Y. Pauleau and P. B. Barna (Eds), Kluwer
Academic Publishers, The Netheriands 1997, 185-196.
89. D. Cooper, F. A. Davies, and R. J. K. Wood, J . Phys. D: Appl. Phys., 1992, 25, A195-A204.
90. A. Edemir and G. R. Fenske, Proceedings of the 51st Annual meeting of the S T L E , Cincinnati -
Ohio, May 1996, 1-8.
91. R. A . Hay and J. M. Galimbarti, Handbook of Industrial Diamonds and Diamond Films, M. A.
Prelas, G. Popovici, and L. K. Bigelow (Eds), Marcel Dekker Inc, New York, Basel, Hong
Kong, 1998, Chapter 33, 1135-1 147.
92. M. J. Neale (Ed.), Tribology Handbook, Butterworths, London, 1973 (2nd Edition 1994).
3 Diamond-like Carbon Films

C.-P. Klages and K. Bewilogua
3.1 Introduction
The term ‘diamond-like carbon’ was coined in the early 70s in order to denote amor-
phous carbon films, prepared by physical vapor deposition techniques, which were
expected to approach genuine diamond, as far as density and mechanical properties
are concerned. Today the expression (frequently abbreviated to ‘DLC’) is used for a
fairly wide range of amorphous films containing carbon atoms in different states of
hybridization and possibly also further elements such as hydrogen, silicon, or metals
(‘Si-DLC’, ‘MeeDLC’). There are presently no generally accepted criteria for the
use of the term DLC or for a clear distinction between ‘diamond-like’ on the one
hand and ‘graphite-like’ or ‘polymer-like’ on the other. This review will therefore
pragmatically emphasize those kinds of amorphous carbon-based films which
have found the greatest interest from a scientific or an applicational point of
view. Aside from hydrogen-free amorphous carbon (a-C) and especially highly
tetrahedrally coordinated amorphous carbon (ta-C), deserving the denotation
‘diamond-like’ most of all, hydrogenated films (a-C :H) and metal-containing
hydrogenated films (MeC: H) will be treated in more detail, owing to the technical
importance for mechanical-tribological applications which these materials have
gained. Section 3.2 will present the deposition methods which have been used or
are being used for the preparation of DLC films. Microstructures and properties
will be treated in sections 3.3 and 3.4 and the final section will deal with applica-
tional aspects of DLC coatings.
3.2 Preparation Methods for Diamond-like Carbon

Films
3.2.1 Hydrogenated Amorphous Carbon (a-C:H)
In 1976 Holland and Ojha [l] reported the deposition of a-C: H films using a glow
discharge in hydrocarbon gases excited by a radio frequency (RF, 13.56MHz)
powered substrate electrode. The counter-electrode was the grounded wall of the
deposition chamber. Because this technique was relatively simple in operation the
work of Holland and Ojha can be seen as the starting point of intensive research
and development efforts aiming at practical applications of DLC films. Hydro-
genated amorphous carbon films have relatively low electrical conductivities,
typically less than lop6 R-’ cm-’. The operation of the substrate with RF allows
624 3 Diamond-like Carbon Films
deposition of a-C: H films with thickness of several micrometers on conducting as

well as on insulating substrates without surface charging effects. On the other
hand, attempts to excite the substrate electrode in the glow discharge only with
DC (direct current) power were reported (e.g. by Whitmell and Williamson [2]
and later by Grill and Pate1 [3]). There are indications that these DC deposition
processes arc sustained by secondary electrons generated at masks or shield sur-
rounding the substrates [4, 51. These secondary electrons compensate the surface
charges. But at least for thicker films the charging of the films becomes an essential
problem if their resistivity is higher than about lo8 Rcm. This can be explained by a
simple estimation of the voltage drop over the film (see below). Today such two-
electrode processes with only DC excitation have nearly no importance. However,
there are intensive efforts to find alternatives to the RF processes, because this tech-
nique has only a limited potential for economic up-scaling of the coated parts
dimensions. One alternative is the substitution of R F in the MHz range by lower
frequencies, especially medium frequencies up to 100 kHz. In 1986 Catherine and
Couderc [6] revealed that films prepared with 13.56MHz or with 50kHz, respec-
tively, had nearly the same properties. At present the medium frequency technique
is interesting above all from the viewpoints of up-scaling and costs.
Following the work of Holland and Ojha several other methods for preparation
of a-C:H were developed, for example: sputter deposition [7, 81, ion plating
[8, 91, cascade arc plasma jet [lo], and simultaneous operation of a RF and a micro-
wave (2.45 GHz) discharge [l I].
Important parameters of all a-C : H deposition processes are the ion energy, com-
monly expressed by the negative substrate bias voltage Ub. the power density at the
substrate, the pressure of the used hydrocarbon gas or vapor, the ion and neutral
flux densities, and the substrate temperature. In contrast to the diamond film deposi-
tion the substrates must not be heated additionally. Therefore it is an essential
advantage of DLC films that they can be deposited on temperature-sensitive
parts. In Table 1 typical parameter ranges of different a-C : H deposition processes
are summarized. Besides the technique, the type of precursors, gas pressures and
bias voltage, the power density at the substrate and the linear growth rate arc
considered. Data on particle flux densities can be found only in few papers. All
the mentioned parameters influence structure and properties of the films. It
should be noted that before starting the deposition process, the substrates must
be cleaned by sputter etching with noble gas ions in order to achieve a good
adhesion.
The following discussion of characteristic features of a-C :H deposition processes
will be concentrated on the R F and the ion plating method.
Figure 1 shows a scheme of the R F glow discharge deposition method. The RF
power (the mostly applied frequency is 13.56MHz) is capacitively coupled to the
substrate electrode, the counterelectrode commonly is the wall of the vacuum
chamber having ground potential. In the frequency range of some MHz only the
electrons, but not the ions, can follow the R F voltage. Due to the large differences
in the mobilities of electrons and ions a high negative DC self-bias U, will be gen-
erated at the substrate electrode if the area of the powered electrode is markedly
smaller than that of the grounded electrode (asymmetric system, see e.g. [12]). In
625
3.2 Preparation Method.yfor Diamond-like Carbon Films
Substrate
Process pressure: 1 to 10 Pa
Self-bias: -50 to -800 V
Power density: 0.05 to 0.5 W cm-2
-
Current density: 0.6 mA cm-’
Deposition rate: - 1 pm h-’
Figure 1. Schematic drawing of the radio frequency (RF) plasma technique for deposition of a-C: H
films with some typical process parameters.
practice this condition is usually fulfilled. The self-bias drops across the plasma
sheath in front of the negatively biases substrate electrode down to the plasma
potential Up. The maximum ion energy Eimax of the ions arriving at the substrate
surface is e( Up - u b ) , where e is the unit electron charge. In asymmetric systems
the mean plasma potential is a few volts positive. Therefore u b is a sufficiently
good measure for the maximum ion energy. However, the ions can lose energy
if their mean free path in the plasma sheath is smaller than the thickness of
the sheath. For the pressure range typical for RF de osition of a-C:H ( z 10Pa)
P
Bubenzer et al. [13] derived the relation Ei M U b / P‘ ’ . Catherine and Couderc [6]
confirmed this relation for the R F process. The self-bias u b varies with the discharge
parameters according to Ub (W/P)’’’ where W is the R F power and P the
N
pressure of the process gas [6]. For the film growth rate the relation R z U b P was
found experimentally [ 131. The deposition rate also depends on the type of precursor
C,H, used. The highest rates can be realized with benzene (C6H6)and cyclohexane
(C6HI2),the lowest with methane. One possible explanation is that the rate
decreases with increasing ionization potential of the source gas [14].
The negatively self-biased substrate surface is bombarded with ions which are
generated in the discharge (C,H; and noble gas ions if noble gases are added).
However, not only the C,H; ions introduce carbon into the film. Comparing the
calculated mass deposition rates of major ions (CH; for CH4 gas and C 6 H l as
well as C6Ht for C6H6 precursor) with the experimentally determined deposition
rates Catherine and Couderc [6] found clear discrepancies in the correlations
between rate and R F power and between rate and pressure. In order to overcome
these, the adsorption of activated species and their incorporation into the growing
film must be taken into account additionally to explain the experimental data.
3.2 Preparation Methods for Diamond-like Carbon Films 627
Concerning the details of a-C:H deposition mechanisms there is still a lot of

uncertainties. A comprehensive discussion of the situation in this field was given
by Moller [15].
Besides pressure and power the substrate temperature can influence the growth
rate of a-C:H films. Commonly the rate decreases with increasing temperature
[15, 161,owing to a declining contribution of adsorbed molecules to the film
growth rate [I 51. Thus the observed temperature-dependence of the growth rates
is an additional support for the adsorption model.
In the R F glow discharge process the plasma generation and the film deposition
will be realized by the same electrode. The advantage of the ion plating method is
that plasma generation and film deposition are separated. Some typical process
parameters for this method are noted in Table 1. The plasma can be generated by
fast electrons ( ~ 1 0 eV)
0 from a hot cathode-anode arrangement [S, 91 or by micro-
waves [17]. Positive ions from the plasma then are accelerated to the negatively
biased substrate. This is done with DC voltage only. A simple approximation
shows that the use of DC becomes problematic for film resistivities > lo8 Rcm.
The voltage drop over the film is estimated as AU = pdfji where p is the resistivity
of the film, df the film thickness and the current density at the substrate. For typical
'
values df = 3p m, ji = 0.1 mA cm-ji and p = 1O8 Rcm [S] follows A U = 3 V. For

substrate voltages of lOOOV and more such a voltage drop is negligible [8, 51. For
higher resistivities or for insulating substrates this deposition method becomes
unsuitable.
If benzene vapor is used, about 70% of the ions striking the substrate consist of
C6H: (x 5 6). As in the case of R F plasma deposition not only ions contribute to
the film growth. The contribution of the hydrocarbon ions to the film deposition is
not more than 50%, further contributions are due to adsorbed and polymerized
species [5].
The properties of the deposited a-C: H films, to be discussed in sections 3.4 and 3.5,
strongly depend on the chosen process parameters, especially on pressure, power
density, bias voltage and temperature. Therefore the deposition conditions must be
carefully adjusted for preparation of films on parts or tools to be utilized in industrial
practice. Today both R F plasma and ion plating deposition methods are in use for
research and development and for prototype production in the industry.
3.2.2. Hydrogen Free Amorphous Carbon (ta-C)

There is a considerable interest in hydrogen free amorphous carbon films with high
sp3 contents (high contents of tetrahedrally bonded carbon atoms), because - com-
pared to a-C :H - such films have properties much closer to diamond. For the pre-
paration of ta-C films a source of energetic carbon species is needed, either with or
without mass filtered carbon ions. Already in 1971 Aisenberg and Chabot [18]
deposited carbon ions of 40-100eV separated from a special source creating
argon and carbon ions by sputtering a graphite target. In their pioneer work they
introduced the designation diamond-like carbon for the deposited material. This
was the first case of preparation of hydrogen free DLC. At present several different
methods to prepare such films are known [19, 201:

(i) Ion beam deposition (with or without mass selection).
(ii) Sputtering from a graphite target.
(iii) Arc evaporation from graphite cathodes (with or without filtered beams).
(iv) Laser ablation from a graphite target.
(v) Ion-beam assisted deposition (IBAD: simultaneous bombardment of the grow-
ing film with noble gas ions).
Mass selected ion beam deposition is a special method with well defined conditions
concerning ion masses and energies. Hofsass et al. [21] for example used I2C-' ions,
in the most cases with deposition energies of 100 eV. This method is interesting for
basic research, but seems to have no perspectives for industrial uses.
From the viewpoint of practical applications especially the sputter and arc
evaporation techniques are interesting. Using the sputter deposition from graphite
targets with argon and an argon-hydrogen mixture, respectively, Savvides and
Window [22, 231 prepared amorphous carbon films with high contents of sp3
bonded carbon. The sp3/sp2ratio in the films increased with increasing energy per
carbon atom. The increase of the sp3 content in the case of hydrogen-containin
sputter gas was explained by a preferred reaction of atomic hydrogen with sp 8
bonded carbon as known from diamond film deposition.
The vacuum arc evaporation technique seems to have a high potential for indus-
trial applications. Arc plasmas are characterized by a high degree of ionization
which is especially suitable to obtain high sp3 contents in amorphous carbon
films. A disadvantage of arc evaporation is the occurrence of graphitic micro-
particles. These particles can be avoided by filtering the particle beam [24]. However,
such filter systems reduce the deposition rates drastically. As an alternative method
which allows to control the arc process and to reduce the microparticle emission
the laser arc technique was developed by Scheibe et al. [25]. Recently this technique
was presented as a new industrial technology for high rate deposition of hard
amorphous carbon [26].
In laser ablation systems a laser beam strikes on a graphite target surface and
produces a flux of carbon material from the target to the substrate where the film
grows (for a review see [27]). The emitted particles have a broad range of energies
and can be utilized to prepare both hydrogen free as well as hydrogenated amor-
phous carbon films.
Ion beam assisted deposition techniques are characterized by a generation of
carbon from a target by evaporation, laser ablation or sputtering [28] and a
simultaneous bombardment of the growing film with argon ions. The low
energy ion bombardment (up to some 100eV) leads to dramatic changes in the
film structure and the creation of dense carbon with high concentration of sp3
bonds.
Several models are presently under discussion to explain the formation of highly
tetrahedrally coordinated carbon ta-C. According to one of these theories the ta-C
deposition can be described as a subplantation process [19,29]. The principle of this
process is that carbon ions penetrate the first atomic layers and enter interstitial
positions, thereby increasing the local density and inducing local compressive
3.2 Preparation Methodsfor Diamond-like Carbon Films 629
stress. The densification can cause a change of the hybridization from sp2 to sp3.
This process occurs in an energy range which obviously depends on the preparation
method used [19]. Considering the experimental data, high sp3 fractions in the a-C
films ( > 70%) were found in the energy range between 20 and 600eV. For too low
energies the ions can not penetrate, will be stuck at the surface and form sp2-carbon.
For too high energies the ion dissipates the energy in a ‘thermal spike’ and relaxation
of the densified regions occurs. At present several methods to prepare ta-C films are
known, but mostly on a laboratory level. Therefore the deposition technique is still
not suitable for broad technological applications as they are already realized in
different fields for a-C: H or MeC: H coatings.
3.2.3 Metal-containing Amorphous Hydrocarbon

An essential problem to be solved before introducing coatings to practice is a good and
reliable adhesion. The very promising properties of a-C :H films for a long time could
not be utilized because of their high compressive stress. For thickness of more than 1-
2pm the adhesion, especially on steel, was insufficient. This problem is still more
pronounced for ta-C films. Today there are possibilities to improve the adhesion
markedly, for example by interlayer systems (for more details see Section 3.5).
Another way to overcome the disadvantage of bad adhesion is the incorporation
of metals into the a-C: H films (MeC : H). Such films were first prepared by Dimigen
and coworkers in 1983 using a reactive R F sputter technique [30, 311. A large
number of metals, either forming carbides (like Ti, Ta, and W) or not (Au or
Cu), were investigated [32]. For applications in tribology the carbide-forming
metals are preferable. MeC: H films with suitably adjusted metal contents turned
out to have considerably lower compressive stress than a-C: H, but nearly the
same excellent friction behavior. On the other hand the wear resistance of
MeC:H coatings is lower than that of a-C:H (see Section 3.5.2). As a function
of metal-content, minimum wear rates were measured for metal to carbon ratios
of 0.1-0.2 [33]. With respect to industrial requirements, reactive DC magnetron
sputter techniques were developed for the MeC: H preparation [34, 351. Magnetron
sputtering has a great potential owing to larger deposition rates about one order of
magnitude higher than with the R F method and the excellent possibilities for up
scaling of the coating devices. An industrial DC magnetron batch coater for the
WC:H deposition was presented by Hofmann et al. [36]. Metals (mostly Ti, W)
or metal carbides (WC) will be used as target materials. The carbon content can
+
be controlled by the reactive gas content in the sputter gas (e.g. Ar C2H2).Like in
the a-C:H deposition the substrates must not be heated additionally. A substrate
bias is not necessary to deposit MeC:H films, but a moderate bias voltage
(Or, = - 100 to -200 V DC) clearly improves the coating quality. Other typical process
parameters to prepare MeC: H coatings with metal to carbon ratios of 0.1 to 0.2 are:
- power density at the target: 10 W cmp2.
- acetylene concentration (ratio of gas flows [C2H2]/[C2H2+ Ar]): 3 0 4 0 %
- total gas pressure: 0.3-0.8 Pa
- substrate sputter cleaning: Ar ions ( U , = -600 V DC)

- substrate current density: > 1 mA cmP2 (unbalanced magnetron mode)
- substrate holder temperature: <200°C
- time for deposition process: about 100 min
Concerning the mechanism of MeC: H sputter deposition two components must be
considered [35, 371.
(i) metal and carbon are sputtered from the target in a Ar-hydrocarbon gas mix-
ture. The carbon material consists of loosely bonded hydrogenated carbon
which will be deposited and resputtered continuously at the target surface.
Such a mechanism can explain the very high deposition rates (more than
10 pm h-' on substrates stationary in front of the target) reached with the DC
magnetron [35].
(ii) Hydrogenated carbon is deposited directly from the plasma surrounding the
substrate (a-C:H process similar to that discussed in Section 3.2.1). In the
case of magnetron sputter deposition this contribution is small ( ~ 1 0 % ) .
Today MeC:H coatings are widely used in industrial practice, for example for
machine parts improved with wear resistant surfaces exhibiting low friction. On
the other hand the wear resistance of MeC :H has still not reached the level of a-
C: H. Therefore a further improvement of the MeC :H quality as well as a higher
productivity and reliability of a-C: H and ta-C coatings will be necessary.
3.3 Microstructure and Bonding of Diamond-like

Carbon
3.3.1 Amorphous Carbon and Hydrogenated Amorphous Carbon
Among the Group IV elements carbon takes an exceptional position owing to its
ability to exist in bonding configurations formed by different hybridizations of its
s and p atomic orbitals [38]: The tetrahedral sp3 hybridization with four o bonds
to neighboring atoms is realized in the diamond allotrope, graphite is formed by
trigonally sp2-bonded carbon atoms, while polyacetylene ( Q n is an example of a
compound with sp-hybridized carbon with two o bonds arranged to form a straight
line. The description of DLC structures even if based solely on carbon (a-C),
~
without any hydrogen or metal incorporation - is complicated by the fact, that

all kinds of hybridizations will in general be present in the amorphous matrix
which may also contain nanocrystalline carbon phases. Additionally, there can be
a chemical ordering, such as by clustering of sp2-bonded atoms to longer chains
or planar graphite-like segments within a predominantly sp3-bonded matrix [39].
In hydrogen-containing DLC films as grown by plasma CVD or reactive PVD
methods in hydrogen atmospheres, different amounts of H atoms can be bonded
to the sp, sp2 and sp3 bonded carbon atoms, resp.
3.3 Microstructure und Bonding of Diamond-like Curbon 63 1
DOS t valence
bands
conduction
bands
EF Energy
Figure 2. Schematic drawing of electron density (DOS) of states of amorphous carbon phases.
EF = Fermi energy (after [14]).
The most important methods which have been in use to reveal the structure and
bonding of a-C and a-C:H films have been summarized in review articles [39, 401.
In order to understand these methods and the electrical and optical properties of
DLC, a closer look to its electronic structure is necessary. Figure 2 is a simplified
schematic diagram of the electron density-of-states (DOS) of amorphous carbon, as
it is frequently shown in the literature (e.g. in [ 141). A pair of o and o*bonding and
antibonding states is due to the bonds forming the skeleton of the a-C or a-C:H
network. Owing to the weaker n: bonds, the n:/n:* couple is located more closely to
the Fermi energy E F .(Strictly speaking, the nomenclature of o and n: molecular orbi-
tals would require the presence of a mirror plane within a molecule [38]. Nevertheless,
this simplified concept is of certain value for a discussion of bonding in DLC.)
The hydrogen content in a-C:H films can be determined by a number of methods,
including classical chemical combustion analysis, nuclear reaction analysis (NRA),
elastic recoil detection (ERD), nuclear magnetic resonance (NMR) spectroscopy, or
an analysis of C-H band intensities in infrared vibrational absorption spectroscopy
(IR). The most reliable information about the relative amounts of sp2- and sp3-
P-
bonded carbon (s bonded C atoms usually do not play an important role) comes
from solid state 1 C-NMR spectroscopy [41] and electron energy loss spectroscopy
(EELS) [4244] (see also the references to other original papers in [39] and [40]). The
results of Weissmantel rt al. already pointed to the existence of a special carbon
structure in films deposited either by dual beam sputtering or by condensation of
species from a low pressure ionization system operated with benzene. Fink et al.
used R F plasma deposition of a-C: H from benzene and measured the EELS spectra
for samples grown at different substrate bias voltages before and after annealing up
+
to 1000°C. The o n plasmon energy positions and intensities are significantly
dependent on the sp3 and sp2 fractions. The concentration of carbon in sp3 config-
uration was found to be roughly two third in as-grown films and increased after
annealing. The hydrogen content declined strongly from 60 to 25 at-%, when the
bias voltage was raised from 200 to 12OOV, accompanied by a significant density
increase (1.49 to 1.75g ~ m - ~A) strong
. dependence of the sp2 carbon and the hydro-
gen content, in a-C:H films grown by R F or DC plasma and ion beam deposition,
was also noted in the NMR measurements performed by Kaplan et al. [41].
A compilation of the sp3 percentages and H contents of typical a-C and a-C:H
films was given by Robertson [45]. Table 2, containing also density, hardness, and
Table 2. Structural characteristics and properties of various kinds of carbon, data taken from an
earlier comuilation 1451.
Yo sp3 at-% H Density, Hardness, Gap

d (g ~ m - ~ ) H (GPa) (eV)
Diamond 100 0 3.515 100 5.5

Graphite 0 0 2.267 -
-0.04
Glassy carbon -0 1.3-1.55 2-3 0.01
Polyethylene 100 67 0.92 0.01 6
a-C, evaporated 1 - 1.9-2.0 2-5 0.4-0.7
a-C, sputtered 2-5 - 1.9-2.4 11-15 0.40.7
a-C, MSIB (ta-C) 90 f 5 <9 3.0 30-1 30 0.5-1.5
a-C : H, hard 30-60 1040 1.6-2.2 10-20 0.8-1.7
a-C: H, soft 50-80 40-65 0.9-1.6 <5 1.w
~
MSIB = mass selected ion beam deposition.
band gap energy values, compares these special forms of carbon and hydrogenated
carbon to a few 'conventional' materials. It is important to note, that a 100% sp3
content, accompanied with a band gap energy in excess of 5 eV, is found in diamond
as well as in polyethylene, representing the extremes of the hardness and density
scales. In DLC, a high amount of tetrahedral bonding is present in very soft, poly-
mer-like a-C :H films with high H content and low density as well as in amorphous
a-C films, grown by mass selected ion beam deposition (MSIB), which have a low or
zero H content and densities as well as hardness values approaching those of dia-
mond. The latter, ta-C (tetrahedral amorphous carbon) films, which can also be
grown by filtered vacuum arc deposition [46], have in recent years gained consider-
able scientific interest. Several models have been proposed to explain the formation
of these very dense amor hous carbon phases [39], [47]. A prerequisite to obtain
i:
maximum density and sp content is a suitable average kinetic energy of carbon
ions in the 20-600eV region. If the energies deposited per carbon atom during
film growth are much smaller, e.g. during evaporation of graphite ( E M 0.1 eV) or
plasma deposition of an a-C:H film from a hydrocarbon at small substrate bias
voltages, soft films of low density and hardness will be deposited, consisting
predominantly of sp2-bonded carbon (in a-C) and strongly hydrogenated sp3-
bonded carbon (in a-C : H).
Among the methods mentioned above, IR spectroscopy is by far the most
inexpensive and - if the analysis of absorption band intensities is done with great
care - a powerful tool for structural investigations of hydogenated amorphous
carbon films. The interrelation of deposition parameters and structural data of
Table 3 which were reported by Dischler [48], are partly based on IR spectroscopic
data (namely the ratios of sp"-hybridized carbon atoms and ratios of CH,-bonded
segments) and can be considered as typical for plasma deposition of a-C :H. Growth
of a dense, hard phase with high refractive index is favored by high ion energies of
carbon-containg species deposited on the substrate, which are achieved under con-
ditions of high bias voltages and relatively low pressures. In case of the softer a-C: H
film, a large amount of hydrogen is present in methyl groups which do not contri-
bute to the rigidity of the film. For the harder sample, methyl groups are virtually
3.3 Microstructure and Bonding of Diamond-like Carbon 633
Table 3. Deposition conditions and structural properties of hard and soft,

polymer-like a-C: H films (after [48]).
a-C :H (hard) a-C : H (soft)

~~ ~
Frequency (MHz) 2.3 2.3

Self bias voltage (kV) 1.O 0.1
Benzene pressure (Pa) 3.2 6.1
Estimated kinetic ion energy (eV) 100 10
Deposition temperature ("C) 50 25
sp3 : sp2 :sp (%) 68:30:2 53:45:2
CH3 : CH2 :CH (%) 0 :40 :60 25:60: 15
Optical gap (eV) 1.3 3.0
Refractive index 2.0 1.65
Density (g ~ m - ~ ) 1.65 1.3
absent and all sp3-bonded carbon atoms make at least two bonds to neighboring
carbon atoms, thus contributing to the cross-linking of the network.
Quantitative investigations of the connectivity in a-C: H random networks were
introduced by Robertson [49] and Angus [50], based on the pioneering work on
random covalent networks of Phillips, Thorpe and Dohler et al. [51-531. In consid-
erations of this kind, the number of translational degrees of freedom of atoms (= 3
for one atom in three-dimensional space), forming the random network, are
balanced out against the number of constraints N,,, on their movement imposed
by bonds to neighboring partners. It can be shown that in three-dimensional net-
works whose bonding is dominated by directed valence bonds to nearest neighbors
the number of constraints for one atom is
Ncon(r)= r2/2 for Y 5 2 and Ncon(r)= 5 r / 2 - 3 for r 2 2
(1)
r is the coordination number. As long as there are less constraints than degrees of
freedom in a random network, atoms can still move in order to minimize strain
energy, the network is 'floppy'. As soon as the degrees of freedom are exhausted
by an increased number of bonds, the network will be fully constrained, additional
cross-linking will then increase the strain energy rapidly. For a monatomic network
this happens at an average coordination number r,, = 2.4. For a completely con-
strained network consisting of sp2- and sp3-bonded carbon atoms as well as
hydrogen atoms with molar ratios x3,x4 and x I , Angus [50] deduced an equation
for the sp3/sp2ratio (x4/x3)as a function of the hydrogen content xl :
xq/x3 = (8x1 - 3)/(8 - 13x1) (2)
In order to be composed exclusively from sp3 carbon atoms (corresponding to a
zero denominator in Eq. (2)) and still be not overconstrained, a network should,
according to Eq. (2), contain a hydrogen atom fraction x1 of 8/13 = 0.615 [54].
Modification have to be introduced, however, in order to take account of the clus-
tering of sp2-bonded carbon [55]. In early papers on DLC structure and properties,
extended clustering of the sp2-bonded carbon atoms to graphite-like aromatic
planes was invoked in order to explain, among other properties, the small band
gaps of a-C and a-C:H. For a band gap of, say, 1.2eV, typical for hard a-C:H,
clusters involving up to 25 benzene rings are required according to this assumption

[40]. Modern atomistic simulations of the detailed electronic structure of a-C and a-
C :H, however, point towards much smaller, mainly chainlike clusters, containing
only 2-10sp2-bonded atoms. The small band gap is a result of the distortion of
these structural elements, as discussed in more detail in [55]. In any case, the fact
that ta-C films can be prepared, with sp3 carbon fractions in excess of 80% and vir-
tually no hydrogen, demonstrates, that the range of accessible sp3 contents extends
far beyond the limits suggested by the considerations of network connectivity.
The highly cross-linked, dense microstructure of amorphous carbon films, grown
under the assistance of energetic particles, becomes also evident from diffusion and
permeation experiments. As already pointed out by Angus and Jansen, the fact, that
even after several years a virtually unchanged content of Ar atoms, entrapped
during plasma or low-energy ion beam deposition from Ar-containing gas phases,
can be detected in a-C:H films, shows that the Ar diffusion coefficient, at least in
certain portions of the film, must be lower than lo-**cm2 s f l . In conventional poly-
mers, the corresponding figures are in the lop9 to lop6cm2s-' region [54]. Using
depth profiling of deuterium by SIMS after exposition of a-C:H samples, grown
by RF plasma deposition from acetylene, to D 2 0 vapor (160h at 85°C and 80-
85% relative humidity), Klages et al. [56] showed, that a virtually hermetical sealing
against humidity can be achieved: In films of 120 nm to 1.9 pm thickness, deposited
with a substrate bias of at least -5OV, D incorporation could be detected only
within the outer 20 to 50 nm thick regions. Only in a 150 nm thick film grown at a
bias voltage of -1OV, a diffusion of D through the whole film thickness was
observed. Conventional organic polymers and coatings as used for corrosion
protection, on the other hand, are easily permeable to water, oxygen and ions.
3.3.2 Metal-containing Amorphous Carbon Films

Structural investigations of metal-containing amorphous hydrogenated carbon films,
prepared by co-evaporation or cosputtering of a polymer and a metal, plasma poly-
merization of a metal-containing monomer, or reactive sputtering of a metal or metal
carbide in a hydrocarbon atmosphere, have generally pointed to a granular structure
of the deposits, containing the metal or (in case of carbide formers) a metal carbide as
nanocrystalline segregates in an amorphous matrix. A review of the relevant literature
until 1989 was published by Klages and Memming [32].
Structural information came from transmission electron micrographs TEM and
mainly from X-ray diffraction (XRD) measurements, revealing the crystallographic
phases formed and (by application of the Scherrer equation D = h/r cos 8) allowing
an estimation of the average size of the particles D from the X-ray wavelength h and
the width r of a reflection on the 28 scale of scattering angles. Reactively sputtered
MeC :H films containing not too high concentrations of refractory metals Me = Ti,
V, W, Mo, Nb were found to contain carbide particles MeC with sizes between
about 1 and 10 nm typically. For iron-containing films the presence of the carbide
Fe5C2 in superparamagnetic segregated nanocrystals was inferred from magnetic
and Moljbauer spectroscopic measurements.
3.3 Microstructure und Bonding of Diamond-like Carbon 635
The microstructure of the amorphous a-C: H matrix surrounding the metal or

metal carbide particles in MeC:H, especially its sp3/sp2 ratio, is still largely
unknown. The application of IR absorption spectroscopy, for example, is possible
only for very small metal contents, owing to the rapidly increasing optical absorp-
tion induced by the metal. A dominance of hydrogen bonding to sp3 carbon atoms
was indicated, for example, in Ta-C: H films containing 4 at-% Ta. The systematic
shift of the G-line in Raman spectra of W-C: H towards larger wavenumbers (1 528
. . . 1564cm-') with increasing W contents (7 . . . 2 5 at-%) was used by Schiffman to
argue in favor of a decreasing sp3 content [57]. The hydrogen concentration
generally decreases with growing metal content: In Ta-C : H, a H/C ratio of about
0.48 is found by Rutherford backscattering analysis (RBS) for a virtually Ta-free
film grown from a methane-containing atmosphere, declining to 0.22 in films with
a Ta/C atomic ratio in excess of 0.66 [32]. Using secondary ion mass spectrometry,
similar results have been obtained for other metals, too [57, 581.
In recent years the application of small angle X-ray scattering (SAXS) working
with synchrotron radiation has turned out to be valuable tool for an analysis of
the nanostructure of MeC:H films [57, 59-61]. SAXS allows an investigations of
structural inhomogeneities of solids, resulting in a modulation of the local electron
density, on length scales in the 1-100nm region. Figure 3 shows SAXS curves as
obtained from measurements on reactively sputtered WC: H films with a varying
content of W [57]. In all cases investigated so far in detail (Me = Au, Pt, W, Fe)
the scattering intensity at large values of the scattering vector q (Porod region)
(typically q > 3.0nm-') decreases as qP4, indicating a relatively sharply defined,
smooth surface of the particles. In the medium q region (Guinier region, typically
10 < qnm < 3), a maximum or shoulder is usually observed, permitting a rough
estimate of the inter-particle distance d = 27c/q,,,. The frequently observed increase
of intensity towards the lowest q values used points to the existence of a relatively
small number of particles having a much larger diameter and therefore a significant
W-C:H
0,Ol 0,'
q
Figure 3. Small angle X-ray scattering (SAXS) curves for WC: H films with differing metal contents.
Continuous lines = simulation, see text. (From [57] by courtesy of K. I. Schiffmann).
volume fraction, among a majority of smaller ones. The continuous lines in Figure 3
are simulated curves, based on a model of hard homogeneous spheres in a
homogeneous matrix. In order to represent the particle sizes, a combination of
two log-normal distributions was used, one main distribution representing the
majority of particles with a relatively small width, plus an additional broad back-
ground distribution representing few, much larger particles. Particle radii and dis-
tances ranging from 0.6 to 0.8nm and 2 to 3nm, resp. could be deduced from
these investigations for W-C:H films with W contents between 7 and 25 at-%.
Metal and carbide particle distances and radii in MeC:H films have also been
measured by applying scanning tunnel microscopy (STM) imaging. In order to
take into account tip convolution effects, leading to an appearant particle enlarge-
ment and to particle hiding, statistical methods were proposed by Schiffmann et d.,
using an off-line tip radius determination by analysing appearant particle radii and
a Monte Carlo model to compute fractions of hidden particles. A very good
agreement was generally observed for radii and distances as measured by STM,
TEM, SAXS and (for the radii only) XRD [57, 601.
The anisotropies in two-dimensional SAXS measurements as observed by Fryda
[59] using sample tilting with respect to the incoming synchrotron radiation beam
are, according to TEM micrographs of film cross sections [57], not due to an
elongated shape of the segregated particles, but possibly to a chain-like ordering
along the growth direction. The particles themselves are usually nearly spherical
and do not exhibit a pronounced shape anisotropy along the growth direction.
The average distance d of the particles, at a given volume fraction directly related
to their average size, correlates quite closely with the melting point T, of the
corresponding bulk metal. The plot against T i 1 in Figure 4 is based on the evalua-
tion of the position of the maximum or shoulder in the Guinier region of SAXS
scattering curves for a number of MeC:H films [59] plus the data for Au from
[57]. Qualitatively, an increasing mobility of the atoms of lower melting metals
80
70
60
.a 50
\
TJ 40
30
20
10
2 3 4 5 6 7 8
T~
; I I O - K-?
~
Figure 4. Plot of the average center of mass distances d (in A) of segregated metal and carbide par-
ticles, respectively, in MeC: H films against the reciprocal melting points l/T, of the corresponding
bulk metals, according to data taken from [59] and, for Au, [57].
3.4 Physical Properties of DLC Filnis 637
during film growth and segregation of the metal or metal carbide phases can be
made responsible for their increasing sizes and distances.
3.4 Physical Properties of DLC Films

3.4. Electrical and Optical Properties
The oom temperature electrical conductivity of amorphous carbon varies over
more than 15 decades from around 1 to R-l ern-', depending on the
growth methods and growth parameters used [49, 621. A fairly exhaustive review,
including data, mostly measured over a range of temperatures, for samples prepared
by evaporation, sputtering, ionized carbon and hydrocarbon deposition, laser
evaporation, arc discharge deposition, and plasma deposition has been published
by Frauenheim et al. [63]. The temperature dependence of conductivity has fre-
quently been discussed in terms of Eq. ( 3 ) [64, 6.51:
o ( T )= ooexp{-(To/T)n}; n = I . . .1 /4 (3)
The value of n depends on the conduction mechanism: n = 1 represents conduction
by activation to extended states as in typical crystalline semiconductors, or by
hopping (i.e., phonon-assisted quantum-mechanical tunneling) between localized
states on nearest-neighbor sites, frequently found in highly doped crystalline and
in amorphous semiconductors at sufficiently high temperatures. For decreasing tem-
peratures, it will become increasingly difficult to find a nearest-neighbor site with a
suitable energy separation and it becomes more favorable for the electrons to make
larger hops to more distant sites, in order to find a site within a suitable energy
distance: The geometry of the most probable path will de end on the temperature,
the value of n becoming smaller than 1. Mott's famous TPc4 law will be obtained as
a limiting case for a constant DOS around the Fermi energy. In nonhydrogenated
amorphous carbon films with small sp3 content, the transition to n = 1 behavior
will happen only at relatively high temperatures, n < 1 has therefore been observed
up to 300-500K. For hydrogenated films, on the other hand, n = 1 prevails down
to about 200-300 K temperatures and only at lower temperatures a transition to
variable range hopping occurs [63].
Meyerson and Smith were the first to grow and investigate a-C:H films doped
with boron and phosphorus, by additions of diborane or phosphine to the hydro-
carbon gas used for deposition [66]. With a dopant concentration of lo%, the
room temperature conductivity increased by about five orders of magnitude to
lo-' R-' cm-l. However, as shown by Thiele et al., this effect ist not a true p- or
n-type doping in the conventional sense (i.e. a shift of the Fermi level towards the
valence or conduction band) but an increase of localized states near the Fermi
level, supporting a hopping conduction mechanism [67].
The mechanism of electrical conductivity in highly sp3-bonded, virtually hydrogen-
free ta-C, having the highest band gap among the dense amorphous carbon phases,
638 3 Diamond-likeCarbon Films
and the question of its dopability is presently under discussion. Data published for
undoped ta-C exhibit a large spread in room temperature conductivity, from to
10-12 0-1 cm-l
, even for quite similar sp3 fractions [39]. Amaratunga et al. were the
first to grow ta-C (filtered arc deposition) with additions of N, P, and B [68]. The
conclusions drawn in this paper and in subsequent publications from this group
have recently been critically reviewed by Hofsiiss [69] on the basis of their own
experimental results. According to his investigations, the conduction in undoped
ta-C is mainly due to hopping in band-tails states and (at higher electrical fields)
Poole-Frenkel excitation of electrons from traps into the conduction band.
Doping with N or B leads to an increase of conductivity by 3 4 orders of magnitude
at room temperature, presumably due to a generation of localized states related to
three-fold coordinated B or N sites, contributing to hopping as well as Poole-
Frenkel conduction. This view is supported by completely symmetrical I- V curves
measured for layered diode-like structures containing N- and B-doped material.
Electrical properties of MeC :H films, deposited by reactive sputtering from
metal or metal carbide targets in hydrocarbon atmosphere, have been described
elsewhere [32]. In case of Ta as a typical refractory, carbide-forming metal, the
electrical behavior can be divided into three categories, depending on the metal
content:
- a 'dielectric' region with thermally activated conductivity according to Eq. (3) at
less than about 13 at-% Ta,
- a transition region with metallic behavior (i.e. residual conductivity at vanishing
temperature) but still a positive or zero slope of log(o)-T curves and
o > lo2R-' cm-' at medium Ta contents (14 and 22 at-% Ta) and
- a metallic region with with negatively sloped curves log(o)-T and
o > lo3R-' cm-' for 44 and 94 at-% Ta.
In contrast to Au- and Cu-C:H, for example, showing a percolation-type sudden
increase of room temperature conductivity at a medium metal content of several
10 vol-%, the o ( R T ) values in Ta-C:H films tend to increase continuously by
about six orders of magnitude right from the lowest concentrations in nearly
metal free films up to less than 10 at-% Ta. It is only at low temperatures, that
an insulator/metal transition is observable as a four orders of magnitude jump in
o(3 K) between 11.5 and 14.3 at-%. A detailed discussion of the electrical properties
of MeC :H films is given elsewhere [70].
Similar to the electrical conductivities optical properties like absorption
coefficients and refractive indices are critically dependent on the preparation
conditions used to grow a-C or a-C :H. Except for absorption coefficients below
about lop4 cm-', the photon energy E dependence of the absorption coefficient M
can generally be described by the Tauc relation (4) [71]:
EM)"^ = G ( E - Eo) (4)

The slope G varies relatively weakly among different kinds of amorphous carbon films,
the optical gap, however, depends uite strongly on the H-content and sp3/sp2ratio of
9
the films, owing to the fact, that sp -related 71: states (see Fig. 2) are mainly responsible
3.4 Physical Properties of DLC Films 639
for absorptions in and near to the visible region. Typical values for a-C and a-C: H are
shown in Table 2. While an upper limit of 1.5eV is still quoted in this survey for ta-C,
prepared by mass selected ion beam deposition, considerably higher values approach-
ing or even exceeding 3 eV seem now feasible according to the review [39]. Thus, the
very hard and wear resistant hydrogen-free ta-C films might gain importance for opti-
cal applications in the visible region.
Refractive indices for ta-C films are typically close to the value of diamond
(n = 2.4) [39]. Incorporation of hydrogen lowers n to values between about 1.8
and 2.3 in a-C:H. In plasma deposition of these films, hydrocarbon pressure P
and substrate bias voltage UB are the most important growth parameters allowing
a tuning of n. Dense films with high refractive indices are favored by low P and high
UB,To a first approximation, holding for n up to 2, constant refractive indices can be
achieved for a constant ratio U B / P[62].
3.4.2 Mechanical Properties

For tribological applications of diamondlike carbon films, hardness, H, and stiff-
ness, measured as the Young modulus, E, are key physical properties. The hardness
of a material is its resistance against permanent geometric deformation under
mechanical loading. In indentation testing, it is equivalent or proportional to the
average pressure under a sharp indenter and can be calculated from the applied pres-
sure divided through the (projected) contact area between the indenter and the
material tested. For hardness measurements on thin films of less than 1 to several
pm thickness, conventional microhardness measurements with microsocopic eva-
luation are difficult due to the small sizes of the indents. Therefore nanoindentation
techniques working with a simultaneous registration of force and indentation depth
with resolutions of a few pN and nm, resp. have been introduced [72]. Aside from
the hardness, Young modulus can be estimated, using these methods, from the
slope of the unloading curve [73]. Jiang et al. applied the method for an investigation
of a-C:H films grown by R F plasma deposition from methane and acetylene [74,
751. Hardness values between 5 and 20 GPa were typically found, correlating closely
with Young moduli of the films, H z 0.1 E . In MeC: H coatings, the metal content is
the prime factor determining H a n d E [76,35]: For WC : H and Ta-C : H, H increases
from around 20 GPa for films with virtually zero or only a few vol-% of MeC pre-
cipitation to about 40 GPa at about 50 vol-%. Over the same range of metal con-
tents E grows from 100 to 200 GPa ( H I E = 0.2).
Even higher H and E values, approaching those of diamond (z100-120 and
1 100 GPa) are characteristic for hydrogen-free, highly tetrahedrally coordinated
ta-C films as to be obtained by ion beam or vacuum arc deposition, owing to the
high average coordination numbers r,, 2 3.6 in these networks [39].
Based on their characteristic chemical bonding and microstructure, as well as
mechanical properties, a gross division of technically important materials into the
three categories metals, polymers and ceramics is frequently made. It is tempting
to compare the spectrum of properties of DLC films to those of these main material
groups. In [33] this has already been done for the narrower class of tribologically
interesting MeC: H films with low metal contents up to about 20 at-%. However, the
essential results basically hold also true for DLC coatings in general. According to
these considerations, the exceptional mechanical-tribological properties of DLC can
be attributed to
~ hardness values (~10-100GPa) similar or even exceeding those of ceramic
inorganic materials (10-20 GPa) and by far larger than for metals (< 8 GPa),
- elastic moduli (~100-1000GPa) similar or exceeding metals and ceramics
(< 600 GPa), combined with
- elasticities as measured by H / E ratios (0.1-0.2) which are throughout larger than
for metals (<0.03) and ceramics (<0.06) and approach the lower limit of a range
which is more typical for organic polymers (0.2-0.5), and
- for hydrogenated film surfaces, surface energies (30-50 x N/m) in a region
typical for polymers and much less than for metals (300-2000 x N/m) and
ceramics (200-1300 x lop3N/m) (see [33] for references to the sources used).
Friction coefficients, being not a physical property in the strict sense, but depending
on the contact partners and many other external parameters, will be reported in the
next section. The combination of low friction coefficients, typically 5 0.2, with
mechanical and surface properties which are partly polymer-like (elasticity, surface
energy) and partly ceramic- or metal-like (hardness, Young modulus) qualitatively
explains the outstanding position of DLC films as low-friction, highly wear-resistant
coatings on which most of their present applications are based [77].
3.5 Applications of DLC Films

Applications of DLC films in several fields like optics, magnetic disks, tribology and
machine elements or biomedical purposes have been reported in the literature. A
review on this topic was given by Matthews and Eskildsen [78].
The dominating field of application at present is that of tribological coatings on
machine parts and tools. This shall be discussed in more detail. However, for all
applications of the coatings a sufficient adhesion is required. Therefore this question
will be treated first in the next section.
3.5.1 Adhesion of DLC Films

The high compressive stresses in hard a-C: H and ta-C films (up to 10 GPa, depend-
ing on the deposition conditions, especially the bias voltage) [79] frequently lead to
film delamination and thus limit the attainable film thickness. For many years thick-
nesses of more than 1-2 ym could not be realized on technically relevant substrates
like steel, hardmetal or glass. Considerable improvements were achieved by
applying intermediate layers consisting of aSi:H [SO], metals (Al, Ti, Ni) [81, 821,
or multilayers (Ti-TiCN-Tic) [83], for example. For the characterization of
3.5 Applicutions OfDLC Films 641
. =.
b
0
0
0 o without
0 ' I I
0 2 4 6 8 1 0 1 2 1 4
6[ P ~ I
Figure 5. Critical load L, against a-C: H film thickness df for deposition with Al- and Ti- interlayers
or without interlayers, respectively, on polished steel substrates.
adhesion the critical load, L,, or the crack pattern caused by a Rockwell indentor is
frequently used. The critical load is a measure determined by scratch tests [84],
utilizing a diamond tip moving over the film with continuously increasing normal
load. L, is the load where an intensive acustic emission occurs, the friction coefficient
between tip and sample drastically increases or a failure of the film can be observed
by a microscope. All three criterions are in use. Therefore a comparison of data from
different laboratories is difficult. Figure 5 shows the effect of aluminum and titanium
interlayers on the critical load as a function of the a-C:H film thickness df.L, was
determined with the second and third criteria. The values determined with both
methods were nearly identical. These films were prepared by R F plasma deposition
from acetylene after magnetron sputtering of the metal interlayers (thickness about
0.5 pm) onto polished steel substrates in the same coating machine. In particular in
the range df > 3 pm a pronounced improvement of the adhesion was found. The
hardness of these coatings was about 25GPa. Even films thicker than lOpm
having excellent adhesion can be achieved by this method. It should be noted that
the effect of interlayers is still more pronounced if substrates with rougher surfaces,
like ground substrates are to be coated.
However, the reliable and reproducible preparation of well-adhering a-C :H films
is commonly not a simple task and requires a good control of several process param-
eters. Traces of oxygen in the films, to give just one example, can cause a drastic
deterioration of the film adhesion.
On the other hand, there are very reliable industrial processes for deposition of
MeC: H coatings. MeC: H films have lower stresses than a-C: H (< 1 GPa) [34, 851
Therefore their tendency to delaminate is considerably reduced. Furthermore,
intermediate layers can be prepared very easily using non-reactive sputtering from
-
ap 120
I
140[
E 100
a
L
80-
Q
'=m
>
- 60-
PI
K
40 -
"
Hard steel Ti N MeC D LC
metal (100Cr6)
Figure 6. Abrasive wear of DLC coatings compared to other coating and substrate materials. As
abrasive medium S i c with a particle dimension of 12 Fm was used.
the metal target (see Section 3.2.3.), before the MeC: H deposition process is started.
3.5.2 Tribology of DLC Coatings

After satisfying solutions for the adhesion problem had been developed, DLC has
become very interesting for tribological applications owing to the very favorable
combination of low abrasive and adhesive wear with low friction coefficients. The
following discussion concerns typical data on wear and friction of DLC films, but
has to be taken into account that these properties can be influenced by the deposi-
tion conditions like bias voltage, gas pressure etc. Figure 6 shows the rates of abra-
sive wear for different types of hard coatings and reference materials. MeC: H and
still more pronounced a-C: H exhibit excellent values. In Fig. 7 adhesive wear rates,
measured in a pin on disc experiment, are presented. As pin a ball of bearing steel
(100Cr6) was used. Figure 7a shows the results for MeC :H and TIN coated discs as
well as for a steel disc, Fig. 7b the results for the 100Cr6 counterpart (pin). In the
case of MeC:H the extremely low wear of the counterpart is an evidence of the
low material transfer between steel ball and coating. It should be mentioned that
the wear rates of MeC:H films depend on the metal content. However, for different
metals [32, 851 and both for R F and DC magnetron sputter deposition the minimum
values of abrasive wear were determined to be in the concentration range of 0.1 to
0.2 (metal to carbon ratio).
Friction coefficients, p, against steel (steel 100Cr6) are shown in Fig. 8. While
the friction coefficient of titanium nitride strongly increases with growing load
due to a material transfer of steel onto the coating, the p values of WC:H (Me/
C ratios in the above mentioned range) even decrease with increasing load. The
friction coefficients of metal-free a-C : H are very close to those of MeC :H. Very
low friction coefficients (< 0.2 against 100Cr6) were also determined for laser-arc
3.S Applications of DLC Films 643
wear of coatings wear of steel

and steel disk (a) counterpart (b)
1ooc
0
w
0 looor w
.-: 1oc
s-
?!E
4-
gF
L
1c
V
:g
.->'
aJ
v)
1
r
a
-0
0.1 AI E
Y
r" Y2
Figure 7. Adhesive wear measured in a ball on disk arrangement: (a) MeC: H and TiN coated discs
and a disc of ball bearing steel 100Cr6 and (b) wear of the counterpart (100Cr6 ball). Measurements
were performed under ambient conditions.
deposited films [86].

The friction coefficients of DLC films can considerably be lowered further by
incorporation of silicon. Minimum p values were reported for films with 10-
20 atom-% Si [87, 881. Another important factor influencing the friction properties
is the humidity. For a-C:H and W C : H (W/C < 0.4) films the friction coefficients
o'2
01
0.01
I 0.1 1
I
10
Normal load "1
Figure 8. Friction coefficients against normal load for MeC:H, TiN, TiAlN and steel measured
against ball bearing steel under ambient conditions (up to about 50% relative humidity).
drastically declined from p > 0.2 to p < 0.05 with decreasing relative humidity
(< 1%) [89,90]. For silicon-containing films very low p values (<0.1) were measured
even up to relative humidities of 75% [88, 911. A possible reason for the different
friction behavior could be a material transfer from the film to the steel ball (decreas-
ing p) or in the opposite direction (increasing p) [90]. In the case of silicon-
containing films reactions of Si with the atmosphere were assumed. The generated
silicon ceramics are known to have extremely low friction coefficients in water
[911. However, there are still several open questions concerning the friction
mechanisms of DLC films. It should be noted here that any published values of
friction coefficients must be considered with caution because measuring conditions
like humidity, loads, material of counterparts or sliding distance, which can have a
strong influence on the result, can be extremely different [78].
3.5.3 Tribological Applications

DLC coatings have been applied successfully for industrially used components and
tools to minimize wear and friction and in some cases to avoid the use of lubricants.
For example gears, ball bearing races, valves, knives and other parts have been
coated with WC:H, see [78]. Roth et al. [17] coated blades for fibreboard material
and stamping tools with a-C: H. There is a great application potential for DLC films
in the field of forming tools, such as for aluminum- and titanium-based alloys [92].
With coated tools the cost intensive and ecologically harmful employment of
forming lubricants can be avoided or at least diminished.
By incorporation of certain elements the wetting behavior of DLC can be
modified considerably. The surface tension of a-C :H can be decreased by additions
of fluorine or silicon and oxygen, while it is enlarged by the incorporation of boron
and nitrogen [81]. Low surface tension coatings have been applied successfully to
reduce the adhesion of organic material on tools used for polymer fabrication.
Another interesting application is the improvement of the heat transfer in steam
condensation heat exchangers, thanks to the fact, that the water vapor does not
form a heat-insulating water film on the coated heat exchanger surface, but it
condenses in form of droplets which easily run down the vertically arranged heat
exchanger, exposing a virtually water-free surface.
3.5.4 Other Applications

Due to their chemical inertness, high abrasion resistance, low friction and good bio-
compatibility DLC films are very promising candidates for biomedical applications.
For example, femoral heads of hip protheses have been coated successfully with
DLC to reduce the production of small wear particles, which can lead to reactions
on a cellular level. Furthermore DLC was shown to be a potential material for
artificial heart valves where the films must be non-thrombogenic and must have a
long-time stability in contact with blood [93]. Mitura et al. [94] investigated DLC
References 645
films prepared by ion beam decomposition, R F plasma and pulsed arc deposition,
for coating implants. In spite of different film properties all three materials turned
out as very suitable.
Ion beam deposited DLC films are used to increase the abrasion and chemical
resistance of sunglass lenses, of float-glass windows in supermarket laser scanners
and of infrared windows [95].Very thin amorphous carbon films are used to protect
magnetic disks and heads. Recently it was found that especially nitrogen containing
carbon films (CN,) have an excellent wear resistance (about four-fold value of a-C)
[96]. One reason for this difference is presumably the different roughness of the films.
Using an atomic force microscope (AFM) roughness values r.m.s. of 0.25 nm and
0.63 nm, respectively, were measured for CN, and for a-C.
For the application of coatings, also cost and benefit aspects must be considered. It
shall be noted here that Matthews and Eskildsen [78]discussed this topic in detail for
four different DLC production methods taking into account prices for deposition
equipments to be obtained in the UK.
References
1. L. Holland and S. M. Ojha, Thin Solid Films 1976,38,L17-LI9.
2. D.S. Whitmell and R. Williamson, Thin Solid Films 1976,35,255-261.
3. A. Grill and V. V. Patel, Diamond Films Technol. 1992,4,219-233.
4. L.Holland, J . Vac. Sci. Technol. 1977,14, 5-15.
5. K. Bewilogua and D. Wagner, Vacuum 1991,42,473476.
6. Y.Catherine and P. Couderc, Thin Solid Films 1986, 144, 265-280.
7. C.Weissmantel, K. Bewilogua, K. Breuer, D. Dietrich, U. Ebersbach, H.-J. Erler, B. Rau, and
G. Reisse, Thin Solid Films 1982,96, 3144.
8. C. Weissmantel, in Thin Films from Free Atoms and Particles, J. Klabunde (Ed.), Academic,
New York, 1985,pp. 153-201.
9. T. Namba and Y. Mori, J . Vac. Sci. Technol. 1983,Al, 23-27
10. M. W. Kroesen, D. C. Schram, and M. J. F. van de Sande, Plasma Chem. Plasma Proc. 1990,
10,49-69.
11. L.Martinu, A. Raveh, A. Domingue, L.Bertrand, J. E. Klemberg-Sapieha, S. C. Gujrathi, and
M. R. Wertheimer, Thin Solid Films 1992,208,4247.
12. B. Chapman, Glow Discharge Processes, Wiley, New York, 1980,Chapter 5.
13. A. Bubenzer, B. Dischler, G. Brandt, and P. Koidl, J . Appl. Phys. 1993,54,45904595.
14. J. Robertson, Surf. Coat. Technol. 1992,50, 185-203.
15. W. Moller, Appl. Phys. A 1993,56, 527-546.
16. H. Kersten and G. M. W. Kroesen, J . Vac. Sci. Technol. 1990,A8, 3842.
17. D. Roth, B. Rau, S.Roth, J. Mai, and K.-H. Dittrich, Surt Coat. Technol. 1995,7475,637-641.
18. S.Aisenberg and R. W. Chabot, J . Appl. Phys. 1971,42,2953-2958.
19. Y. Lifshitz, Diamond Re/. Muter. 1996,5, 388400.
20. J. J. Cuomo, D. L. Pappas, J. Bruley, J. P. Doyle, and K. L. Saenger, J . Appl. Phys. 1991,70,
1706171 1.
21. H. Hofsass, H. Binder, T. Klump, and E. Recknagel, Diamond Rel. Muter. 1993,3, 137-142.
22. N.Savvides, Thin Solid Films 1988, 163, 13-32.
23. N . Savvides and B. Window, J . Vac. Sci. Technol. 1986,A4, 504-508.
24. A. Anders, S. Anders, and I. G. Brown, Plasma Sources Sci. Technol. 1995,4, 1-12.
25. H.-J. Scheibe, B. Schultrich, and D. Drescher, Surf. Coat. Technol. 1995,74-75, 813-818.
26. H.-J. Scheibe, D. Drescher, B. Schultrich, M . Falz, G. Leonhardt, and R. Wilberg, Surf: Coat.
Technol. 1996, 85,209-214.
27. A. A. Voevodin and M. S. Donley, Su$ Coat. Technol. 1996,82, 199-213.
28. F. Rossi and B. Andre, Jpn. J. Appl. Phys. 1992, 31, 872 -879.
29. J. Robertson, Diamond Rel. Muter. 1993, 2, 984-989.
30. H. Dimigen and H. Hubsch, Philips Tech.Rev. 1983/84,41, 186-197.
31. H. Dimigen and H. Hiibsch, R. Memming, Appl. Phys. Lett. 1987,50, 10561058.
32. C.-P. Klages and R. Memming, Muter. Sci. Forum, 1989, 52-53, 609-644.
33. H. Dimigen and C.-P.Klages, Surf. Coat. Technol. 1991,49, 543-547.
34. E. Bergmann and J. Vogel, J . Vac. Sci. Technol. 1987, A5, 70-74.
35. K.Bewilogua and H. Dimigen, Surf. Coat. Technol. 1993,61, 144-150.
36. D. Hofmann, H. Schuessler, K. Bewilogua, H. Hubsch and J. Lemke, SwF Coat. Technol. 1995,
73, 137-141.
37. J. T. Harnack and C. Benndorf, Diamond Rel. Muter. 1992, 1, 301-306.
38. F. L. Pilar, Elementary Quantum Chemistry, MacGraw-Hill, New York, 1968.
39. Y. Lifshitz, in The Physics of Diamond, A. Paoletti and A. Tucciarone (Eds), 10s Press, Amster-
dam, 1997, p. 209.
40. J. Robertson, in Diamond and Diamond-like Films and Coatings, R. E. Clawing et al. (Eds),
Plenum, New York, 1991, p. 331.
41. S. Kaplan, F. Jansen, and M. Machonkin, Appl. Phys. Lett. 1985, 47, 750-753.
42. C. Weissmantel, K. Bewilogua, D. Dietrich, H.-J. Erler, H.-J. Hinneberg, S. Klose, W. Nowick,
and G. Reisse, Thin Solid Films 1980, 72, 19-31.
43. S. Miihling, K. Bewilogua, and K. Breuer, Thin Solid Films 1990, 187, 65-75.
44. J. Fink, Th. Muller-Heinzerling, J. Pfluger, B. Scheerer, B. Dischler, P. Koidl, A. Bubenzer, and
R. E. Sah, Phys. Rev. B 1984, 30,47134718.
45. J. Robertson, Diamond Rel. Mater. 1992, 1, 397406.
46. D. R. McKenzie, D. Muller, B. A. Pailthorpe, Z. H. Wang, E. Kravtchinskaia. D. Segal, P. B.
Lukins, P. D. Swift, P. J. Martin, G. Amaratunga, P. H. Gaskell, and A. Saeed, Diamond Rel.
Muter. 1991, 1, 51-59.
47. H. Hofsass, H. Feldermann, R. Merck, M. Sebastian, and C. Ronning, Appl. Phys. A , 1988,
A66, 153-181.
48. B. Dischler, in Proceedings of E-MRS Symposium on Amorphous Hydrogenated Carbon, Sympo-
sium C, Strasbourg, France, Les Editions de Physique, Paris, 1987, XVII, pp. 189-201.
49. J. Robertson, Adv. Phys. 1986,35, 317-374.
50. J. C. Angus, in Proceedings of E-MRS Symposium on Amorphous Hydrogenated Carbon,
Symposium C , Strasbourg, France, Les Editions de Physique, Paris, 1987, XVII, pp. 179-188.
51. J. C. Phillips, Phys. Rev. Lett. 1979, 42, 153-181.
52. M. F. Thorpe, J. Non-Cryst. Sol. 1983, 57, 355-370.
53. G. H. Dohler, R. Dandoloff, and H. Bilz, J. Non-Cryst. Sol. 1980, 42, 87 96.
54. J. C. Angus, F. Jansen, J. Vac. Sci. Technol. 1988, A6, 1778-1782.
55. J. Robertson, Diamond Rel. Mater. 1997, 6, 212-218.
56. C.-P. Klages, A. Dietz, T. Hoing, R. Thyen, A. Weber, P. Willich, Surf. Coat. Technol. 1996,80,
121-128.
57. K. I. Schiffmann, Thesis, University of Hamburg, published as Fraunhofer-IST-Berichte aus
Forschung und Entwicklung, Fraunhofer IRB Verlag, Stuttgart, 1997, Nr. 1.
58. P. Willich and R. Bethke, in Secondarji Ion Mass Spectroscopy ( S I M S X i , A. Benninghoven,
B. Hagenhoff and H. W. Werner, Wiley, New York, 1996.
59. M. Fryda, Fortschritt-Berichte VDI, Reihe 5 , VDI Verlag, Dusseldorf, 1993, 303, p. 20.
60. K. I. Schiffmann, M. Fryda, G. Goerigk, R. Lauer, and P. Hinze, Ultramicroscopy 1996, 66,
183-192.
61. K. I. Schiffmann, M. Fryda, and G. Goerigk, Mikrochinz. Acta 1997, 125, 107-1 13.
62. J. C. Angus, P. Koidl, and S. Domitz, in Plasma Deposited Thin Films, J. Mort and F. Jansen
(Eds), CRC, Boca Raton, FL, 1986, p. 89.
63. Th. Frauenheim, U. Stephan, K. Bewilogua, F. Jungnickel, P. Blaudeck, and E. Fromm, Thin
Solid Films 1989, 182, 63-78.
64. H. Overhof, P. Thomas, Electronic Transport in Hydrogenated Amorphous Semiconductors,
References 647
Springer Tracts in Modern Physics, Vol. 144, G. Hohler (Ed.), Springer Verlag, Berlin, 1989, p.
10.
65. R. Zallen, The Physics ofAmorphous Solids, Wiley, New York, 1983, p. 252.
66. B. Meyerson and F. W. Smith, Solid State Commun. 1982, 41, 23-27.
67. J. U. Thiele, B. Rubarth, P. Hammer, A. Helmbold, B. Kessler, K. Rohwer, and D. Meissner,
Diamond Re/. Muter. 1994, 3, 1103-1 106.
68. G. A. J. Amaratunga, D. E. Segal, and D. R. McKenzie, Appl. Phys. Lett. 1991, 59, 69-71.
69. H. Hofsass, in Proc. 1st Int. Specialist Meeting on Amorphous Carbon ( S M A C '97).
Amorphous Carbon: State of the Art, S.R.P. Silva et al., (Eds.), Singapore: World Scientific,
1998, p. 296&310.
70. H. Koberle, PhD Thesis, University of Hamburg, 1989.
71. J. Tauc, R. Grigorovici, and A. Vancu, Phys. Stat. Sol. 1966, 15, 627-637.
72. P. Wierenga, A. J. J. Franken, J . Appl. Phys. 1984, 55, 4244-4248.
73. J. L. Loubet, J. M. Georges, 0. Marchesini, and G. Meille, J . Tribol. 1984, 106, 4 3 4 8 .
74. X. Jiang, K. Reichelt, and B. Stritzker, J . Appl. Phys. 1989, 66, 5805-5808.
75. X. Jiang, K. Reichelt, and B. Stritzker, J . Appl. Phys. 1990, 68, 1018-1022.
76. M. Fryda, K. Taube, and C.-P. Klages, Vacuum 1990,41, 1291-1293
77. E. Rabinowicz, Friction and Wear of Materials, Elsevier, Amsterdam, 1987.
78. A. Matthews and S. S. Eskildsen, Diamond Re/. Muter. 1994, 3, 902-91 1.
79. M. Tamor, in Proc. Third Int. Conf on Applications of Diamond Films and Related Materials, A.
Feldman et a/. (Eds), NIST Special Publications 885, 1995, p.691.
80. R. Butter, M. Allen, L. Chandra, A. H. Lettington. and N. Rushton, Diamond Re/. Muter. 1995,
4, 857-861.
81. M. Grischke, K. Bewilogua, K. Trojan, and H. Dimigen, Surf: Coat. Technol. 1995,7475,739-
745.
82. H. Ronkainen, J. Vihersalo, S. Varjus, R. Zilliacus, U. Ehrnsten, and P. Nenonen, Surf. Coat.
Trchnol. 1997, 90, 190-196.
83. D. P. Monaghan, D. G. Teer, P. A. Logan, I. Efeoglu, and R. D. Arnell, Surf. Coat. Technol.
1993, 60, 525-530.
84. P. J. Burnett and D. S. Rickerby, Thin Solid Films 1988, 157, 233-254.
85. M. Grischke, K. Bewilogua, and H. Dimigen, Mater. Manuf. Proc. 1993, 8, 407417.
86. H.-J. Scheibe and D. Klaffke, Surf. Coat. Technol. 1993, 57, 111-1 15.
87. S. Miyake, Surf: Coat. Technol. 1992, 5455,563-569.
88. K. Oguri and T. Arai, Thin Solid Films 1992, 208, 158-160.
89. K. Enke, H. Dimigen, and H. Hiibsch, Appl. Phys. Lett. 1980,36,291-292.
90. R. Memming, H. J. Tolle, and P. E. Wierenga, Thin Solid Films 1986, 143, 3 1 4 1 .
91. K. Oguri and T. Arai, J . Mater. Res. 1992, 7, 1313-1316.
92. K. Taube, M. Grischke, and K. Bewilogua, Surf. Coat. Technol. 1994, 68/69, 662-668.
93. R. S. Butter and A. H. Lettington, Proc. Third Int. Con6 on Applications of Diamond Films and
Related Muterials, A. Feldman et aI. (Eds), NIST Special Publications 885, 1995, p.683.
94. S. Mitura, P. Niedzielski, D. Jachowicz, M. Langer, J. Marciniak, A.Stanishevsky, E. Tochitsky,
P. Louda, P. Couvrat, M. Denis, and P. Lourdin, Diamond Re/. Muter. 1996, 5, 1185-1 188.
95. F. M. Kimock and B. J. Knapp, Surf. Coat. Technol. 1993,56, 273-279.
96. D. Li, E. Cutiongco, Y.-W. Chung, M.-S. Wong, and W. D. Sproul, Diamond Films Technol.
1995,5, 261-272.
4 Ceramics Based on Alumina: Increasing the

Hardness for Tool Applications
A. Krell
4.1 Recent Trends in the Application of Ceramic Tool

Materials
New trends in the development and application of hard ceramic grinding and
cutting materials will be discussed in this Chapter. For both groups of tools,
modern technical demands drive the development of submicrometer microstruc-
tures that exhibit significantly increased hardness and reliability. Manufacturing
approaches and resulting properties will be described for both advanced single
phase sintered alumina materials and for composite ceramics.
The fundamental difference between grinding and cutting tools is the existence of
a geometrically well-defined cutting edge for the latter group, whereas grinding is
performed by a plurality of geometrically not defined edges and tips. Most grinding
grits are inorganic nonmetallic materials (diamond, cubic boron nitride cBN, fused
or sintered ceramic grinding materials based on alumina A1203 or silicon carbide
Sic). For cutting wheels the share of these materials in large-scale manufacturing
is reported to be more or less constant at only l4-l6YO, compared with about
80% for hard metals 111. Table 1 reviews the most important ceramic tool materials
that are commercially available for the machining of metals, but this contribution
will not discuss materials with limited applications (e.g. garnet grits for grinding
wood and glass) or with super-hard characteristics (diamond or cBN).
In recent years, the intensity of research intended to find basically new solutions
was rather different for two main sectors. For grinding metals there was an
avalanche of patent applications in the early eighties associated with the develop-
ment of the first polycrystalline (sintered) grits, and this stream of innovations
continues now. Here, among the great number of new applications there is only a
small percentage of ‘speculative’ inventions concerning materials with unsure
scientific and technical prospects (e.g. an invention claiming diamond reinforced
alumina without describing the sintered dense body, thus without determination
of technically relevant properties). On the contrary, there is a remarkably large
number of inventions which are regarded as so important that an international
(European or PCT (Patent Cooperation Treaty)) application is filed without a
foregoing national priority. The similarly large number of oppositions on the one
hand and of reproaches for the violation of claims demonstrate that such patent
applications describe commercially highly sensitive subjects. Most of the recent
inventions concern the technology of materials and tools based on polycrystalline
sintered corundum (a-A1203),but there are also propositions to manufacture Sic
grits with a polycrystalline microstructure (where the importance of S i c grits is
4.1 Recent Trends in the Application of Ceramic Tool Materials 649
Table 1. Commercially available tool materials for machining metals. HVl (testing load 1 kg) and
HVlO (10 kg load) hardness values are given for grinding and cutting materials to avoid the misuse
of microhardness data
Vickers hardness Fracture toughness Bending strength

(GPa) KIC(MPam’”) (MPa)
Ceramic grinding grits HV 1

Fused aluminas
& >, .
Brown corundum (ALO,: immrities) 17.0-18.5 1.O-2.5 -
White corundum (kl2O3) 19.0-20.5 2.0-2.2 -
Ruby (AI2O3:Cr) 18.5-20.5 2.4-3.1 -
AI2O3/ZrO2eutectic 15.5-17.5 4.0-5.0 -

Sintered corundum (sol/gel)
with nearly equiaxed crystal shape 14.0-22.5 2.94.1 -
with platelet-shaped crystals 18.7- 19.O 3.3-3.7 -

Sic 25.5-27.5 2.5-3 .O -
Cutting tool ceramics

A1203 HVlO
+ 4% ZrOz 18.O-20.0 3.6-4.5 500-600
+ 15% Zr02 17.0-1 9.0 5 .O-6.0 600-750
AI2O3+ Ti(C,N) [- 30%] 19.0-22.0 4.0-5.5 550-800
A1203+ Sicwhisker[- 20%] 17.O-20.0 6.0-8.0 650-950
Si3N, 15.0-18.0 6.5-9.0 750-1000
cBN cutting tools 27.040.0 3.5-6.6 600-800
Diamond (polycrystalline) > 4500 6.0-9 .O 750-1000
generally less compared with alumina because of its poor ability to machine ferrous
workpieces).
On the other hand, there has been some stagnation in the field of tools with
geometrically defined cutting edges after the development of advanced A1203/
ZrO, and A1203/TiC composites at the beginning and of tough Si3N4 cutting
tools and Sic-whisker reinforced grades at the end of the eighties (high-strength
tetragonal zirconia polycrystals cannot be used for machining metals). At least
until 1997 it was the position of most manufacturers that they can satisfy the current
requests of their customers with the available ceramics. Consequently, many labora-
tories are busy performing extended screening tests with commercial grades. A
somewhat special position has to be attributed to efforts intended to improve the
chemical resistance of Si3N4 by surface coatings. A very recent tendency is the
increasing interest in a new class of materials with submicrometer microstructures,
a trend which is discussed in Section 4.4.
The strong motivation for these developments in grinding materials is the real
possibility of new inventions able to surpass the present state of the art by 5&100%
(in terms of rate of material removal or maximum time of use of a tool), the chance
to find materials that redouble the productivity of the industrial grinding operation
and reduce substantially the time-out for the change of tools.
Significant recent changes in the market shares of the different kinds of cutting
ceramics shown in Table 1 should be noted. There has been an increase in the
use of ceramic cutting tools, attributed to increasing use of silicon nitride tools.
650 4 Ceramics Based on Alumina: Increasing the Hardness,for Tool Applications
Two different reasons exist. First, the requirement for high reliability in automated
manufacturing lines (without running a tool to the possible maximum time of use)
favors the application of tool materials with a high fracture toughness (like Si3N4)
tolerating, in the mean time, their lower hardness. Second, a shift in the ratio of
machined materials from steel towards globular and vermicular cast iron also
requires tougher tools like silicon nitride, especially when high feed rates are neces-
sary. A drawback of nitride tools is their rather high chemical reactivity with iron:
against steel and cast iron (like GG25), at feed rates of about 0.3 mm rev-' the wear
of Si3N4is usually higher than for A1203-basedceramics. Hence, even for machining
cast iron, nitride tools can be applied with advantage only when it is impossible to
use the less tough AI2O3/ZrO2or A1203/TiCcomposites due to the required high
feed rates. Attempts have been made to overcome this problem by TIN and/or
oxide coatings, but real success was obtained only in some very special cases.
There is another trend, which has not yet altered market shares, because the new qua-
lities are not commercially available: this is the substitution of certain grinding opera-
tions by highly precise cutting approaches, especially in machining very hard materials.
It is expected that turning may cut the machining time to 1/3 compared with grinding,
and is sometimes associated with significant savings of consumed energy, and even the
compromise of first turning with subsequent final grinding reduces the time to 2/3 com-
pared with an all-grinding process. Additional advantages are the less expensive recy-
cling of turning chips compared with the waste disposal of the grinding silt and
investment costs of the equipment which can be reduced by a half for precise turning
[ 1,2]. On the other hand, investment in new machine tools is imperative for precise cut-
ting of hard materials (the mechanical stability of the most machine tools used today in
the automotive industry is probably not sufficient for the new demands). More wide-
spread use of cutting rather than grinding is restricted by the limited lifetime of today's
cutting ceramics and the increase in workpiece roughness after rather short cutting
paths. Investigations by large-scale manufacturers indicate a lack of real alternatives
to cBN for turning hard workpieces with high feed rates, and drilling and milling is
the field of hard metal tools, but for precise finishing AI2O3/TiCcomposites exhibit
an advantage in the cost-performance ratio [2]. It is expected that such findings will sti-
mulate the development of significantlymore fine grained and harder cutting tool cera-
mics based on corundum. For example, it was in 1997 that a German company
launched a new submicron A1203/Ti(C,N)composite. Indeed, such A1203/Ti(C,N)
tools show the fewest problems when compared with other commercial ceramics
(which, however, all have not been developed to meet the demands of precise machining
of hard surfaces), but an analysis of the technical demands under thls special condition
points to better opportunities with other ceramics (cp. Section 4.4.4).
4.2 Technological Essentials for Producing Hard and

Strong Tool Ceramics
It is beyond the scope of this paragraph to give a review of ceramic technologies,
but the development of tools with an advanced technical performance needs an
4.2 Technological Essentialsfor Producing Hard and Strong Tool Ceramics 651
understanding of the required basic mechanical parameters, and of the real chances
new technologies offer for their improvement. Also, it must be emphasized that the
properties of different new tool materials can be compared only if the effect of their
different microstructures is not obscured by processing-related imperfections: the
investigation of new materials requires samples that are different in their micro-
structures but with similarly small frequency and size of defects. Much care has
to be taken to perform this recent research (the results of which are described
below) because for extremely fine-grained microstructures this demand is not
easy to meet.
4.2.1 Typical Defects in Ceramics Tool Materials:

The State of The Art
It is known that defects reduce the strength of brittle materials due to stress concen-
trations, which cannot relax by plastic deformation, as in metals. However, there is
also evidence that the flaw population has a similarly strong impact on the hardness.
The most promising way to develop new tool materials with improved hardness,
wear resistance, and reliability is to reduce the grain size in the sintered microstruc-
tures: an approach which requires the use of increasingly fine-grained raw materials,
be it within the framework of advanced powder technologies or of sol/gel or precur-
sor approaches.
Unfortunately, it is just this trend towards extremely fine-grained, nanoscale, raw
materials that often prevents the realization of the expected progress in the mechan-
ical properties, simply because with the reduced particle sizes there is an increased
risk of defects in the green microstructure.
4.2.1.1 Powder Processing

All ‘dry’ powder pressing technologies require the powder to be granulated before
compaction, otherwise the specific volume of non-agglomerated particles in or
beyond the micrometer range is so great that on compaction the decrease of
volume becomes excessive, causing extended elastic relaxation (and associated crack-
ing) on removing the load. The key is to produce granules (by spray drying or other
approaches) that are large and strong enough for technical handling but soft enough
to be completely destroyed on compaction to avoid relicts of the granules in the
pressed and finally sintered bodies (Fig. 1). Cold isostatic pressing can significantly
reduce the size and frequency of such defects (provided there is optimized slurry pre-
paration and granulation [3]), but it is impossible to avoid them entirely.
Liquid shaping is a promising alternative. Avoiding granulation, a slurry of a
dispersed powder is poured into a mold. A problem with liquid shaping is that
homogenization and pouring introduces air into the slurry (the more the more
fine-grained the powder). The complete elimination of gas from the slurry is difficult,
and some porosity may appear in the green bodies that cannot be removed in solid
state sintering.
652 4 Ceramics Based on Alumina: Increasing the Hardness for Tool Applications
Figure 1. Relicts of hard powder granules (from spray drying) in pressureless sintered alumina cera-
mics (solid state sintering).
4.2.1.2 Sol-gel Technologies

One of the reasons for developing sol/gel and precursor technologies is the use
of raw materials with particle sizes < 30-S0nm, enabling the production of very
fine-grained microstructures. However, the raw material exists in a phase com-
position which is different from the final product, and a complex sequence of
phase transformations takes place before or during sintering. Most of these
transformations are controlled by the formation of (or pre-existence of artificially
introduced) seeds. Without seeding, the transformation starts at a few isolated
locations, extends from these nuclei in three dimensions, and may produce a
textured microstructure with a preferential crystallographic orientation in rather
large subregions and with a porosity that is hard to remove even at high sintering
temperatures (promoting grain growth and destroying the benefit of using the
nanoscale raw material).
But even with seeding the fundamental problem is that a seed is required from a
phase formed in a later transformation step, and the seed particles should be as fine-
grained as possible to be homogenized (at a low concentration) with small distances
between them. However, seed particles are always much larger than the particles of
the precursor or in the sol (if they were available more fine-grained, people would
use them as a raw powder). As a consequence, the coarser character of the seed
determines the spatial homogeneity of the transformation and of the resulting
microstructure. With a typical size of seed particles between 0.05 and 0.3 ym and
seed concentrations between 1 and S%, the homogeneity cannot be better than in
the micrometer range. Therefore, even advanced sol/gel derived microstructures
contain defects like that in Fig. 2. On the one hand, submicrometer materials
with a narrow grain size distribution are easily formed and can be rather perfect
over distances of several dozens of micrometers. On the other hand, it is almost
impossible to avoid small (0.5-5pm) clusters of very small pores, the spacing of
which seems to correlate with the spacing of seeds in the sol/gel process.
4.2 Technological Essentials for Producing Hard and Strong Tool Ceramics 653
Figure 2. Microstructure with typical cluster of micropores in a submicrometer sintered alumina

ceramic produced from a boehmite sol with corundum (a-Al,O,) seeds <0.2 pm.
Typical flaws like the small pores in Fig. 2 are too small to be seen with the optics
of most hardness testers. Hence, an evasive positioning of indents is impossible,
and the testing gives a hardness which will be considerably lower than the inherent
hardness of the perfect submicrometer microstructure. In fact, this reduced hardness
represents a ‘true’ technical property of the material (because the flaws will affect
other hardness-related properties as well, e.g. the tribological behavior), and the
real drawback comes from the limitation of assessable properties and not from
the influence of measuring procedures.
4.2.2 Recent Trends in Ceramic Technologies Related to

Tool Ceramics
4.2.2.1 Technological Background
The drawbacks of known industrial technologies have initiated a number of new
developments in both powder synthesis and powder processing. Within the first
5-10 years, it is often difficult to distinguish which of the problems of a new technical
approach are simply associated with the present, not optimized state of the art, and
which are really characteristic of the new technology. Therefore, in the following,
the discussion is focused on examples where the original reference work shows
that the investigations avoided such technological problems in the fabrication of sin-
tered parts, as far as it is generally possible within the framework of a given (opti-
mized) approach. Of course. disadvantages which are inherent in a technology in the
present state of the art can never be excluded. For this reason dzflerent
654 4 Ceramics Based on Alumina: Increasing the Hardness f o r Tool Applications
flaw populations are characteristic of the various technical approaches and cause
processing-related property-limits which often cannot be overcome by materials
development [31.
As to new raw materials, the most significant step has been the development of
chemically derived substitutes for materials of fundamental industrial importance
like alumina or zirconia (most grades of which were originally manufactured
from mineral sources). Since the late eighties, new corundum powders have been
commercially available with fine particle sizes (rz 0.2 pm), which offers chances
for developing extremely fine-grained microstructures with a high density. The
small size and the increased curvature of the particle surface are expected to
increase the driving forces of solid state diffusion on sintering, to reduce the
required temperature, and to reduce grain growth. However, these potential
improvements are often offset by the increased tendency of the fine-grained
powders to agglomerate on sintering a body which contains agglomerates, densifi-
cation proceeds first inside these agglomerates (where the particles are in closer
contact), and, as a consequence, with the shrinkage of the agglomerates their
boundaries will move away from each other [4]. Then, to eliminate the sintering-
related new (and rather coarse!) interagglomerate porosity, much higher tem-
peratures are required than expected for the fine-grained powder used. And, of
course, the impact of the high temperatures on these fine particles is an undesired,
strong grain growth (first starting inside the agglomerates). Therefore, for many
years the published progress achieved with using these new raw materials was
much less than hoped.
Success came from increasing understanding of the relationship between sinter-
ing temperature, obtained density, and grain size as a function of spatial homoge-
neity and of interfaces, similarly to the microstructural dependence of the strength
and the hardness of ceramics [3,4]. The general idea concerning the mechanism of
sintering has been outlined above. The expertise starts from the simultaneous or
sequential application of several dispersal approaches to bring the extremely
fine-grained powders into a state that gives the right basis for the adjusted use
of different shaping procedures. Successful experiments were reported with direct
coagulation casting (DCC) [5] and centrifugal compaction [6] as well as with
cold isostatic pressing of freeze dried granules, pressure filtration, and gelcasting
[3]. Cold isostatic pressing easily gives dense sintered products with an average
crystal size of 0.5 pm. Gelcasting means shaping by casting a (preferentially
aqueous) slurry which contains a small percentage of a polymerizable additive.
After polymerization, the soft but geometrically stable body is transferred to
thermal processing. A bending strength of 700-900 MPa was reported for
pressureless sintered single phase alumina bodies (which is about twice the conven-
tional value) [3].
It is interesting to note that it is really the improved homogeneity of compaction
of the powder particles in the ‘green’ bodies that is the key to smaller grain sizes
with an increased hardness, wear resistance, and strength: in spite of a similar
green density (similar starting porosity at the beginning of sintering) of all
bodies produced by different shaping, the sequence dry (uniaxial) pressing +
cold isostatic pressing + pressure filtration + gelcasting gives a dramatic decrease
4.2 Technological Essentials f b r Producing Hard and Strong Tool Ceramics 655
of the sintering temperature by about 250°C (always using the same raw powder
and avoiding any doping additives). Therefore, the improved homogeneity of com-
pacted particles in the given sequence is the one possible reason for the improved
sintering behavior.
For most machining operations where really hard tools are required (e.g. grinding
or cutting of hardened steel), either more complex strength properties than a
maximum bending strength of compact bodies are required, or the macroscopic
strength is generally less important owing to the applied small feed rates (as in
precise cutting). Therefore, advanced principles for obtaining a high macrostrength
for sintered shaped bodies manufactured from these tool ceramics [ 3 ] will not be
discussed here in detail, whereas special emphasis shall be given to prospects for
improved hardness and wear resistance.
4.2.2.2 New Prospects for the Hardness

It seems obvious that this increased homogeneity is also the reason for the (quanti-
tatively evaluated [ 3 , 7 ] ) reduced frequency and size of defects and increased
strength. Equally clear is the influence of the reduced sintering temperature on
the reduced crystallite size, the increased hardness, and wear resistance. Figure 3
shows the increase in hardness obtained by decreasing crystallite sizes (under the
condition of a residual porosity 51%): single phase alumina ceramics can be
produced by pressureless sintering with a hardness that exceeds the hardness of
hot pressed TIC reinforced composites with a high carbide concentration of
v
Figure 4. Microstructure of a single phase submicrometer alumina grinding material manufactured

by powder processing (average crystallite size 0.4 pm, hardness HVlO (ground) = 22.4 f0.7 GPa).
45 vol-%; Fig. 4 shows such a submicrometer sintered alumina microstructure.

Further improvement is possible with small MgO additives (note that microhardness
measurements give artificially higher values related simply to the smaller size of the
indents; these microhardness data are, however, not representative for the technical
behavior of tool ceramics).
Dry Air (4 % r.h.) Humid Air (50 % r.h.) Water

-
2*1o'6-
1'1 0-l6
-
8'1 0 --
1~
6*1Oei7-
4*10-17-
2*10-j7
0.3 0.6 1.0 3.0 0.3 0.6 1.0 3.0 0.3 0.6 1.0 3.0
D @ml D WnI D @mI
Figure 5. Crystallite size and environmental effects on fretting wear of Al2O3-Al2o3couples.
657
Figure 6. Increasing grain size promotes grain boundary fracture and grain pull-out (fretting wear
in dry air; cf. Fig. 5 ) . Average grain sizes: (a) 3 pm; (b) 1.6pm; (c) 0.6pm (bottom) [I I].
4.2 Technological Essentials for Producing Hard and Strong Tool Ceramics
658 4 Ceramics Based on Alumina: Increasing the Hardness,for Tool Applications
When comparing data from different sources, the condition of the surface inves-
tigated has to be noted: machining introduces high dislocation densities [8], and the
hardness of ground ceramic surfaces is usually higher than measured on polished
surfaces, an effect that is the more significant the more fine-grained the microstruc-
ture is (Fig. 3).
4.2.2.3 A Leap in Wear Resistance

The grain size-hardness relationship in Fig. 3, published first in 1995 [9], has opened
new prospects for improved tool materials. Until that time it had not been
clear whether reducing the grain size of sintered alumina ceramics would cause a
technically significant increase in hardness; in the USA it was accepted that the
hardness increases down to crystallite sizes of 2 4 pm, whereas smaller grain sizes
do not contribute to an improved hardness [lo]. But even people with a different
opinion did not have a chance to test it because more fine grained microstructures
could not be prepared until the early nineties.
Figure 5 shows investigations of the fretting wear of alumina: depending on the
environmental conditions, submicrometer ceramics may exhibit a wear rate which
is an order of magnitude lower than measured for conventional A1203.It is impor-
tant to understand that this improvement is not only a simple consequence of the
increased hardness, but is also associated with reduced grain pull-out in the wear
track of the submicrometer material (Fig. 6) [ll].
4.3 Tool Materials with Undefined Cutting Edge:

Sintered Grinding Materials
Grinding is an indispensable process in industrial machining. With the exception of
special technologies like the grinding of wood or glass with garnet grits, it is gener-
ally assumed that high quality grits exhibit a high hardness. The consequences of a
high hardness are rather complex, and we will show that a more than proportional
increase of the grinding rate with a rising hardness of Al2O3 grits (Fig. 7 [12])
demonstrates only one among several positive aspects of a high hardness. Also,
hard grinding materials are advantageously applied not only for grinding metals
or very hard surfaces: hard S i c grits are likewise used for machining glass, some
heavy metals, lacquers, chipboards, and plastics. For working steel, on the other
hand, corundum grits (alumina, a-A1203)are superior to (the much harder!) Sic
due to their extreme chemical stability. Alumina grits are manufactured by melting
tons of raw powders in an electric arc (‘fused’ or ‘electro-’corundum). Depending on
the impurity content and additives, brown, white and Zr-corundum are different
types of fused grits where each of the quasi-macroscopic grains (available in the
size range 0.01-0.5 mm) has an internal structure close to that of a small single crys-
tal. As to the goal of a high hardness, the advantage of the absence of pores is to
some degree offset by the disadvantage of a microplastic deformability that is
4.3 Tool Materials with Undefined Cutting Edge: Sintered Grinding Materials 659
10 15 20 25 30
Vickers hardness HV1 [load = 1 kg] (GPa)
Figure 7. Influence of the hardness of the grinding grit on the rate of removed material on grinding
the cross-section of a steel tube (flexible disc, different alumina grits; the results have been normal-
ized for a unified grit toughness of KIc = 3.6MPam”’ to demonstrate the hardness effect without
the additional influence of Klc) [12].
lower than in polycrystals owing to the lack of grain boundaries. Zr-corundum, in

fact, contains such micro-interfaces because with higher concentration of zirconia
(Zr02) the grits are a eutectic of finest A1203 and ZrOz lamellae, but as the
hardness of the zirconia subregions is much lower than that of corundum the
macroscopic hardness of the eutectic is not improved but even somewhat reduced.
On the other hand, the eutectic structure increases the fracture toughness to twice
that of grits which are free of zirconia (here, toughness is a parameter which may
be understood as a strength related to a unified defect size). On machining cold
rolled steel sheets, the increase of the grinding rate by about 20-30%, in spite of
the reduced hardness, has to be attributed to this increase of the toughness in Zr-
corundum grits.
For many applications fused grits are unsatisfactory, and efforts started at the
beginning of the eighties to develop much harder (single phase) corundum grades
by producing grits with a polycrystalline internal structure. This idea would affect
not only the hardness, but also the fracture toughness, the microfracture behavior
and the formation of chips. A difficult problem to be solved was that most of the
conventionally known approaches for the manufacture of compact polycrystalline
sintered ceramics cannot be used for producing irregularly shaped small grits
(e.g. uniaxial pressing or cold isostatic pressing of powders). Hence entirely new
procedures had to be found to obtain a ‘green’ density of the intermediate (not yet
sintered) product that is high enough to enable sintering until a high density with
less than 1-2% of residual porosity is reached. Presently approaches that associate
minimum crystal sizes with a maximum density of the sintered grits are the primary
target of research.
Efforts to increase the hardness by using composite materials with harder addi-
tives are much less important for grits than in the field of compact ceramics. The
reason is the chemical nature of available hard materials (carbides like Sic, Tic,
but also nitrides or, in a few experiments, diamond). The inability of Sic to
grind ferrous materials is well known, and a similar observation had to be made
for carbide reinforced composite alumina grits, in spite of their improved hardness.
Further, a carbide additive in an oxide matrix does not always result in an
increased hardness because the low diffusion of the carbides (dominated by
covalent bonds) often gives increased porosity or (after sintering at a higher
temperature) a coarser microstructure. On the other hand, highly pure (and
therefore chemically very stable) sintered corundum grades have been developed
recently, which after pressureless sintering exhibit a hardness similar to or
even higher than the hardest hot-pressed composites with high carbide concen-
trations (cp. Fig. 5). All these features limit the motivation for new composites
and increase the trend towards advanced, thermodynamically stable sintered
grits on a pure oxide basis.
4.3.1 Technical Demands for Grinding Materials

The following discussion is focused on grinding materials for machining steel and
cast iron, because these are most important industrially. Two different requirements
are to increase working efficiency and to minimize undesirable interactions of the
tool with the ground surface.
A frequently used criterion in patent applications is an increased rate of material
removal. Another parameter is the quality of the ground surface: the tool should run
for a long time with insignificant changes in the roughness of the ground workpiece.
With new grinding materials that increase the grinding rate by 30-100% (see
paragraph 4.3.2 and 4.3.3) or which give a roughness inside of the tolerance limit
for a doubled working path, the economic benefit is even larger than given by the
percentage of the reduced time per process (or per workpiece). The increasing
output of the manufacturing lines increases the importance of down-times, affected
also by the frequency of tool exchanges. The extended life of grinding tools gives an
additional contribution to the economic efficiency of a line if, besides the reduced
time in the individual grinding step per workpiece, an increasing number of
workpieces can be manufactured with one tool.
In many grinding operations the development of heat is very important: undesir-
able changes in the microstructure and the properties of the ground material have
to be avoided (e.g. in hardened steel), the introduction of residual surface stresses
on machining is tolerable within given limits only. Since grinding tools can give
high material removal rates at fairly low contact pressure (minimizing the process
temperature), the final grinding of functional surfaces in ball bearings can hardly
be substituted by other approaches today. The interest in grinding materials that
exhibit a high grinding power at low process temperatures has been further pro-
moted by the need to improve the environmental compatibility of manufacturing,
a trend that is manifested by efforts to decrease or to avoid the consumption of
coolants (reduced coolant concepts, dry grinding). There are manifold ways to
develop such grinding materials. One approach is microstructural engineering:
the shape of the crystals in the microstructure governs the fracture behavior of
the grinding grain, the shape of the cutting edges and tips, and, finally, the chip
formation. One of numerous ideas was the composition of grinding tools with
hollow sphere-shaped bodies that, during grinding form always new (sharp)
grain edges.
4.3.2 Advanced Commercial Products: Sol/gel-derived

Corundum
Obviously, to develop grinding grains between 10 pm and 500 pm in size and with
a polycrystalline internal structure is possible only if the grit is manufactured as a
polycrystal with crystallite sizes much smaller than the dimension of the grinding
grain, preferentially < 1 ym. However, the available corundum raw powders have
particle sizes of about 0.5 pm, and until now it is impossible to create microstruc-
tures with crystallite sizes of less than about 2 pm from such powders by pressure-
less sintering. Success came with the application of sol/gel procedures that use
chemical precursors of the final corundum phase as extremely fine-grained
(usually nanoscale) raw materials. Most sol/gel approaches are expensive, but
the grinding industry has benefited from the unique situation with alumina
where sols can be prepared from boehmite (an aluminum monohydrate
A100H) readily available from other chemical processes (the price is about
$5 per kg or less). Unfortunately, the sequence of chemical and crystallographic
transformations from the precursor to the final phase generates substantial
risks in forming inhomogeneous, porous microstructures, running counter to
the goal of high hardness. Patent applications suggest different doping [ 131 and
seed [14,15] additives intended to control these transformations. As a result it
is, in fact, possible to improve the strength, and few examples offer significant
advantages in the hardness compared with white fused corrundum. Nevertheless,
on grinding rolled steel sheets such grits exhibit an increased power expressed, for
example, by a rate of removed material which is 140-150°/~ compared with brown
corundum or 115-125% compared with the fused A1203-Zr02 eutectic. The com-
parison with fused grades in Fig. 8 includes two intermediate qualities of sol/gel
derived (sintered) alumina grits together with an advanced material of the sol/gel
group the grinding rate of which is increased by about 150% related to fused
A1203-Zr02 [16].
In Fig. 8, the improved technical performance of the polycrystalline grinding
materials in spite of their almost unchanged hardness seems to disagree with the
250
200-
1r
- sintered AI,O, (soVgel)
K,
(MPadrn)
HVI
Wa)
- 0 *. * *. .O fused eutectic AI,OJZrO,
2.83.9 15-20
4.2 16.3
1 .O-2.5 17.5
I I I I I I I I I I I
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
time (min)
Figure 8. Comparison of the material removal rate of sol/gel derived sintered and of fused grits on
the basis of AI2O3(flexible fiber disc, 6000min-’, contact pressure 30N; workpiece: steel tube [ST W
22 DIN 15431 with diameter 195 mm, wall thickness 1 mm).
concept of developing more powerful grits by increasing the hardness (due to a

fine-grained microstructure). Obviously, there are other strong influences, and
first of all strength properties, modifications of the microfracture behavior are
expected to govern the results in Fig. 8. It has been recognized for a long time
that the strength of ceramics depends on microstructural inhomogeneities
(‘defects’, ‘flaws’) which locally increase the externally applied mechanical load,
causing sudden, ‘brittle’ failure. What is important for grinding grains is, however,
not a generally high but an adjustable (depending on the specific grinding process)
strength: on the one hand, some global minimum strength under given loading
conditions is imperative to run the grit for some time but, on the other hand,
originally sharp edges and tips of a very strong grain would be worn and
would deteriorate the grinding efficiency within less than a minute: an effect
that is readily observed with very strong (e.g. carbide reinforced) ceramic
grains. Therefore, the strength must be low enough to enable microfracture pro-
cesses on a local microscale for re-generating new sharp edges during the grinding
process.
Since it is impossible to evaluate the strength of a geometrically irregular grinding
grain by standard approaches, the fracture toughness is measured by indentation
procedures on mounted samples. The method derives the toughness from the
length of indentation cracks formed at a given load. The analysis of a larger
number of grains gives the statistical distribution of the loading capacity in different
grades of grinding materials, and these statistics can be correlated with the technical
4.3 Tool Materials with Unde$ned Cutting Edge: Sintered Grinding Materials 663
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
K (MPdrn) K, (MPaJrn)
Figure 9. Microstructural dependence of the fracture toughness in grinding grains on the basis of
alumina prepared by sol/gel approaches (left) and in fused grits (right).
grinding performance of the grits under different conditions [16]. Figure 9 shows
different toughness distributions obtained in sol/gel by producing microstructures
with different porosity, crystallite size, and shape (as exemplified by Fig. 10).
Obviously, the polycrystalline nature of sintered materials is a strong tool for modi-
fications of the mechanical properties which are impossible in fused corundum grades.
From Fig. 9 it can be understood that the sol/gel derived grits owe their increased
grinding efficiency (Fig. 8) first of all to their improved global stability (fracture
toughness, strength). On the other hand, it is very instructive to understand the
grinding performance of the A1203-Zr03 eutectic: with a very high toughness
(Fig. 9) and with a hardness which is smaller than observed for zirconia-free
fused grits but in no way inferior to the sol/gel grades (Fig. 8), it is impossible to
attribute the only intermediate grinding efficiency of this material (Fig. 8) to deficits
in its toughness or hardness. On the contrary, the A1203-Zr02 eutectic is a very clear
example for the power-limiting effect of too high a strength of grinding grains,
decreasing the frequency of microfracture processes which are needed to re-generate
sharp cutting edges.
Of course, the concept of an adjustable strength that enables an optimum fre-
quency of microfracture events does not mean tolerance of hardness deteriorating
defects. From Fig. 2 it can be shown that within the framework of sol/gel
approaches the frequency and size of such defects can be minimized but hardly
avoided. In spite of the high number of patent applications devoted to improved
sol/gel-derived grinding materials, it has been impossible to overcome the funda-
mental limit associated with this technology: the discrepancy between the unique
chances to use nanoscale raw materials and the spatial heterogeneity introduced
by the indispensable but always coarser seeds. With the obviously negative impact
of flaws like those in Fig. 2 on the hardness and wear resistance even if the porosity is
no more than 1-2%, it has to be expected that sol/gel derived products will not be
able to exceed some limiting level of properties.
Figure 10. Microstructure of sol/gel derived sintered alumina grits: (a) platelet shaped microstruc-
ture (‘Cubitron’, manufactured by MMM, USA), (b) bimodal grain size distribution after seeding
with transitional alumina nuclei, (c) anisotropic grain growth (Ce doping) [ 161.
4.3.3 Sintered Alumina Grits Produced by Powder Processing

Approaches
Sol/gel-derived grinding materials probably have been the first example where aprecur-
sor approach was successfully used in the industry for a manufacturing volume of
several tons per year. The scientifically fundamental importance of this comes from
the circumstance that it was also the first example of highly dense and submicrometer
ceramics with advanced mechanical properties. This progress was first achieved in the
industry and only later on investigated in scientific institutes in greater detail. And
again, as to the mentioned limitations of the sol/gel approach, it was the requirements
of the industry which promoted a new idea: to complete the family of fused and sol/gel
derived grinding ceramics by a third group which due to a generally different manufac-
turing process should exhibit not a stepwise improvement but another class of quality,
similarly as the sol/gel grits have shown a generally improved performance compared
with the fused materials. Success came with an apparent return to well-known powder
processing technologies, but on a rather different level, as described in Section 4.2.
Considering that such basic influences like grain size and porosity effects on hardness
are often a subject of controversy and are not completely understood, it is not a surprise
that only recently has hardness been discussed as a function of defects [17]. Figure 2
shows small defects (2 pm or less) which will significantly deteriorate the hardness
parameter which is effective for the grinding process. On the other hand, larger flaws
like cracks or pores of about 5-20pm do not necessarily decrease the grinding effi-
ciency: on grinding, such flaws will become nuclei of microfracture processes (which
are useful to re-generate sharp cutting edges), and these fracture processes will eliminate
thejaws from the microstructure of the mesoscopic cutting edge (the hardness of which
is determined by remaining smaller defects). Hence, the 'effective' (local) hardness may
be higher than a value measured with the influence of coarser pores.
All microstructural understanding, however, can give rise to a new class of
grinding materials only if the basic mechanical properties are adjusted to some appli-
cation-related optimum. Therefore, for grinding as for cutting tools, investigation of
technical performance under defined conditions is necessary. Figure 11 shows the
results of test series performed by Hermes Abrasives, Hamburg (Germany). The
most interesting result is the large progress in the grinding efficiency obtained by
the powder derived material compared with the advanced sol/gel grits (all sub-
micrometer microstructures); the difference is of similar magnitude to the advantage
of the soligel grits compared with fused brown corundum. Compared with the
A1203-Zr02 grinding grit, powder processing gives a single phase alumina grade
which exhibits twice the material removal rate of the eutectic. It has to be
emphasized that this improvement does not depend on the extremely small
crystallite size of the microstructure (e.g. 50.5 pm, for which range Fig. 3 indicated
a decreasing grain size influence in the hardness): the result in Fig. 11 relates to a
powder-derived material with a grain size of about 1 pm [16] and was equally
reproduced with crystallite sizes between 0.5 and 1 pm. An example of a powder-
derived tool material with a submicrometer microstructure was given in Fig. 4.
With a fracture toughness of 3.34MPa'12m for all sintered corundum microstruc-
tures independent sol/gel or powder processing, the additional data included in Fig. 11
HV1lnhei
250- sintered alumina Wa)
(by powder processing) 21-25
200 -
1
0) sintered alumina
-
v
z
.-
$150-
E
-0
a,
>
E 100-
??
50-
01 I I I I I
0 1 2 3 4 5 6
time (min)
Figure 11. Comparison of the grinding performance of known (fused and sol/gel derived) grinding
materials on the basis of A1203with a new single phase corundum grit produced by a powder
processing approach (same test conditions as given for Fig. 8) [16].
demonstrate the importance of the ‘effective’ (local) hardness on technical perfor-

mance and especially for the stepwise increase of the grinding power in the ranking
sequence fused corundum + sol/gel derived sintered alumina -+ sintered alumina
by powder processing. This effective hardness (which is inherent to the microstruc-
ture) is diminished by small defects like that in Fig. 2, and with their elimination by
substituting sol/gel by powder approaches the hardness increases (without decreasing
the crystallite size!) and increases the grinding efficiency. With a sufficiently large
number of tests, results like that given in Fig. 7 can be obtained which demonstrate
a general trend of an increasing grinding power with increasing hardness (under the
condition of an always equal chemical interaction of all grits with the ground metal).
4.4 New Trend for Cutting Hard Workpieces:

Submicrometer Cutting Ceramics for Tools with
Defined Cutting Edge
Characteristic for the late nineties was the increasing use of tough grades of silicon
nitride, first of all when high feed rates are requested. With their rather low hard-
ness, however, Si3N4 ceramics are outside the scope of the present discussion.
4.4 Neiz, Trend for Cutting Hurd Workpieces 667
Most of the machining of lzavdened steel and hard cast iron, on the other hand, is done
by grinding, a process that today is necessary for the final finish of functional surfaces,
like in ball bearings. Nevertheless, there are several limitations of grinding which
motivate the search for other methods: compared with cutting operations (whenever
applicable) grinding is a time-consuming process (up to a factor of three), the equip-
ment is more expensive, and disposal or recycling of the grinding silt is difficult from
both ecological and economic points of view. It has been shown that even without a
complete substitution the working time can be cut to 2/3 by combining a first turning
step with final grinding [2]. If turning operations shall be used for the final finish as
well, important tool life criteria are the surface roughness of the machined workpiece
and the stability of the geometrical position of the cutting tip (which governs the accu-
racy of the produced piece). This paragraph deals with the technical demands on cut-
ting tools which might be able to give fresh impetus to such efforts without increasing
the costs (as it is the case with tools manufactured of diamond or cubic boron nitride).
Therefore, the discussion will be focused on ceramics on the basis of alumina as
outlined already in Section 4.1.
4.4.1 Demands for Cutting Materials Used for Turning

Hard Workpieces
Table 2 illustrates the primary technical demands on cutting ceramics for machining
hardened steel or hard cast iron. These demands are a consequence of the required
wear resistance in dynamic contact with the abrasive carbide particles in the
microstructure of the cut metals. As to more fundamental properties of the cutting
plate, there is the requirement for the combination of maximum hardness and
sufficiently high microscopic fracture toughness of the grain boundaries to minimize
the frequency of grain pull-out at the cutting tip.
From the understanding of the influences of the microstructure on hardness
and wear properties, an average grain size of substantially less than 1 pm and a
low frequency of microflaws are imperative to meet these demands. On the other
Table 2. Demands on basic properties of cutting tool ceramics for machining hardened steel:
x = important, .xx = very important
On precise machining
Technical demands Hard cast iron Hardened steel
Wear resistance (especially against the abrasive

impact of carbide particles) S.X SX
Low thermal conductivity S x
Low creep rate (at high temperature) .x xx
High oxidation stability x
High chemical stability (preventing solid state
reactions with constituents of the iron-based
workpieces) xx
hand, however, special measures may be required to balance the increasing high-
temperature creep usually associated with small grain sizes. Reinforcement with
whiskers (Sic whiskers are most common [18]) is one way to reduce the creep
rate [19], an approach that also increases the thermal shock resistance [20]. Another
way to reduce the creep rate is grain boundary engineering. Besides the prevention
of impurity associated amorphous grain boundary films it was observed that a small
concentration of yttria (0.1%) can reduce the creep rate of sintered alumina [21].
As it is also known from other fields of working metals, the different forms of the
chips from machining cast iron and steel cause a much more intense interaction with
the cutting plate for the latter material. Therefore, the demands concerning the
chemical and oxidation stability are more stringent on working hardened steel
than on machining hard cast iron (Table 2).
If the wear resistance is increased by decreasing the grain size of the ceramic
microstructures, an additional advantage comes from a feature illustrated by
Fig. 6. The reduced wear rate is associated with a decreasing amount of grain
pull-out at smaller grain sizes, and the same effect can improve the quality of the
ground ceramic surfaces. With a constant grinding procedure in finishing alumina
ceramics (diamond wheel, grit size 4&50pm), the roughness of a ground 0.6pm
alumina ceramic is about one half the value measured on a ground 3.6 pm micro-
structure [7]. Obviously, the improved quality of the ceramic cutting edge increases
the chances for tool life at a small roughness of the machined hardened steel.
To prevent spontaneous microstructural changes in the workpiece and to obtain a
high efficiency of the machining process, all approaches try to realize conditions with
low cutting forces and a low heat input into the finished surface. Whereas this is gen-
erally true for both grinding and cutting operations, it is also accepted that the use of
cutting ceramics is promoted if local temperatures are high enough for some softening
of the (usually brittle) surfaces of hardened steel or hard cast iron [2]. To this end,
ceramics with a low thermal conductivity are preferred. Table 3 reveals little differ-
ences with the exception of the most expensive tool materials (cBN, diamond)
which are on the less favorable side (in fact, the high thermal conductivity of the
latter gives an improved thermal shock behavior which, however, is considered as
less important when small feed rates are applied on precise machining hardened
steel). Much more significant differences exist for the hardness of the different
ceramics (Table 1) and for their mechanical resistance at high temperatures. As to
its oxidation stability and its chemically almost inert behavior in contact with iron
or steel, pure sintered corundum is second to no other ceramic.
Table 3. Thermal conductivity of ceramic cutting tool materials.
Cutting tool ceramics Thermal conductivity [W m-' K-'1
A1203+ x 15% Z r 0 2 = 28
A1203 x 30
A1203+ x 30% Ti(C, N) x 35
Si3N4 30-35
Cubic boron nitride (cBN) 40-120
Diamond 500-600
4.4 New Trend for Cutting Hard Workpieces 669
4.4.2 Carbide Reinforced Composite Ceramics Based on A1203

Previous investigations of precise cutting operations with hardened steel or hard
cast iron were focused on black composite ceramics of A1203 with about 30%
Ti(C,N). No positive results are known for grades based on Si3N4 or
A1203/Zr02; probably, hardness and chemical stability of silicon nitride and of
zirconia do not meet the demands of this machining procedure.
For example, a commercial AI2O3/Ti(C,N) tool was used for the final (precise)
machining of the inner diameter of a ball bearing ring [2]. The average crystallite
size of such cutting ceramics is about 1-1.5 pm. With the large body of evidence
for the improvement of hardness and wear resistance by decreasing grain sizes, it
is not surprising that most manufacturers have started to develop these tools with
a submicrometer microstructure, but the covalent nature of the carbide bonds pre-
vents pressureless sintering of the composites at temperatures of 1600°C or less.
With TIC concentrations of 25-35%, the most fine grained of these composites exhi-
bit average sizes of A1203and TIC subregions of about 0.8-1 pm associated with a
HVlO hardness up to 23 GPa (measured at a testing load of 10 kgf). Compared with
today's commercial grades with HVlO < 21.5GPa, this is an increase of almost
10%. Nevertheless, these tools showed only similar behavior on fine turnin gray
cast iron (GG25, lOOOmmin-'), hard cast iron (60-65 Shore, 150mmin- ) and F
on turning a hardened steel 100CrMo6 for ball bearings (180mmin-', Fig. 12).
The latest patent applications try to obtain more fine grained microstructures out-
side the known compositions A1203/TiC or A1203/Ti(C,N). An alumina composite
with 25% (by weight) of a mixed carbide/carbonitride consisting of about 90%
Ti(C,N) and 10% WC was hot pressed at 1600°C yielding a microstructure with
hard cast iron hardened steel 100CrMo6

(60-65Shore)
v = 150 d m i n v = 180 m/min

e f = 0.25mm/rev f = 0.1 mm/rev
-
L
Corn S
E
-
Y
c
0
S
E 10-
.-
Corn
5
c
E
z
5-
0
Figure 12. Turning hard workpieces with different ceramic composite tools. Corn = advanced
commercial AI20,/Ti(C,N) grade, concentration of Ti(C,N) about 25-30 vol-%. S = laboratory
grade: average sizes of A1203 and TIC subregions 0.8-1 .O pm, TiC concentration 35 vol-%.
670 4 Ceramics Based on Alumina: Increasing the Hardness Jor Tool Applications
an average grain size of 0.5 pm [22]. On turning hardened steel (X63CrMoV5) with a
discontinuous cutting regime, the comparison with an unspecified commercial tool
showed lifetime (or cutting path) that was three times greater. Another recent devel-
opment uses oxycarbide and oxycarbonitride compositions (Ti(C,O), Ti(C,O,N))
with different stoichiometry instead of TIC and obtains crystallite sizes of about
0.6 pm (after sintering at 1700°C and hot-isostatic post densification at 1435°C);
the hardness is HVlO = 22.5-23 GPa [23]. Figure 13 shows the microstructure of
these materials which is much more fine grained than an advanced (also claimed
submicrometer) commercial material with the same carbide concentration (on the
market since 1997). As another example, Fig. 13 includes a new material with an
increased concentration of 35 vol-% of the covalent phase which can be manufac-
tured with the same small grain size in spite of the high carbide loading.
The cutting performance of these new materials was tested in a direct comparison
with similarly hard but single phase (absolutely oxidation resistant) pure alumina
ceramics. The results are discussed in Section 4.4.4.
4.4.3 Single Phase Sintered Corundum

From Table 2, an advanced cutting ceramic for the precise machining of hard work-
pieces should associate a maximum hardness with finest grain sizes (to minimize
grain pull-out at the cutting edge) and high chemical and oxidation resistance.
With today’s knowledge, it is not clear if carbide or carbide reinforced tools will
ever be able to meet the thermodynamic and chemical side of these demands: in
fact, the (negative) free enthalpy of T i c and TiN is only half of the value of
Si3N4 [24], and even the silicon nitride is unable to cut mild steel (not to mention
hardened steel). On the other hand, from the data in Fig. 3 there exist pure sintered
corundum ceramics with a hardness that even after pressureless sintering is in no
way inferior to hot-pressed carbide reinforced composites, with a strength of 800-
900 MPa [3] (after HIP up to 1300 MPa [25])that equals or even exceeds the strength
of the composites, and with the additional great advantage of highest chemical and
oxidation resistance.
Cutting tests performed with these recently developed alumina grades are rare,
some first examples are compared in Section 4.4.4 with the performance of commer-
cial and of submicrometer composite tools.
4.4.4 Comparative Cutting Studies with Submicrometer Ceramics:

A1203 and Composites Reinforced with Ti(C,N) and Ti(C,O)
4.4.4.1 Mild Conditions: Machining Cast Iron and Steel
First investigations of the cutting performance of submicrometer tools have been
performed in Asia [26-281. Unfortunately, all these tests describe only the machining
of unhardened steel and cast iron. On milling cast iron, the submicrometer sintered
AI2O3 showed almost no advantage and was inferior to 70A1203/30TiC and
Figure 13. Latest tool composites based on A1203:(a) commercial ‘submicronstructured’ Al2O3/
671
Ti(C,N) composite and new laboratory grades [23] with (b) Ti(C,O;N), (c) and with an increased
4.4 N ~ MTrend,for
J Cutting Hard Workpieces
concentration of about 35 vol-% Ti(C,O).

"""1
f = 0.2 mm/rev
a = 1.5 mm
0.0 0 1 2 3 4 5
Cutting time (min)

Figure 14. Wear flank width on turning steel (Japanese code S45C). G = sintered corundum with
grain size 4 pm. Adapted from Takahashi et al. [26].
60A1203/40Zr02composites, a result which is hardly surprising: in this application,

the differences in the tool hardness are less important, and the improvement in the
toughness of submicrometer A1203microstructures is so small ( < 4 MPa m1j2[3,29])
that it is technically unimportant. On the contrary, on turning cast iron or steel
(Fig. 14) the progress obtained by the submicrometer alumina was significant. It
should be emphasized that an advantage of the pure alumina tool compared with
carbide reinforced composites and with zirconia toughened alumina was observed
just under these conditions where during continuous cutting an increased processing
temperature at the cutting edge applies.
4.4.4.2 Machining Hard Cast Iron and Hardened Steel

The behavior of the tools described in the last section supports suggestions of a pre-
ference of submicrometer alumina ceramics for the precise machining of hardened
steel and hard cast iron. First investigations in this field started in 1998 at the Dres-
den Institute of Ceramic Technologies and Sintered Materials. All turning tests were
performed with a feed rate, f = 0.1 mm min-' and a depth of cut, a = 0.2 mm. The
size of the cutting inserts was SNGN120412 (i.e. 12.7 x 12.7 x 4.76mm3 with a
1.2mm radius) with a 20" chamfer (width 0.2mm). The plates were positioned
with a rake angle y = 6", an inclination angle h = -4", and a large entering angle
x = 45" chosen to maximize the selectivity of the tests (as to the tool materials)
4.4 New Trendjor Cutting Hard Workpieces 673
Table 4. Properties of new tool ceramics compared with advanced commercial grades
Composition Average grain size Hardness HVlO

(composites: size of (Vickers, 10 kg testing
single phase load, ground
subregions) surfaces)
(w) (GPa)
Composites
SH1 (CeramTec) Al2O3+ N 33 vol-YO Ti(C,N) 1.5 21.16 f 0.23
commercial
‘submicron’ A1,03+ N 33 vol-YOTi(C,N) <1 21.22 f0.54
AT60A Al2O3+ N 33 vol-% Ti(C,O) 0.7 22.18 f0.77
AT60 A1203+ N 37 vol-YOTi(C,O) 0.67 22.45 f 0.85
Pure alumina
AC4 1 A1203 0.56 22.08 f 0.90
by high loads at the cutting tip. A new commercial grade of an A1203/Ti(C,N) tool
ceramic was claimed ‘submicronstructured’ by the producer and introduced into the
market 1997 especially for machining hard materials; it was compared here with an
advanced standard grade SH 1 (CeramTec AG, Plochingen, Germany) and used as a
reference to evaluate the cutting power of new laboratory grades (Table 4). Both the
submicrometer and the SHl tools are composites based on A1203with about 30%
of Ti(C,N).
On the turning hardened steel at high cutting velocities > 250 m min-’ there is a sig-
nificant input of heat not only into the formed chips but also into the cut surface of the
shaft. Depending on the state of wear of the tool, cutting forces and process tempera-
tures may increase to an extent that causes softening of the hardened steel, and the
hardness was observed to drop from HRC = 57-60 to values of less than 50. There-
fore, the decrease of the hardness has to be recorded as an additional parameter in
such investigations, and an intermediate machining operation with a low velocity
’
‘u = 180 m min- was required before every new run to restore the original hardness.
4.4.4.2.1 Intermediate Demands: Machining Hard Cast Iron Figures 15-17 show
the results for continuous cutting chilled globular cast iron G-X300CrMo 153
(Rockwell hardness HRC = 43-45). The flank wear of the tools is given in Fig. 15.
The high hardness of the oxycarbide reinforced submicrometer composite reduces
its wear by an amount similar to the progress of the commercial submicrometer
tool compared with the standard material SHI. Much more surprising is the quali-
tative leap in the lower wear of the pure alumina insert which cannot be explained by
its hardness: the hardness is only just intermediate between the (lower) hardness of
the commercial composites and the (higher) value of AT60A. Also, the grain size of
the sintered corundum material is not very different from the microstructural data
given in Table 4 for the A1203/Ti(C,0)composite AT60A.
Therefore, Fig. 15 suggests that reducing the grain size of TIC, Ti(C,N), or
Ti(C,O)-reinforced composites may indeed increase the hardness and reduce the
wear in turning hard cast iron, but the effect is limited, probably because of chemical
0.3.
0.25.
E
E
! 0.2-
m
>
5
50.15-
Y
c
m
-
c
5 0.1.
3
0.05.
i
p
if
0.0. I I I I I I
0 10 20 30 40 50 60 70
Cutting time (min)
Figure 15. Flank wear width of submicrometer ceramic tools on turning hard globular cast iron
(Rockwell hardness HRC = 43459, see Table 4 for tool characterization.
interactions between the iron and the carbide phases of the tools. To eliminate this
interaction by using a pure alumina tool can greatly improve the wear resistance as
far as the process temperature at the cutting tip is not too high to initiate additional
wear due the low high-temperature creep resistance of the submicrometer alumina
microstructure.
The cutting edge displacement (CED) characterizes the obtained precision in
turning hard materials. Figure 16 (CED) shows that complex wear processes may
give a different ranking of the tool materials compared with Fig. 15 (flank wear).
Again, all of the new laboratory submicrometer grades exhibit much less wear
than the advanced commercial tools. However, with the criterion of the cutting
edge displacement it is the ceramic with the highest hardness which shows the
best performance (composite AT60A, Table 4).
Another important criterion for the quality of tools in precision turning of hard
materials is the obtained roughness of the cut metal surface. The graphical
presentation in Fig. 17 gives the average roughness R,. A fixed ratio between the
statistically averaged maximum depth of roughness RZD and R, was observed in
all the test (Fig. 17). The arbitrary scatter of data was rather large, and no significant
difference was observed between the surfaces cut by the SHl or by the commercial
submicrometer tools. However, all other inserts show the same ranking as observed
for the flank wear in Fig. 15. Improved results are provided by the oxycarbide
reinforced composites (AT60), but a more significant progress results from the
use of pure submicrometer alumina inserts.
4.4 New Trend f o r Cutting Hurd Workpieces 615
70
cast iron HRC = 42 - 44
v = 250 mlmin =
mm
f = 0.1 mmlrev
a = 0.2 mm K = 45"
E
c
c
E
-
$40-
Q
0)
.-
U
g30-
73
g 20-
0,
.
0
= i AT6OA
10- (Al,OJri(C,O))
0 I
I)-.. ..
I
....4
I I I I
Cutting time (min)
Figure 17. Surface quality of cut surface. Average grain sizes of the tool microstructures (in com-
posites: sizes of single phase subregions) are D(SH1) = 1.5pm, D(AT60) = 0.67pm, D(AC41) =
0.56 p m (see Table 4).
4 Ceramics Based on Alumina: Increasing the Hardness for Tool Applications
60.0
hardened steel criterion:
HRC = 57 - 60 flank wear width
40.0
'. VB = 0.15 mm
.E20.0
c
h
e
3
0
c
0
E
-
.-
10.0
8.0
6.0
4.0 ,
150 200 250 300 350
v (m/min)
Figure 18. Life time of advanced ceramic inserts on machining hardened steel (see Table 4 for
materials characterization). The arrows indicate frequently observed deviations.
Whereas turning of hard materials is usually performed with asymmetrically

positioned inserts (entering angle x. = 10-30"), it was mentioned above that
K = 45" was chosen in these investigations to maximize the selectivity of the tests
(as to the tool materials) by high loads at the cutting tip. For geometrical reasons,
a lower roughness is expected for both smaller (10-30") and larger (60-80") entering
angles. Therefore, it should be noted that in Fig. 17 the increase of R, towards
values between 2 and 4 pm is directly associated with this very special orientation
of K = 45" and does not indicate a disadvantage of the tested tool materials.
Indeed, in an additional test performed with SH1 and with the commercial sub-
micrometer tool at K = 75" all R, values were < 1.5pm for the whole time of cut
(i.e. up to 0.2mm flank wear at least).
4.4.4.2.2 Extreme Conditions: Continuous and Discontinuous Cutting of Hardened

Steel Figures 18-19 give the results for continuous cutting of hardened steel
90MnCrV8. Its high hardness (HRC = 57-60) and the choice of tool materials
with a low thermal conductivity cause a large input of heat into the formed metal
chips. These appear rather flaming than red-glowing, an effect which is optically
more spectacular on machining hardened steel with pur corundum inserts than
using composite tools the thermal conductivity of which is about 15% higher (see
Section 4.4.1, Table 3). On machining hardened steel with composite tools, the
lower temperature gives more faceted chips.
For the reference tools (SH1 and the commercial submicrometer ceramic), the
high temperature is associated with a critical influence of the cutting velocity
4.4 New Trendjor Cutting Hard Workpieces 677
(al 70 (bl 70
hardened steel HRC = 57 - 60 hardened steel HRC = 57 - 60
60-
I
E
150-
c
E
8 40-
n
$
g30-
P
5 0I
20-
0
10-
0 I
*
I
.
... .-.
I I I I
0 5 10 15 20 25 30
Cutiing time (rnin)
Figure 19. Cutting edge displacement on turning hardened steel with advanced ceramic tools: (a) at
v = 220mmin-' (above), (b) at 300mmin-' (below); see Table 4 for materials characterization.
between 200 and 250 m min-' . Flank wear is small at v = 220 m min-l , but already
at 'u = 235 m min-l a greater crater-wear is observed. At v = 235 m min-I, more
than 50% of the tested commercial inserts exhibited a sudden, strong wear increase
during the first 10-20min associated with local or global fracture at the cutting edge,
and it becomes difficult to record a continuously increasing flank wear up to
VB > 0.2mm.
Whereas the cutting edge displacement was immeasurably small during the first
5 min, it was overtaken by fracture soon later and could not be determined optically
(even at v = 220 m min-' the cutting edge displacement of commercial tools was
often shadowed by local features of microfracture). Hence, it was impossible to
measure the cutting edge displacement of the reference tools at v = 235 m min-].
Therefore, tests started with v = 220mmin-', but some of the new laboratory
grades exhibited extremely few wear at this velocity, and additional experiments
had to be performed at higher rates. It should be noted, however, that already at
v = 235 m min-' the reference curves for SH1 and the commercial submicrometer
material characterize but a minority of inserts standing this condition (about 50%
failed at shorter times by cutting edge fracture). Therefore, these data (Fig. 18) over-
estimate the real behavior of the commercial grades at this velocity.
Whereas even at the lower velocity of u = 220mmin-' almost 50% of both
commercial grades failed by fracture during the first 5-1 5 min of cutting hardened
steel, no one such event happened with the laboratory grades (where crater-wear
was also observed but to a lesser extent compared with the commercial ceramics).
This behavior correlates with another surprising observation made on preparing
the hardened shaft for the tests. This preparation is difficult with shafts of hard
cast iron or hardened steel because it associates the high hardness of the counterpart
with discontinuous cutting conditions at a changing frequency and power of
impacts. Cubic boron nitride inserts (dreborid$/, Lach company, Hanau/Germany)
failed on machining the hard (globular) cast iron, and hard metal inserts (WC/
6% Co, Vickers hardness M 16 GPa, Kc M 9 MPam'/2) were not able to cut the
rough outer shell of the hardened steel shaft at velocities between 50 and
100mmin-'. The same observation had to be made with the commercial ceramic
composites (SHI and the submicrometer grade fractured after t < 2min). On the
contrary, both the pure alumina insert AC41 and the composites AT60A and
AT62 were successfully used with w = 120 m min-', feed rate f = 0.3 mm rev-',
and a depth of cut a = 1.5mm. These new, submicrometer cutting ceramics did
not only not fracture macroscopically during one hour of severe cutting (with one
tip) but retained microscopically nearly perfect cutting edges providing a high sur-
face quality of the hardened shaft in the first cut ( R , < 1 pm for t 5 15min [32]).
Surprisingly, under these severe conditions the general performance of the pure
alumina ceramics was even more prospective than the behavior of the new sub-
micrometer composites [32].
For SH 1, the manufacturer's data sheet gives a strength of 600 MPa (no available
data for the submicrometer grade). The bending strength of the Ti(C,O) reinforced
composite AT60A is about 800MPa [23],and a lower strength of about 650-
700MPa was observed for the pure submicrometer alumina ceramic [3]. The
fracture toughness of all of the laboratory grades is 3.3-3.8 MPa m1/2, whereas
surprisingly high values of 5.5 MPa m'I2 (SHI) and 6.6 MPa rn'" (submicron com-
mercial composite) are given by manufacturer's information for these references.
None of these data explain the much smaller fracture risk of the submicrometer
laboratory grades on machining hardened steel. Obviously, complex wear-induced
flaw-generation processes will have to be considered to understand the improved
global stability of the new laboratory grade microstructures.
No differences were observed in the roughness of the hardened steel surface
when different ceramics were used. At 'u = 235 rnmin-' ( f = 0.1 mm rev-',
a = 0.2mm), R, was 5 1 pm for the first 10-15min of cut for all of the tested
ceramics, increasing to about 1-1.5 pm after 30 min (K = 45": see comment at
the end of Section 4.4.4.2.1!). On the other hand, the tools showed very different
flank wear and cutting edge displacements. Figure 18 gives the influence of high
cutting velocities on the life of advanced cutting ceramics on machining hardened
steel (flank wear criterion VB = 0.15mmin-'). As already mentioned, 'u = 200-
220 m min-' is the upper limit of reliable use for the latest generation of commer-
cial A1203/Ti(C,N)composites. At this velocity, best results are obtained with the
new submicrometer composites with Ti(C,O) which exhibit the highest hardness of
all of the tested ceramics (see Table 4). At higher cutting velocities, however, the
relationship compared with the pure submicrometer alumina changes, and pure
A1203 again gives the greater advantage. A similar smaller effect of the cutting
velocity on A1203 inserts compared with composites was noticed on cutting hard
cast iron (Fig. 15, Fig. 17).
Contrary to the new ceramics, most commercial submicrometer inserts fractured
within the first 5-10min when used at w = 300mmin-', and the one continuous
wear curve measured up to 15min and displayed in Fig. 19b was obtained at a
softer part of the steel shaft (HRC = 52-57).
At all cutting velocities the new laboratory grades are clearly superior to advanced
commercial inserts. Contrary to a sometimes different behavior of flank wear and
cutting edge displacement on turning hard cast iron, both parameters were well
4.4 New Trend,for Cutting Hard Workpieces 679
correlated on machining hardened steel. Comparing the new submicrometer A1203/

Ti(C,O) composite with submicrometer pure corundum, Fig. 19 shows a velocity-
dependent different ranking of the cutting edge displacement (CED) of these tools
in perfect correlation with the data in Fig. 18 (flank wear); the CED curves for the
Al,O,/Ti(C,O) composite and for pure A1203were observed close to each other at
an intermediate velocity v = 235 m min-' (as suggested by the flank wear data in
Fig. 18).
Comparing this behavior on machining hardened steel with the turning of
hard cast iron, again both the pure alumina tool AC41 and the laboratory grade
composite AT60A exhibited less wear than the commercial cutting ceramics, but
the changing ranking of AC41 and AT60A ceramics is worth some comments. It
is obvious that the technical ranking of both tool ceramics will be affected by
some important basic properties:
(i) For A1203/TiC composites, chemical interactions cause preferential (local)

wear of the covalent carbide constituents of the microstructure even at
room temperature and with almost inert conditions (dry air, fretting wear
against alumina), with the consequence that at equal crystallite sizes the
pure sintered corundum is more wear resistant than the composite [30,31].
The same ranking is observed for the flank wear of ceramic tools on turning
hard cast iron (Fig. 15).
(ii) On machining hardened steel with v = 220 m min-l, flaming chips indicate
much higher process temperatures than on turning hard cast iron at
u = 250 m min-' . Whereas pure submicrometer alumina exhibits intense
creep at temperatures > 1 100"C, it can be assumed that the continuous network
of covalent crystals reduces the creep rates in composites with more than 25 vol-
% of TIC, Ti(C,N) or Ti(C,O). Indeed, comparing at u 5 250mmin-' the
machining of cast iron and steel, the increasing temperature increases the
wear of alumina more than the wear of the best composite AT60A and changes
their ranking: at a similar submicrometer grain size, AT60A exhibits the lowest
flank wear and the smallest cutting edge displacement.
(iii) Of course, this advantage of the composite will be lost when at yet higher tem-
peratures oxidation of the Ti(C,O,N) phases or more intense reactions with the
steel start. With the results in Fig. 18 and Fig. 19, the commercial composites
with Ti(C,N) are more susceptible to such processes than the new laboratory
grades with Ti(C,O). In fact, no information about the specific nature of
such reactions can be obtained from the present results, but it is obvious that
on machining hardened steel an increase of the cutting speed (and of the process
temperatures) causes a more intense deterioration of the wear resistance of even
the best composite AT60A than observed for the pure alumina tool AC41
(Fig. 18, Fig. 19): already at v = 235 m min-' both ceramics exhibit the same
wear, and with u = 300 m min-l the sintered corundum again shows the best
performance.
At a velocity of v = 300mmin-', the increasing with time wear deteriorates the

cutting edge to an extent that obviously causes an increasing input of heat into
680 4 Ceramics Based on Alumina: Increasing the Hardness j o r Tool Applications
the surface of the machined steel. For example, on machining the hardened steel
with the submicrometer composite ceramic AT6OA the hardness of the cut surface
decreased from originally HRC = 59-60 to values of 57-59 (25min) and 55-56
(30 min) [32]. This influence of high-velocity turning on the hardness of workpieces
is important for different aspects in the development of new machining technologies.
On the one hand, the softening of the hardened steel promotes unexpectedly high
materials removal rates. In this way, turning of hardened steel with cutting ceramics
that are much less expensive than cubic boron nitride (cBN) seems possible now. If,
on the other hand, for the final step of precision machining a constant high hardness
of the cut surface is required, critical limits exist for the cutting velocity (about
2 5 0 m m K ' in the investigations described above) or for the tolerable tool wear
(here at ZI = 300 m min-' : VB 5 0.15 mm). These technical demands are, of
course, affected by the choice of the ceramic tool depending on its thermal and
wear properties.
4.5 Summary
Recent developments show somewhat different trends for sintered ceramic grinding
materials and for cutting tools.
(i) In the field of grinding grits, extended innovative work is directed towards a sub-
stantial further increase of the grinding efficiency (high materials removal rates)
and increased tool life within narrow tolerances for the surface quality of the
ground workpiece.
(ii) For cutting tools, innovation is focused on two aspects: the development of very
tough and highly reliable grades for high feed rates and cutting depths (e.g. for
globular and vermicular cast iron), and of extremely hard and chemically resis-
tant microstructures for the precise machining of hardened steel and hard cast
iron.
Apart from temperature dependent transformation toughening (Zr02 additives),
high values of the fracture toughness are obtained first of all by anisotropic crystal
shapes (whisker reinforcement, rod shaped crystals with high aspect ratio in
Si3N4, platelets); generally new mechanism are not likely to be found in the
near future.
Both the demands for advanced grinding tools and for the precise machining of
hard cast iron and hardened steel with cutting ceramics are expected to promote the
further development of powder technologies for the manufacture of highly perfect
submicrometer microstructures for applications in fields where diamond or cBN
tools are too expensive.
For grinding materials, corundum (a-A1203) ceramics are dominant at present.
Here, contrary to other fields of ceramic technologies, precursor (sol/gel) approaches
represent an advanced industrial state of the art. Extended developments are yet in
progress, but, on the other hand, it is just the wide industrial use of these procedures
Rejkrences 68 1
which gives insight to some general limitations. New powder routes revealed such a
surprisingly large increase of the observed grinding efficiency of the grits compared
with sol/gel derived materials, that another leap in the quality has to be expected
similar to the progress enabled ten years ago by the introduction of sol/gel
approaches.
At present, submicrometer cutting ceramics for machining hard workpieces are
developed in two classes of materials. In the industry, first tests in Europe are
known for carbide reinforced composites based on A1203,whereas more fundamen-
tal consideration (Sections 4.4.1 and 4.4.3) and first laboratory results from Asia
and Europe (Section 4.4.4) indicate advantages of new grades of pure corundum
tools for cutting both mild steel and for machining hard cast iron/hardened steel.
Besides of precision machining, first surprising results suggest that pure A1203
cutting tools may be applied for the machining of hardened steel even under
conditions where at present only cubic boron nitride can be used (discontinuous
cut). On turning hardened steel with pure alumina, high materials removal rates
are possible if at high cutting velocities some softening of hardened steel workpieces
can be tolerated in a preliminary machining step.
References
1. A. Raab and W. Zielasko, Schneidstoffeinsatz in der Groaserienfertigung, Lecture at the meet-
ing of the VDI-Division Schneidstoffanwendung, 2 I . 1 I . 1996, Siegmaringen.
2. G. Wellein and J. Fabry, Hartdrehen ist mehr als der Einsatz von CBN, Werkzeug-Technik
1996, 45, 7-10.
3. A . Krell and P. Blank, The influence of shaping method on the grain size dependence of strength
in dense submicrometre alumina, J . Europ. Ceram. SOC.1996, 16, 1189-1200.
4. F. F. Lange, Sinterability of agglomerated powders, J . Am. Ceram. Soc. 1984, 67. 83-89.
5. T. J. Graule, F. H. Baader, and L. J. Gauckler, Shaping of ceramic green compacts direct
from suspensions by enzyme catalyzed reactions, cfi Ber. Dtsch. Keram. Ges. 1994, 71, 317-
323.
6. S. Tashima, Y. Yamane, H. Kuroki, and N. Narutaki, Cutting performance of high purity
alumina tools formed by a high speed centrifugal compaction process, J . Jpn. Soc. Powd.
Powd. Metall. 1995, 42, 1464-1468.
7. A. Krell, Fracture origin and strength in advanced pressureless sintered alumina, J . Am. Ceram.
SOC.1998, 81, 1900-1906.
8. B. J. Hockey, Observations by transmission electron microscopy on the subsurface damage
produced in aluminum oxide by mechanical polishing and grinding, Proc. BY. Ceran?. SOC.
1972, 20, 95-1 15.
9. A. Krell and P. Blank, Grain size dependence of hardness in dense submicrometer alumina,
J . Am. Ceram. Soc. 1995,78, 1118-1 120.
10. S. D. Skrovanek and R. C. Bradt, Microhardness of a fine-grain-size A1203,J . Am. Ceram. Soc.
1979,62, 215-216.
11. A. Krell, Improved hardness and hierarchic influences on wear in submicron sintered alumina,
Matrr. Sci. Eng. 1996, A209, 156-163.
12. A. Krell, Fortschritte in der spanenden Metallbearbeitung durch Keramiken mit Submikrometer-
Gefiige, in Hartstofle, Hartstoflychichten, Werkzeuge, Vrrschle&chutz, Proc. of the Hagen
Symposium for Powder Metallurgy, 13-14 Nov. 1997, R. Ruthhardt (Ed.), Werkstoff-
Informationsgesellschaft, Frankfurt, 1997, pp. 57-76.
13. H. G. Sowman and M. A. Leitheiser, Non-fused aluminium oxide-based abrasive mineral, a

process for its production and abrasive products comprising the said abrasive mineral,
European Patent Application EP-24 099 B1, Int. C1.6 C09K3/14, 25.01.1984.
14. T. E. Cottringer, R. H. Van de Merwe, and R. Bauer, Abrasive material and method for
preparing the same, U S Patent 4 623 364, C09C1/68, 18.11.1986.
15. G. Bartels, G. Becker, and E. Wagner, Process for the production of a ceramic polycrystalline
abrasive, U S Patent 5 034 360, C04B35/10, 23.07.1991.
16. A. Krell, P. Blank, E. Wagner, and G. Bartels, Advances in the grinding efficiency of sintered
alumina abrasives, J . Am. Ceram. Soc. 1996,79, 763-769.
17. A. Krell and P. Blank, Al,03 sintering material, process for the production of said material, and
use thereof, European Patent 756 586, Int. C1.6 C04B35/111, publ. 19.10.1997.
18. N. G. L. Brandt and Z. D. Senesan, Ceramic cutting tool reinforced by whiskers, US Patent
Application 4 867 761, Int. Class. C04B35/117, 10.09.1989.
19. A. G. Evans and B. J. Dalgleish, Creep and fracture of engineering materials and structures, in
Proc. 3rd Int. Con$ at University College, B. Wilshire and R. W. Evans (Eds), The Institute of
Metals, London, 1987, pp. 929-955.
20. T. N. Tiegs and P. F. Becher, Thermal shock behavior of an alumina-Sic whisker composite,
J . Am. Ceram. Soc. 1987,70, C109--C111.
21. J. D. French, J. Zhao, M. P. Harmer, H.-M. Chan, and G. A. Miller, Creep of duplex micro-
structures, J . Am. Ceram. Soc. 1994, 77, 2857-2865.
22. K. Friedrich, S. Lehmann, M. Fripan, D. Klotz, and H. Ziegelbauer, Mischkeramik auf
Aluminiumoxidbasis, European Patent Application 755 904 Al, Int. C1.6 C04B35/117,
29.01.1997.
23. A. Krell, L.-M. Berger, E. Langholf, and P. Blank, Hartstoffverstarkte A1,03-Sinterkeramiken
and Verfahren zu deren Herstellung, Germanllnternational Patent Application DE-196 46 344
Al, Int. CL6 C04B35/117, 14.05.1998 / W0-98/21161 A l, C04B35/117, 22.05.1998.
24. M. Fripan and U. Dworak, Keramische Hochleistungsschneidstoffe in der CUR-und Stahlzer-
spanung, in Schneidstoffe H. Kolaska (Ed.), Verlag Schmid, Freiburg, 1988, pp. 194-214.
25. J. Koike, S. Tashima, S. Wakiya, K. Maruyama, and H. Oikawa, Mechanical properties and
microstructure of centrifugally compacted alumina and hot-isostatically-pressed alumina,
Muter. Sci. Eng. 1996, A220, 26-34.
26. T. Takahashi, Y. Katsumura, and H. Suzuki, Cutting performance of white ceramics tools
having high strength, J. Jpn. Soc. Powd. Powd. Metall. 1994, 41, 33-37.
27. S. Tashima, Y. Yamane, H. Kuroki, and N. Narutaki, Cutting performance of high-purity
alumina ceramic tools formed by a high-speed centrifugal compaction process, J. Jpn. Soc.
Powd. Powd. Metall. 1995, 42, 464-468.
28. G. K. L. Groh, L. C. Lim, M. Rahman, and S. C. Lim, Effect of grain size on wear behaviour of
alumina cutting tools, Wear 1997, 206, 24-32.
29. A. Muchtar and L. C. Lim, Indentation fracture toughness of high purity submicron alumina,
Acta Muter. 1998, 46, 1683-1690.
30. A. Krell and D. Klaffke, Effects of grain size and humidity on the fretting wear in fine-grained
alumina, AI,O,/TiC, and zirconia, J. Am. Ceram. Soc. 1996, 79, 1139-1 146.
31. V. N. Koinkar and B. Bushan, Microtribiological studies of Al,03. AI2O3/TiC,polycrystalline
and single-crystal Mn-Zn ferrite, and Sic head slider materials, Wear 1996, 202, 110-122.
32. A. Krell, P. Blank, L.-M. Berger, and V. Richter, Alumina tools for machining chilled cast iron,
hardened steel, Bull. Am. Ceram. Soc. 1999, 78(12), 65-73.

K. A. Schwetz
5.1 Introduction
S i c is considered by many as the most important carbide, because of its extreme
properties and the variety of present and potential commercial applications.
According to Kieffer [l] silicon carbide belongs to the group of nonmetallic hard
materials, that is materials whose great hardness and high-melting point result
from a high fraction of covalent bonding. Superhard compounds are obviously
formed by combination of the four low atomic number elements: boron, carbon,
silicon, and nitrogen to form a quarternary system (see Fig. 1); carbon as diamond,
boron-nitrogen as cubic boron nitride, boron-carbon as boron carbide, and silicon-
carbon as silicon carbide belong to the hardest materials hitherto known.
5.1.1 History
Berzelius [2] first reported the formation of silicon carbide in 1810 and 1821. It was
later rediscovered during various electro chemical experiments especially by
Despretz [3], Schiitzenberger [4], and Moissan [5]. However it was Acheson [6],
who first realized the technical importance of the carbide as a hard material and,
by 1891, had managed to prepare silicon carbide on a large scale. Originally,
Acheson had completely different things in mind: he wanted to produce artificial
diamonds by recrystallizing graphite in an aluminum silicate melt; at a time when
high-pressure and high-temperature chemistry was unknown. His efforts proved
fruitless, but he was rewarded by the discovery of a far more versatile material,
silicon carbide.
Believing that this was a compound of carbon and corundum, he named the new
substance ‘carborundum’ (Fig. 2). Today, silicon carbide has become by far the
most widely used nonoxide ceramic material, with an annual world production of
about 700 000 tonnes. Owing to its great hardness, heat resistance, and oxidation
resistance it has become firmly established as an abrasive as well as a raw material
for making refractories such as firebricks, setter tiles and heating elements. Another
major use of silicon carbide is as a siliconizing and carburizing agent in iron and steel
metallurgy.
Self-bonded silicon carbide materials in molded form and of high S i c content in
order to make full use of the excellent material properties, have been available only
from the late 1960s onwards. They are being increasingly used as refractory
materials as well as for structural components in mechanical engineering, and
have proved highly successful for use under extreme conditions involving abrasion,
corrosion, and high temperatures.
684 5 Silicon Carbide Based Hard Materials
( raphite)
(%iamond)
Figure 1. Nonmetallic hard materials in the system B-C-Si-N.
5.1.2 Natural Occurrence [7]

Naturally occurring silicon carbide was first discovered by Moissan [8] in a meteor-
ite from the Diablo Canyon in Arizona, and subsequently several small deposits
have been found [9-111. Extraterrestrial occurrences of silicon carbide are even
Figure 2. Crystalline Sic grain from the Acheson process.

5.2 Structure und Phase Relations of S i c 685
rarer [7]. Its existence in interstellar space has been shown by spectral measurements,
and it has also been found in samples of moon rock.
5.2 Structure and Phase Relations of S i c

As shown in the phase diagram in Fig. 3, silicon carbide is the only binary phase in
the siliconxarbon system, with the composition 70.05 weight-% Si, 29.95 weight-%
C. Silicon carbide does not have a congruent melting point. In a closed system at a
total pressure of 1 bar it decomposes at 2830 f 40°C into graphite and a silicon-rich
melt (Fig. 2). This is the highest temperature at which silicon carbide crystals are
formed. In an open system, silicon carbide starts to decompose at =2300"C with
the formation of gaseous silicon and a residue of graphite. S i c and Si form a
degenerate eutectic at 1413°C and 0.02at% C. The solubility of carbon in liquid
silicon is 13 at% C at the peritectic temperature [ 121.
Silicon carbide exists in several modifications being polymorphic and polytypical
and crystallizing in a diamond lattice, like silicon [13].
Polycrystalline silicon carbide obtained by the Acheson process exhibits a large
number of different structures (polytypes), some of which dominate. These can be
classified in the cubic, hexagonal, and rhombohedra1 crystal systems (Table 1).
4000
3826
3232
-
3000
L+C
-
Y
v
-5 1.5 >99.97
2
3
;2000
L
Q) -0.5
? 1414
1413 *1 Si + S i c
-
49.4 e 0.6 /
1000
Si + Sic
0 20 40 80 )0
Carbon (at. %)
Figure 3. Phase diagram of the silicon-carbon system at 1 bar total pressure after Kleykamp and
Schumacher [ 121.
Table 1. Dominant SIC types
Modification Polytype
clSiC (high-temperature modification) 6H

15R
4H
PSiC (low-temperature modification) 3c
Since silicon carbide predominantly exists in its beta form at temperatures below
2O0O0C,this is referred to as the low-temperature modification. Cubic PSiC is meta-
stable and, in accordance with Ostwald' s rule, is formed initially in Sic production
from silicon dioxide and carbon. PSiC (polytype 3C) can also be prepared at about
1450°C from simple mixtures of silicon and carbon or by hydrogen reduction of
organosilanes at temperatures below 2000°C. Above 2000"C, only the hexagonal
and rhombohedra1 types are stable if there are no stabilizing influences which
raise the transformation temperature [ 14,151. Thus, in a nitrogen atmosphere at
high pressure, PSiC is the stable form, so that above 2000°C aSiC transforms to
PSiC. On the other hand, a hexagonal polytype, the wurtzite or 2H modification
of Sic, has been obtained at temperatures as low as 1300-1600°C by the decom-
position methyltrichlorosilane CH3SiCI3 [ 161. This 2H polytype, which does not
occur in commercial Sic, and the most common 3C, 4H, 6H and 15R structures
are called short-period polytypes. Long-period polytypes are much less common
and consist of blocks of short-period polytypes which are broken by regularly
occurring stacking faults.
The basic element of the silicon carbide structure is the tetrahedron [17] due to sp'
hybridization of the atomic orbitals. This tetrahedron consists of a silicon or a
carbon atom at the spatial center, surrounded by four atoms of the other kind.
The Sic- bond is 88% covalent. The tetrahedra are arranged in such a way that
units of three silicon and three carbon atoms form angled hexagons which are
arranged in parallel layers as shown in Fig. 4.
The layer sequences can repeat themselves in the cycles ABC, ABC . . . (zinc
blende, type 3C) or AB, AB . . . (wurtzite, type 2H), according to cubic or hexagonal
close packing. In addition, numerous others stack sequences are formed in the case
of silicon carbide, resulting in many similar polytypes.
About 180 different polytypes are currently known. They all have the same
density, 3.21 g ~ r n - ~
The
. formation and stability of the various polytypes are not
yet clearly understood. Written polytype nomenclature [ 181 indicates the number
of layers in the repeating layer pack by a numeral, while the crystal system is denoted
by the letters C, H, or R.
The amounts of the most frequently encountered polytypes 3C, 4H, 15R, and 6H
can be quantitatively determined by X-ray diffraction techniques [ 191, which can
indicate the temperature of formation of the silicon carbide since the stability of a
given polytype depends on, among other things, temperature. According to Inomata
et al. [20] 2H-Sic seems to be stable below 14OO0C, 3C between 1400 and about
1600"C, 4H between 1600 and 2100"C, 6H above 2100°C and 15R above 2200°C.
5.2 Structure and Phase Relations o j S i C 687
u - S i c l6H)
Figure 4. Crystal structure of the SIC polytypes 3C and 6H.
There is no doubt, however, that impurities also play a part in the formation of the
various polytypes, together with the surrounding gas atmosphere and growth-
kinetic processes. Preferential stabilization of the 4H polytype is achieved with
additions of aluminum up to concentrations of ~ 0 . weight-%, 4 which is thought
to substitute for silicon in the silicon carbide lattice [21-241. According to Lundquist
[21] pure CcSiC mainly crystallizes as 6H, with small amounts of aluminum 15R also
occurs, and above 0.1 weight-% aluminum, 4H predominates (see Fig. 5). Also it is
well known today that nitrogen can be incorporated into the S i c lattice, whereby it
stabilizes not only the cubic PSiC form but also the wurtzite 2H polytype [14,25].
Whereas the nitrogen content obtained in 3C-Sic by gas/solid diffusion is low,
e.g. only < 1 atom YO,extremely high nitrogen contents of up to = 14 atom YOwere
A l content. w t % -
Figure 5. Proportions of polytypes in technical-grade S i c as a function of Al content
688 5 Silicon Carbide Bused Hard Muterials
recently found for nitrogen-stabilized 2H Sic when formed from carbothermal

reduction of silicon nitride, Si3N4,in the presence of a liquid phase [26].
Transmission electron microscopy using high resolution lattice imaging tech-
niques allows polytype analysis within single grains in the microstructure of dense
S i c bodies [27].
5.3 Production of Sic

5.3.1 The AchesonlESK Process
Abrasive and refractory grade silicon carbide is produced industrially according to
the following reaction
Si02 + 3C = Sic + 2CO
from high-purity quartz sand (99.5% SO2) and petroleum coke, in electrically
heated resistance furnaces [28]. The raw materials should have a maximum grain
size of 10 mm. In addition to the reactants, sawdust is sometimes added to increase
the porosity of the mix and to facilitate the venting of the carbon monoxide formed.
The reaction is strongly endothermic: the heat of formation is 618 kJ mol-' Sic, cor-
responding to 4.28 kWh kg-I Sic. The process takes place at temperatures of
between 1600°C and 2500°C and is far more complex than the above equation
because of the intermediate reactions, some of which occur with participation of
the gas phase [29], [36].
Acheson Sic furnaces are up to 25 m long, 4 m wide, and 4 m high, consuming up
to 5000 kVA, and running for about 130 h. The largest furnaces operate at up to
25000A. The furnace has a rectangular cross section and consists of two water-
cooled graphite electrodes at the end and two movable side walls, whose purpose
is to retain the raw material mixture. When about half of the mixture has been
charged to a depth even with the electrodes, a column of tamped graphite powder
or a solid graphite rod is inserted to connect the electrodes. This so-called core is
then covered with additional reaction mixture. The electrodes are then connected
to the transformer, and the current is switched on. The silicon carbide is formed
in the hot core in the form of a polycrystalline, compact cylinder, the so-called
roll. Temperatures within the furnace reach 2500°C causing the silicon carbide to
recrystallize. The surrounding mixture acts as a thermal insulator. The carbon
monoxide byproduct escapes through the surface of the reaction mixture and
through slits in the side walls, where it burns.
After the current is switched of, the furnace is allowed to cool for several days.
Then the side walls are removed, and the excess reaction mixture is carried off.
The roll of silicon carbide contains not only the original resistance core, but also
graphite formed by decomposition of silicon carbide. The boundary between the
graphite and the Sic is very sharply defined. The innermost Sic zone is of the
highest quality, as the purest and largest crystals can form in this region. Moving
5.3 Production of S i c 689
30 m
I- 4
Resistance core
Reaction mixture
Current conductors
Resistance core n / Reaction mixture
/
~
I
Gas collection duct Porous bed
Figure 6. Silicon carbide production by the ESK process [7, 301:

(a) longitudinal and (b) radial cross-sections through the furnace
outwards, with decreasing temperature, the crystal growth rate decreases and the
S i c roll becomes more and more finely crystalline. The outer zone is composed of
very fine crystals of SSiC. While in the case of the classic Acheson furnaces, with
their laterally positioned electrodes and side walls, the carbon monoxide burns
away on the surface, polluting the atmosphere inside and outside the furnace
building, the ESK (Elektroschmelzwerk Kempten) process [30] developed in the
1970s (Fig. 6), is ecologically safe. It operates with floor electrodes, which enter
the reaction mass from below a porous floor, and the carbon monoxide is collected
by a gastight plastic canopy and transported through gas collecting pipelines to a
power station where it is converted into electrical energy after burning. This
energy is recycled and amounts to 20% of the energy necessary for the production
Figure 7. SIC roll at the ESD facility in Delfzijl, The Netherlands.
of silicon carbide. The silicon carbide cylinders inside the linear or U-shaped resis-
tance furnaces are up to 60 m long and weigh about 300 tonnes after a run of 10 days
(see Fig. 7). A comparative review of some Acheson plants has been published [3I].
The lumps of silicon carbide obtained on dismantling the furnace are taken to a
sorting area and broken down with hydraulic equipment and compressed air
hammers. The finely crystalline outer zone (the so-called amorphous zone) is
further size reduced (e.g., in a jaw crusher) and is usually returned to the furnace
process by adding it to the coke/sand mixture. Alternatively, this 9&95% Sic material
is a valuable raw material for the production of grain refiners for the ferrous metal
industry.
The coarsely crystalline compact pieces from the inner zone of the Sic roll are
sorted according to criteria such as color, structure, etc. This top-grade material
forms the largest fraction of the Sic yield from a furnace charge.
Further processing of the silicon carbide into the various macro- and microgrits
depends upon such factors as the required particle size distribution [32] and shape,
and involves breaking, grinding, sieving, air classification, and elutriation.
For applications that require a particularly high grade of silicon carbide, such as
industrial ceramics, the refractory industry, and ceramic-bound abrasives, the
granular product must be washed with alkali or acid to remove adhering traces of
elemental silicon, metals, metal compounds, graphite, dust, and silica.
Commercial Sic grades vary in composition from 90% to 99.5% Sic. Light green
S i c is 99.8% pure, and, as the purity decreases to 99%, the color changes to dark
green, and, at 98.5% to black.
Before aSiC is converted into molded parts by sintering, it must be intensively
milled into the submicrometer range, that is to a mean particle size of less than
1 pm (see Fig. 8b), in attritors [33,34] or ring-gap mills. Adherent impurities
which interfere with the sintering process, such as SiOz or wear from milling
equipment, are removed by wet chemical purification [35].
5.3 Production of' S i c 69 1
Figure 8. Silicon carbide submicron powders:

(a) PSiC powder, specific surface area: 12.6 m2 g-' [40] and (b) CtSiC powder, S = 15.8m2g-' [35].
5.3.2 Other Production Methods

5.3.2.1 PSiC Powder
5.3.2.1.1 Carbothermal Reduction of Silica This process is based on the following
-
equation
Si02 + 3C < 1.900'C PSiC + 2CO.
According to the reaction mechanism proposed by Weimer [36] the reaction takes
place in several steps:
SiO,(S) + C(S) + CO(,)
+ + C02(g)
+
Si02(,) CO(,) + SO(,)

C(S) + C02(g) 2CO(,)
SiO(,) + 2C(,) + +
Sic(,) CO(,).
The initial reaction with formation of gaseous SiO at points of direct contact
between C and Si02 requires temperatures above M 1600°C at atmospheric
pressure (solid/solid reaction). The high rate of SiO(,) formation suggests that the
further two gas/solid reactions make the final carbon reduction of the SiO to Sic
possible once the C/Si02 contact points are consumed. Carbon crystallite diameter
was found to have a substantial influence on the rate of reaction and the size of
synthesized Sic.
This process has been used in the early 1960s to fabricate pure ultrafine PSiC
powders especially suited for the semiconductor and pigment industry [37,38].
During the first step of the General Electric Process [37] a silica gel is formed by
hydrolysis of silicon tetrachloride in an aqueous solution of sugar (sucrose).
Upon dehydratization at 300°C the sugar pyrolizes and an intimate amorphous
carbon/amorphous silica mixture is obtained which is subsequently fired at
1800°C in an inert atmosphere to form PSiC.
Later the process was modified for the fabrication of PSiC sintering powder
[39, 401, using lower reaction temperatures and as starting materials the com-
binations tetraethylsilicate/sugar and fumed amorphous silica/sugar, respectively.
Particle size control corresponding to powder surface areas in the 10-20 m2 g-'
range (see Fig. 8a) was necessary, because ultrafine powders have high oxygen
contents (surface silica), and often low green densities result due to their high surface
area.
Since 1981 a continuous version of the Acheson process has been commercially
operated by the Superior Graphite Company [41]. A mixture of silica sand and
petroleum coke is fed continuously to the furnace and heated electrothermally to
a temperature of 1900°C. The run-off-furnace product is a free flowing grain
generally smaller than 3 mm, comprising agglomerates of PSiC crystallites and
free carbon. The grain is friable and is readily processed to make sized microgrits
and sintering powder by dry milling, wet grinding, classification, and wet chemical
purification (removal of 5 weight-% free carbon).
5.3.2.1.2 Synthesis from the Elements Prochazka [39] synthesized submicron PSiC
powder with a surface area of 7 m2 g-' using direct reaction of high-purity silicon
powder with carbon-black at temperatures between 1500 and 1650"C, but saw
little densification with boron and carbon additions, suggesting that free silicon is
detrimental to densification.
The in situ formation of silicon carbide according to the reaction
Si(l,,) + C(S) = SIC(,)
is mainly used in the manufacture of reaction sintered Sic components. After
shaping a mixture of fine-grained Sic powder and carbon and exposure at high
temperatures to liquid or vapor-phase silicon, the carbon is reacted to secondary
silicon carbide, which bonds together the primary Sic grains. Excess silicon
(about 10vol-Yo) is usually left to fill the pores, thus yielding a nonporous body
(Si-Sic).
5.3.2.1.3 Deposition from the Vapor Phase Submicron PSiC can be continuously
produced by decomposing gaseous or volatile compounds of silicon and carbon
in inert or reducing atmospheres, at temperatures above 1400°C [42]. Particle size
5.3 Production ojsic 693
and morphology depend considerably on the reaction temperature and on the

composition in the gas phase. Powders obtained below 1100°C are amorphous. A
great variety of reactants and several methods of heating, such as d.c. arc jet
plasma [43], high frequency plasma [44],laser [45], and thermal radiation are possi-
ble. However, large scale commercialization of these processes has been limited by
the very high cost of production.
Small particle sizes are produced, 0.01-0.5 pm, with greater than 70% efficiency
using silicon tetrachloride (SiC14) and methane (CH4) or methyltrichlorosilane
(CH3SiC13)as the gaseous precursors in the presence of hydrogen as the plasma gas:
SiC14(,) + CH4(g)--%
H
+
Sic(,) 4HC1(,).
If boron trichloride is used for the boron dopant together with a small excess of the
carbon feedstock, sinterable Sic powders are obtained.
The laser synthesis allows laboratory scale production of high-purity S i c powder
with controlled particle sizes. Using a C 0 2 laser and a mixture of silane (SiH4) and
methane (CH4) or acetylene (C2H2) PSiC particles in the size range 5-200 nm have
been obtained. An advantage of using SiH4, rather than SiC14, as a reactant is that it
has a strong absorption band near the wavelength of the laser (10.6 pm).
Amorphous S i c powders with metal impurity levels below 60 p.p.m. total have
been synthesized by thermal decompostion of a methylsilane gas stream at tempera-
tures above 600°C according to the equation
CH3SiH3 Sic + 3H2
in an induction heated graphite reactor [46].
5.3.2.2 SIC Whiskers

Sic whiskers are discontinuous, monocrystalline hair or ribbon-shaped fibres in the
size range 0.1-5 pm in diameter and 5-100 pm in average length (see Fig. lOa).Their
aspect ratio (length/diameter) ranges from 20 to several 100 and is an important
figure. Sic whiskers are all grown from the gas phase on solid or liquid substrates
under conditions that result in a small defect size (0.14.4 pm) in the whisker.
Such small defect size results in very high strength, of about 9-17 GPa. Whiskers
grown at temperatures below 1800°C are composed of mostly PSiC, those grown
above 2000°C are of clSiC.
According to Shaffer [47] preparation methods for whiskers can be divided into
three routes:
(i) vapor-liquid-solid process (VLS),
(ii) vapor phase formation and condensation (VC),
(iii) vapor-solid reaction (VS).
5.3.2.2.1 The Vapor-Liquid-Solid Process In the VLS process silicon and carbon-
rich vapors (usually CH4, SiO, or SiC14) react at 1400°C to from S i c on a liquid
,,.,.
<
Vapor Feed Temperature = 1400°C Vapor Feed
sio
c+sI CSolid
Sic Clystals
Liquid Catalyst
Supersaturated with Si+C
Solid Catalyst
(Steel 30vrn)
\1
6%
/////// / / / / / / / / / / / /
Substrate (Carbon)
I 77
Figure 9. VLS process for S i c whisker growth (1400°C).
alloy:
+
Si(d) C(d)
1400°C
___t Sic.
(d) dissolved in liquid iron alloy
Microscopic particles of an alloy are distributed on a substrate (graphite) then

exposed to silicon- and carbon-rich vapors. The presence of a liquid catalyst, such
as a transition metal or usually an iron alloy, distinguishes this method from all
other whisker preparation methods.
Although the VLS-method yields near-perfect whiskers and has been investigated
since the early 1960s, it has not been developed to economical commercial practice
up to now.
Figure 9 illustrates the VLS whisker growth sequence. At 1400°C the solid catalyst
melts and forms the liquid catalyst ball. Carbon and silicon from the gas phase
dissolve into the liquid catalyst, which soon becomes supersaturated and solid S ic
precipitates at the interface with the carbon substrate. Continued dissolution of gas
species into the liquid catalyst allows the whisker to grow, lifting the catalyst ball
from the substrate as additional S i c precipitates. These VLS whiskers are typically
larger in diameter (4-6 pm) than those formed by the vapor-solid process (VS-
whiskers). They frequently grow to length of tens of millimeters.
5.3.2.2.2 Vapor Phase Formation and Condensation To grow whiskers by this

process, bulk S i c is vaporized by heating to very high temperatures (>22OO0C)
usually under reduced pressure. Upon cooling aSiC whiskers form on nucleation
sites.
Sublimation
Sichulk A Sicwhisker
5.3 Production of Sic 695
The addition of lanthanum, yttrium, neodymium, or zirconium leads to an increase

in grow rates. However, today no whiskers are commercially produced by this
sublimation process.
5.3.2.2.3 Vapor-Solid Reaction At present the principal commercial method for

SIC whisker production is the carbothermic reduction of low cost silica sources at
temperatures of 1500-1 700°C. The reaction for the formation of VS-cubic PSiC
whiskers occurs in two steps:
SiOz + C + CO T,
-
-+ SiO
SiO + 2C <18oo-c DSiC + CO .
Many different VS-processes are known according to the starting raw materials and
catalysts species involved. In all cases, a catalyst is used for whisker formation to be
favored versus particulate S i c formation. Commercially available VP-whisker
grades are typically less than 1 pm in diameter (submicron) and up to 30 pm in
length and show strength/Young modulus values of 1-5 GPa/400-500 GPa.
During whisker handling and processing, rigorous protection has to be provided
since smaller whiskers (< 3 pm diameter) can become lodged in the lungs and
represent a health hazard. An inexpensive VS-process utilized calcined rice hulls,
which contain both silica and carbon, as a precursor to S i c whiskers.
The characteristics of commercial S ic whiskers as claimed by producers are given
in Table 2.
PSiC whisker have been added to a wide variety of ceramics (A1203,Si3N4,AlN,
MoSi2, mullite, cordierite, and glass ceramics) in an effort to achieve increased
toughness. In most cases, hot pressing has been required to achieve near-theoretical
density of the matrix. A real success story in the advanced ceramic market was the
introduction of a 25-30 weight-% S i c whisker reinforced hot-pressed alumina for
the high-speed cutting of high nickel content alloys. Thus, by incorporation of
S i c whiskers into the alumina matrix a composite with increased strength (by
over %YO),fracture toughness (by loo%), increased fatigue and thermal shock
resistance was obtained [48]. In this composite, whisker pull out and crack bridging
have been identified as the major toughening mechanisms. Crack bridging consists
in bridging the crack behind the crack tip by unfractured whiskers or fibers. The
effect depends on the strength of the fiber and on its aspect ratio. Inhibiting crack
Table 2. Comparison of the properties of PSiC whisker (supplier's data).
Supplier Tokai Tateho Shin C-Axis Millenium ART

Carbon Etsu Technology Materials Inc.
Phase
Length (pm)
P
20-50
P
5-200
a
3-20
P P
30-200
P
Diameter (pm) 0.3-1.0 0.05-1.5 0.15-0.25 0.3-1.0 1-3 0.5-1.7
Aspect ratio 20-200 20-50 20-50 5-20
Density (gcm-') 3.2 3.18 3.21 3.21 3.21
Table 3. Morphological characteristics of commercial S i c platelets.
Supp1ier C-Axis Technology Millenium Materials Carborundum

SF grade 325 mesh Corporation
Maximum dimension (pm) 5-50 5-70 10-500

Thickness (pm) 2-5 0.5-5 1-5
Aspect ratio 8-10 8-1 5 10-1 5
opening reduces the stress intensity at the tip of the crack and stops or slows crack
propagation.
The fiber or whisker pull out has the potential for the greatest toughening effect
and consists in a debonding of fiber from the adjacent matrix consuming large
energy. This mechanism depends strongly on the aspect ratio of the second phase
and is favored by weak interface bonds.
Sic whisker also can be combined with metals (MMCs, metal matrix composites)
to increase the high temperature strength of the material as well as provide a
comparable substitution for heavier traditional materials, such as steel. MMCs
can be fabricated by infiltrating a Sic whisker preform with aluminum or by
addition of S i c whiskers to molten aluminum.
5.3.2.3 Sic Platelets

As a reinforcement material for all matrix materials, ceramics, metals and plastics, Sic
platelets offer a similar potential as whiskers but at lower cost and without health
hazard. Platelets are single-crystal, plate-like aSiC particles with an aspect ratio of
about 8-15. Sic-platelets typically range in size from about 5 to 100 pm in diameter
and 1 to 5 pm in thickness (see Table 3 and Fig. lOb).They are commercially produced
from inexpensive raw materials (silica and carbon, micron-sized PSiC powders) at
temperatures of 1900-2100°C under inert atmosphere [49]. Due to the presence of
boron and aluminum dopants added, platelet-shaped crystals are formed. Aluminum
enhances the growth in the [OOOl] direction and decelerates the growth perpendicular
to the [OOOI] direction. Boron enhances the growth notoriously perpendicular to the
[OOOl] direction [50]. As impurity elements in Sic platelets from Millenium Materials,
aluminum of 0.0440.45weight-%, boron and nickel of 0.44.8 weight-% (each), and
free silicon of 0.3-3.6 weight-% were analyzed [51]. With the aid of high resolution
AES most impurity elements were found to be present on the outermost surfaces of
the platelets, for example boron was found as about a 3 pm thick layer of B203,as
about 1 pm thin layers consisting of boron or silicon borides and as 3-5nm thick
segregations near to growth steps [51].
Dense Sic platelet reinforced silicon carbide composites were recently fabricated
[52]. The highest sintered densities of 97-98% TD were achieved with 20% platelet
contents using hot molding as the thermoplastic forming technique [53].
In addition to reinforcing ceramics, Sic platelets are also used to increase the
strength, wear resistance and thermal shock performance of aluminum matrices,
and to enhance the properties of polymeric matrices.
Figure 10. (a) PSiC whisker, and (b) CtSiC platelets

(Courtesy S. Weaver, Millennium Materials, Inc., Knoxville, USA).
5.3.2.4 SIC Fibers

Together with boron filaments and carbon fibers, Sic fibers are important continu-
ous inorganic reinforcement materials with high modulus. Owing to their high-tem-
perature properties and the resistance to oxidation they are particularly well suited
for structural parts where high stiffness at high temperatures is required. Three dif-
ferent types of S i c fibers are now available: substrate-based fibers (CVD-fibers),
polymer-pyrolysis derived fibers (PP-fibers), and sintered powder derived fibers
(SP-fibers). Defect size in amorphous or polycrystalline fibers is very small so that
strength can be high. In Table 4 some commercially available Sic fibers are listed
along with their mechanical characteristics.
5.3.2.4.1 Substrate-based Fibers (CVD-Fibers) PSiC filaments of 100-150 pm

thickness can be prepared by chemical vapor deposition onto tungsten or carbon
Table 4. Mechanical Characteristics of various S i c fibers
Supplier Textron Nippon Carbon UBE Carborundum

SCS-6 Nicalon Tyranno Sintered-Alpha
Crystal phase PSiC PSiC amorphous CcSiC

Diameter (pm) 140 10-20 8-10 20-1 50
Tensile strength (MPa) 3920 > 3000 > 3000 up to 1380
Young modulus (GPa) 406 > 420 > 200 400
monofilaments of 40 pm thickness, which act as hot substrates during heterogeneous

nucleation. Various carbon containing silanes have been used as reactants. Owing to
high process costs and the large diameters of the resulting fibers, the method is disad-
vantageous. It has been found that faster Sic filament production is possible on a
carbon filament substrate. The US company Textron produces PSiC fibers on a
carbon-fiber-core with a surface layer of pyrolytic carbon, which is itself coated with
silicon carbide. These fibers are given the designation SCS. The surface carbon layer
provides a toughness-enhancing parting layer in composites having a brittle matrix.
5.3.2.4.2 Polymer Pyrolysis Derived S i c Fibers (PP-fibers) As shown 1976 by

Yajima [54], PSiC fibers with a smaller diameter (8-30 pm) and without a central
core can be manufactured by solid state pyrolysis of a polycarbosilane (PCS) precur-
sor fiber. The melt-spun PCS fiber is first cured at 200°C in air to produce a thin
layer to protect from melting later on, then heated up in inert atmosphere to
1500°C to convert the PCS in crystalline PSiC. The steps leading to the production
of Sic can be summarized as follows:
(-Si(CH3) - CH2-),
I
-
1300°C
PSiC + CH4 + H2.
H
Dimethyldichlorsilane is dechlorinated with metallic sodium in the presence of
xylene to polydimethylsilane (R2Si),, which is converted to polycarbosilane (PCS)
by polycondensation with introduction of silicon-carbon bonds into the chain.
The pyrolysis of the PCS resin to PSiC is accomplished by a slow heating to
1300°C and is accompanied by an appreciable loss of mass (=60%) and consider-
able shrinkage. This melt spinning-curing-pyrolysis route has been adopted by
Nippon Carbon to produce a fiber called Nicalon [55] which has a composition
almost identical to Si3C40thus, it is not pure Sic, but a Si-C-0 composite, in
which nanometer-sized PSiC, Si02 and carbon are uniformly dispersed. These
impurities affect the thermal stability of the fiber and its strength, which has been
reported to degrade above 1200°C. However, these fibers have successfully been
used to reinforce aluminum, refractories and Li-Al-Silicate glass ceramics. The
coating of fibers, for instance by carbon, allows a decrease in the interface bonding
with the matrix, which increases the strength and the toughness by favoring pull-out
mechanism.
UBE Chemicals synthesized amorphous Si-Ti-C-0 fibers from the PCS-titanium
alkoxide compound polymer. These so called Tyranno fibers show excellent proper-
ties and can be spun thinner than the Nicalon fibers (see Table 4). However heating
above 1000°C results in a crystallization of the fiber. In the mid 1990s the German
company Bayer AG synthesized an amorphous Si-B-N-C fiber, by pyrolysis of a
polyborosilazane polymer [56]. This SiBN3Cfiber (see Fig. 11) has a tensile strength
of 3 GPa and maintains its amorphous character up to 1800°C.The advantage of the
production route from liquid to solid to produce S i c has also attracted attention for
5.3 Production of sic 699
Figure 11. SEM of SiBN3C fibers made by Bayer AG, Germany (courtesy H.P. Baldus, [56]).
the S i c film production in microelectronics or as protection layers. Amorphous, and

polycrystalline films of high purity produced by dip-coating of substrates in PCS
solutions and subsequent pyrolysis in inert gas atmosphere have been prepared [57].
5.3.2.4.3 Sintered Powder Derived aSiC Fibers (SP-fibers) Recently, the Carbo-
rundum Company developed a single phase polycrystalline aSiC fiber having a
diameter of approximately 20-1 50 pm and density of over 96% of theoretical. It
is made by first producing green filaments by melt extrusion or suspension spinning
of plasticized mixtures of sinterable aSiC powder with organic additives, such as
polyethylene or polyvinylbutyral plus novolac, respectively. The green filaments
are subsequently debindered during free fall from the extruder in presintering
furnaces and finally undergo pressureless sintered at 2 100°C in argon atmosphere.
These polycrystalline aSiC fibers are stable up to 1600°C in air and 2250°C in inert
gas, far superior than most of the commerciallly available PSiC based fibers [49].
According to Prochazka [%],an increase in the strength of aSiC fibers is obtain-
able by using refined aSiC powders, having a surface area greater than
20m2gp', a median S i c particle size <0.25pm, and no particles larger than
1.5 pm.
5.3.3 Dense Sic Shapes

S i c can be converted into more or less dense parts either by using various bonding
phases to bind the SIC particles or grains together a t elevated temperatures
without dimensional change of the initial compact, or by sintering (pressureless,
sinter/HIP or hot pressing), that is, by mechanisms of mass transport between
particles which lead to porosity elimination together with shrinkage of the overall
body.
5.3.3.1 Ceramically Bonded Silicon Carbide

This product group is based on oxide- or nitride-bonded S i c grains and exhibits an
apparent porosity of up to 20% by volume. A fracture surface of a silicon nitride
bonded Sic is shown in Fig. 12a.
Coarse silicon carbide powder fractions are mixed with clay, or pure oxides like
Si02 and/or A1203,formed to 'green' shapes by conventional ceramic methods, and
fired at about 1400°C.The fired clay or oxidic raw materials bind the silicon carbides
particles and, although these are only loosely bonded, this is sufficient for many
types of applications such as refractory furnace bricks and abrasive disc. Firing mix-
tures of silicon carbide particles and elemental silicon in a nitrogen atmosphere, with
or without the addition of air and possibly A1203produces silicon nitride-, silicon
oxynitride- or SiAlON (silicon aluminum 0xynitride)-bonded Sic products with
enhanced high-temperature strength [59-6 11.
5.3.3.2 Recrystallized Silicon Carbide

Silicon carbide shapes with a high green density made by slip-casting or press-mold-
ing silicon carbide powders of bimodal particle size distribution are fired in an elec-
tric furnace at temperatures of up to 25OO0C, under the exclusion of air [62,63].
Evaporation and condensation [64] take place at temperatures above 2 100°C,result-
ing in a selfbonded structure without shrinkage. The initial (green) density and final
density remain the same and solid S i c bonds are produced between the crystals. The
final RSiC product, which shows transgranular fracture (see Fig. 12b), consists of
100% Sic and can have densities of up to 2.6gcm-', with a porosity of approxi-
mately 20%.
5.3.3.3 Reaction-Bonded Silicon Carbide

In reaction sintering or reaction bonding silicon carbide is produced to some extent
by chemical synthesis in the molding. A preform consisting of silicon carbide,
carbon, and a carbon-containing binder is made by one of the normal ceramic
shaping techniques (dry press molding, slip casting, extrusion, or injection molding,
see Table 5). The elemental carbon in the preform is then reacted with gaseous or
liquid silicon. The process can be controlled in such a way that a porous but pure
Sic part is produced, or one whose pores are filled with excess silicon [65-731.
Since the latter (SiSiC), which is porefree dense (high density, showing no residual
porosity, see Fig. 12c), is much stronger it is this which is most frequently made,
usually by infiltration with liquid silicon. Silicon infiltration can also be used to
fill the pores of recrystallized Sic moldings at a later stage [74].
The amount of free silicon in infiltrated silicon carbide is usually between 10%
and 15%. It is difficult to bring this below 5%. Infiltration is relatively easy, because
of the good wettability of the material, although there are problems if the bodies are
large or thick, because of the exothermic reaction and the expansion of silicon on
freezing, which make perfect impregnation difficult.
Reaction sintering can, however, usually be controlled so that the change in
volume is negligible, resulting in dimensionally accurate components. This is
5.3 Production of’ S i c 701
Figure 12. Microstructure of various S i c ceramics; (a) KSiC, nitride bonded S i c (AnnaSicon 25);
(b) RSiC, recrystallized Sic (AnnaNox CK); (c) SiSiC silicon infiltrated B C (AnnaNox CD), (d)
SSiC, BjC doped (Ekasic F), (e) LPSSiC, equiaxed SIC grains (Ekasic T), (f) LPSSiC, platelet
S i c grains (ESK), (g) HPSiC,Al-doped (ESK), (h) HIPSIC, undoped (ESK).
Figure 12. Continued

102
5.3 Production of’ S i c 703
Figure 12. Continued
achieved by choosing the right S i c and carbon contents in the initial mix, and ensur-
ing correct porosity in the preform and by proper temperature control.
5.3.3.4 Sintered Silicon Carbide
5.3.3.4.1 Solid-state Sintered S i c Once it was thought that silicon carbide was
unsinterable. It was only in the early 1970s that it became possible to sinter silicon
carbide without applied pressure and thus to achieve densities more than 95% of the
theoretical density. This was largely due to the pioneering work of Prochazka [75] at
the General Electric Corporation Research Center in Schenectady. The starting
materials were submicrometer PSiC [76] or later aSiC [77] with the simultaneous
addition of up to 2% carbon and boron. A minimum of 0.3 weight-% boron was
required, although larger additions did not improve the density, which Prochazka
attributed to the need to exceed the solubility limit of boron in S i c (0.2 weight-%
according to Shaffer [78]). Carbon is added in excess to ensure complete deoxida-
tion, that is removal of the Si02 film from the S i c grains, which in turn leads to
the presence of carbon inclusions. Additions of excess carbon are observed to inhibit
104 5 Silicon Carbide Bused Hard Materials
Table 5. Shaping techniques for manufacture of S i c preforms (green bodies).
Moulding Diagram function Feasible molds, Type of tool, Yield for

process scheme recommended tool life average parts
dimensions in pieces/8 h
Automatic or Pellet-like parts in Cemented 5000 pieces

dry pressing pressing directions not carbide molds using single
higher than three times more than lo6 mold of
the diameter; hole pieces average
distances not less than dimensions
0.2 diameter; hole
diameter > 0.1 height of
part
Extrusion Extrusions which can be Extrusion die Approx.

cut to size, diameters made of steel, 300 kg of
1. 540 mm wall minimal wear compound
thickness > 1 mm,
lengths up to 1OOOmm
Slip casting Complex shapes with Plaster molds 1-10 castings

unit weights of up to 5 kg from wooden per mold,
and measuring up to patterns, up depending on
400 mm to 40 castings wall thickness
depending on
mold
Injection Two-part molds, Usually About 500

molding suitable for unit weights multicavity pieces per
of up to 250 g molds made single-cavity
of steel, up to mold
50 000 shots
Isostatic Preferably axially Simple rubber About 15

pressing and symmetric bodies, molds isostatically
dry grinding maximum diameter press molded
320 mm, maximum parts
length 800 mm
grain growth, and produce a more equiaxed grain structure [79],probably as a result
of the energy required from the grain boundaries to overcome the carbon inclusions.
Instead of boron and its compounds [80-831 one can also use aluminum and alumi-
num compounds [84-901, or beryllium and its compounds [91], the net result being
the same. Sintering of AljC doped [92] or A1203-doped[90] Sic powders is believed
to occur in two stages: transient liquid phase sintering; then solid state sintering
during the final stage.
This opened up an inexpensive method of producing dense and complex parts
consisting of pure silicon carbide (SSiC). The powders (see Fig. 8) can be molded
into a green body by any of the methods used in ceramic molding [93,94], depending
on the shape required and the number of pieces involved (see Table 5). Sintering is
carried out in an inert gas atmosphere or in a vacuum at temperatures between

1900°C and 2200°C.
Linear shrinkage exceeds 15% and depends on the green density of the molding.
During the sintering process, there is Sic-polytype transformation as well as grain
growth, whose extent depends on the type and amount of sintering additives, as well
as the sintering temperature. Boron-doped PSiC powders tend towards secondary
recystallization (exaggerated grain growth [95] during 3C + 6H polytype transfor-
mation) whereas, if one starts with CtSiC powders, a homogeneous, fine-grained
microstructure can be obtained (see Fig. 12d). clSiC powder, which predominantly
consists of the 6H-type, usually shows 6H + 4H polytype transformation during
sintering with aluminum, as well as boron containing dopants. Aluminum promotes
6H + 4H transformation at lower sintering temperatures (2050"C), boron most
easily at higher temperatures (2200°C). According to investigations of Schwetz
et al. [96] there seems to be no fundamental correlation between Sic-polytype trans-
formation and microstructure development during sintering of S i c (see Table 6).
Improved reliability in sliding bearing and rotating seal application have been
achieved with microstructure-taylored solid-state sintered S i c materials, which
contain pores [97-1001, a combination of pores and graphite [loll, and/or very
large grained S i c crystallites [ 1021.
5.3.3.4.2 Liquid-phase Sintered S i c Liquid-phase sintered S i c (LPSSiC) is a very

new, dual-phase silicon carbide ceramic, with typically M 95% S i c content and
ultra-fine microstructure (grain size M 1 ym, see Fig. 12e). Y203-A1203 [103-1081
or Y2O3-A1N [109-1111 based compositions were found to be the most effective
densifying aids for even standard grade S i c powders and a pressureless- or gas-
pressure sintered material can be produced with properties matching the best hot-
pressed silicon carbide (see Table 8). Compared with SSiC, LPSSiC exhibits an
increase of up to 100% in fracture toughness to 6.0 MPam1I2 and an increase of
70% in flexural strength to 730MPa. Microcracking at the SIC-YAG interface
ahead of a propagating crack is proposed as the dominant toughening mechanism.
The major part of the secondary phase is YAG (yttrium aluminate, Y3Al5OI2)
which crystallizes together with reprecipitated S i c on solidification from the
liquid [112]. The fracture toughness of LPS S i c has been recently further improved
by anisotropic grain growth, that is growth of larger, platelike S i c grains (in situ
toughening) in a matrix of finer equiaxial S i c grains [113, 1141, see Fig. 12f.
Although LPSSiC may not be suitable as an universal tribology material like
SSiC, the initial assessment of properties suggests that it will be suitable for high-
stress engineering components and in many wear applications, where strength,
reliability, and toughness are demanded.
5.3.3.5 Hot-pressed Silicon Carbide

Dense parts made by hot-pressing are generally considered to have a very fine micro-
structure (see Fig. 12g) and the best mechanical properties. Completely pure silicon
carbide does not exhibit any noticeable plastic behavior up to its decomposition
temperature and can therefore be densified only under diamond synthesis conditions
4
Table 6. Microstructure and Sic-polytype phase composition in sintered silicon carbide (B/C, AljC and B/Al/C doped SSiC materials).
s?
Powder T Sintered body
b
Sample SiC- Addition to Sintering Chemical analysis Polytype analysis Micro- ir!
3
no. modification SIC (weight-%) temperature (weight-%) (weight-%) structureb 0
(colour) ma A1 B 6H 15R 4H 3c n
Q
1 P 1B,2C 2050 - 1.1 13.3 6.4 1.2 79.1 FG 5
2 P 2.5 AIN, 2 C 2050 1.o - 3.9 8.9 74.6 12.6 FG f
tn
3 a-black 2 c 2050 0.44 - 5.6 1.5 92.9 - FG
-
e2L
4 a-black 1 B, 2C 2050 0.65 0.86 - 6.5 93.5 FG
- - Q
5 a-black 2 AIN, 2 C 2050 1.34 - 2.9 97.6 FG QQ
- -
6 a-green-I 1.2 B, 2C 2050 - 1.12 100 - FG
7 a-green-I 0.2 B, 2 AlN, 2 C 2050 1.03 0.22 73.2 .- -
%
26.8 FG f?
8 cc-green-I 0.2 B, 1 Al, 2 C 2050 0.38 0.21 92.6 - 7.4 - FG T
- - 6'
9 a-green-I 0.2 B, 2C 2050 - 0.25 100 - FG 6
-
10 a-green-I1 1B,2C 2050 0.92 83.5 1.3 15.1 - FG
11 cc-green-I1 1 B,2C 2200 - 0.93 40.9 14.3 44.8 - AGG
12 a-green-I1 1 AIN, 2 C 2050 0.49 - 85.9 2.1 12.0 - FG
13 cL-green-TI 1 AIN, 2 C 2200 0.45 - 83.2 1.2 15.6 - FG
14 a-green-I1 2.5 AlN, 2 C 2050 1.20 - 63.7 5.3 31.0 - FG
-
15 a-green-I1 2.5 AIN, 2 C 2200 1.29 48.8 7.0 44.1 - FG
16 a-green-I1 1 Al, 2 C 2050 0.76 - 76.1 2.4 21.4 - FG
-
17 a-green-I1 1 Al, 2C 2200 0.74 68.5 3.2 28.3 - FG
(a) 1 bar argon, graphite-tube-resistance furnace, 15 min hold
(b) FG . . . FINE GRAINED AGG . . . ABNORMAL grain growth
Characterization of Starting Powders (Specific Surface Area: 15-20 m2 g-I):
P carbotherm. red. silica, 100 w/o 3 C, 20 p.p.m. Al-content.
-
a-black Milled Acheson SIC, 89.9 w/o 4 H - 2.5w/o 15 R 7.6w/o 6 H, 0.71 w/o Al -
a-green-I Milled Acheson - SIC, 100 w/o 6 H; 30 p.p.m. Al

-
a-green-I1 Milled Acheson - Sic, 6.5 w/o 4 H 6.4 w/o 15 R - 87.9 w/o 6 H; 200 p.p.m. Al
(35 kbar, 2300°C) to 100% of the theoretical density [115]. In axial hot-pressing with
graphite tools it is, however, necessary to use small amounts of sintering aids: 0.5-
3% boron, aluminum, aluminum oxide, beryllium oxide, yttrium oxide, or tungsten
carbide are added to fine uSiC powder before molding [26, 116-1231. The molding
temperature is 1900-2000°C, depending on the particle size of the powder and the
amount of sintering additives used. The pressure that can be used is limited by
the strength of the graphite, the maximum being 50 MPa. The conventional
method requires large amounts of energy and mold material and can normally be
used only for parts with simple, uncomplicated shapes. Owing to this limitation, pre-
cision parts can only be produced by machining with diamond tools [124] these, of
course, are expensive.
A more convenient method than conventional hot-pressing, also still more
expensive, is hot-isostatic pressing (HIP). During the development of this tech-
nique [ 125-1271 it proved possible to be prepare high-purity S ic products with
silicon carbide contents of more than 99.5% by hot isostatic pressing of S i c
powder or SIC preforms in vacuum-sealed casings to a final density exactly equiva-
lent to the theoretical density and with an uniform, fine-grained microstructure (see
Fig. 12h). Owing to the higher isostatic pressure of 2 kbar (argon gas), compared
with normal hot-pressing, no sintering aids need to be added; that is, the thin film
of silica on the silicon carbide particles (which represents residual oxygen impurity
of the S i c powder) is sufficient to achieve complete consolidation.
The postdensification of SSiC or LPSSiC by hot isostatic pressing is less compli-
cated because no gastight encapsulation is necessary. Not only is it possible to
achieve more than 99% of the theoretical density, but the variation of density
and strength can also be reduced. Pressureless sintering and hot isostatic pressing
can now be performed in the same cycle (see Fig. 13). This new process, called
sinter-HIP, offers economic benefits [ 128-1 301.
The importance of hot-pressed silicon carbide has decreased considerably since the
introduction of pressureless sintering, although it is currently still the most suitable
method of obtaining the best mechanical properties for pure, monolithic SIC.
5.3.3.6 Chemical-Physical-Vapor-DepositedSic
CVD-SIC is obtained by the chemical reaction of volatile silicon- and carbon-
containing compounds in the presence of hydrogen and in the temperature range
1000-1800°C, e.g.
CH3SiC13 2 SIC + 3HC1.
Besides the formation of S i c powders, S ic fibers, and S i c thin or thick films, nowa-
days monolithic bodies in sections up to 25 mm thick can be prepared by this method.
CVD-SIC is of high purity (99.999%), cubic p-modification, very fined grained,
and shows the lowest of oxidation rates. Bulk cubic phase CVD-SIC is produced
for applications such as electronic packaging, components for wafer handling,
kiln furniture for diode manufacture, laser optics for high temperatures, and
substrates for computer storage media [131, 1321. It can be polished to variations
of less than 0.3 nm. Coatings of CVD-SIC were initially developed in the 1960s
708 5 SiEicon Carbide Based Hard Materials
I
I
Shaping 1
Encapsul.
- HIP Containerless-HIP
Figure 13. HIP processing routes (courtesy R. Oberacker, IKM-Karlsruhe/Germany, [ 1301).
for nuclear fuel particles employed to reduce the diffusive release of metallic fission
products from the fuel kernel.
Today Sic thin films (< 1 pm thickness) can also be produced by physical vapor
deposition (PVD), for example by sputtering, which method allows lower substrate
temperatures, but works more slowly. Electrically conductive B/N-doped sintered
clSiC with up to 9weight-% free carbon has been developed as target material,
[133]. Novel applications for PVDSiC include films for computer storage media,
protective coating for lenses, and microwaveable packaging for food.
5.3.3.7 Sic Wafers

Since 1990 high quality S i c wafers of 35 mm diameter are commercially available
from single crystal 6H-Sic bodes, produced via a seeded-sublimation growth tech-
nique [ 1341. In this process, nucleation occurs on a Sic seed crystal located at the top
or bottom of a cylindrical growth cavity. As in the Lely process [135], S i c sublimes
from a polycrystalline source at temperatures > 2200°C under vacuum to form Si,
Si2C, and Sic2 vapors
The gases diffuse through a porous graphite retainer and along carefully programmed
thermal and pressure gradients. The primary gaseous species of silicon reacts with the
graphite walls of the growth cell to from additional Si2C and Sic2 which recombine
on the growing crystal according to a ‘double condensation’ reaction
2SiC2 + 2Si2C = = 6H Sic.
Hence the silicon plays the role of a carbon-transporting agent. Water-clear boules
of the pure 6H polytype having diameters > 50 mm and lengths > 60 mm have been
grown by this technique. It is believed that by the year 2001 the size of the boules will
have increased to 15cm in diameter.
Based on the growth of epitaxial thin films of single crystal polytypes on boule-
grown substrates, S i c is now becoming the material of choice for high-power,
high temperature (> 5OO0C), and high-frequency devices [ 1341.
As a spin-off, currently colorless S i c gemstones, 1/2 to 1 carat in size, cut from
6H-Sic wafers are entering the jewellery market at about 10-1 5% of the price of dia-
monds.
5.3.3.8 S i c Nanoceramics
Sintering nanosized powders with the aim of flaw avoidence from nanosized sintered
bodies (grain size 5 100 nm) is recognized as a promising way of improving mechan-
ical properties and reliability of S i c ceramics. However, nanosized S i c powders are
not easy to process [42,45,136], and several difficulties must be overcome: the
powder flows badly, exhibits low oxidation resistance [40], has low filling and
compaction density [40] and is currently too expensive for large scale use.
5.3.3.8.1 Fabrication by Solid State Sintering In 1991 Vassen et al. [ 1371showed that
polycrystalline PSiC bodies with a density of at least 95% of the theoretical density
and a fine grain size of 150nm can be prepared by encapsulated HIPing of BjC-
doped laser-synthesized powders with particle sizes below 20 nm. Necessary HIP-
temperatures were 1500°C, which is 250°C below the temperature needed to densify
conventional submicron powders. This fabrication process was later optimized
[138] by preheating the shaped S ic bodies in vacuum during an annealing step
before encapsulation. Owing to this additional annealing step, residual oxygen
contents were minimized and a mean grain size of only 60nm could be obtained in
the HIPed Sic shapes. It was found that the reduced final grain size has a strong
effect on mechanical properties; as a consequence of grain size reduction from 1 pm
to 150nm the Vickers hardness increased from 2000 HVlO to 2500 HV10, whereas
fracture toughness decreased from 4 to 3 MPamI”, respectively [139]. However, by
adoption of a bimodal grain-size distribution, with introduction of larger sized S ic
grains into a nano-sized matrix, an incrase in fracture toughness to 6 MPa m1I2was
achieved [140,141]. For a fine grained HIPSiC (300nm) even under a stress of
100 MPa at 1600°C very moderate creep rates of 1 x s-’ were measured [142].
5.3.3.8.2 Fabrication by Liquid Phase Sintering The preparation of PSiC nano-

ceramics with an average grain size of 110 nm by liquid phase sintering has been
710 5 Silicon Carbide Based Hard Matevials
demonstrated by Mitomo et al. [143], who were subsequently awarded a patent

called ‘Superplastic S i c Sintered Body’ [ 1441.
Ultrafine PSiC powder with an average particle size of 90nm was axially hot-
pressed with additions of A1203,Y203,and CaO at 1750°C. The S i c nanoceramic
showed large deformation with high strain rate: 5.0 x lop4sC’ at 1700°C. On the
basis of their results [ 1431, the maximum temperature and the minimum deforma-
tion rate for nano-SSiC might be defined as 1800°C and 10-4sC’, respectively.
The superplastic deformation at temperatures as low as 1700°C is based on the
fine-grained microstructure and the presence of a glassy phase at grain boundaries.
Thus a new technology has been developed whereby nano-Sic parts can be
subjected to plastic deformation as in the case of metals, and can be made into
complicated shapes with near net-shape quality, that is without the need for an
expensive postprocessing stage such as diamond machining.
5.3.3.9 Sic-based Composites

Alloying of S i c is and has been done basically for two reasons: either to improve
properties (toughness, wear, etc.) by the formation of tailored composites/solid
solutions or to improve processing.
Improvements in processing can occur: (1) in solid state sintering, where the
second phase acts simultanously as a sintering aid for Sic, accelerating material
transport by grain boundary and/or lattice diffusion; (2) in reactive liquid sintering
due to reduced sintering temperatures (‘transient liquid phase sintering’).
In the latter case S i c and/or additions are reacted to an intermediate liquid which
not only provides densification at reduced temperatures but since it is consumed in
the reaction yields a Sic-based material without glass at the grain boundaries.
Fabrication of S i c composites by second-phase dispersion is widely applied to
improve material toughness. The various toughening mechanisms [93] that have
high potential to reduce crack extension in Sic- composite materials are: crack
deflection, microcrack formation, crack bridging by reinforcement with metallic
ligaments (e.g. TIC, TiB,), and crack bridging and pull out by platelet- or fiber-
reinforcement .
Tensile fracture in Sic-based composites will only occur after a large enough load
is applied to exceed the compressive stress in the process zone formed by cracking
mechanisms along the crack path (see Fig. 14).
To achieve increased crack deflection and crack-wake interaction in S i c the
microstructure can be modified in various ways:
~ by addition of a second phase with an elongated or fibrous grain structure (SiC-
based composites),
~ by reinforcement with S i c platelets [ 521 or continuous Sic-fibers,
- by inducing growth of elongated S i c grains (in situ toughening of LPS-Sic
[ l l l , 113, 1141).
For optimum toughening by crack deflection Telle et al. [145] pointed out that
geometric factors like grain size, volume fraction, orientation and morphology of
the added or in situ grown phases as well as the grain boundary strength have to
5.3 Production of S i c 71 1
Figure 14. Potential toughening mechanisms for S i c based ceramics: 1 crack deflection, 2 micro-
crack formation, 3 crack bridging with metallic ligaments, e.g. TiB2, 4 crack bridging and pullout
by platelet or fiber - reinforcement.
be considered (see Fig. 15). In the following sections today’s state of the art in some
more or less important Sic-nonoxide composites is reviewed.
5.3.3.9.1 Sic-Tic Very promising composites have been developed in the SIC-TIC
system with S i c as the matrix phase [146-1481. Dispersed TIC particles significantly
Grain Boundary Strength
Dispersion, Volume Fraction
Orientation
- worse- -better-
Figure 15. Microstructural design for optimized crack deflection in composites
(Courtesy R. Telle, RWTH Aachen/Germany [ 1451.
Figure 16. SEM micrograph showing crack deflection and crack bridging in Sic-Tic and SiC-TiBz
composites: light areas = TIC or TiB2, dark matrix = SIC
(Courtesy D.Ly Ngoc, MPI Stuttgart/Germany [149].
improve both the strength and the toughness. Although an addition of T i c does not
reduce the densification temperature significantly below 2 1OO'C, the coarsening of
S i c is completely retarded which raises the strength to 700-800 MPa [146-1491.
The increase in KIc to 6.5-7.5 MPam'I2 is attributed to the misfit of the thermal
expansion coefficients of T i c and Sic, introducing considerable radial tensile stres-
ses at the phase boundaries and hoop compressive stresses in the matrix. These stres-
ses enable crack deflection (see Fig. 16), crack branching, and microcracking above
a critical particle size of = 3 pm. The optimum volume content of T i c ranges
between 20 and 30 vol-%.
5.3.3.9.2 SiC-TiB2 Sic-based composites with transition metal diboride (TiB2,

ZrBz, etc.) particulates have been developed for electroconductive applications such
as heating elements and ignition 1150,15 I], and also as wear resistant structural
parts for high temperatures such as valve-train components and rocker arm pads in
super-hot running engines [ 1471. These composites combine the high thermal and
electrical conductivity of TiBz and ZrB2 with the oxidation resistance of Sic.
Additionally, due to thermal mismatch stresses of the order of 2 GPa toughening
mechanisms such as crack deflection and stress-induced microcracking with a pro-
nounced process zone as well as flank friction have been proven to occur. Cai et al.
[ 1521 and Faber et al. [ 1531 have presented a detailed analysis of the contributions of
5.3 Production of' Sic 7 13
the particular mechanisms to the total fracture toughness, stating that the stress
induced microcracking is operational in a process zone of approximately 150 pm width.
Typical conditions for densification by axial hot-pressing are 2000-21 OOOC, at a
pressure of 20-60MPa for 30-60 min which results in 96-99.8% density. The
particle sizes of the matrix and dispersed phases range between 1-5 and 4--8pm,
respectively. An optimum volume fraction of reinforcing particles of 25-30 vol-Yo
has been reported, yielding a flexural strength of 710 MPa and a fracture toughness
of 5.0-5.7 MPam'/2 [149]. Composites with a lower TiB2 content of 15 vol-Yo exhi-
bit a mean strength of 485 MPa combined with a KIc of 4.5 MPa [147].
The strength of SIC based materials with 50 vol-YOZrB2, HfB2,NbB2 or TaB2 par-
ticles also ranges between 400 and 500 MPa [ 1501. Similar strength values (480 MPa)
combined with a exceptionally higher fracture toughness of 7-9 MPa m1/2have been
reported for large scale lots of pressureless sintered 16 vol-YO TiB2 composites
[ 1513. Since the sintering was carried out with temperatures exceeding 2000°C yielding
98-99% of the theoretical density and an average particle size of 2.0 pm, it is obvious
that the reinforcing phase also acts as a grain growth inhibitor for Sic. The high
temperature strength of SiC-TiB2 and SiC-ZrB2 composites was found to remain
nearly constant at 480MPa up to 1200"C, and is hence superior to that of many
sialons [150,151].
Tani and Wada [ 1541fabricated optimized SiC-TiB2 composites by reactive sintering
starting from an intimate mixture of Sic, Ti02, B4C, and C powders. The mixture was
either hot pressed or pressureless sintered and post-HIPed at temperatures of > 1900°C.
Titanium diboride was formed in situ according to the reaction
Ti0 2 + 0.5B4C+ I .5C t TiB2 + 2C
during an intermediate heating step at 140CL15OO"Cin vacuum or argon atmosphere.
Overstoichiometric amounts of B4C and C (1-2 weight-% each) can be adjusted to aid
sintering. The primary advantages claimed for this reaction sintering process are the
use of water in powder processing due to the disuse of highly reactive, preformed TiB2
powders, and the very small size of reinforcing TiB2 particles formed in situ due to the
use of ultrafine Ti02, B4C, and C starting powders.
Effective reactive pressureless sintering of SiC-TiB2 composites was recently
reported by Blanc, Thevenot, and Treheux [155]. In addition they studied the
tribological behavior using a pin on flat configuration (flat: Sic, pin: SiC-TiB2).
In dry conditions the composites showed less wear resistance than monolithic
Sic, however, with water as lubricant the opposite was the case.
In the very recent study of Kuo and Kriven [156]indentation-strength tests were used
to determine the retained strength, flaw tolerance, and toughness-curve characteristics
of two kinds of SiC-TiB2 composites. BSiC-TiB2 composites which were hot-pressed
with an A1203sintering aid, were compared with the well-studied aSiC-TiB2 compo-
sites, which were pressureless sintered with boron and carbon additives. TiB2 (I 5 vol-
Yo)in the B- and C-doped aSiC only increased the retained strength without a
significant improvement in the toughening. On the other hand, TiB2 (3ovol-%)
along with the effect of AI2O3 sintering aid for the BSiC-TiB2 composite greatly
improved properties with a higher retained strength in long crack regions, better
flaw-tolerance behavior and a sharply rising toughness vs. crack size curve. The
7 14 5 Silicon Carbide Based Hard Materials
S/HIPS WHIPS S/HIPS S/HIPS

80
t
76.8
19.2
3.7
58.6
37.0
42
39.0
560
4,7
20.0
74,5
5,5
'/. c c
a./
=/e
SIC
6
'
Figure 17. Flexural strength of pressureless sintered (= S, unhatched bars) and post-HIPed
(=HIPS, hatched bars) SiC-B4C particulate composites [ 1571.
different toughening behaviors for rx and pSiC-TiB2 were related to the weak nature of
the Sic-Sic and SiC-TiB2 interfaces as well as the fraction and size of TiB2.
5.3.3.9.3 Sic-B4C In the mid 1980's pressureless sintering and posthipping were
developed by Schwetz et al. [157] to produce 100% dense Sic-B4C composite
materials having SiC:B4C weight ratios within the range of from 90: 10 to 10:90
and a free carbon content of 4-5 weight-%.
These composites combine the good thermal shock resistance and oxidation resis-
tance of silicon carbide, with the hardness, wear resistance, and low specific gravity
of boron carbide. In this way a maximum strength of 550 MPa (four point bend) was
achieved for a composite of 59 weight-% Sic-37 weight-% B 4 C 4weight-% C (see
Fig. 17). The composite can be used in oxidizing atmospheres up to 1200 "C. Its micro-
structure is characterized by equiaxial B4C and graphite grains of < 1 pm diameter,
which were embedded in a matrix of S ic grains with an average grain size of 1.5 pm
(see Fig. 18). However, no improvement in fracture toughness was achieved, since the
fracture mode was almost 100% transgranular. Similar results on sintered Sic-B4C
5.3 Production of’ S i c 7 15
composites were obtained by Thevenot [158] and later by Tomlinson et al. [l 591, who
observed a 20% increase in strength when 25 vol-% B4C was added to Sic.
Excellent tribological properties for SiC-B4C-C composite materials were
recently encountered by Kevorkiijan et al. [ 1601.
In this study SiC-B4C-C seal rings for magnetic pumps were prepared by
pressureless sintering and characterized by a pin-on-disc method (medium: water,
pressure: 16-25 MPa, speed: 35-75 m sC’). The introduction of a lower level of
B4C (5-20 weight-%) particles into the S i c matrix resulted in an almost linear
decrease of wear rate. Moreover, further addition of B4C ( 2 0 4 0 weight-%) led to
an almost parabolic wear rate response. For example, with 40weight-% B4C a
decrease of the relative wear volume of seal rings by more than 55% was achieved.
These results recommend use of wear resistant SiC-B4C composites for heavily
loaded mechanical face seals in the pairing hard/hard against themselves. They
may be likewise suitable for the production of shaft protection sleeves and compo-
nents for sliding bearings whose wear resistance is to be improved.
5.3.3.9.4 SiC-AlN A series of solid solutions between S ic and A1N over the whole
composition range was concurrently discovered at the Universities of Utah and
Newcastle upon Tyne [ 1611, and has since received considerable attention [I 62-1 651.
A 2H wurtzite-type structure is formed by the carbothermal reduction of fine Si02
and A1203with a carbon source under nitrogen atmosphere at 1600°C:
3 s io 2 + 0.5AI2O3+ 4.5C + N2 1600°C ___i. 3SiC.AlN + 1.5CO + 0.5N2.
Kinetically favored is the carbothermal reduction of ct’ SiAlONs or a’ SiAlON -
CaSi9A130N15 1OC +
+
precursor mixtures (3Si3N4 3A1N CaO):
- +
1800 c
3(3SiC.A1N) + Ca + CO + 6N2.
Because the diffusion coefficients in covalent solids are extremely small, solid
solution was thought unlikely to be obtained by heating and annealing of the pow-
dered solid components.
wt. ‘Yo AIN
Figure 19. Tentative SIC-A1N phase diagram after Zangvil and Ruh [166].
Zangvil and Ruh [166] however, obtained SIC-AIN solid solutions by hot
pressing powder mixtures
(1 -x)SiC+xAlN+ (SiC),-x(AIN)x
According to the phase diagram proposed by Zangvil and Ruh (see Fig. 19) at
temperatures above 2O0O0C, a 6H-4H-2H series of solid solutions appears with
increasing amounts of AlN. AlN strongly stabilizes specific polytypes (4H and
2H) at certain composition ranges, enabling the engineering of single-phase Sic
materials with discrete physical properties. At temperatures below about 19OO0C,
a miscibility gap was first proposed by Rafaniello et al. [163], a suggestion supported
by several later studies. Xu et al. [165] obtained strong SIC-AlN materials with
flexure strength up to 1 GPa. Several mechanisms of grain refinement resulting
from Sic-polytype transformations into a wurtzite (2H) solid solution, were
reported. Kuo et al. [167] discovered the formation of modulated structures
within the miscibility gap and Lee and Wei [168] reported that pressureless sintering
with 2weight-% Y 2 0 3 as sintering aid at 2050°C produced a duplex structure
composed of large (Sic), (AIN), grains and small Sic grains. Further solid solution
treatment (> 2225°C) followed by annealing within the miscibility gap (1860°C)
resulted in spinoidal decomposition, giving various duplex/modulated structures
with improved fractured toughness of the alloys.
In addition to their possible use as high temperature structural ceramics, materials
in the system SIC-AIN have potential as wide band-gap semiconductors and for
opto-electronic applications [ 1661.
5.3.3.9.5 SiC-A120C Extensive SiC-Al20C solid solutions have been found by

Cutler et al. [161] for 1-100% A4OC. Moreover, they showed that wurtzite 2HSiC
can incorporate substantial amounts of AIN and A120C in solid solution and coined
the acronym ‘SiCAION to describe these materials by analogy with ‘SiAlON’ceramics.
Jackson et al. [169] have sintered Sic at temperatures between 1850°C and 1950°C
using a transient liquid phase produced by the carbothermal reduction of A1203by
5.3Production Qf Sic 717
A14C3.The resulting ceramic was fine grained (average grain size less than 5 pm) and
consisted of S i c (starting polytypes) and A120C as the two major phases. The
properties of the hot pressed ceramics varied with the amount of A120C, but at
an optimum composition of about 5-10 wei ht Y ' A120C, the strength (660 MPa)
- O.
and fracture toughness (KIc = 3.1 MPa m I$) obtained were comparable or super-
ior to the corresponding properties of commercial grades of sintered Sic. Huang
et al. [170] found encapsulation to be necessary for effective sintering with additions
of A1203 and A14C3, the densification occurring above M 1860°C. They attributed
densification to a transient liquid phase in the system A1203-A1404C in the route
to forming A120C with an eutectic temperature of ~ 1 8 4 0 ° C Four . point bend
strength, hardness and fracture toughness for the SiCAlON materials have been
reported [170]. The strength decreased with A4OC content in hot-pressed samples,
from ~ 6 0 MPa0 at 10 weight-% A120C to around 250 MPa at M 50% A120C. Most
significantly, the fracture toughness of some SiCAlON com ositions appeared to be
higher than that of S i c ( ~ 4 . compared
2 with 3.0MPam I f : ), using an indentation
technique. Lihrmann and Tirlocq [ 17I] proposed fabrication of sintered or hot-
pressed S i c based composites containing S i c as well as 5-30weight-YO solid
solutions composed of A120C and AIN. Starting from SiC-A1N-Al4C3-Al2O3
powder mixtures, densification was greatly enhanced by occurrence of a transient
liquid phase originating in the A1203-A14C3 system at temperatures above
1800°C. The composites with 5-10% (Al2OC-A1N) solid solution exhibit a mean
grain size < 2 pm and a mean strength of 620-670 MPa combined with a K,, of
5.1-6.8 MPam'/2. Strength and fracture toughness both retained their values up
to 14OOcC, before weakening, thus demonstrating the highly refractory nature of
the A120C-A1N second phase.
5.3.3.9.6 S i C S i C and SiC-C (Continuous Fiber Reinforced S i c Matrix Composites)
Fabrication by Chemical Vapor Infiltration In chemical vapor infiltration (CVI)

silicon carbide is vapor deposited inside a porous preform (4&60% porosity)
made of layers of woven cloth, from high strength C- or S i c fibers (CVI-SiCfib,,/
Sicmatrix or CVI-Cfber/SiCmatrix). Isothermal CVI infiltration of the fibers fills the
porosity with pure S i c and leads to a composite with up to 90% of the theoretical
density (TD), which fractures in a noncatastrophic mode at a typical flexural stress
of 300400MPa and a toughness of over 20MPam'I2. Using the forced flow
thermal-gradient CVI process developed at Oak Ridge National Laboratory,
USA, the infiltration time is reduced from weeks (isothermal CVI at SocietC
Europeenne de Propulsion/France) to less than 24 h (ORNL) and final densities
for composites of >go% T D 11721 are reached.
Material data of CVI-SiCjSiC and CVI-C/SiC composites available from MAN
- Technologie, Germany, are listed in Table 7 [173]. Figure 20 shows the fracture
surface and fiber pullout of a CVI-SiCjSiC composite.

Prior to the matrix infiltration one ore more layers of pyrolytic carbon or boron
nitride are usually applied to the fibers to provide a means of fiber debonding and
toughening by pull-out and crack bridging.
Table 7. Mechanical and physical data for continuous fiber reinforced S i c matrix composites
(MAN-Technology, Germany [ 1731.
Property Units Gradient Gradient LPI

CVI CVI CjSiC
SiCjSiC CjSiC
Fiber fraction Vol% 4247 4247 4244

Density g cm-3 2.3-2.5 2.1-2.2 1.7-1.8
Porosity YO 10-15 10-15 15--20
Tensile strength MPa 280-340 300-320 240-270
Strain % 0.5-0.7 0.6-0.9 0.8-1.1
E-modulus GPa 190-210 90-100 60-80
Flexual strength MPa 450-550 450-500 330-370
Compressive strength MPa 600-650 450-570 43W50
ILSa MPa 45-18 4448 35
CTE~ K-' 11 4 3 3
I 4 5 4
(a) ILS = Interlaminar strength. (b) CTE = Coefficient thermal expansion.
At high temperatures (> 600°C) the composites degrade in strength and toughness
due to oxidation of C fibers and/or these interface layers and prevention of fiber
pullout (brittle fracture mode). Studies are in progress to increase oxidation resistance
by use of a CVD-SIC overlayer which seals the surface of the porous composites.
Fabrication by Liquid Polymer Injiltration ( L P I ) In the first step of the LPI

process, a carbon-fiber preform is infiltrated with resin (e.g. polycarbosilane), to
bind the fibers together. Then the polymer is pyrolized to form Sic. These process
steps are repeated a number of times until the pores are narrow enough that further
Figure 20. Fracture surface of a CVI-SiCjSiC composite showing fiber pullout

(Courtesy M. Leuchs, MAN TechnologyjGermany).
5.4 Properties of Silicon Carbide 7 19
infiltration ceases [174]. Finally the body is heated to temperatures between 1000
and 1500°C for crystallization of the S i c matrix (LPI-C/SiC).
Fabrication by Liquid Silicon Infiltration (reaction bonding) ( L S I ) A leading

candidate for use in industrial gas turbine engine is a Sic matrix composite
named ‘toughened Silcomp’ [175]. It is produced by melt infiltration of molten
silicon into a porous preform containing carbon as well as BN-coated S i c fibers
(e.g. Textron SCS - 6). The composites thus produced consist of a fully dense
+
matrix of SIC Si, reinforced with continuous SIC fibers. Moreover, the melt
infiltration process is net shape and fast. Ultimate strength and strain at ultimate
strength are 220 MPa and 0.8%, respectively at room temperature (LSI-SiCjSiC-Si).
5.4 Properties of Silicon Carbide

5.4.1 Physical Properties
5.4.1.1 Color
Pure aSiC is colorless while the cubic [, modification is yellow. The only other ele-
ments that can be included in the S i c crystal lattice in amounts > 1p.p.m. are N, Al,
and B. Nitrogen gives a green color to 3C and 6H, and a yellow color to 4H and 15R.
The presence of the trivalent elements boron and aluminum gives all the modifica-
tions and polytypes a blue-black color [ 1761.
5.4.1.2 Optical properties

OrSiC is birefringent due to its crystal structure: no = 2.648-2.649, and n E = 2.688-
2.893 (Na 589 nm, 20°C) [177]. For PSiC, a refractive index of ~ 2 . 6 (Li
3 671 nm) has
been reported [7].
5.4.1.3 Electrical properties

Silicon carbide is a semiconductor. The most important electronic properties of Sic
are its wide energy band gap of 3.26 eV for 4H-Sic and 3.03 eV for 6HSiC, high
breakdown electric field of 2.2 x lo6Vcm-’ for l0OV operation, and high saturated
electron drift velocity of 2 x 10’ cm s-’. Doping with the trivalent elements aluminum
and boron gives the Sic p character, whde the pentavalent element nitrogen produces
y1 character when incorporated into the Sic crystal lattice. The resistivity can be varied
between 0.1 R cm and lo’* Ocm, depending on the concentration of the dopant [178].
Whereas compact, homogeneous SIC obeys Ohm’s Law, aggregates of S i c grains
show nonlinear current-voltage behavior. At low applied voltages they behave as
insulators, but when the applied voltage is increased above a certain value the
current increases exponentially. Thus, the points of contact between the grains
cause the electrical resistance to be voltage dependent [178,179].
5.4.1.4 Thermal and Calorific Properties

For a ceramic material, silicon carbide has an unusually high thermal conductivity:
150W mK-' at 20°C and 54 W mK-' at 1400°C [ 1SO]. The high thermal conductivity
and low thermal expansion (4.7 x K-' for 2&1400"C) explain why the material
has such good resistance to thermal shock.
The s ecific heat capacity of Sic is 0.67 J g-' K-' at room temperature, and
1.27Jg-'K1 at 1000°C. The standard enthalpy of formation AH&,, is
-71.6 f 6.3 kJmol-', and the entropy 5$98K is 16.50 f 0.13 Jmol-' K-' [181].
5.4.1.5 Mechanical Properties

Silicon carbide is noted for its extreme hardness [ 182-1 841, its high abrasive power,
high modulus of elasticity (450 GPa), high temperature resistance up to above
15OO"C, as well as high resistance to abrasion. The industrial importance of silicon
carbide is mainly due to its extreme hardness of 9.5-9.75 on the Mohs scale. Only
diamond, cubic boron nitride, and boron carbide are harder. The Knoop micro-
hardness number HK-0.1, that is the hardness measured with a load of 0.1 kp
(=O0.98N),is ~ 2 6 0 0(2000 for aA1203, 3000 for B4C, 4700 for cubic BN, and
7000-8000 for diamond). Silicon carbide is very brittle, and can therefore be crushed
comparatively easily in spite of its great hardness. Table 8 summarizes some typical
physical properties of the Sic ceramics.
Since the microstructural grain size (Fig. 12a-h), pore content, and chemical com-
position of the various ceramic products differ considerably, it follows that the prop-
erties are also different.
Recrystallized Sic is much stronger than the ceramically bonded material, but its
high residual porosity imposes limits as far as mechanical strength is concerned
[185]. Reaction-sintered SiSiC is still stronger, but only up to 1400°C (Fig. 21),
the softening point of the accompanying silicon phase [ 1861.
The best mechanical strength is exhibited by sintered S i c and hot-pressed
materials [109,187,188].
Solid-state sintered, hot-pressed, and isostatically hot-pressed materials offer
considerable advantages over all other ceramic materials in plastic deformation
under a sustained load (creep), because of the low content or almost complete
absence [189,190] of sintering aids.
Compared to solid state sintered silicon carbide (SSiC) and dense alumina
(A1203), liquid phase sintered Sic (LPS-Sic) features improved edge toughness
(see Fig. 22) close to the edge flaking resistance of sintered silicon nitride (Si3N4).
The latter property indicates the sensitivity of edges against mechanical chipping,
a quantity which is most important for safer handling of ceramics in grinding,
clamping, transportation etc.
5.4.2 Chemical Properties

One of the outstanding characteristics of silicon carbide is its chemical resistance,
which is due to the high affinity of silicon for oxygen. The reaction of silicon with
Table 8. Physical properties of various S i c ceramics.
Thermal Thermal Flexural strength

Sic Sic Young expansion conductivity
material content Density Porosity modulus 30-1500°C at 600°C 20°C 1400°C
type (weight-%) (g cm-') (YO) (GPa) K-I) (W mK-') (MPa) (MPa)
Ceramic bonded S i c up to 95 2.55 20 100 5.8 16 30 20
KSiC
Recrystallized S i c I00 2.60 20 250 5.0 28 100 100
RSiC
Reaction bonded S i c 100 2.60 20 250 5.0 25 250 250
RBSiC
Infiltrated Sic 90 3.12 <1 400 4.3 60 350 200
SiSiC
Sintered S i c 98 3.15 <2 410 4.9 50 430 450
(solid state)
SSiC
Sintered S i c 95 3.21 <I 420 4.5 50 730 400
(liquid phase)
LPSSiC
Hot-pressed S i c 98 3.20 0 450 4.5 55 640 650
HPSiC
Hot-isostatic-pressed >99.5 3.21 0 450 4.5 75 640 610
Sic
HIPSiC
HIP-post densified 98 3.19 0 430 4.8 50 450 450
SSiC
HIPSSiC
I 900 1
SIC-CERAMICS
800 -
600 -
solid st sintered
500 ---------------
..-- _/--
reaction banded
400 - (with free Si)
1 ,
\ ,
I
recrystallized
__
---------1---------'
100
ceramic-bonded
- -
-
- L
0
~ ~
0 200 400 600 800 1000 1200 1400 1600

TEMPERATURE "73
Figure 21. Flexural strength of S i c ceramics as a function of temperature.
1
Load P
Distance d
Load P
0 0.1 0.2
Distance from Edge d, mm
0.3 0.4 0.5
Figure 22. Edge toughness of LPS-Sic (EKasic T) compared to sintered Al,03 and Si3N4:
b
(a) flaking load against distance from edge, (b) schematic of experimental setup (Courtesy
Edge Flake
L.S.Sig1, ESK-Kempten/Germany)
oxygen in an aqueous medium causes passivity and, if exposed to thermal oxidation,

for example, in air, leads to the formation of glassy silica films [191-1931. The
oxidation of pure S i c begins at around 600°C, forming a coating of Si02 on the sur-
face of the SIC that prevents further oxidation [ 1941:
Sic + 2 0 2 + Si02 + C02.
The reaction rate varies with time according to a parabolic law [191]. The kinetics
are determined by the diffusion of oxygen through the SiOz layer. The temperature
dependence of oxidation follows the Arrhenius equation.
The ‘active’ oxidation of SIC is distinguished from the ‘passive’ oxidation reaction
described above:
Sic + O2 + SiO + CO.
Active oxidation takes place under conditions of oxygen deficiency above 1000°C
and leads to decomposition of the SIC and formation of silicon monoxide
[194, 1921. The two forms of corrosion (active, passive) and the conditions for the
boundary have been recently discussed by Nickel et al. [I951 using examples from
Sic- and also Si3N4-based ceramics. A likely high temperature boundary for SIC
is z 1700-1 8OO0C,where a secondary active-to-passive transition by bubble forma-
tion and spalling occurs.
Thus silicon carbide is attacked and decomposed by oxidizing agents (e.g.,
+ +
Na2C03 Na202or Na2C03 KN03) if the protective layer of Si02 is removed,
thereby enabling the reaction to proceed unhindered.
Pure silicon carbide is twice as oxidation resistant, even at 1500”C, than the best
current superalloys at their maximum service temperature of 1200°C.
Silicon carbide is resistant to most chemicals, resisting acids and alkalis and even
aqua regia and fuming nitric acid.
A mixture of hydrofluoric adic, nitric acid, and sulfuric acid slowly attacks silicon
carbide PSiC being somewhat more reactive than &Sic[196]. However, complete disso-
lution only takes place if the SIC is very finely divided, under pressure, and at a elevated
temperature (e.g., 250°C for 16h). Alkali melts [ 197, 1981will attack it in the presence of
oxidizing agents. Oxides, molten metals, and water vapor have a destructive effect on
Sic at temperatures of 1000°C and over. Chlorine reacts exothermally with silicon
carbide above 800°C with the formation of silicon tetrachloride (Sic&) and carbon.
Silicon carbide behaves in various ways towards molten metals. It is not attacked
by molten zinc or zinc vapor [199]. Molten aluminum attacks SIC slowly, forming
Al,C, and silicon, but as the silicon content increases, the reaction eventually
ceases because an equilibrium is established [200]. Molten iron desolves Sic, form-
ing iron carbide and iron silicide.
5.4.3 Tribological Properties

5.4.3.1 Review on Tribology Work in Sintered &Sic Ceramics
Tribology is the study of adhesion, friction, wear and lubricated behavior of
materials in solid state contact. It was in the early 1980s that the advantage of
sintered S i c as seal face material was determined [201,202]. Since then, the mechan-
ical engineering industry has focused attention on this material and tribological
studies have been simulated to understand better those physical and chemical prop-
erties of SSiC that will affect its behavior when in contact with itself, other ceramics,
or metals. Model investigations involving Sic have shown the coefficients of fric-
tions against various materials, even itself, to be a function of contact stress [203]
and have documented the anisotropic wear behavior of monocrystalline S i c crystals
[204]. Lashway et al. [205] found that a controlled amount of porosity improved the
ability of sinteredSiC to retain a hydrodynamic film with lower friction. Seal tests
also indicated lower power dissipation at varying pressure-velocity values for com-
binations with sinteredSiC seal face materials.
Consistently positive practical experience gathered by Knoch et al. on sintered-
Sic seal rings [206,207] has helped the material to gain popularity rapidly for use
in situations involving wear problems.
Excellent results have been achieved in developing sliding bearings for hermeti-
cally sealed pumps [208]. All of them are absolutely leak-proof, whether in operation
or shut down and they are therefore of great value environmentally. They all have
the same design requirement that the sliding bearing of the pump shaft must be
flushed and lubricated by the pumped medium. Traditional materials simply
could not cope with the harsh conditions of the chemical industry and were quickly
destroyed by corrosion and abrasion. Sintered silicon carbide solved the problems.
Liquids containing abrasive particles do not restrict the use of sintered Sic.
Many sliding-wear problems occurring in the field are attributable to interruption
of the ideal (that is properly lubricated) running conditions, in which case the sliding
faces of the bearing or seal in question make contact with each other, giving rise to
solid-state or dry friction marked by a pronounced increase in the coefficients of fric-
tion. Local frictional heat leads to peak thermal stresses that may be of such inten-
sity as to cause a breakout of microstructural constituents. Then, when lubrication
(and cooling) is restored, the material is in danger of cracking or fracturing due to
thermal shock. Sintered silicon carbide, however, is better able to cope with such
situations than other ceramic materials, because it is stronger and has a lower ther-
mal expansion coefficient, and a higher thermal conductivity. Consequently, brief
periods of dry running can be survived.
The nonlubricated wear behavior of sintered silicon carbide under unidirectional
sliding at room temperature was studied by Derby et al. [209] as a function of load
and sliding time. At low loads polishing and ploughing mechanisms were observed
and the microstructure revealed an etched appearance. With increased loads and slid-
ing times microcracking at the grain boundaries occurred leading to subsequent grain
pull-out. In the study of Cranmer [210] it was shown that for sintered silicon carbide
sliding against itself also surface plastic deformation, ploughing and cracking were
operative as wear mechanisms. Miyoshi et al. [211] carefully studied friction and
wear of the pair SiC/iron in vacuum environment at temperatures up to 1500°C.
By using X-ray photoelectron spectroscopy it was found that the surface chemistry
of sintered silicon carbide (graphite and Si02) as well as the alteration of this chem-
istry by the temperature are very influential in dramatically changing the friction co-
efficient. Breznak et al. [212] quantified the role of initial surface roughness on the
friction behavior of sintered silicon carbide rubbing against itself in cyclic oscillatory
motion. It was shown that the coefficient of friction decreased from an initial value of
0.40 to 0.25 as a result of surface polishing. Further work by Breval et al. [213]
indicated that adding graphite to Sic reduces the wear of SiCjSiC couples somewhat,
but improving their initial surface finish has the opposite effect: it leads to a greater
loss of material during running in. Wear debris of SiCjSiC couples exhibited a bi-
modal particle size distribution: some particles were micrometer size and the others
ranged in nanosize from 5 to 50nm. Using pin on disk, as well as abrasion wheel
test, Wu et af. [214] observed that the amount of wear increases with increasing
grain size for sintered silicon carbide.
Smythe and Richerson [215] conducted experiments in which the dynamic sliding
contact behavior at temperatures greater than 1000°Cwas studied. It was found that
surface film formation at higher temperatures governs the friction behavior. It is
believed that Si02 or Si02 modified by impurities from the SIC or the environment
with flow properties contributed to the lowering of the friction coefficient [216].
Tomizawa et al. [217] performed pin-on-disk experiments on friction of SSiC against
itself in water at room temperature. The authors found a friction coefficient of 0.26 and
noted that wear of sintered silicon carbide occurs by a combination of tribochemical
dissolution and the formation of pits by fracture of Sic grains. The amount of material
removal varied from one Sic grain to the other, due to a strong dependence of tribo-
chemical wear on crystallographic orientation of Sic grains.
Knoch and Kracker [2 181 observed that this anisotropic tribological behavior is
most conspicuous with EKasic D sintered silicon carbide, which has a bimodal
grain structure: about 3ovol-% of larger platelets of hexagonal a grains (about
l00pm long) and about 70vol-Y0 of smaller a grains (about 10pm long) as shown
in Fig. 23a. This material shows superior performance particularly if paired against
a softer carbon material. Figure 23b shows the surface topography developed in a
finely lapped surface under in-service conditions, as water lubricated EKasicj
carbon seal. This relief structure, with the depressions in a textured surface, obviously
acts as reservoirs for lubricant, thereby improving the emergency running properties
of the components. Figure 23c shows a sketch of the process. Boch et al. [219] studied
the dry friction behavior of SiCjSiC couples (fixed ball rotating disk) at temperatures
from 20 to 1000°C and observed a rapid decrease in the coefficient of friction from
0.45 to 0.16 when temperatures increases over 400°C. From 400 to 700°C, the
debris agglomerates in the form of SiOz rolls, which arrange themselves perpendicu-
larly to the sliding direction. These rolls act as minute roller-bearings decreasing the
coefficient of friction and preventing the formation of cracks in the wear track.
Habig and Woydt [220] studied unlubricated friction and wear of selfmated SSiC
sliding couples (EKasic D) at temperatures between 22 and 1000°C using a station-
ary pin and a rotating disk assembly. As shown in Fig. 24, at room temperature the
coefficient of friction decreases with increasing sliding velocity fromf = 0.8 to 0.6.
At temperatures >400°C a scatter of friction coefficients between 0.2 and 0.7 is
observed. By using small-spot-ESCA (electron spectroscopy for chemical analysis)
it was shown that the low friction coefficients of SiCjSiC were due to formation of
thin oxide layers formed by tribo-oxidation, while the higher friction coefficients
accompanied by higher wear coefficients were related to thicker oxide layers.
Figure 23. (a) SEM micrograph of a polished and plasma-etched section of sintered Sic, ‘EKasic D’,
note: bimodal grain size distribution. (b) SEM micrograph of EKasic D sliding face after about
4000 h of successful service as a mechanical seal face (SiC/carhon pair, lubricated with water,
maximum pressure 50 bar). (c) Schematic of the operational condition of a relief structured sliding
surface in the lubricated state and after breakdown of the lubricant film.
These results are in excellent agreement with those of Yamamoto et al. 12211, who
observed a decrease in the coefficient of friction when SiCjSiC couples had been
heated for 1 h at 1000°C with formation of a thin oxidation layer. However, at a
higher oxidation level the coefficient of friction increased rapidly. A low coefficient
T = 22oc T = 4OOY T = 800°C T = looooc

Figure 24. Friction coefficientJ; and wear coefficient, k , of SiC,ISiC sliding pairs as a function of
sliding velocity with 10 N for various temperatures (sintered SIC ‘EKasic D’, after Habig and
Woydt, BAM Berlin/Germany [220]).
of friction for the SiCjSiC couple was also measured by Martin et al. [222] who
studied friction under an oxygen partial pressure of 50 mPa, which should admit
only a very thin oxidation layer.
Sasaki [223] studied the influence of humidity on friction and wear behavior of
SiCjSiC couples. It was shown that humidity decreased the coefficient of friction
from 0.5 (dry air) to 0.2 (wet air), simultaneously a decrease in the coefficient of
wear was observed from (dry air) to mm3 N-l m-I.
Denape and Lamon [224] considered the important contribution of wear particles to
the wear mechanisms. The main findings were: polishing at low loads (5 N) is due to
fine individual wear particles smaller than 1 pm circulating in the sliding interface,
and abrasion and grain pull-out (at high loads of 20N) are associated with large
accumulations of particles adherent to the sliding phases. Wear tests performed
under water showed again an anisotropic wear of individual S i c crystallites. It
was concluded that the circulation of wear debris in the sliding interface is control-
ling both the wear rate and the friction response.
Woydt et al. [225] evaluated the tribological behavior of SIC-TIC and SiC-TiB2
ceramic-matrix composites using a special high temperature tribometer. In an
normal ambient environment these composites are able to form ‘lubricious oxide’
reaction layers on the hard substrate; the low friction coefficient of the couple
Sic--TiC/SiC-Tic at room temperature with values of f = 0.2. and 0.3 was
explained by the formation of self-lubricating layers of TiOz and SiC,O,,. The
friction coefficient of self-mated SiC-TiB2 sliding couples was in the range 0.56-0.72
and compared to sintered aSiC materials no improvement was observed in the wear
resistance of the SiC-TiB2 composite [226].
In a study on the friction and wear behavior of lubricated ceramic journal bear-
ings it was shown by Maurin-Perrier et d.[227] that both SiCjSiC and SiC/Si3N4
couples provide a significantly better behavior in terms of film stability compared
with the classical materials used for water- or hydrocarbon-lubricated bearings. A
self-improvement of surface roughness during the running-in period by tribochem-
ical reaction significantly increases the range of stability of the film; consequently it
is not necessary to produce a highly expensive surface finish to increase the perfor-
mance of the SiCjSiC couple. The lifetime of the components is mainly determined
by the behavior during the start and the stop phases where boundary lubrication
occurs. Sliding wear on sintered silicon carbide leads to a very smooth surface
between residual porosity. Further it was found [228] that in dry run situations,
sintered Sic of bimodal grain structure (EKasic D) outperforms other Sic
materials. The bimodal EKasic D shows the lowest development of heat, and there-
fore it has the lowest coefficient of friction. In a mayor US study on 'Tribological
Fundamentals of Solid Lubricated Ceramics' [229] the bimodal grain size/shape
distribution was judged best in terms of wear resistance, when compared with
other sintered aSiC materials.
Loffelbein et al. [230] tested self-mated sliding couples of SSi3N4,HIP-RBSi3N4,
SSiC, SiSiC, MgO-Zr02 and A1203 in different aqueous solutions (H20, NaOH,
KOH, NH40H, HN03, H2S04, H3P04, CH3COOH, HCl, HC104) under condi-
tions of boundary or mixed lubrication. The best frictional behavior was observed
with couples of SSiC, with steady state friction coefficients of 0.05. The lowest wear
coefficients were measured for couples of the two types Si3N4,SSiC and A1203with
values of approximately 10-7mm3N-'m-'. If low friction and low wear are
required, couples of SSiC seem to be the best choice in most aqueous solutions. A
recent study by Anderson and Blomberg [231], based on tests with sintered silicon
carbide sliding unlubricated on itself in point (pin-on-disc), line (journal bearing)
and plane (mechanical face seal) contacts. Tribo-oxidation and surface fracture
were identified as the dominating deterioration mechanisms. The oxidation
products formed silicon dioxide and, within narrow operational regimes, silicon
monoxide. The highest wear rates occured in the pin-on-disc configuration, while
the lowest rates were obtained in the journal bearing tests.
Kitaoka et al. [232] investigated the effects of temperature and sliding speed on the
tribological behavior of sintered silicon carbide (0.1 weight-% B/l .O weight-% C) by
sliding on the same material in deoxygenated water from room temperature to
300°C under high vapor pressures (120°C/2 bar, 3OO0C/85 bar). The wear mechan-
ism appears to consist of hydrothermal oxidation of Sic according to the equations:
Sic + 2 H 2 0 + Si02 + CH4,
S i c + 4H20 4 Si02 + C 0 2 + 4H2
S i c + 2H20 Si02 + 0.5C2H6+ 0.5H2,
4
S i c + 2H20 Si02 + C + 2H2,

4
and dissolution of reaction products such as silica. Fine mirrorlike worn surfaces
were observed without wear debris under all sliding conditions.
For description of the tribotechnically used machine-elements and more informa-
tion about tribology, the book ‘Tribology Handbuch Reibung und Verschleip’
~
[233] is recommended to the reader.
5.4.3.2 Sintered S i c Material Development in Sliding Wear Applications

The coefficients of friction of hard and wear resistant ceramic materials and of most
surface coatings are always greater than 0.1 under dry run conditions. This value is
far too high for dry running bearings to be designed. Frictional heat develops, and if
loads are high then very high temperatures are generated. These will not affect the
sintered silicon carbide but the housing and the overall structure may well distort
and become damaged leading to failure [234]. Since materials research shows that
it is impossible to develop hard ceramic material with coefficients of friction
below 0.01 (which is the typical value for a lubricated bearing) then the targets
for product development must include:
- stabilizing the hydrodynamic lubrication film,
~ preventing dry running,
~ reducing friction and wear if the lubrication film breaks down.
There are obviously two routes towards stabilizing the hydrodynamic film. One is to
optimize the design of the component including the most suitable bearing surface
characteristics. The other is to optimize the silicon carbide material itself. In follow-
ing this second route, it was assumed that the tribological performance can be
improved by tailoring the microstructure of SSiC. This can be done by introducing
isolated pores which act as lubricant pockets or incorporating graphite particles,
which act as dry lubricants.
Fig. 25 shows typical sliding faces (left) as well as the microstructures (right) of the
improved materials in comparison with standard EKasic D. The material contain-
ing pores (diameter about 40pm) was named TRIBO 2000, the second material
containing pores and graphite particles, in the range 40-60 pm, was named
TRIBO 2000-1. Properties of these S i c materials are shown in Fig. 26 in comparison
with EKasic D.
Both pores and graphite inclusions act as internal flaws, so it is no surprise that
strength is reduced. On the other hand since these imperfections are of regular
size it follows that the Weibull modulus is increased which is important for
design. The Young modulus is slightly reduced. Friction under boundary lubrica-
tion conditions decreases as intended, while the corrosion resistance is unchanged.
TRIBO 2000 and TRIBO 2000-1 were tested in product lubricated bearings and
mechanical seals in comparison with EKasic D as the basis sintered S i c material.
The Stribeck testing method was used to evaluate the coefficient of friction as a func-
tion of sliding velocity (see Fig. 27). This is an axial bearing test in which a circular
segment sits (lubricated) on a rotating ring of the same material. The torque between
the segment and ring is measured. The load bearing capacity of the hydrodynamic
lubricant film is the key factor. The better tribological material is the one which
Figure 25. SEM micrographs of sliding faces (left) and polished sections (right) of sinteredSiC
grades: EKasic D (top), TRIBO 2000 (middle), and TRIBO 2000-1 (bottom). The graphite in the
TRIBO 2000-1 microstructure was revealed as white particles after coating the polished section
with a carbon film.
shows stable hydrodynamics down to slower sliding velocities, before the lubricant
film breaks down and the bearing comes into solid state contact. Figure 28 is an
example.
In this case, perhaps unexpectedly, the TRIBO 2000-1 material shows both a
higher hydrodynamic friction and a deviation with increasing velocity when com-
pared with TRIBO 2000 and EKasic D.
5.4 Properties qj" Silicon Cavhidp 73 1
250
-5 200
a
-
150
a
.->
5 100
50
00
0
Flexural Strength Weibull Fracture Youngs Thermal Thermal Corrosion coeffof ~ ~ ~ c t ~
4-Point Modulus Toughness Modulus Conductlvity Expansion Resistance khaml Seal
p.25 bar v=9 4 in's
Figure 26. Comparison of properties of the pore-containing material (TRIBO 2000) and the pores-
plus-graphite-particles-containing material (TRIBO 2000- 1) with EKasic D
Summarizing all Stribeck test data (load, media, and up to 10000 repeats) the
TRIBO 2000 material shows the best performance in terms of stabilizing the hydro-
dynamic film down to the lowest velocities. This is important for sliding bearings
and has been confirmed many times. The poorer performance of the TRIBO
2000-1 material in this typical sliding bearing test is not clearly understood, but it
Lubricant
stationary body
Surface roughness R
FOUR MODES OF LUBRICATION

1 Boundary friction ( h ---> 0)
2 Mixed friction (h = R )
h 3 Elastohydrodynamic (h > R )
Film thickness lubrication
4 Hydrodynamic lubrication (h >> R )
0
f 0.1' STRIBECK - CURVE

Coefficient
of friction
0,OI
Viscosity . Rubbing speed v
__f
load
Figure 27. Schematic of a Stribeck curve and the four modes of lubrication
5
a
7
=c 4
.-
c
0
.o 3
IL
o ~ ~ ' ' ' i ' ' " i ' ' " ~ ' " ' ~ ' ' " l
0 5 10 15 20 25
Sliding Velocity [m/s]
Figure 28. Coefficient of friction as a function of sliding velocity for self-mated sliding pairs of sin-
tered S i c materials.
may have something to do with the poor wetting behavior of graphite in water under
normal pressure.
In mechanical seal applications, under differential pressures of less than 15 bar the
same ranking of materials is observed in terms of friction and wear. Under extreme
situations, however, like very high differential pressure (producing continuous
boundary lubrication) or pump cavitational run (causing breakdown of the hydro-
dynamic film under frequent violent loads), the TRIBO 2000-1 material showed the
best performance (see Fig. 29).
To diminish the effects of hydrodynamic grain-boundary corrosion in applica-
tions with hot water, such as chipping out of fine grains in hot spots, tribochemical
reaction with water
S i c + 2H20 = Si02 + CH4
1.2 I I
.a
.-0 - +
m n
a"
Y
.-
0
In
0
Figure 29. Relative wear volume on seal rings of sintered Sic material (self-mated couples).
Figure 30. Microstructure of EkasicW: coarse-grained SSiC with platelet structure for mechanical
seals and bearings (hard/hard combinations) in hot water applications.
and formation of damaging SiOz layers on the sliding surfaces, another two new
modified SSiC materials with a predominantly coarse-grained, bimodal platelet
structure have been developed [235]. Catastrophic failure of coponents is avoided
since the large S i c platelets near the surface are anchored to a depth at which
there is no grain-boundary corrosion.
EKasic W silicon carbide is a dense material with a predominantly coarse-grained
bimodal platelet structure. The effectiveness of the coarser microstructure (see
Fig. 30) in improving the corrosion resistance has been clearly demonstrated in prac-
tical tests on a mechanical seal test rig. Even after 500 h of testing (deionized water,
60°C, 6 bar, hard/hard couples) no SO2 layer was formed on the functional surface.
Figure 31. Microstructure of EKasic HW: coarse-grained SSiC with platelet structure and graphite
particles (black phase 4 0 4 0 pm in diameter) for mechanical seals and bearings (hard/hard combi-
nations) in hot water applications.
EKasic HW silicon carbide, on the other hand, is a dense material with a

predominantly coarse-grained, bimodal platelet structure and graphite particles of
40-60pm in size (see Fig. 31). This new tribological material, which is still in its
test phase, should be particularly suitable for mechanical seals and sliding bearings
in applications in contact with hot water containing solids.
5.5 Quality Control

The S i c content of silicon carbide products is now usually determined by measuring
the carbon contents. The total carbon content is determined by combustion of the
sample in a stream of oxygen at 1050°Cin the presence of lead borate. The COz pro-
duced is absorbed in Ba(C104)z solution and determined by coulometry [236,237].
An alternative technique is to oxidize the S i c with oxygen in a high-frequency
induction furnace containing a flux metal, and to detect the C 0 2 produced by IR
absorption.
Free carbon can be directly determined coulometrically by combustion at 850°C
without any additives, provided that S i c itself is not appreciably oxidized. If the S i c
samples are very finely divided, that is, if they have large specific surfaces, and if long
heating times are required due to high free carbon contents, this method can not be
used. In such cases, the weight change on combustion is measured and the decrease
in total carbon content is determined. This enables the effect of the unavoidable oxi-
dation of the S i c to be eliminated, and the free carbon content can be calculated
[236,237]. For the determination of free carbon in S i c sintering powders or in
more-or-less contaminated Sic, containing iron or silicon impurities or small
amounts of sintering aids etc., the use of a wet chemical method, based on the
wet chemical dissolution of free carbon in a hot chromic-sulfuric-acid mixture
proved to be very useful [238].
Analyzers with IR detection are also suitable for the direct determination of free
carbon provided that they allow precise temperature control during combustion of
the free carbon.
The S i c content is calculated from the difference between total and free carbon:
weight-% S i c = (weight-%C,,,,, - weight-%Cfree)x 3.3383
The free silicon content is determined by measuring the volume of hydrogen
produced on treatment with sodium hydroxide solution. Alternatively, metallic
silver is precipitated from a silver fluoride solution, dissolved in nitric acid, and
determined by a standard method [236,237]. In the chemical analysis of S i c abrasive
grits, emphasis is placed on the determination of accompanying materials such as
carbon, silicon, silicon dioxide, and metallic oxides. The S i c content (plus the
free carbon) is found by weighing the residue after treatment with a mixture of
hydrofluoric, nitric, and sulfuric acids [237]. For special grades of S i c for refractory
and metallurgical applications, volatile components can interfere with the S i c
determination. In this case, the sample must be annealed under argon [239].
5.5 Quality Control 735
In order to characterize S i c powders and sintered ceramics the total oxygen and
nitrogen content as well as the contents of metallic impurities are analyzed. Total
oxygen and nitrogen contents are usually determined by an inert gas fusion
method (Leco T C 436) using powdered samples, whereas metallic impurities (Na,
K , Ca, Mg, V, Fe, Ti, Al, Cr and Ni) and boron content are determined in acidic
solutions by inductive plasma emission (ICP) spectroscopy [240-2421.
Besides chemical analysis, physical properties such as particle size, particle size
distribution, and bulk density are also important.
The particle size distribution of coarse abrasive materials (5 mm-50pm) is
determined by sieving. For particle sizes less than 50 pm, sedimentation and laser
diffraction methods are used.
Both the particle size distributions and the methods for their determination are
standardized. The standard for bonded abrasive applications [243] differ from
those for coated abrasives [244]. The bulk density depends both on the particle
size distribution and on the particle shape. It is measured by weighing a known
volume of S i c grains [245]. The packed density is often determined instead of the
bulk density [246].
According to the analysis of the microstructure [247] and the physical properties
measurements, the reader is referred to the literature [93,248-25 11. The major
sequential steps in conducting analysis and properties evaluation for SIC ceramics
are shown in Fig. 32.
Ceramic Powder Processing

Productioi MixingISpray Drying
Bulk Chemical and Microstructural Molding
Physical Analysis Analysis Analysis, SEM Sintering
TEM, EELS, AES
Raw Materials
to Ceramic Materials
Figure 32. Analysis and properties determination for S i c materials.

136 5 Silicon Carbide Based Hard Muterials
5.6 Toxicology and Occupational Health

Silicon carbide is nontoxic, and is therefore a nonhazardous material as defined by
GefStoffV [252].
For fiber-free finely divided Sic, the MAK value is 4mgmP3 [253]. If the MAK
(‘Maximalzulassige-Arbeitsplatz-Konzentration’)is exceeded a dust mask with a PI
filter (for inert dusts) must be worn. Recent investigations, however, have shown
that these Sic dusts are not fibrogenic [254], so that the need for a MAK value is
questionable.
Sic whiskers have no known routes of entry into the body except as airborne
particles.
Conclusive experimental studies on the effects of respirable fibrous S i c dusts
(whiskers) are not yet available [255]. In the United States ASTM has published
recommendations for safe handling procedures related to all ceramic fibers,
including silicon carbide whiskers [256].
5.7 Uses of Silicon Carbide

Of the M 700 000 tons Sic produced per year, about 33% is used in metallurgy as a
deoxidizing plus alloying agent, and about 50% in the abrasive industry [257]. The
remainder is used in the refractory and structural ceramics industries and to a small
extent also in electric and electronic industries as heating elements, thermistors,
varistors, light-emitting diodes, and attenuator material for microwave devices.
In its loose granular form silicon carbide is used for cutting and grinding precious
and semiprecious stones and fine grinding and lapping of metals and optical glasses
[258-2601. Bound with synthetic resins and ceramic binders S i c grits are used in
grinding wheels, whetstones, hones, abrasive cutting-off wheels, and monofiles for
machining of metals, ceramics, plastics, coal-based materials, and so on [26 11.
Coated abrasives include abrasive paper and cloth in sheet or band form. They are
produced by strewing the Sic grains onto a substrate coated with glue or bonding
resin and then covering with a second layer of bonding agent [257].
The addition of silicon carbide during the melting of cast iron aids carburization
and siliconization, and improves the quality of the cast iron as a result of its seeding
action [262]. In the production of steel in an arc furnace, silicon carbide acts as
deoxidant and helps in slag melting.
The need to control thermal expansion and to increase the strength and Young
modulus of aluminum alloys produced new Al-Sic composites alloys containing
up to 50 vol-’YOSic particles [263]. The market is poised for rapid expansion in this
area over the next years as developmental products move into commercialization
stages. For reinforcement applications Sic is also used in the form of whiskers
[264], platelets and fibres.
The resistance of ceramically bonded and recrystallized silicon carbide to thermal
shock, oxidation, and corrosion is utilized in its use as a refractory construction
material, for example, in linings and skid rails for furnaces and hot cyclones, and as
a kiln furniture, especially in saggers [61,265-2691. The good electrical conductivity
of the material at high temperatures, coupled with its outstanding oxidation resis-
tance, led to its early use in the electric heating industry [270-2721, which markets
its products mainly in the form of rods and tubes that operate up to 1500°C.
Recrystallized S i c igniters are used in home gas appliances, replacing pilot lights.
High-purity S i c shapes are used in the electronic industry as furnace components
for processing of silicon wafers. The thermoelectric properties of S i c suggest the
use of sintered S i c rods as high-temperature thermoelements [273] and as Seebeck
elements [274] for high-temperature thermoelectric energy conversion. Voltage-
dependent resistors (varistors) consist of ceramic- or polymer-bonded S i c and are
used in overvoltage protection equipment.
Silicon carbide is an outstanding material for the construction of electronic equip-
ment. Blue light-emitting diodes having an improved 470 nm peak wavelength are
being produced and marketed as the first commercial S i c semiconductor device.
The continual development of the deposition of S i c thin films and of large
diameter single crystal S i c wafers, the associated technologies of doping, etching
and electrical contacts have culminated in a host of new solid state devices including
field effect transistors capable of operation up to 650°C [275].
High-density, high-strength SiSiC, SSiC, HP-, and HIPSiC materials, which have
been on the market only since the 1970s, have opened up a new field of application
[276], namely, in mechanical and high-temperature engineering. SiSiC and,
especially SSiC, are displacing the chemically less resistant tungsten carbide (hard
metal) and the thermal-shock-sensitive aluminum oxide in modern mechanical
seals where they are used in the form of slide rings (see Fig. 33). The excellent
Figure 33. SSiC slide rings (ESK-Kempten, Germany).

Figure 34. SSiC sliding bearings (ESK-Kempten, Germany).
wear resistance of sintered silicon carbide, its excellent chemical resistance, and
outstanding tribological characteristics ensure that mechanical seals made of this
material last longer, resulting in much reduced maintenance and production costs
for pump-dependent processes in the chemical industry [277].
SiSiC seal rings (in contrast with those made of pure SSiC) can only be used in acid
media because of the accompanying silicon phase, which is attacked by alkalis [207].
For similar requirements involving radial loads, sliding bearings (see Fig. 34) are
manufactured from SSiC. The erosion and chemical resistance of S i c enable the
designer to position the bearings in the medium to be transported; that is, to elim-
inate lubrication and sealing problems. Other components include shaft protection
sleeves for waste gas exhaust fans, and precision spheres for dosing and regulating
valves.
Due to the tailored properties of liquid phase sintered silicon carbide (LPSSiC) it
is used as dewatering elements in the paper machinery and as rings for highly
stressed gas seals. It is a price competitive alternative to silicon nitride materials
and outperforms alumina and tungsten carbide materials. In addition, LPSSiC is
proposed as neutral matrix in ceramic matrix composites containing plutonium to
burn the world’s stockpiles of military plutonium in thermal or fast reactors [278].
Hot-pressed SIC is the preferred material to replace oxide ceramics for rods
fixtures, and punches in high-temperature strength testing equipment. In view of
the low level of plasma contamination, the low induced radioactivity, and excellent
high-temperature resistance of silicon carbide, it is the ideal material (especially in its

isostatically hot-pressed form) for use in fusion reactors [279].
Combustion tubes made of slip-cast SiSiC [280] have better resistance to corro-
sion, high-temperatures, and thermal shock, so that they will last far longer than,
for example tubes made of heat resistant steel. The heat treatment industry has
begun using SSiC radiant tubes in indirect gas-fired heat treating operations. In
such systems, the tubes are internally heated by combustion burners and radiant
heat to some external work load, such as an ingot of alloy, which is isolated from
the combustion atmosphere [281]. SiSiC and SSiC are destined for use in heat
exchanger systems because of their high-thermal conductivity and corrosion resis-
tance [282-2863.
Sintered silicon carbide bonded to plastic laminate substrates reinforced with
glass or Kevlar TM fabrics can be used as ceramic armor to defeat armor piercing
projectiles [287].
Sintered and isostatically hot-pressed S i c materials, as well as silicon nitride
(Si3N4) are playing an important role in the development of ceramic components
(see Fig. 35) for motor vehicle engines and gas turbines [288-2931. Real technical
success with sintered S i c components in the field of high temperature gas turbines
has not yet been achieved [294,295]. However, S i c and Si3N4gas turbine parts are
in field tests and strong development efforts continue in several countries.
Developments for the application of continuous fiber reinforced S i c matrix
composites (CMCs) have started with and are concentrating on hot components
in military and space technology: hot gas ducts and thermal heat shields for space
Figure 35. Monolithic SSiC gas turbine rotor (courtesy A.Lipp, ESK-Kempten/Germany).
reentry vehicles [173]. Due to the high thermal shock capability of this class of
materials and to the high fracture toughness of some of these composites some
civil applications are gaining importance: highly loaded brake discs, for high
speed trains, and highly loaded journal bearings; in both applications the con-
ventional materials (metals for brakes, monolithic ceramics for bearings) are not
applicable because of the thermal loads and the brittle failure mode respectively.
Acknowledgments
Some parts of the text appeared in the earlier review by K.A.S. (Silicon Carbide in:
Encyclopedia of Advanced Materials pp. 2455-2461,1994) and Elsevier Science Ltd.,
Kidlington OX5 16B, UK is kindly thanked for permission to use them.
References
I. R. Kieffer and F. Benesovsky, Hartstoffe, Springer Verlag,New York, 1963.
2. J. J. Berzelius, Am. Phys. Chem. 1824, 1, 169-230.
3. C. M. Despretz, Compt. Rend. 1849, 29, 709-724.
4. P. Schutzenberger and A. Colson, Compt. Rend. Akad. Sci. 1881,92, 1508.
5. H. Moissan, Der elektrische Ofen, Verlag M. Krayer, Berlin, 1900.
6. E. G. Acheson US Pat. 492 767 (28 Feb. 1893), DE 76629 (1894), DE 85195 (1896).
7. Gmelin Handbook of Inorganic Chemistry, 8th edition, Silicon Supplement Vo.B 3, System No.
15, H. Katscher et al. (eds.), Springer Verlag, Berlin, 1986.
8. H. Moissan, Compt. Rend. Akad. Sci. 1905, 140, 405.
9. A. J. Regis,Land . B. Sand, Bull. Geol. Soc. Am. 1958, 69, 1633.
10. A. P. Bobrievich, V. A. Kalyuzhnyi, and G. I. Smirnov, Proc. Akad. Sci. USSR: Geol. Sci.
1957, 115, 757.
11. J. Bauer, J. Fiala, and R. Hrichovi, Am. Miner. 1963, 48, 620-34.
12. H. Kleykamp and G . Schumacher, Ber. Bunsenges. Phys. Chem. 1993,97, 799-805.
13. A. Dietzel, H. Jagodzinski, and H. Scholze, Ber. Dtsch. Keram. Ges. 1960, 37, 524-537.
14. R. Kieffer, E. Gugel, P. Ettmayer, and A. Schmidt, Ber. Dtsch. Keram. Ges. 1966,43,621-623.
15. N. W. Jeeps and T. F. Page, J. Am. Ceram. Soc. 1981, 64, C177-C178.
16. R. F. Adamsky and K. M. Merz, Z. Kristallogr. 1959, 111, 350-361.
17. H. Ott, Z. Kristallogr. 1925, 61, 515-532; 62, 210-218; 63, 1-19.
18. L. S. Ramsdell, Am. Miner. 1947, 32, 64-82.
19. J. Ruska, L. J. Gauckler, J. Lorenz, and H. U. Rexer, J . Muter. Sci. 1979, 14, 2013-2017.
20. Y. Inomata, Z. Inoue, M. Mitomo, and H. Suzuki, Yogyo-Kyokai-Shi, 1968, 76, 313-319.
21. D. Lundquist, Acta. Chem. Scand. 1948, 2, 177-191.
22. P. T. B. Shaffer, Mater. Res. Bull. 1969, 4, S13-24.
23. M. Mitomo, Y. Inomata, and M. Kumanomido, Yogyo-Kyokai-Shi1970,78, 365-369.
24. M. Mitomo, Y. Inomata, and H. Tanaka, Muter. Res. Bull. 1971, 6, 759-764.
25. M. M. Patience, Silicon Carbide Alloys, PhD. Thesis, University of Newcastle upon Tyne,
1983.
26. D. Foster, Densification of Silicon Carbide with Mixed Oxide Additives, University of
Newcastle upon Tyne, 1996.
27. N. W. Jeeps and T. F. Page, J . Microscopy 1979, 116, 159-171; 1980, 119, 177.
References 741
28. Anon, Ind. Heating, 1954, 992-1004.

29. W. Poch and A. Dietzel, Ber. Dtsch. Keram. Ges. 1962, 39, 413426.
30. K. Liethschmidt, Siliciumcarbid in Chemische Technologie, Band2, Anorg. Techn. I , 4 . Aufl. K.
Winnacker and L. Kuchler (Eds), Hanser-Verlag, Munchen-Wien, 1982, pp. 626-629.
31. K. H. Mehrwald, Ber. Dtsch. Keram. Ges. 1992, 69, 72-81.
32. A. Graf von Matuschka and G. Schonbrunn, Stand der Mikrokorngropenbestimmung von
Elektrokorund und Sic, ZwF 1981, 76, 3 8 4 5 .
33. W. Boecker, H. Landfermann, and H. Hausner, Powder Met. Int. 1981, 13, 37-39.
34. S. Prochazka, Techn. Report 86-CRD-158, General Electric, Schenectady NY, 1986.
35. P. Matje and K. A. Schwetz, Proc. 1st Internat. Conf. Ceramic Powder Processing Science,
Nov. 87, Orlando FL, Gary Messing et al. (Eds), Ceram. Trans. IY88, 1, 46W68.
36. A. W. Weimer, Carbothermal reduction synthesis processes, in Carhide, Nitride and Boride
Materials Synthesis and Processing, A. W. Weimer (Ed.), Chapman & Hall, London, 1997,
pp. 75-180.
37. J. S. Prener, US Patent 3 085 863, 1960.
38. T. L. O'Connor and W. A. McRae, US Patent 3 236 673, 1963.
39. S. Prochazka. Final Report SRD 72-171, General Electric. Schenectady N Y . 1972.
40. K. A. Schwetz and A. Lipp, Radex-Rundschau, 1978, 2, 479498.
41. W. M. Goldberger, A. K. Reed, and R. Morse, Synthesis and characterization of HSC silicon
carbide, in S i c '87, J. D. Cawley (Ed.), Ceram. Trans. 1989, 2, 93-104.
42. F. Hahn, G. Rudakoff, and H.J. Tiller, Hermsdocfer Techn. Mitt. 1990, 79, 2546-2550.
43. H. R. Baumgartner and B. R. Rossing, Pressureless sintering and properties of plasma
synthesized S i c powder, in S i c '87, J. D. Cawley (Ed.), Ceram. Trans. 1989, 2, 3-16.
44. F. G. Stroke (PPG), US Patent 4 295 890, 1981.
45. W. R. Cannon, S. C. Danforth, J. H. Flint, J. S. Haggerty, and R. A. Marra, J . Am. Ceram.
Soc., 1982, 65, 324330.
46. W. Bocker and H. Hausner, Ber. Dtsch. Keram. Ges. 1978, 55, 233-237.
47. P. T. B. Shaffer, Sic Whiskers, in Handbook of Advanced Ceramic Materials, Advanced
Refractory Technologies, Buffalo, NY, 1994.
48. T. N. Tiegs and T. F. Becher, Am. Ceram. Sac. BuN. 1987, 66, 339-342.
49. W. D. G. Bocker, S. Chwastiak, F. Frechette, and S. K. Lau, Single phase CtSiC reinforcements
for composites, in: SIC '87, J. D. Cawley (Ed.), Ceram. Trans. 1989, 2, 407420.
50. P. A. Kistler-De Coppi, and W. Richarz, Int. J . High Techno/. Ceram. 1986, 2, 99-113.
51. B. Meier, R. Hamminger, and E. Nold, Microchim. Acta 1990,II, 195-205.
52. R. Lenk, J. Adler, Influence of forming technique on the Sic platelet orientation in a liquid-
phase sintered Sic-matrix, in Fourth Euroceramics, Vol. 2, Part II. Basic Science, C. Galassi
(Ed.), Faenca Editrice Italy, pp. 407414.
53. R. Lenk, A. F. Kriwoschepov, and K. Grope, Heipgiepen von drucklos gesintertem Sic,
Sprechsaal Ceram. Mater. 1995, 128, 17-20.
54. S. Yajima, J. Heyashi, M. Omori, and K. Okamura, Nature, 1976, 261, 683.
55. T. Ishikawa, S i c continuous fiber, Nicalon, in SIC Ceramics-2, S. Somiya and Y. Inomata
(Eds), Elsevier, Amsterdam, 1991, pp, 81-98.
56. H.-P. Baldus and M. Jansen, Angew. Chem. 1997, 109, 338-354.
57. D. Heimann, T. Wagner, J. Bill, F. Aldinger, and F. F. Lange, J . Mater. Res. 1997, 12, 3099-
3101.
58. S. Prochazka, Sintered Si c and method of making, US Patent 566 8068, 1997.
59. K. K. Kappmeyer, D. H. Hubble, and H.W, Powers, Am. Ceram. Soc. Bull. 1966,45, 1060-1064.
60. M. E. Washburn and R. W. Love, Am. Ceram. Soc. Bull. 1962, 41, 447449.
61. A. Fickel, S i c Materials, in Refractory Materials - Pocket Manual, G. Routschka (Ed.),
Vulkan Verlag, Essen, 1997, pp. 74-80.
62. R. Van der Beck and J. O'Connor, Ceram. Ind. 1957, No. 3 , 9 6 9 8 .
63. R. A. Alliegro, Processing and fabrication of non-hotpressed Sic, in Ceramics,for High Per-
formance Applications, J. J. Burke et a/., (Eds), Metals and Ceramics Inf. Center, Columbus,
OH, 1974, pp. 253-263.
64. J. Kriegesmann, Interceram. 1988, No. 2, 27-30 and Powder Metall. Int. 1986, 18, 341-343.
65. P. Popper and D. G. S. Davies, Powder Metall. 1961, 8, 113.
66. K. M. Taylor, Improved K TSiC for high temperature parts, Muter. Meth. 1956, Vol. 44, No.
4, 92-95; US Patent 3 189 472, 1965.
67. R. Kieffer, E. Gugel, and A. Schmidt, Verfahren zur Herstellung eines dichten Formkorpers
auf Basis Sic, DE-OS 1 671 092, 1967.
68. P. Kennedy and J. V. Shennan, Engineering applications of REFEL silicon carbide, in Silicon
Carbide 1973, R. C . Marhall et a/. (Eds), Univ. of South Carolina Press, Columbia, 1974,
pp. 359-366.
69. G. Q. Weaver, H. R. Baumgartner, and M. L. Torti, Special Ceramics 6, P. Popper (Ed.), Brit.
Ceram. Res. Association, UK, pp. 261-281.
70. W. B. Hillig, R. L. Mehan, C. R. Morelock, V. J. De Carlo, and W. Laskow, Am. Ceram. Soc.
Bull. 1975, 54,1054-1056.
71. T. Hase et al., J . Nucl. Muter. 1976, 59, 4248.
72. Anon. A super ceramic called REFEL, Interceram. 1982, No. 1, 50-51.
73. G. Willmann and W. Heider, Werkstofftechnik1983, 14,158.
74. K. Taylor, Dense silicon carbide, US Patent 3 205 043, 1962.
75. S. Prochazka, The sintering process for Sic, a review, Techn. Report 81-CRD-314, General
Electric, Schenectady NY, 1981, pp. 16.
76. S. Prochazka, Sintering of Sic, in Ceramics for High Performance Applications, J. J. Burke et al.
(eds.), Metals and Ceramic Info. Center, Columbus, OH, 1974, pp. 239-252.
77. J. A. Coppola and G. H. McMurtry, Substitution of ceramics for ductile materials in
design, Nut. Symp. on Ceramics in the Service of Man, Carnegie Institution, Washington
DC, 1976.
78. P. T. B. Shaffer, The Si c phase in the system SiC-B4C-C, Muter. Res. Bull. 1969,4, 213-220.
79. R. Hamminger, Carbon inclusions in sintered Sic, J . Am. Ceram. SOC.1989, 72, 1741-1744.
80. Y. Murata and R. H. Smoak, Densification of Sic by the addition of BN, BP and B4C, and
correlation to their solid solubilities, in Proc. Int. Symp. Densijcation and Sintering, Oct. 1978,
Hakone/Japan, S. Somiya and S. Saito (Eds), Gakujutsu Bunken, Tukyu-Kai, Tokyo, 1979,
pp. 382-399.
81. W. Boecker and H. Hausner, Powder Metall. In?. 1978, 10, 87-89.
82. D. R. Stutz, S. Prochazka, and J. Lorentz, 1. Am. Ceram. SOC.1985, 68,479482.
83. T. Fetahagic and D. Kolar, Ceram. Acta 1990, 2, 31-37.
84. K. A. Schwetz, A. Lipp, The effect of boron and aluminum sintering additives on the proper-
ties of dense sintered aSiC, Science of Ceramics 10, Berchtesgaden Sept. 1979, H. Hausner
(Ed.), Verlag Deutsche Keramiske Ges 1980, pp. 149-158.
85. W. Grellner, K. A. Schwetz, and A. Lipp, Fracture phenomena of sintcred alphasic, in Proc.
7th Symp. on Special Ceramics, Bedford College, London, Dec. 1980, D. Taylor and P. Popper
(Eds), Br. Ceram. Res. Assoc., Stoke on Trent 1981, pp. 27-36.
86. W. Boecker, H. Landfermann, and H. Hausner, Powder Metall. Int. 1979, 11, 83-85.
87. H. Hausner, Pressureless sintering of non-oxide ceramics, in 4th CIMTEC, Energy and Cera-
mics, St. Vincent, Italy 1979, P. Vincenzini (Ed.), Elsevier Scientific Publ. Comp., Amsterdam,
1980, pp. 582-595.
88. H. Tanaka, Y. Inomata, K. Hara, and H. Hasegawa, J . Muter. Sci. Lett. 1985, 4,315-317.
89. K. Suzuki, Pressureless sintering of SIC with addition of AI2O3,Rep. Rm. Lab. Asahi Glass Co.
1986,36, 25-36.
90. K. Suzuki, Pressureless sintering of S i c with addition of AI2O3,in SIC Ceramics - 2, S. Somiya
and Y. Inomata (Eds), Elsevier, London, 1991, pp. 163-182.
91. R. H. Smoak, Pressureless sintering beryllium containing SIC powder composition, DE-OS 27
51 851, 1977.
92. A. Mohr, Untersuchungen zur Minimierung der Additivgehalte fur die drucklose Sinterung
von aSiC, Diplomarbeit Institut fur Keramik im Maschinenbau, Universitat Karlsruhe, 1989.
93. D. W. Richerson, Modern Ceramic Engineering, Dekker, New York, 1992.
94. F. Thummler and R. Oberacker, Introduction to Powder Metallurgy, The Institute of Materials
Series, University Press, Cambridge, 1993.
95. C. A. Johnson and S. Prochazka, Microstructures of sintered Sic, in Ceramic Microstructures
1976, R. M. Fullrath and J. P. Pdsk (Eds), Westview Press, Boulder, CO, 1977, pp. 366-378.
96. K. A. Schwetz, F. Isemann, and A. Lipp, Injection molded sintered turbine components of Al-
References 143
doped alpha-Sic; in Proc. 1st Internat. Symp. Ceramic Comp. For Engine, S. Somiya et a/.
(Eds), KZK Scientific, Tokyo, 1984, pp. 583-594.
97. Showa Denko, DE 3 927 300, 1988.
98. Ceramiques et Composites, European Patent 486 336, 1991.
99. Carborundum Company, US Patent 5 589 428, 1996; US Patent 5 635 430, 1997.
100. ESK-GmbH, European Patent 685 437, 1995.
101. Carborundum Company, European Patent 145 496, 1984.
102. J. Greim, L. Sigl, and H. Thaler, Sintered SIC for high-performance bearings and seals, Maga-
zine of Wacker Chernie-GmhH Werk+ Wirken: Int. Edn, 1997, 1, 23-25.
103. M. Omori and H. Takei, J. Am. Ceram. Soc. 1982, 65, C-92; and US Patent 5 439 853,
1995. .
104. E. Kostic, Powder Metall. Int. 1988, 20, 28-29.
105. R. A. Cutler and T. B. Jackson, Liquid phase sintered Sic, in Ceramic Materials and Components
for Engines, Proc. 3rd. Int. Symp., V. J. Tennery (Ed.), Am. Ceram. Soc., 1988, pp. 309-318; and
US Patent 4 829 027, 1988.
106. W. Boecker and R. Hamminger, Interceram. 1991, 40, 520-525.
107. F. K. Van Dijen and E. Mayer, J. Europ. Ceram. Soc. 1996, 16, 413420.
108. C. Wolf, H. Hubner, and J. Adler, Mechanical behavior of pressureless sintered SIC at high
temperature, in 3rd Euro-ceramics, Vol.3, P. Duran and J. F. Fernandez (Eds), Faenze Editrice
Iberica, Castellon, Spain, 1993, pp. 465470.
109. K. Y. Chia and S. K. Lau, Ceram. Eng. Sci. Proc. 1991, 12, 1845-1861.
110. K. A. Schwetz, P. Matje, M. Mohr, and K. P. Martin, Development of a pressureless sintering
process for ceramics in the system SiC--AlN (0),Final Report on German BMFT Project No.
03-zC-153, 1986 (unpublished).
111. I. Wiedmann, M. Nader, M. J. Hoffmann, and F. Aldinger, in Symposium 7, Material-
wissenschaftliche Grundlagen, Werkstqfimhe '96, F. Aldinger and H. Mughrabi (Eds),
DGM Informdtionsgesellschaft mbH, Oberursel, 1997, pp. 51 5-520.
1 12. L. S. Sigl and H. J. Kleebe, J. Am. Ceram. Soc. 1993. 76, 773-776.
113. M. P. Padture, J. Am. Ceram. Soc. 1994, 77, 519-523; 2518--2522.
114. S. K. Lee and C. H. Kim, J. Am. Ceram. Soc. 1994, 77, 1655-1658.
115. J. S. Nadeau. Ceram. Bull. 1973, 52, 170-174.
116. R. A. Alliegro, L. B. Coffin, and J. R . Tinklepaugh, J. Am. Ceram. Soc. 1956.39, 386389.
117. S. Prochazka and R. J. Charles, Am. Ceram. Soc. Bull. 1973, 52, 885-891.
118. J. Kriegesmann, Ber. Dtsch. Keram. Ges. 1978, 55, 391-397.
119. F. F. Lange, J. Mafer. Sci. 1975, 10, 314320.
120. J. M. Bind and J. V. Biggers, J. Appl. Phys, 1976, 47,5171-5174.
121. D. Broussaud, Independence of composition of hot-pressed Sic, in Ceramic Microstructures
'76, R. M. Fullrath and J. A. Pask, (Eds), Westview Press, Boulder, CO, 1976, pp. 679-688.
122. T. Iseki, K. Arakawa, H. Matsuzaki. and H. Suzuki, Yogyo-Kyokai-Shi1983, 91, 349.
123. K. Nakamura and 0. Asai, Kagaku Kogyo 1982, 33, 977.
124. H. Kessel and E. Gugel, Industrie Diamanten Rundschuu 1978, 12, 180-185.
125. J. Kriegesmann, K. Hunold, A. Lipp, K. Reinmuth, and K. A. Schwetz, European Patent
71241. 1981.
126. T. J. Whalen, R. M. Williams, and B. N. Juterbock, HIP of Sic and Si3N4structural ceramics,
in IOth Plansee Seminar Proceedings, Reutte, Tyrol, 1985, p. 783.
127. K. Hunold, Powder Metall. Int. 1984, 16, 236-238; 1985, 17, 91-93.
128. K. Hunold, Sinter/HIP of Sic, in Proc. Adv. Mat. Technology Ceramic Workshop No. 4 ,
Advances in Materials, Processing and Manufacturing Science, Nagoya Japan, March 3 4 ,
Japan Fine Ceramics Center, 1988, pp. 49-62.
129. R. Oberacker, A. Kiihne, and F. Thummler, Powder Metall. Int. 1987, 19, 43-50.
130. F. Fetahagic, R. Oberacker, and F. Thummler, Process development for sinter-HIPing of Sic,
in Ber. KFA-Jiihlich 1989, Juel. Conf: -77, Emerging Mat. Adv. Process., 9th German-Yugosla-
vian Meeting on Materials Science Development, April 1 6 1 9 , 1989, pp. 313-325.
131. T. Hirai and M. Sasaki, S i c prepared by CVD, in S i c Ceramics-I, S. Somiya and Y. Inomata
(Eds), Elsevier, London, 1991, pp. 77-98.
132. Anon, Material Innovations: CVD scaled up for commercial production of bulk Sic, Am.
Ceram. Soc. Bull. 1993,12,74-78.

133. M. Tenhover, I. Ruppel, S. S. Lyle, and L. J. Pilione, DC-magnetron sputterable Sic, in Proc.
36th Annual Techn. Conf: Society of Vacuum Coaters, 1993,pp. 362-365.
134. Anon, S i c electronic materials and devices, in M R S Bull. 1997,22,No. 3 (Special Issue); Cree
Research Inc., World Patent 97/28297,Growth of colorless S i c crystals, 1997.
135. J. A. Lely, Ber. Dtsch. Keram. Ges. 1955,32, 229-231.
136. R.Fantoni, E. Borsella, S. Piccirillo, R. Ceccato, and S. Enzo, J. Muter. Res. 1990,5,143-150.
137. R. Vassen, D. Stover, and J. Uhlenbusch, Sintering and grain growth of ultrafine amorphous
Sic-Si-Powder mixtures, in Euro-Ceramics II, Vol. 2 , Structural Ceramics and Composites,
Proc. 2nd European Ceram. SOC.Conf. Augsburg, Sept. 1991, G. Ziegler and H. Hausner
(Eds), Deutsche Keram. Ges., Koln, pp. 791-797, 1993.
138. R. Vassen, H. P. Buchkremer, and D. Stover, Verfahren zum Herstellen feinkristalliner
Siliciumkarbidkorper, DE 196 42 753 (1999).
139. R. Vassen and D. Stover, Philos. Mag. B 1997,76, 585.
140. J. Forster, R. Vassen, and D. Stover, J . Muter. Sci. Lett. 1995, 14, 214-216.
141. R. Vassen, J. Forster, and D. Stover, NanoStructured Muter. 1995,6, 889-892.
142. A. Kaiser, R. Vassen, T. Stover, and H. Buchkremer, NanoStructuredMater. 1997,8,489497.
143. M.Mitomo, Y.-W. Kim, and H. Hirotsuru, J . Muter. Res. 1996, 11, 1601-1604.
144. M. Mitomo, H. Hirotsuru, and Y.-W. Kim, US Patent 5 591 685,7 Jan. 1997.
145. R. Telle, R.J. Brook, and G.Petzow, J . Hard Muter. 1991,2, 79-114.
146. G. C.Wei and P. F. Becher, J . Am. Ceram. SOC.1984,67,571.
147. M. A. Janney, Am. Ceram. SOC.Bull. 1987,66,322.
148. D. J . Jiang, J. H. Wang, Y. L. Li, and L. T. Ma, Muter. Sci.Eng. 1989,A109, 401.
149. D.Ly Ngoc, Gefiigeverstarkung von Sic Keramiken, doctoral thesis, University of Stuttgart,
Germany, 1989.
150. R.Jimbou, K. Takahashi, and Y. Matsushita, Adv. Ceram. Muter. 1986, 1, 341.
151. C.H. McMurtry, W. D. G. Backer, S. G. Seshadri, J. S. Zanghi, and J. E. Garnier, Am.
Ceram. SOC.Bull. 1987,66,325.
152. H. Cai, W. H. Gu, and K. T. Faber, in 5th Techn. Con$ On Composite Materials, Proc. Am.
SOC.Comp., 1990,pp. 892-901.
153. K. T. Faber, W. H. Gu, H. Cai, R. A. Winholtz, and D. J. Magleg, in Toughening Mechanisms
in Quasi-Brittle Materials, S. P. Shah (Ed.), Kluwer, Dordrecht, 1991, pp. 3-17.
154. T. Tani, S.Wada, European Patent 303192, 1988.
155. C. Blanc, F. Thevenot, and D. Treheux, Elaboration and characterization of submicronic SiC-
TiB2 composites by reactive pressureless sintering, in Key Eng. Muter. 1997,132-136,968-971.
156. D. H.Kuo and W. M. Kriven, J . Europ. Ceram. SOC.1998,18,51-57.
157. K. A. Schwetz, K. Reinmuth, and A. Lipp, World Ceramics, 1985,2, 70-84.
158. F.Thevenot, Sintering of B4C and B4C-SiC two-phase materials and their properties, in Proc.
9th Int. Symp. on Boron, Borides and Related Compounds, H. Werheit (Ed.), University of
Duisburg Germany,l987, pp. 246-256.
159. W. J. Tomlinson and J. C. Whitney, Ceram. Int. 1992,18,207-21 1.
160. V. Kevorkiijan, A. Bizjak, J. Vizintin, F. Thevenot, G. Interdonato, and C. Reimondi, B4C-
S i c based material for wear applications, in 4th Euro Ceramics Vol.4, Basic Science, A. Bellosi
(Ed.), Faenza Editrice, 1995,pp. 209-216.
161. I. B. Cutler, P. D. Miller, W. Rafaniello, H. K. Park, D. P. Thompson, and K. H. Jack, Nature
1978,215,434-435.
162. W. Rafaniello, K. Cho, and A. V. Virkar, J. Muter. Sci. 1981, 16,3479-3488.
163. W. Rafaniello, M. R. Plichta, and A. V. Virkar, J . Am. Cerm. SOC.1983,66, 772-76.
164. R. Ruh and A. Zangvil, J . Am. Ceram. Soc. 1982,65,260-265.
165. Y. Xu,A. Zangvil, M. Landon, and F. Thevenot, J. Am. Ceram. Soc. 1992,15,325-333.
166. A. Zangvil and R. Ruh, Alloying of S i c with other ceramic compounds: A Review, in Silicon
Carbide '87, J. D. Cawley and L. E. Semler (Eds), Am. Ceram. SOC.,Westerville, OH, 1989,
pp. 63-82.
167. S. Y.Kuo, A. V. Virkar, and W. Rafaniello, J. Am. Ceram. SOC.1987,70, C-125.
168. R. R. Lee and W. C. Wei, Ceram. Eng. Sci. Proc. 1990, 11, 1094-1121.
169. T. B. Jackson, A. C. Hurford, S. L. Brunner, and R. A. Cutler, Sic-based-ceramics with
References 745
improved strength, in Silicon Carbide '87, J. W. Cawley and C. Semler (Eds), Am. Ceram. Soc.,
Columbus, OH, 1989, pp. 227-240.
170. J. L. Huang, A. C. Herford, R. A. Cutler, and A. V. Virkar, J . Muter. Sci. 1986,21, 1448-1456.
171. J . M. Lihrmann and J. Tirlocq, Process for producing dense products based on S i c and
composite products thus obtained, World Patent, PCT, WO 97j06119, 1997.
172. D. P. Stinton, R. A. Lowdon, and R. H. Krabill, Mechanical property characterization of
fiber-reinforced Si c matrix composites, ORNL/TM- 11524, April 1990.
173. H. Wurtinger and A. Miihlratzer, Cost effective manufacturing methods for structural ceramic
matrix composite (CMC) components, A S M E Paper 96-GT-296, 1996.
174. E. Fitzer and R. Gadow, Fiber reinforced Sic, Am. Ceram. Soc. Bull. 1986, 65, 326335.
175. G. S. Corman, M. K. Brun, P. J. Meschter, K. L. Luthra, and R. Eldrid, Toughened silcomp
ceramic composites for gas turbine applications, Proc. 39th Internat. S A M P E Symposium,
April 1 1-1 3 1994, pp. 2300-23 13.
176. Ullmanns Encyclopedia of Technical Chemistry (E. Bartholome et al., eds.), 4th edn., 21,
Verlag Chemie, Weinheim, 1982, 43 1438.
177. P. Wecht, Feuerfest-Siliciumcarbid, Applied Mineralogy, Vol. 11, Springer Verlag, New York,
1977.
178. K. Ziickler, in Halbleiterprobleme 111, W. Schottky (Ed.), 1956, pp. 207-229.
179. S. H. Hagen, Ber. Dtsch. Keram. Ges. 1970, 47, 630-634.
180. E. Gugel, P. Schuster, and G. Senftleben, Stahl Eisen 1972,92, 144-149.
181. Anon, JANAF Thermochemical Tables, 2nd edn., NSRDS-NBS37, Washington DC, June
1971.
182. K. Niihara, Am. Ceram. Soc. Bull. 1984,63, 1160-1164.
183. T. F. Page, Proc. Brit. Ceram. Soc. 1978, 26, 193-208.
184. W. Kollenberg, B. Mossner, and K. Schwetz, VDI-Berichte 1990, Nr. 804, 347-358.
185. M. E. Washburn and W. S. Coblenz, Ceram. Bull. 1988,67, 356-363.
186. F. Thummler, Sintering and high temperature properties of Si3N4 and Sic, in Sintering
Processes, G. C. Kuczynski (Ed.), Plenum, New York, 1980, pp. 247-277.
187. J. Kriegesmann, A. Lipp, K. Reinmuth, and K. A. Schwetz, Strength and fracture toughness
of SIC, in Ceramics for High Performance Appl. 111, E. M. Lenoe et al. (Eds), Plenum, New
York, 1984, pp. 737-75 1.
188. R. Hamminger, R. G. Grathwohl, and F. Thummler, Microchemistry and high temperature
properties of sintered Sic, Proc. 2nd Int. ConJ Science of Hard Materials, Rhodos Sept.
1984, Inst. of Physics Conf. Series No. 75, Hilger, Bristol, 1986, pp. 279-292.
189. G. Grathwohl, Th. Reets, and F. Thiimmler, Creep of hot-pressed and sintered S i c with
different sintering additives, in Sci. Ceram. 1981, 11, 425431.
190. J. L. Chermant, R. Moussa, and F. Osterstock, Rev. Int. Hautes Temp. Refract. 1981,18,5-55.
191. G . Wiebke, Ber. Dtsch. Keram. Ges. 1960, 37, 219-226.
192. J. Schlichting, Ber. Dtscli. Keram. Ges. 1979, 56, 196199 and 256261.
193. J. Schlichting and K. Schwetz, High Temp. High Press. 1982, 14, 219-223.
194. B. Frisch, W. R. Thiele, R. Drumm, and B. Miinnich, Ber. Dtsch. Keram. Ges. 1988,65,277-284.
195. K. G. Nickel, Z. Fu, and P. Quirmbach, Trans A S M ;J. Eng. Gas Turb. Power 1993,115,7682.
196. K. Konopicky, I. Patzak, and H. Dohr, Glus Email Keramo. Tech. 1972, 23, 81-87.
197. D. W. McKee and D. Chatterji, J . Am. Ceram. Soc. 1976, 59,441444.
198. N. W. Jepps and T. F. Page, J . Microscopy 1981, 127, 227-237.
199. P. T. B. Shaffer, Ceram. Age 1966,82, 4244.
200. T. Iseki, T. Kameda, and T. Maruyama, J . Muter. Sci. 1984, 19, 1692-1698.
201. T. Labus, Lubr. Eng. 1981,37, 387-394.
202. J. H. Eisner, Chemie-Anlagen + Verfahren, 1982, p. 46, p. 51, p. 54.
203. D. W. Richerson, Muter. Sci. Res. 1981, 14, 661-676.
204. K. Miyoshi, Proc. Int. Con$ Wear of Mat., ASME, San Francisco 1981.
205. R. W. Lashway, S. G. Seshadri, and M. Srinivasan, Various forms of Sic and their effects on
seal performance, 38th Ann. Mtg. of the ASLE, Houston, Texas, 1983.
206. H. Knoch, J. Kracker, and A. Schelken, Chemie-Anlagen + Verfahren, 1983, 28-30.
207. H. Knoch, J. Kracker, and A. Schelken, Cliemie-Anlagen + Veufahren, 1985, 101-104.
208. H. Knoch, J. Kracker, and A. Schelken, World Ceramm 1985, Vol. 2, 96-98.
209. J. Derby, S. G. Seshadri, and M. Srinivasan, Fracture Mechanics of Ceramics, Vol. 8, R. C.

Bradt et a/. (Eds), Plenum, New York, 1986, pp. 113-125.
210. D. C. Cranmer, J. Muter. Sci. 1985, 20, 2029-2037.
211. K. Miyoshi, D. H. Buckley, and M. Srinivasan, Ceramic Bull. 1983, 62, 494500.
212. J. Breznak, E. Breval, and N. H. Macmillan, J . Mat. Sc
213. E. Breval, J. Breznak, and N. H. Macmillan, J . Mat. S
214. C. C. Wu, R. W. Rice, B. A. Platt, and S. Carrir, Ceram. Sci. Eng. Proc. 1985, 6, 1023.
215. J. R. Smythe and D. W. Richerson, Ceram. Sci. Eng. Proc. 1983, 4, 663473.
216. L. J. Lindberg and D. W. Richerson, Ceram. Sci. Eng. Proc. 1985, 6, 1059-1066.
217. H. Tomizawa and T. E. Fisher, A S L E Tans. 1987,30,4146.
218. H. Knoch and J. Kracker, cfi Dtsch. Keram. Ges. 1987,64, 159-163.
219. P. Boch, F. Platon, and G. Kapelski, J . Europ. Ceram. Soc. 1989, 5,223-228.
220. K. H. Habig and M. Woydt, in Proc. 5th Znt. Congress Tribology, Vol. 3, K. Holmberg and
I. Nieminen (Eds), Lansi-Savo Oy, St Michel, Finland, 1989, 106.
221. Y. Yamamoto, K. Okamoto, and A. Ura, in Proc. 5th Znt. Congress Tribology, Vol. 3,
K. Holmberg and I. Nieminen (Eds), Lansi-Savo Oy, St Michel, Finland, 1989, 138.
222. J. M. Martin, T. LeMogne, H. Montes, and N. N. Gardos, in Proc. 5th Znt. Congress Tribology,
Vol. 3, K. Holmberg and I. Nieminen (Eds), Lansi-Savo Oy, St Michel, Finland, 1989, 132.
223. S. Sasaki, Wear 1989, 134, 185-200.
224. J. Denape and J. Lamon. J . Muter. Sci. 1990, 25, 3592-3604.
225. M. Woydt, A. Skopp, and R. Wasche, Proc. 4th Internat. Symp. Ceramic Materials and Compo-
nents for Engines R. Carlsson eta/. (Eds), Elsevier, Amsterdam, 1992, pp. 1219-1239.
226. 0.0.Ajayi, A. Erdemir, R. H. Lee, and F. A. Nichols, J . Am. Ceram. Soc. 1993,76,511-517.
227. P. Maurin-Perrier, J. P. Farjandon, and M. Cartier, Wear Muter. 1991, 2, 585-588.
228. P. Maurin-Perrier, BRITE-EURAM Project Proposal P-2231, Contract No. RJ-lb-295, 1992.
229. N. N. Gardos, Determination of the tribological fundamentals of solid/lubricated ceramics,
WRDC-TR-904096, Hughes Aircraft, El Segundo, CA 90245, 1990.
230. B. Loffelbein, M.Woydt, and K. H. Habig, Wear 1993, 162-164, 220-228.
231. P. Andersson and A. Blomberg, Wear 1994, 174, 1-7.
232. S. Kitaoka, T. Tsuji, T. Katoh, Y. Yamaguchi, and K. Kashiwagi, J . Am. Ceram. Soc. 1994,
77, 1851-1856.
233. H. Czichos and K.-H. Habig, Tribologie Handbuch Reibung und Verschleip, Vieweg Verlag,
-
Braunschweig/Wiesbaden, 1992.
234. H. Knoch, L. Sigl, and W. D. Long, Product development with pressureless sintered Sic, 1
Oct. 1990, 37th Sagamore Army Mat. Res. Conf. AMTL-Watertown MA.
235. J. Greim, H. Thaler, and L. Sigl, Gesintertes Siliciumcarbidfur hochbeanspruchte Komponenten
in der Lager- und Dichtungstechnik, in VDZ- Bericht 1331, Dusseldorf, 1997, pp. 153-159.
236. DIN 51 075 part 1-5, 1982, Beuth Verlag GmbH, Berlin/Germany.
237. I S 0 9286, 1997, Beuth Verlag GmbH, Berlin/Germany.
238. K. A. Schwetz and J. Hassler, J . Less Common Metals, 1986, 117, 7-15, and DIN 51079-3,
1997.
239. DIN 51 076, part 1, 1991, Beuth Verlag GmbH, Berlin/Germany.
240. F. F. van Dijen, Interceram 1993, 42, 92-94.
241. J. A. C. Broekdert, R. Brandt, F. LeiP, C. Pilger, D. Pollmann, P. Tschopel, and G. Tolg, J . Anal.
Atom. Spectr. 1994,9, 1063-1070.
242. G. Zaray, F. Leis, T. Kantor, J. Hassler, and G. Tolg, Analysis of silicon carbide powder by
ETV-ICP-AES, Fres. J . Anal. Chem. 1993,346, 1042-1046.
243. DIN-IS0 8486, part 1-2, 1997, Beuth Verlag GmbH, Berlin/Germany.
244. DIN 69 176, part 1-3, 1985.
245. FEPA 44-D-1986, Fachverband Elektrokorund-und Sic-Hersteller, Fr.a.M.
246. I S 0 787, part 11, 1991, Beuth Verlag GmbH, Berlin/Germany.
247. V. Carle, U. Schafer, U. Taffner, F. Predel, R. Telle, and G. Petzow, Ceramography of high
performance ceramics, Part I1 silicon carbide, Pruct. Met. 1991, 28, 420-434.
248. W. Kollenberg, Prufverfahren keramischer Hochleistungswerkstoffe und deren Grundlagen,
KfA-Bericht Nr. 470, Zentralbibiothek der Kernforschungsanlage Julich GmbH, 1988.
249. L. L. Hench and R. W. Gould, Characterization of Ceramics, Dekker, New York, 1971.
References 747
250. 0. Van der Biest, Analysis of High Temperature Materials, Applied Science Publishers,
London and New York, 1983.
25 1. H. J. Hunger, Werkstqffinalytische Verfahren, Deutscher Verlag fur Grundstoffindustrie,
Leipzig/Stuttgart, 1995.
252. FEPA-ESK, S i c material safety data sheet, according to 9ljl55 EEC and ISO-Standard
11014, Edition 1, issued 22-05-97, 5 pages.
253. D. Henschler (Ed.), Gesundheitsschadliche Arbeitsstoffe, Toxikologisch-arbeitsmedizinische
Begriindung von MAK-Werten, VCH, Wcinheim, 1987.
254. J. Bruch, Institut fur Hygiene und Arbeitsmedizin, Universitatsklinikum Essen, 1993.
255. I. D. Birchall et al., J . Muter. Sci. Lett. 1988, 7, 350-352.
256. ASTM, Standard Practice for Handling S i c whiskers, E 1437-91, Subcommittee E 34.70,
American Society for Testing and Materials, Philadelphia, USA, I99 1.
257. L. Coes, Jr., Abrasives, in Applied Mineralogy, Vol. I , Springer Verlag, Wien, 1971.
258. K. Martin, Neue Erkenntnisse iiber den Werkstoffsabtragsvorgang beim Llppen, Fachber-
ichte Oberflachentechnik 1972, 10, 197-202.
259. G. Spur and I. Sabotka, ZwF 1987,82, 275-380.
260. G. Spur and D. Simpfendorfer, ZwF 1988, 83, 207-212.
261. H. B. Britsch, Ber. Dtsch. Keram. Ges. 1976, 53, 143-149.
262. T. Benecke, S. Venkateswardn, W. D. Schubert, and B. Lux, GiePerei 1993,80, 256-662.
263. S. V. Nair, J. K. Tien, and R. C. Bates, Int. Met. Ref. 1985, 30, 275-290.
264. A. Lipp, Feinwerktechnik 1970,74, 150-154.
265. E. Gugel, Ber. Dtsch. Keram. Ges. 1966, 43, 354359.
266. A. F. Fickel, Sprechsaal 1980, 113, 517-531; 1980, 113, 737-747.
267. R. Rasch and H. Maatz, Maschinenschaden 1978, 51, 145-147.
268. Z. Stavric and M. Hue, Crystar, Keram. Z. 1975, 27, 125-128.
269. G . Bierbauer, Keram. Z . 1972, 24, 142-145.
270. 0. Rubisch and R. Schmitt, Ber. Dtsch. Keram. Ges. 1966, 43, 173-179.
271. E. Buchner and 0. Rubisch, Corrosion behavior of S i c heating elements, in Silicon Carbide -
1973, R. C. Marshall et al. (Eds), Univ. South Carolina Press, Columbia, SC, 1974, pp. 428434.
272. Y. Nakamura and S. Yajima, Am. Ceram. Soc. Bull. 1982, 61, 572-573.
273. W. Huether, Thermoelement zur Temperaturmessung und Verfahren zur Herstellung dessel-
ben, European Patent 72 430, 1982.
274. K. Kuomoto, M. Shimohigashi, S. Takeda, and H. Yanagida, J. Muter. Sci. Lett. 1987, 6,
1453-1 455.
275. R. Davis, Recent advances regarding the definition of the atomic environment, film growth
and microelectronic device development in Sic, in The Physics and Chemistry of’ Carbides,
Nitrides and Borides, R. Freer (Ed.), Kluwer, Dordrecht, 1990, pp. 589-623.
276. M. Srinivasan, The silicon carbide family of structural ceramics, in Structural Ceramics Trea-
-
tise on Materials Science and Technology, Vol. 29, J. B. Wachtman (Ed.), Academic, San
Diego, CA, 1989, pp. 100-159.
277. D. Zeus, How the use of advanced ceramics as tribomaterial has effected the evolution of
mechanical seals, cfi Ber. Dtsch. Keram. Ges. 1991, 68, 3-5.
278. V. D. Krstic, M. D. Vlajic, and R. A. Verrall, SIC ceramics for nuclear applications, in
Advanced Ceramic Materials, H. Mostaghaci (Ed.), Key Eng. Muter. 1996, 122-124, 387-396.
279. F. Porz, G. Grathwohl, and R. Hamminger, J . Nucl. Muter. 1984, 124, 195-214.
280. G. Heuschmann and G. Willmann, Interceram. 1986, 1, 24-29.
281. D. P. Butt, R. E. Tressler, and K. E. Spear, Discontinuous phase formation and selective
attack of S i c materials exposed to low oxygen partial pressure environments, in Corrosion
of Advanced Ceramics, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 153-164.
282. R. A. Penty and J. W. Bjerklie, S i c for high-temperature heat exchangers, in Ceram. Eng. Sci.
Proc. 1982, 3, 120-127.
283. J. Heinrich, J. Huber, S. Foster, and P. Quell, Advanced ceramics as heat exchangers in
domestic and industrial appliances, in High Tech Ceramics, P. Vincenzini (Ed.), Elsevier,
Amsterdam, 1987, pp. 2427-2440.
284. M. C. Kerr, Advanced ceramic heat exchangers utilizing Hexology-SA, single phase S i c tubes,
in High Tech Ceramics, P. Vincenzini (Ed.), Elsevier, Amsterdam, 1987, pp. 2441-2449.
285. W. Heider, Verwendung von SiSiC im Apparatebau, Proc. Techkeram. '87, WiesbadenlFRG,
S . Schnabel and J. Kriegesmann (Eds), Demat Exposition Managing, 1987, pp. 18.01-18.19.
286. W. Hof, Hexoloy S i c SA ein neuer Werkstoff fur Rohrbiindel-Waremetauscher, Chemi.
-
Techn. 1991,20, 18-22.

287. B. Matchen, Applications of ceramics in armor products, in Advanced Ceramic Materials,
H. Mostaghaci (Ed.), Key Eng. Mat. 1996, 122-124, 333-342.
288. R. S. Storm, R. W. Ohnsorg, and F. J. Frechette, Fabrication of injection molded sintered
alpha-Sic turbine components, Trans. ASME: J. Eng. Power. 1982, 104, 76.
289. K. A. Schwetz, W. Grellner, K. Hunold, A. Lipp, and M. Langer, HIP-treated sintered Sic
turbocharger rotors, in Proc. 2nd Intern. Symposium Ceramic Materials and Components for
Engines, W. Bunk, H. Hausner (Eds), DKG Verlag, 1986, pp. 1051-1062.
290. R. W. Ohnsorg and M. 0. Ten Eyck, Fabrication of sintered a-Sic Turbine Engine Comp., in
Silicon Carbide '87, J. D. Cawley and C. E. Semler (Eds), Ceram. Trans., Vol. 2, Am. Ceram.
SOC.,Westerville, OH, pp. 367-386.
291. R. Westerheide, T. Hollstein, and K. A. Schwetz, Tension-compression testing of HIP-treated
sintered Sic for gas turbine applications at temperatures between 1400 and 1600"C, Proc. 6th
Intern. Symp. on Ceramic Materials and Components for Engines, October 1997, Arita, Japan,
(K. Niihara et al., eds.), 253-258, 1998.
292. Ch. Gutmann, A. Schulz, and S. Wittig, A new approach for a low-cooled ceramic nozzle
vane, ASME-Paper 96-GT-232, 1996.
293. M. Dilzer, Ch. Gutmann, A. Schulz, S. Wittig, Testing of a low-cooled ceramic nozzle vane
under transient conditions, paper presented at 43th ASME Gas Turbine and Aeroengine Tech-
nical Congress, June 2-5 1998, Stockholm/Schweden, Paper No. 98-GT-116, ASME Atlanta/
Georgia.
294. G. Andrees, Entwicklung eines keram. Werkstoffes zur Auskleidung thermisch hochbean-
spruchter Brennraume und Heipgasfiihrungen, Final Report, Project BMFT 03M-2028,
MTU Motoren- und Turbinen-Union Miinchen GmbH, p. 114.
295. L. S. Sigl and K. A. Schwetz, Fabrication and properties of HIP-treated sintered S i c for
combustor liners of stationary gas turbines, in Symposium 3, Werkstoffefiir die Energietechnik,
Werkstoffwoche '98/Miinchen, A. Kranzmann and V. Gramberg (eds.), Wiley-VCH, Wein-
heim, NY,1999, pp. 15-24.

M. Herrmann, H. Klemm, Chr. Schubert
6.1 Introduction
Silicon nitride ceramic inaterials have been intensively studied for many years
because of their great potential for use in structural applications at room and
high temperatures. This is due to their excellent mechanical properties in combina-
tion with good corrosion and thermal shock resistance.
Over the past decade a continuous increase in the number of application
fields has been observed. Besides parts for the automotive industry (cam rollers,
valve plates for coninion rail systems). which have not been produced in the
volume that was predicted 10 years ago, applications such as cutting tools. ball
bearings, wear parts. applications at high temperatures, and application in the
electronic industry are becoming increasingly important [1-4]. A new cooking
system based on a silicon nitride plate is penetrating into the market at the
moment [51.
This wide range of applications leads to an increase in the variety of materials
with different microstructures (see Fig. I ), making precise control of the microstruc-
ture and propertics of the materials necessary (Table 1).
The materials differ in the production technology used to make them, their phase
content and properties. The different production technologies used are summarised
in Table 2. A inore detailed overview is given in [ l . 61.
The reaction-bonded materials (RBSN) produced by the nitridation of compacts
of silicon powder (see Table 2) are not dense. whereas the other kinds of materials in
the table are dense.
The range of applications for RBSN and sintered reaction-bonded silicon nitride
has shrunk over the last few years. This is connected with the decreasing cost of
silicon nitride powders. reducing the production costs of SSN and the lower strength
and lifetime of RBSN in comparison to SSN in inany applications. For this reason,
these materials will not be explained in detail in this work; reference is made to
special reviews [ 6 , 7 ] . The following paragraphs concentrate on dense silicon nitride
materials made from silicon nitride powders.
The most common sintering technology used for silicon nitride materials is gas-
pressure sintering. This technology lcads to improved reliability and strength in
comparison to norinal gas-pressure sintering and is accompanied by only a moder-
ate iiicrease in cost. The HIP processes are only used for special applications (e.g.
balls for ball bearings) due to the high cost of these technologies. Hot pressing is
used for evaluation of materials or for some special applications requiring simple
geometries and only ;I very limited number of parts.
Besides the different production technologies, the materials differ in their compo-
sition and microstructure (see Fig. I ) . The wide variation of the microstructure and
Figure 1. Microstructures of different gas pressure sintered dense Si,N4 materials.

750
6.1 Introduction 75 1
Table 1. Overview of the microstructure-property relationships for dense Si,N, materials.
Property Microstructure
grain sizejshape grain-boundary phase
high strength up fine-grained/needle-like grains median content

to 1000°C
high strength at fine-grainedjneedle-like grains AI2O3-freewith special compositions
T > 1200°C
high fracture large, needle-like grains, low AI2O3and SiO, content
toughness or large, needle-like grains in a fine
matrix
high hardness fine-grained or low additive content
d-SiAION
high fatigue fine-grained/needle-like grains
strength (cyclic
mechanical load)
high large grains no components which can be
heat conductivity incorporated into the silicon nitride
crystal lattice, nitrides (e.g. Al, Be),
low impurity content
high creep large, needle-like grains, composiles A1203-freewith special compositions
resistance with S i c or refractory silicides or
no sintering additives
high oxidation large, needle-like grains, composiles AI2O3-freewith special compositions
resistance at with S i c or refractory silicides or
T > 1200°C no sintering additives
high corrosion special compositions depending on
resistance the corrosive media
good wear fine-grained microstructure homogeneous distribution

behaviour
phase content, which is possible in all of the above-mentioned production

technologies, is the reason for the wide variation of material properties produced
by each of these production technologies.
Dense silicon nitride materials are not monophase materials because, for
densification, additives, which remain as an amorphous or partially crystallised
grain-boundary phase in the material, are necessary (see Section 6.3).
This grain-boundary phase has both positive and negative consequences for the
material properties (see Section 6.5). The softening of the grain boundary has a
negative influence on the high-temperature properties, but on the other hand the
grain-boundary phase is a main factor causing the high strength and fracture
toughness of Si3N4. The liquid-phase-sintering process and the wide range of
possible additives make the tailoring of the microstructure and the adaptation of
the microstructure to a given application possible.
The aim of this paper is to show how the different microstructures and composi-
tions influence the properties at room and elevated temperatures.
Table 2. Different production technologies and resulting materials.
Material type Reaction Sintered RBSN Sintered silicon Gas pressure Hot-pressed Sinter HIP HIP-ed silicon
bonded silicon (SRBSN) nitride sintered silicon silicon nitride silicon nitride nitride
nitride (RBSN) (SSN) nitride (HPSN) (Sinter-HIPSN) HIP-SN
(GPSN)
Starting metallic Si Si + additives Si3N4,+ Si3N4 + Si3N4 + Si3N4, + Si3N4 +

components additives additives additives additives additives
Heat treatment 1250-1450°C nitridation as for 170&I80O0C 1750-2000°C 150&I80O0C 1750-2000°C 1750-2000°C gas
up to 100 h RBSN, Sintering nitrogen pressure uniaxial pressure gas pressure pressure up to
as for GPSN up to 10 MPa in a graphite die up to 200 MPa 200 MPd
Relative density 7&88% 95-99% 95-99% 98-100% 100% 100 % >99%
Main advantages no shrinkage low shrinkage, low sintering different shapes good different shapes Materials
low cost raw cost, possible, densification, possible, without additives
material different shapes better reliability high reliability, sintering of can be produced
possible than SSN; lower materials with
additive content low additive
can be used content (low
sinterability),
high reliability
Main low strength expensive Lower strenght only simple high sintering high costs for
disadvantages (1 50-350 MPa) nitridation than GPSN, shapes, costs encapsulation
good low productivity
sinterabillity
necessary
Applications refractories the same as wear parts cutting tools, mainly used for balls for ball used for
SSN/GPSN wear parts, ball evaluation of bearings, wear materials with
bearings, seals, materials, parts very low additive
valves, turbo prototypes with content or
charger rotor simple geometries without additives
6.2 Crystal Structure and Properties qf the SijN, Modifications 753
6.2 Crystal Structure and Properties of the Si3N4

Modifications
Silicon nitride has three modifications. The a and p modifications can be produced
under normal nitrogen pressure. Recently a high-pressure modification was
produced under 15 GPa pressure by the technique of laser heating in a diamond
cell [8] (see Table 3 ) . The a-phase is metastable under sintering conditions (e.g. at
1400-2000°C and 0.1-100 MPaN2), but is the main phase in the starting silicon
nitride powders. The a and p modifications are based on SiN4 tetrahedra connected
at the corners. Every N belongs to 3 tetrahedra, i.e. the structure is made up of a
three-dimensional network of tetrahedra. Only one layer of SiN4 tetrahedra
exists in p-Si3N4 (see Fig. 2), whereas two layers shifted with respect to each
other exist in a-Si3N4.This leads to a doubling of the c-axes in the a-Si3N4crystal
lattice in comparison to the p-Si3N4 lattice.
The high-pressure phase has a completely different structure. On the basis of
powder diffraction patterns, it is found that this phase has a spinel-type structure,
in which one silicon atom is coordinated by four nitrogen and two silicon atoms
by six nitrogen atoms (octahedra; see Fig. 2d).
The Si and N atoms in p-Si3N4can be replaced by Al and 0 atoms to form the so-
called p’-SiAlONs with the formula: Si6-ZAl,N8-,0,. The range of the p’ solid
solution extends from z = 0 to 4.2 at 1750°C [I, 31, i.e. it includes pure Si3N4 (see
Fig. 3 ) . That is why a clear distinction between SSN and P-SiAlON materials
cannot be made. Every Si3N4 material containing A1203 as a sintering additive
(i.e. nearly every commercial material) is some kind of p-SiAlON. Most commonly
the terminology P-SiAlON is used for z values >0.5. A similar solid solution can be
Table 3. Crystal structures of the modification of the Si,N, modification and the resulting materials
names.
Property Modifications
P-Si3N4 cc-Si,N, high pressure

modification [8]
Space group P63:m No. 176 [9, 101 P31c No. 159 [9] Fd3, or Fd3 m
Lattice parameter
a, nm 0.7586 0.766 0.7738
c, nm 0.2902 0.5615
Hardness (1 00) plane: 2 100 (1 10) plane: 2250 20-40 GPa
(001) plane: 1326 (001) plane: 2200
HV 0.025 [12] HV 0.300 [ 1 11
Coefficient of 3.39 3.64
thermal expansion 1131 [I31
0-1000”C, 10-6/K
materials based on nearly all SSN-, cc’-SiAION ?
the modification GPSN-materials, cc’/P’-SiAION
P'-Si AlON
754 6 Silicon Nitride Based Hard Materials
SI (OCt)
SI (tetr).
(4
c7 b
a
Figure 2. Crystal structures of the Si3N4modifications
formed with the addition of BeO/BeN [13]; however, materials made from these
solid solutions have no practical use due to the toxicity of Be.
Analysis of the a-Si3N4crystal structure shows that there is an empty position
+
with a coordination number of eight (7 1). This position can be partially occupied
6.3 DensiJicntion 755
2iR207)
A
/
/
\‘ ‘, 4/3(AIN A1,03)
\
/ ‘
i _- >2(A1203) ,,
/ ’ -
3(Si02)e
\
3/2(Sl*N20) ‘\
‘4
i
I I
S-SlAlON Si3N4 I
SALON-
Ct-SiAION Polytypolds
Figure 3. Stability range of a’- and 0-SiAIONs in the system Y -Si-AI-O-N (see also Section 6 4.2
[ I , 14,151)
by ions with an atomic radius of about 0.1 nm. Additionally, Si and N must be
replaced by A1 and 0 to obtain electroneutrality. The resulting phase is the so-
called a-SiAION with the formula: R,Si12-n-mAlm+nN16-nOn. Occupation of the
+
(7 1)-coordinated position leads to stabilisation of the metastable a-Si3N4
phase. The lowest x value for trivalent cations is 0.33, for Ca 0.3 [14-161. This
means that the a’-SiAlON solid solution does not include the composition of pure
a- Si3N4.
The different kinds of the materials based on the different modifications are given
in Table 3 .
The hardness values of the crystals are given in Table 3. These data show that the
p modifications have much lower hardnesses than the a modifications. N o data for
the chemical stability of the high-pressure modification have been published up to
now. First measurements of the microhardness show that this structure must have
a high hardness which is similar to that of diamond and c-BN.
The microhardness was determined for the different crystal planes of B-Si3N4.The
data differ widely in different orientations.
6.3 Densification
The high energy of the covalent chemical bonds in the Si3N4 crystals, which is
the basis for the excellent properties, presents a disadvantage in the fabrication of
the materials. The self-diffusuion coefficient is very low in comparison to other
ceramic materials. The mass transport necessary for sintering can only be achieved
at temperatures where the decomposition of silicon nitride occurs to a great extent
even at high pressure (i.e. >2O0O0C). Dense materials without sintering additives can
only be produced by encapsulation-HIP processes. The green bodies which are thus
formed are sealed in a dense glass container and densified at 1800-1900°C at 2000
bar [l]. The resulting material consists of Si3N4,with some Si02 at the grain bound-
ary. These materials have good oxidation and creep resistance (see Section 6.5.2) but
low fracture toughness and strength (see Section 6.5.1).
Usually the densification of Si3N4 is achieved through the acceleration of mass
transport by the formation of a liquid phase during sintering. In the liquid-phase-
sintering process, silicon nitride can be sintered in a temperature range of 1550-
1900°C. The liquid phase used for sintering is an oxynitride liquid formed by the
reaction of the sintering additives with the Si02 existing on the surface of the
Si3N4powder particles. The most commonly used additives are A1203 in combina-
tion with Y203, La203, lanthanoids or MgO. For high-temperature applications,
pure rare-earth additives are used. AlN additives are necessary for the production
of a’-or P’-SiAlON.
Under sintering conditions stable oxides or nitrides, which form a liquid phase
with Si3N4 and Si02 during sintering, can be used as sintering additives. For
example, alkalis and alkaline earths beside MgO can be used as sintering additives.
The problem in using these elements is their high vapour pressure under sintering
conditions.
Besides Y203,Sc203and La203,the oxides of the d-elements, Zr02 or HfO,, can
be used as sintering aids. Under sintering conditions, Ti02 reduces to TIN; hence
Ti02 can only accelerate the sintering during the initial stage of sintering. The d-
elements in the 51hto gth group form silicides under sintering conditions, forming
separate inclusions.
The amount of liquid phase formed and the resulting grain-boundary phases can
be predicted by the phase diagrams (see Fig. 3). Detailed information about the
phase diagrams are given in [l, 3,13,17].
Densification is improved when the liquid is formed at low temperatures and has
a low viscosity. For this reason, the densification behaviour worsens in the order
MgO/A1203< MgO/R203;MgO; R203/A1203 << R 2 0 3(R = Y, Sc, La and lantha-
noids).
The mechanisms of liquid-phase sintering of ceramic materials are explained in
detail in [19]. Figure 4 shows typical shrinkage curves of Si3N4 materials with
different amounts of liquid phase of the same composition. Densification starts at
1200-1300°C. At these temperatures, the chemical reaction involving the additives
starts and the formation of the liquid is initiated. The first maximum in the
densification rate is connected with the rearrangement of the Si3N4 particles. At
1500-18OO0C, densification by the solution/diffusion/precipitation mechanism
takes places. The phase transition of a-Si3N4 to P-Si3N4 occurs simultaneously.
In [18] it was shown that the a/P-Si3N4 ratio has practically no influence on the
densification when the grain sizes of the a and P phases are the same. This is an
indication that the solubilities of the a- and P-Si3N4 modifications in the liquid
do not differ significantly [20].
6.3 DensiJication 151
dLILo, 9
21.0
18.0
15.0
12.0
9.0
6.0
3.0
0.0
1200 1300 1400 1500 1600 1700 6 12 18
Temperature, "C Time, min
dLIdt, %bin
1200 1300 1400 1500 1600 1700 6 12 18

Temperature, "C Time, min
Figure 4. Typical densification curves for Si3N4 materials (1-3.5 vol.% additives, 2-6 vol.%
additives; 3-8.5 vol.% additives; AI2O3/YzO3weight ratio 1 : 2).
The sintering behaviour is improved with increasing additive amount (see Fig. 4).
However, for additive contents higher than 15-20 vol.%, formation of gas bubbles
hinders densification. The bubbling is connected with the low solubility rate of Si3N4
and the increased formation of gaseous SiO [21,22 1.
After sintering, the additives are situated at the grain boundary (3-20 vol.%). The
majority of the grain-boundary phase is concentrated in the triple junctions. The
thickness of the grain boundary films between two grains is in the range of 0.8-
1.5 nm [41].
The grain-boundary phase significantly influences the material properties. There-
fore, one usually attempts to reduce the amount of additives. However, a lower addi-
tive content also leads to a lower sinterability, a lower reliability and a broader
distribution of the properties. Thus for materials for room-temperature applications
/\
LT
'f ...........................
...............
1 P
I 'dp
dt
- t
Figure 5. Time pressure sintering regime for gas-pressure sintering
(e. g. wear parts), the additive content usually lies between 6 and 15 vol.%, representing
a compromise between reliability and properties. For high-temperature applications,
special refractory grain boundary compositions are used (see Section 6.5.2).
The most common method used for sintering silicon nitride materials is gas-
pressure sintering. The principle of gas-pressure sintering is the same as that of
the sinter-HIP process (see Fig. 5). The material is sintered to closed porosity
( > 95% theoretical density) at a low gas pressure. Then an outer gas pressure of
up to 10 MPa (GPS: gas-pressure sintering) or 100-200 MPa (sinter-HIP) is applied.
The difference between the outer pressure and the pressure in the closed pores is an
additional driving force for densification, leading to a better healing of defects. The
increase in cost for the GPS process in comparison to sintering under 1 atm (SSN) is
relatively low. In many cases these costs can be offset by the better reliability and
properties of the materials.
6.4 Microstructural Development

6.4.1 Microstructural development of P-Si3N4materials
The adaptation of silicon nitride materials to various applications leads to an
increase in the variety of microstructures (see Fig. 1). This was the impetus for
the large number of investigations into the mechanisms of microstructural forma-

tion and controlling of the same in recent years.
The most important commercial silicon nitride powders have u-Si3N4, i.e. the
phase which is metastable at high temperatures, as the main constituent [ 11. The con-
sequence of this is that during densification, a phase transformation takes place,
having a great influence on the subsequent formation of the microstructure.
It was shown that high fracture toughness and strength can be achieved when the
grains have a needle-like shape. Characterisation of the shape is usually done using
two parameters: the thickness of the grains and the aspect ratio, i.e. the ratio of
needle length to thickness [23-261.
For this reason the influence of different parameters such as additive content,
grain size of the powder, p- and oxygen content on the sintering and microstructural
formation was investigated [2439]. In basic works from the 1980s [26-291, it was
formulated that a needle-like microstructure, which is necessary for high fracture
toughness, can only be achieved using starting powders which contain a high
amount of ct-SijN4. The reason for this was suggested to be the needle-like
grain growth of a limited number of p nuclei [28] or the homogeneous nucleation
of P-Si3N4 [26,27]. Later it was shown that needle-like microstructures can also
be produced from fine p-Si3N4 powders [30,32-341. Despite intensive research, no
direct evidence of homogeneous nucleation in the oxynitride liquid or heterogeneous
nucleation of P-Si3N4 on a-Si3N4could be found [30,36-381. The only process of
epitaxial growth of one modification on the other was found in SiAlON systems
with a high degree of substitution [40]. The lattice constants differ less in these
materials than in pure a/P-Si3N4 due to the incorporation of other ions.
The reason for the absence of formation of P-Si3N4nuclei during sintering seems
to be connected with the minimal differences in the free energies of formation of p-
and a-Si3N4, leading to a quite high critical size of nuclei which can grow (60-
600nm [38]). Thus the critical nucleation size is the same or even larger than the
grain size of the P-Si3N4 crystallites in the powder. In recent works, it could be
shown experimentally that the addition of small P-Si3N4 nuclei to a commercial
u-Si3N4powder does not result in a change in the microstructure, whereas the addi-
tion of P-Si3N4nuclei of the same size as the nuclei existing in the starting a-Si3N4
powder alters the microstructure significantly (see Fig. 6). The result is a fine struc-
ture which is very similar to the structure made from the P-Si3N4powder. In [20], it
was shown that at low sintering times and temperatures, the microstructure corre-
lates with the crystallite size in the starting powder. This provides indirect evidence
of the absence of nucleation during sintering. The measurement of the p crystallite
size in materials sintered at different temperatures shows a continuous increase in
the p crystallite size, indicating that no nucleation takes place (see Fig. 7).
A detailed analysis of different microstructures [29-3 1,331 shows that they can be
explained by the number of [3 nuclei in the starting powder, even at very low P-Si3N4
contents. In Fig. 8 the calculated volume fractions of growing nuclei which are
necessary in the starting composition to produce two different fine microstructures
in the sintered materials are given. This calculated content is the lowest (j-Si3N4con-
tent in the starting powder required to explain the microstructure. For crystallite
sizes between 50 and 100 nm [20,35], the necessary amount of the p fraction in
Grain thickness in the material, pm

Figure 6. Change in the grain size distribution of materials with different p-Si3N4 amounts and
crystallite sizes (material 1 produced from cc-Si3N4 powder with 4% S-Si3N4;crystallite size p-
Si3N4 46nm; material 2 produced from a mixture of 60% cc-Si3N4 powder and 40% p-Si3N4
powder (the p-powder contain 33% p-Si3N4; crystallite size 18nm); material 3 made from 60%
a-Si3N4powder 40% p-Si3N4 powder (the p-powder contain 80% p-Si3N4;crystallite size 55 nm);
material 4 produced from p-Si3N4 powder (crystallite size 18 nm) (according to [62]; corrected
device function of the XRD-equipment).
the starting powder must be in the range >0.5-1%. This is the case for commercially
available a-Si3N4 powders. In [39], it was found that at normal heating rates
the nitrogen content in the oxynitride liquid during heating is lower than at low
heating rates and low additive contents. This means that for normal heating rates,
the liquid phase is undersaturated with respect to (j-Si3N4and no nucleation can
take place.
These experimental data suggest that (j-Si3N4nucleation has no influence on the
microstructure of silicon nitride materials under common sintering conditions and
using common powders.
The consequence of this is that the different microstructures are a result of the
different amounts and sizes of pre-existing p nuclei as well as the growing conditions.
The amount of growing (j-Si3N4nuclei depends on the (j content in the starting
powder, the crystallite size and the sintering conditions. In Fig. 9 the dependence
of the equilibrium concentration in the oxynitride liquid during sintering on the
:I
6.4 Microstructurul Development 76 1
0
, ;J:,
vl
53 /
c /+
10
__.--
0 7
1200 1300 1400 1500 1600 1700
Temperature, "C
100 j
3
E
C
'1
20 i
1200 1300 1400 1500 1600 1700
Temperature, "C
Figure 7. Change in crystallite size and phase content as a function of sintering temperature,
amount of sintering additives and starting powder 1. diimid powder 5% Y2O,/Al2O3; 2. gas
phase synthesised powder 5% Y2O3/Al2O3;3. gas phase synthesised powder 11.6% Y2O3/AI2O3.
crystallite size of a- and P-Si3N4is shown. Nuclei with a crystallite size lower than a
critical value have a higher equilibrium concentration than the Si3N4concentration
in the liquid. These crystals dissolve in the liquid and cannot growth. During the a/P
transformation under isothermal conditions, the critical size of P-Si3N4 is deter-
mined by the crystallite size of a-Si3N4(r(crtl), see Fig. 9). During heating, espe-
cially for high heating rates, the oxynitride liquid is undersaturated (see Fig. 9,
lines 2 and 3).
As a result of the undersaturation, the critical crystallite size becomes infinitely
large and only the largest nuclei survive because they require more time to dissolve
than smaller nuclei. The resulting microstructure of materials heated at high heating
rates in the temperature range of 1200-1600°C is coarser than for materials heated
at low heating rates [20,29,39].
After the alp transformation, the critical crystallite size is determined by the
crystallite size distribution. The critical crystallite size increases with increasing
sintering time and temperature, i.e. only the largest crystallites which were present
in the starting powder determine the microstructure.
Crystallite size, prn

Figure 8. Calculated volume content of growing nuclei as a function of the size of the nuclei, if the
resulting microstructure has a mean grain thickness of 0.2 pm (I) and 0.3 (2) and an aspect ratio of 5.
The needle-like grain shape seems to be kinetically determined and therefore not
the equilibrium crystal shape [30,38]. The estimated equilibrium aspect ratio (ratio
of grain length to thickness) is 1.3, not as high as 20, as was found to be the case for
the silicon nitride materials. The microstructural formation must therefore be
explained on the basis of growth mechanisms.
The explanations for the microstructure given in the ’80s were based on the
assumption that needle-like grain growth can take place only during the ct/p trans-
formation. Later it could be shown that anisotropic grain growth could also take
place after the a l p conversion [31,32,42] (see Fig. 10).
low heating rate (2)

----------- high heating rate (3)
r( crit3) = = -
in the liquid
0.6
0 0,025 0.05 0,075 0.1 0,125 0,15 0,175 0,2
Crystallite size, pm
Figure 9. Dependence of the equilibrium concentration in the oxynitride liquid during sintering on
the crystallite size of a- and P-Si3N4,according to [63].
6.4 Microstvuctuvul Developnzent 763
Figure 10. Microstructure of a material hot pressed at 1800°C (starting powder: fine P-Si3N4)and
the same material after additionally heat treatment at 1900°C. It is observed that anisotropic grain
growth takes place after a/P conversion.
Anisotropic grain growth occurs due to the different growth mechanisms in the
different directions of the B-Si3N4 crystals. The (001) basal planes of the needle-
like (j-Si3N4 crystals are atomically rough; therefore, their growth is diffusion-
controlled [4244]. Transport occurs by two mechanisms: diffusion through the oxy-
nitride liquid and surface diffusion in the (100) planes [44]. The (100) prismatic
planes are atomically flat and therefore the growth is interface-controlled. The
growth rate of the (001) plane has been assigned values of 30 [38] and greater
than 27 [35]; thus, it is at least ten [42] times higher than the growth rate of the
(100) plane. The first value seems to be the more exact one because it was determined
by growth observation on crystals in oxynitride glasses. The different growth
mechanisms are supported by TEM observations [43,44], theoretical calculations
[38] and by the observed crystal habits [4547]. The relationship between growth
rate and oversaturation is shown schematically in Fig. 11. For a high degree of
oversaturation, the (1 00) planes become atomically rougher, causing the growth
rates to approach the rate of the other planes, the growth to be isotropic and the
grain morphology to be equiaxed. At the lower saturations found in common sinter-
ing conditions, extended anisotropic grain growth occurs. The boundary between
the two regimes shifts to lower oversaturations with increasing temperature.
For a given crystal, the oversaturation depends on the size and is different for the
different crystal planes. Using the differences in surface energy calculated by the
broken bond method [38], the oversaturation as a function of grain thickness is
given in Fig. 12. The equilibrium concentration above the basal planes depends
only on the thickness of the grains, whereas the equilibrium concentration above
the (100) planes depends on both the thickness and the length of the grains.
Using these data and the dependencies of the growth rate on the oversaturation,
it can be shown that the largest crystals with the highest aspect ratios grow
anisotropically, whereas the small crystals with a low aspect ratio dissolve. Thin
crystals with a high aspect ratio grow stably whereas crystals of the same thickness
but a low aspect ratio dissolve. This behaviour was also determined experimentally
[30,48].
At first these results and the growth mechanisms discussed seem to be in contra-
diction to the results of the works from the 1980s showing a reduction in the aspect
region of anisotropic
grain growth - * .-.-
grain growth //
//
/
.-.-
region of isotropic,/'
,
/
/
A
0
c ,
e
$e I !
i
1
i
/interface controlled
Oversaturation
o(1OOyD; o(100YD + o(oo1)n
Figure 11. Growth rate of the basal plane (diffusion-controlled) and the prismatic planes (interface-
controlled) as a function of oversaturation in the oxinitride liquid. The oversaturation after the alp-
transformation depends on the surface energy (o), the grain thickness (2D) and the length of the
needle like grains (2L).
1.1:
(C(100) A-I
1.1
Y
0
1.05
Grain thickness, pm
Figure 12. Calculated dependence of the oversaturation of the basal (001) plane and the prismatic
(100) planes on the thickness of the grains, according to [30] using data [38,63] (A = aspect ratio).
ratio with increasing p-Si3N4content and at longer soaking times. These differences
can be explained by taking the crystallite size of the P-Si3N4in the starting powder
into consideration. The powders used for these investigations were produced by
direct nitridation of silicon. The higher p content of these powders was a result of
the faster nitridation and larger crystallite size. The larger crystallite size has two
consequences: 1. the oversaturation, which is inversely proportional to the grain
size, decreases and consequently the grain growth rate also decreases; and 2. the
thicker crystallites must undergo more growth to become needle-like. Both pro-
cesses lead to a reduced aspect ratio. Additionally, the powders had a wide grain
size distribution. With increasing sintering time, the critical crystallite size (see
Fig. 9) increases and only the largest crystals survive. However, the largest
crystallites had lower aspect ratios. The aspect ratio is the result of two competing
processes: anisotropic grain growth (preferred growth in the direction of the needle),
and the dissolution of small elongated grains. When the second process is the
dominating process, the overall aspect ratio decreases.
An additional effect influencing the different microstructures which are formed is
connected with the different grain sizes of the materials. The oversaturation depends
on the mean grain size. With increasing grain size, the curve characterising the
dependence of the oversaturation on grain size is flatter (see Fig. 9); for a given
mean grain size, the oversaturation cannot compensate for the difference in surface
energy between the (001) and (100) planes. As a result, anisotropic grain growth
cannot take place and the grain shape approaches the more equiaxed equilibrium
grain shape. Based on the investigations of microstructural formation, it can be
concluded that these processes are only significant for mean grain thicknesses of
0.7-1.5 pm or larger. The materials investigated in the 1970s and early 1980s were
characterised by mean grain thicknesses of 0.7-2 pm, whereas today’s materials
are based on finer powders having normally finer grain sizes.
Recent investigations [49-521 show that needle-like microstructures can be
achieved from p-Si3N4 powders when the crystallite size of the starting powder is
small enough or the sintering temperature and soaking time are large enough.
Materials with high fracture toughnesses of up to 10-12 MPam1/2[48] are observed.
Low-cost powders with higher P-Si3N4 contents, produced by special nitridation
processes which allow the exothermic nitridation to be controlled [53], have
relatively small crystallite sizes. Materials with strengths higher than 800 MPa and
fracture toughnesses similar to materials based on a-Si3N4powders can be produced
from these powders [49].
At high sintering temperatures, abnormal grain growth can take place [35,54,55].
The reason for this is the presence of large p-Si3N4 nuclei (seeds) which can exist
in the starting powder, formed during densification [35,56,62] or added as seeds
[57-591. The formation of large seeds during sintering can be intensified by low
heating rates or dwell times at temperatures higher than 1500°C but lower than
the temperature at which rapid densification occurs [56]. The amount of additional
seeds needed to produce a reinforced material ranges from 1-5 or 10 vol.%. The
seeds are usually between 10 and 100 times larger than the nuclei in the starting
powder [56-58, 60, 611. The addition of larger seeds is intended to increase the
fracture toughness without drastic decreasing the strength (see Section 6.5.1).
166 6 Silicon Nitride Based Hard Materidv
The mechanism of accelerated grain growth can be explained by adapting the

ideas developed for metals and hard materials [4546]. According to [45], abnormal
grain growth can take place under sintering conditions when the grain growth is
interface-controlled (two-dimensional nuclei on an atomically flat suface) and the
degree of supersaturation is high, sufficient to realise accelerated growth. The
region of necessary oversaturation is given in Fig. 11. This grain growth is aniso-
tropic and leads to the formation of large P-Si3N4 needles. With increasing
temperature, the probability of formation of two-dimensional nuclei on the (100)
surface (rate determine step of the growth) at a given oversaturation increases. As
a result, the accelerated grain growth is more pronounced at higher sintering
temperatures than at lower temperatures.
The oversaturation above a large nucleus depends on the mean grain size of the
matrix (see Fig. 12); therefore, abnormal grain growth occurs to a greater extent in
materials with a fine matrix.
When the density of large nuclei is too high, the mean oversaturation is reduced
and no abnormal grain growth takes place. Thus the addition of more than 1&15
vol.% of large p-Si3N4 seeds to a starting a-Si3N4 powder does not result in a
bimodal microstructure but in a coarse microstructure consisting of large grains
with low aspect ratios.
If the seeds added have the same size as the crystallites in the starting powder, the
resulting microstructure consists of fine P-Si3N4grains. If the size is smaller than in
the starting powder, there is no change in the microstructure.
The microstructural changes resulting from the addition of P-Si3N4 seeds are
shown schematically in Fig. 13. The microstructural formation depends on the
mean
crystallite size
in the
starting
powder (r,,)
1 10 50 100
amount of added seeds, Vol %

Figure 13. Microstructural changes resulting from the addition of P-Si3N4seeds (schematically).
density of p nuclei. The density is proportional to the ratio of amount of p-Si3N4 to

the mean crystallite size. Fig. 13 is based on the crystallite size of the common a-
Si3N4 powders, which lies in the range of 50 to l00nm. A change from one type
of microstructure to an other can be achieved by changing the sintering conditions.
To example a fine grained homogeneous microstructure can be changed in a bimo-
dal microstructure by high temperature sintering (Fig. 10).
The scheme shown in Fig. 13 can be used for the design of the microstructure
in p-Si3N4. The use of p powders with small grain size distributions allows the
microstructure of the P-Si3N4 materials to be controlled more precisely. Presently
such fine P-Si3N4 powders are only available in limited quantities and are
not widely used [63]. By hot pressing or using special shaping technologies
(tape casting), an anisotropic orientation of the needle-like grains can be
achieved. This results in anisotropy of the fracture toughness, strength and
thermal conductivity.
The microstructure, i.e. grain thickness and aspect ratio, depends on the type and
amount of sintering additives used. The influence of additive type on the morphol-
ogy was determined from a large number of investigations; only some important
tendencies can be given as a result of these investigations. With increasing A1203
content and constant overall additive content, the Si3N4 grains become finer and
more equiaxed. [24,26-27,641. At the same time, partial incorporation of A1 in
the Si3N4 grains occurs. The incorporation of A1 in the grains can be expressed
by the reaction:
0 . 5 ~A1203+ (2 - 0.25z)Si3N4@ Si6p_Al,N8-,0, + 0.252 Si02

This process leads to a reduction of the A1203 content in the grain-boundary
phase and to an increase in the Si02 content; i.e. AI2O3 functions as a buffer of
the grain-boundary phase composition. The more Si02 evaporates during sintering,
the more A1 dissolves in the silicon nitride grains and reproduces Si02 [64]. The
incorporation takes place according to the local equilibrium in the liquid. The
diffusion in the silicon nitride grains is slow. Therefore A1 gradients in the silicon
nitride grains can be observed [64,67].
The incorporation of A1 in the silicon nitride crystal lattice leads to a reduction of
the thermal conductivity of the materials [65,66]. This means that the A1203/A1N
content in the sintering additives can be used for the tailoring of the thermal conduc-
tivity; values between 10 and 150 W/mK are possible.
Investigations of the growth of the p-Si3N4 grains in oxynitride liquids with
different rare-earth additives show that with decreasing melt viscosity (Yb to La),
the aspect ratio of the grains increases and the thickness of the grains decreases.
This was found in materials with rare earths and alumina as sintering additives
and in materials with only rare-earth additives [68-691. Very fine, needle-like micro-
structures were obtained for Sc203-containingadditives. [70]. CaO and MgO were
found to accelerate the densification and the anisotropic grain growth [71,72], but
were found to be detrimental to the high-temperature properties. The use of rare-
earth additives or mixtures with MgO results in the formation of microstructures
consisting of grains with high aspect ratios [ 13.
6.4.2 Microstructural development of a’-SALON materials

a’-SiALONs are formed during liquid-phase sintering by the reaction of Si3N4,
AlN, A1203 and an appropriate cation which can enter the structure of a’-
SiALON (R,Si12-n-mAlmtnN16-nOn). The amount of liquid available for densi-
fication is quickly reduced due to the formation of a-SiAlON solid solutions
(Fig. 14). This reaction starts during heating at temperatures above 1450°C
4/3 (AINxA1203)
Si3N4 m=2 m=3 RN x3A1N
4/3 (AINxA1203)
area of elongated
grain growth
R203 x 9A1N
7
Si3N4 m=1 m=2 m=3 RN x3AlN
b)
Figure 14. Stability area of the a’-SiAlON solid solution R,Si,,_,~_,A1,+,N16_nO~ a) in the systems
with R = Nd, Sm, Dy, Y and Yb at 1800-1900°C (after [14]) and b) in the area where elongated grain
growth takes place in the a’-SiAlON/P’-SiAlON-plane(after [14,15,77,79]).
[73,74]. When the a’-SiAlON formation is completed, no more liquid exists for
densification. This is the reason why pure a’-SiAlON materials with low degrees
of substitution n and m are difficult to densify to full density. Better densification
can be realised at higher n and m values (Fig. 14) [75-771 or in composite dip’-
SiALON materials [78].
The resultant materials had microstructures consisting of equiaxed grains, lead-
ing to reduced fracture toughness and strength. This was a reason for the limited
use of these materials [I]. In the past few years materials with elongated grains
have been produced by different research groups. The materials with elongated
a-SiAION-grains have relatively high n and m values (greater than 1.2) (Fig. 14)
or mixed cations (rare earth’s and Sr or Ca [14,75]). In early studies it was sug-
gested that elongated grain growth occurs only when p-Si3N4 powder is used [76].
However, later investigations have shown that the use of a-Si3N4 powder also
leads to elongated grain growth [75,14, 15,791. The reason for the anisotropic
grain growth of a’-SiAlON is not completely clear because it was not investigated
in the same manner as the p-Si3N4 grain growth. The a’-SiAlON materials with
low n and m values and equiaxed grains possess a low grain size. This is the
consequence of the fast disappearance of the liquid. In contrary, the microstruc-
tures of the a’-SiAlON materials with elongated grains show a coarser micro-
structure; i.e. more intensive grain growth takes place. This must be connected
with the amount of liquid existing during sintering, which is higher when
higher substitution levels of the d-SiAIONs are used. Recently it could be
shown that materials with elongated a’-SiAlON and low n and m values can
be produced by gas pressure sintering using an appropriate liquid phase (see
Fig. 15) [80]. This suggests, that the needle like grain shape is governed by the
grain growth kinetics.
The stability of a’-SiAlON at elevated temperatures (1 100-1 500°C) depends on
the nature of the cations R entering the a’-SiAION structure. For the trivalent
rare-earth cations (i.e. La-Yb; Y), the stability increases with decreasing atomic
radius [14,81]; i.e. the temperature below which a phase transformation a’ -+ p’
can thermodynamically take place decreases. The stability also increases with
increasing substitution level. Thermodynamic stabilisation of the a’-SiAlON
with respect to the p’-SiAION can be achieved by the use of mixed cations [14].
Besides the nature of the cation which is incorporated in the a’-SiAlON, the rate
of a’-$-transformation depends on the existence of $-SiAlON nuclei and the
amount and viscosity of the amorphous grain-boundary phase. The transformation
between a’-SiAlON and D’-SiAlON is reconstructive and must occur via a liquid in
which one phase can dissolve and the product precipitate. Therefore the transforma-
tion can be retarded by the absence of a glassy phase or by a very high viscosity
[14,81].
Better understanding of the microstructural formation of the a’-SiAlON
materials in the future will be necessary for the improvement of the mechanical
properties, especially fracture toughness and strength, and the wider application
of this group of materials, having higher hardnesses than the B-Si3N4 and p-
SiAlON materials (see Fig. 16).
Figure 15. Polished and etched micrograph of an a’-SiAION material with n,m < 1; a) plasma-
chemically etched material (dark area: P-Si3N4;grey area: d-SiAION; bright area: grain-boundary
phase); b) chemically etched material.

770
6.5 Properries of Si, N4 Materials 771
110 HV10
- 1000
cl
iE
3
Ei
-900 $
,
''
,
' 5
-3
E
-800
t20 to
c
3 .
2
5
-700
--2"
I
m
;a
.-
II
- 600
- 500
T - 400
B-Si3N4 - Conventional High toughness
materials a '-SiAlON a '-SiAION
materials materials
Figure 16. Comparison of the properties of the conventional a'-SiAION materials with P-Si3N4
materials and with high toughness a'-SiAION materials with elongated grains [80].
6.5 Properties of Si3N4Materials

6.5.1 Mechanical properties at room temperatures
The strength of commercial silicon nitride materials lies in the range of 800 to
1400 MPa. The strength-determining defects are usually pores, inclusions such as
iron silicides or agglomerates of glasses especially for low medium strengths. Besides
these defects, large, elongated grains can be strength-determining defects for higher
strengths or in specially developed, high-fracture-toughness materials. For strength
levels higher than lOOOMPa, special surface finishing of the samples is necessary
because surface defects play a decisive rule in determining the strength.
The strength of a brittle material is proportional to the fracture toughness and
indirectly proportional to the square root of the defect size. The defect size of the
materials can be reduced by optimised processing. [55,82]. The highest measured
mean three-point-bending strength for silicon nitride was 2000 MPa [85]. This
corresponds to a defect size of about 5 pm. Materials with a strength level of 1400
to 1500 MPa usually have a defect size of 10 pm [55,83], i.e. these materials have
[871
[551 seeds oriented
" .0
5 10 1
Fracture toughness, MPa rnA1/2

Figure 17. Dependence of the tree point bending strength on the fracture toughness.
to have a grain size lower than that of the defect, because larger grains can act as
strength-determining defects. This means that high-strength materials must be
fine-grained. On the other hand, the fracture toughness increases with the square
root of the grain size [36,69,87,88]. These two adverse dependencies produce the
relationship between strength and fracture toughness shown in Fig. 17. For low
fracture toughness values, flaws other than the largest grains are the strength-
limiting defects. In this region of KIC, further improvement of the strength is
possible. For high KIc values, the strength is determined by large elongated
grains. Thus the structure itself determines the strength, not processing-dependent
defects. Materials with large, needle-like grains in the matrix are usually called
insitu-reinforced materials. Attempts to improve the strength of such materials by
seeding a matrix with large, needle-like grains with a narrow size distribution are
underway [57,59,60]. Results for tape-cast silicon nitride materials with large,
oriented P-Si3N4needles represent a possible approach [60] (Fig. 17) to increase
the strenght and fracture toughness. However, such materials exhibit anisotropic
properties, e.g. low strength in the direction of orientation of the needles.
Additionally, the insitu-reinforced silicon nitride materials exhibit an increased
Weibull modulus [86]. This has two reasons: firstly, the concentration of the
large, strength-limiting grains is so high that the defect size has a narrow size distri-
bution; and secondly, these materials exhibit pronounced R-curve behaviour, which
makes the material tolerant of larger cracks. The high thermal shock resistance of
these materials also seems to be connected with this fact [30].
The subcritical crack growth rate in silicon nitride materials is relatively low in
comparison to other ceramic materials. The growth exponent is in the range of
6.5 Properties of Si3N, Materials 773
N = 30-300 for static loads and n = 20-50 for cyclic loading [90,91,97-1001. Up to
now, no clear correlation between the microstructure and the growth exponent
could be made. For cyclic loading it can be shown that materials with the lower
grain size have the higher growth exponent, i.e. have a lower degradation of strength
during loading and a higher lifetime at a given strength level [91,88].
The fracture toughness depends strongly on the microstructure. Two main factors
influence the fracture toughness: 1. the grain shape and size, and 2. the composition
of the grain-boundary phase.
The high fracture toughness of silicon nitride materials in comparison to other
ceramic materials is connected with toughening mechanisms which are similar to
those in whisker-reinforced composite materials: grain bridging and pull-out,
crack deflection and grain branching around large, elongated grains [88,89,94].
Due to these mechanisms, the fracture toughness increases with increasing
volume fraction and square root of the mean grain thickness of the elongated
grains (i.e. grains with aspect ratio >4) [37,70,88,89].
Recent investigations show that fine-grained materials have a higher toughness
than coarse-grained materials in the small crack region but lower toughness in the
large crack region [82,84,90]. This can be important for the use of such materials
under conditions in which high local stresses exist (e.g. for ball bearings). The
dominant toughening mechanism depends on the grain thickness of the elongated
grains. Elastic bridging and pull-out were observed for thin, needle-like grains
(with thickness < 1 pm). Crack deflection was mainly observed for thick, elongated
grains (with thickness > 1 pm), whereas frictional bridging was detected indepen-
dently of the grain size [89].
The toughening mechanisms described above can only operate when the domi-
nant fracture mode is intergranular. The ratio of transgranular to intergranular frac-
ture depends on the relative strengths of the grain boundary and the grains. For a
material with a high toughness, the grain boundary must be weak in comparison to
the grains (see Fig. 18).
The strength of the grain boundary is connected with two different mechanisms:
local residual stresses [92,101] and special chemical interactions between the grain-
boundary phase and the silicon nitride grains [93,94].
The amorphous or partially crystallised grain-boundary phases generally have
different thermal expansion coefficients than silicon nitride. When the thermal
expansion coefficient of the grain boundary is higher than that of the silicon nitride,
the grain-boundary phase is under tensile stress and the fraction of intergranular
fracture increases. As a consequence the fracture toughness increases (see Fig. 19).
Similarly, a material with a grain-boundary phase under compression (e.g. HIPSN
without sintering additives) shows a low fracture toughness [ 1021 due to a high per-
centage of transgranular fracture (see Fig. 18a).
The residual stresses can be influenced by the sintering cycle. This can be brought
on by a change in the composition of the grain boundary (e.g. by evaporation of
S O 2 ) during sintering, by crystallisation of the grain boundary or by partial relaxa-
tion of the stresses (e.g. through slow cooling). These changes are usually out-
weighed by the grain size and shape or the starting composition of the material,
which have a greater influence on the stress state and KIc.
Figure 18a and b. Crack path in silicon nitride materials a) HIPed material (only S O 2 sintering
additive, mainly transgranular fracture; chemical etched), b) material with Y20,/A120,additives
(mainly intergranular fracture, plasma etched).
Figure 19 shows the additional influence of the grain size (see curves 1-3). The
composition of the grains also influences the fracture toughness by determining
the special chemical interactions that occur between the grains and the grain
boundaries. The formation of P-SiAlON layers on the Si3N4 grains was found to
result in an increase in the amount of transgranular fracture and a decrease in the
fracture toughness [93, 1001. This can be a reason for the lower fracture toughness
of the P-SiAlONs with a high degree of substitution, in addition to the lower
aspect ratio (see Fig. 18d).
The recently developed a-SiAlON materials have a microstructure consisting of
relatively coarse, elongated grains (see Fig. 15) and show a high percentage of inter-
granular fracture; thus, these materials exhibit a high fracture toughness (see Fig. 16).
Summarising the fracture toughness data, it can be concluded that the fracture
toughness of the silicon nitride materials depends strongly on the microstructure.
For high fracture toughness values, elongated grains and a grain-boundary phase
which promotes intergranular fracture are necessary.
The hardness of different silicon nitride materials is given in Fig. 20. Materials
with a high a-Si3N4 content, i.e. materials with a high amount of metastable a-
Si3N4which did not transform during sintering (normally produced by hot pressing
or HIP’ing [l, 59]), show a hardness which is as high as that of the d-SiAlON
materials.
Analysis of the hardness values of different P-Si3N4 materials reveals that the
hardness increases with decreasing amount of grain-boundary phase and decreasing
6.5 Properties q f S i 3 N 4 Mureriuls 775
Figure 18c and d. Fracture surfaces of silicon nitride materials c) fracture surface of material with
Y2O7/AI2O1additives (mainly intergranular fracture), d) fracture surface of a P’-SiAION material
with a high degree of substitution z (z = 4) (mainly transgranular fracture).
grain size; i.e. the dependence of hardness on the microstructure is opposite to that
of the fracture toughness.
Usually the hardness values of b-Si3N4 materials (HVlO) are in the range of
12 GPa (coarse grain size, high additive content, residual porosity) to 16 GPa (fine
grain size, low additive content). The hardness of the mixed a/b-SiAlONs change
lineary with the phase content [ 1311.
The hardness at elevated temperatures depends additionally on the softening of
the glassy grain-boundary phase. Materials with MgO/Al2O3as sintering additives
show a faster degradation of the hardness than materials with more refractory grain-
boundary phases (Fig. 20).
Common sintering additives’

/
Glass
Si02
i
ain boundary
Silicon nitride
---___
----_--__ /
,i h e r m a l expansion coefficient, 10-6/K

,
SiO2-grain
boundary phase
Si3N4
2J
I
0 2 4 6 8
Thermal expansion coefficient, 10-6/K
Figure 19. Residual stresses of grain-boundary phase and the silicon nitride grains as a function of the
thermal expansion coefficient of the grain-boundary phase (calculated after [92]); used constants for
Si3N4:E = 320, v = 0.27, s( = 3.39 x K; and for the glass: E = 140, v = 0.29, volume fraction
of glass = 0.15; a) and dependence of the fracture toughness on the thermal expansion coefficients
(data 1-3 after [92]) 1. mean thickness of the grains 0.55; 2. 1-1.3; 3. 1.5-1.7 (data 4 after [64]).
6.5 Properties of Si,Nd Materials 777
25
20
15
10
5
500 loo0 1500
Temperature, "C
Figure 20. Hardness of different Si3N4materials; 1 . additive free material with remaining a-Si3N4
[59], 2. hot pressed with 10% Ce02 and 6.6% AIN contain 20% a-SiAION [84]; 3. sintered p-silicon
nitride material with 3% Y 2 0 3and 2 % Al2O3;4. d-SiAlON.
6.5.2 High-temperature properties of silicon nitride materials

Silicon nitride is among the most promising materials for high-temperature applica-
tions because of its combination of excellent mechanical properties at room and
elevated temperatures, oxidation resistance, low coefficient of thermal expansion
and low density in comparison to refractory metals. However, for the majority of
Si3N4materials these properties can only be obtained by a purposeful tailoring of
chemical composition and microstructure. Special emphasis needs to be placed on
the intergranular grain-boundary phase which is formed during the sintering
process.
As already mentioned, Si3N4 materials are usually densified by the addition of
sintering aids. They form a glassy silicate phase with the silica always present as
an impurity in the Si3N4raw powder; this silicate phase remains as an amorphous
or crystalline grain-boundary phase between the Si3N4grains after sintering.
While the properties of the Si3N4 grains do not change up to temperatures of
16OO0C,as a consequence of the high degree of strong covalent bonding in Si3N4,
the grain-boundary phase starts to become weak at elevated temperatures. Depend-
ing on the amount, composition and condition of the grain-boundary phase between
the silicon nitride grains, various processes (diffusion, creep, slow crack growth,
oxidation, corrosion) occur at elevated temperatures. The consequence of these
high-temperature processes is the generation of a new defect population which
determines the failure behaviour and ultimately limits the lifetime of the silicon
nitride material. The extent to which these processes occurs is mainly influenced
by the softening point and viscosity of the amorphous grain-boundary phase.
Table 4. High-temperature properties of various Si3N4 materials (01400’~ bending strength at

1400°C; i - creep rate at 1400”C, 200MPa; Am,, - weighted gain during oxidation at 1500°C;
~
o,, - residual strength after oxidation).
1 HIP-Si3N4 500 9x 0.7 (2500 h) 490 1

2 SN 5% Y203/3% A1203 450 ~ 4 (100h) < 200 < 0.2
3 SN 8% Y203/0.6%Ai203 720 6x 3.7 (1000h) 210 0.2
4 cL’/D’-Sialon (Y) 380 5 x lo-* 10 (100h) < 200 < 0.2
5 SNS%Y203 730 2 x lo-’ 2.2 (2500 h) 340 0.4
6 Si3N4-SiC (SY” Y203) 750 2 x lo-’ 3.3 (2500 h) 640 0.8
7 Si3N4-MoSi2 (So/, Y203) 730 2x 3.1 (25ooh) 630 0.8
In the following chapter fundamentals of the high-temperature behaviour of

silicon nitrides and silicon nitride-based composites will be presented. Depending
on the sintering additive used, silicon nitride materials with different high-tempera-
ture properties can be obtained. Some typical materials are chosen from the broad
variety of Si3N4ceramics that exist and summarized in Table 4.
The first Si3N4ceramic (Table 4) is a material without additional sintering aids.
The only liquid phase appearing during sintering is the silica from the high-purity
Si3N4 powder. Because of the small amount and the high softening point of the
grain-boundary phase, these materials can only be densified by encapsulation hot
isostatic pressing (HIP). The materials exhibit excellent behaviour at elevated
temperatures as the consequence of a clean grain-boundary phase consisting only
of silica [102]. The oxidation resistance is used to demonstrate the excellent high-
temperature properties of this material. Figure 21 shows the microstructure of a
HIP’ed Si3N4 without additives after 2500 h oxidation at 1500°C. The oxidation
behaviour of silicon-based nonoxide ceramics is mainly influenced by the protective
layer which is formed at the surface of the material as a result of the oxidation
Figure 21. SEM images of polished cross sections of the Si3N4material 1 (Table 1) after 2500 h of
oxidation treatment at 1500°C.
6.5 Properties of' Si3N4 Muteriuls 779
process and the ability of this layer to prevent oxygen from diffusing into the
material. In this case the formation of a surface layer of pure silica causes the rate
of oxygen diffusion into the material to be very low [103, 1041.
A more refractory grain boundary can only be obtained by removing the silica
from the grain boundaries and triple junctions. Each addition of only a small
amount of impurities or sintering additives will weaken the material by producing
a change in the chemistry of the grain-boundary phase [105-1071. The result is
degradation of the high-temperature behaviour of the materials. Besides the
costly fabrication, the main disadvantage of these material is their inadequate
mechanical properties (especially toughness) at room temperature. The microstruc-
ture, which is mainly globular, and the strong bonding of the intergranular phase do
not allow toughening mechanisms like crack deflection or bridging to be active.
With the addition of sintering aids these disadvantageous properties can be
improved; however, the improved densification, the microstructure with elongated
grains, and the improved toughness are the consequence of a changed grain-boundary
phase. Depending on the kind of sintering additive used, a silicate phase will be
formed during sintering and will remain as an amorphous or partially crystallized
phase in the grain boundaries and triple junctions between the Si3N4grains. Besides
the chemical composition and the amount, special emphasis is placed on the crystal-
linity of the intergranular silicate phase. With a high amount of crystallized silicates in
the grain boundary, the amorphous phase is minimized; this is found to be especially
beneficial to the creep and slow crack growth behaviour of the Si3N4 materials.
Glass-forming and stabilizing sintering additives which, with the silica of the Si3N4
powder, form a silicate phase with a low softening point and viscosity such as MgO,
AlzOi or AlN are principally not useful in the fabrication of high-temperature silicon
nitride materials. More refractory silicates that form an intergranular phase with a
high crystalline content were obtained by using yttria or the rare-earth oxides as sin-
tering additives. Examples for these materials are given in Table 4. When A13+ions are
present, the softening point and viscosity of the grain boundary are lowered [ l , 31,
with the consequence of degradation of the mechanical properties at elevated tem-
peratures. Figure 22 summarizes the creep curves of the alumina-containing silicon
nitride, material 3 (Table 4), in comparison with the additive-free material and the
material with yttria as a sintering aid. The HIP'ed material exhibits the highest
creep resistance. Superior creep behaviour is also found for Si3N4with Y203 as a sin-
tering additive. This is the consequence of the high degree of crystallized grain-bound-
ary phase having the apatite structure. The creep resistance of the Si3N4materials is
found to be lowered by an amount which depends on the amount of A13+ions present.
With the addition of a small amount, the crystalline grain-boundary phase does not
change significantly; however, a degradation of the creep resistance is observed,
caused by the concentration of A13+ ions in the residual amorphous grain-boundary
phase. In the material with 5% Y 2 0 3and 3% A1203,the grain-boundary phase is
found to be mainly amorphous, with a lowered softening point and viscosity. The
result is a very poor high-temperature resistance: the material failed after only a
few minutes during creep testing at 1400°C and 200 MPa.
A similar tendency is found by comparison of the oxidation behaviour of the
materials. With the addition of the sintering aid the composition of the protective
0,Ol
0,008
0,006
0,004
1/
0,002
0
0 20 40 60 80 100
Figure 22. Comparison of the creep behaviour of the materials (1 in Table 4) HIP’ed Si3N4without
additives, (3 in Table 4) Si3N4with 8% Y203/o.6%A1203and (5 in Table 4) Si3N4with 8% Y203as
sintering additives. The tests were conducted at 1400°C in air at 200 MPa.
oxidation layer formed during oxidation changes, with the consequence of a higher
oxygen diffusion into the materials. The weight gain of the Si3N4with Y2O3 is about
three times higher than that of the HIP’ed Si3N4without sintering aids, and that of
the material with the small amount A1203about one order of magnitude higher than
that of the HIP’ed Si3N4without sintering aids. Figure 23 shows the microstructure
+
of the materials 3 (8% Y2O3 0.6% A1203) and 5 (8% Y203)after oxidation at
1500°C. Oxidation processes are observed in the bulk of both materials and lead
finally to alteration of the microstructure. However, due to the addition of A1203,
the oxidation of material 3 (in Table 4) occurs faster. Already after 1000 h oxidation
the microstructure is found to be strongly degraded. The same damaging processes
occur with the Y203-containingSi3N4material; however, this material exhibits, due
to its more refractory grain-boundary phase, a higher oxidation resistance (see
Figure 23. Microstructural damage in the surface region of the Si,N4 materials with Y2O3/Al2O3
(3 in Table 4) after 1000 h (A) and Y 2 0 3(5 in Table 4) after 2500 h (B) oxidation at 1500°C.
6.5 Properties of Si3N4 Materia1.r 78 I
microstructure in Fig. 23 after 2500 h). As a result of the oxidation processes, degra-
dation of the mechanical properties is observed; this is found to be dependent on the
extent of oxidation as represented by the residual bending strength after oxidation
(see Table 4).
Interesting materials with respect to high-temperature behaviour are mixed a’/p’
SiAlON [3,78]. Due to the ability of the a’-SiAlON grains to incorporate cations
from the sintering aids into their lattice, it is possible to modify the grain boundary
of the silicon nitride material. In a material with a high amount of a’-SiAlON, a
skeleton from strong grain boundaries between the a’-SiAlON is formed due to
the incorporation of the ions of the sintering aids into these grains, with the
consequence of an improved creep resistance. The oxidation resistance cannot be
improved by the increase in the d-SiAlON content. The high amount of A13’
ions in the material is the reason for the formation of a weak surface layer of alumo-
silicates during oxidation, allowing a high rate of diffusion of oxygen into the
material 1781.
One goal in recent studies in the field of high-temperature Si3N4materials was the
improvement of the behaviour at elevated temperatures by the formation of compo-
sites. Typical examples which have been reported in the literature are Si3N4-Sic
micro- or nanocomposites or Si3N4-MoSi2 composite materials. In the case of
Si3N4-SiC nanocomposites, a structural synergism between the matrix Si3N4and
the nano-Sic was supposed to be responsible for the high level of the properties
of these composites at elevated temperatures [ 1091. Recent studies, however,
showed that the improvement of the mechanical properties at high temperatures
was the consequence of a chemical modification of the composition of the grain-
boundary phase and a significant improvement of the oxidation resistance
[104, 108,110,ll I]. Similar behaviour was found for Si3N4-MoSi2 composite
materials [112]. After long-term oxidation tests at 15OO0C, a considerably less
damaged microstructure was observed. Figure 24 shows the comparison of the
material 5 (Si3N4 with 8 % Y203) with a Si3N4-MoSi2 composite (material 7).
Apart from the 10% MoSi2 in the composite, both materials had the same composi-
tion. The microstructural development of the materials was found to be different. A
significantly lower amount of damage was observed in the Si3N4-MoSi2 composite,
although the oxidation processes also occurred in the bulk of the composite.
Figure 24. Comparison of the microstructure of (A) the Si3N4material ( 5 in Table 4) and (B) the
Si3N4-MoSi2 composite (7 in Table 4) after 5000 h oxidation at 1500°C in air.
1-1
bending stress / MPa
900
i
700
500
300
I00
time I h
Figure 25. Comparison of time to failure behaviour of silicon based nonoxiode materials at 1400°C
in bending (curve 1: material 7; curve 2: material 6; curve 3: material 5; curve 4: SBC).
This behaviour was the consequence of different processes in the surface region
of the materials, resulting in a changed oxidation mechanism in the composites.
The important feature of the Si3N4--MoSi2and Si;N,-SiC composite was the
increased formation of Si20N2 instead of SO2, producing an additional layer
between the oxidation surface and the bulk of the silicon nitride composite materials.
The significantly reduced defect size gives an idea of the potential of these
composite materials in terms of stability and time-to-failure behaviour at elevated
temperatures (Fig. 25). Principally, the amount, composition and condition of the
grain-boundary phase are the key factors which must be considered for the suc-
cessful development of Si3N4materials for applications at elevated temperatures.
Materials without sintering additives exhibit the highest stability at elevated
temperatures due to the high refractory grain-boundary phase of pure SO2. The
addition of sintering aids ultimately leads to a degradation of the properties at
elevated temperatures. However, by a purposeful design of the grain-boundary
phase and the formation of high temperature stable composites which are stable
at high temperatures, it is possible to obtain silicon nitride materials which can be
used up to temperatures of 1500°C.
6.5.3 Wear resistance of Si3N4 materials

The excellent wear properties of silicon nitride materials are the basis for many
applications of these materials, such as ceramic or hybrid rolling bearings, cutting
tools, wear parts in automotive engines (valves, cam rollers, valve plates for
common rail systems), and parts for metal shaping.
Wear is a property of the system, not only of the material. Therefore a few special
applications will be presented to show the influence of the microstructure on the
wear properties.
6.5 Properties of Si3N4 Materials 783
Figure 26. Typical cracks on the track of ball bearings (dry running under 1.75GPa Hertzian
pressure. 18 turns).
Silicon nitride is the preferred ceramic material for use in ceramic and hybrid ball
bearings due to its high strength, fracture toughness and high resistance to sub-
critical crack growth. The advantages of ceramic ball bearings in comparison to
metal ball bearings are: reduced wear, higher hardness, better temperature and
corrosion resistance, lower specific weight and higher electrical resistance. This
combination of properties makes lubrication of such bearings either almost or
completely unnecessary [2, 113, 1 15, 1171. The field of application of such bearings
is fast growing. The bearings show a good performance and a high reliability. To
improve the behaviour a development of silicon nitride materials for ball bearing
was started in the last few years [ I 13-1 16,1181. The main failure mechanism of
ceramic ball bearings are connected with local and temporal changing high tensile
and compressive stresses during operation. These local and temporally changing
stresses result in the formation of cracks perpendicular to the rolling direction of
the balls in the ball bearing (see Fig. 26) [I 141.
The lifetime and load capacity of a ball bearing can be increased by optimisation
of the geometry of the track [ 1 13-1 15,1181, by reduction of the material damage due
to cyclic Hertzian contact, and by reduction of the friction coefficient [I 14,1161.
The influence of the microstructure on the Hertzian contact damage was investi-
gated in detail [119, 1201. Coarse-grained silicon nitride materials are subjected to an
accelerated fatigue under cyclic Hertzian pressure, unlike fine-grained materials.
Therefore materials for ball bearings have to be fine-grained. Apart from that, the
materials must exhibit high strength (i.e. >800 MPa) and a high Weibull modulus
because every defect on the track can be a starting point for failure.
The influence of the friction coefficient on the load capacity is connected with the
fact that, besides rolling, sliding always exists in the ball bearing [114, 1151. This
friction of the ball on the track is the reason for an increase in the local tensile
stress in the track during operating. Therefore reduced friction will increase the
load capacity and the lifetime. The load capacity of ball bearings operating in wet
conditions is more than 2 times higher than in dry conditions. This is due to the
lubricating effect of the water [I 151. The development of Si3N4 composites with
small amounts of submicro-TiN or BN inclusions leads to an increase in the load
capacity of up to 60% under dry run (see Fig. 27) [115]. Further improvements
can be expected through surface modification by ion implantation [I 171, the use
Self lubricant SN-based composites
Friction Figures p
0.02
Idry run in air
water lubrication
0,015
0,Ol
0.005
0
Standard
Standard Si3NUBN Si3N3/TN Silicon nitride
Silicon nitride (grease lubriicated )
Figure 27. Friction figure (maximal value) of angulare contact bearings 7006 E under dry running in
air and in water at 500N preload [115].
of special thin coatings or by the development of Si3N4materials with high strength

and low grain sizes < 100 nm with improved wear resistance [63].
The friction of silicon nitride materials depends strongly on the wear conditions.
At low loads the grain boundary is mainly affected (see Fig. 28a), whereas at higher
loads silicon nitride grains can break (see Fig. 28b). The wear products can
form lubricating layers or rolls that can reduce the friction (see Fig. 28c) [121-
1241. Additions of TIN and BN can further reduce the friction and wear under
special conditions. An overview of the wear properties under sliding conditions of
different Si3N4 materials is given in [121-124,631. Up to now, the silicon nitride
materials have not been optimised for use under different wear conditions. Optimi-
sation of the grain size and composition and amount of the grain boundary can lead
to improved wear behaviour of silicon nitride materials in the future.
Silicon nitride materials are used as cutting tools for cutting cast iron and super-
alloys [ 125-1 3 11. Ceramic tools and particularly silicon nitride cutting tools exhibit
higher hardness at high temperatures than cemented carbides. That allows the use of
ceramic tools for machining at higher cutting speeds than for cemented carbides.
The typical increase in metal removal rate due to the use of Si3N4in comparison
t o cemented carbide tools is more than 200% [129].
The high temperatures and mechanical load that are generated at the cutting edge
cause wear by a variety of mechanisms:
- abrasive wear due to hard inclusions in the workpiece or chemical interaction with
the workpiece or the ambient atmosphere,
6.5 Properties of' Si,N4 Materials 785
Figure 28. Surface of silicon nitride materials after oscillating sliding tests.
- thermal deformation or degradation by thermal shock, subcritical crack growth

and, consequently, fracture of the tool. Tool failure due to fracture must be
avoided because it leads to an unpredictable behaviour of the tool. Therefore
the cutting materials must have high strength, fracture toughness and thermal
shock resistance.
The first two mechanisms lead to more or less continuous wear of the cutting tool
over its lifetime.
The temperature of the cutting tool edge can reach about 800°C for turning cast
iron [ 130, 1341; temperatures of up to 1350°C were measured for the cutting of steel
[134]. The enhanced chemical interaction and softening of the grain-boundary phase
due to the high temperatures during steel cutting are the reasons why silicon nitride
cutting tools cannot be used effectively for the cutting of steel.
Chemical interaction is also a dominant wear mechanism during turning and
milling of cast iron [ 127,1301. On the surface of the cutting tool, different silicates,
which have a low hardness and which can therefore be easily removed during
machining, can be formed. Chemical interaction seems to be the reason for the
relatively rapid rounding of the cutting edge at the beginning of the cutting process.
This produces an increase in the cutting forces and is the reason for using silicon
nitride tools only for rough machining.
The combination of the different wear mechanisms during cutting makes under-
standing of the cutting behaviour difficult. Up to now, many open questions exist
and the cutting behaviour cannot be predicted using knowledge of the mechanical
properties and microstructure. However, some overall relationships between cutting
behaviour and material properties can be given. The cutting materials must exhibit
786 6 Silicon Nitride Based Hard Mutevials
high strength, hardness and fracture toughness. These properties are necessary but
not sufficient for a good cutting behaviour: there are several Si3N4materials which
fulfil these conditions, but which have a bad cutting behaviour due to enhanced
chemical interaction with the workpiece.
A wide range of substances in different ratios are used as sintering additives for
cutting tools: CaO, MgO, A1203,Y 2 0 3 ,Zr02 and Hf02 [2,171,125-1311. The use
of sintering additives that reduce the high-temperature strength is surprising, but is
less so when one considers that the softening of the glassy grain-boundary phase
starts at temperatures above 800”C, whereas the temperature at the cutting edge
in the machining of cast iron is lower than 800°C. An increased Si02 content in
the grain-boundary phase reduces the lifetime of the cutting tools [133]. This
seems to be connected with the acceleration of the chemical interaction.
A decrease in the amount of sintering additives improves the cutting behaviour
[125]. The problem of optimising the cutting tool material is that on the one hand
it is necessary to minimise the amount of the grain-boundary phase and on the
other hand it is necessary to attain a reliable densification to achieve full density,
high strength and high reliability. One possibility to improve the densification of
the materials with a low additive content is to use MgO-containing additives
[71,129]. Another way to improve the sintering behaviour without increasing the
amount of the residual grain-boundary phase is to form p’- or P’/a’-SiAlONs
[130]. The substitution grade z in the P-SiALON must be small because high z
values cause the fracture toughness and strength to be reduced. In [ 1321 an optimum
for Si6-,Al,0,N8_, is found with z = 1-1.5, but most of the commercial cutting
tools have an lower substitution degree. p’/a’-SiAlON composite materials are
used especially for cutting superalloys. They have a slightly reduced strength and
fracture toughness [128, 1311.
A low grain size is favoured for cutting tools because the hardness increases with
decreasing grain size. Nevertheless excellent cutting performance is found for self-
reinforced Si3N4materials with large, elongated grains [7 11.
The examples given above show that it is necessary to find a compromise between
the different material properties. The compromise strongly depends on technologi-
cal possibilities (e. g. improvement of the sintering behaviour of the powder) and on
changes in the metal machining. The chemical stability of the cutting tools can be
increased through the use of coatings, e. g. A1203/TiN[129].
6.5.4 Corrosion resistance of Si3N4

Silicon nitride-based ceramics are promising engineering materials for application
under corrosive and wear conditions [135-142]. However, the use of silicon nitride
materials in hot acids under hydrothermal conditions and in bases is often limited
by degradation of the materials. This is the impetus for current efforts for better
understanding of the influence of the microstructure on the corrosion behaviour
of silicon nitride materials and to improve the materials’ behaviour on its basis.
Up to now, only a small amount of data about the mechanisms of corrosion and
pitting formation exists.
6.5 Properties o f S i 3 N 4 Muterials 787
Table 5. Classification of corrosion conditions and main features of the corrosion behaviour
Conditions Corrosion Literature
Organic components (oil, Work as lubricants, reducing the 113, 115

hydrocarbons) wear
Acids (HCI, H2S04, H N 0 3 . . . ) Main attack at the grain 135-146, 148
boundary phase
Corrosion resistance can be
improved significantly by
tailoring the composition
Media-intensive solving SiOz Dissolution of the grain 135, 144-146, 149
protective layers, H F alkaline melts; boundary and the Si3N, grains
concentrated alkaline solutions at
high temperatures > 100 to 150°C;
hydrothermal conditions at 2250°C Intensive corrosion
Bases at medium temperatures Main attack at the grain 135, 138, 140,
<lo0 to 150°C boundary phase 144, 149, 150
Hydrothermal conditions at Main attack at the grain 136, 141, 151--155
T 5 200°C boundary phase
Nevertheless, some correlation’s between the microstructure and the corrosion

behaviour are known. The corrosion behaviour of silicon nitride materials in
liquids is mainly controlled by the stability of the grain-boundary phase. Therefore
the corrosion resistance of the silicon nitride materials in different media can be
altered by more than two orders of magnitude by changing the composition of
the material. The corrosion behaviour can be organised into a few main classes
(see Table 5).
Standard silicon nitride materials with Y203/Al2O3sintering additives degrade
quite strongly in hot acids of medium concentration (see Figures 29-31). With
increasing temperature the corrosion resistance decreases, whereas with decreasing
additive content the corrosion resistance increases. The best corrosion resistance in
acids was obtained for a HIP’ed Si3N4material without additives. Si3N4materials
are more stable in concentrated acids (>5 N) than in solute acids [135, 136, 138, 1481.
The MgO-containing materials possess better corrosion resistance than the Y 2 0 3 /
A1203-containing materials. The reasons for this are not clear [136, 1411. The
corrosion behaviour is different in H 3 P 0 4than in other acids due to formation of
a protective phosphate layer [138, 1441.
Corrosion in HF-containing solutions is much more intensive than in other acids.
This is connected with the ability of H F to dissolve the SiO, protective layers. There-
fore H F solutions will not only attack the grain-boundary phase but also dissolve
the Si3N4grains.
Figure 29. a) SEM micrograph of an corroded in H2S04 100 h 90°C material with Y203/A1203
additives and b) SEM micrograph of an material with optimized grain boundary corroded in
H2S04 500 h 90°C material (plasmaetched).
Recent investigations show that, besides the amount of the additives, the main
parameter governing the corrosion resistance of the silicon nitride materials is the
Si02 content in the grain-boundary phase (see Figures 29-31). A material with a
high S O z content in the grain-boundary phase is more stable in acids than one
with a low Si02 content. This correlates with the stability of the grain-boundary
phase. The acid corrosion of glasses shows a similar behaviour [147,155]. The net-
work modifier is leached from the glass network by the acids and the glass network
hydrated. Therefore a network with a low amount of network modifiers has a
higher stability [ 1471. Additional to the factors influencing the corrosion of glass,
the corrosion of silicon nitride is affected by the fact that corrosion takes place
only in small channels between the Si3N4 grains. During corrosion the Si3N4
grains also produce hydrated Si02, which can reduce the corrosion rate and
change the corrosion mechanism. This seems to be one of the reasons for the
low corrosion rate of Si3N4 materials with a low amount of sintering additives
[141].
6.5 Properties of Si,N4 Muterials 789
P
I OY 203/A1203- 1
I
m
. 5Mg0
10 MgO/Y203/A1203
*- *-
<
N
1
5
;;4
E
-
v1
0
- -- 3,2 Vol.%
0 100 200 300 400
Time, h
b)
105 1
751 \
3,2 Val.%,
65
0 100 200 300 400
Time, h
C)
Figure 30. Weight loss of different Si3N4-materialsin H2S04 at 60°C (additive composition is given
at the curves) a) and weight loss b) and residual strength c) of different materials with different
amount of Y203/A1203sintering additives in HCI at 60°C.
190 6 Silicon Nitride Bused Hard Muterials
0,li ,
25 30 35 40 45 50 55 60 65 70
' ' " ' ' ' 8
Si02-content, Mol%
Figure 31. Weight loss of silicon nitride materials with different Si02 content in the grain boundary
(200 h corrosion in H2S04 at 60°C).
Additionally, the concentration of the other constituents in the glassy grain-

boundary phase has an influence on the corrosion behaviour more detailed
investigations are necessary to confirm this,
The kinetics of the corrosion of Si3N4materials in acids and the corrosion mechan-
isms are not clear. Data about diffusion and reaction-controlled mechanisms exist. It is
possible that the mechanism changes, depending on the corrosion depth, the acid con-
centration, temperature and microstructural parameters, but further investigations are
necessary to confirm this [141]. The reduction of strength after corrosion in acids for a
short time (pitting formation) cannot be correlated with the weight loss. Acid corro-
sion have an influence on the subcritical crack growth of the materials [138].
The corrosion resistance of selected Si3N4materials in NaOH is given in Fig. 32.
The silicon nitride materials are less attacked by bases than by acids. The extent of
corrosion increases with increasing temperature and concentration of the bases
[150]. Materials which are less stable in acids are more stable in bases and under
hydrothermal conditions.
The most stable materials under hydrothermal conditions are materials with a low
Si02 content in the grain-boundary phase (see Fig. 33). The HIP'ed materials with-
out additives are, besides the MgO-containing materials, less stable than the Y 2 0 3 /
A1203-containing materials. Under hydrothermal conditions the grain-boundary
phase of these materials is dissolved completely, leading to removal of the silicon
nitride grains from the surface. This behaviour is different to the corrosion in
acids, where the grain boundary is not completely dissolved and the corroded struc-
ture is quite strong and stable. This is connected with the stability of the glass
network. Under corrosion conditions in acids mainly the network modifiers are
dissolved, whereas under hydrothermal corrosion conditions and in strong basic
solutions the SiO, network dissolves. Therefore materials with a high SiOz content
6.5 Properties of Si3N4 Muterials 79 1
7%Y203/A1203
Time. h
7%Y 203/Al203
I O%Y203/A1203
0 50 100 I50 200
Time , h
Figure 32. Weight loss and residual strength of silicon nitride materials in 1 N NaOH at 60°C.
in the grain boundary are less stable under hydrothermal conditions than materials
with a low Si02 content in the grain-boundary phase.
Under hydrothermal corrosion at 270"C, a significant dissolution of the Si3N4
grains takes place. In materials with Y203/A1203as sintering additives, the dissolu-
192 6 Silicon Nitride Bused Hard Mutevials
Si3N4
MgOiA1203
r4
‘E 1-
HIPSN
E
Sl
y:
without additives
-0
0,014
10 100 1000
Time, h
Figure 33. Weight loss of silicon nitride materials under hydrothermal conditions at 210°C.
tion rate of the grains is higher than that of the grain-boundary phase (see Fig. 34).
The weight loss due to hydrothermal corrosion depends linearly on time. Only
materials with high corrosion rates (MgO-containing and Si02-rich materials)
show an increase of the corrosion rate with time.
Investigations of the corrosion behaviour of silicon nitride materials in acids,
bases and under hydrothermal conditions show that a main parameter governing
the corrosion resistance is the composition and amount of the grain-boundary
phase. Materials with a high Si02 content in the grain-boundary phase are stable
in acids and less stable in bases and under hydrothermal conditions. Materials
with a low Si02 content in the grain-boundary phase are stable in bases and
under hydrothermal condition but less stable in acids.
Up to now no materials which are simultaneously stable in acids, bases and under
hydrothermal conditions exist.
A more detailed understanding of the corrosion mechanisms and the influence of
the medium on the lifetime of the materials is necessary before silicon nitride
materials can be used in a broad range of applications under corrosive conditions.
The results of the investigations indicate that tailoring of the microstructure and
the grain boundary composition is necessary for the application of silicon nitride
ceramics in corrosive environments.
6.6 Conclusions/Further potential of silicon nitride

materials
Silicon nitride ceramic materials exhibit a high potential for structural applications
at room and elevated temperatures. This is due to their excellent mechanical
properties in combination with good corrosion and thermal shock resistance.
6.6 ConclusionslFurther potential of silicon nitride rnateriu1.y 793
Figure 34. Corrosion layer of Si3N4 materials a) MgO/AI2O3-additive system, hydrothermal

corrosion conditions: 210'C, 200 h under at 250°C; b) Y2O3/AI2O3-additivesystem, hydrothermal
corrosion conditions: 270"C, 200 h (the grain-boundary phase solves less than the Si,N4 grains and
are visible as bright areas).
The development of the silicon nitride materials in the future will be divided in
two main directions; the increased penetration in the market of the state of the
art silicon nitride materials by improving the technology and reliability and decreas-
ing the production costs and the development of new materials with improved wear,
corrosion, high temperature properties.
In the last decade a continuously increase in the number of application fields was
observed. The technical feasibility and the reliability of the materials and compo-
nents were shown in many applications, such as valves and other parts for the
automotive industry, bearings, household applications. For the wider penetration
into the market there is a engineering work necessary to optimise the production,
stabilise the reliability and reduce the production cost. First steps in this direction
were made by using low cost powders and optimising processing, sintering and
finishing.
The majority of applications can be realised with strength levels lower 1000-
1200 MPa.
Higher strength level need a special more expensive finishing procedure increasing
the costs of parts. Additionally a strength level of 1200 MPa corresponds to a defect
size of 10-15 pm (at the existing KTClevels) which is in the range of defects created
during operating of the materials.
Whereas the direction of further material development mentioned above is
connected with optimising the technology, the second direction is connected with
adapting of the microstructure and properties of silicon nitride materials on differ-
ent applications. As like as exists different kinds of steel, which are well designed for
the given applications, as well in the future will exist different classes of silicon
nitride materials aimed for applications under different conditions e.g. wear, corro-
sion, operation at high temperature and so on (see Table 1).
These differentiation processes stay only at the beginning and need as a bases a
controlled formation of the microstructures and the knowledge of the relation
between the microstructure and properties.
The main factors and processes influencing the microstructure of Si3N4materials
are known and the microstructure can be tailored in a wide range. In the future the
use of different p-silicon nitride powders as seeds for a precise tailoring of the micro-
structure will be used more intensively. New aspects can be expected by the
development of fine grained materials. The correlation between microstructure,
strength and toughness are well understood, whereas the understanding of the
correlation between microstructure on the one side and wear, corrosion properties
and long term behaviour on the other side need further intensive interdisciplinary
investigations.
An intensive development of high fracture toughness a’-SiAlON-materials can be
expected. The better understanding of the microstructure formation of these
materials and a further improvement of strength and toughness by a reduced
grain size are main topics of the investigations.
Frequently, there is a need to make a compromise between the different properties
due to the opposite dependencies on the microstructure. How such compromise
looks like depends on the technological possibilities and on the powder properties,
which developments offer new possibilities.
An improvement of the materials behaviour can be expected by the development
of homogeneous composite materials based on silicon nitride, as it was shown for
Si3N4/MoSi2or Si3N4/SiCcomposites for the high temperature long term applica-
tions. Especially in this area the use of metallorganic precursors can lead to an
improvement of the materials. The reduction of sintering temperature and in
References 795
some modifications the shrinkage (concept of active filler) can allow the production
of composites of thermodynamic unstable components. The development in this
direction is only at the beginning.
Acknowledgements
The authors are grateful to the colleges of the IKTS which take part in the technical
completion of the paper especially A. Bales and D. Hermannutz. They are indebted
to the coworkers in the IKTS especially to Dr. I. Schulz, Dr. T. Reich, A. Bales and
G. Michael involved in research projects in the field of silicon nitride materials which
results were included in this paper. The work is based on different projects supported
by the BMBF, AiF and the DFG in the field of silicon nitride materials.
References
1. S. Hampshire, “Nitride Ceramics in Structure and Properties of Ceramics”, ed. R. W. Cahn,
P. Haasen and E. J. Kramer, Materials Science and Teclznology, 11, 121-168.
2. R. N. Katz, “Application of Silicon Nitride based Ceramics in the U.S.”, Mat. Res. Soc. Symp.
Proc., 287, 1993. 197.
3. K. H. Jack, “SiALON Ceramics: Retrospect and Prospect”, Mat. Res. Soc. Symp. Proc., 287,
1993, 15-18.
4. J. G. Heinrich and H. Kruner, “Silicon Nitride Materials for Engine Applications”, M. J.
Hoffmann and G. Petzow (eds.), Tailoring of’ Mechanical Properties of Si,N4 Ceramics,
Kluwer Academic Publishers, 1994, 1941.
5. “Der Kochherd der Zukunft arbeitet mit Si3N4-Platten”,cfi/Ber. DKG, 76, 1999, D 18-20.
6. J. Heinrich, “Siliciumnitridkeramik”, editor: J. Kriegsniann, DKG Technisclze Keramisclze
Werkstofe, 1999, 4.3.1 .O.
7 . H. M. Jennings, “Review on Reactions between Silicon and Nitrogen, Part 1 Mechanisms”,
Journal of Material Science, 18, 1983, 951 -967.
8. A. Zerr, G. Miehe, G. Serghiou a. o., “Synthesis of Cubic Silicon Nitride”, Nature, 400, 1999,
340-344.
9. W. Ching, Y. Xu, J. Gale and M. Riihle, “Ab-Inito Total Energy Calculation of cc- and p-sili-
con nitride and the Derivation of Effective Pair Potentials with Application to Lattice
Dynamics”, J . Am. Ceram. Soc, 81, 1998, 3189-3196.
10. I. Kohatsu and J. W. McCauley, “Re-Examination of the Crystal structure of a-Si3N4”,Mat.
Res. Bull, 9, 1914, 9 17-920.
11. H. Suematsu, J. J. Petrovic and T. E. Mitchell, “Plastic deformation of silicon nitride single
crystals” Materials Science und Engineering, A 209, 1996, 97-102.
12. J. Ihsza, T. Eschner and K. Rundgren, “Hardness anisotropy in bimodal grained gas pressure
sintered Si3N4”,Journal of Materials Science Letters, 16, 1997, 16641667.
13. Andrievskii, Spivak I. I. , “Nitrid kremnija i materiali na evo osnove”, Moskau, Metallurgijr,
1984, 19-20.
14. H. Mandal, D. P. Thompson and K. H. Jack, “cc H f3 Phase Transformations in Silicon
Nitride and Sialons”, preprint.
15. L.-0. Nordberg, “cc-Sialon Ceramics and Y-cc-Sialon Composites; Composition, Microstruc-
tures and Properties”, Doctorial Dissertation, 1997, 1-124.
16. K. H. Jack, “a-SiAION Ceramics”, Nature, 274, 1978, 880-882.

17. T. Y. Tien, G. Petzow, L. J. Gauckler and J. Weiss, “Phase Equilibrium Studies in Si3N4-
Metal Oxides Systems”, Progress in Nitrogen Ceramics, 1983, 89-99.
18. D. D. Lee, S. J. Kang, G. Petzow and D. N. Yoon, “Effect of c( to p Phase Transition on the
Sintering of Silicon Nitride Ceramics”, J. Am. Ceram. Soc., 73, 1990, 767-9.
19. M. N. Rahaman, “Ceramic Processing and Sintering”, Marcel Dekker Znc., Department of
Ceramic Engineering University of Missouri, 1995.
20. M. Herrmann und Chr. Schubert, “Grundlagen der Gefugeausbildung in 0-Siliciumnitridwerk-
stoffen”, editor: J. Kriegsmann, DKG Technische Keramische Werkstoffe,Sep. 1999, 5.1.3.1.
21. P. Greil, “Einflup der intergranularen Glasphase auf die Hochtemperatureigenschaften von P-
SiALONen” Dissertation Stuttgart 1982.
22. M. Herrmann, G. Putzky, S. Siege1 and W. Hermel, “Einflup von Zersetzungsreaktionen auf
die Sinterung”, c$/Ber DKG, 69, 1992, 375-382.
23. P. Obenaus and M. Herrmann, “Methode zur quantitativen Charakterisierung von Stengelk-
ristalliten in Siliziumnitridkeramik”, Prakt. Met., 27, 1990, 503-5 13.
24. H. Bjorklund, L. K. Falk, K. Rundgren and J. Wasen, “a-Si3N4Grain Growth, Part I: Effect of
Metal Oxide Sintering Additives” Journal of the European Ceramic Society, 17, 1997, 1285-1299.
25. H. Bjorklund and L. K. L. Falk, “a-Si3N4 Grain Growth, Part 11: Intergranular Glass
Chemistry”, Journal of the European Ceramic Society, 17, 1997, 1301-1308.
26. G. Wotting, B. Kanka and G. Ziegler, “Microstructural Development, Micccrostructural
Characterization and Relation to Mechanical Properties of Dense Silicon Nitride”, Nonoxide
Ceramic, 1986, 83-95.
27. G. Ziegler, J. Heinrich and G. Wotting, “Relationship between Processing, Microstructure
and Properties of dense and Reaction-Bonded Silicon Nitride”, J. of Mat. Sci., 22, 1987,
3041-3086.
28. F. F. Lange, “Fracture Toughness of Si3N4as a Function of the Initial a-Phase Content”,
J. Americ. Ceram. Soc., 62, 1979, Nr. I , 428429.
29. G. Woetting, H. Feuer and E. Gugel, “The Influence of Powder and Processing Methods on
Microstructure and Properties of Dense Silicon Nitride”, Mat. Res. SOC.Symp. Proc., 287,
1993, 133-146.
30. S. Kessler, M. Herrmann and Chr. Schubert, “The a-P-Transformation and anisotropic
growth of 0-grains during sintering of Si3N4”, Mat. Sci. Forum, 94-96, Trans. tech. Publ.,
1991, 821-827.
31. G. Petzow and M. J. Hoffmann, “Grain Growth Studies in Si3N4-Ceramics”, Mat. Sci.
Forum, 113-115, 1993, 91-102.
32. M. Herrmann, S. Kessler and Chr. Schubert, “Microstructural Design of Dense Si3N4”,Euro-
Ceramics IZ, V01.2,Edited by G. Ziegler and H. Hausner, 1991, 847-851.
33. W. Hermel, M. Herrmann and Ch. Schubert, “Have the limits of the development of P-Si3N4-
Materials Already been reached”, In: Third Euro-Ceramics, V3, edited by P. Duranand, J. F.
Fernandez, Faenca Edititrice Iberica S. L., 1993, 391-396.
34. W. Hermel, M. Herrmann and I. Schulz, “Sintering and Microstructure of Si3N4-Materials”,
Silicon Nitride 93, Edited by M. J. Hoffmann, P. F. Becher, G. Petzow, Trans tech Publications,
1993, 181-186.
35. W. Dressler, H.-J. Kleebe, M. J. Hoffmann, M. Ruhle and G. Petzow, “Model Experiments
Concerning Abnormal Grain Growth in Silicon Nitride”, J. qf’ the Europ. Ceram. Soc., 16,
1996, 3-14.
36. M. Kramer, M. J. Hoffmann and G. Petzow, “Grain Growth Studies of Silicon nitride
Dispersed in an Oxynitride glass’’, J. Am. Ceram. Soc., 76, 1993, 2778-84.
37. W. Dressler and R. Riedel, “Progress in Silicon Based Non-Oxide Structural Ceramics”, Znt.
J. of Refractory hard Materials, 15, 1997, 1347.
38. M. Kramer, “Untersuchungen zur Wachstumskinetik von P-Si,N4 in Keramiken und Oxini-
tirdglasern”, Dissertation, Stuttgart, 1992.
39. G. Ziegler, W. Lehner and H.-J. Kleebe, “Nucleation affecting microstructure development of
Si3N4ceramics”, Brit. Ceram. Proc., 60, 1999, 5-7.
40. S.-L. Hwang and I.-W. Chen, “Nucleation and Growth of a’-SiAION on a-Si3N4”, J . Am.
Ceram. SOC.,77, 1994, 1711-18.
RLferences 797
41. H.-J. Kleebe, “Structure and Chemistry of Interfaces in Si,N4 Ceramics Studied by Transmis-
sion Electron Microscopy”, Journal of’the Ceramics Society of Japan, 105, 1997, 453475.
42. T.-Y. Tien, “Silicon Nitride Ceramics-Alloy Design”, Mut. Res. Soc. Symp. Proc., 287, 1993.
43. T. Y. Tien and C. J. Hwang, “Microstructural Development in Silicon Nitride Ceramics”,
Mater. Sci. Forum, 47, 1989, 84109.
44. L. L. Wang and T. Y. Tien, “Morphology of Silicon Nitride Grown from a Liquid Phase”,
J. Am. Ceram. Soc, 81, 1998, 10, 2677-2686.
45. D. Y. Yoon, Korean Advanced Institute of Science and Engineering, Report 16.6. 1999 in the
IKTS.
46. Y.-J. Park. N.-M. Hwang and D. Y. Yoon, “Abnormal Growth of Faceted (WC) Grains in a
(Co) Liquid Matrix”, Metal/. Mater. Trans., 27A, 1996, 1-1 1.
47. D.-D. Lee, S.-J. L. Kang and D. N. Yoon, “Mechanism of Grain Growth and a-p’ Transfor-
mation During Liquid-Phase Sintering of ff-Sialon”, J. Am. Ceram. Soc., 71, 1988, 803-806.
48. N. Hirosaki, Y. Akimune and M. Mitomo, “Microstructural Design by Selective Grain
Growth of a-Si3N4”, Mat. Res. Soc. Symp. Proc., 287, 1993, 405410.
49. M. Herrmann, I. Schulz and J. Hintermayer, “Materials From Low Cost Silicon Nitride
Powders”, Proc. 4th EcerS Conf ., Riccione, Ed. by C. Galassi, Gruppo Editoriale Faenza
Editrice, 1995, 21 1-216.
50. N. Hirosaki, M. Ando, Y. Akimune and M. Mitomo, “Gas-Pressure Sintering of p-Silicon
nitride Containing Y z 0 3 and Nd z0 3” ,J. of Ceranz. Soc. ofJapan, 100, 1992, 817-820.
51. M. Mitomo, M. Tsutsumi, H. Tanaka, S. Uenosono and F. Saito, “Grain Growth During
Gas-Pressure Sintering of p-Silicon Nitride”, J . Am. Ceratn. Soc., 73, 1990, 244145.
52. M. Mitomo, H. Hirotsuru, H. Suematsu and T. Nishimura, “Fine-Grained Silcon Nitride
Ceramics Prepared from P-Powder”, J . Am. Ceram. Soc., 78, 1995, 21 1-14.
53. J. Hintermayer, G. Ernst, W. Gmohling, G.Schrol1 and W. Kobler, “Verfahren zur Herstel-
lung von Siliziumnitrid” EP 0377132B1, 1994.
54. E. Hidseyuki and M. Mitomo, “Control and Characterization of Abnormally Grown Grains
in Silicon Nitride Ceramics”, Journal of the European Ceramic Society, 17, 1997, 797-804.
55. M. Mitomo, “In-situ Microstructural Control in Engineering Ceramics”, Key Engineering
Materials, 161-163, 1999, 53-58.
56. P. Branvall and K. Rundgren, “Grain growth mechanisms in in situ reinforced Si3N4”,Brit.
Ceranz. Proc., 60, 1999, 7-9.
57. K. Hirao, T. Nagaoka, M. E. Brito and S. Kanzaki, “Microstructure Control of Silicon
Nitride by Seeding with Rodlike 8-Silicon Nitride Particles”, J . Am. Ceram. Soc., 77, 1994,
1857-62.
58. K. Hirao, T. Nagaoka, M. E. Brito and S. Kanzaki, “Mechanical Properties of Silicon
Nitrides with Tailored Microstructure By Seeding”, Journal of the Ceramic Society of
Japan, 104, 1996, 55-59.
59. R. A. Andrievski, “Some high-temperature properties of silicon nitride”, High Temperature -
High Pressure, 26, 1994, 451455.

60. H. Emoto and H. Hirotsuri, “Microstructure Control of Silicon Nitride Ceramics Fabricated
from a-Powder Containing Fine 8-Nuclei”, Key Engineering Materials, 161-163, 1999, 209-
212.
61. K. Hirao, H. Imamura, K. Watari, M. E. Brito, M. Toriyama and S. Kanzaki, “Seeded Silicon
Nitride: Microstructure and Performance”, Key Engineering Materials, 161-163, 1999, 469-
474.
62. M. Herrmann, I. Schulz, Chr. Schubert and W. Hermel, “Silicon Nitride Materials with Low
Friction Coefficients”, Key Engineering Materials, 161-163, 1999, 599-602.
63. M. Herrmann, I. Schulz, Chr. Schubert, I. Zalite and G. Ziegler, “Ultrafine Si3N4-Material
with Low Coefficient’s of Friction and Wear Rates”, c j / B e r . DKG, 75, No 4, 1998, 3845.
64. G . Riedel, H. Bestgen and M. Herrmann, “Influence of Sintering Additives with Differing
Proportions of Y2O3/AI2O3on the Sintering and Material Properties of Si,N4 Ceramics”,
cJ, Ber. DKG, 75, No. 1-2, 1998, 30-34.
65. K. Watari, K. Ishizaki, S. Cao and K. Mori, “The Relationship between Thermal Conduc-
tivity and Microstructure in Si3N4 Ceramics”, Key Engineering Materials, 161-163, 1999,
2 13-2 16.
66. N. Hirosaki, Y. Okamoto, M. Ando, F. Munakata and Y. Akimune, “Effect of Grain Growth
on the Thermal Conductivity of Silicon Nitride”, Journal of the Ceramic Society ofJapan, 104,
1996, 50-54.
67. D. A. Bonnell, M. Riihle and T.-Y. Tien, “Redistribution of Aluminum Ions During Proces-
sing of Sialon Ceramics”, J. Am. Ceram. Soc., 69, 1986, 623-27.
68. M. Kitayama, K. Hirano, M. Toriyama and S. Kanzaki, “Anisotropic Ostwald Ripening in 0-
Si3N4With Different Lantanide Additives”, G . L. Messing, F. F. Lange, S.-I. Hirano: Ceramic
Processing Science, 5 17--24.
69. M. J. Hoffmann, “Analysis of Microstructural Development and Mechanical Properties of
Si3N4Ceramics”, M . J. Hoffmann, G . Petzow, Tailoring of Mechanical Properties of Si3N4
Ceramics, Kluwer Academic Publishers, 1994, 59-72.
70. H. J. Kleebe, G. Pezzotti and G. Ziegler, “Microstructure and Fructure Toughness of Si,N4
Ceramics”, J. Am. Ceram. Soc., 82, 1999, 1857-1860.
71. A. R. Prunier and A. J. Pyzik, “Self reinforced Silicon Nitride for Cutting Tool Application”,
Key Eng. Mat., 89-91, 1994, 129-134.
72. A. J. Pyzik and D. R. Beaman, “Microstructure and Properties of Self-Reinforced Silicon
Nitride”, J. Am. Ceram. Soc., 76, 1993, 273144.
73. M. Menon and I.-W. Chen, “Reaction Densification of a’- Sialon: I, Wetting Behavior and
Acid-Base Reactions”, Journal of the American Ceramic Society, 78, 1995, 545-552.
74. M. Menon and I.-W. Chen, “Reaction Densification of d-Sialon: 11, Densification Behavior”,
Journal of the American Ceramic Society, 78, 1995, 553-559.
75. H. Mandel and M. J. Hoffmann, “Novel developments in a-SiALON ceramics”, Brit. Ceram.
Proc., 60,1999, 11-13.
76. Schutzrecht USA WO 98/23554 (1998-06-04).
77. I-W. Chen and A. Rosenflanz, “A tough SiAlON ceramic based on cc-Si3N, with a whisker-
like microstructure”, Nature, 389, 1997, 70 1-704.
78. H. Klemm, M. Herrmann, T. Reich, C. Schubert, L. Frdssek, G. Wotting, E. Gugel and
G. Nietfeld, “High-Temperature Properties of mixed a’/$-Sialon Materials”, Journal of the
American Ceramic Society, 81, 1998, 1141-1 148.
79. T . 4 . Sheu, “Microstructure and Mechanical Properties of In Situ 0-Si3N4/d-SialonCompo-
site”, Journal of the American Ceramic Society, 77, 1994, 2345-2353.
80. M. Herrmann, annual report 1998, IKTS, 1998.
81. N. Camascu, D. P. Thompson and H. Mandal, “Effect of Starting Composition, Type of Rare
Earth Sintering Additive and Amount of Liquid Phase on a’/p-Sialon Transformation”,
Journal of the European Ceramic Society, 17, 1997, 599-613.
82. K. Urashima, Y. Ikai and S. Iwase, “Features of a Superior Strength Si3N4 Ceramic and its
Applications”, 6th International Symposium on Ceramic Materials and Components for
Engines, 1997, 167-112.
83. G. Riedel and H. Kruner, “Si3N4-material with high strength (1400MPa)”, Third Euro
Ceramics, Madrid, 1993, 453-458.
84. R. F. Silva and J. M. Vieira, “Hot hardness of Si3N4-basedmaterials”, Journal of Material
Science, 30, 1995, 5531-5536.
85. M. Yoshimura, T. Nishioka, A. Yamakawa and M. Miyake, “Grain Size Controlled High-
Strength Silicon Nitride Ceramics”, J. Ceram. SOC.Japan, 1995, 407.
86. T. Nishioka, K. Matsunuma, T. Yamamoto, A. Yamakawa and M. Miyake, “Development of
High Strength Si3N4Material for Automobile Parts”, Sumitomo Electric Technical Review, 36,
1998, 77.
87. N. Hirosaki, Y. Akimune and M. Mitomo, “Effect of Grain Growth of 0-Silicon Nitride
on Strength, Weibull Modulus and Fracture Toughness”, J. Am. Ceram. Soc., 76, 1993,
1892-94.
88. P. F. Becher, H. T. Lin, S. L. Hwang, M. J. Hoffmann and I. W. Chen, “The Influence of
Microstructure an the Mechanical Behaviour of Silicon Nitride Ceramics”, Mat. Res. Soc.
Symp. Proc., 281, 1993, 147-157.
89. P. Sajgalik, J. Dusza and M. J. Hoffmann, “Relationship between Microstructure, Toughen-
ing Mechanism, and Fracture Toughness of Reinforeced Silicon Nitride Ceramics”, J. Am.
Ceram. Soc., 78, 1995, 2619-24.
References 799
90. C. W. Li, C. J. Gasdaska, J. Goldacker and S. C. Lui, “Damage Resistance of in Situ Rein-
forced Silicon Nitride”, Mat. Res. SOC. Symp. Proc., 287, 1993, S. 473-480.
91. B. Speicher, G. A. Schneider, W. Drepler, G. Lindemann, H. Boder and V. Knoblauch,
“Reliability of Ceramic Valve plates for Common-Rail Injection Pumps”, preprint Materialica
1999, 1999.
92. I. Peterson and T. Tien, “Effect of the Grain Boundary Thermal Expansion Coefficient on the
Fracture Toughness in Silicon Nitride”, J . Am. Crram. Soc., 78, [9], 1995, 2345-52.
93. E. Y. Sun, P. F. Becher and K. P. Plucknett, “Microstructural Design of Silicon Nitride with
Improved Fracture Toughness: 11, Effects of Yttria and Alumina Additives”, J . Am. Ceram.
SOC.,81, 1998, 2831-2840.
94. A. G. Evans, “Perspective on the Development of High-Toughness Ceramics”, J . Am. Ceram.
SOC.,73, 1990, 187-206.
95. H. Kessler. H. Kleebe, R. Cannon and W. Pompe, “Influence of Internal stresses on Crystal-
lization of Intergranular Phases in Ceramics”, Acta metall muter., 40,1992, 2233-2245.
96. I. Tanaka, G. Pezotti, Y. Miyamoto and T. Okamoto, “Fracture Toughness of Si3N4and its
Si3N4 Whisker Composite Without Sintering Aids”, J . of Mat. Sci., 26, 1991, 208-210.
97. J. Zhihao and Q. Guanjun, “Fatigue Behavior of some Advanced Ceramic Materials”, 6th
International Symposium on Ceramic Materials and Componentsfor Engines, 1997, 760-765.
98. T. Ogasawara and Y. Mabuchi, “Effect of Humidity and Cyclic Loading Condition on
Fatigue behavior of Silicon Nitride”, J . Ceram. Soc. Jap., Int. Edition, 101, 1993, 1122-1 127.
99. D. Jacobs and I. Chen, “Mechanical and Environmental Factors in the Cyclic and Static
Fatigue of Silicon Nitride”, J . Am. Ceram. Soc., 77, 1994, 1153-61.
100. P. F. Becher, E. Y. Sun, K. P. Plucknett, K. B. Alexander, C. H. Hsueh, H. T. Lin, S. B.
Waters and C. G. Westmoreland, “Microstructural Design of Silicon Nitride with Improved
Fracture Toughness: I, Effects of Grain Shape and Size”, J . Am. Ceram. SOC., 81, 1998,
2821-30.
101. R. Schober, M. Herrmann, W. Krehler, H.-J. Richter, W. Hermel and G. Naumann, “Verdich-
tungsverhalten und Eigenschaften von nachgesintertem reaktionsgebundenem Siliziumnitrid
(SRBSN)”, editor: J. Kriegsmann, DKG Technische Keramische Werkstofe, 3. Erg.-Lfg. 1999.
102. I. Tanaka, G. Pezzotti, T. Okamoto, Y. Miyamoto and M. Koizumi, “Hot Isostatic Press
Sintering and Properties of Silicon Nitride at Elevated Temperatures”, J . Am. Ceram. SOC.,
74, 1992, 752-59.
103. H. Klemm, M. Herrmann and Chr. Schubert, “High Temperature Oxidation and Corrosion of
Silicon-Based Nonoxide Ceramics”, A S ME Turbo Expo ‘98,Stockholm, Sweden 1998, ASME
Paper 98-GT-480.
104. H. Klemm, M. Herrmann and Chr. Schubert, “High Temperature Oxidation of Silicon Nitride
Based Ceramic Materials”, Proc. 6th Internat. Symp. on Ceram. Muter. & Comp.for Engines,
Arita, Japan, 1997, 57681.
105. I. Tanaka, G. Pezzotti, K. Matsushita, Y. Miyamoto and T. Okamoto, “Impurity Enhanced
Cavity Formation in Si3N4at Elevated Temperatures”, J . Am. Ceram. SOC.,74, 1992,752-59.
106. H. Klemm and G. Pezzotti, “Fracture Toughness and Time-Dependent Strength Behavior of
Low- Doped Silicon Nitrides for Applications at 1400C”, J . Am. Ceram. Soc., 77, 1994,553-61.
107. G. Pezzotti, K. Ota and H.-J. Kleebe, “Viscous Slip along Grain Boundaries in Chlorine-
Doped Silicon Nitride,” J. Am. Ceram. SOC.,80, 1997, 234148.
108. M. Herrmann, Chr. Schubert, A. Rendtel and H. Hiibner, “Silicon Nitride/Silicon Carbide
Nanocomposite Materials: I, Fabrication and Mechanical Properties at Room Temperature”,
J . Am. Ceram. Soc., 81, 1998, 1095-108.
109. K. Niihara, “New Design Concept of Structural Ceramics - Ceramic Nanocomposites”, The
Centennial Memorial Issue of the Ceramic Society of Japan, 99, 1991, 974.
110. M. Herrmann, H. Klemm, Chr. Schubert and W. Hermel, “Long-Term Behavior of Sic/
Si3N4-Nanocomposites at 1400-1500”C”, Key Engineering Muter., 132-136, 1977-80, 1997.
11 1. G. Pezzotti and M. Sakai, “Effect of a Silicon Carbide ‘Nano-Dispersion’ on the Mechanical
Properties of Silicon Nitride”, J . Am. Ceram. SOC.,77, 1994, 303941.
112. H. Klemm. K. Tangermdnn, Chr. Schubert and W. Hermel, “Influence of Molybdenum
Silicide Additions on High-Temperature Oxidation Resistance of Silicon Nitride Materials”,
J . Am. Ceram. Soc., 79, 1996, 2429-35.
113. R. Sternagel and M. Popp, “Innovationen fur Gleitlager, Walzlager, Dichtungen und Fuhrun-
gen”, VDI-Ber., 1331, 1997, 131-138.
114. M. Rombach, R. Schofer and W. Pfeiffer, “Tragfahigkeit und Lebensdauer fur Wdzlager aus
Hochleistungskeramik”, Report of a BMBF- project 03 M2107B, 1996.
115. M. Popp, R. Sternagel and G.Wotting, “Hybride- and Ceramic Rolling Bearings with modi-
fied Surface and low Friction Rolling Contact”, Euromat 1999, 1999.
116. Ch. Schubert and M. Herrmann, “Materials development of silicon nitride for ceramic ball
bearings”, Report of the BMBF- project Nr. 03 M2107B4, 1996.
117. M. Herrmann, I. Schulz, Chr. Schubert and W. Hermel, “Siliziumnitridwerkstoffe mit
niedrigem Reibkoeffizienten und hohem Verschleipwiderstand”, Werkstqfwoche 98, Vol. V ,
Symposium 6 , 1998, 241-247.
118. M. Popp and R. Sternnagel, “Neue Entwicklungen bei Walzlagern aus Hochleistungskera-
mik’, Werkstoffwoche 98, Vol. V , Symposium 6 , 1998, 247-253.
119. B. R. Lawn, S. K. Lee and K. S. Lee, “Fracture and Deformation Damage Accumulation in
Tough Ceramics”, Key Engineering Materials, 161-163, 1999, 3-8.
120. S. K. Lee and B. R. Lawn, “Role of Microstructure in Hertzian Contact Damage in Silicon
Ntride: 11, Strength Degradation”, Journal qf’ the American Ceramic Society, 81, 1998, Nr.
4, 997-1003.
121. S. W. Lee, “Tribological Characterization in Ceramic Based Composite”, Key Engineering
Materials, 161-163, 1999, 593-598.
122. K. H. Habig, “Tribologisches Verhalten von Ingenieur-Keramik”, Ingenieur- Werkstoffe, I,
1989, 78-83.
123. A. Skopp, “Tribologisches Verhalten von Siliziumnitridwerkstoffen bei Festkorpergleitrei-
bung zwischen 22°C und lOOO”C”, Dissertation 1993.
124. S. K. Lee and B. R. Lawn: Role of Microstructure in Hertzian Contact Damage in Silicon
Ntride: 11, Strength Degradation. In: Journal of ihe American Ceramic Society, 81, 1998,
997-1003.
125. H. Tanaka, “A Recent Tendency of Si3N4Cutting Tools”, Advanced Materials: Ceramics,
Powders, corrosion and Advanced Processing, 14A, 1994, 541-545.
126. H. K. Tonshoff and B. Denkena, “Wear of Ceramic Tools in Milling”, Journal of the Society of
Tribologists and Lubrication Engineers, 47, 1991, 772-778.
127. H. K. Tonshoff, Chr. Blawit, M. Rodewald and H.-G. Wobker, “Entwicklung und
VerschleiPverhalten von Si3N4-Schneidkeramik”, Mat.-wiss. u. Werkstofftech., 26, 1995,
255-262.
128. G. Brandt, “Development of ceramic cutting tools”, Maieriaux & Techniques, 85, 1997, 3-12.
129. R. D. Nixon and P. K. Mehrotra, “Optimization of Sintering Aids in Silicon Nitrides for
Cutting Tool Application”, Ceramic Engineering & Science Proceedings, 18, 1997, 457464.
130. H. K. Tonshoff and C. Blawit, “Wear mechanism of silicon nitride in dependence of workpiece
material”, BMFT- Project, Final Report No. 03T0028A, 1993, 1-96.
131. H. Miao, L. Qi and G. Cui: Silicon Nitride Ceramic Cutting - Tools and their Applications”,
Key Engineering Materials, 114, 1995, S. 135-172.
132. G. Brandt, “Wear- and thermal shock-resistant SIALON cutting tool material”, EP 97-850038
9703 1 1.
133. US- Patent 432 325.
134. Y. R. Liu, J. J. Liu, B. L. Zhu, Z. R. Zhu, L. Vincent and P. Kapsa, “Wear Maps of Si3N4
ceramic cutting tools”, J . Mater. Eng. and Perform., 6, 1997, 671-675.
135. Gmelin Handbook, Si-Suppl, b5d2., 11&117; 1995, 184-210.
136. K. Komeya, T. Meguro, S. Atago, C.-H. Lin, Y. Abe and M . Komatsu, “Corrosion Resis-
tance of Silicon Nitride Ceramics”, Key Engineering Materials, 161-163, 1999, 235-238.
137. M. Herrmann, A. Krell, J. Adler, G. Wotting, T. Hollstein, W. Pfeifer and M. Rombach,
“Innovationen fur Gleitlager, Walzlager, Dichtungen und Fuhrungen”, VDZ-Ber., 1331,
1997,251-258.
138. T. Hollstein, T. Graas, K. Bundschuh and M. Schutze, Keramische Zeitschrijt, 50, 1998, 416 -
421.
139. T. Sato, Y. Tokunaga, T. Endo, M. Shimada, K. Komeya, et al., “Corrosion of Silicon Nitride
Ceramics in Aqueous Hydrogen Cloride Solutions”, J. Mat. Science, 23, 1988, 3440-3446.
RL.firences 80 1
140. S. Iio, A. Okada, A. Akira, A. Tetsuo and Y. Masahiro, “Corrosion of silicon nitride ceramics
in solutions. Part 3. Corrosion behavior in hot sulfuricacid and microstructure of the corroded
layer”, Journal oftlie Ceran?ic Society of Japan, h t . Edition, 100, 1992, July, 954-956.
141. M. Herrmann, Chr. Schubert and G . Michael, “Korrosionsstabile keramische Werkstoffe fur
Anwendungen in Walzlagern und im Anlagenbau”. Fortschrittsberichte der DKG, 14, 1999,
13~151.
142. M. Herrmann and G. Michael, “Corrosion Behaviour of Engineering Ceramics in Acids and
Basic Solutions”, Britidl Ceramic Proceedings, 60, extended Abstracts, Vol I, 1999, 455457.
143. A. Okada and M. Yoshimura, “Mechanical Degradation of Silicon Nitride Ceramics in Cor-
rosive Solutions of Boiling Sulphuric Acid”, K q Engineering Materials, 113, 1996, 227-236.
144. Y. G. Gogozi and V. A. Lavrenko, “Corrosion of High Performance Ceramics”, Springer
Verlag, Berlin, 1992, 7 6 7 8 .
145. M. Shimada and T. Sato, “Corrosion of Silicon Nitride Ceramics in H F and HCI Solutions”
Ceram. Trans., 10, 1989, 355-365.
146. Q. Fang, P. S . Sidky and M. G . Hocking, “The Effect of Corrosion and Erosion on Ceramic
Materials”, Corrosion Science, 39, 1997, 51 1-527.
147. A. Paul, “Chemical Durability of Glasses: a Thermodynamic Aproach”. J . Mat. Sci., 12, 1977,
2246-2268.
148. K. Kanbara, N. Uchida, K. Uematsu, T. Kurita, K. Yoshimoto and Y. Suzuki, “Corrosion of
Silicon Nitride Ceramics by Nitric Acid”, Mat. Res. Soc. Symp. Proc., 287, 1993, 533-538.
149. T. Sato, Y. Tokunaga, T. Endo, M. Shimada, K Komeya, K. Nishida, M. Komatsu and
T. Kameda, “Corrosion of Silicon Ceramics in Aqueous H F Solutions”, J . q f M a t . Sci., 23,
1988, 3440-3446.
150. T. Sato, S. Sato, K. Tamura and A. Okuwaki, “Corrrosion Behaviour of SiliconNitride
Ceramics in Caustic Alkaline Solutions at High Temperatures”, Br. Ceram. Trans. J., 91,
1992, 117-120.
151. T. Sato, T. Murakami, E. Shimada, K. Komeya, T. Komeda and M. Komatsu, “Corrosion of
Silicon nitride ceramics under hydrothermal conditions” J . Mat. Science, 26, 1991, 1749-1754.
152. M. Yoshimura and S. Yamamoto, “Corrosion and alteration of ceramic materials in high
temperature-high pressure water”, Serumikkusu, 30, 1995, 995-998.
153. K. Oda, T. Yoshio. Y. Miyamoto and M. Koizumi, “Hydrothermal corrosion of pure, hot
isostatically pressed silicon nitride”, J . A m . Cerum. Soc., 76, 1993, 1365-1368.
154. K. Oda and T. Yoshio, “Properties of YzO3-Al2O3-SiO2Glasses as a Model System of Grain
Boundary Phase of SijN4 Ceramics (Part 2)”, J . Ceraniic Soc. qfJupan, 99, 1991, 11 10-1 112.
155. S. Iio, A. Okada, T. Asano and M. Yoshimura, ”Corrosion Behavior of Silicon Nitride
Ceramics in Aqueous Solutions (Part 3)”, J . o j t h e Ceramic Soc. ofJc~pan,100, 1992,965-967.

R. Telle, L. S. Sigl, and K. Takagi
7.1 Introduction
Materials based on boron compounds have been explored for many decades because
of their exceptional properties in respect to chemical bonding, crystal structure, and
phonon and electron conduction. Especially in the field of energy conversion, elec-
tron emission, and neutron absorption, borides occupy many niches of application
for which no other material can be employed. Until approximately 1980, the main
interest in borides always came, however, from basic research aimed at the
understanding of their electronic structure, being either responsible for the unique
transport properties or the peculiarities in chemical bonding. It is, therefore, no
wonder that the most information about borides was at that time created from
the viewpoint of physicists and chemists.
Although even some of the interesting thermal and mechanical properties of
borides, e.g., the generally high melting point and high hardness, have already
been exploited and led to application as wear-resistant parts or grinding grits a
long time ago, the interest in borides increased dramatically together with the
fundamental understanding of technical ceramics. The exploitation of materials
with exceptional mechanical, chemical, electrical and thermal properties yielded
boron compounds as potential candidates for “high-tech” applications besides the
well-developed oxide ceramics, silicon nitride, silicon carbide, and hard metals or
cermets.
For a successful economic application, these advanced ceramics should be
available without any limitations; prepared by relatively simple methods, they
should exhibit a low specific gravity, high reliability, long lifetime and of course
they should be available at low costs. Unfortunately, most boron-based materials
do not satisfy all of these requirements and, hence, disadvantages and disappoint-
ments have to be tolerated in development and application. In particular, densifica-
tion of powder-derived parts is extremely difficult due to their high amount of
covalent bonding resulting in low diffusion coefficients and, therefore, costly.
Furthermore, oxide impurities on surfaces and grain boundaries create undesired
and unexpected effects on the properties whereas the affinity of boron to oxygen
limits its use at higher temperature in an oxidizing atmosphere. Additionally,
phase diagrams and crystal structures of the particular compounds are the subject
of permanent revision since improvement of the purity of raw materials, more
sophisticated synthesis procedures and characterization methods, such as precur-
sor-based processing, advance chemical analysis, atomic-resolution transmission
electron microscopy, and high-temperature X-ray diffraction, yielded more precise
but also more confusing information about the true characteristics of the boride
structures and phases.
Besides the relatively high costs a lack in understanding is the main reason why
boride-based materials are not yet in as widespread use as the other ceramics
materials, although some properties such as extreme hardness make them superior
to oxides and nitrides. Thus, some “eras” of boride exploration can be distin-
guished, one supported by the needs of the nuclear energy and weapon industry
around 1960, another between 1970 and 1980 for thermo-electric power generation
and coming up again at the present, the next driven by the need for exceptional wear
properties between 1980 and 1990, and finally between 1985 and 1992 triggered by
aerospace and military research. The interest in borides rising from a political and
strategic point of view has contributed to the fact that information about borides is
still difficult to obtain and even more difficult to interpret since most of the data have
been restricted for a long time and their sources and quality cannot be ascertained in
all cases. Thus, sometimes contradictory data are available that cannot be easily
discerned as being right or wrong.
Besides the “high-tech” applications of novel materials, it should not be forgotten
that there was and still is a tremendous market for borides in the metallurgy of steel
and iron, e.g., for antioxidizing additives in refractory linings or as alloying ingredi-
ents for the metals.
Thus, this contribution is aimed at the state of the art in boride ceramics with their
problems in densification, microstructural peculiarities and exceptional mechanical
properties. Starting with the unique interaction of metallic, covalent and ionic types
of bonding and the crystal structures of technically important compounds, phase
diagrams will be presented as far as they are of technical interest. The major part
consists of the description of the synthesis and properties of ceramics and cermets,
reflecting the development of suitable sintering procedures and the consequent
improvement of the thermal and mechanical properties.
7.2 Chemical Bonding and Crystal Chemistry of Borides

The nature of the chemical bonding is the key to the physical and chemical behavior
of matter. Borides possess exceptional properties due to a high amount of covalent
bonding in combination with small band gaps or even metal-like transport proper-
ties. The unique interaction of metallic, covalent and ionic types of bonding results
in a high melting point combined with semiconductivity or metallic electric and
thermal conductivity and excellent wetting by metallic melts. The formation of
many unique crystal structures found only in borides reflects again the outstanding
nature of these materials.
7.2.1 Chemical Bonding of Borides

The nature of the chemical bonding in boron compounds is governed by the well-
known two-electron-three-center bond, i.e., three boron atoms share two common
804 7 Boride-Based Hard Materials
electrons. These electrons are thus more or less delocalized. The resulting sp2 hybri-
dization leads to the plane B3X3 hexagon as the main structural element in BN,
B203,H3B03 and related compounds, and to the B3 triangle as a fraction of the
typical five-fold symmetric icosahedron of elemental boron, the group of boranes
and their derivatives. Depending on the saturation of the electron deficiency, soft
and non-conducting, salt-like compounds or semimetallic to metallic materials of
exceptionally high melting point and hardness and excellent electrical conductivity
exist. As pointed out in the following section, the latter boron compounds may
contain ionic, metallic and covalent fractions of bonding forming very stable
compounds due to the well-balanced electron transfer between metal and boron
sublattice.
7.2.2 The Crystal Structure of Borides

Similar to silicates, the crystal structures of borides can easily be classified according
to the arrangement of the boron atoms. Boron may occur as an isolated atom or
form B-B bonds with an increasing degree of interconnection in the chains,
double chains, layers and frameworks and combinations thereof (Fig. 1). Due to
the strong covalent bonding between the boron atoms and the electron deficiency
of the three-center bond a number of complex and unique structures result which
1-
i
Single Multiple
x
0-0
8
0-0
p7GG-j LUnits helping to form 3-D frameworks]
sp2-Type B,-TYPe
Figure 1. Structural units in borides (after [64]).

have been the subjects of investigation for many years [1,2]. In general, compounds
with a boron-to-metal ratio of less than 1.O are built up of isolated boron atoms or
pairs with a low B-B interaction (e.g., Ni3B, Ru7B3,Fe2B, Cr5B3),in zigzag chains
with additional, isolated B (e.g., o-Ni4B3).At a ratio of 1.0 to 1.3, infinite chains are
formed which may be parallel to one or even two crystallographic axes (e.g., m-
Ni4B3, FeB, CrB, MOB), whereas in M3B4borides double chains are predominant
(e.g., Cr3B4). With increasing boron content, two-dimensional nets are stable,
yielding preferential stoichiometries between M2B3 and MB4. The most important
structure type group thereof is the A1B2 type, which is covered in more detail later
on. Three-dimensional frameworks exist in so-called higher borides with typical
stoichiometries of MB4, MB,, MBI2,and MB,,. Channels with rectangular cross-
sections were found in, e.g., CrB4 and MnB4, which is unique for the three-center
bond of boron [3]. A rigid boron skeleton consisting of B, octahedra is a character-
istic of the CaB6 structure group (important member: LaB,), whereas the UB12
structure contains BI2cubo-octahedra. Other borides of MB6 and MB12 stoichio-
metry or a higher boron-to-metal ratio, especially the main group element borides,
can be derived from the trigonal rhombohedra1 a-boron or (3-boron structure with
the B12 icosahedron as the most important structural unit. The SiB6 structure
consists of a special type of boron arrangement built up by 18 boron icosahedra
(BIZ units), 4 icosihexaedra (BI5 units) and 8 single atoms where some of the
boron positions are occupied by Si [4]. A structural curiosity are the yttrium borides
with B/Y > 25. The member being richest in boron, YB66crystallizes in cubic sym-
metry and contains 1584 boron atoms and approximately 24 yttrium atoms (Fig. 2).
The boron is arranged in 27 (B12)13 units made up by 13 interconnected icosahedra
and in 8 Bso clusters being occupied by approximately 42 B atoms. The yttrium is
arranged in Y-Y pairs of most probably Y3+ state, providing all the electrons neces-
sary to stabilize this framework [5, 61. Even this compound is of high technical
importance for monochromators of synchrotron radiation.
Some general systematics on chemical bonding and crystal chemistry have been
published by Matkovich and Economy [7], and Aronsson et al. [8, 91, who also
refer to the structural similarities in silicides and phosphides.
7.2.2.1 A1B2-Type Structures

The transition metal borides of the AlB2 structure type group are of great technical
interest for ceramics, as are the ternary a,p and T type borides as compounds for
cermets and coatings. The A1B2 structure is conveniently described as a sequence
of alternating metal and boron layers of hexagonal symmetry. The metal layers
are close-packed and stacked in an A-A-A sequence, resulting in a basal-centered
unit cell. The boron atoms are six-fold coordinated and situated in the center of
trigonal prisms of metal atoms ( H position). Hence they generate a planar primitive
hexagonal, two-dimensional, graphite-like network (Fig. 3). The total stacking
sequence is then A H A - H A H . .. and belongs to the space group P6lmmm. The
unit cell contains one formula unit, MB2. Since this structure is very versatile at
accommodating metal atoms of various sizes and electron configurations, M
could be Mg, Al, group IVa, Va, VIa, actinide or lanthanide elements.
: Y-Y pair
Figure 2. Structural units in YB66 (after [6] ).
Furthermore, other transition metal borides of various stoichiometries can be

derived from the AlB2 structure type by introducing the metal layer positions B
and C in analogy to close packings and the boron layer types K and K’, which
are slightly puckered. By allowing stacking sequences such as AHAK-BHBK-
CH CK . . . (“Mo2B5”-type), AHAK‘-BHBK‘. . . (“W2B5”-type), or AH‘AK‘-
BH‘BK‘. . . (Ru2B3-type)and AK‘BK‘. . . (ReB2-type), and vacancies in the boron
K layers, other structures and symmetries can be generated (Fig. 4) [9, 101. The
particular molybdenum and tungsten borides have formerly often been denoted
M02B5 and W2B5,respectively, but there is evidence that the homogeneity range
is narrow and close to the 1 : 2 stoichiometry [9-12,1411. High-resolution TEM
micrographs of WB2-containing ceramics show, however, that the puckered B
layer really exists regularly alternating with the plain boron layer of the A1B2 struc-
ture (Fig. 5), yielding the space group P6/mnzc. Thus, the stoichiometry range from
WB,,, [ 131 to WB2.27[ 141 arises from a boron deficiency in both kinds of layers. As
co I
0 @.
Metal Boron
A
H C
Figure 3. The AIBz structure type
an exception, A1B2-type WB2 has been produced under non-equilibrium conditions

by chemical vapor reaction of WCI6 with a boron wire at 800°C [I51 with entirely
different unit cell dimensions.
Calculation of the band structures of A1B2-type compounds shows that no
band gaps are present, and all the compounds are predicted to be electron conduc-
tors, which is in agreement with experimental results. For the main-group element
K’ A, B: metal
a1 -c A H, K’: boron
Figure 4. The Mo2BS structure type (after [ l I]).

808 7 Boride-Bused Hard Muterials
Figure 5. High-resolution TEM micrograph of WlB, stacks, view of [110] zone axis, distance of
stacks: 13.9 A (courtesy, B. Freitag [368]).
diborides the boron 2p0 and 2p7t orbitals are the main constituents of the states at
the Fermi edge, while for the transition metal diborides it is the localized metal 3d
orbitals which are the predominant component of the valence and conduction
bands. Since the boron sublattice is electron deficient all diborides exhibit an elec-
tron transfer from the metal atom to the boron, which gives rise to a strong ionic
contribution to the bonding. In the transition metal diborides, the charge transfer
decreases from 2.28 electrons in ScB2 to 1.09 electrons in MnB2 [16]; lower values
have been presented by Samsonov and Kovenskaya [17, 181. The additional elec-
trons occupy the 2p7t orbitals of the boron where the electrons are involved in
both the B-B bonding as well as the metal-boron interactions. Cluster calculations
of main group element diborides show that the metal-metal bonds are weak, the
metal-boron interaction is significant and the boron-boron interactions are very
strong. In the transition metal diborides the metal-metal bonds within the layers
are considerably stronger than in main-group diborides and reaches a maximum
for VB2. This internal bonding within the layers is clearly of a metallic type and is
thus responsible for the metallic transport properties. The metal-metal interlayer
bonding, as well as the metal-boron interactions also increase from ScB2 to
MnB2, whereas the contribution of the boron-boron bonding decreases in this
order. Due to the existence of vacancies in the boron layer and the possible occupa-
tion of interstitial sites by additional boron atoms, the boron sheets may also exhibit
some metallic or semimetallic conductivity. The considered metallic fraction does
not, however, account very much for the transport properties. In contrast, the inter-
action between metal and boron layers contains a more efficient metallic portion,
which explains the electric conductivity along the c-axis [19]. In ideal boron
7.2 Chemical Bonding and Crystal Chemistrv of Borides 809
layers, the donor capability of the metal governs the extent of electron localization
in the sp states of the boron atoms. Thus the covalent character of the B-B bonds
decreases from group IV to group VI metal diborides [20].
As it has been established that the boron network is rather rigid and governs the
lattice expansion in the a direction whereas the lattice dilatation perpendicular to the
metallic layers strongly depends on the metal species, it seems likely that the metal
atoms are distorted in some cases [21, 221. The c / a ratio is thus a function of the
Y ~ ~ ratio
~ and~ depends
~ / furthermore
Y ~ ~on the
~ valency
~ ~ electron concentration
~91.
Since the layers of the A1B2 type structure consist of very distinct kinds of
material, namely metals and boron, exhibiting entirely different binding character-
istics, more attention was recently paid to the (0001) terminating surface layers of
single crystals. Hayami et al. [23] and Souda et af. [24] have studied HfB2, TaB2,
and WB2 single crystals by impact-collision ion scattering spectroscopy at
5 x lo-' Pa inert gas pressure, which gives information on the ultimate surface
structure by atomic shadowing effects of scattered He+ and Lif ion beams at low
incident angles. As a result, the HfB2 basal plane is shown terminated by metal
atoms whereas B is entirely absent even in some layers down further. In contrast,
TaB2 and WB2 basal planes exhibit boron surface layers emerging from boron
diffusion to the surface at above approximately 500°C, and compensating for the
boron defects due to ion bombardment. This effect is attributed to the compara-
tively stable graphite-like boron layers and the higher affinity of Hf to oxygen
compared to Ta and W.
7.2.2.2 Crystal Structure of Boron Carbide and Isotopic Compounds

Boron carbide, referred to as (B4C) in brackets indicating the solid solution in
contrast to B13C2,for example, meaning the stoichiometry, crystallizes in the
trigonal-rhombohedra1 space group R3m. The unit cell is shown in Fig. 6. The struc-
ture may be described as a cubic primitive lattice elongated in the direction of the
space diagonal with almost regular icosahedra at the corners. Parallel to the space
diagonal, which becomes the c-axis in hexagonal notation, a linear chain consisting
of three atoms interconnects the adjacent icosahedra. Thus the unit cell contains
twelve icosahedral sites and three sites on the linear chain. If B atoms are attributed
to the icosahedral positions and C atoms are considered to be situated in the linear
chain, a stoichiometry of BI2C3,i.e., B4C, results. The icosahedra exhibit two topo-
logically different positions, first the B1 position (also known as 6 h l ) which consists
of a planar array of three atoms perpendicular to the linear chain around the outer
atoms. Thus, this position occurs six times in the unit cell. The second distinguish-
able icosahedral site is the B2 (or 6h2) position, which is situated in the middle of the
edges of the rhombohedra1 unit cell and accounts for a further six atoms. A special
position is the centrosymmetric B3 (or 16) site in the linear chain, which is consid-
ered to be preferentially occupied by larger atoms such as A1 and Si that form solid
solutions [25-291. For the binary solid solution series the question is whether the
linear chain is formed by a C-C-C, a C-B-C, a C-C-B, a B-C-B, or a B-B-C
array throughout the homogeneity range and how to explain the variations in the
8 10 7 Boride-Based Hard Materials
- --
Atom Symbol Site
!
trigonal
OXIS k)
B,C: @ 6h, B(1) !
B,C,Si : 0
B,C?,A1:
Figure 6. Crystal structure of boron carbide.
stoichiometry [3042]. At that time, the majority of the authors agreed that the
linear chain is of the C-B-C type and does not change with the C content, which
fits the stoichiometry of the most stable compound in the system, B13C2.Accord-
ingly, the linear chain thus contains a closed shell of ten valence electrons, which
is achieved by a charge transfer to the B12icosahedra to which 38 valence electrons
are formally assigned. The charge difference between B and C within the chain
results in even stronger electrostatic binding forces [35], but the energetic differences
estimated by density-of-state calculations are very small [43]. Since there is a
deficiency of one electron in the icosahedron, additional C as an electron donor
preferentially replaces B in one of the icosahedral sites [42, 44, 451. The B2 site
was established as the most favorable position for such a replacement [41, 43, 46-
491. Thus the total structure can be written as (BI1C)-(CBC)+.
To re-investigate the carbon-rich limit of solid solubility, boron carbide being
in eutectic equilibrium with graphite was prepared by arc fusion or diffusion couples
and subsequently analyzed with a microprobe [50] yielding a composition of B4.3C
(i.e., a maximum carbon concentration of 18.87 at.-%). Optical absorption spectros-
copy of this material indicated that 8 1.4% of the linear chains had a C-B-C structure
7.2 Chernicul Bonding and Crystal Chemistry of Boridrs 8 11
Table 1
Phase Sublattice structure Reference
and 18.6% consisted of a C-B-B array; the chains were statistically distributed [51].
Concluding from Fourier transform infrared (FTIR) spectroscopic data, Kuhlmann
rt al. [51] argue that in contrast to the generally accepted continuous substitution of
B I 2 for B l l C icosahedra and C-B-B for C-B-C chains with decreasing carbon con-
tent, the structure consists of statistical mixtures of these units. With decreasing C
content, a growing portion of unit cells without any central linear chain is formed.
Earlier, the boron-rich side of the homogeneity range was established at 8.7 mo1.-%
C (i.e., B : C = 10.4) by Bouchacourt and Thevenot (1981) [49] who assigned the
structural composition B12_\-C\.(C-B.yCI - ,-C) with 15.33 atoms per unit cell to
the resulting formula B10.4C.The density of boron carbide at the boron-rich end
is with 2.465 g cmP3lower than that at the carbon-rich corner with 2.51 g cmP3.
Together with the assessment of the binary B-C phase diagram the homogeneity
range of boron carbide and the other compounds was modeled by means of the
Compound Energy Formalism using the following sublattice models (Table 1)
where vacancies are denoted as V:
In comparison to Lim and Lukas [52] who still consider the existence of a linear
C-C-C chain, Kasper [53] favors the model taking vacancies in the chain into
account. His calculation reveal a prediction of the occupation state of the icosahe-
dral and chain position (Fig. 7) being in accordance with experimental results on
IR absorption bands and heat capacity.
As already mentioned, the linear chain can accommodate other main group
elements such as Al, Si, P, As, and 0 without a change in the structure type.
Solid solutions formed with (B4C), however, are only known for A1 and Si, which
can partially occupy one of the positions within the C-B-C chain, e.g., A1 is
placed in the centrosymmetric B3 site which causes a slight kinking of this linear
array, whereas Si replaces up to one third of the carbon sites at the ends of the
chain. The total solid solubility of both species is comparatively small (max. 2.5
at.-%) [28, 541, although a complete solid solution series between BI2(C,B, Si),
and the silicon boride “SiB4” was considered [55]. Due to the large size of the Si
atom, however, the formation of a two-atom chain is favored, as in the cases of
P, As, and 0. The isotopic binary borides of these elements thus fit into the general
stoichiometry B12(XVX). This arrangement favors a charge transfer to the BI2
icosahedron. The ideal structure of BI2O2 may be written as (B12)2-(00)2+,
which exactly yields a filled-band configuration. In reality, for all these compounds
a considerable disordering has to be taken into account, which results in the compo-
sitions B12P1,8, B12A~2.0,and B1201.82 (corresponding to B6.60)[25]. In the case of
“SIB4”, the real stoichiometry is (B10,4Sil.6)(SiSi) = SiB2.89,which is attributed to
a Si-Si chain and two sites in the icosahedron being partially occupied by Si
[7, 21, 561. Sublattice structure modeling carried our by Kasper [53] to establish
XC
Figure 7. Calculated occupation of sites in the icosahedra and the chain of B4C solid solution [53].
the ternary B-C-Si system considered that the Si-Si chain may also at least partially
be the structural unit for the Si-containing boron carbide solid solution [Eq. (l)]:
B4+8C(Si) = (B, C)12(C2B,CB2, B2V, Si2) (1)
The solid solubility limit was then simulated by a statistical combination of all the
kinds of chain occupancies known.
7.3 Phase Systems

Knowledge of the phase diagrams for compounds of technical interest and of the
environmental phases in contact with these compounds is the key for materials
development and for the understanding of materials behavior in application. Not
only can the thermal stability of particular phases be calculated by means of
thermodynamic data, but suitable sintering procedures can also easily be
considered, and decomposition in aggressive media can be predicted. Generally
recommended data books on binary and ternary systems are, e.g., those by
Hansen [57], Elliott [58], Shunk [59], Mofatt [60, 611, Massalski [62], and Petzow
and Effenberg [63].Nowadays, the thermodynamic data of most of the important
phases are available in publications or databases and can be readily used for thermo-
chemical calculations.
In the following paragraphs, phase diagrams of the most important boron-
containing phases will be presented and discussed, starting with the binary systems,
7.3 Phase Systems 8 13
then selected ternary systems which are of technical interest. In the subsequent
sections particular phase systems will be treated in respect to sintering of B4C and
TiB2 or in the context of microstructural design and mechanical strengthening.
7.3.1 Binary Phase Diagrams of Technically Important Systems

Many attempts have been made to correlate the binary metal boride phase equili-
bria, the boride crystal chemistry, and the ranking of the elements in the Periodic
Table [64]; larger metals and those with unfilled d-shells favor the formation of
boron-rich phases with two- or three-dimensional boron frameworks. The smaller
metals having a high number of d-electrons prefer the formation of metal-rich
phases with only a few boron-rich bonds, whereas more noble metals such as Ru,
Rh, and Pd tend to generate defect structures. Systematic work on crystal structures
and stabilities has been performed by Kiessling [l], Aronsson et al. [8], and Lund-
strom [2, 651. Spear [64, 661 presented a systematic compilation of binary phase
diagrams resulting in predictions of phase relations. Guillermet and Grimvall
[67,68] systematized thermodynamic data of transition metal diborides with empha-
sis on the enthalpy of formation, the vibrational entropy, and the melting tempera-
ture, in order to account for the transition from stable to metastable phases with
increasing atomic number. As a result, an entropy-related free energy term was
introduced which correlates linearly with cohesive energies and melting points. By
this means, stabilities of boride, carbide, and some nitride phases have been success-
fully predicted. In the following sections some binary systems of technical interest
are presented.
7.3.1.1 The B-C System

In contrast to early publications by Samsonov and Schuravlev [69] and Schuravlev
and Makarenko [70] considering several boron carbide phases it is generally
accepted today that only one binary phase B13C2* exists with a wide homogeneity
range of 8.8 to 20.0 at.-% C, depending on temperature. This phase melts congru-
ently at 2450°C [58] at the composition B13C2(18.5 at.-% C, 20.2 weight-% C). For
the B-rich corner of the phase diagram, Bouchacourt and Thevenot [71] proposed a
degenerated peritectic with elemental boron at 2075"C, according to measured ele-
ment distribution coefficients. In this diagram the melting point of boron is placed at
2020°C. Since the melting point of B accepted today is 2092°C the resulting reaction
with boron carbide should be an eutectic one, assuming that the non-variant equili-
brium at 2075°C is correct. The maximum carbon content is usually given as 20.0
at.-%, corresponding to the stoichiometry of B4C. Beauvy [72] suggested a
carbon content steadily increasing with temperature from 21.4 at.-% (20°C) to
23.1 at.-% (2375°C). Recent microprobe analyses by Schwetz and Karduck [50]
indicated, however, that the maximum carbon content of fused boron carbide
being in equilibrium with graphite is only 19.2 at.-% at the eutectic with carbon,
and 18.5 at.-% at IOOO"C, corresponding to the formula B4.3C. The eutectic with
8 14 7 Boride-Based Hard Materials
2600
2400
- 2200
2
- 2000
a,
L
3
3 1800
L
a,
1600
a,
I-
1400
1200
1000
B 10 20 30 40.0C
(a) -C (atom-%)-
2800
I;)
t 2400
/
-
! ! l 2000
a
E
F 1600 Boron +
boron
carbide
1200
8.6 1 18.81 I I
B 10 20 30 40
(b) Carbon, atom% ---+
Figure 8. The B-C phase diagram (a) calculated, (b) according to Schwetz and Karduck [50].
carbon is given at 2375 f 5°C and 29-31 at.-% C, which is in a good agreement with
thermodynamic calculations [53, 731 stating 29.1 at.-% as the eutectic composition.
The phase diagram is presented in Fig. 8 comparing the calculated version (Fig. 8a)
with the one (Fig. 8b) suggested by Schwetz and Karduck [50]. The reaction of
boron carbide with elemental boron was modeled as a eutectic at 0.2 at.-% C and
2073"C, i.e., 2 K below the melting point of boron.
Table 2. Experimentally observed and calculated reactions in the B-Si system.
Reaction type Temperature “C
Calculated Experimental
Diirner Lim and Kasper Armas Olesinski and Telle
-+-
[78] Lukas [52] [53] [82] Abbaschian [80] [83]
SiB, 0-B + liquid 2020 2020 2054 1929 2020 2060
-
SIB, SIB, + liquid 1898 1850 1850 1989 1850 1900
SIB, (Si)sol,dH liquid 1385 1384 - 1385 1385
SiB, SiBb + liquid 1377 - - 1377 -
SiB, + (Si) H liquid 1340 - 1340 -
++ +
SIB3 (Si),,l,, SiB6 - 1270 1198 1270 I358
7.3.1.2 The B-Si System

The B-Si system is of particular importance for the understanding of Si as a sintering
aid for boron carbide because of its chemical relationship to B and C since hot
pressing of Si bearing boron carbide results in a significant reduction of the grain
size and, therefore, in an improved fracture toughness and strength of 500-
600 MPa [74, 751. Moreover, the silicon borides have been periodically studied for
their thermoelectric and thermomechanical properties.
Uncertainties in the interpretation of the system arise from the three silicon
borides, SiB2.89-3.65, SiB6, and SiB12-14.Both the homogeneity range and the decom-
position temperatures have not yet been completely established (e.g., Elliott [58];
Ettmayer et al. [76]; Lugscheider [77]; Dorner [78]; Olesinski and Abbashian [SO]).
Experimental problems in both binary and ternary phase studies are related to
the comparatively high vapor pressure of Si at temperatures exceeding 1400°C.
The Si-richest SiB2.89-3.65, often also denoted SiB, or SiB4, can be derived from
the B4C structure by an arrangement of B12p\-Si,y icosahedra and a Si-Si chain. It
is not known for certain whether “SiB3” decomposes peritectoidally at 1270°C to
form SiB6 and solid Si or shows a eutectic reaction with Si at 1340-1385°C and
decomposes peritectically to SiB6 and a boron-rich Si melt at 1377°C.
SiB6 crystallizes in an own structure type of space group Pnnm, which is built up
by 18 boron icosahedra (BI2units), 4 icosihexaedra (BI5units) and 8 single atoms
where some of the boron positions are occupied by Si [4]. In spite of this versatility
in B-Si exchange, its solid solubility for Si and B is generally considered less than 1
at.-%. It is agreed upon that SiB6 reacts peritectically to the next higher boride
SiB12-14and a Si liquid containing 65.3 at.-% B.
SiB, with n = 12-14.. .23 comprises a homogeneity range between 93.3 and 97 at.-%
B and can structurally be derived from the P-B structure. The close similarity to elemen-
tal boron led formerly to the conclusion that SIB, is in fact a Si-rich solid solution of
boron. SIB, undergoes a peritectic reaction with B containing 3 at.-% Si and liquid con-
taining 9.3 at.-% Si at 2020 f 15°C as compiled by Olesinski and Abbashian [80] and
calculated by Lim and Lukas [52]. Table 2 presents the experimental and calculated
data of the various reactions, Figs. 9a,b show the corresponding types of phase diagrams.
2200
1200-
- SiB, SiB, + Si
-
1000 , I ' I ' I ' I ' I ' I I I '
Si
X 60Bro [3BlMal
A BOVia X &Via
0 .2 .4 .6 .0 1.0
B Molar Fraction Si
Figure 9. (a) The binary B-Si system according to experimental results with powders of technical
purity. Homogeneity fields of the silicon borides according to Ettmayer et al. [76], Lugscheider
et al. [77], Armas et al. 1981 [82]. (b) The binary B-Si system according to recent calculations by
Lim and Lukas [52].
7.3.1.3 The TCB System

The most recent compilation on the Ti-B system was published by Murray et a/.
[84]. The assessed phase diagram (Fig. lo), being in good agreement with thermo-
dynamic calculations, consists of three intermediate phases, orthorhombic TiB
(FeB type structure), orthorhombic Ti3B4 (Ta3B4 structure), and hexagonal TiB2
(AlB2 structure). While TiB and Ti3B4 decompose peritectically at 2180 and
22OO0C, respectively, TiB2 melts congruently at 3225 f 25°C [85]. TiB has a
narrow homogeneity range of about 49-50 at.-% B [86] and reacts eutectically
with Ti solid solution at 1540°C and 7 at.-% B. The existence of the Ti3B4
phase was proven by Fenish [86] and its peritectic reaction with TiB2 and liquid
was placed at 2020°C. Rudy and Windisch [85], however, omitted this phase
from their binary diagram, probably because it could not be observed in melt-
derived samples close to the decomposition point of TiB. In 1981, Ti3B4 was re-
investigated by Neronov et al. [87] in reaction layers between Ti and B; later, in
1986, it was confirmed by Spear et al. [88] by arc melting and annealing
studies. TiB2 reacts eutectically with elemental boron at 2080 f 20°C and
approximately 98 at.-% B. Batzner [89] re-calculated the system with slightly dif-
ferent liquidus lines and omitting the little solid solubilities of TiB and TiB2,
respectively, reported by Murray et al. [84]. The congruent melting point of
TiB2 was set to 3216"C, the eutectic between TiB2 and B to 2059°C instead of
*
2080 20°C after Murray et al.
7.3.1.4 The Zr-B System and Other Transition Metal-Boron Systems

Similar to the Ti-B system, ZrB2 is an important phase having an A1B2-type
structure and a melting point of 3250°C (Fig. 11) [62]. It reacts eutectically with
elemental Zr at approximately 1680°C and 86 at.-% Zr. In contrast to the Ti-B
system, no ZrB or Zr3B4phases exist, but there is a ZrB12phase with a UB12struc-
ture which melts incongruently at 2030°C forming ZrB2 and liquid. ZrBlz forms a
eutectic with elemental boron at approximately 1990°C which is not yet firmly
established.
Figure 10. The Ti-B phase diagram (after [84]).

m a ss % Zr
03050 70 80 90 100
---
0 10 20 30 LO 50 6 0 7 0 80 90 100
B at.% Zr Zr
Figure 11. The Zr-B phase diagram (after [62]).
In the other transition metal boron phase systems of groups 111, IV, and V, the
MB2 phase is also the dominating compound with respect to the high melting
point. Exceptions are the Y-B, lanthanide metal-B and actinide metal-B systems,
which possess very stable MB4- and MB6-type compounds. In the Y-B system
additional higher borides of the stoichiometry YBI2, YBZ5,and YB6, exist. In
group V the number of known phases with a B/M ratio < 2 increases, and the
tendency for the MB phase to be more stable than MB2 is obvious on advancing
from V to Ta. Destabilization of the A1B2 structure to the benefit of the MB struc-
ture is also evident in group VI where the MB2 phase forms an individual structure
type that can be derived from the A1B2 structure by the introduction of variations in
the stacking sequences. In the VIIth and VIIIth group metal-boron systems the
melting points of the borides decrease becoming significantly lower than the melting
points of the elements. A summary of the known boron-containing binary systems is
given by Spear [64].
7.3.2 Ternary and Higher Order Systems

Many ternary, quaternary and higher systems containing borides have been inten-
sively investigated basically for three reasons: firstly, to elaborate suitable sintering
systems for these high-melting and thus difficult to densify compounds; secondly, to
avoid probable chemical complications such as phase changes and decomposition
during application; and thirdly, to investigate ways of optimizing materials
properties by the fabrication of tailored composites or solid solutions. Most of
these investigations have been concerned with military-, nuclear- or aerospace-
related research for new high-temperature materials, fabrication of cutting tools
(e.g., transition metal borides for hard metals and cermets), or wear resistant
parts (high-strength and high-toughness structural ceramics based on composites).

For the last decade, more systematic studies related to the edition of alphabetic
volumes on ternary systems, or to the investigation of peculiarities of chemical
bonding in multicomponent phases, have also been started. Hence the data available
can be divided into three groups: boron-carbon-metal/semi-metal systems (basi-
cally ceramics: sintering systems, composites), transition metal boride systems
with low melting metals (for densification of hard metals, cermets and other cutting
tools), and systems with two transition metals and boron (development of solid solu-
tions of exceptional properties or with emphasis on the substitution of tungsten and
other strategic metals). In the following sections some ternary systems are selected as
technically important examples of these three categories.
7.3.2.1 Boron-Carbon-Metal Systems

Aluminum is an effective sintering aid for B4C and Sic ceramics if combined with
elemental boron and carbon. Phase relations in the B-C-A1-Si system may hence
indicate suitable procedures to initiate transient liquid phase or enhanced solid-state
sintering. Furthermore, liquid A1 may be used to infiltrate porous B4C bodies acting
as a reinforcing phase.
Although the binary boundary phase diagrams of the AI-B-C system are rather
well established there is only limited experimental information on the ternary equi-
libria [73, 901. Six ternary phases have been discovered, B4,,A1C4 and B48A12C8,
which have a B4C structure and thus are probably B12(B,C,A1)3solid solutions,
orthorhombic BS1A12C8,hexagonal and orthorhombic B4*AI3Cs,and hexagonal
B4A18C, denoted T. The temperature stabilities of these phases are not known.
The ternary solid solubility of B12(B,C,A1)3was discovered by Lipp and Roder
[27] and described in more detail by Neidhard et al. [32]. An isothermal section
calculated by Lukas [73] neglecting all boron-rich ternary phases except T is pre-
sented in Fig. 12, stating that B12(B,C,A1)3is in equilibrium with a boron-rich A1
melt at above 1000°C.
The B-C-Si system was first treated by Kieffer et al. [91]. The experimental data
indicated a ternary equilibrium between B4C, SiB6 and SIC up to temperatures
exceeding 1900°C.
Thermodynamic calculations by Darner [78], Lukas [93] and Lim and Lukas [52],
however, clearly demonstrated the existence of a binary phase equilibrium of boron
carbide and a Si- and B-containing melt above 1560°C. The theoretical results were
confirmed by hot pressing, liquid phase sintering and infiltration experiments by
Lange and Holleck [75], Telle [83], Telle and Petzow [94], and Telle [54], which
also yielded more details on the extension of the homogeneity field of boron carbide
towards the Si-rich corner of the system B-C-Si.
Discrepancies between previous and recent experiments as well as calculations
arise from the binary B-Si system, in particular from the various plausible equilibria
at the Si-rich corner. Furthermore, experimental problems in the ternary phase
studies are related to the high Si vapor pressure at > 1400°C.
The characteristics of the B-C-Si system as assessed today, comprise the stability
of a BI2(B,C,S& solid solution with a maximum of 2.5 at.-% Si [28, 82, 951 being
Mol-% A1
Figure 12. The B-C-A1 system at 1400°C (after [73]).
in equilibrium with a boron-saturated silicon melt [54, 751. No ternary phase

exists. The unlimited solid solubility between B4C and SiB2.89,as postulated by
Meerson et al. [55], could not be verified. According to Secrist [96] and Shaffer
[97], the B4C-SIC section is of a quasi-binary type with a eutectic equilibrium
between 2250 and 2420°C and 3&35mol.-% Sic. This section is, however, of a
real ternary type since the eutectic melt postulated by Secrist and Shaffer is in
fact a binary equilibrium between solid carbon and Si-B-C liquid. The formation
of a solid solution of B12(B,C,Si)3is accompanied by the precipitation of P-SiC
which melts eutectically with SiB6 and residual solid Si above 1380°C. As a Si
incorporation into the boron carbide lattice always results in the release of C or
the simultaneous formation of Sic if Si is present in excess, it may be concluded
that Si substitutes for C assuming that there is no carbon in boron carbide on
interstitial sites. This assumption is supported by X-ray analysis and electron
energy loss spectroscopy in TEM. Lim and Lukas [52], and Kasper [53] have
refined again the B-C-Si system by thermodynamic calculations. A consistent
data set is now available which reproduces the maximum solid solubility of
boron carbide for silicon and the peritectic decomposition of the particular silicon
borides. The calculated phase diagram shows some differences to that in [54], aris-
ing from the new information about the B-Si system where now the eutectic
+ +
reaction SiB6 Si H liquid is preferred rather than the SiB3 Si H liquid
(Fig. 9b and 13). This means that SiB3 is formed peritectoidally by
-.
W
m+ @--
; SiB, + Si
w
B I
z
w
W SiB3 + Si
Figure 13. Two possible view of the B-Si system (after [53]).
+
SiB3 H SiB6 Sisolid.Fig. 14 shows a calculated isothermal section of the system
at 2000K according to Lim and Lukas [52].
The ternary B-C-Ti system was intensively studied by Rudy et al. [99] and has been
considered for the fabrication of ceramic cutting tools [loo-1031. The most recent
critical assessment was published by Duschanek et al. [104, 1051 who, in contrast
to the previous authors, take the re-established Ti3B4phase into account. No tern-
ary phases have been discovered. The ternary solid solubility of the particular binary
compounds except TiC0.81B0.17 is generally less than 1 at.-%. TiB2 coexists with
822 7 Boride-Based Hard Muteriuls
S Ji
C
(B4C) Molar Fraction C
Figure 14. Isothermal section of the B-C-Si system at 2000 K (calculated by Lim and Lukas [52]).
T i c l --x in a quasi-binary eutectic equilibrium at 2620 f 15°C (calc.: 2643°C) and

57 f 2mol.-% Ticl --x (Fig. 15, [99]; calculated by Duschanek et al. [104, 1051).
TiB2 also forms a quasi-binary eutectic with B4C at 2310°C (calc.: 2381°C) and
88 f 3mol.-% B4C, as well as with C at 2507°C (calc.: 2456°C) and 33 f 3 at.-%
C. This means, that T i c is not stable in the presence of B4C but reacts to form
+
TiB2 C. On the other hand, TiB2-B4Ccomposites can be fabricated from TIC
Table 3. Experimentally observed and calculated reactions in the B-C-Si system.

Reaction type Temperature "C
Calculated Experimental
Dorner [78] Kasper [53] Secrist [96] Telle [83]

~~ ~ ~~ ~~ ~~~~
B4C + S i c H C + liquid 2166 2286 >2250 n.d.

+
B4C SIB, ++ B + liquid 2058 2038 n.d.
B4C + SiB6 H SiB, + liquid 1896 1840 1900
S i c + SiB, ti B4C + liquid 1560 - I560
B4C + Sisolldti Sic + liquid 1396
Si + SiBs + Sic ++ liquid
~
1380 - 1380
si + SiB6 H SiB, + sic 1345 - 1345
I I I I
3500 -\
3000 -
2500 -.
s
TiB2+TiC ,.x
2000 -
1500 -
Ti, B, +TIB,+TiC,.,
1000 I I I I
A o TiB,
0.1 0.2
xc
0.3 0.4
TIC
Figure 15. IsoplethTiB2-TiC1-,(calculated by Duschanek ef al. [IOS]).
and B by reaction sintering. Composites of T i c , --x and TiB2 have been investigated
for coherent grain boundaries [103]. Furthermore, Tic, --y is stable together
with TiB up to approximately 2160°C (calc.: 2113°C) and then decomposes to
Ti3B4 and liquid. The previously established ternary peritectic at 21 60°C
+ +
(TiB2 T i c l p.y + TiB liquid, [99]) has also to be replaced by three transition
equilibria due to the existence of Ti3B4. Table 4 presents the ternary reactions
accordingly [ 1051.
Table 4. Ternary Reactions in the B-C-Ti system [104,105].
Reaction Type Reaction Experimental Calculated
-
temperature ["C] temperature ["C]
quasibinary eutectic TiB, + TIC, - liquid 2620 2643

TiB2 + C tf liquid 2507 2456
~
quasibinary eutectic
quasibinary eutectic liquid H TiB2 + B4C 2310 2381
eutectic liquid ++ TiB, + TIC, -, + C 2400 2394
eutectic liquid TiB2 + B4C + C 2240 2246
transition liquid + TiB2 tf Ti3B4+ T i c l 1; 2180 2121
transition liquid + Ti3B4 tf TiB + TIC, -, 2160 21 13
eutectic TiB2+ B4C,,+ B tf liquid 2016 2058
eutectic TiB + T i c , ~~z+ PTi u liquid 1510 1535
transition PTi + T i c , -,tf TiB + aTi 890 927
1.0 *
pTitTiBtTiC l.x
Figure 16. Isothermal section of the B-C-Ti system at 1400°C [105].
Figures 16 and 17 show isothermal sections of the Ti-B-C system at 1400°C and
21OO0C, respectively [104, 1051. The diagrams based upon the work of Rudy et af.
[99] are wrong below 2100°C as they neglect the Ti3B4phase. Technically interesting
isopleths are shown in Fig. 18 (TIC-B), Fig. 19 (TiB2-B4C),Fig. 20 (B4C-Ti), and
Fig. 21 (B4C-Tic). The equilibria of the various titanium borides with TIC1-,
have been studied in more detail by Brodkin and Barsoum [106].
The quaternary B-C-Si-Ti system has not yet been established completely.
Evidently TiB2 and Sic form a eutectic and can be cast after arc-melting
[107]. TiB2-BI2(B,C, Si)3-SiC composites are chemically stable and may form a
quasiternary eutectic. Composites of these three hard materials can be prepared
either by direct hot pressing or by reactive infiltration of porous B4C bodies with
an Si-TiSi2 eutectic melt [108, 1091. Of technical interest is the ternary boundary
system C-Si-Ti studied by Brukl [I lo], Borisova et af. [l 1 I], and Holleck [112],
calculated by Touanen et al. [ 1 131 and experimentally re-examined by Wakelkamp
et af. [114].
7.3.2.2 Ternary Systems with Boron and Metals

High-temperature equilibria of the extraordinarily hard borides with metallic
melts bring about the opportunity for a pressureless liquid phase sintering and
the fabrication of hard and simultaneously tough composites similar to hard
metals but are also of interest in ceramic systems or coatings. In this section
emphasis is put on binary and ternary borides which are in equilibrium with transi-
tion metals.
825
7.3 Phase Systems
1.o
1 .o
C
B
Figure 17. Isothermal section of the B-C-Ti system at 2100°C [105].
0.8
0.8
I
Mole fraction C
0.6
0.6
I
X.B
L B
Figure 18. Isopleth along Tic-B [105].

TIC,.,
0.4
I
0.2
0.2
I
Tio.5Co.5
Ti
0
I
3500
A
1 .o
Ti
-
0.8
I
0.6
Ti
I
TiB, +B4C
Figure 20. Isopleth along B4C-Ti [lo51

0.4
I
0.2
A o
2500 -
2000 -
1500 -
.
826
+
Y
3500 i 7
L+Bd C
Figure 21. Isopleth along B,C-Tic [105].
The Ti-Fe-B System

The fabrication of TiB2-based cermets was recently achieved by using Fe as a
binder phase [I 15-1 191. The ternary Ti-Fe-B system was studied first by Federov
and Kuzma [120] who established that no ternary phases exist. Below llOO°C,
TiB2 is in equilibrium with FeB, Fe2B, a- or y-Fe, Fe2Ti, and FeTi, whereas
TiB is stable together with FeTi and a-Ti (Fig. 22). Although there are still
some controversies concerning the phase diagram at higher temperatures [ 121-
1241, TiB2 is in an eutectic equilibrium with liquid Fe at 1340°C (eutectic concen-
tration 6.3 mo1.-% TiB2), which enables liquid phase sintering. Discrepancies exist
for the phase equilibria at temperatures between 1100 and 1300°C because of the
problem of whether the observed, undesired brittle Fe2B is an equilibrium phase or
results from impurities in the starting materials used (Figs. 23 and 24). It is, how-
ever, obvious that oxygen and carbon contaminants introduced by the manufac-
turing processes of the starting powders significantly affect the composition of
the liqyid phase by the precipitation of Ti-rich oxides and TIC, respectively.
Since the solid solubility of B in 6-Fe is less than 0.5 at.-%, any slight Ti deficiency
of the liquid phase composition will move the overall composition from the binary
TiB2-Fe equilibrium into the ternary TiB2-Fe-Fe2B field. Both constituents,
oxygen and carbon, do, therefore, indeed cause dramatic changes of the sintering
kinetics [119].
Figure 22. Isothermal section of the Fe-Ti-B system at 1050°C (calculated by Golczewski and
Aldinger [320]).
The Ti-Ni-B System

In order to achieve lower sintering temperatures, also many other metallic additives
such as Ni, Co, and Cr, or borides of these elements have been used at higher
concentrations to allow liquid phase sintering of TiB2. These transition metals
1500
2 1300
E
a,
E
a,
I- 1200
1100
0 20 40 60 80 100
Concentration TiB,, Mol0/o

Figure 23. Isopleth of the Fe-TiB, section [121].
829
100
7.3 Phase Sq’stems
Figure 25. Isothermal section of the Ti-Ni-B system at 800°C in at.-% (after [125]).
T: Ni,,Ti,B,
80
Concentration TiB,, Mol%

Figure 24. Isopleth of the Fe-TiB, section [124]. both 1992 refs?
20
0 60
40
1100
1200
1300
1400
1500
Ti
0
I ' ' ' I ' 1 ' I I 1 ' 1 " ' 1 ' 1 I ' I I
20 40 60 80
TiB, a t % Ni Ni
Figure 26. Isopleth of the Ni-TiB, section (after [127]).
react forming various metal borides with a low melting point (approx. 90&1 100°C)
and a suitable wetting behavior.
In contrast to the very convenient Ti-Fe-B system, the Ti-Ni-B phase diagram
contains a congruently melting ternary phase, denoted z, with a composition of
Ni21Ti2B6and a Cr23C6structure (Fig. 25) [125-1271. At 800°C, the z-phase is in
equilibrium with Ni, Ni3B, Ni3Ti, and TiB2. A pseudo-binary eutectic with Ni
exists at 1077 5"C, whereas the relationships in the TiB2-rich corner are more
complicated. An isopleth (Fig. 26) across the line Ni-TiB2 reveals a solid-state equi-
librium below 980°C involving TiB2, z,and Ni3B [127]. A liquid phase forms above
that temperature because of the decomposition of z. Above llOO°C, Ni3B also
decomposes completely in the presence of TiB2, which increases the amount of
liquid phase and thus accelerates the densification.
The Boron-Transition Metal Systems of Group (IV-VI) and Group VIII Elements
Cemented borides with a metallic matrix have also been fabricated successfully from
the ternary transition metal borides of so-called z-, cp- and a-types since these com-
posites can easily be liquid phase sintered with metallic melts.
The z-phase with a general composition of Mi1Mi'B6 or (MI, Mr1)23B6 has been
observed in ternary systems with M' = group VIIIB elements such as Fe, Co, or Ni
and M" = group IVB-VIB elements such as Ti, Zr, Hf, Nb, Ta, Mo or W, and with
M' as the liquid phase [125-1281. It forms in coatings or particulate reinforced
metals as a consequence of the reaction between diborides such as TiB2, ZrB2,
Figure 27. Isothermal section of the B-Co-Mo system at 1000°Cwith indicated field of liquid phase
formation [128].
HfB2, NbB2, and TaB2 and Ni-Co-Cr based alloys. The refractory and extra-
ordinarily hard MB2 phases are hence in equilibrium with a comparatively soft
and brittle ternary compounds.
The ternary systems of these kinds of metals with boron reveal a more complex
structure because of the presence of many other ternary phases denoted to as p-
and o-phases. The stoichiometries of these p- and o-phases are M'MI'B and
M;M"B2, respectively, where M' represents Cr, Mo, Ta, or W and M" holds for
Fe, Co, or Ni and solid solutions thereof. Other ternary phases have the composi-
tion and M;M1'B2\-, e.g., TaNiB2, Mo2FeB4,and Mo3CoB6.As an example, an iso-
thermal section of the B-Co-Mo system is shown in Fig. 27 in which both the t- and
the p-phases are linked with Co as the binder [128]. However, in systems with Fe
replacing Co, a p-phase does not exist. Hence o is in equilibrium with liquid
metal and is thus likely to form a cermet material with Fe like the z-phase
Mo2Fe13B5(Fig. 28). Phase compositions located in the pseudo-binary equilibria
with a metal can easily be pressureless liquid phase sintered at temperatures between
1500°C and 1700°C. Wear-resistant parts have been developed from Mo2FeB2-Fe
cermets with Ni or Cr additives [129-131, 3071. Figure 29 presents an isothermal
section of the Ni-Ta-B system at 950°C [126] with three ternary phases where
only z is in equilibrium with metallic Ni.
Quasibinary Systems of Ti, Cr, and W Diborides

Boron-based ternary phase diagrams with two transition metals have been investi-
gated for exploitation of the diborides with an A1B2 structure, which have high
hardness and a high melting point. Due to the identical crystal structure, most of
the transition metal diborides have been considered of high mutual solubility
Figure 29. Isothermal section of the Ta-Ni-B system at 950°C (after [126]).
Ta
Figure 28. Isothermal section of the Mo-Fe-B system at 1000°C [128].
100
80
Mo
TapNi
60
YoTa
TaNi
A
B
At
40
B
Mo,Fe,
TaNi,TaNi,
,
Ni
J
m-Ni4B3
o-Ni, B,
20
8
p"
"O
\
0
832
Ni
Fe
[ 132-1 361. Precise experimental data are, however, rare or are not readily available.
Makarenko [137] simply mentions that most of the transition metal diborides of
group IV-VI are fully soluble between 2000 and 3000°C and probably also at
room temperature. As the only limiting factor the difference in atomic radii was con-
sidered which should not exceed 15% like in case of V and Cr, and Hf and Zr [132].
This general statement should be taken with care since low-temperature miscibility
gaps have been observed for (Cr,Nb)B2 and (Cr,Ta)B2 [133, 1381 and other quasi-
binary systems were proved of eutectic character. Complete solid solubility with
an miscibility gap at lower temperatures was also proven for the TiB2-CrB2
system, limited low-temperature boundary solubility and large homogeneity
ranges at high temperatures have been observed for the TiB2-WB2, and CrB2-
WB2 systems which will be discussed in more detail.
Studies in the TiB2-CrB2system are difficult due to the comparatively high vapor
pressure of Cr and the little densification of powder blends at temperatures below
2000°C. Hot-pressing at 2000°C for 3 w 2 0 min yields a homogeneous (Ti,Cr)B2
solid solution [136]. Koval’chenko et al. [139] have also studied the sintering
behavior of pre-reacted powders and the physical properties of a (Ti,Cr)B2 solid solu-
tion containing 20 mo1.-YOCrB2. Klimenko and Shunkowski [ 1401 have investigated
the activated sintering of a “titanium-chromium mixed boride” at 2000 and 2200°C
being obviously a completely single phase. Unfortunately, no X-ray nor chemical ana-
lyses have been reported for the particular annealing steps. In spite of the problems
upon obtaining equilibrium conditions in the overall material chemical and X-ray
analyses by Telle et al. [I361 yielded evidence for the existence of an immiscibility
gap below 2000°C in the TiB2-rich corner. In the temperature-concentration range
between 1500°C and 2000°C and correspondingly 0 to 55mol.-% CrB2 no solid
solutions possessing an intermediate CrB2 concentration were observed whereas
homogeneous (Cr,Ti)B2 particles with CrB2 contents exceeding 55 mo1.-% at
1500°C could be detected in contact with (Ti,Cr)B2 phase of a CrB2 content less
than 2mol.-Yo. Samples with overall concentrations of 35, 75, and 90mol.-% CrB2
have been treated by dilatometry and hot-pressing up to 2200°C to prove the
formation of a liquid phase but no melting could be monitored at temperatures
lower than the melting point of CrB2. This indicates that the quasi-binary system
does not contain a liquidus temperature minimum nor is of eutectic character.
From the equilibrium limits a room temperature solubility of TiB2 in CrB2 to 15-
20mol.-%0 is estimated. The melting points of TiB2 and CrB2 in Fig. 30 have been
adopted from Rudy and Windisch [85] and Liao and Spear [142]. These results
confirm the existence of a continuous solid solubility above 2100°C as stated by
Post et al. [132]. Zdaniewski [I351 also reported on properties of a continuous series
of (Cr,Ti)B2 solid solutions, however, without giving details on the temperature range.
Post et al. [132] have investigated the CrB2-WB2 system at 2 1 0 0 f 100°C but
note that no exact limits of the homogeneity range could be determined. Telegus
and Kuz’ma [ 1431 studied 15 compositions in the quasi-binary CrB2-WB2 section
represented as a part of an isothermal section of the B-Cr-W system at 1500°C.
They report a maximum solid solubility of 6 at.-% Cr in W2B5 corresponding to
3mol.-% CrB2 in WB2 whereas the published section shows 16.7mol.-% CrB2
being dissolved in the W2B5 structure. This work allows an extrapolation of
3 500
..
0 10 20 30 40 50 60 70 80 90 100
Ti62 Mole - O h CrBz CrBz
Figure 30. The quasi-binary TiB2-CrB2system.
the homogeneity range of (W, Cr)B2 to about 89 mo1.-Yo CrB2 or less at 1500°C.
Moreover, in contrast to Rudy [104, 1441 the binary homogeneity range of the
“W2B5” phase is more extended. In the CrB2 corner Telegus and Kuz’ma report
less than 2at.-% W in solution which could not be confirmed in the materials
investigated by Telle et al. [ 1361who synthesized the system up to 2070°C. Unfortu-
nately, the liquidus equilibria were not established, so the accurate composition of
the eutectic is not determined yet. At temperatures between 1750°C and 2000°C
samples with overall compositions between 40 and 86 m0l.-YOWB2 consist of two
phases, namely (Cr,W)B2 solid solutions with WB2 concentrations up to 38-
40mol.-% and (W, Cr)B2 phases containing always less than 9-14mol.-% CrB2.
These results allow the formulation of the boundaries of the homogeneity fields
2500 -
Liquid 2365
-_____ -- -
-
*.-;:
,,
I
; ; 2200 -----==--------__
- I
,
’
0
-W
L
2000 - 2030250
3
z
c
W
a
E,
1-
1500 - .. (Cr,W)B, + (W,Cr)B,
1000- I , I I I . I I ,
0 10 20 30 40 50 60 70 80 90 100
CrB2 Mole -YOWB, WB2
Figure 31. The quasi-binary CrB2-WB2system.
3500
3225
- 3000
-
0
0
f 2500
,2365
Y
I
2000
+W
1500 / (Ti.W)B, + (W,Ti)B,
.L AIB,-type W2B,-type *
1000
0 1 0 20 30 10 50 6 0 70 80 90 1 0 0
TIB, Mol% WB, WB,
Figure 32. The quasi-binary TiB2-WB2system.
of (Cr, W)B2 and (W, Cr)B2 phases with AlB2 and W2B5 structure, respectively. As
shown in Fig. 31, the CrB2-WB2 system is of a simple eutectic type
(T, = 2030 f 50°C) with a solid solubility of CrB2 in WB2 of 1Omol.-% and WB2
in CrB2 of 37mol.-% at eutectic temperature. The tungsten content in CrB2
decreases strongly with decreasing temperature whereas the chromium content in
the W2B5 structure is almost temperature independent.
The quasibinary TiB2-WB2 system as a part of the Ti-W-B system has been
treated by Telle et al. [136] up to 2250°C. The borders of the (Ti, W)B2 homogeneity
range have been intensively studied between 1500°C and 17OO"C, at 2000°C and
around the quasi-binary eutectic temperature. Thus a huge solid solubility up to
approximately 63mol.-YOWB2 could be established. In the W-rich corner of the
phase system, the TiB2 concentration in (W, Ti)B2 is with 1-2 mo1.-% almost con-
stant up to 1700°C. At higher temperatures the solid solubility of the W2B5-type
structure for TiB2 increases slightly to approximately 3 mo1.-% at the eutectic tem-
perature. A further decrease in TiB2 along the supersolidus equilibrium could be
established up to 2250°C as expected according the phase rules. Taking inaccuracies
in the temperature measurement of the dilatometric experiments into account, TiB2
*
and WB2 react eutectically at T, = 2230 40°C and 90 f 3 mo1.-% TiB2 (Fig. 32).
The solid solubility of WB2 in TiB2 at that temperature is approximately 63 mo1.-%
whereas the solid solution of (W,Ti)2B5 type contains only 3mol.-% TiB2 at the
eutectic equilibrium. The homogeneity range of the (Ti, W)B2 solid solution narrows
significantly with decreasing temperature and is 46-49 mo1.-Yo at 2000°C and 8-
10mol.-% WB2 at 1500°C.
Post et al. [I 321 have treated this system before and found a homogeneity range of
the A1B2 structure type to at least 50mol.-Yo WB2. The preparation of a pure single
phase WB2 of AlB2 type structure has failed. Pastor [134] also states that the WB2
concentration in (Ti, W)B2 may exceed 50mol.-%. Yasinskaya and Groisberg [145]
have examined the interaction of TiB2 with metallic W up to 2700°C. Although no
data are given in the text a maximum solubility of 5 at.-% metallic W can be taken
Figure 33. The ternary Ti-W-B system at 1400°C (after [146]).
from the given pseudo-binary section TiB2-W. The corresponding points are shown
as a quadrangles in Figs. 32 and 33. The isothermal section of the B-Ti-W system
presented by Kuz’ma et af. [146] reveals a maximum of 6-8 mo1.-% WB2 dissolved
in (Ti, W)B2 and 34mol.-% TiB2 in (W, Ti)B2 at 1400°C (Fig. 33). These data are
more or less consistent with the phase diagram presented by Telle et al. [136] and are
thus shown as triangles in Fig. 32. Kosterova and Ordan’yan [147] have retreated
the ternary B-Ti-W at 1400°C in order to specify more accurately the phase equili-
bria in the metal-rich region but have basically found the same situation in the TiB2-
WB2 section as Kuz’ma et al. According to their isothermal section the TiB2 content
in (W,Ti)B2 is considered in the range of 10mol.-%, which is certainly overesti-
mated. Contrary to Kuz’ma et al. [146], Kosterova and Ordan’yan [147] take the
existence of a WB4 phase into account being in equilibrium with a W-rich TiB2
and WB2. The presentation of a metastable B-Ti-W phase diagram at room
temperature by Ariel et al. [148] notes a maximum W content in (Ti, W)B2 of 2
at.-%, which was obviously measured on samples of the TiB2-W section and thus
corresponds to the data of Kosterova and Ordanyan [147]. Unfortunately, the
more recent publication on the formation of A1B2 structure type solid solutions
[135] does not give any data on temperaturexoncentration relations in this
system. A partial re-treatment of the Ti-W-B system by Ahn et af. [149] for the
development of TiB2-W cermets was based on the isothermal section of Kuz’ma
et al. [146] but proved the existence of a distinct WTiB2 phase in the WB-TiB
monoboride section. Hot isostatic pressing of TiB2 with 1-30vol.-% W at 1900°C
resulted in TiB2-WTiB2 composites whereas a higher amount of W yielded
WTiB2-W cermets with little residual TiB2. The WTiB2 possesses P-WB structure
and contains 25.7 at.-% W, 24.8 at.-% Ti, and 49.3 at.-% B which is close to the
stoichiometry 1 : 1 :2. In spite of this, the isothermal section given by Ahn et al.
[149] show an extension of the homogeneity range to WB.
Current EXAFS- and HR-TEM studies in the TiB2-WB2by Pohl et al. [150,151],
Mitra and Telle [ 152, 1531, and Schmalzried and Telle [ 1541 indicate, however, that
annealing of (Ti, W)B2 solid solutions in the TiB2-“W2B5”two phase region at
lower temperatures results not only in the precipitation of ‘‘W2B5’’ particles but,
besides segregation of metallic glasses and elemental boron, also of metastable p-
WB-type phases and spinodal ex-solutions of W-rich A1B2-type phases which
have formerly been erroneously interpreted as “W2B5” [136, 1551. This is attributed
to the much faster diffusion of boron compared to Ti and W [I501 and shows that
equilibrium conditions are difficult to obtain.
The quasi-ternary CrB2-TiB2-WB2 system is of special interest in respect to
anisotropy effects of the thermophysical properties of the diboride solid solutions
and has been studied between 1900°C and 2100°C by Telle et al. [I361 up to now. It
is obvious that homogeneous ternary (Ti, Cr, W)B2 solid solutions of A1B2 type struc-
ture are formed in a wide range of composition. The miscibility gap emerging from the
TiB2-CrB2edge extends to towards the TiB2-WB2edge but disappears at < 80 mo1.-%
CrB2, in the WB2 rich corner a W-rich ternary liquid was observed. It is concluded
that the quasi-ternary system exhibits a ternary eutectic at 1800 f 50°C. Close to the
TiB2-WB2edge, a relatively small amount of chromium stabilizes a (Ti, Cr, W)B2 solid
solution with a high tungsten content exceeding by far that of the quasi-binary com-
pound. Also the extend of TiB2 solid solubility in “W2B5”is increased by the presence
of CrB2. But also in this system, the previously mentioned metastable phases have been
observed which make the interpretation of the entire equilibria difficult.
7.4 Boron Carbide Ceramics

Boron carbide was first prepared by Joly [ 1561 and labeled as B6C by Moisson [ 1571.
In 1934, Ridgeway [158] suggested the composition B4C, which is still under contro-
versial discussion. As discussed in Section 7.2.2.2 in connection with the crystal
structure, the composition as assessed today ranges from B4,3Cto B10.4C.Due to
the fabrication process where carbon is used to reduce boron oxide, the composition
of the commercially available boron carbide is close to B4C.
7.4.1 Preparation of Boron Carbide

Boron carbide is prepared for most purposes according to a technical-scale process
providing comparatively cheap powder which is used for grinding and lapping grits
or for the boronization of steel, superalloys and other non-ferrous metals. For the
production of ceramics this material must be processed further on by intensive
milling, favorably ball-milling, and subsequent cleaning from impurities. Other
more sophisticated methods, especially vapor phase reactions and synthesis
starting with metallo-organic precursors are very costly and, therefore, restricted
to scientific applications or special purposes of high profit gain.
7.4.1.1 Technical Scale Production

Boron carbide powder is produced on a technical scale by the carbothermic reduc-
tion of boron oxide with graphite or petroleum coke
2B203 + 7 C + B4C + 6CO (2)
The process is carried out in huge electric arc or resistance furnaces and is compar-
able to the Acheson process. The reaction takes place between 1500 and 2500°C, is
strongly endothermic and requires 1812 kJ/mol, i.e., 9.1 kWh/kg [159, 1601. Since
large quantities of carbon monoxide (approximately 2.3 m3/kg) are formed, the
reaction of Eq. (2) is accelerated to the benefit of B4C. Both volatilized boron
oxides and carbon monoxide generate an internal Boudouard equilibrium within
the raw material mixture and thus contribute to a self-propagating purification
process, which can be expressed by
B203+3CO-+2B+3CO2 (3)
2 co * c02 c + (4)
4B + C + B4C (5)
The furnace is usually cooled externally to limit the loss of volatile materials and
hence the outer mantle stays unreacted. The core contains blocky boron carbide
of relatively high purity (total metallic impurities < 0.5 mass-%), reproducible
stoichiometry (B/C ratio = 4.3) [50], and several percent of residual graphite. The
chunks are crushed and milled to the final grain size.
A similar process with lower productivity is used for the synthesis of high-purity B4C
of controlled stoichiometry. At temperatures of 1600-1800°C, hydroboric acid reacts
with acetylene black, high purity sugar, or ethylene glycol in a vented tube furnace:
4H3BO3 + 7 C + B4C + 6 H 2 0 + 6CO (6)
Powders of 0.5-5 pm particle size are obtained [161].
Boron oxide can also be converted to boron carbide by exothermic magnesio-
thermic reduction in the presence of carbon black at 1000-1 800°C [ 1621.
2B203 + 6Mg + C -+ B4C + 6MgO (7)
The process is performed by single point ignition (thermiteprocess) or in a carbon tube
furnace in a hydrogen atmosphere. The problem is the removal of magnesia, mag-
nesium borides, and unreacted magnesium metal which are usually extracted by
hydrochloric or sulfuric acid. Since MgO acts as a grain growth inhibitor, submicron
powders with Mg compounds as the only impurities are produced [163, 1641. Further
chemical refinement by high-temperature vacuum treatment, however, induces an
undesirable coarsening of the particles. The 1990 total annual production of boron
carbide in the western world is estimated at approximately 500-600 t.
7.4.1.2 High-Purity Material

In laboratory-scale production, boron carbide can also be synthesized in the form of
high-purity powders or coatings (e.g., [165]):
(i) from the elements by arc melting at 25OO0C, or self-propagating synthesis above
1100°C:
4B + C + B4C (8)
(ii) by chemical vapor deposition reducing boron trichloride in the presence of
carbon in a hydrogen atmosphere:
4 BC13 + 6 H2 + C -+ B4C + 12 HCI (9)
(iii) by pyrolysis of boron trihalides with methane or carbon tetraiodide as carbon
carriers, in high-frequency furnaces:
4 B(C1, Br)3 + CH4 + 4 H2

900- 1800°C
B4C + 12 HCl(HBr) (10)
4 BI3 + CI4 900-11OO’C

B4C + 8 12
The latter methods yield boron-enriched solid solutions with a maximum of
20.4 mass-% carbon.
Very fine boron carbide powders of spherical shape and 20-30nm in size have
been prepared by chemical vapor deposition according to (iii). In an Ar-H2-CH2-
BC13 atmosphere a radio frequency plasma produces stoichiometries between
BIS& and B3.& [33, 1661. Also laser-induced pyrolysis of similar gas mixtures
with or without acetylene has been employed for the preparation of nano-sized
particles [167]. With similar success, composites of B4C and S i c have been produced
by the pyrolysis of boron-containing polysilanes [ 1681.
The general problem associated with the production of submicron powders by pyr-
olysis is the comparatively low yield of these highly expensive procedures and the
excess of free carbon which cannot usually be avoided. The advantages of high
purity and well-defined composition are limited due to the pick-up of oxygen by the
large and hence extremely reactive surface area of the particles when exposed to air.
7.4.2 Sintering of Boron Carbide

Sintering of covalently bonded materials is generally much more difficult than
densification of oxide ceramics or metals. This is not only due to the low self-
diffusion (poor tendency towards grain boundary and volume diffusion), high
ratio of grain boundary-to-surface energies and high vapor pressure of particular
constituents (strong tendency towards surface diffusion and evaporation reconden-
sation), but also due to their extreme sensitivity to environmental factors such as
sintering atmosphere, traces of contaminants, particle size and shape distribution,
temperature gradients, etc. The phenomenon of a “terminal” density, i.e., the
density obtained after sintering which is far below the theoretical density for pore
closure, above which neither an increase in the temperature nor a prolonged sinter-
ing time would assist further densification, was frequently observed for B4C, S i c
and Si3N4.The reason for this, as proposed by DeHoff et al. [169], Greskovich
and Rosolowski [170], and Prochazka [171], is that upon sintering the decrease in
Density I%]
Figure 34. De-Hoff diagram showing the loss of specific surface area during densification of various
ceramics (after [ 1711).
the specific surface area (driving force for densification) is consumed to a much
greater extent for pore and particle coarsening (Ostwald ripening) than for grain
boundary movement and pore removal. Figure 34 shows a so-called DeHoff
diagram correlating the specific surface area and the fractional density on which
the path of an ideally densifying material is illustrated by the diagonal line. The
lines plotted for A1203, Sic and B4C powders make it obvious that in the first
step of sintering, the surface energy is dissipated very fast due to coarsening [171].
Since a doubling in particle size corresponds to a decrease in the densification
rate by a factor of ten, it is no wonder that densification comes to an end before
pore closure is achieved. As pores are favorably removed by grain boundary move-
ment, it is essential to generate a pore size distribution below a critical size above
which pores are stable or even tend to grow (i.e., the driving force for pore shrinkage
is 5 0) and to induce grain growth at moderate rates so that vacancies may be
suitably removed from the surface of the pores.
Another reason for the poor sinterability is the extraordinarily high vapor
pressure of boron oxides and suboxides. Since boron carbide powders are generally
coated by a B203 layer [172] which quickly reacts to form boric acid, H3B03,in
humid atmosphere, vapor phase reactions are active at higher temperatures, in
particular above 15OO0C, providing a fast transport of boron compounds. Redox
reactions such as
B2°3 vap. + B4Csolid * Bovap. + covap.
may be shifted to the benefit of the left or right side of Eq. (12) depending on the local
chemical potential of the particle surface being defined by the local curvature. The
Gibbs-Thomson equation correlates not only the particular chemical potential p
with the local surface radii but also the local vapor pressure. As a result, B 2 0 3 may
pick up boron from B4C particle areas with convex surface curvatures and move it
7.4 Boron Curhide Ceramics 84 1
as BO to areas of concave curvature where B4C is deposited again due to the lower
vapor pressure compared to the starting point. This reaction releases B203 again,
which starts the process cycle another time. Depending on the oxygen vapor pressure
also other boron suboxides such as B202and B 2 0 may be involved. This mechanism,
denoted as evaporation/reprecipitation in the science of sintering, is known to contri-
bute significantly to grain growth without any shrinkage (e.g., [173]).
In conclusion, the sintering of boron carbide requires (i) oxygen-removing
additives such as graphite, carbon black or organic deoxidation agents such as
formaldehyde resin, (ii) very fine powders of high surface area and therefore high
driving force, (iii) high temperatures to enable grain boundary and volume diffusion,
and, if everything fails, (iv) high pressures.
7.4.2.1 Pressureless Sintering

Densification of pure stoichiometric boron carbide is extremely difficult. Due to
the high fraction of covalent bonding (> 90 YO), pore eliminating mass transport
mechanisms such as grain boundary and volume diffusion become effective at
temperatures above 2000"C, i.e., at temperatures close to the melting point. At
lower temperatures, surface diffusion and the already discussed evaporation-recon-
densation reactions are the favored mechanisms, resulting in neck formation
(increase of contact area), pore coalescence and particle rounding (decrease of
specific surface area), or euhedral growth of particles by vapor phase reactions,
respectively. Grabchuk and Kislyi [ 1741 proved that the regime of predominant
surface diffusion extends from 1500°C to 1800°C, whereas sublimation occurs
above 1800°C with boron being the more volatile species. Only the latter sintering
mechanism causes an enhanced shrinkage of the ceramic body. However, a poor
tendency towards plastic deformation, a high resistance to grain boundary sliding,
and low surface energies hinder considerable particle rearrangement or shape
accommodation before grain boundary or volume diffusion is effective. Even sub-
micron powders cannot thus readily be densified completely by pressureless sinter-
ing if they are not mechanically or chemically activated. The general preconditions
for the densification of pure stoichiometric boron carbide are to start with very fine
powders (preferably << 3 ym) of low oxygen content and to use temperatures in the
range of 2250-2350°C. Above 2000"C, a rapid coarsening occurs, which usually
results in unremovable, entrapped residual porosity. Sintering parameters and
obtained densities have been presented for many compositions in the homogeneity
range of boron carbide by, e.g., Adlassnig [175] (2250-2300°C; 80-87% density;
2450°C: >goo/,) and Grabchuk and Kislyi [176] (2300°C: 99-99.5% density) who
used finer powders. A considerable reduction in temperatures can be achieved by
microwave sintering using 2.45 GHz radiation. After a 12 min treatment at
2O0O0C, 95% of the theoretical density was obtained [177]. The energy conservation
compared to hot-pressing is, however, rather low.
Starting from powders with a smaller particle size, e.g., < 1 ym, results in lower
sintering temperatures and higher final densities. Boron carbide, however, becomes
pyrophoric with increasing specific surface area and is hence strongly oxygen-loaded
or even dangerous in handling.
Activation of grain boundary and volume diffusion and thus densification at

lower temperatures is possible by increasing the density of point defects or disloca-
tions: (i) mechanically by high-energy milling (attrition milling); (ii) by doping with
trivalent ions, which substitute for carbon and thus introduce electron deficiencies
and vacancies, e.g., by adding boron or aluminum; (iii) by introducing sintering
additives, which remove oxide layers on the surface of the boron carbide particles
and thus increase the surface energy, e.g., by adding carbon, aluminum carbide,
silicon carbide or related compounds, which also inhibit exaggerated grain
growth [178]. Other methods make use of additives which possess a comparatively
low melting point and have a suitable wetting behavior on boron carbide to provide
a rapid path for mass transport via the melt and thus to initiate liquid phase sinter-
ing. Dense bodies of boron carbide have also been obtained by liquid phase infiltra-
tion of highly porous powder compacts or presintered ceramic bodies.
Kislyi and Grabchuk [179] reported that volume diffusion is enhanced in the
boron-rich area of the homogeneity range of boron carbide due to the generation
of point defects. In fact, pressureless sintering with boron additives results in an
onset of shrinkage at temperatures which are 300 K lower than those required for
stoichiometric B4C. Since aluminum also substitutes for carbon, a similar mechan-
ism may be activated. Accordingly, 95-99.2% of the theoretical density is obtained
at 2100-2200°C with 3-15%, preferentially < 1% A1 additive (e.g., [ISO]). Other Al-
providing sintering additives are A14C3,A1203 and AlF, [180-1821, which also use
carbon or fluorine as deoxidizing agents. The use of metallic additives is limited due
to the low thermodynamic stability of boron carbide reacting with metals to form
metal borides and free carbon, except in the case of Cu, Zn, Sn, Ag, and Pb. Never-
theless, Mg, Cr, Co, and Ni have been used [181,183,184], with minor success.
Stibbs et al. [185] have proposed additions of 5-10 mass-% Al, Mg or TiB2 to
obtain > 99% density between 21 50 and 2250°C. TiB2, CrB2, and W2B5 additives
inhibit grain growth by grain boundary pinning or, as in the case of W2B5 at
> 2220°C, initiate liquid phase sintering if a eutectic reaction occurs [186]. Sintering
of submicron powder with an addition of 1 mass-% Be2C resulted in 94% density
when sintered between 2200 and 2280°C [187]. Kriegesmann [188] made use of the
previously discussed vapor phase reactions and considered a “recrystallization”
process to produce intentionally porous boron carbide resembling “recrystallized”
Sic. Since the vapor pressure of the boron species increases with higher surface area
an smaller radii of curvature the starting material consists of powders with two
maxima of grain size distribution. The smaller portion of the particles serves as a
feeding material for the volatile boron species, which are deposited at the necks
between the coarser particles. The process is carried out at >2000°C and a high-
purity ceramic of defined pore size and grain size is produced.
The only technically important sintering additive for boron carbide is carbon, as
discovered almost simultaneously by Schwetz and Vogt [189], Henney and Jones
[190], and Suzuki et al. [191], which today allows the routine production of dense
parts. An amount ranging from 1 to 6 mass-% is sufficient to obtain almost
theoretical density. Schwetz and Grellner [ 1921 added phenolic resin (corresponding
to 1-3 mass-% C) to a submicron B4C powder and obtained >98% density at
2150°C. The resins are pyrolized at temperatures up to ~ 1 0 0 0 ° C and they leave a
homogeneous layer of amorphous carbon on the B4C surface, which promotes

sintering. The sinter activation was attributed to an increase in surface energy due
to the removal of oxide layers. Moreover, residual graphite particles which have
been observed at the grain boundaries may inhibit surface diffusion and evaporation
and may also control the grain boundary movement [178,193]. Firing of B4C with 6
mass-% C additive at 2220°C results in a 97% dense microstructure of 1-5pm
equiaxed particles, i.e., almost no coarsening has occurred. Abnormal growth of
individual grains to 10-30 pm starts above 2235°C; at 2250°C, extensive Ostwald
ripening and twinning is observed. The local growth of faceted grains exceeding
500 pm in size was attributed to liquid phase sintering processes due to the presence
of low-melting impurities [ 1781.
The method of in-situ pyrolysis of organic additives such as novolac-type resins to
amorphous carbon was also studied by Bougoin et al. [194]. The advantage of the
precursor method is the improved homogeneity of the carbon distribution and the
extraordinarily fine resulting average grain size of 2 pm and less. Furthermore,
the resin may act as a molding aid upon cold isostatic pressing, or may even be
the plasticizer for injection molding. Thus complicated parts can be fabricated
easily and subsequently pyrolyzed and pressureless sintered.
The drawbacks of the resin route, i.e., the fairly complex handling of the powders
and environmental problems during pyrolysis, have triggered research towards
suitable resin substitutes. The direct blending of amorphous carbon with B4C has
been suggested as an alternative, but until recently rather porous microstructures
were observed when B4C is sintered with carbon black, the suspected reason being
the insufficient distribution of the sintering aid. An improved route of dispersing
amorphous carbon in B4C powders has been developed recently by Matje [195]
and studied further by Sigl and Schwetz [116,197]. Boron carbide powder was
doped with carbon black resulting in 5, 6, and 7 mass-% of free carbon in the
as sintered bodies. Sintering was carried out in a 1Ombar Ar atmosphere for 2 h
at temperatures between 2050 and 21 75°C. All compositions sinter at surprisingly
low temperatures and approach nearly full density at around 2150°C. These sinter-
ing temperatures are significantly lower ( ~ 1 0 0 K )than those reported for the
phenolic resin as a dopant. As shown in Fig. 35 densification is hardly affected by
the amount of free carbon. This observation agrees well with the results of Dole
and Prochaska [178] who found that the density of B4C becomes independent of
the carbon addition above 733 mass-% C.
A typical microstructure of B4C doped with 5 mass-% free carbon and sintered at
21 50°C is shown in Fig. 36. In general the microstructure is very uniform and fine
grained, but some grains have already entered the regime of discontinuous
growth. This process is more clearly reflected in Fig. 37: While the grain size
increases rather slowly below 2150"C, a rapid rise of the mean intercept length is
observed beyond that temperature.
Combined additives consisting of carbon and a metal carbide or horide make use of
both the deoxidizing effect of carbon and the diffusion enhancement by the metal, or
+
the grain growth inhibiting and reinforcing effect of nonreacting phases, e.g., B C,
S i c + C, S i c + Al, or TiB2 + C (e.g., [194, 198,1991). Weaver [200,201] sintered
relatively coarse (average size 9 pm) boron carbide powders with 2 4 0 mass-%
1
5
Q
0.98 i
s
.-
c.
rn 0.96
s
C
free carbon, wt%
a, 0.94
.->
3
-
2 0.92
2000 2050 2100 2150 2200 2250

Sintering Temperature, "C
Figure 35. Densification of boron carbide with 5-7 mass-% C (after 11971).
S i c and 0-10 mass-% A1 additives to > 85% density. In similarity to the decom-
position of an A14C3 addition, metallic A1 is dispersed very homogeneously by
evaporation and condensation in the still porous ceramic body [180,202]. Starting
from submicron powders, Schwetz et al. [203] prepared composite materials consist-
ing preferentially of 9-10 mass-% Sic and 1-3 mass-% C with 97-99.7% density at
2000-2100°C. Residual porosity was removed completely by a post-HIP (hot iso-
static pressing) treatment at 1950-2050°C. Both C and S i c may also be introduced
in the form of organometallic precursors, e.g., by infiltration of a porous B4C body
with polycarbosilane and phenolic resin, or by coating of boron carbide powders
followed by pressing and subsequent pyrolysis. [107,204]. Bougoin and Thevenot
Figure 36. Microstructure of boron carbide pressurelessly sintered with carbon black (after [ 1971).
0 L- ' 1 1 8 1 8 I
2050 2100 2150 2200
Sintering Temperature, 'C
Figure 37. Grain growth In boron carbide pressurelessly sintered with carbon (after [197]).
[205] reported on the fabrication of composite bodies containing 5 mass-% S i c

residue but no free graphite. Sintering for 15 min at 2175°C results in a density of
> 92%. The microstructure of 7.5 mass-% polycarbosilane material exhibits
relatively large, faceted B4C particles (20-50 pm) with entrapped pores and local
enrichments of S i c implying that liquid phase sintering may be active. Increasing
the amount of polycarbosilane to 17.5 mass-% results in a more uniform micro-
structure that is characterized by B-to-a transformed S i c platelets of 50 pm size.
Pressureless sintering with liquid phases was studied in the B4C-A1 and B4C-Si
systems. Since A1 melts below 600°C and exhibits a significant vapor pressure at
only slightly higher temperatures, the equilibrium between 1000 and 188O"C, at
which liquid A1 is stable with an Al-saturated B12(B,C,A1)3 solid solution
(Fig. 12) [73], cannot readily be utilized for liquid phase sintering. Moreover,
problems in wetting due to oxide layers on the surface of both A1 and B4C
powder particles have to be overcome. As shown by Halverson et al. [206], it is
more effective to infiltrate compacted or presintered porous B4C bodies with
liquid Al. Since the resulting material is a metal-reinforced B4C cermet rather
than a liquid-phase sintered B4C ceramic, it will be treated in detail in one of the
following sections.
According to the B-C-Si phase diagram, liquid phase sintering of B4C should
generally be possible above 1560°C with a B-rich Si liquid [54,94]. Starting from
powder mixtures of B4C, B, and Si, the first unit is generated at 138OoC,which is
in equilibrium with SiB6 and S i c and thus may cause the partial decomposition
of B4C. Above 1560"C, however, a B12(B,C,Si)3solid solution is in equilibrium
with the liquid. Besides the complications due to iterative changes of the wetting
behavior due to dissolution and precipitation reactions upon heating, a strong
limitation on the final densification arises from the continuous evaporation of Si,
which may cause degassing channels and thus even open porosity.
Recently, boron carbide was successfully densified with TIC by pressureless sin-
tering [207]. Since TIC reacts with B4C by the formation of TiB2 and free carbon,
’05 1
In lo4 + 35 vol%
m U 25.1 vOI%
I
n
E ----.f- 13.7 VOl%
103
v
u 4.5 vol%
A
.-
c
v) lo2
0
0
.-In
> 10’
loo
0 2 4 6 8 10 12 14
PH
Figure 38. Viscosity as a function of pH and volume fraction of B4C in aqueous suspension (after
[2081).
densification is enhanced by the same mechanisms described before. Details about

the resulting B4C-TiB2composites will be discussed later on.
Since pressureless sintering to relatively high density allows the fabrication of
complex parts compared to hot-pressing research on appropriate molding
procedures for boron carbide has increased since 1990. Slip casting was studied
by, e.g., Williams and Hawn [208] using aqueous solutions. Surface contamination
by hydroboric acid was removed by washing the powders with water or alcohol.
Electrophoresis indicated that B4C is negatively charged in water above pH 1.
Stable dispersions with > 30 v0l.-YOsolid fraction and low viscosity were obtained
and successfully slip-cast at pH > 6 but boric acid destabilized suspensions above
pH 7. A slight reoxidation after storing the powders in air for 6 months caused
an increase of viscosity at pH > 9 but yielded > 58% green density compared to
45% of as-received powders. The highest casting density of 63% was obtained
after washing of the powder. Fig. 38 shows the viscosity as a function of pH and
volume fraction of solids whereas Fig. 39 presents the influence of surface conditions
to the viscosity for a 35 v01.-YOB4C suspension.
The zeta potential of B4C is generally negative above pH 1, indicating the isoelectric
point is at pH << 1 (Fig. 40). The results by Williams and Hawn [208]correspond to
similar findings by Pyzik et al. [209] who have prepared slips at pH values between
9 and 10. This behavior is attributed to the strong shielding of surface charges by
dissolved boric acid or, at higher pH, by the formation of polyborates.
Injection molding of boron carbide with 2-5 mass.-% carbon black was developed
by Schwetz et al. [210]. Like in conventional processes known for oxide and nitride
ceramics, the spray-dried powder blend was mixed with 18 mass-% organic binder
and molded at 120°C and 45 MPa. Dewaxing was accomplished by heating in an
atmosphere-controlled furnace at 100 mbar. The binder components decomposed
thermally by cracking and evaporated within four days and temperatures up to
-sr
a
105
a
E
4
lo4 _C_ As Recvd
6 Months Old
.-I*n
L
lo3 _f_ freshly Dried
0 __o_ freshly Washed
.-
0
In
7 lo2
10'
4 5 6 7 8 9 1 0 1 1
PH
Figure 39. Influence of surface conditions to the viscosity for a 35vol.-% B4C suspension (after
[2081).
450°C. Vacuum sintering between 2125 and 2225°C for 2 h yielded >96% of
theoretical density which was improved to > 99.5% by subsequent hot isostatic
pressing at 2050°C at 200 MPa.
7.4.2.2 Hot Pressing and Hot Isostatic Pressing

Since pressureless sintering allows the fabrication of complex shapes but results in
coarse microstructures and some residual porosity, this process is only applicable
--t- As Recvd
+ Washed
- U 0.2M B(OH),
-
>
-
E
.-
m -20 -
-
c
C
aJ
-40-
a
0
c
m
aJ -60 -
N
-I
L
- s o ! . I ' I ' 1 . ' ' 1 ' 1 . 1
0 2 4 6 8 10 12 14
PH
Figure 40. Zeta potential of B4C (0.01 mass-% suspension) as a function of pH [208]
70 -
99-100
60 - 97-98
h
A 95-96
CI A
"i*
O \
90-94
-
L 50- o
2 o 80-89
Q, 40- A 70-79
L
v 60-69
3
30-
2! % th. 0.
0- 20-
10 -
0 ' I ' ' ' I ~ ' ' I ' " I '

1600 1800 2000 2200
Temperature (O C)
Figure 41. Hot-pressing parameters and densities obtained from undoped boron carbide (literature
survey).
for wear parts or shieldings which are not subjected to high stresses because these
materials exhibit a low strength ( C T ~< 300MPa) and a low fracture toughness
(Klc< 3MPa m1I2). Hence, for high densification at reasonable temperatures a
hot-pressing treatment that causes particle rearrangement and plastic flow is
required. Grain boundary sliding, strain-induced twinning, creep and, at a later
stage, bulk diffusion combined with recrystallization were identified as the mechan-
isms of mass transport [211-2131. Densification maps and diffusion diagrams of B-
rich boron carbide and C have been established by Beauvy and Angers [214], and
Bouchacourt et al. [215]. Suitable preconditions are (i) the use of submicron pow-
ders, (ii) temperatures in the range 2100-22OO0C, (iii) pressures of 2 5 4 0 MPa, (iv)
15-20 min hold, and (v) a vacuum or an argon atmosphere. To resist the high pres-
sures at these temperatures and to provide carbon as a sintering aid, the use of boron
nitride-coated, graphite crucibles is favored. In Fig. 41, literature data on the
obtained fractional densities of undoped B4C are related to the particular hot-press-
ing conditions. It is obvious that both high temperatures and high pressures are
required to achieve a density of >95%. Only formation of sinter necks is obtained
at 20 MPa pressure and 2000°C (Fig. 42). However, a strong coarsening has to be
taken into account at higher temperatures and average particle sizes of > 100 pm
in commercial ceramics are not rare.
Similar to pressureless sintering, additives may be used for hot pressing of boron
carbide to reduce temperatures required for grain boundary and bulk diffusion and
to retard grain growth. Figure 43 shows a typical microstructure of a hot-pressed B-
doped material with strain-induced polysynthetic twinning. Suitable dopants are B
[216,217], almost doubling the strength, C [192], Mg, Al, Si, Ti, V, Cr, Fe, Ni, and
Cu [94,183-185,2161. As demonstrated by Telle and Petzow [218], combined B-Si or
B-Si-Ti additions lubricate the grain boundary sliding and prevent coarsening by
Figure 42. Formation of sinter necks in undoped boron carbide after hot-pressing at 20 MPa pres-
sure and 2000°C.
forming a thin Sic or TiB2 grain boundary phase (Fig. 44), which pins the grain
boundary movement and controls the surface diffusion. Compounds used as addi-
tives are various glasses, alumina, sodium silicate with Mg(N03)2,and Fe203,which
may reduce the hot-pressing temperature down to 1750°C [219]. MgF2, AlF3 [181]
and ethyl silicate [220] are other additives which are active at particle surfaces and
grain boundaries: Hot pressing with 1-5 mass- % of the above-mentioned additives
usually requires a temperature of 1750-1900°C to obtain >95% density. In most
cases, grain size refinement and distributed second phases result in improved
mechanical properties such as strength and fracture toughness.
Figure 43. Hot-pressed B-doped boron carbide with strain-induced polysynthetic twinning.
Figure 44. S i c intergranular phase in hot-pressed B,C-Si-B powder blends.
Hot isostatic pressing (HIP) of boron-containing ceramic powders creates special

difficulties due to the choice of the canning material. In general, containers made
from metals or usual glasses cannot be used because of reactions with the sample
material. In the presence of metals, boron carbide decomposes forming metal
borides plus graphite, which embrittles the capsule. In the case of silica glass,
boron diffusion from the outer layers of the specimen into the glass strongly changes
the viscosity and the glass transformation temperature. Hence the softening of the
container and the pressure transfer to the specimen cannot be controlled reliably.
Moreover, boron oxide gas may be released from both the capsule and the
sample and hence result in blowing of the container.
Promising techniques have been developed by Asea Cerama AB, Sweden, and
Elektroschmelzwerk Kempten, Germany, using diffusion barriers and a special
type of boron oxide glasses [221]. These methods are also applicable to silicon nitride
and silicon carbide ceramics and make the fabrication of complex parts, e.g., injec-
tion-molded sand blasting nozzles, in large-scale production feasible. In the case of
boron carbide, this treatment was applied to additive-free submicron powder
obtained by sedimentation of commercial powder in an aqueous suspension by
changing the pH value from 10 for dispersion to 3 for flocculation. The sedimented
powder exhibited a particle diameter << 3 pm and yielded a final density of 100%
after HIP above 1700°C for 60 min at a hydrostatic pressure of 200 MPa. In
Fig. 45, a fracture surface is shown which illustrates that no grain growth has
occurred during the heat treatment. A three-point bending strength of 714 MPa
was reported with a Weibull modulus m of 8.3 [221]. The increase in strength com-
B
pared to normal hot-pressed material is almost three-fold. The fracture tou hness
was, however, not influenced at all since values ofKrc = 2.5-3.2MPam” have
been measured in all cases by the indentation method.
Generally, all presintered or hot-pressed materials with closed porosity (i.e.,
> 95% density) can be fully densified by a post-sintering HIP-treatment [210,222].
Best results for C-Sic-doped B4C are obtained at 2000°C and 200MPa isostatic
pressure. Post-HIP of injection molded B4C-C powder mixtures yielded > 99.6%
density. The four-point bending strength of 470435MPa for 2% C samples
sintered between 2150 and 2175°C was improved to 560-580 MPa. At higher sinter-
ing or post-HIP temperatures the strength decreases due to the exaggerated grain
7.4 Boron Carbide Ceramics 85 1
Figure 45. Fracture surface of pure B4C hot isostatically pressed at 1700°C
growth to 350MPa and less. The fracture toughness as studied by Schwetz et al.
[2 101 by means of the bridge method (indentation and subsequent quasi-three-
point loading) was found to be grain size dependent and ranges between 2.5 and
3 MPa m1/2 at particles sizes < 2 p m and between 3.7 and 4 MP a m1l2 at 10m
mean grain size, which can be attributed to grain size dependent residual stresses
in the microstructure causing crack deflection.
7.4.3 Properties of Boron Carbide

The physical and chemical properties of boron carbide have been reviewed by Lipp
[159], Thevenot and Bouchacourt [256], Thevenot [164,165], and Schwetz (1999)
[223]. Special problems while presenting the physical properties arise from the
large homogeneity range of boron carbide. Furthermore, its poor sinterability
requests additives that are usually unspecified and results in residual porosity and
various grain sizes which are often also not considered in the publications. Most var-
iation and discrepancies in the properties reported come from the undefined compo-
sition of the materials studied.
The microstructure and dopant-dependent mechanical properties have been
addressed in the previous section. The strong variability of Young's modulus,
strength, hardness and toughness become more comprehensive if compared with
the sintering temperature and, e.g., free carbon content and grain size (Figs. 4 6
48). The intrinsic mechanical properties given in Table 5 are based on the most
reliable measurements essentially by Schwetz et al. [210]. An excellent overview
of hardness measurements as a function of sample preparation and testing load
was presented by Bouchacourt and Thevenot [71] stating that the hardness
values rise exponentially with decreasing load. A more or less constant hardness
is measured at loads > 20 N, which results in a large indentation diameter integrat-
ing across the average microstructure. Depending on porosity and free carbon con-
tent, however, parasitic cracks may be generated by this and higher loads which
make the evaluation of the indents again impossible. At very low loads, e.g., at
0.25 N hardness values in the range of 58 GPa have been measured with, however,
a high standard deviation. Figure 49 compares the data by Bouchacourt and
500 2
400:
rn I
1
350 1-
3
9
300
0 1 2 3 4 5 6 7
Free Carbon in HIPS Boron Carbide, wt%

Figure 46. Young's modulus vs. free carbon in post-HIP B4C sintered at 2175°C (after [210]).
Thevenot [7 11 with those of Si-doped both single crystalline and polycrystalline

B4C [83].
At room temperature the hardness of B4C is only inferior to diamond and cubic
boron nitride, which tend to weaken above 500-600°C due to the beginning of the
transformation from the diamond structure into the graphite structure. Above
2.100 2.150 2.200 2.250

Figure 47. Four-point bending strength vs. sintering temperature and free carbon content (after
[2101).
"1 o
5 Wi-%C
0
0 2 4 6 8 10 12
Mean Grain Size , prn
Figure 48. Fracture toughness (Bridge method) vs. mean grain size and free carbon content (after
[2 101).
11OOOC, and in a non-oxidizing atmosphere, B4C is the hardest compound known

up to now. B4C is thus used for wear-resistant parts and inserts for mortars and
ball mills, wear plates, sand blasting nozzles, dressing tools for grinding wheels,
lightweight armor plates for helicopters, tanks, and in composites of glass fiber-
reinforced plastics as bullet-proof protection for personnel.
Depending on the B : C ratio boron carbide possesses remarkable conduction prop-
erties. It is a high-temperature p-type semiconductor with a forbidden band width of
0.8 eV. The electric resistivity is with 0.1-100 Rcm in the range of Sic. The temperature
coefficient of electric resistivity is negative but it shows a high increase of thermoelectric
power (e.g., [224]). The extraordinarily high thermoelectric power of boron carbide
was subject of research for many years for direct power conversion plants. The appli-
cation of this unique property is today restricted to high-temperature thermocouples in
connection with a graphite electrode. Temperatures up to 2300°C can reliably
measured by this device. Thermal, optical, and electrical properties have been inten-
sively studied in order to understand the nature of atomic bonding and the homogene-
ity range by, e.g., Bouchacourt and Thevenot [225], Werheit et al. [226], Wood et al.
[227], Emin [79], and Aselage et al. [39,228] in hundreds of papers since 1980 so that
no general view can be given in this book. The data listed in Table 5 have been
taken from these few publications listed above.
The large cross section for thermal neutrons makes boron carbide an interesting
candidate for absorption or retardation of neutron radiation in power plants and as
first-wall coating in fusion reactors. The cross section for 'OB is approximately 4000
barn, which is naturally present in boron carbide at 19.9%.
Table 5. Physical properties of boron carbide.
Property Unit High carbon High boron
Formula -
B4.3C B10.4C
Density g/cm3 2.52 2.465
Composition B, C at.% 81.1, 18.9 91.3, 8.7
Crystal structure Space R3m R3m
group
Lattice constants a, c nm 0.5607, 1.2095 0.5651, 1.2196
Cell volume nm3 0.32894 0.33938
Melting point "C 2380 (eutectic with C) 2250 (liquid + C-rich)
Congruent melting point "C 2450
Thermal expansion 10-'/K 2.6 + 4.5 . T (25-8OO0C)
coefficient CL 4.5 (600°C)
4.6 (25HOO"C)
3.016 + 4.3. T - 9.18. lo-' T~
Thermal conductivity X W1m.K 35 (25°C) 4.0 (20°C) BqC
28 (200°C) 4.7 (200°C)
23 (400°C) 5-7 (400°C)
19 (600°C) 5-6.9 (600°C)
16 (SOO'C) 5-6.7 (80O0C)
Thermal diffusivity CL cm2 s-' 1 .O x lo-' (400 K)
3.0 x lo-' (600 K) 1 x lo-' (600K)
2.1 x (800K) 0.8 x (800K)
Electric resistivity p Rcm 0.1-100
Thermal coeff. of electric
conductivity K-' 3.2 10-~
Seebeck coefficient of
thermoelectric power PVIK 100 (400 K) 220 (400 K)
140 (750 K) 233 (750 K)
196 (1000K) 256 (1000K)
225 (1300K)
Self diffusion coefficient:
boron Cm-* 5.18 x lo-'' (lO0OT)
neutron absorption cross
section barn 400-750 at 0.025 eV
Fracture energy ys mJ 5.45 f0.44
Hardness HKo,, kp/mm2 2950
Strength (4-pt. bending) MPa 450-550
Fracture Toughness MPam'I2 2.5-3 for < 2 pm grain size
(bridge method) 3 . 5 4 for > 2 pm grain size
Young's modulus E GPa 440460 (25°C)
430 (250C)
420 (500°C)
415 (750'C)
400 (lO0OoC)
Porosity dependence of E GPa +
Ep = E, [(l - P)/(1 2.999P)], E, = 460 GPa
Poisson ratio - 0.15
Sound velocity ms-l 14.000
- 167 -
60
50
40
30
20
I B12(C,6 ,Si)3- single crystal
10 B12(C,6,SiJ3 + Si B,
B,C, +Si
0 i '
4
' " " "
7 2
lo0 i01 i02
Load (N)
Figure 49. Hardness vs. test load (after [71,83]).
7.4.4 Chemical Properties and Oxidation of Boron Carbide

The chemical properties are characterized by a pronounced stability in acids or
alkali liquids. Boron carbide reacts slowly with HF-H2S04 or HF-HN03 mixes,
i.e., in strongly oxidizing environment. Due to this stability, impurities of metal
or other boron compounds can readily be removed by chemical leaching. In contact
with molten alkalis boron carbide reacts to form borates. A disadvantage of the
application of boron carbide ceramics is their instability against metals a t high tem-
peratures, in particular with metallic melts. Depending on the affinity of the metal to
boron or carbon, the particular metal borides and carbides are formed. If no stable
metal carbide exists, free carbon is released by the reaction. Boron carbide is also
capable of reducing many oxides to form metal borides and carbon monoxide
under reducing conditions. In air, the particular metal borates are generated. In
hydrogen it reacts slowly above 1200°C to form borane and methane, which prohi-
bits sintering in hydrogen atmosphere. In nitrogen a decomposition to BN occurs
above 1800°C. Boron halides are evolved by the reaction with gaseous chlorine
and bromine above 600-C and 8OO"C, respectively.
Boron carbide in air is immediately, i.e., within seconds, coated by a B203 layer,
in presence of water by a hydroboric acid layer [172,229]. Oxidation starts at 500-
600°C and accelerates significantly above 800-IOOO'C, depending on the humidity.
The weight gain of boron carbide powder being surface-cleaned in glove boxes of
1 ppm O2 and 1 ppm H 2 0 and subsequently Si-sputtered, Ar' ion implanted, or
46 C/g Ar
_____----
-t
untreated
I
r-.
iIu1
---*---
0 100 200 300 400

1000 2000
Exposure to Air [hJ
Figure 50. Weight gain of ion-implanted B4C powders at 20°C.
left untreated at 20°C in air was monitored by Heuberger [230] and reported by Telle
[231,232]. Similar experiments were carried out by Matje and Schwetz [172]. Con-
trolled exposure of Ar-sputtered and implanted powders resulted in a strong,
immediate weight gain which was considerably higher for the treated powder
compared to the untreated (Fig. 50). After approximately 20 h, the untreated and
the Si-sputtered material was stable whereas the implanted powder exhibited a
continuous weight gain with a final oxygen content of almost 4 mass-% compared
to the contamination of 1.6 mass-% in the as-received powder. The studies by Matje
and Schwetz [172] proved a similar parabolic weight-gain in atmosphere of 92%
humidity, whereas the increase in weight was linearly dependent on time in air
of 52% humidity (Fig. 51). The tremendous oxygen pick-up by Si-sputtered and
-f 4 t B4C- powder 92 % rel.humidity
3t
El
x
B
O' 0 ' I00 ' 200 ' 300 ' 400

/
" I000 ' 2000 '
Exposure to Air [h]
Figure 51. Weight gain of B4C powders in dry and humid air [172].
Ar-implanted materials at room temperature is attributed to the enlarged surface

area as well as due to the formation of structural defects which aid oxygen diffusion.
7.4.5 Boron Carbide-Based Composites

Since boron carbide is brittle and susceptible to oxidation in air a combination with
other materials such as Sic, TiB2, ZrB2 was considered beneficial. As discussed
earlier, boron carbide tends to grow exaggeratedly at the temperatures required
for high densities. According to the theory of sintering isolated inert particles dis-
persed in the boron carbide matrix would also inhibit coarsening by pinning the
grain boundaries.
7.4.5.1 Boron Carbide-Silicon Carbide Ceramics

In boron carbide-based composites, silicon carbide can be dispersed as isolated
particles, e.g., by simple powder mixing 12031, mechanical alloying or as a grain
boundary phase which is formed in situ by liquid phase reactions [233]. Another
method of coating B4C with S i c which was mentioned in the previous section is
the deposition of a polysilane precursor on powder particles prior to sintering
which can be converted to S i c by a pyrolytic heat treatment [204]. In all the exam-
ples, the presence of S i c retards the strong coarsening of the matrix at temperatures
above 1900°C (Figs. 44 and 52). In general, B4C and S i c matrix ceramics can be
toughened by the incorporation of S i c whiskers, but polytype changes are encoun-
tered because of the high temperatures required for complete densification, and the
decomposition temperature of 2160°C may easily be reached in the B4C-SIC system.
Pressureless sintering of B4C-SiCfibre composites is difficult because of back-stresses
which cause porosity in the vicinity of the fibers. Moreover, the toughening and
strengthening effect is not very large since the thermal expansion coefficients of
matrix and inclusions are about the same. Thus, misfit stresses are small, and
only load transfer mechanisms due to differences in the Young’s modulus may be
operational rather than crack deflection. Thus the combination of B4C and S i c
Figure 52. Microstructure of a B4C-SiC composite derived from polysilane-coated B4C powders.
are, for the time being, only useful for the fabrication of corrosion and oxidation
resistant parts with, however, comparatively small fracture toughness.
The oxidation behavior of single-phase B4C and B4C-Sic composites in dry and
in humid air was studied by Telle [2311 in more detail depending on the microstruc-
ture. B4C-SIC composites were prepared for this purpose (i) by the conventional
powder route yielding isolated Sic particles in B4C matrix and (ii) by the precursor
route [204] giving Sic layers between the B4C particles with the same volume
fraction. Figure 53 illustrates the weight change for the heating period and the
5 - 0.5 -
.-m
- 1.0 I ~ I I , ~
0 100 200 300 400

Time at 700 O C Iminl
-E 4 - BJ+SiC lprec.)
0 200 400 600

Temperature
800 1000 1200
LOCI
U
2 -2
.-m
5 -4
o
-6 0 100 200
Time at 1000°C Iminl

Mo 400
0 100 200 300 1 3

Time a t 1200 OC [rninl
Figure 53. Oxidation behavior of B4C and B,C-SiC composites in humid air.
7.4 Boron Carbide Cerumics 859
- A: Bulk untreated -----

&- 8 B . Bulk Si-implanted
-
-
E ; C . Powder Si-implanted
\ 0 : Powder Al- implanted
ISI
E E : Liquid phase reacted .........
I -
U
L
L
2 -
o
.- r -
so&- .
0 500 1000 1500

Temperature ["[I
Figure 54. Weight gain of Si-sputtered, ion-implanted, and liquid-phase sintered B,C in air.
subsequent isothermal annealing procedures between 700°C and 1 2 0 0 T in humid

air (dew point 0°C). Upon heating to 700"C, B4C and powder-derived B4C-SIC
composites react with a slight sinusoidal weight gain and weight loss followed
by a strong weight loss at higher temperatures. Precursor-derived B4C-SiC compo-
sites exhibit a slight increase in weight starting from the very beginning but a
decrease in rate is observed above 1000°C. The isothermal treatment at 700°C
shows clearly that B4C and B4C-SiCpowder ceramics behave similarly, undergoing
a parabolic weight gain whereas the B4C-SiCpre,,,,or material exhibits less increase
in weight. At 1000°C, the initial weight loss of single phase B4C during heating is
compensated again by a slight weight gain approaching steady-state conditions.
B4C-SiCpowder samples again show almost the same reaction of stabilizing weight
change whereas precursor-derived composites undergo a continuous weight gain.
At 12OO0C, all materials suffer from a slight weight loss after 50-l00min of
exposure.
Monitoring the weight change of continuously heated single-phase and Sic-
containing boron carbide ceramics in dry air (dew point -20°C) yielded again a
step of weight gain at 650°C for all materials except one with continuous S i c
grain boundary phase prepared by liquid phase sintering, followed by a strong
increase in mass at above 1200-1250°C (Fig. 54).
A XPS analysis of the energies of Sizp and B,, bonds in oxidation layers from
humid air compared to untreated reference materials revealed that in B4C-
Sicpowder composites the binding energy of Si stays constant between 700°C and
1200°C and is somewhat lower than that of crystalline silica, but a peak splitting
is observed at 1200°C indicating that a boro-silica glass is additionally formed.
The B binding energies are close to that of glassy and crystalline boron oxide. Pre-
cursor-derived B,C-SiC composites show a decreasing binding energy of Si with
increase of oxidation temperature, indicating that a boro-silica glass forms at low
temperatures becoming rich in Si with increasing temperature and time. Finally at
1200°C, the binding energy is close to that of silica, and, indeed, isolated cristobalite
and tridymite crystals can be found on the surface.
The B bonds decrease only slightly in strength confirming that a boro-silica glass
of changing composition is present. X-ray analysis of oxidized single-phase B4C and
B4C-SiCpowder composites reveals also the existence of H3B03and HB02 after cool-
ing if treated at 700°C and IOOO'C, respectively, whereas no derivatives of boric acid
are found at room temperature after heating to 1200°C except of crystals formed
newly during storage. In contrast, B~C-SiC,,,,,rs,, did not exhibit any indication
of the presence of boric acid.
It may be concluded that in the initial stage the oxidation of all materials is
governed by the build up of a boron oxide layer, which in the case of single-phase
B4C and B4C-SiCpowderimmediately undergoes hydrolysis in humid air yielding
boric acid, which is removed by evaporation and hence is responsible for the subse-
quent weight loss according to the reactions. Only in dry systems a stable liquid
B203 layer may be formed before volatile suboxides are generated at locally low
oxygen partial pressures and temperatures exceeding 1000°C:
dry air humid air

B203(g1ass) + B202(vap) + 1/202 B203 + 3 H20 + 2 H3B0qliq) (14%b)
B2°3(glass) B2°(v2p) -k O2 B203 + H20 4 2 HB02(liq) (15a,b)
B2°3(glass) ---$ B203(liq) + B2°3(vap) B2°3 + H3B03 liquid f H2° b,
In the case of B4C-SiCpOwd,, material, the B4C matrix is exposed to air like pure B4C,
thus oxidizes according to Eqs. (13)-( 16) and hence governs the weight change of the
composites whereas a Si02,,,,,,, passivation layer is deposited on the isolated S i c
particles. Thus, the oxygen attack is not retarded at all (active oxidation) unless a
boro-silica glass is formed at 1200°C which tends to release B2O3 as a vapor
phase. In contrary, B4C particles which have been coated with precursor-derived
Sic form a continuous boro-silica glass layer at low temperatures (400°C), which
results in a progressive weight gain (passive oxidation) although B2O3 may also eva-
porate from the melt and cause a relative enrichment of SO2. At 1200"C, B203and
Si02 eventually segregate in all materials and form distinguishable Si02 and boro-
silica glass areas. The oxidation then is accompanied by increasing loss of CO
(pC0 = at 1200"C), generating small fumaroles in the protecting cover. Sur-
prisingly, thermodynamic calculations predict the occurrence of elemental boron
as a stable reaction product. At temperatures above approximately 14OO0C, SiO
evaporates at the expense of the boro-silica layer whereas Sic is newly formed
due to active carbothermic reactions close to the oxidation front involving Si02
and CO.
The rate-controlling mechanism after the formation of boro-silica glass layers is
molecular oxygen diffusion through that layer. The activation energy for that diffusion
in silica is reported to range between 83 and 125kJ/mol[234] whereas that in B203-
Si02 glass is 139 kJ/mol. The activation energy of B4C oxidation in dry air was deter-
mined by Telle [231] and Litz [235] to be 134 and 108 kJ/mol, respectively, which is
considerably higher than that determined for humid air in that work (51 kJ/mol).
As expected, B4C-Sic composites exhibit an intermediate activation energy of

77 kJ/mol, whereas precursor-derived B4C-SiC composites with 112 kJ/mol are
close to oxygen diffusion in S i c or boro-silica glass, which confirms this mechanism
to be the rate-controlling step in dry systems. For comparison, the formation of Si-0-
H . . . -0-Si bonds (i.e., H2 diffusion in Si02(glass))requires 71-75 kJ/mol for activa-
tion [236] and may become important for humid environments.
The microstructural appearance of S i c in B,C-SIC composites clearly influences
the oxidation kinetics. The formation of a continuous boro-silica layer at rather low
temperatures is helpful in preventing active oxidation during which always new
material is exposed to oxygen attack by the evaporation of volatiles. Humid atmos-
phere is generally more detrimental to the oxidation behavior of B4C-containing
materials than a dry one. A uniform boro-silica glass layer is only formed if S i c
is homogeneously distributed as an intergranular phase, which can be accomplished
by (i) liquid phase reaction sintering, (ii) coating of B4C powder with polysilane-
derived Sic, and (iii) ion beam assisted sputtering of Si or S i c on B4C powders if
the powders are processed under inert gas atmosphere.
7.4.4.2 Boron Carbide-Transition Metal Diboride Ceramics

Boron carbide-based composites with transition metal diborides - in particular with
TiB2 - have been extensively studied for cutting tools and wear parts [71, 98, 100,
103, 218, 237-2401. Since both phases are thermodynamically stable up to 2300°C
composites can be prepared either by pressureless sintering with an Fe additive at
2175°C [241], or by hot-pressing and HIP without additives. Nishiyama and Ume-
kawa [240] have obtained full density by pressureless sintering blends of 20-60 v01.-
% TiB2 at 2100°C for 1 h in a vacuum. A maximum three-point bending strength of
620 MPa was measured on composites with 35 v01.-% TiB2 combined with hardness
HRA of 93.8. During cutting of a A125Si alloy, the composites exhibited the same
performance as cBN and K15 hard metal tools and were clearly superior to
Ti(C,N)- and alumina-based materials.
Another method of densification makes use of reaction hot-pressing or self-
propagating combustion sintering of MC-B powder mixtures under pressure:
MC + 6 B + MB2 + B4C (17)

Since sintering of MB2-B4Cpowder mixtures yields similar complications to the sin-
tering of the pure compounds due to favored surface diffusion and evaporation-
recondensation reactions, the combustion route is more likely because heat is
generated inside the sample due to the exothermic conversion, the bulk diffusion
is significantly enhanced and a grain size refinement occurs as the carbide phase
decomposes. A certain risk is the evaporation of volatiles such as CO, C 0 2 , B203
as deoxidation products, or even of Bgas, which may be formed because of the
high heat release. Temperatures exceeding 2300°C have been reported during fabri-
cation of TIC and TiB2 from the elements [242-2461. The reaction velocity can be
retarded by the addition of the final conversion product to the starting powder
which then behaves as if inert. Thus in the case of TiC/B mixtures, TiB2 is added
862 7 Boride-Based Hard Matevials
Figure 55. SEM micrograph of a reaction hot-pressed Tic-B powder blend. Light: TiB2, dark: B4C.
or B4C, which also takes part in the reaction:

2MC + B4C -+ 2MB2 + 3 C (18)
3C+12B+3B4C (19)
In this case, B4C also undergoes a grain size refinement which is very beneficial for
the mechanical properties. In Fig. 55, a micrograph of a reacted TIC-B powder
mixture is presented, which still exhibits TiB2-B2C agglomerates of the size of the
initial T i c particles. Note that the average particle size of both reaction products
is approximately 1 pm. Generally, this reaction can be employed for most of the
transition metal boride-boron carbide composites since the borides are usually
more stable than the particular carbides [92].
Dense composites of MB2 and B4C, in particular of TiB2 and B4C, regardless of
their fabrication technique exhibit improved mechanical properties compared to the
particular single-phase materials. The increase in strength of hot-pressed or HIPed
materials to ob = 600-800MPa is mostly attributed to a retardation of the grain
growth, whereas the improved tou hness is due to crack deflection around TiB2 par-
,g
ticles. K,, values of 5-7.3 MPa m' have been reported for B4C-based composites
with LaB6, TiB2, ZrB2, NbB2 and W2B5 [75]. At 2150°C, pressureless sintered
B4C/TiB2 composites with 1 mass-% Fe additive exhibit a maximum bending
strength of ob= 420 MPa at an optimum volume fraction of 20% TiB2. The
lower strength compared to the hot-pressed material is mainly attributed to the
embrittling FeB intergranular phase. With increasing sintering temperature and
amount of additive, the strength even drops to 100-250 MPa due to the exaggerated
coarsening of the B4C matrix by one order of magnitude.
Another example of successful materials development is B4C-TiB2-W2B5compo-
site ceramics prepared by reaction hot pressing [218,232,237]. The initial powders
consisted of B4C, B, and Si and were mechanically alloyed with WC, TIC, and
Co. During sintering or hot pressing, the carbides react with B4C and free boron
to the desired transition metal borides. This reaction is accompanied by a strong
bloating. Above the eutectic a boron- and metal-containing Si melt is generated
which promotes liquid phase sintering depending on the volume fraction. Generally,
hot pressing is, however, required to overcome the swelling. The reaction was
- 100 I 1
,\"
v
x 8o
.-
4-
In
C
-,-
C
60
H
I
Y
8 LO
a
Q)
.-> 20
4-
+
0
a,
L O
0 200 LOO 600 800 1000 1200 ILOO 1600 1800 2000
Temperature ("C )
Figure 56. X-ray analysis of sintering reactions during hot pressing of B4C-Si-B-TiC-WC-Co
blends. Holding time at the particular temperature step 1 min.
monitored by heat treatments at temperatures ranging from 600°C up to 2100°C

with pressure of 47 and 65MPa and isothermal sintering times ranging from
1 min to 8 h. As expected from the phase diagrams, WC and TIC react with elemen-
tal boron and B4C-forming borides such as WB4, W2B5,and TiB2.As shown by the
relative X-ray intensities of the phases in Fig. 56, this reaction takes place in the
temperature range from 1000°C to 1200°C. Si coming from the additive and Co
resulting from the binder material of the hard metal milling balls also react forming
a B-rich Si-Co melt at approximately 1100°C. At the same temperature, unknown
phases denoted as U, and U2 appear which are possibly silicides or ternary borides
and decompose at 1600°C. Above 1600"C, the dominating phases are B4C, W2B5,
and TiB2. Individual Si or Co phases are not detectable any more. Si and Co segre-
gate at or close to the grain boundaries between the transition metal borides and the
boron carbide matrix where they dissolve into TiB2 and W2B5 particles. A solid
solubility of TiB2 for Si of at least 8 mass-% was documented in model experiments.
Free carbon can be observed only in powders which were milled for 3 h if the initial
boron content was not increased. The strong weight loss observed during heating is
due to the evaporation of Si, Co, and B species as well as due to deoxidation
products such as H3B03, B203,CO, and CO2.
The following chemical reactions may occur:
-
900-1 IOO'C
2WC+ 1 3 B W2B5+2B4C
~ (20)
TIC +6B 900-1 100°C
TiB2 + B4C
8 WC + 7 B4C
1000-1 100°C
4W2B5 + 2 B4C + I3 C (22)
864 7 Boride-Bused Hard Materials
Figure 57. Microstructure of hot-pressed material. Dark matrix: boron carbide, gray: TiB2-solid
solution, white: W2BS.
1000-1100°C
2TiC+B4C- 2TiB2+3C (23)
> 600°C
4B+C-B4C (24)
These reactions are strongly exothermic and belong to the type of materials synthesis
denoted as “self-propagating high-temperature combustion synthesis”. The standard
Gibbs Free Energy of the reaction (21) is with AGO = -200 kJ/mol B4Cinsensitive to
the MC compound involved and fairly independent of temperature. Furthermore,
W2B5 tends to decompose above 1600°C in the presence of TiB2 forming a solid solu-
nTiB2 -
tion (Ti,W)B2 with increasing solubility of W by increasing temperature:
+ mW2B2> 1600°C (Ti,W,)B2 with n = 1 - m and mmax= 0.63 at 2230°C
(25)
The microstructure of a hot-pressed sample is characterized by a homogeneous
distribution of B4C, W2B5, and (Ti,W)B2 (Fig. 57). The average grain size is 0.8-
1.O pm. The final phase composition consists of 72 v01.-YO B4C, 20 v01.-YO
(Ti,W)B2, and 8vol.-% W2B5. The porosity is less than 3vol.-%. Boron carbide
and titanium diboride particles are of more or less equiaxed morphology whereas
it is a characteristic of the W2B5 phase to grow in an elongated shape. Hot pressing
at temperatures above 1900°C results in an exaggerated grain growth of boron
carbide and the transition metal borides. The average grain size may reach more
than 10pm if fired at 2000°C. Boron carbide grows most probably by transient
liquid film sintering, consuming smaller particles, whereas the transition metal
borides grow by the dissolution of W2B5 in the liquid film and precipitation as
(Ti,W)B2 solid solution. The liquid phase consists possibly of a Si-Co-Ti-W-B
alloy which is consumed by the formation of B12(B,C,Si)3and (Ti,W,Co,Si)B2
solid solution in the areas close to the grain boundaries.
Table 6. Mechanical properties of B4C-TiB2-W2B5composites hot

pressed at standard conditions.
Property Value Unit
Hardness HV,,
4-pt. bending strength
*
32.1 3.2
830 rt 113
GPa
M Pa
Fracture toughness
ISB 4.2 f 0.4 MPa ml’*
ICL,lON 5.0 f0.8
Young’s modulus 425 zt 10 GPa
Shear modulus +
185 5 GPa
Poisson’s ratio 0.15 zt 0.01 ~
The hardness (HV,,) reaches 32 GPa as a maximum. The maximum bend strength
of 830 MPa is obtained by hot pressing at 1820°C. Compared to single-phase boron
carbide, the strength is increased more than two times. In comparison to other TiB2-
reinforced boron carbide ceramics, the increment in strength is still 150%. Higher
hot-pressing temperatures result, however, in a decrease to 600 MPa due to coarsen-
ing. Fracture toughness was also improved from 2.5 to 4.2-5.0 MPam’/2 by crack
deflection. The properties are summarized in Table 6. Figure 58 shows the micro-
structure of the composite in Fig. 57 failed by dynamic fracture toughness measure-
ment applying the multiple impact test (courtesy of B. Ilschner and R. Zohner).
The microstructure of hot isostatically pressed samples does not differ so much
from that of hot-pressed material. The average grain size is slightly smaller than
Figure 58. Fracture surface after dynamic fracture toughness test.

Table 7. Mechanical properties of B4C-TiB2-W2B,composites HIPed at 17OO0C,200 MPa, 60min.
Silicon Porosity Hardness 4-pt. Bending Fracture Young's Poisson

content strength toughness modulus ratio
[mass-%] [vol.-%,I HVI [GPd] [MPa] [MPa [GPa]
17.8 f 2.4 102 f 965 4.5 f0.2 384 0.17

*
0.0 3.4
3.0 1.7 37.3 f 0.2 894 73 3.5 f 0 . 7 412 0.17
1.5 0.2 29.8 f 2.4 1129 f 85 5.2 rtO.8 431 0.17
that of hot-pressed material, i.e., in the range of 0.8pm if treated at 1600°C and
1.2pm if HIPed at 1900°C. Porosities of <0.55% at 1600°C have been achieved.
The bending strength is further enhanced to 1129 i 85 MPa. The fracture toughness
exhibits a remarkable maximum of 5.2 MPam'I2 at 1700°C HIP temperature. The
entire data set is listed in Table 7.
A further improvement of both hot-pressing cycles and additives, in particular
concerning the transition metal diborides resulted in a microstructural optimization
with in-situ grown W2B5 particles. After annealing, the average of the diameter of
the W2B5 particles was about 4.5-7 pm (Fig. 59).
According to the proceeding growth of W2B5platelets an increase of toughness
with increasing sintering temperature and time was observed. As predicted by the
theory of particle reinforcement of brittle matrices introduced by Faber and
Evans [247] the increasing aspect ratio makes crack deflection a more efficient tough-
ening mechanism (Fig. 60).
In Figs. 61 and 62 the data of high-temperature strength and K,, measurements
are shown. At 1000°C the strength is only half of the room temperature strength.
The toughness at this temperature is still more than 60% of the room temperature
toughness and stays constant for higher temperatures. Since crack deflection
responsible for the relatively high fracture toughness is mainly caused by internal
stresses emerging from the misfit of thermal expansion between matrix phase and
Figure 59. SEM micrograph of annealed B4C-TiB2-W2B5sample. Note the elongated W2B5
particles.
7.4 Boron Carbide Cerumics 867
6-
E0
a 5-
2E
U
U
U
Y
3
1.0 1!5 2.’0 2
Aspect Ratio a/c

Figure 60. Dependence of the toughness of B4C/diboride composites on the aspect ratio of platelets.
dispersed particles, it is evident that during an increase in temperature the stresses

are reduced to zero at 1000°C. At that temperature no stresses contribute to the
deviation of the crack path, and the composite possesses the same toughness as
the unreinforced boron carbide matrix.
Simple stress calculation according to Selsing [248] in the B4C-TiB2-WB2-CrB2
system have shown possible stress variations from -10 to 2500 MPa if the thermal
20/7mm,51rin air
500
-a 400
300 _ _ _ _ _ _ _ _ _ _ _ - - _\ _ -4_ _ _ _
+-\*
Ll
f. unreinforced matrix
b” 200 \I\
100
200 400 600 800 1000 1200

Temperature I”C1
Figure 61. High-temperature strength of hot isostatically pressed B4C/TiB2/W2B5composite.
5.0 ISB, Srnin air
% O { , . . , , , , , , , ,
200 400 600 800 1000 1200

Temperature P C I
Figure 62. High-temperature fracture toughness of hot isostatically pressed B4C/TiB2/W2B5com-

posite measured by the ISB method.
expansion of the diborides is changed by the formation of solid solutions

[232,249,250]. Since measurements show that the thermal expansion of (Ti, Cr)B2
solid solutions do not obey the linear rule of mixture and that the anisotropy of a
may even change in sign [251], the radial matrix stresses around these kinds of
particles may be tensile or compressive in nature. Calculations of the stress levels
indicate, however, that microcracking is not likely to occur due to the comparatively
small grain size of the hot-pressed and HIPed materials [250].
As introduced above, the reaction of boron carbide with metal carbides can be
used to fabricate metal borides or metal boride/boron carbide composites in a con-
trolled way during densification if boron carbide or free boron is used in excess, or if
carbon is bonded by another additive. Although the incompatibility of B4C and
metal carbides is well known, many attempts have been undertaken to produce com-
posites or coatings thereof but failed as soon as equilibrium conditions were
approached. Physical or chemical vapor deposition of B4C on hard metal substrates,
or WC coatings on boron carbides are typical problems (e.g., [252]). In both cases,
interlayers of graphite form and hence result in an unsatisfactory adhesion of the
deposited coating to the substrate.
Recently, Sigl [207] made use of this conversion and reacted B4C with TIC to
fabricate a B4C-TiB2composite while using the emerging free carbon for the reduc-
tion of oxide layers and thus for activated pressureless sintering of boron carbide
powders. The most striking advantage associated with the processing of MC
sintering aids is their chemical similarity to boron carbide. Thus, unlike resins or
amorphous carbon, metal carbides exhibit significantly fewer problems with long-
term stability, dispersability of the sintering agent, or flow behavior of spray-
dried granules into die cavities. Furthermore, the in-situ synthesized TiB2 is
anticipated to support sintering similar to the beneficial effects such as grain
growth inhibition that have been observed in SiC-TiB2 or B4C-TiB2 composites.
Submicron boron carbide powder was doped with various amounts of fine TIC to
yield samples with nominal contents of 1.5, 3, 4.5, and 6 mass-% free carbon
after the reaction. The B4C-Tic reaction took place in a 60min holding step at
1250°C. The specimens were finally sintered in a 10 mbar Ar atmosphere for 2 h
Figure 63. Microstructure of B4C-TiC sintered with 3 mass-% C. White: TiB2, black: free C (after
~071).
at various temperatures (2125,2150,2175, and 2200°C). Samples with closed poros-

ity were further densified in a post-HIP cycle at 2050°C (30 min., 200 MPa Ar).
Figure 63 shows representative microstructures of the composite sintered at
2150°C with 3 mass-% C. It is notable that the 1.5 mass-% C composites consist
of a small fraction of pores, two major solid phases, i.e., TiB2 grains in a B4C
matrix, and traces of the dark graphite. Significantly more graphite becomes visible
in the high carbon materials. Figure 63 shows, that TiB2 particles and most graphite
are in direct contact, indicating the simultaneous formation of both compounds
consistent with the reaction considered. The mean grain size of B4C in all composi-
tions is plotted in Fig. 64. As expected, there is a pronounced tendency for grain
/- B,C sintered with 4

7
0 1 1 - 1 1
2050 2100 2150 2200 2250
Figure 64. Dependence of the grain size of B4C on free carbon content and sintering temperature
(after [207]).
Table 8. Mean strength and toughness of sintered and post-HIPed B,C-TiB2 composites as a func-
tion of sintering temDerature; n.d. = not determined.
Residual Flexural strength [MPa] Fracture toughness

C content Sintering temperature ["C] [MPaJmI
[mass-%]
21 50°C 2175°C 2200°C 21 50°C 2 175°C 2200°C
1.5 292 286 266 4.2 4.0 3.7

3.0 368 n.d. n.d. 3.6 n.d. n.d.
4.5 454 n.d. n.d. 3.2 n.d. n.d.
6.0 502 n.d. n.d. 2.9 n.d. n.d.
growth in B4C with increasing sintering temperature. However, the coarsening is

suppressed by an increasing amount of sintering aid as documented in Fig. 64.
This plot, together with the observation that the number of second phase particles
at grain boundaries increases with initial TIC content, provides evidence that both
TiB2 and graphite precipitates are effective in pinning moving grain boundaries. The
TiB2 particles coarsen with increasing sintering temperature as well. This coarsening
happens most likely by coalescence [253].
All compositions, except the 1.5 mass-% carbon material, exceed a relative
density of 95% at temperatures above 2150°C. In high carbon materials, density
increases further as the sintering temperature is raised to 22OO0C, whereas the
density of materials with 1.5 and 3 mass-% carbon declines at 2200°C. Composites
exceeding 95% of the theoretical density were post-HIPed to full density, virtually
independent of the initial sintering temperature.
The range of strength and toughness in B4C-Tic-derived composites is generally
comparable to B4C materials sintered on the resin route. Data for fracture tough-
ness, K,, and flexural strength, of, are disclosed in Table 8 as a function of the
nominal carbon content. An inverse trend becomes, however, evident from Fig. 65
a
I
B,C-TIC
L 0 L
0 200 400 600 800
Flexural Strength, MPa
Figure 65. Fracture toughness and strength of B,C-TiC materials (after [207]).
4 t I*--
1 /
/
F
i
1
Figure 66. Fracture toughness-grain size relation (after [207]).
where the flexural strength is plotted versus toughness: This plot reveals a general
trend for the strength of boron carbide materials to drop with increasing toughness.
Figure 66 shows a plot of the asymptotic fracture toughness of B4C materials
versus grain size. The fracture resistance data of B4C-TiC materials closely follow
a scatter band, which is also characteristic for B4C doped with amorphous
carbon. KI, starts at a toughness of 2 MPaJm, reflecting the inherent fracture resis-
tance of B4C, and increases up to a shallow maximum of = 4 MPaJm at a grain size
of about 10-15 pm, until KI, gradually drops again. From a microstructural point of
view, this behavior is accompanied by a continuous switch from inter- to trans-
granular fracture. This trend is outlined in Fig. 67, which summarizes the
dependence of flexural strength on grain size for B4C materials doped both with
TIC and amorphous carbon. As in many ceramic materials, the strength data
could be fitted to a grain size dependence, where strength scales with the inverse
square root of grain size, d-'I2.
Since toughening mechanisms by changing the crack path from transgranular to
intergranular fracture depend on both internal stresses and interfacial strength it is
worth while studying the grain boundary structure in B4C-TiB2 materials in more
detail as well as to explore the possibility for microcrack formation. In contrast
to Telle and Petzow [218] who have used transient liquid film hot pressing providing
a relatively strong B4C-MB2 interface, Sigl and Schwetz [254] and Sigl and Kleebe
[255] have exploited the opportunity for crack deflection and microcracking by
weakening the B4C-TiB2 phase boundary by the incorporation of free carbon. It
was suggested that microcracking accounts most for the toughness increment.
Sintering and post-HIP of B4C blends 20 and 40vol.-% TiB2 and 0, 1.2, and
4.8 mass-% free carbon derived from phenolic resins yielded full density samples
which have been examined by SEM and TEM. A SEM micrograph of a 20vol.-
Yn TiB2 composite is shown in Fig. 68. Despite the fact that 1.2 mass-% free
>
800 7117
(d B4C-Tic
n 0
z 600
Strength/Grain-Size Dependence
of B,C sintered with amorphous C
L
Q, 400
5
L
3 200
X
a,
LL
0 I J - I
0 5 10 15 20
Grain Size, pm
Figure 67. Strength-grain size relation (after [207]).
carbon should be present according to the chemical analysis, only two phases,
i.e., TiB2 (white) and boron carbide (gray), are visible. Particulate carbon
precipitates are obviously absent in B4C-TiB2composites with < 2 mass-% free
carbon. Discrete graphite particles appear in 4.8 mass-%-free carbon composites
Figure 68. SEM micrograph of post-HIPed 1.2% free carbon containing B4C-TiB2 composite.
Arrows indicate circumferential microcracks after cooling from sintering temperature (after [255]).
7.4 Boron Carbide Crrumics 873
Figure 69. TEM micrograph of microcracked B4C-TiB2grain boundary (after [255]).
(Fig. 69). Additionally, the analysis of hetero-phase boundaries by TEM yields

evidence of thin carbon interlayers (Fig. 70) with a thickness between z 5 and 10 nm.
Composites with free carbon contain microcracks which have formed spon-
taneously upon cooling from sintering temperature. Such cracks are predominantly
detected at B4C/TiB2phase boundaries (Fig. 68), and in rare cases also at TiB2/TiB2
grain boundaries. Notably, microcracks are not present at all phase boundaries
but restricted to a few interfaces feeling residual tension above a critical threshold
stress. Remarkably, microcracking coincides with the position of carbon interlayers
which is in accordance with the fact that microcracks were not observed in
composites without free carbon. Another origin of microcracking is occasionally
observed inside graphite particles, which are usually located along B4C-TiB2
Figure 70. TEM micrograph of C interlayer between B4C and TiBz particles (after [255]). The
inserted diagram presents EELS-analyses of the corresponding points 1-3 of the micrograph.
81 BT 80120
P
7
plain Boron Carbide
0
0 2 4 6 8
Relative Carbon Content, Cfre$B,C, wt%
Figure 71. Fracture toughness-free carbon relation (after [255]).
phase boundaries. Figure 7 1 compares the fracture toughness of B4C-TiB2compo-

sites with plain boron carbide as a function of free carbon content. Two effects
obviously control the toughness of these composites: (i) Particle effect: The tough-
ness of plain boron carbide varies around 2.2-2.5 MPam'l2. Upon the addition of 20
and 40vol.-% TiB2, the fracture resistance increases to 3.0 and 3.6 MPam'/*, respec-
tively, with the trend to increase with rising volume fraction of particles. (ii) Free
carbon effect: A further increase in toughness is observed upon the addition of
free carbon such that the fracture resistance exceeds 6.0 MPam'/2 at elevated
carbon contents. At still higher carbon quantities, toughness tends to remain
constant. In conclusion, strong B4C-B4Cinterfaces with comparatively low tensile
stresses coexist together with weak but highly stressed B4C-TiB2phase boundaries
with carbon interlayer or graphite precipitates which allow not only crack deflection
but also microcracking. Cracking and the relatively large graphite particles
obviously contribute to a decrease in strength since no data are provided by Sigl
and Kleebe [255].
7.5 Transition Metal Boride Ceramics

Transition metal boride ceramics are mainly based on TiB2 or ZrB2 due to their high
hardness and relatively high toughness. Because of their poor sinterability, additives
have been employed consisting of other transition metal diborides, carbides or iron-
group metals and their borides. Since the introduction of a second phase as usually
beneficial effects to the mechanical properties, a large variety of composites has been
created to reinforce TiB2-based ceramics but, vice versa, TiB2 was also used to
strengthen other matrix phase materials. The fabrication of cemented borides is
presented separately in Section 1.8.
7.5 Trunsition Metal Boride Ceramics 875
7.5.1 Preparation of Transition Metal Borides

Large-scale production of metal borides occurs preferentially in electric furnaces by
the following high-temperature reactions:
(i) Carbothermic reduction of the metal oxide, graphite or carbon black
MOz + B203 + 5 C -+ MB2 + 5CO (26)
The carbothermic method yields carbon contaminated powders and is suitable for
materials in which a C content of up to 3 mass-% can be tolerated. For instance,
TiB2, ZrB2, and the technically important hexaboride CaB6 are synthesized by
this method.
(ii) Reduction of metal oxides with carbon and/or boron carbide, known as the
boron carbide process
2MO2 + B4C + 3C + 2MB2 + 4 C O
M203+3B4C+MBs+3CO
where M = rare earth elements.
The boron carbide process can also start from blends of metal carbides, metal
hydrides, boron oxide, boron carbide and carbon black:
3 MO2 + B4C + B203 + 8 C + 3MB2 + 9 C O T (29)
MC + MO;! + B4C + 2MB2 + 2 C O T (30)
This material usually contains only small amounts of residual carbon or boron
carbide but no metals, and is thus the favored process for the technical synthesis
of less contaminated borides. The process is carried out in tunnel furnaces under
hydrogen or in a vacuum at 1600-2O0O0C, i.e., below the melting point of the
boride. It is thus a reaction sintering procedure yielding a high-porosity product
which can easily be crushed and milled. Additional refinement is obtained by
multiple vacuum treatments with metallic or B4C additives to compensate non-
stoichiometries. The final product is then called “vacuum quality”.
(iii) Aluminothermic,silicothermic, magnesiothermic reduction of mixtures of metal
oxides and boric acid
MOz + B 2 0 3+ AI(Si, Mg) + MB2 + A1203(Si02,MgO) (31)
The yield is usually contaminated by residual metals or oxides and thus has to be
purified by subsequent leaching, or a high-temperature vacuum treatment.
High-quality borides of the transition metals with-defined stoichiometry and
crystal structure are synthesized by the following laboratory-scale methods:
(i) From the elements or metal hydrides by fusion in an arc or resistance furnace,
or by diffusion during sintering or hot pressing
M + 2 B + MB2 (32)
MHz + 2 B + MB2 + H2 (33)
876 7 Boride- Based Hard Materials
(ii) Borothermic reduction of metal oxides

MO2 + 4 B + MB2 + B202 t (34)
(iii) Conversion of metal carbides with boron and/or boron carbide yielding
powder mixtures or, carried out during powder metallurgical densification, i.e., sin-
tering or hot pressing, composites
MC + 2 B + MB2 + C (35)
MC + 6 B MB2 + B4C
4 (36)
2 M C + B4C 2MB2 + 3 C
4 (37)
(iv) Electrolysis of fused salts containing metal oxides, boron oxide or hydroboric
acid plus alkaline borates and fluorides.
(v) Molten metal/boron dissolved in Al, Cu, Sn, or Pb melts (auxiliary-metal bath
method). This procedure is based upon the growth of large particles at the expenses
of small particles of high surface area which are dissolved in the melt. Furthermore,
solid solutions may be precipitated starting from a mixtures of the pure materials.
The grain size can readily be controlled by the ultimate temperature of soaking
and the cooling rate. Upon cooling down the solubility of the feedstock material
gets smaller, which is hence preferentially precipitated. The auxiliary bath method
is well-suited for the fabrication of single crystals of borides.
(vi) Chemical vapor reaction of metal halides and boron halides in a hydrogen
atmosphere under plasma conditions. This method, however, yields material of
varying stoichiometry and crystallinity.
Limiting factors for commercial fabrication are the relatively high costs of
elemental boron and the low production rate in the reactors.
7.5.2 Densification of Transition Metal Borides

The densification of single-phase and pure ceramics of transition metal diborides
is complicated by two characteristics of these compounds, the high melting point
and the comparatively high vapor pressure of the constituents. As a rule, sintering
temperatures exceeding 70% of the absolute melting temperature have to be
applied.
7.5.2.1 Pressureless Sintering

Titanium diboride, TiB2 (T, = 325OoC),requires firing temperatures of the order of
1800-2300°C to initiate grain boundary and volume diffusion, and thus to obtain
more than 95% of the theoretical density. One disadvantage is that the borides
undergo a similar abnormal grain growth at high temperatures to B4C or Sic.
Furthermore, at lower temperatures evaporation of B and boron suboxides enhance
the grain growth without shrinkage by evaporation and recondensation mechan-
isms. Since TiB2 is strongly anisotropic due to its layered structure well-faceted
particles are formed and, together with pore trapping inside large grains, the final
densities of pure TiB2 do not exceed 90% even if heated between 1900 and 2500°C
(e.g., [257,258]). Thus it is nearly impossible to achieve completely dense bodies by
pressureless sintering, as no shrinkage between 1900°C and 2100°C occurs, and
further densification by volume diffusion and plastic flow, is accompanied by
exaggerated grain growth. This behavior was attributed by Coble and Hobbs
[259] and Kislyi and Zaverukha [260] to the competing mechanisms of evapora-
tion-recondensation and volume diffusion exhibiting the same rate of mass trans-
port, whereas gas-transport reactions are favored due to the lower activation
energy. The powder size-dependent sintering behavior of TiB2 was studied by
Kislyi et al. [26 I]. Starting with high-purity submicron-size powder, synthesized
from TiC14 and BC13 in hydrogen in a plasma-arc heater, Baumgartner and Steiger
[262] achieved densities of 98.4-99.4% at 2000-2100°C combined with a com-
paratively fine microstructure (average grain size 1-18pm) due to TIC and T i 0
inclusions. Further heating or a prolonged holding time generate exaggerated
grain growth to 80pm, whereas the density does not improve. This is attributed
to entrapped porosity, which can only be removed by volume diffusion. In contra-
diction to reports of the other authors, no significant weight loss was observed even
after several hours hold at 2250"C, which could be related to active evaporation
reactions. Since the small grain size of 1 pm could be retained up to 2O0O0C, the
porosity was preferentially removed by grain boundary diffusion. Thus, contrary
to carbothermically produced TiB2 powder, high densities have been obtained
below the critical temperature promoting rapid grain growth. This implies that
both the initial particle size and the presence of impurities significantly influence
the densification kinetics. Baik and Becher [263] have studied the effect of oxygen
contamination of submicron TiB2 powders and concluded that in the case of
hot-pressing between 1400 and 1700"C, oxygen promotes grain coarsening by
enhanced evaporation-recondensation of B203. Upon pressureless sintering
between 1700 and 2050°C oxygen remains primarily as titanium oxides and sub-
oxides, which increase the surface diffusivity and thus the pore and particle
coarsening rather than the densification. A maximum total amount of oxygen of
less than 0.5 mass-% in the powder or reducing additives such as carbon is recom-
mended. Sintering is usually carried out under vacuum in a resistance furnace with
a graphite, tantalum or tungsten resistor or in a high-frequency furnace with a
graphite susceptor. Ar or H2 gas atmospheres can also be used. If carbon crucibles
are used, boron nitride diffusion barriers have to be inserted to prevent eutectic
melting of the borides and carbon in the temperature range of 20O&250O0C.
The considerable losses of volatile boron or boride species may be reduced by
powder-bed sintering [ 1341.
7.5.2.2 Hot Pressing and Activated Sintering

Densities above 95% have been achieved by axial hot-pressing at pressures exceed-
ing 20 MPa and temperatures above 1800°C. The microstructures consist typically
of particles of > 20 pm in size. Another problem is related to the hexagonal layered
structure of the A1B2-type borides. Because of the strong anisotropic behavior of
the physical properties, especially of the coefficients of thermal expansion, the
878 7 Boride-Based Hurd Materials
coarsening can be very detrimental to the mechanical properties, by producing spon-

taneous microcracking and residual strains.
Pressureless sintering and hot-pressing of transition metal borides can be gener-
ally activated either physically by starting from submicron powders or by extensive
milling, i.e., by increasing the specific surface area and introducing defects, or chemi-
cally by doping with small additions (0.3-3 mass-%) of transition metals such as Fe,
Ni, Co, Cr, Pt, or their halides. The mechanisms of sinter activation by doping are
not yet readily understood but an increase of driving forces and volume diffusion
and a retardation of evaporation seems to be likely. Crystallographic studies on
Co-, Nb-, Cr-, and Re-doped ZrB2 by Czech et al. [264] indicate that the metals sub-
stitute for Zr in the metal sublattice, which is also confirmed for Mo and W [265].
Other borides such as VB2, NbB2, TaB2 or W2B5 and Mo2B5 could not be satis-
factorily densified by pressureless sintering. For an extensive survey on powder
molding, compaction and sintering of various transition metal borides, containing
detailed descriptions of additives, the reader is referred to [ 1341.
Reaction sintering starting from chemically incompatible compounds may also
lead to high densities, especially if combined with hot-pressing, since that synthesis
is strongly exothermic and provides high internal temperatures [242]. The so-called
self-propagating high-temperature combustion synthesis was used for TiB2, ZrB2,
NbB2, and TaB2 starting from the elements, B4C mixtures with metallic Ti, Zr,
Cr, or Nb [266], and Ti-B-TiB2 blends [244,267]. In the case of other reactants,
e.g., blends of metal oxides and boron carbide or Al-Ti02-B203mixtures [243],
the formation of gaseous by-products often prevents complete densification but
may result in bodies of well-defined porosity. The kinetics of the combustion synth-
esis of TiB2 from the elements have been studied by Holt et al. [245]. Ouabdesselam
and Munir [268] investigated the sinterability of directly synthesized TiB2 powder
but could not find any significant difference to carbothermically produced powders.
7.5.3 Properties of Transition Metal Borides Ceramics

Transition metal borides are mainly explored for their mechanical properties. Since
they exhibit metallic transport properties such as high electric and thermal conduc-
tivity with a negative temperature coefficient they are also of interest as electrode
materials, for heating elements and sensors.
7.5.3.1 Single Phase Ceramics

Single phase TiB2 ceramics with high density have been prepared almost exclusively
by hot-pressing. Applying temperatures between 1800 and 2300°C densities of
>95% may be achieved [134]. Small amounts of additives may, however, signifi-
cantly improve the sintering behavior. While between 1950 and 1965 the fabrication
of single phase TiB2 was aimed at applications in the nuclear industry, multiphase
ceramics produced since then have basically been aimed at wear applications. But
also the electric properties have been studied extensively and led to highly developed
electrode materials ready for use in, e.g., aluminum production.
Table 9. Physical properties of TiBz and ZrB2
Property Unit TiBz ZrB,
Density g/cm3 4.52 6.09-6.17

Crystal structure Space group C6/mmrn C6jmmm
Lattice constants a, c nm 0.3028, 0.3228 0.3167, 0.3529
Standard enthalpy of
formation kJ/mol 279.49 308.78
Melting point “C 322 5 520 324 f 518
Thermal expansion
coefficient a,, 10-6/K 5.107+ 1.997 1-3 T
Thermal expansion 10-6/K 7.443 + 2.261 T
coefficient a,, 25-1600°C
Thermal conductivity h W/m.K 24-59 23-24
Electric resistivity p pf2 cm 20.4 (25°C) 9.2 (25°C)
26 (200°C) 10 (200°C)
36 (400°C) 11 (400°C)
46 (700°C) 13.5 (700°C)
56 (1000’C) 17 ( 1000°C)
68 (1300°C) 20 (1300‘C)
Thermal coeff. of electric
conductivity K-’ 4.76 x 6.32 x
Debye temperature 6 K 807-820 585
Hardness HK,,, kp/mm2 2600 (25°C) 2100 (25°C)
ceramics, >95% dense 2400 (200°C) 1850 (200°C)
1800 (400°C) 1000 (400°C)
1050 (600°C) 900 (600°C)
700 (1OOO°C) 800 (IOOO’C)
Young’s modulus E GPa 560 490
Shear modulus G GPa 490 220
Poisson ratio -
0.327 0.3
The physical properties of diborides have been reviewed in many papers, e.g., by
Clougherty and Pober [269], Samsonov et al. [270], and Castaign and Costa [271].
The reported value depend, however, on purity, final porosity, grain size, and
other factors that are usually not well-documented. Therefore, a comparison of
the data is difficult. Especially the mechanical properties differ very much as the ani-
sotropy of thermal expansion eventually results in microcracking during cooling
from sintering temperature, or at least to high stress concentrations at grain bound-
aries. Hardness and fracture toughness, for instance, vary therefore with grain size
and testing conditions. Table 9 presents a data set for TiB2 and ZrB,. Only data
from polycrystalline materials with densities > 95% and preferentially small grain
size, if available, have been used. Thermal expansion of some diborides have been
measured by high-temperature X-ray analysis by Lonnberg [272] and by Fendler
et al. [250] and Telle [232].
The electric resistivity in the solid solution system TiB2-ZrB2 was studied by
Rahman et al. [273]. Billehaug and 0 y e [274] present a study of several transition
metal diborides for cathode materials in Hall-H&-ould cells and come to the conclu-
sion that TiB2 should be an excellent candidate because of its stability against the
aluminum-kryolithe melt and high thermal and electric conductivity. The sensitivity
to thermal shock, the infiltration of liquid phase along grain boundaries and, finally,
the high price are, however, the limiting factors. In combination with other, cheaper
or properties-adjusting materials, there are excellent chances for this kind of appli-
cation. Ssrlie and 0ye [275] favor today a dense TiB2 coating on graphite to reduce
costs and to gain full advantage of the corrosion resistance and energy saving of
TiB2. In connection with BN and AlN the excellent electric conductivity of TiB2
and the perfect wetting by A1 has been used for the manufacturing of resistance-
heated evaporator boats for the metal deposition on plastic sheets. Data about
the corrosion behavior have been presented by Bannister and Swain [276].
7.5.3.2 TiB2-TiC Composites

As shown in the phase diagram in Fig. 19, the combination of TIC and TiB2 is
thermodynamically stable up to 2500°C undergoing a quasi-binary eutectic reaction
[I011 In that system, excellent wear-resistant materials have been produced by hot-
pressing or even by pressureless sintering of eutectic compositions at 1600-1 700°C
[103]. A Vickers hardness of HV2 = 23 GPa was measured at room temperature,
which is lower than that of the pure materials with values of 27.5GPa for TIC
and 28.5GPa for TiB2. At 600”C, however, the hardness of the composite,
8.3 GPa, far exceeds the hardness of monolithic TIC and TiB2, which decrease to
6.8 GPa and 7.8 GPa, respectively. The fracture toughness is notably improved to
7.1 MPa m’I2. The significant decrease in wear during turning or milling of steel
compared to the monolithic materials was mainly attributed to “phase boundary
toughening” due to the favored occurrence of common coherent (1 1
(0001)TiB2particle interfaces [103]. Besides this very sophisticated toughening
effect, mechanisms which influence the crack propagation such as crack deflection
or crack impediment due to thermal misfit effects between the boride and carbide
phases certainly contribute to the increase in toughness, whereas grain growth retar-
dation due to the pinning of grain boundaries by incorporated particles affects the
strength positively. As another example, WC is used for grain size refinement of
TiB2, and, vice versa, TiB2 is used as an additive for WC-based materials [257,277].
High-temperature reinforcement by in situ precipitation of TIC and TiB, from
supersaturated solid solutions has already been used with interesting results. In
the Tic-TiB2 system, either TIC or TiB2 can be the host crystal for the correspond-
ing minority phase or the precipitate [278-2801. The addition of a small fraction of
boron to T i c can increase the critical resolved shear stress at 1600°C by a factor of
six if TiB2 precipitates are formed at the (1 1 I) slip plane of TIC.
7.5.3.3 TiB2-B4C Composites

Kang and Kim [281] have investigated the improvement of TiB2 with a dispersion of
B4C particles. Using 1 mass-% Fe as a reactive additive, hot-pressing at 1700°C for
60 min at 35 MPa resulted in 99% dense composites with a clear maximum in
strength of 700MPa at IOvol.-% B4C and in KIc of 7.6MPa m1/2at 20~0l.-%
B4C. This optimizing effect was attributed to both grain growth inhibition and
7.5 Transition Metal Boride Ceramics 88 1
change in fracture mode from transgranular to intergranular by the B4C addition.

Since studies on the B4C-rich side of this system also indicate optimum properties
at approximately 60-70 v01.-YO B4C a change in strengthening and toughening
mechanisms most occur at a composition between 40 and 50vol.-% B4C. The
total system was investigated by Nishiyama and Umekawa [240] by pressureless sin-
tering of ultrafine B4C and TiB2 powders. Besides other properties such as oxidation
and wear resistance, they report a maximum in strength of 650 MPa at 35 v01.-YO
TiB2 and an optimum hardness of HRA = 94 at 2Ovol.-% TiB2. Pressureless
sintering of B4C with additions up to 16.7 mass-% TIC to initiate the reaction to
TiB2 and the simultaneous release of C for deoxidation was studied by Sigl [207]
and discussed already in Section 7.4.4.2 like all other B4C-TiB2 composites with
B4C as the majority compound.
7.5.3.4 Transition Metal Diboride-Sic Composites

Reinforcement of TiBz by dispersed S i c particles is generally possible since both
materials are chemically compatible. As a result, crack impediment is obtained
but the increase in strength and toughness is small [282]. This composite material
has, however, not yet been studied extensively.
Silicon carbide-based composites with transition metal boride particulates have
been developed for electroconductive applications such as heating elements and
igniters [283-2851 but also as wear-resistant structural parts for high temperatures
such as valve-train components and rocker arm pads in super-hot running engines
[286,287]. These composites combine the high thermal and electric conductivity of,
e.g., TiB2 and ZrB2 with the oxidation resistance of Sic. Additionally, due to ther-
mal mismatch stresses of the order of 2 GPa, toughening mechanisms such as crack
deflection and stress-induced microcracking with a pronounced process zone, as well
as crack flank friction have been proven to occur. Cai et al. [288] and Faber et al.
[289], have presented a detailed analysis of the contributions of the particular
mechanisms to the total fracture toughness, stating that stress-induced microcrack-
ing is operational in a process zone of approximately 150 pm width.
Typical conditions for densification by axial hot-pressing are 2000-2 1OO'C, at a
pressure of 2&60MPa for 30-60min which results in 96-99.8% density. The
particle sizes of the matrix and dispersed phases range from 1-5 and 4-8 pm, respec-
tively. An optimum volume fraction of reinforcing particulates of 25-30 v01.-YOhas
been reported, yielding a flexural strength of 710MPa and a fracture toughness of
5.0-5.7 MPa rn'/*, as shown in Fig. 72 [282]. Composites with a lower TiB2 content
of 15 v01.-YOexhibit a mean strength of 485 MPa combined with a K,, of 4.5 MPa
m1I2[287]. The strength of Sic-based materials with 50vol.-% ZrB2, HfB2, NbB2
or TaB2 particles also ranges between 400 and 500MPa [283]. Similar strength
values (480 MPa) combined with an exceptionally higher fracture toughness of
7-9 MPa m'l2 have been reported for large-scale lots of pressureless sintered
l6vol.-% TiB2 composites [284]. Since the sintering was carried out with tem-
peratures exceeding 2000°C (no details given) yielding 98-99% of the theoretical
density and an average TiB2 particle size of 2.0 pm, it is obvious that the reinfor-
cing phase also acts as a grain growth inhibitor for Sic. The high-temperature
A
-
0 TIC 1,5um K =3.LMPa\liii
o T i c 8,Oym ICO
T 70 -
- TiB 4,Oum
3
aJ
60 -
& 50 -
3 to
!-
-
30 -
.I-
0
a -
W
20
2
2
c-(
10 -
0 c
0 10 20 30 LO 50
TiB and Tic-Content in Sic [vol.-%]
Figure 72. Volume-dependent mechanical properties of SiC-TiB2composites [282].
strength of SiC/TiB2 and SiC/ZrB2 composites was found to remain nearly con-
stant at 480MPa up to 1200”C, and is hence superior to that of many sialons
[283,284]. SiC-TiB2 composites have been prepared a in situ synthesis by Ohya
et al. [253] by adding TIC and B powder to Sic. If the dopants are well-balanced,
these incompatible compounds react between 1000 and 1600°C to form isolated
and homogeneously distributed TiB2particles. The reaction is, however, accompa-
nied by a strong expansion, which can be overcome by isothermal soaking at
1500°C and final pressureless sintering at 2200°C for 30 min. The fracture tough-
ness of the >98% dense samples is around 3.7MPa m112 at lOvol.-% TiB2.
A systematic study on the rule of mixtures for the mechanical and electrical prop-
erties of TiB,-TiC-SiC composites was carried out by de Mestral and Thevenot
[290]. They modeled “iso-property” curves in the quasi-ternary phase diagram for
mechanical parameters by fitting 20 independent coefficients of a third-order poly-
nomial developed by Phan-Tan-Luu et al. [291] to the results of experimental test
points. Calculated iso-bend strength curves as well as tests on hot-pressed materials
indicate a maximum of 1 100 MPa close to the TiB2-Tic binary edge of the system
(Fig. 73). The best fracture toughness value of 6.4MPa m112was obtained on the
binary SIC-TIC edge (measured and calculated), on .the binary TiB,-Sic edge and
in the ternary region close to the TiB2 phase (67mol.-% TiB,, and 16.5mol.-%
S i c and Tic, respectively) (Fig. 74). The calculated rule of mixtures could also be
confirmed in the case of hardness measurements.
Sic-based composites with W2B5have been discussed by Telle [232]. The fabrica-
tion method was similar by starting with mechanically alloyed WC-B-containing
S i c powders. Due to the strong tendency of W2B5 to grow anisotropically in a
platelet shape crack deflection and crack bridging was efficiently applied, Several
kinds of diboride additives have been studied by Tanaka and Iyi [292] to reinforce
Sic. Pressureless sintering of P-SiC composites with 15-17 vo1.-YO NbB,, TaB,,
TiB2, and ZrBz at > 2000°C resulted in > 99% density. Surprisingly, the additives
strongly influenced the p-cl transformation of Sic during sintering. TaB2 addition
Ti 8
Si C Ti C
Figure 73. Iso-bend strength curves in SiC-TiC-TiB2composites [290].
clearly stabilized the 3C polytype up to 2200°C whereas ZrB2 addition extended the
stability of the 15R structure. NbB2 doping results in a very smooth transition from
3C to 6H between 2000 and 2200°C. The 4H polytype was only found in TaBz
containing materials at a larger content. Although the grain size and shape of the
S i c matrix is heavily influenced the increment in toughness reaches only 20%.
The highest vaIue was obtained for ISvol.-% TaB2 with KI, = 4.751MPam”~.
7.5.3.5 TiB2-TransitionMetal Diboride Composites

Combinations of diborides of different transition metal borides have been studied,
especially in the TiB2/CrB2 and TiB2/W2B5systems, for wear applications and to a
minor extent for electrodes in Hall-Herould cells [135, 139, 140,293-2951. Since the
transition metal diborides crystallize in the same structure type, namely the AlB2
884 7 Boride-Bused Hard Muterials
Ti B ,
Si C Ti C
Figure 74. Iso-toughness curves in SiC-TiC-TiB2composites [290].
layered structure, the formation of solid solutions has been extensively investigated
and used for hardening effects. As an example, the quasi-binary system CrB2-TiB2
exhibits a continuous mutual solid solubility approximately between 2000°C and
2100 f 50°C (Fig. 30) [132, 1361, but there is evidence of a solubility gap below
2000°C where the solubility of TiB2 in CrB, is about 40mol.-% at 1500°C and
the solubility of CrB2 in TiB2 is less than 1 mol.-% below approximately 1800°C.
The presence of CrB2 aids the densification of TiB2 due to its higher diffusion
coefficient. Above 21OO0C, CrB2 containing materials partially melts, which is due
to an almost horizontal solidus line between approximately 40 mol.-% CrB, and
pure CrB,. This fact enables liquid phase sintering of TiB2 but with the risk of
exaggerated grain growth and evaporation of chromium and chromium borides,
since the vapor pressure of Cr is four orders of magnitude higher than that of Ti.
Pre-reacted and hot-pressed materials of that system exhibit a flexural strength of
350-500MPa [139,140].
Figure 75. SEM micrograph of large W2B5 precipitates white and thin TiWB, layers in (Ti,W)B,
solid solution matrix.
In the TiB2-W2B5system, the borders of the (Ti,W)B2 homogeneity range have

been intensively studied between 1500°C and 1700"C, at 2000°C and around the
quasi-binary eutectic temperature. TiB2 and WB2 react eutectically at
T, = 2230 f 40°C and 90 f 3 mo1.-% WB2 (Fig. 32). The solid solubility of WB2
in TiB2 at this temperature is approximately 63 mo1.-%, whereas the solid solution
of the (W,Ti)2B5-typecontains only 3 mol-% TiB2 at the eutectic equilibrium. The
homogeneity range of the (Ti,W)B2 solid solution narrows significantly with
decreasing temperature and is 46-49 mol-% at 2000°C and 8-10 mol-% WB2 at
1500°C.
A high-temperature treatment of TiB2-W2B5 powder mixtures inside the solid
solubility range of (Ti,W)B2 at above, e.g., 2000"C, for 30-720 min results theoreti-
cally in a uniform, single phase microstructure. Subsequent annealing at, e.g., 1500-
1700°C, causes the epitaxial precipitation of very fine platelets of the metastable
WTiB2 monoboride phase with p-WB structure onto the prism plains of the host
crystal (Fig. 75). After 30-240 min annealing, these precipitates measure 0.5-5 pm
in length and 0.01-0.2pm in thickness and can be aged by prolonged heating
or by the choice of a higher temperature (Fig. 76). The growth of W2B5 platelets
can also be initiated by heterogeneous nucleation close to a grain boundaries of
the host crystals [81, 152,1531. The precipitate is then able to grow across the
grain boundary into a neighboring W-rich (Ti,W)B2 grain and thus create an
Figure 76. Aged in-situ reacted (Ti,W)B2-WB2composite with grown WB, particles.
Figure 77. Interlocking grain boundaries bridged by WB2 particles. Note the thin TiWB, stacks in
the host crystals.
interlocking microstructure, as shown in Fig. 77. Crack propagation studies confirm

that crack deflection is operational around the W2B5-typephases. This process is
assisted by differences between the Young's moduli of the particular phases, the
differences in the thermal expansion coefficients and their anisotropic behavior gen-
erating residual misfit stresses. Of similar importance for crack interactions are the
grain boundaries of the W-depleted host crystals and neighboring W-rich solid
solutions. Here, an active crack deflection was observed, which indicates that
both the elastic constants and thermal misfit stresses of TiB2-type solid solutions
vary significantly with composition. High-temperature X-ray diffraction measure-
ments of the lattice constants of (Ti,W)B2 solid solutions confirm this observation.
Hot-pressed com osite materials developed from the more complex systems of the
g
type TiB2-MTB2-M with M' being Hf, V, Nb, Ta, Mo, or Mn and MI' being sinter-
ing additives such as Co and Ni, exhibit bending strengths between 850 and 1000 MPa
which are due to the grain growth inhibiting influence of the 1-5 mass-% of M"B2
particulates (Fig. 78) [293-2961. During liquid phase sintering in a Co- or Ni-boride
melt, both TiB2 and MI are partially dissolved and reprecipitated as a solid solution.
The effect of grain growth retardation as well as of strength and hardness increments
is attributed to stresses at the TiB2/(Ti,M')B2 phase boundaries generated by the mis-
match of the lattice parameters between the unreacted TiB2 acting as a nucleus and
the epitaxially precipitated (Ti,M')B2 solid solution. In the case of a TiB2-5 mass-
YOW2B /TaB2 material with 1 mass-% COBbinder the lattice strain ranges between
9 x 10- d and 14 x depending on the hot-pressing temperature [293]. The addi-
tion of 1.7% TIC to the above-mentioned base composition reduced the porosity from
0.3-0.7 to 0.1-0.2vol.-% after hot-pressing at 1500°C and a pressure of 20MPa for
1 h. The improved sintering behavior was achieved by intensive ball milling resulting
in an average particle size of 1 pm, but increased oxygen contamination. Watanabe
and Shoubu [297] reported the formation of a (Ti,Ta)(C,O) solid solution which is
considered to initiate the improved densification resulting in a flexural strength of
l000MPa. In a similar multiphase system, transition metal carbides were used as
additives for pressureless sintering of TiB, yielding composites of binary and ternary
borides [218,296]. Attrition milled powder mixtures of TiB, with 3-10 mass-% Co or
Ni and 20-35 mass-% WC have been sintered in a vacuum at temperatures between
7.5 Transition Metal B o d e Ceramics 887
1200 I
600 -
400 - "\ " t o -
200 -
, r l , , , l l l l , I I I ~
Figure 78. Strength-grain size relation of various TiB2-MBZ-Mcomposites (data from [293]).
1500 and 1700°C for 60 to 120min yielding 98-99% relative density. Densification
starts above 980°C due to the formation of a liquid phase in the Ti-B-Co/Ni system
(Fig. 25). At this early stage, a rigid skeleton of TiB2 and WC develops. Due to dis-
solution and reprecipitation, a (Ti,W)B2 solid solution grows on the residual TiB2
particles. Subsequently, crystals of w-phase (see Section 7.8.3) form with compositions
of WCoB or W2NiB2,respectively. Upon cooling, the residual liquid phase crystal-
lizes as C-and Ti-enriched Co3B or Ni3B solid solution. A typical microstructure is
shown in Fig. 79. Sintering at 1700°C for 2 h yields an average particle size of
Figure 79. SEM micrograph of a (Ti,W)B2-WzNiB2-Ni3Bcomposite. Light: W2NiB2,intermediate:

Ni,B, dark: TiB2.
0.8 m resulting in a flexural strength oh= 600-680 MPa. The KIc of 6.5-7.5 MPa
mIP is mainly attributed to crack deflection because of the weak Co- or Ni-boride
intergranular phases.
7.6 Multiphase Hard Materials Based on Carbide-

Nitride-Boride-Silicide Composites
Ternary composites of Ti(C,N)-TiB2-MoSi2 were studied by Shobu and Watanabe
[298] in order to improve the oxidation resistance of Ti(C,N) TiB2 materials. Full
density was obtained after sintering composites with less than 80 mass-% TiB2
and less than 60 mass-% MoSi2 at 1750°C. The oxidation resistance above
1000°C was good for small Ti(C,N) concentrations, i.e., when all the carbonitride
particles were surrounded by a phase of either TiB2 or MoSi2. The formation of
rutile (Ti02) and silicate glass was observed and considered to prevent further
oxygen diffusion. TiB2-20 mass-% MoSi2 composites sintered at 1800°C in a
vacuum exhibited a flexural strength of 600MPa and a hardness of HV 2100,
whereas the fracture toughness was only 3.7 MPa m'I2. Composites of 70 mass-%
Ti(C,N)-30 mass-% TiB2 showed a three-point bending strength of 800MPa and
a KI, of 5MPa m1/2.With a hardness of HV > 2500, the material was tested as a
cutting tool and exhibited a longer lifetime upon machining plain carbon steel at
300 m/min than conventional hard metals or cermets [298].
7.7 Boride-Zirconia Composites

The most important toughening strategy for oxide ceramics, namely the so-called
transformation toughening by dispersed tetragonal zirconia particles, is not applic-
able to borides and carbides to a similar extent. Chemical interactions between
Zr02, with its pronounced tendency for oxygen loss, and, in presence of oxygen,
the thermodynamically less stable borides lead to the formation of boron oxides
or carboxides, respectively, which in some cases result in the total degradation of
the composite. This is particularly the case under reducing conditions and at high
temperatures, which are both required for a successful densification of hard
materials. For example, boron carbide decomposes in the presence of zirconia
according to Lange and Holleck [75]:
B4C + 2 Zr02 2 ZrB2 + CO + B203 (38)
A pronounced bloating of the samples is observed due to the release of gaseous
compounds. Eq. (38) resembles the so-called boron carbide route for the production
of the particular transition metal borides. Another limiting factor is the chemical,
geometrical and mechanical destabilization of tetragonal zirconia if combined
7.7 BorideeZirconia Composites 889
Figure 80. SEM micrograph of a (Ti,Zr)B2-(Zr,Ti)02material. Note the core-rim structure of the
dark TiBz particles.
with transition metal diborides. Stabilizing additives such as MgO or Y z 0 3 tend to

migrate into grain boundaries since these bivalent or trivalent cations in the zirconia
lattice are substituted by the more favored, such as Ti4+.A geometrical destabiliza-
tion results from the strong coalescence of ZrOz causing a particle coarsening due to
the high sintering temperatures. Large crystallites exceeding a critical size cannot be
retained in the tetragonal modification upon cooling to room temperature and
hence transform spontaneously to the monoclinic modification. Moreover, if asso-
ciated in clusters, a transforming zirconia particle may trigger the transformation of
all the other crystals by an autocatalytic reaction. This mechanical destabilization
results from the anisotropy of the thermal expansion of the diborides, which intro-
duces radial tensile stresses in the vicinity of the zirconia inclusions. This initiates
the spontaneous tetragonal-to-monoclinic transformation or at least reduces the
contribution of the ZrOz volume expansion during stress-induced transformation
to toughening [299].
TiBz-ZrOz ceramics have been studied intensively as possible candidates for
active transformation toughening [239,300-3051. Composites with ZrOz additives
show an improved densification behavior and a grain growth inhibiting effect for
the TiB2 (Fig. SO). Hot-pressing of composites with 22-60 mass-% ZrOz between
1700°C and 1900°C at 20 MPa yields densities exceeding 99.8% [300,305], whereas
98% of the theoretical density is obtained by pressureless sintering at 2100°C [239].
Volume fractions of unstahilized ZrOz between 15 and 30% result in a significant
increase in both the strength and the toughness. Depending upon the microstructure
and the density, a maximum oh of 700-800MPa is measured at 22 or 35vol.-%,
respectively, and the maximum KI, varies between 5 and 9.5MPa m'/* (Figs. 81
and 82). The hardness decreases linearly with the amount of Zr0,additive and
is thus of the order of 16-18 GPa at 20-30 mass-% Z r 0 2 being optimum for both
strength and toughness (Fig. 83). This improvement in the mechanical properties
is attributed to enhanced sintering and grain size refinement of TiB2, active transfor-
890 7 Boride-Bused Hard Muteriuls
-0
900
Q
E
u
800
5 700
C
2”v, 600
- 500
2
3
400
LL
300
200 I I I I 1
0 10 20 30 40 50
Z r 0 2 [voI.-% ]
Figure 81. Strength of TiB2-Zr02composites (data from [218,300]).
mation toughening [300],crack deflection and microcracking [239,301]. Miiller [305]

quantified the transformable amount of tetragonal Zr02 by X-raying of as-sintered
and as-ground samples. Depending on the processing and sintering conditions, the
I
remaining tetragonal fraction ranges between 0 and SO%, which could almost be
entirely transformed to monoclinic during machining. Between 20 and 30 mass-%
ZrOz, there is, however, about 15 mass-% untransformable tetragonal Zr02. The
presence of these untransformable tetragonal ZrOz particles without an yttria addi-
12.0
1f I+
11.0 1
10.0
-
E
c
9. 0
8. 0
t
a
3 7. 0
::
Y 6.0
5. 0
4. 0
3. 0
0 10 20 30 40 a
Vol% ZrO,
Figure 82. Facture toughness of TiB,-ZrO, composites (dots: data from [300],squares: data from
P181).
7.7 Boride-Zirconia Composites 89 1
Figure 83. Hardness vs. Z r 0 2 content [305]
tion can be explained by a pronounced mutual Ti and Zr interdiffusion, resulting in

(Ti,Zr)B2 and (Zr,Ti)02 solid solutions. The (Zr,Ti)02 may contain 14-16 mol-YO
Ti02 at 1700°C in outer layers of the Z r 0 2 particles and hence result in a stabiliza-
tion of the tetragonal modification. The maximum ZrB2 content in TiB2 at 1700°C
was found to be 3.2mol-%. Moreover, after sintering between 2000°C and 2100°C
there is evidence of an intergranular phase between adjacent TiB2 and Z r 0 2 particles
consisting of zirconium titanate (Zr,Ti)*04which forms peritectically from a Ti02-
Z r 0 2 melt [239,306]. This embrittling phase can be avoided by the substitution of
TiB2 by ZrB2 since zirconium zirconates do not exist. These composites are, how-
ever, also very sensitive to a spontaneous tetragonal-to-monoclinic transformation
resulting in extensive microcracking (Fig. 84).
If yttriu-stabilized Z r 0 2 is applied, the contribution of stress-induced transforma-
Figure 84. TEM micrograph of spontaneously cracked ZrB2-Zr02composites.

1.400 -
a
0
z 1200 -
f
m
-E 1000-
C
VI
m
5
C
800-
m
600 -
0 20 LO 60 80 100
Tetragonal ZrO, content (mass%)
Figure 85. Mechanical Properties of TiB,-YTZP composites sintered at 1500°C [301].
tion to the toughening is more likely. An average bending strength of 1250MPa

(maximum ob= 1500MPa) is obtained a 40 mass-% of 1.94mol-% Y203-doped
Zr02, whereas the fracture toughness is, however, only 4 6 MPa m1/2but increases
with higher Y-Zr02 contents (Fig. 85) [301]. The reason for this comparatively small
increase in toughness was attributed to the high stiffness of the matrix phase, which
reduces the dilatational strain associated with the tetragonal-to-monoclinic trans-
formation of ZrOz [302,308]. In contrast to these results Muller [305] observes a
parabolic increment of toughness as a function of the 3Y-stabilized tetragonal zir-
conia content (Fig. 86). The obtained maximum fracture toughness was 7.5-8 MPa
m1/2at 25-40 mass-% Zr02 if the sample was hot-pressed at least 1700°C. Muller
reports a maximum strength of 750 MPa at 20-25 mass-% Zr02. At higher 3Y-TZP
fractions the strength decreases drastically, at > 40 mass-% the samples ruptured
spontaneously during cooling from hot-pressing temperature. X-ray analysis of
as-sintered and as-ground samples proved that during cooling 20-90% of the Y-
TZP transforms to the monoclinic state which explains the severe cracking of the
material. The residual tetragonal phase can almost be fully transformed to mono-
clinic by grinding [305]. The reasons for the spontaneous transformation are the
diffusion of Y2O3 from the Zr02 solid solutions into the grain boundaries where,
especially at temperatures > 18OO0C,a Y203-B203eutectic is found. On the other
hand, the formation of Zr02 clusters with increasing sintering temperature and
7.7 Boride-Zirconia Composites 893
8.5
7.5
B
6.5
4.5
3.5
0 10 20 30 40 50 60
Mass -% ZrO,
Figure 86. Parabolic increase in toughness in TiB2-3YTZP-composites [305].
time is obvious [239,305]. Composites of comparably fine grain size and homoge-
neous phase distribution, i.e., isolated ZrO2 particles of z 1-2 p size, undergo a con-
tinuous phase redistribution and coarsening if hot-pressed up to 1900°C. Zirconia is
then arranged in clusters with favored Zr02-Zr02 interfaces and has grown to
> 5 pm size. As in the case of unstabilized Zr02, Ti is found in the surface-near
areas of the Z r 0 2 particles and, vice versa, Zr is also present in the outer rims of
the TiB2 particles. Thus, a grain boundary diffusion of both Zr and Ti, and presum-
ably also B and 0 is most likely. Plasma etching makes an intergranular phase
clearly visible. This kind of coalescence resulting in Z r 0 2 clusters incorporated by
a contiguous TiB2 matrix with an optimized number of TiB2-TiB2grain boundaries
is considered the reason for geometrical and mechanical destabilization of the tetra-
gonal phase. The spontaneous t-m transformation is assisted by the large ZrOz grain
size and the radial tensile stresses at the contact to TiB,. Since these interfaces are
the weakest anyway, microcracking occurs here preferentially, and the residual tet-
ragonal Zr02 particles in the cluster transform by autocatalysis.
At higher temperatures, coarsening continues up to 2100°C where the material
decomposes by internal oxidation [239]. Both boron and oxygen are then very
mobile and exhibit a high vapor pressure so that the material is blown up drastically.
Volume changes of more than 200% can be observed. The ceramic then consists of
Ti02, titanium borates, and oxygen-deficient Zr02-Ti02 solid solutions whereas
B 2 0 3 vapor creates round entrapped pores causing the strong bloating. Rupture
was not observed, most probably because the grain boundary phase is already
liquid or at least viscous at this temperature.
The high enthalpy of formation of both Z r 0 2 and transition metal borides can be
used to enhance densification by a chemical driving force starting from, e.g., Ti02
and ZrB, [303,304]:
a94 7 Boride-Based Hard Materials
Ti02 + ZrB2 + TiB2 + Zr02 (39)

This reciprocal salt-couple reaction yields a tremendous grain size refinement since all
powder particles are involved in the conversion. After reaction the microstructure is
characterized by an average grain size of 1-2 pm, approximately, and the preforma-
tion of solid solutions. Thus Eq. (39) has to written more precisely as follows:
Ti02 + ZrB2 + (Ti, Zr)B2 + (Zr, Ti)02 (40)
The mechanical properties of the reaction product have not been exploited as
yet.
Besides TiB2, also WC, ZrC, TIC and mixtures thereof have been transformation-
toughened with Y-stabilized Zr02 [309]. Hot-pressing of WC-TZP blends with up to
0TI (C, Nj-50 wt % T I 8,-a% Zr 0,

c]TI (C, N)-50 wt % TI 8, -a% 2Y Zr 0,
m
Q
E
---
c
F
65
600
._
U
a
l
c
m
400
I L
200 ' I
10 20
I
30
I
40
I I
50
Zr 0 , content (wt%)
Figure 87. Mechanical properties of Ti(C,N)-50 mass-% TiBz-ZrOz composites (after [3lo]).
5Ovol.-% ZrOz at 1600°C yields almost dense composites with a flexural strength of
2300MPa and a hardness of HV 1800 (HRA = 93.4). A reaction between zirconia
and tungsten carbide has not been observed but the appearance of a liquid phase
is reported. A similar behavior of the mechanical properties has been observed
for composites of Ti(C,N) containing Y-doped tetragonal zirconia composites
where a significant increase in the strength and hardness with Y-Zr02 content is,
however, accompanied by a modest increment in the toughness. In Ti(C,N)-50
mass-% TiB2-Zr02a decrease in hardness and strength combined with a parabolic
increase in toughness was measured with increasing Z r 0 2 content [310]. The best
performance was achieved by 12% ceria-stabilized ZrOz (Fig. 87). A characteristic
of all the Zr02-reinforced borides is the comparatively high scatter of the mechan-
ical properties measured.
7.8 Cemented Borides

Due to their remarkable hardness, borides are interesting candidates for the devel-
opment of metal-matrix composites resembling the hard metals based on cemented
carbides or cermets such as WC-Co or Ti(C,N)-Ni composites. However, since
metal-boron systems usually contain many binary and ternary borides a suitable
choice of materials is difficult and requires full knowledge of the particular phase
diagrams and compound properties. Moreover, to gain full advantage of the
chosen metal-boride system, problems in densification such as wetting have to be
overcome by appropriate dopants, and, finally, the comparatively little corrosion
and oxidation resistance of borides and matrix phase at high temperatures have
to be taken into consideration for an application.
7.8.1 Boron Carbide-Based Cermets

The fabrication of metal matrix cermets with boron carbide as a dispersed phase is
very limited under equilibrium conditions since B4C reacts with all metals, except
Ag, Cu, Sn, and Zn, forming metal borides and graphite or metal carbides (e.g.,
[31 11). In systems with slow reaction kinetics, however, complex low-temperature
materials with interesting mechanical properties have been investigated. A develop-
ment from Kiew makes use of a Ti-containing bronze as a binder phase in which
the reaction of Ti with B4C to give TiB2 is employed for active brazing and
improvement of the wetting behavior. The use of pure Cu, Sn, or Zn, or alloys
thereof for the infiltration of B4C power compacts usually fails since the wetting
behavior is rather poor (wetting angle >90"), but this can be improved by adding
Cr or other metals, which may react with the B4C when approaching equilibrium
conditions.
Other metal matrix composites with B4C particulates have been obtained using
aluminum because of slow reaction kinetics. The process is based on a n infiltration
of liquid A1 into a porous body of B4C at temperatures between 700°C and 1200°C.
Since A1 melts at 600°C and exhibits a significant vapor pressure at only slightly
higher temperatures, the equilibrium between 1000 and 1880°C at which liquid A1
is stable with an Al-saturated B12(B,C,A1)3-solidsolution [73] cannot readily be uti-
lized for liquid-phase sintering with small volume fractions of liquid. As shown by
Halverson et al. [206], it is more effective to infiltrate compacted, or presintered,
porous B4C bodies with liquid Al. The resulting material is a metal-reinforced B4C
cermet rather than a liquid-phase sintered B4C ceramic. The wetting behavior is
strongly influenced by oxidation layers formed on the surface of the B4C particles
[312], but can be improved by superheating the melt. Between the melting point of
A1 and approximately 1000°C, wetting angles of 100-150" are observed, which
decrease to reasonable values with prolonged soaking for thousands of hours
[206]. Hence, in that temperature range, only hot-pressing or hot isostatic pressing
result in high-density cermets. Above 1000-1200°C, a suitable wetting behavior is
obtained within minutes of annealing. Due to capillary forces and phase reactions
both densification and adhesion of the metal-ceramic interface are excellent.
During infiltration, reactions of A1 with B4C occur. Below 1200"C, A14BC, A1B2,
A1B12, and AIBI2C2are formed within tens of hours whereas above 1200°C the
generation of A14C3,A1B12,and A1B24C4is more favored [206]. If the composite is
prepared by fast heating, infiltration, and rapid cooling, most of the aluminum
matrix is retained unreacted. The matrix can then be hardened by a subsequent
heat treatment at 800°C for 20 hours due to the precipitation of aluminum carbides
and borides. Since the mechanical properties are determined by the A1 matrix, a K,,
of 5-16 MPa m1I2and a flexural strength of 200-680 MPa can be obtained, depend-
ing on the quality and volume fraction of the metallic binder. The Vickers microhard-
ness of 15.7 GPa for a 3 1 v01.-YOA1 composite is improved by annealing to 19.4 GPa.
A re-investigation of the Al-B-C system with special emphasis on B4C-A1cermets
was published by Pyzik and Beaman [90]. The composites consisted of both isolated
and interconnected B4C particles and A1 matrix and only a small amount of ternary
aluminum carbo-borides by fast firing and a post-sintering heat treatment between
600 and 1050°C. Cermets with a high B4C contiguity had the highest strength of
550 MPa if treated at < 600°C. Then the strengths drops to 400420 MPa and
remains constant at higher heat treatment temperatures while structures with iso-
lated B4C particles show a maximum strength of 420 MPa after a 600°C treatment
with a similar decrease at higher temperatures. As expected, the latter composite
with the continuous Al-matrix phase gives the higher fracture toughness of
11.7 MPa m1/2 after a 600°C annealing for 20 hours compared to 8.5 MPa m1I2
for the contiguous B4C microstructures, being best without any heat treatment
(Fig. 88).
Reactions of B4C with metals have also been employed for the bonding of boron
carbide-based ceramics to metallic substrates. Nishiyama and Umekawa [3131 have
studied the intermediate phases in B4C-Fe couples with A1 and Al/Mo interlayers.
Depending on the treatment temperature and thickness of the foils, B12C2A1and
intermetallic phases are formed that allow reactive brazing with a tensile strength
of 55-60 MPa. In a similar way, porous boron carbide of several B:C ratios was infil-
trated by Si-TiSi2alloys with a 1330°C melting point [94,109,233]. After wetting and
0 200 400 600 800 1000 1200

Heat-Treatment Temperature, "C
Figure 88. Effect of heat-treatment temperature on fracture toughness of B4C-A1composites (after
POI).
infiltration, the liquid phase reacts with boron carbide to form TiBz and SIC. There-
fore, these composites were discussed in Section 7.4.4.2.
7.8.2 Titanium Diboride-Based Cermets

TiB2 is a candidate material for wear-resistant parts and cutting tools because of its
high hardness, very high Young's modulus, considerable high-temperature strength
and remarkable chemical stability. The liquid-phase sintering of TiB2 with suitable
metallic binder phases is intended to produce cermets which combine the desired
properties with sufficient toughness. Although TiB2 is stable together with liquid
A1 and Cu, these systems have not been studied seriously for cermets but extensively
for particle hardening of aluminum and copper alloys by TiB2. In the Al-Ti-B
system it is, for instance, still uncertain whether a continuous (Ti,Al)B2 solid solu-
tion exists which is in equilibrium with A1 [89,3 141 or whether AlB2 and TiB2 coexist
as separate phases [315]. Pastor [I 341 summarizes the results of other systems based
on the iron group and other transition metals such as Cr, Zr, Mo, W, and Re. Fe,
Co, and Ni have, however, the highest potential for this purpose.
7.8.2.1 TiB2-Fe Composites

In various studies Fe was investigated as a binder phase for TiB2 [316,317], but
Samsonov [3181 and Kieffer and Benesovsky [ 3 191mentioned, that Fe2Bis generated
during the liquid phase sintering of TiB2-Fe powder blends. Therefore, the fabri-
cation of suitable TiB2-Fe cermets was considered impossible. Other studies by
1500 I
I I
I I
I I
-
1400
L + Ti&
h
0
v -
9
3
+ -
2 1300
1268 'C
a,
Q -
E L + TiB, +a
1200 1183'C -
L + TiB, + y 1167 "C
-
y + Fe,B y + TiB, + Fe@

I I I I I I I I I
1100
0 20 40 60 80 100
Concentration TiB,, (Mol0/o)

Figure 89. Isopleth of TiB,-Fe section (after [320]).
Federov and Kuzma [ 1201 and Shurin and Panarin [ 1211, however, showed evidence
for the existence of a pseudo-binary Fe-TiB2 equilibrium, which is characterized by
a eutectic point at a temperature of 1340°C and 6.3mol.-% TiB2 (Fig. 23). After
Smid and Kny [122] the two phase equilibrium is limited at 800°C by the tie-lines
TiB2-Fe2B and TiB2-Fe (5 at.-% Ti). Ottavi et al. [123,124] finally proved that
not pure Fe is in equilibrium with TiB2 but a Fe solid solution with Ti. The bound-
+
ary tie-line interconnects TiB2 with Fe 0.5 at.-% Ti. Accordingly, a small excess of
boron by result in the formation of FeB2. The first liquid phase in the pseudo-binary
section appears at 1170°C coming from the Fe-FeB2 eutectic, the ternary eutectic is
set to 1240°C (Fig. 24). The contradictions in these experiments have been attributed
by Sigl and Schwetz [116,119,196] to the instability of the system TiB2-Fe in the
presence of C or B4C impurities, which originate from the TiB2 synthesis by the
carbothermic reduction of Ti02-B203or Ti02-B4C mixtures. These impurities
react with Fe and TiB2 and form Fe2B and TIC. Thermodynamic calculations by
Golczewski and Aldinger [320], however, take the y-a transition of iron more
precisely into account and indicate that TiB2 is compatible with Ti-bearing ferrite
only below 900°C (Figs. 89 and 90). The pseudo-binary section TiB2-pure Fe there-
fore contains FeB2 below 1167"C, and between 1 167 and 1268°C solid y- or a-Fe
instead plus liquid. At above 1268"C, the two phase equilibrium TiB2-liquid is
entered (Fig. 89) whereas in the TiB2-Fe (2 at.-% Ti) section FeB2 and y-Fe are
absent (Fig. 90). Although there are still some uncertainties on the phase diagram,
which the synthesis of pure two-phase cermets is based upon, the authors agree that
the presence of oxygen and carbon impurities is detrimental to the wetting behavior
1600 I
1500
L + Ti&
1400
1300
'-
L + TiB, +a
1100 '
0
I I
5
I I I I I I1
I
10 15
2o
Concentration TiB,, (Mol%)
Figure 90. Isopleth of TiB2-(Fe + 2 at.-% Ti) section (after [320]). Note the difference to Figure 89
concerning the stability of CL 4 y Fe and the appearance of the liquid phase.
and responsible for the presence of the embrittling but hard Fe2B phase, which also
controls the sintering behavior and thus the properties.
After the development of cemented borides with a Fe matrix by Funke et al. [316]
and Funke and Yudkovskii [32 11, this cermet system was recently intensively studied
again by Yuriditsky [115], Sigl and Schwetz [116, 1961, and Jiingling et al. [322,323],
and, in connection with an enhanced European research activity, independently by
Ottavi et al. [123, 1241, Ghetta et al. [117], Pastor et al. [324,325], and Sanchez et al.
[326].
Liquid Phase Sintering of TiB2-Fe Composites

A disadvantage of TiB2-Fe composites is their delicate densification behavior,
which has been attributed to the oxygen impurities present in commercial TiB2
powders. At solidus temperature, oxide impurities may give rise to evaporation
and recondensation reactions involving volatile boron suboxides, which result in
an exaggerated grain growth of TiB, without densification. Coarsening by means
of vapor phase or surface reactions consumes surface energy and, therefore, sinter
activity. Consequently, much effort was put into the removal of oxide contamina-
tion from TiB2 powders, e.g. by reduction with BC13 [263,327].
As already mentioned before, oxides and carbon impurities are also responsible
for the wetting behavior of the liquid. Figure 91 illustrates both complete and
imperfect wetting of TiB2 by liquid iron. Thin Ti02 layers covering parts of the
TiB2 grains are assumed to account for the poor wetting [116,196]. Experiments
900 7 Boride-Bused Hard Materiuls
+
Figure 91. SEM micrograph of an 8Ovol.-% TiB, 2Ovol.-% Fe cermet sintered at 1600°C. Note
both the incomplete and complete wetting of TiB, by liquid Fe.
by Ghetta et al. [328] have demonstrated that the spreading of liquid Fe in fact wor-
sens with increasing oxygen content of TiB2 substrates (Fig. 92) whereas powders
with < 0.26 mass-% oxygen show a wetting angle of < 5" after 10 min soaking time.
Generally, transition-metal diborides should be wetted well by liquid metals due
to their missing gap between the valency band and the conduction band whereas
oxides exhibit bad wetting behavior due to their partially ionic and partially
covalent character in bonding [329-33 11. Although an improvement of wetting by
liquid Fe may be obtained by addition of Cr or Ni [332], breaking up the Ti02-
layer is thus considered the key to well-processible TiB2 powders.
80 7 7 - r -1
Oxygen content in TIB,
70
0
a- 60
50
40
30
20
0 500 1000 1500 2000
Time, sec.
Figure 92. Oxygen contamination affecting the wetting behavior of Fe liquid on sintered TiB, at
1300°C (redrawn after [117]).
7.8 Cemented Borides 90 1
Figure 93. SEM micrograph of a TiB2-Fe cermet with a nominal composition of 80voI.-%
TiB, + 20vol.-% Fe. Note the existence of Fe,B in the binder phase (after [119]).
It also appears that the impurities in commercial TiB2 promote the formation of
iron boride, Fe2B. A TiB2-Fe alloy with a typical two-phase microstructure of the
binder, i.e. with both Fe and Fe2B making up the matrix between the TiB2 grains,
is shown in Fig. 93. Fe2B is a hard but fairly brittle compound [333] and its forma-
tion consumes a substantial fraction of the ductile Fe phase such that Fe2B is
believed to embrittle the composite seriously. Avoiding that phase has consequently
been a major goal for developing tough TiBz cermets.
Commercial TiB2 powders are typically produced by the carbothermic reduction
of Ti02 with B2O3 or B4C. Both reactions
Ti02 + B 2 0 3+ 5 C + TiB2,+5 CO (41)
and
2Ti02 + B4C+ 3 C + 2TiB2 + 4CO (42)
are being industrially utilized. A typical analysis of the as-received TiB2 is listed in
Table 10.
Since this powder originates from the B4C-processingroute, i.e. Reaction (41), it
is not surprising that about 90% of the total carbon is present as boron carbide and
only 10% is free carbon. Assigning the oxygen content to particular compounds is
not unambiguous, but it is estimated that approximately 1/2 of the oxygen is present
as Ti02 and B203,respectively. Nitrogen is completely bonded as TIN, as BN is
usually not found. It is therefore concluded that B4C, B2O3, free carbon, and
Table 10. Chemical composition of the as-received TiB2 and Fe powders
Powder Spec. Total 0 Total C Total N B4C Free C TiO, BZ03

surf. [weight-%] [weight-%] [weight-%] [weight-%] [weight-%] [weight-%] [weight-%]
Im2/nl
TiB 3.2 2.1 0.20 0.05 0.85 0.02 2.6 1.5
Fe 0.5 0.2 0.05 - ~ ~ - ~
Table 11. Chemical compatibility of the compounds in TiB2-Fe powder mixtures
Fe C B4C TiOp B2°3
Ti02, generally make up the critical impurities in the TiB2 powder. More carbon is
introduced to the powder blend when organic binders are added and it should be
noted that the oxygen content increases significantly upon milling [118].
The key questions for a successful liquid phase sintering of TiB2 with iron are,
therefore, (i) how to break up the Ti02-films on the TiB2 grains and (ii) how to
avoid the formation of Fe2B.
+
Sigl and Jiingling [119] have shown that powder compacts of TiB2 20 v0l.-% Fe
heated to 1000°C contained Fe2B and Ti203.Notably, their concentration did not
increase significantly at temperatures up to 1600"C, suggesting that the major part
of the reaction had been completed at that time already. Since Fe2B and Ti203
already appear at low temperature, solid-state reactions are considered responsible
for the formation of this compound. Of particular interest are the reactions of Fe
with the boron-containing impurities, i.e. with B4C and B203.According to Table
11 the following reactions are candidates to develop Fe2B:
B4C+ 8 Fe --+4Fe2B + C (43)
2 B203+ 7 Fe 2Fe2B + 3 FeO
---f
(44)
The change in Gibbs Free Energy, AG, has been calculated for Eqs. (43) and (44)
with the data from Janaf [334]. The results are plotted versus temperature in
Fig. 94. Obviously a reaction between Fe and B203is unlikely in the whole tempera-
ture range of interest, which has been confirmed by corresponding model reactions.
It is, however, well known that Fe and B4C react heavily even at low temperatures
forming Fe2B and free carbon [335]. Due to the favorable properties of Fe2B, this
fact is commercially utilized for depositing thin layers on steel surfaces [336].
As mentioned above, borides can be synthesized by the carbothermic reduction of
B203 with carbon and a metal or a metal oxide [337], e.g. according to Eq. (40).
Since free carbon is present in the as-milled powder mixture and also develops during
Reaction (43), the following process must also be considered a potential source of Fe2B:
+
B203 4Fe +3C t 2Fe2B + 3 CO. (45)
Although the AG, which is plotted as a dashed line in Fig. 94 appears to favor
Reaction (45) above 800"C, it is still considered unlikely. According to Sigl and
Jiingling [119], the following steps rather than the gross Reaction (45) are proposed
to operate instead
a,
,f -400
-800 L -
600 800 1000 1200 1400 1600
Temperature, "C
Figure 94. The change in Gibb's free energy for reactions yielding Fe2B at 0.1 mbar total pressure.
B20 + 3CO 2B + 3C02

2B+4Fe + 2Fe2B
3COz+3C+6CO
Given the CO/CO2 equilibrium according to Eq. (48), the Gibbs Free Energy for
Reactions (45) and (47) can be calculated and is plotted in Fig. 94 at a total pressure
of 0. I mbar. Though Fe and B would favorably react, CO is unable to reduce B 2 0 3
below 1000°C and therefore the carbothermal formation of Fe2B is operating only
beyond 1000°C. This hypothesis has been confirmed experimentally: While a B203-
Fe-C powder blend with a composition according to Eq. (45) does not change its
phase composition after annealing the mixture at 1000°C for 1 h, the X-ray diffrac-
tion of identical samples heated up to 1600°C shows ample evidence for Fe2B. This
suggests that processes involving liquid iron can play an important role during the
formation of Fe2B as elaborated below.
In the regime of liquid-phase sintering TiB2 is dissolved in the Fe melt according
to
TiB2 {Ti} + 2 {B} (49)
where the brackets { } denote the dissolved state of an element in liquid Fe.
Obviously impurity compounds such as T i0 2 , Ti203, C, B203, TIN, and Fe2B
will also dissolve in the liquid. During cooling some of these compounds reprecipi-
tate from the melt in their original composition, e.g. TIN or T i2 0 3and hence do not
influence the equilibrium concentration of Ti and B in the liquid. Species such as
carbon, Ti02, and B 2 0 3which introduce a lot of {C}, {0}, and {B} but only little
{Ti} into the melt do, however, precipitate as compounds that are richer in Ti than
the previously dissolved species, e.g. as
Fe Fe,Ti FeTi Ti
-Ti [at.-%] +
Figure 95. Isothermal section of the Fe-Ti-B system at 1000°C. The point indicates the cermet com-
position and the arrow the change in composition if the titanium content in the liquid phase is
decreasing.
{Ti} + 2 {B} + TiB2 (50)

{Ti} + { C} + TIC (51)
2 {Ti} + 3 {0}+ Ti203 (52)
Thus, these processes extract a lot more {Ti}from the melt than has been introduced
-
during the dissolution. Consequently, the solubility product (uTi a;) being constant
at a given temperature forces the composition of the liquid to shift towards low {Ti)
and high {Bj concentrations upon cooling. Finally, a {Bj-rich Fe melt is left behind.
This conclusion is supported by the observation that a liquid phase is stable down to
1 150°C[335], i.e. the eutectic temperature of Fe-Fe2B,although the TiB2-Feeutectic
solidifies at 1350°C. Eventually, the remaining melt crystallizes as a mixture of Fe
and Fe2B (Fig. 93). An experimental confirmation of this conclusion was obtained
by Sigl and Jungling [119]. These processes shift the composition of the cermets from
their intended place in the (TiB2 + Fe) two-phase field into the (TiB2 Fe Fe2B) + +
three-phase equilibrium of the Fe-B-Ti ternary system, either by introducing B or by
extracting Ti (Fig. 95). Thus the Ti concentration in TiB2 cermets is a variable to
be strictly controlled during sintering, very similar to the C content of cemented
carbides, which must be adjusted well to avoid graphite or q-phase formation.
Figure 91 suggests that liquid Fe cannot penetrate in between the TiB2 grains,
unless the Ti02 layer is broken up. Total wetting is nevertheless feasible. Figure 96
indicates that this process is closely related to the precipitation of particulate Ti203.
It is generally accepted that the reduction of Ti02 involves various titanium subox-
ides, i.e. Ti305,Ti203,and TiO, until Ti is finally obtained [334]. The compound
that actually evolves from this process depends on how well the reducing agent
can overcome the affinity of oxygen in the Ti oxides. Since the TiB2 powder of
this study contains only free carbon as a suitable reducing agent, reactions such as
Figure 96. Optical micrograph of a TiB2-Fe cermet containing Ti203emerging from powder con-
taminations.
2 T i 0 2 + C 4Ti203 CO + (53)
need to be studied. The Gibbs free energy for Reaction (52) is plotted in Fig. 97 as a
function of temperature at a total pressure of 0.1 mbar. Though carbon should be
able to reduce T i 0 2 at temperatures above 850°C according to the thermodynamic
calculations, a reaction between a thin oxide film and solid carbon is considered
unlikely. Instead, carbon monoxide rather than free carbon is believed to deoxidize
T i 0 2 according to
2Ti02 + CO 4 +
Ti203 C02 - (54)
since the driving forces for Reactions (52) and (53) are identical. A further reduction
7
Y
)r
F
a
5
a
9
LL
B
0
(3
Temperature, "C
Figure 97. The change in Gibb's free energy for the reduction of titanium oxides by carbon monox-
ide at 0.1 mbar total pressure.
of Ti203according to
+
Ti203 CO + 2TiO + C02 (55)
is also supported by thermodynamic arguments (Fig. 97), but CO seems unable to
reduce T i 0 to Ti at a pressure of 0.1 mbar.
Since experiments have shown that TiB2 powders with more than 2 weight-%
oxygen can be sintered to full density [116, 196,3221 and that Ti203 particles preci-
pitate during sintering, it is considered that the free carbon in the as-received TiB2
powder and the carbon being set free from the reaction of Fe with B4C generates
CO, which eventually breaks up the oxide layers on the TiB2 grains. The result
that Ti203appears instead of TiO, as may be inferred from Fig. 97, is attributed
to the small quantity of carbon normally available. Principally, a carbon quantity
sufficient to form either Ti203 or T i 0 should promote the wetting and consequently
the sintering behavior of TiB2 powders. This conclusion is supported by Fig. 96,
which suggests that Ti203 particles hardly disturb the wetting by liquid Fe.
Excess carbon can, however, lead to the formation of Fe2B via the liquid-phase
processes described above. Since small quantities as in the previous case will be
used up during the reduction of Ti02, the carbothermic formation of Fe2B from
Fe and B203according to Eq. (45) cannot occur. Hence it is concluded that Fe2B
originates mainly from the reaction of Fe with B4C and that the gross reaction
B4C + 2 Ti02 + 8 Fe + 4Fe2B + Ti203 + CO 7' (56)
operates at temperatures below 1000°C.Though a good part of Fe2B is due to solid-
phase reactions, it is anticipated that B203 will also give rise to both Fe2B and
Ti203. However, the role of B203is as yet not fully understood.
In conclusion, deoxidizing the thin TiOz films appears to be an absolute pre-
requisite for the sintering of TiB2 with liquid iron. Though carbon monoxide
cannot fully reduce Ti02 at the sintering conditions (0.1 mbar), the formation of
particulate lower oxides such as Ti203 seems to be sufficient for supplying a good
wetting behavior.
Kinetics of Liquid-Phase Sintering of TiBz with Fe

The sintering kinetics and phase development during sintering were investigated by
Jungling et al. [322]who demonstrated that Fe2B already forms at low temperatures,
i.e. before shrinkage starts (Eq. (55)). The deoxidation of TiB2 grains and the simul-
taneous formation of Fe2B contribute significantly to the densification behavior of
TiB2-Fe-materials, i.e. nearly full densities (>98.5% T.D.) were achieved by pres-
sureless sintering. However, the formation of Fe2B consumes a considerably high
amount of the Fe binder, which deteriorates the fracture toughness of the composite
seriously. For these reasons Jungling et al. [322] and Sigl and Jungling [338] sug-
gested the addition of Ti to balance the excess boron in order to obtain materials
with a ductile binder:
2 Ti02 + 2 Ti + B4C + 2 TiB2 + Ti203 + CO (57)
1800 4
9 1400 0
2i
3
5 1000 -4
8
a, A-
Q d
600 -8
200 -12
0 100 200 300
Time, min
Figure 98. Densification behavior of model alloys with 30 v01.-% binder with and without Ti addi-
tion (after [340]).
This reaction as well as Eq. (56) were confirmed by thermodynamic considerations.

Ghetta et al. [328] and Ottavi et al. [339] concluded from wettability studies that the
addition of Ti deoxidizes the TiB2 grain surface by the formation of a parasitic
Ti(O,C,N) phase. Nevertheless, the sintered density of a material with Fe2Ti addi-
tion was rather poor ( ~ 8 9 %of theoretical density). The addition of NdNi5
enhanced the density to 96.7% of theoretical density and resulted in a parasitic
N d 2 0 3phase.
Jungling et al. [323] report on model experiments to study the sintering kinetics of
+
TiB2 with 20 v01.-YOcarbonylic Fe and 20 v0l.-YOcarbonylic Fe 7 mass-% Ti up to
1700°C in 0.1 mbar Ar. Sintering parameters were optimized for materials with 15
and 20 v01.-YOFe and 20 vol.-% Fe-Cr-Ni binder and for materials with 5-20 v01.-YO
Fe-Cr-Ni binder phase with Ti addition to prevent the formation of the Fe2Bphase.
Materials with Ti addition were pressurelessly sintered at 1650°C to a density of
about 95% T.D. followed by a hot isostatic pressing step at 1460°C with a pressure
of 100 bar Ar.
Figures 98 and 99 illustrate the shrinkage behavior and densification rates of the
alloys with and without Ti additions. Soon after the spontaneous start of densi-
fication, the sintering rate decreases drastically, but eventually begins to increase
again. Without Ti addition shrinkage starts at 1140°C but in the alloy containing
Ti, sintering begins at 1250°C. After this temporary decrease, the densification
rate increases in both materials above 1320°C. Densification comes to an end
after 2 h isothermal sintering at 1700°C.
Unlike the material with pure Fe, the alloy with F e t - 7 mass-% Ti does not
develop any Fe2B. Again Ti203is formed at 1000°C but the intensity ratio Ti203/
TiB2 is smaller than without Ti addition and increases with increasing temperature.
Additionally, X-ray diffraction presents evidence for the formation of very small
amounts of TIC in the sintered sample.
Figure 100 displays the density as a function of the sintering temperature
without holding time. Samples with 15 or 20vol.-% Fe show similar densities in
3 1800
.-r 2 1400 $I
-.
s
E
n!
3
2
1000
a,
Q
-1 200
0 50 100 150 200 250 300 350
Time, min
Figure 99. Densification rate of model alloys with 3Ovol.-% binder with and without Ti addition
(after [340]).
the temperature range of 1500-1700°C and reach nearly 99% T.D. The addition of 7
mass-% Ti results in a strong decrease in the density of as-sintered materials, but at
> 1650°Ca density of >95% T.D. and closed porosity are obtained. Further densi-
fication can be achieved by subsequent hot isostatic pressing. Materials with 5, 10,
and 20 v0l.-YOFe-Cr-Ni-binder with 7 mass-% Ti addition nearly approach their
theoretical densities by a post-HIP treatment at 1460°C with a pressure of 100
bar Ar (Table 13).
The spontaneous shrinkage for materials with Fe2B at 1140°Cis explained by the
formation of a liquid phase and a subsequent rearrangement of particles. This con-
clusion is compatible with the eutectic temperature of 1177°Cin the system Fe-Fe2B
--
"1
100 I --I , , , r
Grade F + carbanyllc Fe
d
95-
c
c
90 - i
+20 VOl%
1450 1500 1550

1
1600
' 1650 1700 1750
Sintering Temperature, OC
Figure 100. Densification behavior of TiB2 + 15 or 20 v01.-% Fe cermets (after [340]).
Table 12. Solid-state and vapor-phase reactions during sintering.
Without Ti additive With 7 mass-% Ti Eq .
B4C + 8Fe + 4Fe2B +C ~
(57)
- B4C + 2Ti --i 2TiB2 + C (58)
c + c02 + 2 c o c + co*+ 2co (59)
2 TiOz + CO + +
Ti203 C 0 2 2 Ti02 + CO + Ti203+ COz (60)
[341]. In this way Fe2B contributes a lot to the excellent densification behavior of
these materials but it also consumes a major part of the ductile Fe-binder phase.
In materials without Fe2Bthe beginning of shrinkage is delayed. A first maximum
in densification rate appears at 125OoC,about 90 K below the eutectic temperature
of Fe-TiB2. Again, this densification peak is attributed to the formation of a
liquid phase, which coincides with the eutectic Fe-Fe2Ti at 1289°C rather than
with the Fe-TiB2 eutectic [342]. This can be explained by the solution of some Ti
in the binder during heating. As proved by X-ray analysis, the addition of 7 mass-
% Ti fully prevents the formation of Fe2B. Smaller amounts of Ti yield
corresponding quantities of Fe2B. Table 12 compares the particular reactions of
materials with and without Ti additions.
After initial particle rearrangement the densification rate decreases in both
materials. It is accelerating again when a sufficient amount of liquid phase is
formed due to the eutectic Fe-TiB2 at about 1320°C. During cooling the liquid
phase finally solidifies at 1140°Cfor samples with and at 1250°Cfor samples without
Fe2B as proved by exothermic reactions producing the small expansion peak in
Fig. 99. The last expansion peak at 800°C is due to the a 4 y transformation of
Fe. As expected, this peak is much higher for the composite that does not contain
Fe2B.
In conclusion, the formation of FeB2 can be avoided by the control of powder
impurities, sintering atmosphere, and compensation of excess boron and carbon
by titanium addition. The wetting behavior depends strongly on oxide surface
layers and is excellent if surface contamination of TiB2 is entirely removed. The sin-
tering temperature and Fe addition must ensure that a suitable amount of liquid
phase is available at the lowest possible vapor pressure. Since the reducibility of
T i 0 2 and the vapor pressure of liquid Fe determine the atmospheric pressure, the
interval (0.01-0.1) mbar < p < 1 mbar is considered best suited. The optimum
amount of liquid phase is generated at 1450 < T < 1650°C if grain growth is
taken into account as another limiting factor. Accordingly, a feasibility diagram
of suitable sintering conditions can be constructed (Fig. 101) showing the optimum
environmental parameters.
Microstructure and Properties of TiB2-Fe Composites

Since the eutectic concentration in the quasi-binary TiB2-Fe system with 14 v01.-%
TiB2 is considerably closer to the metal corner than in the similar WC-Co system
(32 v01.-YO WC), a much smaller amount of liquid phase is generated upon sintering,
b
-.
2
Table 13. Properties of optimized TiB2-hard metals. ICL: Indendation-crack-length method, NB: notch-beam method.
Ptx
g
6.
Starting powders TiB2 + Fe TiB, + Fe-Cr-Ni TiBz + Fe-Cr-Ni + 7 mass-% Ti Q
Binder composition cc-Fe + Fe2B y-Fe-Cr-Ni + Fe2B y-Fe-Cr-Ni

3%.
%
2
.
Binder content [vol.-'Yo] 5 15 20 20 5 10 20 6.
2
Density of pressureless sintered material [% T.D.] 97.8 98.8 99.1 99.2 -95 -95 %95
Density of post-HIP [Yo T.D.] - 100 99.1 99.2
3-point bend strength [MPa] 610 760 720 790 900 1010 950
Young's Modulus [GPa] 510 465 465 465 515 510 465
Hardness [HV,,] 25°C 2260 1830 1510 1610 2040 1800 1450
Hot hardness [HV2]200°C - 1130 1560 1270 1150
Hot hardness [HV,] 400°C - 870 1290 990 850
Hot hardness [HV,] 600°C - 730 1030 810 740
Hot hardness [HV2]800°C - - 480 660 580 500
K,,/ICL [MPam] 5.9 6.9 7.9 8.2-8.7 7.3-8.1 8.1-9.5 12.4
KI,/NB [MPam] 5.5 9.3 10.0 11.4-1 1.9 7.6 9.0 14.0
10'
1 ' 1 ' 1 ' 1 2
reduction of Ti0,
1oo
-ki grain
growth
E
Y 10' -
lo3
1300 1400 1500 1600 1700 1800
Temperature [ "C]
Figure 101. Optimum parameters for the pressureless sintering of TiBz-Fe cermets limited by grain
growth, evaporation, suboxide formation, and suitable liquid phase content.
which makes densification more difficult. A simple increase in temperature cannot

satisfactorily balance the lack of liquid because it is accompanied by accelerated
coarsening of TiB2 due to Ostwald ripening [I 161. The volume fraction of binder
phase thus ranges between 10 and 30%. A typical microstructure is very similar
to that of WC-Co hard metals. Euhedral TiB2 particles are embedded in a
continuous Fe matrix. The densification mechanisms are typically dissolution and
reprecipitation as well as coalescence, i.e., rearrangement and intergrowth of
particles with common faces of the same orientation. The latter mechanism is
active if the volume fraction of liquid exceeds 30% but may result in the growth
of elongated platelets.
The residual porosity after pressureless sintering between 1500°C and 1800°C
depends upon the initial liquid phase composition. At 15OO0C, 88% of the
theoretical density has been obtained for the TiB2-Fe system at lSOO°C),
whereas at 1450°C Ti addition results in 98% and combined Ti-Nb additives
result in 96.7%. Hot-pressing and hot isostatic pressing yield densities >98% at a
lower binder content.
Figure 102 shows the microstructure of a plasma etched sample with 2Ovol.-YO
binder. Besides TiB2 with its hexagonal grain shape and Ti203a binder consisting
of Fe and Fe2B is present. A Fe-Cr-Ni binder phase diminishes the grain sizes nota-
bly. In materials with 7 mass-% Ti addition to the Fe-Cr-Ni binder the FeB2 phase is
absent. The total carbon content of samples without Ti addition strongly decreases
during sintering from 0.11 YO(starting powder) to 0.0 1YO(sintered compact). The
oxygen content slightly decreases from 2.8% to 2.4%.
Table 13 summarizes some properties of TiB2 with 15 and 2Ovol.-% Fe-Fe2B
binder, 20 vol.-% Fe-Cr-Ni-Fe2B binder, and 5-20 v01.-YO Fe-Cr-Ni binder.
Samples with a two-phase binder reach densities of about 99% T.D. by pressureless
sintering. However, alloys with a single-phase binder approach no more than 95%
T.D. by pressureless sintering, yet can be densified to near theoretical density by a
912 7 Boride-Rased Hard Materiuls
Figure 102. Plasma-etched microstructure of a cermet with 20 v01.-% cc-Fe-Fe2B-binder.
post-HIP treatment. As expected, samples with a low binder content are harder than
those with high binder fraction, provided the binder composition remains
unchanged. The observation that materials with Fe-Cr-Ni-Fe2B binders are
harder than materials with Fe-Fe2B binders is attributed to the smaller grain size
of the former materials (Fig. 103). Simultaneously, the fracture toughness increases.
Thus avoiding the formation of Fe2B strongly increases the fracture toughness but
decreases the hardness only moderately, such that the combination of hardness and
fracture toughness is clearly improved [343] compared to WC-Co-based hard
metals.
The bending strength does not vary significantly with increasing binder content
(Fig. 104) but the presence of FeB2 results in considerably lower values. This
not only attributed to the brittleness of this phase but is also due to the residual
2500 1
s> 2000
r
ui
u)
a,
c
E 1500
m
I
conventionalcemented carbi
-
10001' " " " " I ' " ' I ' " ' 1
0 5 10 15 20
Fracture Toughness hC,M P a h
Figure 103. Hardness and fracture toughness of TiB2-Fe cermets in comparison to conventional
hard metals (after [340]).
a 1200
a
.- 1000
a
?t 800
600
a,
L 1
400 1 I
3 Fe-Cr-Nt-Fe,B
0 5 10 15 20 25
Binder Content, vol%
Figure 104. Flexural strength dependence on binder volume of TiB2-Fe cermets with and without
FezB (after [343]).
2 % porosity. A post-HIP treatment may overcome this problem and also yields
a lower scattering of data. Additions of metals such as Mo, Cr, Ni, and Co to
the Fe matrix may be used to fabricate composites with improved mechanical and
corrosion properties. Figure 105 shows significant variations in the bending strength
with increasing amounts of Mo in the binder phases of different volume fractions
[115].
Figure 106 illustrates the hot-hardness of composites with various binder content,
again in comparison to WC-Co and WC-Fe-Co-Ni hard metals [343]. It is evident
that TiB2-based materials with a small amount of binder are superior in the entire
temperature range measured.
Figure 105. Variation of strength with Mo content in the binder phase of TiB,-Fe cermets (after
[115]). Binder volume: A, 12.5vol.-%; M, 15vol.-%; 0 , 17.5vol.-%.
25001 I I I I I I I I I I I I I I I I
1 2000
uj-
1500
a,
v)
C
E
ca
T
1000
0 200 400 600 800

Temperature, "C
Figure 106. Room-temperature hardness HV,, and high-temperature hardness HV2 of TiB2-Fe
cermets and conventional hard metals (after [340]).
7.8.2.2 TiBz-Ni, Co Composites

Transition metals such as Co and Ni are useful for liquid-phase sintering of TiB2-
type borides causing rapid densification and grain growth by Ostwald ripening,
but they react chemically to form M,B,-type phases or even more complex ternary
phases that are very brittle and possess a comparatively low melting point < 1500°C.
No metal-reinforced composites containing solely TiB2 as the hard phase can thus
be produced except where reactions can be at least partially avoided by fast heating
during hot-pressing. In contrast, Co- and Ni-based alloys can be successfully
improved for wear resistance by the incorporation of TiB2 and CrB2 particles if
reaction layers of lower hardness can be tolerated.
In the case of Ni-bonded TiB2, a ternary 7-phase with the composition Ni21Ti2B6
forms by the dissolution of TiB2. At 800°C, the z-phase is in equilibrium with Ni,
Ni3B, Ni3Ti, and TiB2, as shown in the isothermal section in Fig. 25 [125].
During sintering or hot-pressing, the 7-phase is generally not obtained. The forma-
tion of 7 may be suppressed either by the presence of T i 0 or Ti02 [344], or for kinetic
reasons by fast firing and quenching. Hence the residual matrix phase consists
mainly of Ti-containing Ni3B solid solution [218,344,345]. Since other Ni-borides
such as Ni2B and Ni3B4 and even metallic Ni are found after hot-pressing at
1600°C [94], equilibrium conditions are obviously not easy to obtain.
Typical metal contents required for a successful liquid phase hot-pressing of
TiB2 are 5-25 mass-% (i.e. 2-12 at.-%o) Ni or Co. In order to avoid reactions
consuming TiB2, the borides of Ni or Co have also been used. By this method,
the sintering temperatures have been decreased from 2100°C to 1400°C [134,
277, 294, 295, 3461.
Figure 107. Optical micrograph of a TiB2-Ni composite, pressurelessly sintered at 1600C. Note the
incomplete wetting of TiBz by liquid Ni and the formation of Ni borides in the matrix phase.
The liquid phase intensifies the mass transport but causes an accelerated grain
growth. The microstructures of composites prepared by liquid-phase sintering are
similar to those of hard metals. The TiB, particles form a rigid skeleton of faceted crys-
tals whereas the binder, e.g., Ni3B, Ni2B, Ni3B4,or comparable compounds of Fe, Cr,
or Co, is the matrix phase. The TiB2 grain size usually exceeds 20pm (Fig. 107).
Depending upon the wetting behavior, which is influenced by the surface oxidation
of the hard material phase, round pores may accumulate at particle/matrix interfaces
or close to triple junctions that have not been completely infiltrated by the liquid phase.
Moreover, the evaporation of Fe-, Co-, or Ni-borides may cause entrapped gas pores.
Hence, hot-pressing is still required for a homogeneous distribution of the liquid phase,
particle rearrangement, and complete removal of the residual porosity. In contrast to
hard metals, the matrix phase is very brittle, e.g., the K,, of Ni3B equals 1.4-1.9 MPa
m"2 [347], and hence does not improve the mechanical properties.
Nishiyama and Umekawa [348] report on sintering and wear application of a TiB,
cermet with Ni7Zr2 binder. Pressureless sintering between 1450 and 2100°C in
hydrogen results in an incomplete melting of the alloy added with wetting angles
of around 90". The phases detected are Ni4B3, Ni7Zr2,and solid solutions thereof
with Ti. Since TiB2 forms a rigid skeleton the hardness HV, reaches 30-33 GPa at
binder contents of < 8 mass-%, dropping almost linearly to 17 GPa with 30 mass-%
Ni7Zr2 addition. Due to the toughness of the matrix phase, the 3-point bending
strength improves from 500-600 MPa at < 5 mass-% binder to 850-900 MPa with
> 15 mass-% of additives.
7.8.2.3 Transition-Metal Diboride Cermets with Co, Ni, Cr, Mo, and W
The research for high-strength/high-toughnesscomposites in the TiB2-transition
metal systems resulted in many attempts to avoid brittle matrix phases and to
introduce hardening particles while making use of the opportunity of liquid-phase
-
:
x
I
1 0 0 0 - :;9 0 .
0
0
TiB,+lrn%CoB
TtBz+5rn%TaB,+lrn%CoB
TiB +, 5 rn% W B
, st 1m % C o B
2 200-1
m
0 5 10 15 20
A v e r a g e groin size (prn)
Figure 108. Strength-grain size relation in TiB,-COB-based composites (drawn according to data in
[2931).
sintering. Pastor [I341 reports on the manifold of diborides that have been sintered
with Fe, Co, Ni, Cr, Mo, W, and even Re additives, yielding closed porosity or
almost total density. Unfortunately no phase compositions were presented in
most cases. In order to avoid brittle ternary phases, many liquid-phase sintering sys-
tems started with combinations of binder metals, sometimes even with their low
melting boride phases, which will be reviewed in the following paragraph.
Hot-pressed composite materials developed from the more complex systems of
the type TiB2-M1B2-M" with M' being Hf, V, Nb, Ta, Mo, or Mn and MI' being
sintering additives such as Co and Ni, exhibit bending strengths between 850 and
1000 MPa, which are due to the grain growth inhibiting influence of the 1-5 mass-
% of M"B2 particulates (Fig. 108) [293-2961. During liquid-phase sintering in a
Co- or Ni-boride melt, both TiB2 and M' are partially dissolved and reprecipitated
as a solid solution. The effect of grain growth retardation as well as of strength and
hardness increments is attributed to stresses at the TiB2/(Ti,M')B2 phase boundaries
generated by the mismatch of the lattice parameters between the unreacted TiB2
acting as a nucleus and the epitaxially precipitated (Ti,M')B2 solid solution. In
the case of a TiB2-5 mass-% W2B5/TaB2material with 1 mass-% COB binder the
lattice strain ranges between 9 x and 14 x depending on the hot-pressing
temperature [293]. An improved sintering behavior of this material was obtained by
the addition of 1.7% TIC, which reduced the porosity from 0.3-0.7 to 0.1-0.2 vo1.-YO
after hot-pressing at 1500°Cand by intensive ball milling resulting in an average par-
ticle size of 1 pm, but increased oxygen contamination. Watanabe and Shoubu [297]
reported the formation of a (Ti,Ta)(C,O) solid solution, which is considered to initi-
ate the improved densification resulting in a flexural strength of l000MPa. The
strength-grain size correlation in Fig. 108 can be interpreted in the way that the frac-
ture is flaw-controlled at average grain sizes < 7-8 pm and microcrack-controlled at
> 8 pm.
In a similar multiphase system, transition metal carbides were used as additives for
pressureless sintering of TiB2, yielding composites of binary and ternary borides
[218,296]. Attrition milled powder mixtures of TiBz with 3-10 mass-% Co or Ni
and 20-35 mass-% WC have been sintered in a vacuum at temperatures between
Figure 109. SEM micrograph of a (Ti,W)B2-W2NiBz(w)-Ni,B composite. Light areas: W2NiB2;

intermediate: Ni3B; dark: (Ti,W)B,.
1500 and 1700°C for 60 to 120min yielding 98-99% density. Densification starts
above 980°C due to the formation of a liquid phase in the Ti-B-Co/Ni system
(Fig. 26). At this early stage, a rigid skeleton of TiB2 and WC develops. Due to dis-
solution and reprecipitation, a (Ti,W)B2 solid solution grows on the residual TiB2
particles. Subsequently, crystals of o-phase form with compositions of WCoB or
W2NiB2.Upon cooling, the residual liquid phase crystallizes as C-and Ti-enriched
Co3B or Ni3B solid solution. A typical microstructure is shown in Fig. 109. Sintering
at 1700°C for 2 h yields an average particle size of 0.8 pm, resulting in a flexural
strength ob= 600-680 MPa. The K,, of 6.5-7.5 MPa m1/2is mainly attributed to
crack deflection because of the weak Co- or Ni-boride intergranular phases.
As already shown, an interesting characteristic of the mixed diborides is the so-
called core-rim structure, which is well known from complex cermets based on
(Ti,Ta . . . )(C,N) compounds. During sintering or hot-pressing of the particular
carbonitride powder blends solid solutions are formed at the phase contacts but
the annealing time is insufficiently long to obtain equilibrium conditions. Conse-
quently, interdiffusion is stopped, an unreacted core of, e.g., Ti(C,N) remains
whereas an outer layer, the rim, consists of, e.g. (Ti,Ta)(C,N) solid solution with
a very distinct boundary to the residual host crystal. Exactly the same effect is
observed in borides of all kinds of composition if sintered for a not appropriately
long time (Fig. 1 lo). In case of liquid-phase sintered borides, this layer is formed
by epitaxial precipitation of the thermodynamically more stable solid solution
onto a nucleus of undissolved starting material. Recently, Telle et al. [349] observed
dislocations at the interface of TiB2-(Ti,W)B2solid solutions in hot-pressed TiB2-
WB2 composites (Fig. 11 l), indicating that stresses are generated at the epitaxial
interface as considered by Watanabe and Kouno [293].
Another approach to fabricate metal-matrix-based boride and carbide compo-
sites according to the Lanxide process starts with reactive blends of B4C and Ti
or Zr metal. Upon conversion to TiB2 or ZrB2, respectively, a large release of
heat is observed, which can easily lead to partial melting of the composites. Depend-
ing upon the starting composition, residual metallic Ti or Zr, or B4C may be found
after reaction. Interesting microstructures can also be obtained if TIC or ZrC are
added as fillers [350].
918 7 Boride-Bused Hurd Material5
Figure 110. Core-rim structure of (Ti,W)B2 solid solution on a TiB, host crystal.
Similar metal-matrix composites with B4C, Sic, and TiB2 as fillers have been
fabricated by the so-called Lanxide- or Dimox-process (direct metal oxidation)
where an Al- or Ti-based liquid mixed with ceramic particles -preferably of whisker
or platelet shape - is slowly converted in air, oxygen, or nitrogen to alumina or tita-
nium nitride, respectively [351-3531. This self-propagating reaction yields columnar
crystals of the oxide or nitride phase, with B4C, Sic, or TiB2 inclusions and residual
metal-filled channels, which contribute significantly to the strength and tou hness.
SiC-A1203compositeshave an excellent fracture toughness of 8-1 5 MPa rn1Ig and a
flexural strength of 500-800 MPa. They have also been demonstrated to be highly
resistant against erosive wear [354].
Figure 111. Dislocations at the boundary (Ti,W)B,/TiB,.

7.8.3 Cemented Ternary Borides

As discussed before, hard metal-like composites can be prepared by pressureless
sinterin of ternary borides with Fe, Ni, or Co melts. Materials with z-phase
(M:,M2I? B6, where MI = Fe, Ni, or Co, and M" = Zr, Hf, Nb, Ta, or W with MI
as the matrix phase) have not been developed for technical use but Ni-based
alloys with z are in applications as wear- and corrosion-resistant coatings on
steels [355]. The z phase is also used for the improvement of the creep resistance
of Ni-based superalloys.
P
The other kind of ternary hases, cp and o,have M'M'IB and MiM"B2 stoichio-
metry, respectively, where M = Cr, Mo, Ta, or W and MI' = Fe, Ni, Co, or Cr and
solid solutions thereof. In particular, the o-type borides have been developed exten-
sively by Takagi et al. [129-1311 and Komai et al. [356], focusing on molybdenum
rather than W or Ta due to its corrosion resistance.
7.8.3.1 Technically Important Systems and Structures

The crystal structure of the o-type ternary borides M:M"B2was described first by
Rieger et al. [357] as a distinct kind of structure similar to the U3Si2structure. The lat-
tice is orthorhombic, space group Immm, and contains two formula units, i.e. ten
atomic sites. The cell is built up by two metal prisms containing four atoms of M'
and two atoms of M". These double prisms are piled-up along the b-axis. The
boron atoms are arranged in pairs at the edges and in the center of the cell with
their axes parallel to the b-axis. If Cr is inserted into the structure it is considered to
preferentially replace Fe, Ni, or Co and thus occupies the MI1-position. Rietveld ana-
lysis of powder data by Ozaki er al. [358]proves that this kind of solid solutions creates
a distortion of the orthorhombic structure, which then can be indexed according to a
tetragonal lattice where the former c-axis of the orthorhombic cell becomes the tetra-
gonal c-axis. The M' sites are now arranged in a planar quadrangle surrounded by a
tetragonal-face centered MI' coordination shell. Since the edges of the MI quadrangle
are not parallel to the axes of the cell, the boron pairs are also out of the b-alignment.
Thus the space group P4/mbm is of comparatively low symmetry (Fig. 112).
Three kinds of ternary borides with metallic binders have been exploited for
future applications in detail, the WCoB-Co cermet for its high heat resistance, the
Mo2NiB2-Ni cermet because of its excellent corrosion resistance, and the
Mo2FeB2-Fecomposite for its wear resistance. In all cases, these borides are com-
patible with a melt consisting of M" and alloys thereof. According to the particular
phase diagrams, which have been treated in Section 7.3.2.2 (third part), the corre-
sponding liquid phase must, however, contain some portion of M' as well. The
two phase equilibria cp- or o-liquid, respectively, are limited by a tie line with the
brittle z-phase or, in systems in which (does not exist, with the particular binary
M" borides, e.g., with Fe2B, Co3B, etc., which have already been treated as a prob-
lem in TiB2-Fe cermets. The other tie line interconnects the ternary borides of
M',Mj' type intermetallic phases, which may even be beneficial for the hardening
of the consolidated binder phase.
(a) Orthorhombic-MsBz: a=0.70945 b=0.45746 ~=0.31733nma=By=90'
Mo
Ni
.B
(b) Tetragonal-M3B2 : a=b=0.58042 C=O.31367nm a+y=90'
Figure 112. Crystal structure of w-type ternary borides MiM"B2 with M' = Mo and M" = Ni or
Cr. Depending on the Cr content, the orthorhombic structure (a) or the tetragonal structure (b) are
stable (after Ozaki et al., 1994 [358]).
7.8.3.2 Liquid-Phase Sintering and Phase Reactions

Since ternary borides are commercially not readily available the preferred sintering
route is not a simple liquid-phase sintering, making use of powder blends of the
equilibrium compounds. Takagi [359] introduced a reaction sintering process
denoted as reaction boronizing sintering. This method has several benefits, (i) the
starting powders are comparatively cheap, available in fine grain size with suitable
purity and can easily be handled; (ii) the process takes advantage of chemical acti-
vation, i.e. the enthalpy of formation being released for sintering; (iii) the liquid
phase forms after a solid-state reaction and is therefore homogeneously distributed
throughout the microstructure; and (iv) the previous reaction brings about in-situ
grain size refinement and removal of oxide impurities, which enables an almost per-
fect wetting by the liquid phase.
The process starts with a pressed powder mixture of one or more binary borides
of MI-type metals and the metals required for the M" constituent of the ternary
phase and for the binder phase. This method enables the fabrication of tailored
MI-MIr-solid solutions in the ternary boride as well as allows the introduction
of high-performance Fe- or Ni-based alloys with exceptional properties. The
metal powders are usually atomized alloys but may also consist of powder
blends of the particular elements of interest. In Fig. 113 [360] the reaction process
is schematically summarized showing the M" alloy denoted as "metal". In princi-
ple, the first step is the boronization of the M" powder particles by boron diffusion
(1 B o r o n s o u r c e powder
0 0 0 Metal powder
T e r n a r y complex b o r i d e
Liquid
i n i t i a l stage
mixed powders
solid state
diffusional
l i q u i d phase
s i n t e r i ng
final stntered
compact (3 Metal m a t r i x
b o r o n i z ing
reaction
Figure 113. Schematic illustration of the "reaction boronizing sintering process" (Takagi, 1993)
[359].
from the M' boride across the contact points. This solid-state diffusion is enhanced
by the concentration gradient in boron and both M' and M" metals as well as by
the chemical driving force for the formation of the ternary M;M"B2 phase. Both
diffusion and heat release result in the growth of sinter necks between the starting
particles and the precipitation of very small ternary boride particles being homo-
geneously arranged around the M" grains. The source powder for boron is used up
completely by this reaction, and, upon further heating, the liquid phase is created
first on sites where the eutectic composition exists. Since the ternary borides have
been precipitated in situ, no oxide layer is present, preventing the wetting. During
spreading of the melt particle rearrangement by capillary forces is active and
causes shrinkage and a more homogeneous distribution of all phases. Due to the
higher temperature and the enhanced diffusion via the liquid phase, the residual
metal powder dissolves in the melt, filling the residual pores and allowing grain
growth of the stable ternary borides. Finally, after cooling, the cermet consists
of the hard boride phase dispersed in a metallic matrix phase or with a binder
phase at the triple points, depending on the volume fraction of binder. The micro-
structure resembles that of ordinary hard metals based on WC-Co with some pecu-
liarities, however, which will be discussed later.
To enable the solid-state reaction at the beginning and to control wetting at higher
temperatures, oxide impurities on the starting metal powders have to be removed.
This can be performed by reducing agents such as carbon or by evaporation of
boron oxides and suboxides during vacuum sintering. A hydrogen atmosphere
can also be used for sintering. Pressureless sintering is carried out slightly above
the quasi-binary eutectic temperature, usually between 1200 and 1350°C. The sinter-
ing time is comparatively short, e.g. 20-30 min to prevent grain growth by dissolu-
tion and reprecipitation.
A detailed study of the sintering phenomena of a Mo2FeB2-(Fe-alloy)cermet is
presented by Sivaraman et al. [361] who focused on the commercially available
KH-C5O material by Toyo Kohan Co., Ltd. consisting of 42.71 mass-% Mo,
39.28% Fe, 10.20% Cr, 4.96% B (as FeB), and 2.85% Ni. Sintering was followed
by DTA, TG, dilatometry, and X-ray diffraction; the microstructure was observed
after quenching from particular heating steps. As a result, the sintering reactions
could easily be monitored by both shrinkage effects and exo- and endothermic
effects, The first solid-state reaction is the formation of Fe2B at 486°C followed
by the precipitation of tetragonal Mo2FeB2, the reduction of metal oxides, the
Table 14. Sintering reactions of ternary boride cermets.

Temp. "C Reaction Remarks, Microstructure
48G500 Fe + FeB Fe2B

4 Solid-state reaction
700-1000 Sinter neck formation
10o(r1100 2 Mo + 2 Fe2B --+ Mo2FeB2+ 3 Fe Solid state reaction, enhanced growth of
sinter necks, formation of spherical
particles and clusters
1132 yX+ M,O, x M + y XO t
--+ Reduction of oxides,
X = reducing agent, e.g., C
1150 Mo2FeBz+ Cr Mo2(Fe,Cr)B2
4 Solid-state reaction, later also in liquid
state; high-density clusters with rounded
surfaces, initial growth of faceted grains
by surface and volume diffusion
1150-1 180 Fe + Fe2B + Mo2(Fe,Cr)B, + Liquid Ternary eutectic, wetting, spreading,
cluster fragmentation and particle
rearrangement
1200 Redistribution of phases, filling of pores
1225 Grain growth of faceted particles
1245 Continuous homogenization, formation
of boride framework with high
contiguity
formation of the liquid phase, and the incorporation of Cr into the Mo2FeB2.Table
14 summarizes the reaction steps and the resulting intermediate microstructures.
The sintering as monitored by dilatometry is illustrated in Fig. 114 [361]. Evi-
dently, the solid-state reaction to form Fe2B as well as the initial state sintering
by neck growth and particle center-approach yield a small contribution to shrink-
age. The most dramatic effect is, however, the generation of a liquid phase at
above 1150°C which yields a shrinkage rate of 6%/min, i.e., after 5-10min the
final density of 99.6% of theoretical density is achieved. This fast sintering can be
attributed to the sudden rearrangement of the boride particles of less than 1 pm
in size. At temperatures exceeding 1200-1220°C only coarsening is observed by dis-
solution and reprecipitation mechanisms but the particle size remains around 1 pm.
The change in particle size in Mo2NiB2cermets with Cr-Ni binder was investigated
in more detail by Matsuo et al. [362]. They started from MOBas a boron source to
compare the influence of Cr addition on structural changes. Table 15gives the
composition of the boride phase aimed at whereas the binder content was kept
constant at 25 mass-% (Ni f 10% Cr).
Sintering at 1260-1300°C for 20min revealed that MOBwas consumed totally by
the formation of Mo2NiB2, which exhibited an orthorhombic crystal structure,
while the addition of Cr yielded Moz(Ni,Cr)B2with a tetragonal structure. The micro-
structures consist of blocky faceted crystals of somewhat elongated habit in all cases.
It is, however, evident that Mo2NiB2tends to coarsen strongly (mean particle size
(M 8-10 pm) compared to Moz(Ni,Cr)B2 (mean particle size (M 3-5 pm), Fig. 1 15.
Another peculiarity is that the first phase forms euhedral crystals with sharp edges
and corners whereas the latter phase shows rounded edges and corners under exactly
the same conditions. This observation is attributed to the higher surface energy ani-
7.8 Cemented Boridees 923
0 400 800 1200 1600

temperature(%)
Figure 114. Sintering of MozFeBz with Fe,Cr,Ni-binder (after [361]).
sotropy of Cr-free orthorhombic boride by which the formation of equilibrium low-

energy facets are favored, as predicted by Warren [363]. The lower anisotropy in case
of tetragonal Mo2(Ni,Cr)B2is therefore also considered the reason for the observed
retardation of grain growth by Ostwald ripening [358,362].
7.8.3.3 Microstructures and Properties

A further study of Mo2NiB2-based cermets was dedicated to the influence of other
transition-metal additives on the structure and habit of the ternary boride phase
[365]. Taking the previously developed 35.4 mass-% Ni 6 mass-% B 58.6 + +
mass-% Mo cermet as a starting point, 10 mass-% of Cr, V, Fe, Co, Ti, Mn, Zr,
Nb, and W were added at the expenses of Ni. In another series of experiments, V
was added in steps of 2.5 mass-% up to 12.5 mass-%. The mechanical properties
Table 15. Composition of Mo2NiBzwith Cr-Ni binder.
Boride composition B Mo Ni Cr
Mo2NiBz mass- % 7.9 70.5 21.6 0

at.-% 40 40 20 0
Moz(Ni,Cr)Bz mass-% 8.0 71.4 10.6 10.0
at.-% 40 40 9.7 10.3
Figure 115. Microstructure of Mo2NiB2and Mo2(Ni,Cr)B2with Ni and Ni-Cr binder.
obtained clearly indicate that some of the elements have no effect on strength and
hardness at all, namely, Fe and Co, whereas Ti, Mn, Zr, Nb, and W yielded even
lower strength than the cermet without additives. Cr and V, however, caused
more than 40% increase in strength with a slightly higher hardness. The
addition of V gives rise to a steep increase in both strength and hardness if only
2.5 mass-% are added (Figs. 116-1 18).
It was proved that all boride solid solutions except the Cr- and V-containing
one crystallize in the orthorhombic structure type. As mentioned before, the
lower anisotropy of the surface energy of the tetragonal phase is the reason for
the formation of more spherical particles. The inhibition of grain growth was
accordingly found in Cr- and V-containing cermets but also in Nb- and W-bearing
materials (Fig. 119). In the latter cases, however, the microstructure is very inhomo-
geneous, which may be the reason for the comparatively low strength. In case of Co
and Mn addition the particles have grown considerably, which again explains the
low strength since no other brittle phase was detected. If Ti and Zr are added,
third phases have been observed but could not be identified.
A variation of the Cr content similar to the studies of the influence of V on the
properties was carried out by Komai et al. [356].Also in this case, an improvement
of both strength and hardness was achieved as a function of the additive amount.
Figure 116. Strength of Mo2NiB2-based cermets with various additives (after [365]).
However, the strength shows a pronounced maximum at 2300 MPa at 10 mass-%

and a decrease beyond the level of the cermet without Cr at 20 mass-%, while the
hardness steadily increases on the order of 5% (Fig. 120).
Although the ternary phases are hard but brittle, the improved commercial1
available cermets exhibit excellent toughness ranging from 18 to 30MPa m l i :
and strength ( ( J =
~ 1.6-2.2 GPa). Young's modulus reaches 290-350 GPa, and the
hardness 83-89 RA, depending on additives and binder content [359]. The
coefficient of thermal expansion ranges between 8.5 and 1 1 x lop6K-I. The cermets
7 p
Figure 117. Hardness of Mo2NiB2-basedcermets with various additives (after [365]).

92
2.5
2 2.0
-
2
tn
.
ln
0 88 g
E
2+ 1.5 I
86
c
- a4
1.0
0 2.5 5 7.5 10 12.5

V (mass%)
Figure 118. Transverse rupture strength (TRS) and hardness of V-containing Mo2NiB2 (after [365]).
Figure 119. Back-scattered SEM micrographs of a Ni-6 mass-%, B-58.6 mass-%, Mo-10 mass-%
metal cermet. Note the change in grain size and shape due to the metal additive (after [365]).
50-2. 92
90
88
2
-86 \
In
-a4 g
[z
I
82
-
80
0.501 I 78
0 5 10 15 20 25
Cr content (mass%)
Figure 120. Transverse rupture strength and hardness as a function of C r addition to Mo2NiB,
(after [356]).
of the Mo2FeB2 group exhibit an excellent corrosion resistance against organic

and inorganic acids and alkalis [ 1311 being comparable to or better than those of
cemented carbides and stainless steels. Furthermore, the cemented borides are
resistant in various molten metals such as Zn and Al. Consequently, these kinds
of materials can also be used as anti-corrosion layers on other metals [361].
7.8.4 Potentials and Applications

Most of the developments in cemented borides aim at the high wear resistance due to
the superior hardness compared to carbide-based hard metals. One of the most
important applications are, therefore, the cutting tools where cemented boride
have to compete with tungsten carbide, titanium carbonitride, and ceramic
materials as well.
7.8.4.1 Cutting Tools

Bearing the potential application as a wear-resistant material in mind, the
mechanical properties of the new cermets were assessed by measuring key mechan-
ical properties such as fracture toughness and hardness both at ambient and
elevated temperatures. Obviously, boride-based cermets integrate both excellent
hardness and a considerable fracture toughness such that their hardness-toughness
profile is superior to that of commercial hard metals. Of particular importance is the
fact that TiB2-Fe alloys are able to cover the high hardness regime beyond 2000
HVlO.
A test program to evaluate the potential of TiB2 cermets as a cutting tool material
was thus initiated by the European Community and many companies between 1990
Table 16. Mechanical properties of tool materials.
Tool material Hardness Fracture toughness Young’s modulus

HV 10 [MPaJmI [GPaI
TiB2-cc(Fe)-Fe2B 1830 9.3 490

TiB2-y(Fe-Cr-Ni) 1800 9.0 510
A1203-Zr02 1700 4.5 380
Ti(C,N)-Ni 1550 8.1 440
and 1993. The tests were aimed at (i) screening the cutting properties of TiB2-Fe and
(ii) providing useful information for further development of these materials.
The properties of TiB2 cermets suggest that workpiece materials that demand
tools of the lower P, M, and K qualities would be most adequate for cutting
tests. Thus a medium carbon steel Ck 45, a globular cast iron GGG 50, and a
hypo-eutectic cast G-AlSil2CuMgNi alloy were selected by Sigl et al. [366] for
continuous turning experiments having a Brine11 hardness of <210, <220, and
< 115, respectively.
Two series of TiB2 cutting inserts with ferritic iron binders and austenitic steel
binders (nominal composition 18 weight-% Cr, 10 weight-% Ni, bal. Fe) were
tested. Due to a Ti deficiency, the ferritic TiB2 grades with a nominal binder fraction
of 15 v01.-YO contained approximately 1Ovol.-% cc-Fe and 5vol.-% Fe2B. Ti was
added to prevent the formation of Fe2B in the steel cutting grades, such that the
second series of TiB2 cermets could be prepared with 10vol.-YOof fully austenitic
binder. Uncoated ferrite-binder cermets were used to machine the A1 and cast
iron alloys, whereas the Ck 45 steel was treated with both uncoated and TiN-
coated (3 pm CVD-TIN layer) inserts with austenite binder. Both composites
represent “generation-I” TiB2 cermets, in the sense that they feature a much coarser
grain size (on the order of 5 pm) than the composites available today. The tests were
carried out in comparison to conventional cutting tool materials. Thus the cast iron
grade was machined with a commercial alumina-zirconia material (10 weight-%
Zr02), while the carbon steel and the hypo-eutectic A1-Si alloy were treated with
a commercial Ti(C,N)-Ni cermet. Important mechanical properties of the cutting
tool materials are compared in Table 16. The actually tested combinations of work-
piece and cutting tool materials are compiled in Table 17.
Continuous turning tests were performed. The SNGN 120408 tool geometry and
parameters summarized in Table 18. Cutting was interrupted after a cutting length
of 500m for the GGG 50 and after l000m for the Ck 45 and G-AlSil2 alloys,
Table 17. Tested combinations of workpiece and cutting tool materials.
Tool materials G-AlSi 12 GGG 50 Ck 45
TiB2-a(Fe)-Fe2B X X
TiB2-y( Fe-Cr-Ni) X X
A1203-Zr02 X
Ti(C,N)-Ni X X
Table 18. Cutting parameters.
Cutting parameters G-AlSi 12 GGG 50 GGG5O Ck 45

finishing light roughing finishing finishing
Feed ,f [mmjrev.] 0.2 0.3 0.1 0.1

Cutting depth a [mm] 1 .o 1.5 0.5 2.0
respectively. The flank wear, wB, the depth of cut notch depth, 6,the surface rough-
ness of the workpiece material, the cutting forces, and the power consumption of the
machine were monitored at these intervals. The test was performed until the failure
criterion or a total cutting length of 8000 m was reached.
In all cutting materials the flank wear and the cutting forces increased con-
tinuously up to the ultimate lifetime. With very few exceptions of cutting edge
fracture, all inserts faded due to flank wear. Failure due to excess depth-of-cut
notch depth was not observed. Cratering at the rake face did occur, but was in no
instance the reason for insert breakdown. The surface roughness of the workpiece
materials was rated acceptable in all cases. Thus a flank wear, wB of 0.4mm was
defined as lifetime criterion. The cutting results using this criterion are summarized
in Figs. 121-124 for the various combinations of tool and workpiece materials.
Ranking the cutting tool materials, TiB2 with a ferritic binder is superior to the
Ti(C,N) cermet and the alumina-zirconia ceramic for light cutting operations such
as finishing of GGG 50 and G-AlSil2. However, increased thermal loads favor cut-
ting materials with enhanced high-temperature endurance, i.e., the Ti(C,N) cermets
and the alumina-zirconia ceramics. Figures 121- 124 illustrate the details of these
general observations: TiB2 cermets outperform commercial Ti(C,N) tool materials
100
.-c 10
E
0.1
Cutting Speed v,, d m i n

Figure 121. Life-time diagram for finishing G-AlSil2.
I000
100
10
1
10 100 1000
Cutting Speed v, m/min
Figure 122. Life-time diagram for finishing GGG 50.
in A1 cutting over the whole range of cutting speeds up to 700 m/min, and they work
well at medium cutting speeds up to 200 m/min during finishing operations of gray
cast iron. Tool life decreases dramatically, however, upon increased cutting loads,
i.e., for light roughing of GGG 50 and for steel finishing operations (Figs. 123,
124). In particular, the Taylor slope of TiB2-Fe in cast iron cutting is much steeper
than for alumina-zirconia tools (Figs. 122, 123). A similar, yet less pronounced differ-
ence in Taylor slopes is observed between Ti(C,N) and TiB2 steel cutting grades
(Fig. 124). The TIN coating significantly improves the life of TiB2 tools, but it
cannot fill the gap of the exceptional quality of Ti(C,N) cermets.
100
10
10 100 1000
Cutting Speed v,, m/min

Figure 123. Life-time diagram for light roughing GGG 50.
10 100 1000
Cutting Speed v,, m/min
Figure 124. Life-time diagram for finishing Ck 45
The following issues must be considered in a discussion of the cutting results:

0) TiB2 is a very creep and plasticity resistant compound, particularly a t elevated
temperatures [367].
(ii) The ferrite binder in the A1 and cast iron grades contains only Ti and B in solid
solution and its high-temperature strength is expected to be rather limited. The
austenite binder with 18% Cr and 10% Ni should perform significantly better
at high temperatures, but is probably less creep resistant than Ni(Mo) binders.
(iii) Coarse microstructures are known to favor erosive wear, even in very hard and
tough materials.
According to (iii), a generally improved cutting performance of TiB2 cermets, both
at high and low temperatures, is expected from a reduction of the boride grain
size. A further enhancement, specifically at high cutting loads, is anticipated from
binders with improved creep resistance since both high cutting speed and increasing
feed and depth of cut tend to raise the cutting tool temperatures. Thus the reduced
cutting performance at rising loads is primarily attributed to the limited high-
temperature strength of the ferrous binders. This conclusion is supported by the
improved tool life of TiN-coated inserts because the coating reduces the friction
between tool and chip and consequently keeps the temperature level of the cutting
insert low.
In conclusion, TiBz is an excellent hard compound for continuous cutting
operations already matching or surpassing conventional tool materials in appli-
cations where limited temperature loads evolve. Specifically, ferritic binders perform
well in the cutting of cast A1 alloys and cast iron. TiB2 austenite cermets do show
potential for the cutting of steel, but as yet they cannot match the exceptional
performance of Ti(C,N) cermets.
An optimization of the microstructure towards fine grain size of TiB2 and binders
with sufficient high-temperature strength will further improve the cutting capacity of
TiB2 cermets specifically for steel machining operations. The major disadvantage
that has prevented the production of TiB2-Fe composites on a commercial scale is
their delicate processing behavior. Thus the cost-performance relation is still the cri-
tical point for the current market conditions.
7.8.4.2 Wear Parts and Molding Tools

The applications of cemented ternary borides aim at other industrial branches but
also make use of the unique combination of hardness, toughness, and corrosion
resistance. Sliding wear tests carried out by Takagi [359] in a rotating-ring-on-
block arrangement have shown that the wear rate of a Mo2FeB2 (Fe-alloy)
cermet is one order of magnitude less than that of WC-1OCo. The characterization
of the wear debris yielded a small amount of Moo2, B203,and Fe304 at sliding
velocities >0.6m/s, whereas chips produced by the WC-Co hard metal did not
give evidence for any oxide formation. The higher wear rate of WC-Co is attributed
to WC-particle pull-out, providing abrasive grains at the interface between the wear
couples.
Wear-resistant parts such as injection-molding machine parts for fabrication of
fiber-reinforced polymers, metals, and ceramics, drawing dies for metallic wires,
bearings, liners, and other parts for protection against sliding wear have been
developed from Mo2FeB2-Feor Mo2NiB2-Nicermets (Fig. 125). They are commer-
cially available and already widely used in Japan. Moreover, ternary borides have
been successfully applied as coatings on steel for surface hardening and improve-
ment of the corrosion resistance [360]. Thus this new class of materials has to
Figure 125. Injection molding machine parts composed of MolFeB2-based alloys: (a) cylinder; (b)
auger; (c) check valve; (d) seal ring; (e) auger head.
compete with all kinds of hard metals and cermets with respect to performance
and costs.
7.9 Future Prospects and Fields of Application

At the time this article was written, optimism concerning the application of high-
performance advanced ceramics was still considerably low after the boom between
1980 and 1990, especially in respect to structural parts for automotive engines or
substitutes for other high-strength, high-toughness materials. This tendency was
and is not restricted to Europe or the USA., but is also valid for Japan and can
easily be attributed to the following problems: (i) difficulties related to reliable
large-scale production at reasonable cost and quality (problems of “zero-flaw”
processing and part testing), (ii) difficulties in a material-appropriate design of
parts (problems of substitution and new construction taking the relative brittleness
of ceramics into account), (iii) the still comparatively high price of high-quality
starting materials and manufacturing processes (problems of affordable high-
purity and fine grain size), and (iv) the current worldwide economic recession in pro-
duction stopping any activities in materials research and development. Considering
these facts, the following paragraph will give some basic ideas on the actual and
most probable applications, which have already been demonstrated up to now to
be feasible. Besides the well-known application of grits (e.g., as grinding, lapping,
and polishing particles, as oxidation inhibitors in refractories, and for alloying of
special metals), boride ceramics are already widely used in mechanical, chemical,
thermal, and electrical areas.
Most of the applications of borides related to mechanical aspects make use of
their extraordinarily high hardness and wear resistance. Also, their high Young’s
modulus (stiffness) and excellent high-temperature behavior are very attractive
properties. The high-temperature hardness of the most important ceramics and
hard materials is shown in Fig. 126. It is obvious that, besides carbides, borides
are superior to most oxide ceramics. At high temperatures, however, the metallic
transition-metal borides suffer from a strong decrease in hardness due to enhanced
plastic deformability (thermal initiation of slip systems), which is basically due to
their comparatively primitive crystal structure and their high metallic portion in
bonding. Additionally, oxidation has to be taken into account at temperatures
above 600°C.
The poor corrosion resistance of borides may be partially overcome in composites
with S i c if the microstructure allows passive oxidation kinetics. In combination
with S i c the borides retain their high electrical and thermal conductivity and thus
suitable thermal shock resistance. SiC-TiB2 composites have been extensively devel-
oped for wear parts in machinery such as sliding rings, valves, valve seats, roller and
ball bearings, plungers, and rocker arm pads.
Besides the well-known cemented carbides and cermet cutting tools, boron
carbide- and titanium diboride-containing cermets have been proven suitable for
8000
6000
-
N
E
E
\
-.-
m
Y
2 1000
Ln
In
a,
C
?
I
2000
0
0 LOO 800 1200
Temperature 1°C)
Figure 126. High-temperature hardness of ceramic materials in comparison to diamond and cubic
boron nitride.
machining aluminum alloys, whereas ternary boride cermets have been successfully
developed for injection molding nozzles, bearings, wire drawing cones, and other
wear resistant parts that are not exposed to high temperatures. Composites of
TiB2/TiC, TiB2/TiN, TiB2/ZrB2/TaN, as well as B4C/TiB2, B4C/SiC, and B4C/
SiC/TiB2 have been exploited as cutting tools for brass, bronze, and A1 alloys,
drilling tools for rocks, and concrete due to their comparatively high toughness.
An obvious conclusion from the various efforts in strengthening and toughening
of boride-based materials is that the maximum strength obtained by crack-
propagation influencing mechanisms operational in composites is about 1-
1.2 GPa and the fracture toughness usually obtained in optimized microstructures
is about 6-7 MPa m1I2.It is also clear that simple two-phase composites are gener-
ally superior to materials of more complex composition. Metal-matrix composites
may exhibit a much higher toughness than ceramics but are limited in high-
temperature use due to their sensitivity towards creep and corrosion. With the
possible exception of the very promising TiB2-Fe-based and ternary boride-based
cermets, borides are thus currently of more interest for strengthening metals similar
to oxide-dispersed superalloys. A certain challenge is still the transformation tough-
ening of borides by Y-stabilized Zr02, increasing the toughness by a factor of two
compared to simple particulate reinforcement. In competition with oxide and nitride

ceramics the borides have thus been developed to a similar excellent mechanical
performance but still suffer from their considerably high price and poor oxidation
resistance at high temperatures.
Generally speaking, the high reactivity of the boron compounds is the most
important, and sometime the most limiting factor. Although the wear behavior is
very promising due to the high hardness and the high thermal conductivity,
boron compounds, except for some transition-metal diborides, cannot be used in
contact with iron-based alloys but for wear contacts with aluminum and, in some
cases, nickel-based materials. On the other hand, reaction sintering can be widely
employed for the fabrication of unique microstructures, which has not yet been
exploited entirely.
Although mechanical applications are currently by far the most important aims
for the development of materials and parts based on carbides and borides, this
may easily change to the benefit of thermal and electrical applications if there is
an improvement in the understanding and control of the semiconductivity of B4C
and a breakthrough in the manufacture of suitable parts. Moreover, thermal
applications may increase considerably in importance as more efficient energy
conversion or recovery becomes a significant requirement.
Excellent opportunities for applications of transition-metal borides such as TiB,
are thus more a result of their high electrical conductivity and resistance against
metallic melts combined with an excellent wetting behavior. Well-known examples
are TiB,-AlN-BN evaporator dies (e.g., for the evaporation of Al, Cd, Ge, Pb, Bi,
Cr, Cu, Ag, Au, Mg, Fe, Zn, and Sn for the coating of plastics, paper, and parts) and
electrodes for the aluminum electrolysis in Hall-Herould cells. In the latter case,
there is a very high demand for such a material having a high electric conductivity
and extraordinary corrosion resistance against kryolithe melts, but some undesired
grain boundary reactions (infiltration of the melt along impurity-loaded grain
boundaries) and the poor thermal shock resistance (strongly anisotropic thermal
expansion) are the limiting factors for a successful replacement of graphite electro-
des. Other diborides, such as ZrB,, have been studied for use as structural parts,
such as bearings, nozzles, and molds for injection molding, valves, and sealings,
but further development is still necessary for an industrial breakthrough. For
example, ZrB2-MoSi2 composites exhibit remarkable corrosion resistance up to
2000"c.
Functional applications of borides are also related to the high cross-section of
boron for the absorption or retardation of thermal neutrons. B4C sintered products
have been used as moderating elements and radiation-protecting components in
nuclear reactors of several designs. Of course, this market is declining with the
decreasing political support for nuclear power and nuclear research.
In conclusion, boride ceramics are materials of high potential for use in mechanical
and electric fields. Due to the stagnating materials markets, however, most research
on borides is focused on the basic questions of atomic bonding and physical trans-
port properties rather than on application-related problems. Only a few highly
specialized research groups are active worldwide and promote the science rather
than the technology of borides. Depending upon the economic situation and the
demand for future key technologies, there is a chance that these materials may
increase in importance.
References
1. Kiessling, R. Acta Chem. Scand. 1950, 4, 209.
2. Lundstrom, T. Arkiv Kemi 1969, 31(19), 227.
3. Anderson, S. and Lundstrom, T. Acta Chem. Scand. 1968, 12, 3103.
4. Vlasse, M., Slack, G. H., Garbauskas, M., Kasper, J. S., and Viala, J. C. J . Solid State Chem.
1986, 63, 31.
5. Richards, S. M. and Kasper, J. S. Acta Crystallogr. 1969, B25, 237.
6. Higashi, I., Kobayashi, K., Tanaka, T., and Ishizawa, Y. J. Solid Stute Chem. 1997, 133, 16.
7. Matkovich, V. I. and Economy, J. in Boron and Refractory Borides, Matkovich, V. I. (Ed.),
Springer, Berlin, 1977, pp. 77-94.
8. Aronsson, B., Lundstrom, T., and Rundqvist, S. Borides, Silicides and Phosphides: A Critical
Review of Their Preparation, Properties and Crystal Chemistry, John Wiley & Sons Inc., New
York, 1965.
9. Aronsson, B., Lundstrom, T., and Engstrom, I. in Anisotropy in Single Crystal Refractory
Compounds, Plenum Press, New York, 1968, Vol. 1, pp. 3-22.
10. Lundstrom, T. Arkiv Kemi 1969,31(19), 227.
11. Higashi, I. and Takahashi, Y. J . Less-Common Met. 1986, 123, 277.
12. Okada, S., Atoda, T., Higashi, I., and Takahashi, Y., J . Mater. Sci. 1987, 22, 2993.
13. Lundstrom, T. Arkiv Kemi 1968, 30(11), 115.
14. Romans, P. A. and Krug, M. P. Acta Crystallogr. 1966, 20, 313.
15. Woods, H. P., Wawner, F. E., and Fox, B. G. Science 1966, 151, 75.
16. Armstrong, D. R. in Proc. 9th Int. Symp. on Boron, Borides, and Related Compounds, Werheit,
H. (Ed.), University Press, Duisburg, 1987, pp. 125-131.
17. Samsonov, G. V. and Kovenskaya, B. A. in Boron and Refractory Borides, Matkovich, V. I.
(Ed.), Springer, Berlin, 1977, pp. 5-18.
18. Samsonov, G. V. and Kovenskaya, B. A. in Boron and Refractory Borides, Matkovich, V. I.
(Ed.), Springer, Berlin, 1977, pp. 19-30.
19. Aravamudham, R. Z. Metallkd. 1967, 58, 179.
20. Samsonov, G. V., Goryachev, Yu. M., and Kovenskaya, B. A. Izv. Vuz Fiz. 1972, 6 , 37.
21. Hoard, J. L. and Hughes, R. E. in The Chemistry of Boron and Its Compounds, Muetterties,
E. L. (Ed.), Wiley, New York, 1967, pp. 25-154.
22. Spear, K. E. J. Less-Common Met. 1976, 47, 195.
23. Hayami, W., Souda, R., Aizawa, T., and Tanaka, T. Surf. Sci. 1998,415, 433.
24. Souda, R., Asari, E., Kawanowa, H., Otani, S., and Tanaka, T. Surf. Sci. 1998,414(1-2), 77.
25. La Placa, S. and Post, B., Planseeberichtefiir Pulvermetallurgie 1961, 9, 109.
26. Lipp, A. Ber. Deutsche. Keram. Ges. 1966, 43(1), 60.
27. Lipp, A. and Roder, M. Z . Anorg. Allg. Chem. 1966, 343, 1.
28. Lipp, A. and Roder, M. Z . Anorg. Allg. Chem. 1966, 344, 225.
29. Matkovich, V. I. and Economy, J. in Boron and Refractory Borides, Matkovich, V. I. (Ed.),
30. Clark, H. K. and Hoard, J. L. J. Am. Ceram. Soc. 1943,65, 2115.
31. Allen, R. D. J. Am. Chem. Soc. 1953,753582.
32. Neidhard, H., Mattes, R., and Becher, H. J. Acta Crystallogr. 1970, 26B, 315.
33. Ploog, K. J . Less-Common Met. 1974, 25, 115.
34. Will, G. and Kossobutski, K. H. J . Less-Common Met. 1976, 47, 43.
35. Kirfel, A., Gupta, A,, and Will, G. Acta Crystallogr. 1979, B35, 1052.
36. Werheit, H. and de Groot, K. Phys. Status Solidi 1980, 97, 229.
37. Conard, J., Bouchacourt, M., and Thevenot, F. J . Less-Common Met. 1986, 117, 51.
References 937
38. Tallant, D. R., Aselage, T. L., and Campbell, A. N. Phys. Rev. B 1989, 40, 5649.
39. Aselage, T. L., Emin, D., and Wood, C. in Trans. 6th Symp. Space Nuclear Power, Albuquer-
que, NM, January 8-12, 1989, pp. 430433.
40. Aselage, T. L., Tallant, D. R., Gieske, J. H., Van Deusen, S. B., and Tissot, R. G. in The
Physics and Chemistry of Carbides, Nitrides, and Borides, Freer, R. (Ed.), NATO AS1 Ser.
E, Vol. 185, Kluwer, Dordrecht, The Netherlands, 1990, pp. 97-1 12.
41. Aselage, T. L. and Emin, D. in Boron-Rich Solids, Proc. 10th Int. Symp. on Boron, Borides, and
Related Compounds, Emin, D., Aselage, T. L., Switendick, A. C., Morosin, B., and Beckel, C. L.
(Eds), AIP Conf. Proc. 231, American Institute of Physics, New York, 1991, pp. 177-185.
42. Morosin, B., Aselage, T. L., and Emin, D. in Boron-Rich Solids, Proc. 10th Int. Symp. on
Boron, Borides, and Related Compounds, Emin, D., Aselage, T. L., Switendick, A. C., Mor-
osin, B., and Beckel, C. L (Eds), AIP Conf. Proc. 231, American Institute of Physics, New
York, 1991, pp. 193-196.
43. Kleinman, L. in Boron-Rich Solids, Proc. 10th Int. Symp. on Boron, Borides, and Related
Compounds, Emin, D., Aselage, T. L., Switendick, A. C., Morosin, B., and Beckel, C. L.
(Eds), AIP Conf. Proc. 231, American Institute of Physics, New York, 1991, pp. 13-20.
44. Armstrong, D. R., Bolland, J. and Perkins, P. G. Acta Crystallogr. 1983, 39, 324.
45. Emin, D. in Boron-Rich Solids, Emin, D., Aselage, T, Beckel, C. L., Howard, I. A,, and Wood,
C. (Eds), AIP Conf. Proc, American Institute of Physics, New York, 1986, pp. 189-205.
46. Silver, A. H., Bray, P. J. J . Chem. Phys. 1959, 31, 247.
47. Hynes, T. V., Alexander, M. N. J . Chem. Phys. 1974,54, 5296.
48. Becher, H. J., Thevenot, F. Z. Anorg. Allg. Chem. 1974, 410, 274.
49. (a) Bouchacourt, M. and Thevenot, F. J . Less-Common Met. 1981,82,219. (b) Bouchacourt,
M. and Thevenot, F. J . Less-Common Met. 1981, 82, 227.
50. Schwetz, K . A. and Karduck, P., J . Less-Common Metals 1991, 175, I .
51. Kuhlmann, U., Werheit, H., and Schwetz, K. A. J . Alloj's Compd. 1992, 189, 249.
52. Lim, S.-K. and Lukas, H.-L., in Hochleistungskeramiken Herstellung, AU!fbau,Eigenschaften,
-
Petzow, G., Tobolski, J., and Telle, R. (Eds), VCH, Weinheim, and Deutsche
Forschungsgemeinschaft, 1996, pp. 605 -616.
53. Kasper, B. Dr. rer. nut. Thesis, University of Stuttgart, 1996.
54. Telle, R. in The Physics and Chemistry of Carbides. Nitrides, and Borides, Proc. Freer, R. (Ed.),
NATO AS1 Ser. E, Vol. 185, Kluwer, Dordrecht, The Netherlands, 1990, pp. 249-268.
55. Meerson, G. A., Kiparisov, S. S., and Gurevich, M. A. Sov. Powder Metall. Met. Ceram. 1966,
5, 223.
56. Magnusson, B. and Brosset, C. Acta Chem. Scand. 1962, 16,449.
57. Hansen, M. Constitution of Binary Alloys, McGraw-Hill, New York, 1958.
58. Elliott, R. P. Constitution qf Binary Alloys, 1st suppl., McGraw-Hill, New York, 1965.
59. Shunk, F. A. Constitution of Binary Alloys, 2nd suppl., Mc-Graw-Hill, New York, 1969.
60. Mofatt, W. G. The Handbook of Binary Phase Diagrams, General Electric Co., Schenectady,
NY, 1976.
61. Mofatt, W. G. The Index to the Binary Phase Collections, General Electric Co., Schenectady,
NY, 1979.
62. Binary Alloy Phase Diagrams, Massalski, T. B. (Ed.), ASM Int., Materials Park, OH, 1990,
Vol. 1: Alloys, Vol. 2: Phase Diagrams.
63. Ternary Alloys, Petzow, G., Effenberg, G. (Eds), VCH, Weinheim, 1988ff, Vols. 1 ff.
64. Spear, K. E. in Application of Phase Diagrams in Metallurgy and Ceramics, National Bureau of
Standards, Gaithersburgh, 1977, Special Publication SP-496, pp. 744-762.
65. Lundstrom, T. in Boron undRefractory Borides, Matkovich, V. I. (Ed.), Springer, Berlin, 1977,
pp. 351-376.
66. Spear, K. E. in Phase Diagrams, Materials Science and Technology, Vol. IV, Alper, A. M.
(Ed.), Academic Press, New York, 1976, Ch. 1 I , pp. 91-159.
67. Guillermet, A. F. and Grimvall, G. Phys. Rev. B 1989, 40, 1521.
68. Grimvall, G. and Guillermet, A. F. Boron-Rich Solids, Proc. 10th Int. Symp. on Boron,
Borides, and Related Compounds, Emin, D., Aselage, T. L., Switendick, A. C., Morosin,
B., and Beckel, C. L. (Eds), AIP Conf. Proc. 231, American Institute of Physics, New York,
1991, pp. 423-430.
69. Samsonov, G. V. and Schuravlev, N. N. Fiz. Met. Metullovad. Akad. Nauk, S S S R Ural Filial 3
1956, 109.
70. Schuravlev, N. N. and Makarenko, G. N. Ed. Acad. Sci. USSR,OTN, Metallurgia i Toplivo I ,
1961, 133.
71. Bouchacourt, M. and Thevenot, F. J. Less-Common Met. 1979,67, 327.
72. Beauvy, M. in Conf. Abstract, 8th Int. Symp. on Boron, Borides, Carbides and Related Com-
pounds, Tsagareishvili, G. V. (Ed.), Acad. Sci. Georg. S.S.R., Tbilisi, 1984, p. 25.
73. Lukas, H. L. in Constitution of Ternary Alloys 3, Petzow, G. and Effenberg, G. (Eds), VCH,
Weinheim, 1990, pp. 140-146.
74. Ekbom, L. B. Sci. Ceram. 1977, 9, 183.
75. Lange, D. and Holleck, H., Proc. 11th Int. Plansee-Conf., Planseewerke, Reutte, 1985, HM 50,
741.
76. Ettmayer, P., Horn, H. C., and Schwetz, K. A. Microchim. Acta, 1970, Suppl. IV, 87.
77. Lugscheider, E., Reimann, H., and Quaddakers, W.-J., Ber. Deutsche Keram. Ges. 1979,
65(10), 301.
78. Dorner, P. Dr. rer. nut. Thesis, University of Stuttgart, 1982.
79. Emin, D. in The Physics and Chemistry of Carbides, Nitrides and Borides, Freer, R. (Ed.),
Kluwer, Dordrecht, The Netherlands, 1990, pp. 691-704.
80. Olesinski, R. W. and Abbaschian, G. J. Bull. Alloy Phase Diagrams 1984, 5, 478.
81. Mitra, I., Biittner, D., Telle, R., Babushkin, O., and Lindback, T. in Proc. 8th CIMTEC,
Ceramics Charting the Future, Advances in Science and Technology, Vol. 3B, Vincenzini, P.
(Ed.), Techna Srl., Milano, 1995, pp. 787-792.
82. Armas, B., Male, G., and Salanoubat, D. J . Less-Common Metals 1981, 82, 245.
83. Telle, R. Dr. rer. nut. Thesis, University of Stuttgart, 1985.
84. Murray, J. L., Liao, P. K., and Spear, K. E. Bull. Alloy Phase Diagrams 1986, 7, 550.
85. Rudy, E. and Windisch, S. in Ternary Phase Equilibria in Transition Metal-Boron-Carbon-
Silicon Systems, Part I, Vol. VII, Technical Report No. AFML-TR-65-2, Wright Patterson
Air Force Base, OH, 1966.
86. Fenish, R. G . Trans. A I M E 1966, 236, 804.
87. Neronov, V. A,, Korchagin, M. A,, and Aleksandrov, V. V. J . Less-Common Met. 1981, 82,
125.
88. Spear, K. E., McDowell, P., and McMahon, F. J . Am. Ceram. Soc. 1986,69, C4.
89. Batzner, C. Dr.rer.nat. Thesis, University of Stuttgart, 1994.
90. Pyzik, A. J. and Beaman, D. R. J . Am. Ceram. Soc. 1995, 78, 305.
91. Kieffer, R., Gugel, E., Leimer, G., and Ettmayer, P. Ber. Deutsche. Keram. Ges. 1972,49(2), 44.
92. Mogilevsky, P., Gutmanas, E. Y., Gotman, I., and Telle, R. J. Eur. Ceram. Soc. 1995, 15, 527.
93. Lukas, H. L. Final Report, Deutsche Forschungsgemeinschaft Project, Contract No. Lu 283/
2-2, 1992.
94. Telle, R. and Petzow, G. in High Tech Ceramics, Vinzencini, P. (Ed.), Material Science Mono-
graphs, Elsevier, Amsterdam, 1987, pp. 961-973.
95. Werheit, H., Kuhlmann, U., Laux, M., and Telle, R. J . Alloys and Compounds 1994, 209, 181.
96. Secrist, D. R. J. Am. Ceram. Soc. 1964, 47, 127.
97. Shaffer, P. T. B. Muter. Res. Soc. Bull. 1969, 4, 213.
98. Nishiyama, K., Mitra, I., Momozawa, N., Watanabe, T., Abe, M., and Telle, R. J . Jpn. Res.
Inst. Muter. Technol. 1997, 15, 292.
99. Rudy, E. and Windisch, S. Ternary Phase Equilibria in Transition Metal-Boron-Carbon-Silicon
Systems, Part 11. Related Tenary System Vol. VIII, TI-B-C System, Technical Report No.
AFML-TR-65-2, Wright Patterson Air Force Base, OH 1966.
100. Nowotny, H., Benesovsky, F., and Brukl, C. Monatsh. Chem. 1961,92,403.
101. Ordan’yan, S. S. and Unrod, V. V. Sov. Powder Metall. Met. Ceram. 1975, 14, 729.
102. Holleck, H. Planseeber. Pulvermetall. 1982, 2, 849.
103. Holleck, H., Leiste, H., and Schneider, W. in High Tech Ceramics, Proc. 6th CIMTEC,
Vincenzini, P. (Ed.), Elsevier, Amsterdam, 1987, pp. 2609-2622.
104. Rudy, E. Ternary Phase Equilibria in Transition Metal-Boron-Carbon-Silicon Systems, Vol. V,
Compendium of Diagram Data, US Atomic Energy Comm. Publ. AFML-TR-65-2, Wright
Patterson Air Force Base, OH, 1969.
References 939
105. Duschanek, H. and Rogl, P., Lukas H. L. J . Phase Equilibria 1995, 16, 46.
106. Brodkin, D. and Barsoum, M.W. J . Am. Ceram. Soc. 1996,79, 785.
107. Telle, R., Brook, R. J., and Petzow, G. J . Hard Muter. 1991, 2, 79.
108. Taffner, U., Telle, R., and SchHfer, U. Z. Prukt. Metallogr. 1988, 21, 17.
109. Klingler, H., Franz, E.-D., and Telle, R. Spreclisaal 1995, 128, 11.
1 10. Brukl, C. E. Ternary Phase Equilibria in Transition Metal-Boron-Carbon-Silicon Systems Part
11; Ternary Systems, Wright, Patterson Air Force Base, OH, 1965, Vol. VII, Tech. Rep. No.
AFML-TR-65-2.
11 1. Borisova, A. L., Borisov, Yu. S., Polyanin, B. A., Shvedova, L. K., Kalinovskii, V. R., and
Gorbatov, I. N. Sov. Powder Metall. Met. Ceram. 1986, 25, 769.
112. Holleck, H. in Proc. 12th Planseeseminar ofPoivder Metallurgy 3, Bildstein, H., Ortner, H. M.
(Eds), Planseewerke, Reutte, 1989. pp. C3, 1-12.
113. Touanen, M., Teyssandier, F., and Ducarroir, M. in Proc. 7th Eur. Con6 on Chemical Vupour
Deposition, Ducarroir, M., Bernard, C., and Vandenbulcke, L. (Eds), J . Phys. 1989, 50, 105.
114. Wakelkamp, W. J. J., van Loo, F. J. J., and Metselaar, R. J . Eur. Ceram. Soc. 1991, 8, 135.
115. Yuriditsky, B. Y. Refractory Muter. Hard Muter. 1990, 3, 32.
116. Sigl, L. S. and Schwetz, K. A. in Boron-Rich Solids, Proc. 10th Int. Symp. Boron, Borides, and
Related Compounds, Emin, D., Aselage, T L., Switendick, A. C, Morosin, B., and Beekel, C. L.
(Eds), AIP Conf. Proc. 231, American Institute of Physics, New York, 1991, pp. 468472.
117. Ghetta, V., Gayraud, N., and Eustathopoulos, N. Solid State Phenomena 1992, 25/26, 105.
118. Jiingling, T., Oberacker, R., Thiimmler, F., Sigl, L. S., and Schwetz, K. A. Powder Metall. Int.
1991, 23(5), 296.
119. Sigl, L. S. and Jiingling, T. J . Hard Muter. 1992, 3, 39.
120. Federov. T. F. and Kuzma, Y. B. Izvest. Akad. Nuuk S S S R - Neorg. Muter. 1967, 3, 1489.
121. Shurin, A. K. and Panarin, V. E. Izvest. Akad. Nauk. S S S R Metally 1974. 4, 235.
-
122. Smid, I. and Kny, E. I n f . J . Refractory Met. Hard Muter. 1988, 8, 135.
123. Ottavi, L., Chaix, J. M., Allibert, C., and Pastor, H. Solid State Phenomena 1992,25/26, 543.
124. Ottavi, L., Saint-Jours, C., Valignant, N., and Allibert, C. Z. Metullkd. 1992, 83, 80.
125. Schobel, J. D. and Stadelmaier, H. H. M~~tallwissenschuft Tech. 1965, 19, 715.
126. Lugscheider, E., Reimann, H., and Pdnkert, R. Z. Metullkd 1980, 71, 654.
127. Lugscheider, E., Reimann, H., and Pankert, E. Metall. 1982, 36, 247.
128. Haschke, H., Nowotny, H., and Benesovsky, F. Monatsh. Chem. 1966, 97, 1459.
129. Takagi, K., Ohira, S., and Ide, T. in Modern Developments in Powder Metallurgy, Vol. 16:
Ferrous and Non-ferrous Materials, Aqua, E. N., and Whitman, C.I. (Eds), MPI & APMI
Publ., Princeton, NJ, pp. 153-166.
130. Takagi, K. et al. Int. J . Powder Met. 1987, 23, 157.
131. Takagi, K., Ohira, S., Ide, T., Watanabe, T., and Kondo, Y. Met. Powder Rep. 1987,42,483.
132. Post, B., Glaser, F. W, and Moskowitz. D. Acta Mefall. 1954, 2, 20.
133. Kuz’ma, Y. B., Telegus, V. S., and Kovdlyk, D. A. Poroshkovuya Mefallurgiya 1969,5(77), 79;
Sov. Powder Metall. Met. Ceram. 1969, 4, 403.
134. Pastor, H. in Boron and Refractorj, Borides, Matkovich, V.I. (Ed.), Springer, Berlin, 1977,
pp. 457493.
135. Zdaniewski, W. A. J . Am. Ceram. Soc. 1987, 70, 793.
136. Telle, R., Fendler, E., and Petzow, G. J . Hard Muter. 1992, 3, 21 1.
137. Makarenko, G. N. in Boron and Rcrf).actoryBorides, Matkovich, V. I. (Ed.), Springer, Berlin,
1977, pp. 310-330.
138. Kuz’ma. Y . B., Telegus, V. S., and Marko, M. A. Poroshkovaya Metallurgiya 1972,4(66), 215
139. Koval’chenko, M. S., Ochkas, L. F., and Vinokurov, V. B. J . Less Common M e t . 1979,67,297.
140. Kliinenko, W. N. and Shunkowski, G. L. Planseeher. Pulverrnetall. 1981, 2, 405.
141. Okada, S., Satoh, M., and Agata. T. Nippon Kagaku Kuishi 1985, 685.
142. Liao, P. K. and Spear, K. E. Binar?, Alloy Phase Diagrams Vol. I , ASM Internat., Metals Park,
OH, 2nd Ed. 1990,471474.
143. Telegus, V. S. and Kuz’ma, Y. B. J . Sov. Powder Metallurgy and M e t . Ceram., 1968, 2, 133.
144. Rudy, E. and Windisch, S. Ternary Phase Equilibria in Transition Metal-Boron-Carbon-
Silicon Systems, Part I, Related Binary Systems, Technical Report No. AFML-TR-65-2,
Wright Patterson Air Force Base. OH, 1966.
145. Yasinskaya, G. A. and Groisberg, M. S. Sov. Powder Metall. Met. Ceram. 1963, 2, 457.
146. Kuz’ma, Yu. B., Svarichevskaya, S. I., and Telegus, V. S. Soviet Powder Metall. Met. Ceram.
1971, 10,478.
147. Kosterova, N. V. and Ordanyan, S. S. Zvest. Akad. Nauk U S S R Neorg. Muter. 1977, 13,
-
1411.
148. Ariel, E, Barta, J. and Niedzwiedz, S. J. Less-Common Met. 1970, 29, 199.
149. Ahn, D.-G., Kawasaki, A,, and Watanabe, R. Muter. Trans. JIM 1996, 37, 1078.
150. Pohl, A., Kitzler, P., Telle, R., and Aldinger, F. Z . Metallkunde 1994, 85, 658.
151. Pohl, A., Telle, R., and Petzow, G. 2. Metallkunde 1995, 86, 148.
152. Mitra, I. and Telle, R. in Werkstoflwoche 1996: Symposium 7: Materialwissenschaftliche
Grundlagen, Aldinger, F. and Mughrabi, H. (Eds), DGB Informationsgesellschaft, Oberursel,
1996, pp. 521-525.
153. Mitra, I. and Telle, R. J. Solid State Chem. 1997, 133, 25.
154. Schmalzried, C. and Telle, R. in Proc. Werkstoffwoche 98:\Band 7: KerumiklSimulation Kera-
mik, Heinrich, J., Ziegler, G., Hermel, W., and Riedel H. (Eds), Wiley-VCH, Weinheim, 1999,
pp. 377-382.
155. Telle, R. in Materials Science and Technology, Cahn, R. W., Haasen, P., and Kramer, E. J.
(Eds), Verlag Chemie, Weinheim, 1993, Vol. 11, Ch. 4, pp. 175-266.
156. Joly, A. C. R. Acad. Sci. 1883, 97, 456.
157. Moisson, H., C. R. Acad. Sci. 1894, 118, 556.
158. Ridgeway, R. R. Trans. Electrochem. Soc. 1934,66,117; Norton Co. US Patent I8972I4, 1933.
159. Lipp, A. Tech. Rundschau 1965, 57(14), 5; 57(28), 19; 57(33), 5.
160. Lipp, A. Tech. Rundschau 1966, 58(7), 3.
161. Dufek, G., Wruss, W., Vendl, A,, and Kieffer, R. Planseeber. Pulvermetall. 1976, 24, 280.
162. Gray, E. G. Eur. Patent Appl. 1 IS2 428, 1980.
163. Schwetz, K. A. and Lipp, A. in Ullmanns Encyclopedia of Industrial Chemistry, VCH, Wein-
heim, 1985, A4, pp. 295-307.
164. Thevenot, F. J . Eur. Ceram. Soc. 1990,6, 205.
165. Thevenot, F. in The Physics and Chemistry of Carbides, Nitrides, and Borides, Freer, R. (Ed.),
NATO AS1 Series E, Vol. 185, Kluwer, Dordrecht, The Netherlands, 1990, pp. 87-96.
166. McKinnon, I. M. and Reuben, B. G. J . Electrochem. Soc. 1975, 122, 806.
167. Knudsen, A. K. Adv. Ceram. 1987, 21, 237.
168. Walker, B. E., Rice, R. W., Becher, P. F., Bender, B. A,, and Coblenz, W. S. Ceram. Bull. 1983,
62, 916.
169. DeHoff, R. T., Rummel, R. A., LaBuff, H. P., and Rhines, F. N. in Modern Development in
Powder Metallurgy, Hausner, H. H. (Ed.), Plenum, New York, 1966, pp. 31&330.
170. Greskovich, C. and Rosolowski, J. H. J. Am. Ceram. Soc. 1976, 59, 336.
171. Prochazka, S. WhyIs It DifJult to Sinter Covalent Substances? General Electric Corp. Res. and
Dev. Center Technical Information Series, Report No. 89CRD025, Schenectady, NY, 1989.
172. Matje, P. and Schwetz, K. A. in Proc. 2nd Int. Conj: on Ceramic Powder Processing Science,
Hausner, H., Messing, G. L., and Hirano, S. (Eds), Deutsche Keramische Gesellschaft, Koln,
1989, pp. 377-384.
173. Shaw, N. J. Powder Met. Int. 1989, 21, 16; 31.
174. Grabchuk, B. L. and Kislyi, P. S. Sov. Powder Metall. Met. Ceram. 1976, 15, 675.
175. Adlassnig, K. Plunseeber. Pulvermetall. 1958, 6, 92.
176. Grabchuk, B. L. and Kislyi, P. S. Poroshkovaya Metallurgiya 1974, 8, 11.
177. Katz, J. D., Blake, R. D., and Petrovich, J. Met. Powder Rep. 1988, 43, 835.
178. Dole, S. L. and Prochazka, S. in Ceramic Engineering and Science Proc. 6(7/8), Smothers,
W. J. (Ed.), American Ceramic Society, Westerville, OH, 1985, pp. 1151-1160.
179. Kislyi, P. S. and Grabchuk, B. L. in Proc. 4th Eur. Symp. Powder Metallurgy, SOC.Franqaise
Metall., Grenoble, 1975, p. 10-2-1.
180. Kriegesmann, J. Ger. Patent DE 37 I I 871 C2, 1989.
181. Lange, R. G., Munir, Z. A., and Holt, J. B. Muter. Sci. Res. 1980, 13, 311.
182. Kanno, Y., Kawase, K., and Nakano, K. J. Ceram. Soc. Jpn. 1987, 95, 1137.
183. Glasson, D. R. and Jones, J. A. J. Appl. Chem. 1969, 19(5), 125.
184. Janes, S. and Nixdorf, J. Ber. Deutsche. Keram. Ges. 1966, 43, 136.
References 941
185. Stibbs, D., Brown, C. G., and Thompson, R. US Patent 3 749 571, 1973.
186. Zakhariev, Z. and Radev, D. J . Muter. Sci. Lett. 1988, 7, 695.
187. Prochazka, S. U S . Patent 4005235, 1977.
188. Kriegesmann, J. in Teclznische Keramische Werkstoffe, 4th Suppl., Kriegesmann, J. (Ed.),
Deutscher Wirtschaftsdienst, Koln, 1991.
189. Schwetz, K. A. and Vogt, G. Ger. Patent 2 751 998, 1977; U S Patent 4 195 066, 1980.
190. Henney, J. W. and Jones, J. W. S. Br. Patent Application 20141934, 1978.
191. Suzuki, H. and Hase, T. in Proc. Conf. Factors in Densification of Oxide and Nonoxide
Ceramics, Somiya, S. and Saito, S. (Eds), Japan, 1979, p. 345.
192. Schwetz, K. A. and Grellner, W. J . Less-Common Met. 1981, 82, 37.
193. Dole, S. L., Prochazka, S., and Doremus, R. H. J . Am. Ceram. Soc. 1989, 72, 958.
194. Bougoin. M., Thevenot, F., Dubois, F., and Fantozzi, G. J . Less-Common Met. 1985, 114,
257.
195. Matje, P., internal report, Elektroschmelzwerk Kempten GmbH, Germany, 1990.
196. Sigl, L. S. and Schwetz, K. A. Powder Metall. Int. 1991, 23, 221.
197. Sigl, L. S. and Schwetz, K. A. Euro-Ceramics II 1991, 1, 517.
198. Grabchuk, B. L. and Kislyi, P. S. Poroshkovaya Metallurgiya 1975, 7(151), 27. Engl. Transl.:
Sov. Powder Metall. Met. Ceram. 1975, 14, 538.
199. Oh, J. H., Orr, K. K., Lee, C. K., Kim, D. K., and Lee, J. K. J . Korean Ceram. Soc. 1985,22,
60.
200. Weaver, G. Q. U S Patent 4320204, 1982.
201. Weaver, G. Q. UK Patent GB 2093481 A , 1982.
202. Borchert, W. and Kerler, A. R. Metall 1975, 29, 993.
203. Schwetz, K. A,, Reinmuth, K., and Lipp, A. Sprechsaal 1983, 116, 1063.
204. Lorcher, R., Strecker, K., Riedel, R., Telle, R., and Petzow, G. in Solid State Phenomena 8&9,
Proc. Int. Conf: Sintering of Multiphase Metal and Ceramic Systems, Upadhyaya, G . S. (Ed.),
Sci-Tech Publications, Vaduz, Gower Publ. Co., Brookfield, VT, 1990, pp. 479492.
205. Bougoin, M. and Thevenot, F. J . Muter. Sci. 1987, 22, 109.
206. Halverson, D. C., Pyzik, A. J., Aksay, I. A,, and Snowden, W. E. J . Am. Ceram. Soc. 1989,72,
775.
207. Sigl, L. J . Europ. Ceram. Soc. 1998, 18(11), 1521.
208. Williams, P. D. and Hawn, D. C. J . Am. Ceram. Soc. 1991, 74, 1614.
209. Pyzik, A. J., Aksay, A,, and Sarikaya, M. Muter. Sci. Res. 1987, 21, 45.
210. Schwetz, K. A., Sigl, L. S., and Pfau, L. J . Solid State Chem. 1997, 133, 68.
21 1. Kuzenkova. P. S., Kislyi, P. S., Grabchuk, B. L., and Bodnaruk, N. I. J . Less-Common Met.
1979, 67, 217.
212. Ostapenko, I. T., Slezov, V. V., Tarasov, R. V, Kartsev, N. F., and Podtykan, V. P. Porosh-
kovaya Metallurgiya 1979,197,38. Engl. Transl.: Sov. Powder Met. Met. Ceram. 1979,19,312.
213. Brodhag, C., Bouchacourt, M., and Thevenot, F. in Ceramic Powders, Material Science
Monographs 16, Vincenzini, P. (Ed.), Elsevier, Amsterdam, 1983, pp. 881-890.
214. Beauvy, M. and Angers, R. Sci. Ceram. 1980, 10, 279.
215. Bouchacourt, M., Brodhag, C., and Thevenot, F. Sci. Ceram. 1981, 11, 231.
216. Ekbom, L. B. and Amundin, C. 0. Sci.Ceram. 1980, 10, 237.
217. Champagne, B. and Angers, R. J . Am. Ceram. Soc. 1979, 62, 149.
218. Telle, R. and Petzow, G. Muter. Sci. Eng. 1988, A105/106, 97.
219. Vasilos, T. and Dutta, S. K. Ceram. Bull. 1974, 53, 453.
220. Furukawa, M. and Kitahira, T. Nippon 7ungsten Rev. 1979, 12, 55.
221. Larker, H. T., Hermansson, L.. and Adlerborn, J. Ind. Ceram. 1988, 8, 17.
222. Schwetz, K. A., Grellner, W., and Lipp, A. in Science of’Hard Materials, Proc. 2nd Int. Conf.
Science of Hard Materials. Almond. E. A,. Brookes. C. A,. and Warren. R. (Eds’l. Inst. Phvs.
\
Conf. Series, Elsevier, London, 1986, pp. 41 5426.

223. Schwetz, K. A,, “Boron Carbide, Boron Nitride, and Metal Borides”, in Ullmann’.~Encyclo-
pedia of Industrial Chemistry, 6th ed. on CD-ROM, Wiley-VCH, Weinheim, Germany, 1999.
224. Schwetz, K. A,, de Groot, K., and Malkemper, W. J . Less-Common Metals 1981, 82, 153.
225. Bouchacourt, M. and Thevenot, F. J . Muter. Sci. 1985, 20, 1237.
226. Werheit, H. Prog. Crystal Growth and Charact. 1988, 16, 179.
227. Wood, C., Emin, D., and Gray, P. E. Phys. Rev. B 1985, 31, 6811.
228. Aselage, T. L. and Emin D. Proc. 10th Int. Symp. Boron, Borides, and Related Compounds,
Albuquerque, NM, Boron-Rich Solids, AIP Con$ Proc. 1990, 231, 177.
229. Heuberger, M., Telle, R., and Petzow, G. Powder Metallurgy 1992, 35, 125.
230. Heuberger, M. thesis, University of Stuttgart, 1995.
231. Telle, R. in Boron-Rich Solids, Proc. 10th Int. Symp. Boron, Borides, and Related Com-
pounds, Emin, D., Aselage, T. L., Switendick, A. C, Morosin, B., and Beekel, C. L. (Eds),
AIP Conf. Proc. 231, American Institute of Physics, New York, 1991, pp. 553-560.
232. Telle, R. Ceramics for High-Tech Applications, Materials Science, European Concerted Action
COST 503 Powder Metallurgy-Powder Based Materials, Vol. V, Valente, T. (Ed.), The Eur-
opean Communities, Brussels, 1997.
233. Telle, R. and Petzow, G. in Proc. 9th Int. Symp. on Boron, Borides, and Related Compounds,
Werheit, H. (Ed.). University Press, Duisburg, 1987, pp. 234-245.
234. Efimenko, L. N., Lifshits, E. V., Ostapenko, I. T., Snezhko, I. A,, and Shevyakova, E. P. Sov.
Powder Met. and Met. Ceram. 1987, 26(3), 318.
235. Litz, L. M. and Mercuri, R. A. J. Electrochemical Soc. 1963, 110(8), 921.
236. Motzfeld, K. Acta Chem. Scand. 1964, 18, 1596.
237. Hofmann, H. and Petzow, G. J. Less-Common Met. 1986,117, 121.
238. Telle, R. and Petzow, G. in Horizons of Powder Metallurgy 11, Proc. 1986 Int. Powder Metall.
Conf. Exhibit., Kaysser, W. A. and Huppmann, W. J. (Eds), Verlag Schmid, Freiburg, 1986,
pp. 1155-1158.
239. Telle, R., Meyer, S., Petzow, G., and Franz, E. D. Muter Sci. Eng. 1988, A105/106, 125.
240. Kim, D. K. and Kim, C. H. Adv. Ceram. Muter. 1988,3, 52.
241. Nishiyama, K. and Umekawa, S. Trans. Jpn. Soc. Ceram. Met. 1985, 11(2), 53.
242. Rice, R. W., Richardson, G. Y., Kunetz, J. M., Schroeter, T., and McDonough, J. in Proc.
10th Annu. Con$ Composites and Advanced Ceramic Materials, Messier, D. R. (Ed.), American
Ceramic Society, Westerville, OH, 1986, p. 737.
243. Richardson, G. Y., Rice, R. W., McDonough, W. J., Kunetz, J. M., and Schroeter, T. in Proc.
10th Annu. Con$ Composites and Advanced Ceramic Materials, Messier, D. R. (Ed.), American
Ceramic Society, Westerville, OH, 1986, p. 760.
244. McCauley, J. W., Corbin, N. D., Resetar, T., and Wong, P. in Proc. 10th Ann. Con$ Com-
posites and Advanced Ceramic Materials, Messier, D. R. (Ed.), American Ceramic Society,
Westerville, OH, 1986, pp. 538-554.
245. Holt, B., Kingman, D. D., and Bianchi, G. M. Muter. Sci. Eng. 1985, 71, 321.
246. McCauley, J. W. Am. Ceram. Soc. Bull. 1988, 67, 1903.
247. Faber, K. T. and Evans, A. G. Acta Metallurgica 1983, 31(4), 565.
248. Selsing, J. J . Am. Ceram. SOC.1961, 44,419.
249. Fendler, E., Babushkin, O., Lindback, T., Telle, R., and Petzow, G. in Proc. IntSymp.
Ceramic Materials and Components,for Engines, Gothenberg, Sweden, June 10-12, 1991.
250. Fendler, E., Babushkin, O., Lindback, T., Telle, R., and Petzow, G. J . HardMater. 1993,4, 137.
251. Lonnberg, B. J. Less-Common Met. 1987, 141, 145.
252. Rey, J. and Male, G. in Proc. 9th Int. Symp. on Boron, Borides, and Related Compounds,
Werheit, H. (Ed.), University Press, Duisburg, 1987, pp. 419420.
253. Ohya, Y. and Hoffmann, M. J., Petzow, G. J. Am. Ceram. Soc. 1992, 75, 2479.
254. Sigl, L. S. and Schwetz, K. A. Jap. J. Appl. Physics 1994, 10, 224.
255. Sigl, L. S. and Kleebe, H.-J. J . Am. Ceram. Soc. 1995, 78(9), 2374.
256. Thevenot, F. and Bouchacourt, M. L’lndustrie Ceramique 1979, 732, 655.
257. Binder, F. Radex-Rundschau 1975, 4, 531.
258. Fitzer, E. Arch. Eisenhiittenwesen 1973, 44(9), 703.
259. Coble, R. L. and Hobbs, H. A. in Investigation of Boride Compounds for Very High Tem-
perature Applications, NTIS Report AD 428006, Kaufman, L. and Clougherty, E. V. (Eds),
Clearinghouse for Federal Scientific and Technical Information, Springfield, VA, 1973,
pp. 82-120.
260. Kislyi, P. S. and Zaverukha, 0.V. Poroshkovaya. Metallurgiya 1970,7, 32. Engl. Transl.: Sov.
Powder Metall. Met. Ceram. 1970, 7, 549.
261. Kislyi, P. S., Kuzenkova, M. A,, and Zaveruha, 0. V. Phys. Sintering 1972, 4, 107.
References 943
262. Baumgartner, H. R. and Steiger, R. A. J . Am. Ceram. Soc. 1984, 67, 207.
263. Baik, S. and Becher, P. J . Am. Ceram. Soc. 1987, 70, 527.
264. Cech, R., Olivierus, P., and Seijbal, J. Powder Metall. 1965, 8(15), 142.
265. Rasskazov, N. I. in Proc. 3rd Znt. Powder Metallurgy Conf. 2, Karlovy Vary, Czechoslovakia,
1970, pp. 228-238.
266. Krylov, Yu. I., Bronnikov, V. A., Krysina, V. G., and Prislavko, V. V. Sov. Powder Metall.
Met. Ceram. 1976, 15, 1000.
267. Munir, Z. A. Am. Ceram. Soc. Bull. 1988, 67, 342.
268. Ouabdesselam, M. and Munir, Z. A. J . Muter. Sci. 1987, 22, 1799.
269. Clougherty, E. V. and Pober, R. L. Nucl. Met. 1964, 10, 422.
270. Samsonov, G. V., Goryachev, Y. M., and Kovenskaya, B. A. Izv. Vuz Fiz. 1972, 6, 37.
271. Castaign, J. and Costa, P. in Boron and Refractory Borides, Matkovich, V. I. (Ed.), Springer,
Berlin, Heidelberg, New York, 1977, pp. 390412.
272. Lonnberg, B. J . Less-Common Metals 1988, 141,145.
273. Rahman, M. and Wang, C. C., Chen, W., Akbar, S. A. J . Am. Ceram. Soc. 1995, 78, 1380.
274. Billehaug, K. and 0ye, H. A. Aluminium 1980, 56, 642.
275. Serrlie, M. and 0ye, H. A. Cathodes in Alurniniuni Electrolysis, 2nd ed., Aluminium-Verlag,
Dusseldorf, 1994.
276. Bannister, M. K. and Swain, M. V. Ceram. Znt. 1989, 15, 375.
277. Murata, Y., Julien, H. P., and Whitney, E. D. Ceram. Bull. 1967, 45, 643.
278. Venables, J. D. Philos. Mag. 1967, 16, 873.
279. Williams, W. S. Trans AZME 1966, 236, 211.
280. Ramberg, J. R. and Williams, W. S. J . Muter. Sci. 1987, 22, 1815.
281. Kang, E. S. and Kim, C. H . J . Muter. Sci. 1990, 25, 580.
282. Ly Ngoc, D. Dr. rer. nut Thesi.s, University of Stuttgart, 1989.
283. Jimbou, R., Takahashi, K., and Matsushita, Y. Adv. Ceram. Muter. 1986, 1, 341.
284. McMurtry, C. H., Bocker, W. D. G., Seshadri, S. G., Zanghi, J. S., and Garnier, J. E. Am.
Cerani. SOC. Bull. 1986, 66, 325.
285. Takahashi, K., Jimbou, R. Commun. Am. Ceram. Soc. 1987, C-369.
286. Janney, M . A. Am. Ceram. Soc. Bull. 1986, 64(5), 357.
287. Janney. M. A. Am. Ceuam. Soc. Bull. 1987, 66(2), 322.
288. Cai, H., Gu, W-H., and Faber, K. T. in Proc. Am. Soc. Composites, 5th Techn. Conf. on Com-
posite Materials, 1990, pp. 892-901.
289. Faber, K. T., Gu, W.-H., Cai, H., Winholtz, R. A,, and Magleg, D. J. in Toughening Mechan-
isms in Quasi-Brittle Materials, Shah, S. P. (Ed.), Kluwer, Dordrecht, The Netherlands, 1991,
pp. 3-17.
290. de Mestral, F. and Thevenot, F. in The Physics and Chemistry of Carbides. Nitrides, and Borides,
NATO AS1 Series E, Vol. 185, Freer. R. (Ed.), Kluwer, Dordrecht, The Netherlands, 1990,
pp. 457-482.
291. Phan-Tan-Luu, R., Mathieu, D., and Feneuille, D. Mbthodologie de la Recherche Expirimen-
tale, Fascicules de cours, L.P.R.A.I., Universite d’Aix-Marseille, 1989.
292. Tanaka, H. and Iyi, N. J . Am. Ceram. Soc. 1995,78, 1223.
293. Watanabe, T. and Kouno, S. Ceram. Bull. 1982, 61, 970.
294. Watanabe, T. J . Am. Ceram. Soc. 1977, 60, 176.
295. Watanabe, T. Am. Ceram. Soc. Bull. 1980, 59, 465.
296. Petzow, G. and Telle, R. in Advanced Ceramics, Somiya, S. (Ed.), Terra Scientific Publ. Co,
Tokyo, 1987, pp. 131-144.
297. Watanabe, T. and Shobu, K. Yogo-Kyokai-Shi1988, 96(7), 778.
298. Shobu, K. and Watanabe, T. Yogo-Kyokai-Shi1987, 95(1), 991.
299. Telle, R., Meyer, S., Petzow, G., and Franz, E. D. Marer. Sci. Eng. 1988, A105/106, 125.
300. Watanabe. T. and Shobu, K. J . Am. Ceram. Soc. 1985,68, C34.
301. Shobu, K., Watanabe, T., Drennan, J., Hannink, R. H. J., and Swain, M. V. in Proc. 4th Int.
Conf: Sci. Techn. Zirconia Advanced Ceramics, Zirconia 86, American Ceramic Society,
Westerville. OH, 1987, pp. 1091-1099.
302. Swain, M. J . HardMater. 1991, 2, 139.
303. Franz, E.-D., Tumback, M., and Telle, R. Sprechsual 1992, 125, 55.
304. Tumback, M., Franz, E.-D., and Telle, R. Sprechsaal 1992, 125,415.
305. Miiller, R. thesis, University of Stuttgart, 1995.
306. McHale, A. E. and Scott, R. S. J. Am. Ceram. SOC.1986,69, 827.
307. Takagi, K., Komai, M., Ando, T in Sintering ’87, Vol. 2, Proc. 4th Inst. Symp. Science and
Technology of Sintering, Somiya, S., Simada, M. (Eds.), Elsevier, Amsterdam, 1988, pp.
1296-1 301.
308. McMeeking, R. M. J. Am. Ceram. SOC.1986, 69, C-301.
309. Ogata, T. Toray Zndustries Internal Report, private communication, 1989.
310. Watanabe, T., Shobu, K., Tani, E., and Nakanishi, T. in Proc. 11th Znt. Symp. Boron. Borides
and Related Compounds, JJAP Series 10, Uno, R. and Higashi, I. (Eds), Publication Office, Jap.
J. Applied Physics, Tokyo, 1994, pp. 208-21 1.
3 11. Hamjian, H. J. and Lidman, W. G. J. Am. Ceram. SOC.1952, 35, 44.
312. Halverson, D. C., Pyzik, A. J., and Aksay, I. A. Ceram. Eng. Sci. Proc. 1985, 6, 736.
3 13. Nishiyama, K. and Umekawa, S. in Achievements in Composites in Japan and the United States,
Kobayashi, A. (Ed.), Proc. Vth Japan-US Conference on Composite Materials, Tokyo, 1990,
pp. 371-378.
314. Hayes, F. H., Lukas, H. L., Effenberg, G., and Petzow, G. Z . Metallkunde 1989,80, 361.
315. Zupanic, F., Spaic, S., and Krizman, A. Mater.Sci. Technol. 1998, 14, 1203.
316. Funke, V. F., Yudkovskii, S. I., Samsonov, and G. V. Russ. J. Appl. Chem. 1961,34,973.
317. Tangermann, I. Neue Hutte 1963,8, 973.
318. Samsonov, G . V. Metallov. i. Obrab. 1958, 12, 35.
319. Kieffer, R. and Benesovsky, F., Hartstoffe, Springer, Vienna, 1963.
320. Golczewski, J. and Aldinger, F., poster presentation at Annu. Mtg. Germany Society ,fbr
Materials (DGM), Gottingen, 1994.
321. Funke, V. F. and Yudkovskii, S. I. Zhur. Fiz. Khim. 1963, 37, 1557; translation into English:
Russ. J. Phys. Chem. 1963,37, 835.
322. Jiingling, T., Oberacker, R., Thiimmler, F., Sigl, L. S., and Schwetz, K. A. J. Hard Muter.
1991, 2(34), 183.
323. Jiingling, T., Oberacker, R., Thiimmler, F., and Sigl, L. S., in Proc. Znt. Conf.Advances in Hard
Materials Production, 1992, pp. 15/1-15/14.
324. Pastor, H., Allibert, C. H., Ottavi, M. L., Albajar, M., and Castro, F. French Patent Applica-
tion No. 91 08030, 1991.
325. Pastor, H. et al., in HMP’Y2 - Advances in Hard Materials Production, Proc. 4th Int. Conf.,
MPR Publishing Services, Bellstone, UK, 1992, pp. 23/1-23-11.
326. Sinchez, J. M., Barandika, M. G., Gil-Sevillano, J., and Castro, F. Scripta Metall. Muter.
1992, 26, 957.
327. Brynestad, J., Bamberger, C. E., Land, J. F., Fandinch, C. B. J. Am. Ceram. SOC.1983,66, C215.
328. Ghetta, V., Gayraud, V., and Eustathopoulos, N. presented at Sintering ‘ Y I , Vancouver,
August, 1991.
329. Li, J. G. J. Am. Ceram. SOC.1992,75, 3118.
330. Li, J. G. Rare Met. 1992, 11, 177.
331. Savov, L., Heller, H.-P., and Janke, D. Metall. 1997, 51, 475.
332. Panasyuk, D. and Umansky, J. J. Less-Common Met. 1986, 117,336.
333. Schwetz, K. A., Reinmuth, K., and Lipp, A. Radex Rundschau 1981, 3, 568.
334. Janaf, Thermochemical Tables, 2nd ed., NBS, Washington, DC, 1971.
335. Kubaschewski, 0. and Alcock, C. B. Metallurgical Thermochemistry, Int. Series on Materials
Science and Technology, Vol. 24, Pergamon, Oxford, 1979.
336. Matuschka, A. V. Borieren, Carl Hanser Verlag, Munich, 1977.
337. Schwetz, K. A., Reinmuth, K., and Lipp, A. Radex Rundschau 1981, 3, 568.
338. Sigl, L. S., Jiingling, T. Proc. 4th Int. Conf. Sci. Hard Materials (ICSHM4), Funchal, Madeira,
NOV.10-15, 1991, p. 321.
339. Ottavi., L., Chaix, J. M., Allibert, C., and Pastor, H. presented at Sintering ‘91, Vancouver,
August, 1991.
340. Jiingling, T., Dr.-Zng. Thesis, University of Karlsruhe, Germany, 1992, IKM-Series 007.
341. Liao, P. K. and Spear, K. E. in Binary Alloy Phase Diagrams, Vol. 1, Massalski, T. B., et al.
(Eds.), American Society of Metals, Metals Park, OH, 1986, p. 355.
References 945
342. Murray, J. L. in Binary Alloy Phase Diagrams, Vol. 1, Mas-salski, T. B., et a/. (Eds.), American
Society of Metals, Metals Park, OH, 1986, p. 1118.
343. Jiingling, T., Sigl, L. S., Oberacker, R., Thiimmler, F., and Schwetz, K. A. Int. J . Refractory
Met. Hard Muter. 1993194, 12, 71.
344. Angelini, P., Becher, P. F., Bentley, J., Brynestad, J., Ferber, M. K., Finch, C. B., and Sklad,
P. S. in Science of Hard Materials, Proc. 2nd Int. Cohf. on Science of Hard Materials, Almond,
E. A,, Brookes, C. A,, and Warren, R. (Eds), Inst. Phys. Conf. Ser. 75, Adam Hilger, Bristol,
1986, pp. 1019-1032.
345. Sklad, S. and Yust, C. S. in Proc. 1st Int. Con$ Science ofHardMaterials, Hagenmiiller, R. and
Thevenot, F. (Eds), Bordeaux, I98 1.
346. Takatsu, S. and Ishimatsu, E. in Proc. 10th Plansee-Seminar on Powder Metall, Planseewerke,
Reutte, 1981, Vol. 1, p. 535.
347. Finch, C. B., Cavin, 0. B., and Becher, P. F. J . Cryst. Growth 1984, 67, 556.
348. Nishiyama, K. and Umekawa, S. in Proc. Composites '86: Recent Advances in Japan and in the
United States, Proc. 3rd Japan-US Conf. on Composite Materials, Tokyo, Kawata, K., Ume-
kawa, S., Kobayahi, A. (Eds.), 1986, pp. 433440.
349. Telle, R., Schmalzried, C., and Schunck, B, unpublished report to the Volkswagen-Founda-
tion, 1999.
350. Johnson, W. B., Nagelberg, A. S., and Breval, E. J . A m . Ceram. Soc. 1991, 74, 1093.
351. Newkirk, M. S., Urquhart, A. W., and Zwicker, H. R. J . Muter. Res. 1986, 1, 81.
352. Antolin, S., Nagelberg, A. S. J . Am. Ceram. Soc. 1992, 75, 447.
353. Nagelberg, A. S . , Antolin, S., and Urquhart, A. W. J . Am. Ceram. Soc. 1992, 75(2), 455
354. Weinstein, J. in Proc. Int. Symp.Advances in ProcessinE andCharacterization of Ceramic Metal
Matrix Composites, CIMjICM; Vol. 17: Mostaghaci, H. (Ed.), Pergamon, Oxford, 1989,
p. 132.
355. Lugscheider, E. and Eschnauer, H. in Proc. 9th Int. Symp.on Boron, Borides, and Related Com-
pounds, Werheit, H. (Ed.), University Press, Duisburg, 1987, pp. 202-212.
356. Komai, M., Yamasaki, Y., Takagi, K., and Watanabe, T in Advances in Powder Metallurgy
and Particulate Materials 8, Capus, J. M. and German, R. M. (Eds), Metal Powder Industries
Federation, Princeton, NJ, 1992, pp. 81-88.
357. Rieger, W., Nowotny, H., and Benesowsky, F. Monatsli. Chem. 1966, 97, 378.
358. Ozaki, S., Yamasaki, Y., Komai, M . , and Takagi, K. in Proc. 11th Int. Symp. Boron, Borides
and Related Compounds, JJAP Ser. 10, Uno, R., Higashi, I. (Eds.), Publication Office, Jap.
J. Applied Physics, Tokyo, 1994, pp. 220-221.
359. Takagi, K. in Proc. 11th Int. Symp. Boron. Borides and Related Compounds, JJAP Series 10,
Uno, R. and Higashi, I. (Eds), Publication Office, Jap. J. Applied Physics, Tokyo, 1994, pp.
200-203.
360. Sivaraman, K., German, R. M., and Takagi, K. in Proc. Int. Con/ Powder Metallurgy PM'96,
Toronto, 1996, pp. 9/117-9/129.
361. Sivaraman, K., Griffo, A,, German, R. M., and Takagi, K. in Proc. Int. Confi Powder Metal-
lurgy PM'96, Toronto, 1996, pp. 11/67-11/79.
362. Matsuo, S., Ozaki, S., Yamasaki, Y., Komai, M., and Takagi, K. in Sintering Technology,
German, R. M., Messing, G. L., and Cornwall, R.G. (Eds), Marcel Dekker, New York,
1996, pp. 261-268.
363. Warren, R. J . Muter. Sci. 1968, 3, 471.
364. Schmalzried, C. and Telle, R. in Proc. Int. Conf: on Boron, Borides and Related Compounds
( I S B B '99). Dinard, France, October 1999, to be published in J . Solid State Chem.
365. Takagi, K., Yamasaki, Y., and Komai, M. J . Solid State Phys. 1997, 133, 243.
366. Sigl, L. S., Schwetz. K. A., and Dworak, U. Int. J . Refractory and Hard Materials, 1994, 12,
pp. 95-99.
367. Ramberg, J. R., Wolfe, C. F., and Williams, W. S. J . Am. Ceram. Soc. 1985, 68,C78.
368. Freitag, B. and Mader, W. J . Microsc. 1999, 194, 42.
8 The Hardness of Tungsten Carbide-Cobalt

Hardmetal
S. Luyckx
8.1 Introduction
WC-Co hardmetal is a sintered material consisting of brittle tungsten carbide (WC)
crystals bonded by a tough cobalt-based binder. Figure 1 shows the microstructure
of a typical commercial WC-Co hardmetal.
WC-Co is used mostly on account of its hardness which is high for a metallic
material, although other outstanding properties, such an extremely high Young
modulus, a high thermal conductivity, and a low coefficient of thermal expansion,
contribute to its success in a wide range of technical applications.
This chapter reviews the information available on the dependence of the hardness
of WC-Co on microstructural parameters, composition and external conditions,
such as temperature. It also reviews information available on the effect of micro-
structure and composition on the relationships between hardness and toughness
and hardness and abrasive wear resistance.
The hardness of industrial hardmetal grades is characterized preferentially by the
Rockwell A scale [l] in the USA [2] and by the Vickers scale [3] in Europe. Most
research work, however, has been carried out, also in the USA, using the Vickers
scale. Most of the results reported below were obtained using standard pyramidal
Vickers indenters [ 3 ] .
Figure 1. Microstructure of a typical WC-6 weight-% Co alloy.

Table 1. The hardness of three crystallographic

faces of a WC single crystal [5].
WC crystal plane Hardness, HVI
(0001) 2100
(1 100) 1080
(1 101) 1060
8.2 The Hardness of the Two Component Phases

8.2.1 The Hardness of Tungsten Carbide
WC is present in hardmetal in the form of single crystals. These have a hexagonal
structure and are highly anisotropic [4]. Table 1 lists the Vickers hardness of three
WC crystal faces and shows that the hardness can vary from face to face by up to
100% [5]. On account of the superior hardness of the (0001) faces, attempts have
been made to produce WC-Co parts with a (0001) texture [6].
The literature contains contradictory results on the hardness of WC because it has
been measured mostly on sections of hardmetal samples, where the grains are
randomly oriented, rather than on single crystals.
An additional source of contradictory results is the dependence of hardness on
the load applied to the indenter when microhardness is measured (and very low
indenting loads must be used when measuring the hardness of WC grains in hardmetal
samples, because the grains are typically between 1 and 10 pm in size). Figure 2 [7] is a
34
32
-3
5 28
In
m
0)
f 26
m
r
24
22
I
201
1
0
Figure 2. Plot of the microhardness of a (0001) plane of a WC single crystal against the indenting load.
948 8 The Hardness of Tungsten Carbide-Cobalt Hardmetal
15!
10 15 20 25 30 35 40
l/d112(mm-’12)
Figure 3. Plot of the hardness of hot-pressed polycrystalline WC against d - ‘ I 2 , where d is the mean
size of the WC grains. Adapted from reference [9].
plot of the microhardness of the (0001) planes of WC crystals versus the indenting
load. It shows that the microhardness of WC decreases with increasing load, which
means that when measuring the microhardness of WC on sections of WC-Co samples
one obtains different results at different loads.
The hardness of WC has also been measured on polycrystalline samples [8].
Figure 3 reports results from several investigators [9] which satisfy the following
Hall-Petch relationship:
HWC= HOWC + KOWCd-1’2, (1)

where Hwc = hardness of polycrystalline WC, d = mean grain size of the randomly
oriented WC grains, HOwc = average hardness of a WC crystal (over all possible
crystallographic planes), and KOwc= Hall-Petch coefficient, related to the ease of
slip transfer across WC-WC grain boundaries.
From Fig. 3 it has been calculated that Hwc = 1330 kgfmmP2 and Kowc =
24 kgf mm-3/2. It must be noted that the results reported in Fig. 3 were obtained
at room temperature, thus the values of HOwcand KOwcare expected to be those
given above only at room temperature.
8.2.2 The Hardness of Cobalt

The binder in WC-Co hardmetal is not pure cobalt but a solid solution of carbon
and tungsten in cobalt. When cooling the material from the sintering temperature
Figure 4. Plot of the hardness of the cobalt binder in WC-Co alloys against where X is the
mean width of the binder layers, or binder mean free path. Adapted from reference [9].
to room temperature only part of the cobalt transforms from the high temperature
(above 417°C) f.c.c. structure to the low temperature h.c.p. structure, and the
transformed regions are present as hexagonal lamellae in a predominantly cubic
material [lo].
The hardness of pure cobalt has been reported as being between 140 and 240 HV,
depending on the method of sample preparation [l I]. However, a solid solution of C
and W in cobalt can be up to 100% harder than pure cobalt [12], depending on the
composition. It has been difficult to measure directly the hardness of the binder in
WC-Co because the width of the binder regions in the hardmetal is typically of the
order of 0.1 pm.
However, Fig. 4 summarizes measurements by a number of investigators of the
in situ hardness of the cobalt binder [9]. The hardness is plotted versus X-”2,
X being the thickness of the cobalt layer where the hardness was measured, usually
called ‘the cobalt mean free path’. The results in Fig. 4 satisfy the following Hall-
Petch type relationship
Hco = Hoco + KocoX-”2 (2)
where Hco = the in situ hardness of the cobalt layers, X = thickness of the layers,
Hoco = mean hardness of the binder and Koco = Hall-Petch coefficient.
From Fig. 4 it has been calculated that Hot,, = 149kgfmmP2 and Koco =
16 kgfmmp’/2. Koco is related to the ease of slip transfer across Co-WC interfaces.
The contribution of the Co-Co grain boundaries to the hardness of the material can
be neglected because in WC-Co the size of the cobalt grains is of the order of lo3 pm
while the size of the WC grains is of the order of 1 pm. Thus the number of Co-Co
boundaries is negligible compared to the number of Co-WC interfaces.
8.3 Factors Affecting the Hardness of WC-Co

Hardmetal
8.3.1 Cobalt Content and Tungsten Carbide Grain Size
The two main factors affecting the hardness of WC-Co are the cobalt content and
the tungsten carbide grain size. It is well established that the hardness of WC-Co
decreases with increasing cobalt content at all carbide grain sizes and decreases
with increasing grain size at all cobalt contents [13]. Figures 5 and 6 confirm this
trend for the most comprehensive range of cobalt contents tested to date [14,15].
As a result, by selecting appropriate combinations of cobalt contents and carbide
grain sizes, the hardness of WC-Co can be varied from below 800 HV to more than
2000 HV. The most appropriate combination is determined by the properties which,
besides hardness, are required for a specific application. The recent introduction of
grain sizes of the order of lo2 nm has extended the range of possible hardness well
above 2000HV [16]. However, nano-grade hardmetals are not included in the
present review because they always contain grain refiners such as VC or Cr2C3,
while this review is limited to two-phase WC-Co alloys.
Lee and Gurland [8] attempted to express the relationship that exists between the
hardness of WC-Co, its microstructure and its composition by using a simple law of
mixtures of the type
Hwc-co = Hwc vwc + HCO(1 - VWC) (3)
2200
-UF+F *M -++C
2000-
1800-
..................................
.............................................................................................................................
......................................................................................
1000 - ...................................
800- .............................................
600- .....................................................................................................
400 I
0 5 10 15 20 25 30 35 40 45 50
I I I I I I I I
Cobalt content (wt%)

Figure 5. Plot of the hardness of WC-Co alloys against their cobalt content at four different carbide
grain sizes. The grain sizes are indicated in the key where: U F = 0.7 pm; F = 1 .O pm;M = 3.0 pm;
C = 5.0 pm. Reproduced by permission of the Int. J. Refract. Met. & Hard Muter.
8.3 Factors Affecting the Hardness of WC-Co Hardmetal 95 1
2200 I I I I I I I I I i
Figure 6. Plot of the hardness of WC-Co alloys against carbide grain size at the cobalt contents indi-
cated in the key. Reproduced by permission of L. Makhele.
where Hwcpc0= the hardness of the alloy, Hwc and Hco = the hardness of the
component phases as expressed by Eqs (1) and (2), and Vwc = volume fraction of
the WC phase.
However, Lee and Gurland found that this simple law does not agree with the
measured hardness values of WC-Co alloys. In order to reproduce the measured
values, they had to introduce the concept of ‘continuous carbide volume’ or
‘volume fraction of the carbide skeleton’, V,, which is defined as the volume fraction
occupied by that part of the WC grains which are involved in forming infinitely long
chains of connected particles while those parts which are surrounded by the cobalt
binder do not contribute to V,. Lee and Gurland derived that V, is related to the
volume fraction of the carbide phase, Vwc, via the contiguity, C:
v, = cvw, (4)
where the contiguity, C , is defined as the ratio of the grain boundary area and the
total surface area of the carbide phase.
By replacing Vwc in Eq. (3) with V,, Lee and Gurland obtained the following
equation:
Hwc-co = CVWCHWC + (1 - CVWC)HCO (5)
which is in excellent agreement with experimental results (see Fig. 7).
1800
.
iii
E
E
m
& 1400
1600
/c-
I I
v)
5
(I)
1200
c
m
t
Y
0
1000
5
800
5
-0
0
2 600
400
4 0 600 800 1000 1200 1400 1600 1 DO
Calculated hardness (kg/mma)
Figure 7. Plot of the hardness of WC-Co alloys, calculated from Eq. (9,against measured hard-
ness. Adapted from reference [8].
By using Eq. ( 5 ) it is possible to calculate that in commercial WC-Co grades (i.e.

grades with cobalt content ranging from about 6 to about 20 weight-%) the first
term of the equation accounts for up to 80% of Hwcpc0,which means that the
continuous carbide volume accounts for up to 80% of the hardness of the alloy.
8.3.2 Grain Size Distribution and Cobalt Mean Free Path

A wide carbide grain size distribution obviously leads to variations in the micro-
hardness of the hardmetal, but it has not been established what effect it has on the
macrohardness. However, if variations in grain size distribution lead to variations
in the binder mean free path the effect on hardness can be substantial, as shown in
Fig. 8 [17]. Figure 8 shows that at equal mean free path (and so equal binder hardness,
according to Eq. (2)), finer grades have lower hardness than coarser grades.
8.3.3 Binder Composition and Carbon Content

Although the hardness of hardmetal is determined mostly by the carbide phase (see
Section 8.3.1), the hardness of the binder does contribute to the overall hardness of
the material. Therefore the composition of the binder, and specifically the amount of
W dissolved in the binder, affects the overall hardness.
The composition of the binder depends on the total carbon content of the material
since, according to Exner [ 181, there seems to exist a reciprocal relationship between
8.3 Factors Affecting the Hardness qf WC-Co Hardmetal 953
0 2 4 6 a 10 12
mean free path (um)
Figure 8. Plot of the hardness of WC-Co alloys of four different grain sizes against the binder mean
free path U F = 0 7pm, F = 1 Opm, M = 3 Opm, C = 5 Opm Reproduced by permission of the
Int J of Refract Met & Hard Mater
the dissolved W and the dissolved C. The higher the total carbon, the lower is the
amount of W dissolved in the binder and the lower is the hardness of the binder
(Fig. 9) [19].
8.3.4 Porosity
The hardness of materials produced via powder metallurgy is known to decrease
with increasing porosity level. However, hardmetals are almost pore-free materials
and variations in their extremely low level of porosity does not affect their hardness
appreciably.
8.3.5 Effect of Temperature

In most applications WC-Co is subjected to temperatures which can be as high as
900-1000°C (as in metal cutting). Therefore knowledge of the effect of temperature
on hardness is essential for a correct selection of the grades suitable for a specific
application.
The hardness of both WC and cobalt decreases with increasing temperature.
Figure 10 shows the rate of hardness decrease with temperature for hot pressed
polycrystalline WC and for WC single crystals [20] and Fig. 11 shows the rate of
hardness decrease in cobalt [I 11.
90.5
go - ......................................................
~ 89.5 - ......................................................
LII
v)
v)
89 ......................................................
E
88.5 _.....................................................
.
I
_
88 ......................................................
87 7
carbon -
87.5 _ ......................................................
---High Low carbon
0.1 1
mean free path in Co (um)
Figure 9. Plot of the hardness of WC-Co alloys of different carbon contents against the binder mean
free path. Adapted from reference [19].
The most comprehensive work on the effect of temperature on the hardness of

WC-Co has been carried out by Milman and co-workers [21] and has involved
WC-Co grades ranging in grain size from 0.5 to 2.3pm and in cobalt content
from 6 to 15 weight-%. These grades have been tested at temperatures ranging
from 20 to 900°C and the results are summarized in Figs 12 and 13. Figure 14
gives the same results in Hall-Petch coordinates, and shows that at all temperatures
and cobalt contents a linear relationship exists between the hardness of WC-Co,
Hwc-c0,and d-'I2, d being the mean grain size of the WC grains.
The linear relationships shown in Fig. 14 can be explained by combining Eqs (l),
(2), and (5) and by expressing A, the mean free path in cobalt, in terms of d using the
22
.;a 20
n
c
2
In 18
5
In
a,
16
0
.c
8 14
0
4 12
10
8
100 200 300 400 500 600 700 800 900
temperature (
'C)
Figure 10. Plots of the hardness of WC against temperature. Adapted from reference [20].
8.3 Factors Affecting the Hurdness of' WC-Co Hardmetal 955
200
180
160
140
120
100
80
60
40
20
0 100 200 300 400 500 600 700 800
temperature (OC)
Figure 11. Band including possible values of the hardness of cobalt at various temperatures.
Adapted from reference [I 11.
following equation due to Lee and Gurland [8]:
By combining the above equations one obtains
where
and
with
Ho and K y can be considered constant at constant carbide volume fraction, Vwc,

since the contiguity, C, varies with cobalt content but does not vary appreciably
with grain size [8]. Hence the relationship between Hwc-co and d-'I2 in Eq. (7) is
24
22 -- d = 0.5 pm
m
h
a
c7
\
6 - r ~
15 wt% co
4
0 100 200 300 400 500 600 700 800 900 0 100 200 300 400 500 600 700 800
Temperature ("C) Temperature ("C)
0 100 200 300 400 500 600 700 800 900

Temperature ("C)
Figure 12. Plot of the hardness of WC-Co alloys against temperature. The cobalt content of the
alloys and the grain sizes are indicated.
linear at any Vwc, in agreement with the results in Fig. 14. Figure 14 shows that the
above relationship is linear also at all temperatures, which implies that the
relationships between Hwc and d and Hco and X (see Eqs (1) and (2)) are linear
at all temperatures.
Figures 15 and 16 show that Ho in Eq. (7) decreases rapidly with increasing
temperature up to about 600"C, while K,, decreases substantially with temperature
only above 600°C. Since Ho is a combination of the intrinsic hardness of WC and Co
(see Eq. (8)) and Ky a combination of the Hall-Petch coefficients of Eqs (1) and (2)
(see Eq. (9)), Figs 15 and 16 suggest that the softening of WC-Co with increasing
temperature is due to the intrinsic softening of the component phases up to about
600°C but is controlled by the ease of slip transfer across grain boundaries and
interfaces above that temperature.
Milman and co-worker? results [21] have established that the hardness of finer
grained material decreases with increasing temperature at a lower rate than the
hardness of coarser grained material and thus the advantage of using finer grained
material increases with increasing temperature.
8.3 Factors A.fecting the Hardness of' WC-Co Hardmetal 951
24
22
20
18
0
c2 16
g 14
C
$ 12
r
10
8
6
4
0 100 200 300 400 500 600 700 800 900
temperature (C)
20
10 wt% c o
18
-I? 16
14
g 12
C
$
c
10
-
8
6
d = 0.5 urn +- d = 1.2 urn +++ d = 2.3 urn
4
I 100 200 300 400 500 600 700 800 C
temperature ( C)
18
15 wt% Co I
16
-I? 14
12
9.
g 10
C
$ 8
c
6
2 0
temperature ( C)
Figure 13. Plot of the hardness of WC-Co alloys against temperature. The grain size of the alloys
and the cobalt contents are indicated,
8 The Hardness of Tungsten Carbide-Cobalt Hardmetal
-
25 .........
20 -
m
5 '5-
In)
u
acl
2 10-
m
I
5~
15 20 25 30 35 40 45 50
d (rnm)i'nr
ll.l^l"lllll ................................ " "
I
_I___
__
*20 ' C
20 -
0200 ' C
A300 ' C
A400 ' C
0500 ' C
0 600 ' C
m700 ' C
0800 'C
XJOO ' C
~
I
i
j
20 ~
15 wt% co
A300 "C
1 A400 "C
i 0 6 O O "C
15 20 25 30 35 40 45 50
d (mrn)4'n)
Figure 14. Plot of the hardness of WC-Co alloys against d-''2, where d is the carbide mean grain
size, at temperatures in the range 20-900°C. The regression coefficients are all higher than 0.95 and
mostly higher than 0.99.
8.3 Factors Affecting the Hardness of WC-Co Hardmetal 959
700
600
500
.
h
E 400
E
0
Y
Y
0 300
I
200
100
04 I
o loo 200 300 400 500 600 700 aoo 900
temperature ("C)
I -I- WC-10 vol% Co -+ WC-16 vol% Co -+- WC-24 vol% Co
Figure 15. Plot of the parameter Ho of Eq. (8) against temperature for WC-Co alloys of different
cobalt content. The cobalt contents are indicated in the key. Reproduced by permission of the Znt. J .
of Refract. Met. & Hard Mater.
40 I
35
30
E
A
E
25
m
& 20
y"
15
10
5 I
100 200 300 400 500 600 700 aoo goo
temperature ("C)
I -D-- WC-10 vol% Co t-WC-16 vol% Co - WC-24 vol% Co I

Figure 16. Plot of the parameter K y of Eq. (8) against temperature for WC-Co alloys of different
cobalt contents. The cobalt contents are indicated in the key. Reproduced by permission of the
Int. J . of Refract. Met. & Hard Mater.
960 8 The Hardness of Tungsten cur bide^ Cobalt Hurdmetul
L"
-
6 wt% CO
20
19
10 Wt% cc 18
3
a 17
16
w
c E 15
E
2 14
! 3 12
1
d = 0.5 pm d = 1.2pm
14.1 I I 11
-200 -150 -100 -50 0 50 -200 -150 -100 -50 0 50
Temperature ("C) Temperature ("C)
- 16
m
& 15
u)
$ 14
E
2 13
12
11
104 I I
-200 -150 -100 -50 0 50
Temperature ("C)
Figure 17. Plots of the hardness of WC-Co alloys of different cobalt contents against low tempera-
tures. The grain size of the alloys is indicated in the keys.
Milman and co-workers tested WC-Co alloys of grain size ranging from 0.5 to
2.3pm and cobalt content from 6 to 15 weight-% also at temperatures ranging
from -196°C to 20°C. The results are shown in Figs 17 and 18 [21].
8.4 Relationships between Hardness and Other

Hardmetal Properties
In most hardmetal applications it would be desirable to use the hardest possible
grade. However, properties often have reciprocal relationships, i.e. an improvement
in one leads to a deterioration in another. Therefore, in order to select the most
appropriate grades for a specific application, it is desirable to know quantitatively
the relationships between the various properties. So far, extensive work has been
8.4 Relationships betn3een Hurdness and Other Hardmetal Properties 96 1
124
-200 -150 -100 -50
I
0
I
50
temperature ( C)
24
22
- 20
cr
n
18
u)
16
;
I?
c
14
12
10
-200 -150 -100 -50 0 ! I
temperature ( C)
18
17
16
:
cr
15
3 14
c
-0
13
c
12
11
10
-200 -150 -100 -50 0 50
temperature ( C)
Figure 18. Plots of the hardness of WC-Co alloys of different grain size against temperature. The
grain sizes and the cobalt content are indicated.
-t-22pm --yt 6pm

~-
1000 I I
8 10 12 14 16 18 20 22
Fracture Toughness (MPafi)
Figure 19. Plot of the hardness of WC-Co alloys of constant grain size against toughness. The grain
sizes of the alloys are 2.2 pm and 6 pm. Adapted from reference [22].
done to establish the relationships that exist between hardness and toughness and
hardness and abrasive wear resistance.
8.4.1 Relationship between Hardness and Toughness

It is well established that a reciprocal relationship exists between the hardness and
the toughness of WC-Co. However, it has proved difficult to explain this relation-
ship in terms of the material’s microstructural parameters until it was observed that
the relationship between the two properties is linear when the grain size is kept
constant [22]. Figure 19 shows the particular case of alloys having grain size 2.2
and 6pm, but the linearity of the relationship has been confirmed for a wide
range of grain sizes [23].
Therefore it appears that, in general:
Hwc-co = 4 W I C + b ( 4 (1 1)
where KIc = toughness of WC-Co, and a(d) and b ( d ) = functions of the WC grain
size.
8.4.2 Relationship between Hardness and Abrasive Wear

Resistance
Figure 20 shows the results of a comprehensive investigation into the relationship
between the hardness and the abrasive wear resistance of WC-Co, with the wear
resistance being measured according to the ASTM Standard B611-85 for hardmetal,
Acknowledgments 963
1000
........................
.-0IE
E
n l
U- ...................................
+F
............................
.................................. -UF "M *c
0.1
400 600 800 1000 1200 1400 1600 1800 2000
Hardness fKglrnrn2J
Figure 20. Plot of hardness against abrasive wear resistance for a wide range of WC-Co alloys.
U F = 0.7 pm; F = 1.0 pm; M = 3.0 pm; C = 5.0 pm. Reproduced by permission of Int. J . Refract.
M e t . & Hard Mater.
using alumina particles as abradors. The alloys tested ranged in cobalt content from
3 to 50 weight-% and in grain size from 0.6 to 5.1 pm [17]. Figure 20 shows that up to
a hardness of approximately 1000 HV there was a one-to-one correspondence
between hardness and abrasion resistance, which suggests that the main wear
mechanism was plastic deformation. Above 1000 HV the wear resistance of coarser
grained alloys was higher than that of finer grained alloys of equal hardness, up to
approximately 1600 HV. This suggests a wear process controlled by microfracture,
since in that range of hardness coarser alloys are tougher than finer alloys. There-
fore, the abrasive wear resistance of hardmetal increases with increasing hardness
but, in the hardness range which is of interest to most applications, there is not a
one-to-one correspondence between hardness and abrasive wear resistance since
resistance to wear is determined also by other properties, such as toughness.
8.5 Conclusions
The knowledge of the factors affecting the hardness of WC-Co alloys has increased
considerably in recent years, which has allowed a more precise selection of grades.
However, much work remains to be done to quantify the relationships that exist
between hardness and other properties which affect the performance of the material.
Acknowledgments
The author wishes to acknowledge the contribution of several colleagues to the research
work carried out over the years and reported here. Professor Yu. V. Milman,
Mr D. G. F. O’Quigley and Mr I. T. Northrop must be mentioned in particular.

The financial support of the Foundation for Research Development, THRIP
and the Boart Longyear Research Centre, over the years, is also gratefully
acknowledged.
References
1. ASTM Standard B 294-92.
2. J. J. Oakes, Teledyne Advanced Materials, Private Communication.
3. I S 0 Standard 3878.
4. S. Luyckx, F. R. N. Nabarro, Siu Wah Wai, and M. N. James, Acta Metall. Muter. 1992, 40,
1623.
5. T. Takahashi and E. J. Freise, Philos. Mug. 1965, 12, 1.
6. S. Luyckx and J. Katzourakis, Muter. Sci. Technol. 1991, 7,472.
7. S. Luyckx and L. C. Demanet, Unpublished Results.
8. H. C. Lee and J. Gurland, Muter. Sci. Eng. 1978, 33, 125.
9. L. S. Sigl and H. E. Exner, Muter. Sci. Eng. 1989, A108, 121.
10. J. Freytag, PhD Thesis, University of Stuttgart, 1977.
11. W. Batteridge, Cobalt and its Alloys, Chichester: Ellis Horwood, 1982.
12. B. Roebuck and E. A. Almond, Muter. Sci. Eng. 1984, 66, 179.
13. H. E. Exner and J. Gurland, Powder Metall. 1970, 13, 13.
14. D. G. F. O’Quigley, MSc Thesis, University of the Witwatersrand, Johannesburg, South Africa,
1995.
15. L. Makhele, University of the Witwatersrand, Unpublished results.
16. D. F. Carroll, 14th International Plansee Seminar, Plansee AG, Reutte, Austria, 1997, 2, 168.
17. D. G. F. O’Quigley, S. Luyckx, and M. N. James, Int. J. Refract. Met. Hard Muter. 1997,15,73.
18. H. E. Exner, Int. Muter. Rev. 1979, 4, 149.
19. T. Sadahiro, J. Jpn SOC.Powd. Powd. Met. 1979, 26, 33.
20. M. Lee, Metall. Trans. A , 1983, 14A, 1625.
21. Yu. V. Milman, S. Luyckx, and I. T. Northrop, Int. J . Refract. Met. Hard Muter. in press.
22. D. G. F. O’Quigley, S. Luyckx, and M. N. James, Muter. Sci. Eng. 1996, A209, 228.
23. S. Luyckx, V. Richter, D. G. F. O’Quigley, and L. Makhele, Proc. Int. Con$ Deformation and
Fracture in Structural Materials, Institute of Materials Research, Slovak Academy of Science,
Kosice, Slovakia, 1996, 2, 109.
9 Data Collection of Properties of Hard

Materials*
G. Berg, C. Friedrich, E. Broszeit, and C. Berger
9.1 Introduction
Hard materials are used as thin hard coatings of some microns thickness for
wear protection of tools and machine parts because of their high abrasive wear
resistance. For the selection of the coating material the physical, mechanical, and
technological properites of these coatings, required by the application, are decisive.
The following data collection presents fundamental and available material proper-
ties for approximately 130 hard materials as a result of a literature search on
carbides, nitrides, borides, silicides, and oxides.
9.2 Profile of Properties

Each technical application demands a special profile of properties concerning the
material. In tribological applications the material selection of the system com-
ponents plays a dominant role. The existing profile of material properties must, if
possible, go beyond the demanded characteristics. For a functional selection of
materials basic data must be available.
A ‘profile’ does not exist if only one property is known, it is built up from all
relevant material characteristics, which could be called ‘mechanical-technological’,
‘mechanical-tribological’ or ‘physical-chemical’ for mechanical components. Only
the combination and comparison of numerous characteristic data of one material
make it possible to sketch a profile of properties (columns in the following table).
First if there is a full characteristic profile for this material available, it makes
sense to compare it with other materials in complex applications to recognize all
positives and negatives (rows of the table).
Investigations of thin films show that the material properties are not constant.
Responsible for this are different deposition technologies and procedures which
result in different structure properties of the coatings, having various textures or
various densities of defects. For this reason the evaluation of material properites
by different references is of special interest (several data for one characterisic
value).
* This contribution is the revised and translated version of the publication Datensammlung zu
Hartstoffeigenschaften, published in Materialwissenschaft und Werkstofftechnik 1997, 28, 57-76,
0VCH Verlagsgesellschaft mbH, D-69451 Weinheim
966 9 Data Collection of Properties of Hard Materials
The presented data are mainly of a physical-mechanical character. For applica-

tions technological/tribological data (i.e. load bearing capacity of hard coatings in
different tribological systems under corrosive attack or under high temperature
applications) are of special interest. For these uses, in practice physical data may
only give first qualitative hints for decisions concerning a successful selection.
9.3 Organization and Contents of the Data Collection

The following data collection provides an overview of basic properties of hard
materials found in the literature. The list is structured according to the chemical
composition of the materials. To define ‘hard materials’ a minimum hardness of
HV = 1000 is demanded (definition of Schedler [57]).
Values of bulk materials and coatings are listed together. A differentiation would
be useful but is very difficult to achieve because basic information on design and
structure of the layers is not available. For this reason multilayer and multiphase
coatings like (Ti, A1)N or Ti(C, N) have not been taken into account.
Physical values like crystal structure and the Young modulus are cited, which may
play a dominant role in mechanical/tribological loadings. Additionally values for
microhardness and oxidation resistance are given. More than others these two
values are a function of the test system, the test parameters, and the evaluation
method. At this point very strong deviations appear (e.g. hardness of Sic:
HV = 1400 - 6000). In many cases the boundary conditions of the investigations
are not documented (e.g. no load specification for hardness testing). For this
reason values for hardness and oxidation resistance have a qualitative character
more than others.
This points out the necessity of further standardization efforts, to give the results
of the system tests a transferrable and absolute character. In the field of thin films,
the standardization is in progress, including special test procedures, but they are not
yet established and for this reason not used and documented. Giving a guarantee of
properties is only possible if firm regulations exist. For series investigations the
realization of this demand is absolutely necessary and already realized for other
test methods (e.g. macro hardness testing according to different DIN or I S 0
standards).
To reduce the size of the table, the given data are as far as possible compressed.
One line in the data collection includes data from different authors. Mostly in one
reference only single properties of the hard materials are given, so that only the
collection of these properties leads to a profile of characteristic properties for a
single material. The references are quoted in brackets.
The influence of the deposition procedure and exactness o f the test methods on
the profile of properties of hard coatings leads to very different values presented
by the different authors. Especially the properties depending on crystal structure
like thermal conductivity, electrical resistivity or the Young’s modulus (see e.g.
electrical conductivity for NbC or the Young modulus for TIN) differ very much.
Acknowledgement 967
Good correspondence is found for crystal structure, lattice parameters, density and
melting point for one material.
The data collection (Table 1) does not claim to present all existing investigations
on hard materials. The aim of the presented table is to give a representative overview
on a special material. The number of quotaions gives a first impression on the
number of published values (see e.g. TIN which is well known as layer for tribolo-
gical applications). Looking on the five material groups, carbides and nitrides are
dominant.
The database for the data collection comprises actual textbooks, theses, and
publications in technical journals and from conferences in recent years. All in all
approximately 3000 references were searched through. One result is that in new
publications single phase materials are seldom studied. The quotation of older
literature in these publications leads to the repetition of single values by different
authors.
Acknowledgement
Many thanks to Mrs. R. Kurth for listing all the data on PC and thanks to Mr. U.
Petzel as well as Mr. T. Eid for their help in increasing the data base.
Table 1. Data collection of properties of hard materials
Legend
materials: carbides, nitrides, borides, silicides, oxides (mo) monocrystalline (x) density as determines by X-rays, determined by
m. 0, y. S, 6, TI, T2: phases (PO) polycrystalline lattice parameters and atomic mass
HV: Vickers hardness, value after HV gives load in kp according to DIN (i) incongruent melting point (formation of liquified material and a second solid phase) (d) decomposition
HK: Knoop hardness, value after H K gives load in k p according to DIN (per) peritectic melting point (several solid phases and liquified material in equilibrium) (p) “heat proofness”
HU Universal hardness. value after HU gives load in N according to DIN (I) “thermal stability”
Crystal stmcbre:
ortho: orthorhombic cub: cubic fcc: face-centered cubic mono: monoclinic The crystal systems of single materials may he
hex: hexagonal cub-B1: cubic, NaCI-type bcc body-centered cubic tri: triclinic explained by different crystal stuctures
hcp: hexagonal close-packed rhom: rhombohedral/trigonal let’ tetragonal e.g. hexlrhom for AIzO,)
Carbides
No. Symbol Crystal Lattice Density Melting Linear thermal Thermal Electrical Enthalpiy Young Micro Oxidation
Structure parameters point expansion, 3 conductivity, X resistivity modulus hardness resistance
inm) (c: cm-’) i’C) K-’) (W m-’ K - ’ ) ncm) (kl mol-’) (LO’ N mm-’) (10 Nmm?) (IOO’C)
I B4C rhom [XI 0 5631/1.2144 [52] 2.52 [I] 2450 [1.48,57.7716 [I] 27.63 [I] lo6 [I41 72 (51 4.5 [I1 3700 [I. 331 11-14 [ZS]
7 rhom [31] 0.5804,1.2079 [55] 2.5 171 2470 [5] 4.5 [8] 29.3 [XI 5 x 10’ [311 40 [43] 4.48 [lo] 3000 I71 7-8(P) [35]
3 rhom [43] 0.5599/1.20?4 [SS] 2 52 1141 2350 [7] 7.32,’s 33 [30] 10.42 1351 4 x 10’ [38] 71 6 1461 4.4 I141 490&5000 HV IS] ’ L431
4 rhom [SZ] 0 56003/1.2086 [I121 2.52 [251 2350 [XI 4.5 [31] 29 [38] 101-10’ [48] 57.7 [48] 2.96 [XI 2940HK0.05 [I41 x [701
5 hex [I121 2.52 [?I] 2450 [I41 45 [35] 28 [43] 4 x 10’ 1571 71.6 [73] 4.41 [31] 3700 HV 0.05 1251
6 2.52 [35] 2450 [25] [381 35 [48] lo6 [681 45 [38] 300M000HV [3I]
7 2.52 [38] 2430 [30] [431 28 [57] 5 x 10’ I771 4.6 [43] 2800 HK 0 I [33l
8 2.52 1391 2450 [31] 4.5 [48] 27 2 [I061 5 x 10’ [Y71 4.5 [48] 3700-4700 HV [35]
9 2.52 1431 2450 [38] [571 4.48 [S7] 4950 HV 1391
10 2.51 [48] 2447 [39] ~701 42 [70] 2800 HK [44l
11 2.52 [52] 2447 [43] 4.5 [77l 441 [77] 3700 HV [13.44. 57,681
12 2.52 [57] 2500 [55] 4.5 [97] 4.41 [97 35w-4500 HV 0 I [48]
13 2.51 [70] 2450 [88] 4.5 [I061 2.9 [I061 2940 HK 0. I I581
14 2.52 [771 2350 [I061 4.4-4.7 [lox]
15 2.5 Wl 4000-5500
30W3500 HV 0 I [58]
1701
16 2.52 [97l 300&7000 [781
17 2.5 U061 30w-4000 HV [77]
18 2.45 11081 3500 HV 0.2 [881
19 3200 HV [I081
20 Cr,C: ortho [6] I 146/0.552/0.2821 161 6.68 [6l 1900 (per) [Sl 10 3 [XI 18 8 (81 75 [7] 88.8 PI 4 191 1500-2000 I71 12 [71
21 ortho [8] I 147/0.5545/0.2830 [lo] 6.68 [7] 1850 [7] 117 [9] 19 [9] 75 [9] 88 [6] 3.86 [lo] 1300 HV [8,871 11-14 [25]
22 ortho [lo] 1.147/0.5545:0.2830 [Ill 6.68 191 I890 [XI 10.3 [lo] 75 [40] 23 [Ill 3.7 I401 2100 [91 11-12 [53]
23 ortho [Ill 1.147:0.553j0.282 [25] 6 68 [lo] 1810 [9. 10,401 10.3 [40] 75 [75] 109.7 [46] 3.73 [57] 1300 16, 10.40.761
24 ortho [25] I 147/0.5545/0.2839 1281 6.68 [XI 1810 176.77.871 10.3 [57] 75 [77] 85.4 [I021 3.284 1741 100&1400 [I 11
25 ortho 1281 I. 148310.5531/0.2827 [52] 6.68 [391 I895 [I31 10.3 [76] 75 [87 94.2 [73] 37 [76] 2280 HV [I31
26 ortho [46] I. 1488/0.552710.2829 [74] 6 66 [53] 1895 [ZS] 11.7 [77] 75 [I061 4 1771 1300 HV 0.05 ~251
27 ortho [52] 1.14883,O 5527310 28286 [I121 67 [57, 1071 1890 [39,46.90] 10.3 [87] 3.7 [87] 1350 HV 139.901
28 ortho [57] 6.68 1741 1895(d) I531 10.3 [I061 3.73 [I061 2280 HV 1531
29 ortho [74] 6.68 1771 18Y5(d) I571 2280 [571
30 ortho [76] 6 68 [YO. 1061 1850 [I061 1350 1591
31 ortho [I121 2150HV [771
32 Cr7Ci hex [I] 14/ll.45 [I1 69 [I1 1780Id) [I] 106 [I] 190? [I] 3.5 [53] 2200 [I1 7 [53]
33 hex [6] 1.398 0.4523 181 6.85 [61 1780 [46] 178 161 3.726 [74] I600 [Ill
34 hex [lo] I401 0.4525 [lo] 69 1521 1600-1790 [I021 228 2 1461 1900-2200 1531
35 her [Ill I401~04?32 I62 iloi] 1200-1600 H V [SBj
36 hex [25] I 398,O 453 181 3 [731 2200-2400 H V 0 05 [61]
77 hex [46.76] I4006 0.4532 1910HV003 [641
38 rhom [52] 0.7015,1.?153 0.4532 1200-1600 H V 1681
39 ortho [74] 070149'1 2153 045320 I500 H V 10.025 [72]
40 ortho Ill!] 1200 H K 0 01 [1271
41 Cr2,C6 cub [I] I 06 1520 [I] 101 [I] 68 7 [I/ 1650 111
42 fcc [6] I0638 68 6 [8] 1000 [ I 11
43 fcc [lo] I0655 396 5 [73] IROOHV (581
44 fcc [Ill I 06&1 066 l8OOHV [681
45 fcc (251 I065
46 fcc 1521 1.0659
47 fcc 174. 7x1 I066
48 cub (1121 1.06599 [Hi]
49 HfC fcc [8] 0.464 [lo] 122 [7] 3830 151 673 [XI 63 [8] 37 [7] 209 7 [II] 3 55 [9] 2200 H V 2
50 fcc IIO] 0 464 [Ill 1267 (91 3890 [7,8.9.13] 66 [lo] 6 [PI 39 [9] 205 [45] 3 174.21 [lo] 2600
51 cub-Bl [I I ] 0.464 [??I I 2 6 7 ( x ) [lo] 3928 [lo] 678 1301 29.31 [32] 37 [32] 219 [4h] 4.71 [32] 2700 H V
52 cub-BI 1251 0.464 [321 12 2 I251 3890 j25] 66 [32] I3 [38] 40 [38] 2303 [73] 4 [38] 2800
53 cub-Bl [30] 0 4646 [52] I?7 1251 3927 [30] 6.4 1381 37 [57] 3 5 4 6 0 8 [74] 2700
54 cub-BI [52] 0.446 [I091 12 3 [38] 3950 [32] 6.9 [45] 37 [lO6] 4 64 [76] 2276 HK 0.1
55 fcc 1571 0 463765 [I121 -12 [39] 3890 1381 6 (621 3 52 [I021 2200
56 fcc 1741 12 7 [52] 3890 1391 6.6 [76] 1932-2900 H V 0.05
57 cub-Bl 1761 12.2 1741 3387 1451 67 1981 1000 H V
58 fcc [98] I2 1901 3830 [46] 659 IlOh] 2913 H V
59 cub-BI [I091 I22 [I061 3890 1571 3200
60 cub 1112) 122 [I081 3890 i62j 180&2500 H V
61 12.7 [I071 3928 [76] 2600
62 3890 [YO] 180@2500
63 3900 198. 1061 ?913HV
64 2750 H V
65 MoC-., hex [2] 0 29'0 284 2700 [I] 49 [6] 11.7 [5]
66 hex 161 0 2901/02785 2677 IS] 126 [I!]
61 hex [in] o 2896 o 2x09 2692 [6] 10 [73]
68 hex 1521 0.2896\0 2809 2600 [lo]
69 hex [I121 0,2932 I 0 9 7
70 MoC-0 fcc [Ill 0427
71 cub-BI 1251 0.428
72 Mo2C-n ortho [ l o ] 0 7244jO 600410 5199 2670 (71 49:./x2 [lo] 228 [lo] 1600 171 5-8 [2?]
73 ortho [ I l l 04733l06034'05206 2400 [I31 8.5'4 5 5 7 [28] 1500 [101
74 ortho (28. 1121 04732.0 6037 05204 2410 [39] 1479-1800 [I11
75 1950 H V [I31
Table I. Continued
Structure parameters point expansion, a conductivity, X resistivity modulus hardness resistance
(nm) (gcm-7 (“c) K-’) (W m-‘ K - ’ ) (lo-‘ Qcm) (kJ mol-’) (10’ N mm-2) (10 N mm-’) (IOWC)
76 MolC-$ hcp [6] 0.2994/0.4722 [6] 89 [6] 2687 (61 5.6/7 5 1281 97.5 [6] 46 [Ill 533 (571 15OOHVU1410.05 1251
77 hex [10,112] 0.3005/0 4755 [lo] 9 18 1251 2690 1251 7.8 1571 97 (251 46 (461 5.4 1771 1499 HV p91
78 hex [Ill 0.3004/0 4722 I251 9.2 1571 2430(per) [46] 7.8-9 3 1771 133 is71 46.1 (731 5.4 1971 iaoo HK (441
79 hcp 1141 0 300210.4724 1521 9.18 1771 2400Id) 1571 7.8-9 3 1971 57 I771 533 IlO6l ISOOHV
80 hcp 1251 0.414 8.9 1901 2430 1731 7.8 [I061 57 [97] 1950
81 hex 128,461 0 3012 0.47352 9 18 1971 2517 1771 71 [I061 I500 HV
82 hen (521 8.9 [I061 2410 1901 1500
83 hcp 1571 8.Y (107] 2517 1971 1499 HV
84 fcc 1251 8.9 [I081 2697 (1061 I660 HV
85 NbC fcc [I] 045 7.78 [I] 3490 [I] 6.65 111 14.24 [I] 3.C74 [6] 139 8 IS] 3.4 [I] 2400
86 cub-BI 161 0.4171 7 51-7.82 161 3600 [S] 6.2 161 11.25 [6] 34 [?I 140.7 [6] 2.763.45 161 195k2700
87 fcc 1x1 0447 7.82 171 3477-3900 161 6.84 181 14.2 181 74 1251 140.7 [Ill 3.38 [lo] 2000-2400
88 cub-BI [lo] 0.4471 7.79(x) [lo] 3500 I71 66 110) 14 24 [lo] 51 1321 138.2 1461 3 45 1321 2400 HV
89 cub-81 (111 0.4461 7 82 I251 3480 181 52 1301 14 24 1251 35 1571 140.7 1731 3 38 1571 2055-2400
90 cub-Bl 1251 0.447 7.56 1391 3600 [lo] 6.6 I321 14 24 I321 19 166) 3 193-5 10 1741 2400 HV 0.05
91 cub-BI 1301 0.4449 7.8 1571 3500 [I31 67 (571 I4 (1061 19 (771 58 [7?] 1800
92 cub-BI 1461 0.447 76 1741 3500 125) 7.2 1771 19 1971 5.8 1971 1961 HV
93 cub-BI 1521 044698 7.78 1771 3500 1301 7.2 197) 35 11061 3.38 [IOh] 1265
94 fcc 1571 7 56 1901 3600 1321 6.65 (1061 2000
95 ccc 114) 7.6 11061 3480 1391 2004 HV
96 cub [I121 7.9 [I071 3500 (571 2400 HV 0.025
97 7.9 [lox] 3480 j73,901 1800 HV
98 3613 (771 1961 HV
99 3613 197) 1800 HV
100 3480 1106l
101 Nb,C-a ortho [lo. 641 1.092/0497/0.308 [lo] 785(x) 1521 3090 (per) [73l 4.1UIl.OU7 1281 1863 1461
102 ortho [I121 1.092010.4974~0.309 11121 191.5 1731
103 Nh,C-P hex 161 0.540j04974 UOI 7 85 (XI 161 3100 (61 1924-2322 161
I04 hex 110. 1121 0.3116,’04958 11121
105 hex 1281
106 Nb?C-y hcp [6] 03128104974 I61 -3500 (per) 161 6.67.0,8 7 (61 167-195 161
I07 hcp [lo] 0.3127i0.4972 1101 6.816.9 1281
108 hcp 1281 0.311910.4Y59 1521
109 hcp 1521
110 SIC Ir.hcx [22] B.0.4360 [61 3.2 111 2200 [I] 5 68 [I] 15.49 111 10’ [9] a: 71 6 161 4.8 [I] 3500 [I. 331 1617 1251
Ill pfcc [22] n:0.3-7.3,1-1.5 [22] 3.17 16) 27W [7] 53 191 63-155 191 10‘ 1141 p73.3 161 3.94 I 161 1400HV 141 I3-14iP) 1351
II2 a:hex 1311 0:0.43 (221 3.2 171 2760fd) 191 4.8 (141 42 1141 10’-10’? 1221 80 1221 4.8 [9] 2169-2428 161 1: 16 1431
113 B:cub 1311 0 0 4358 1521 3.22 191 2300(d) 1141 a:4.5 [22] 60 (221 10’ 1311 1 70.0 (431 4.7 1141 2600 17. 451 16 1701
II 4 ?:hex [43] X:O 3n73!1 508 (1121 3.22 1141 2700 2. 440 [22] 20W3000
115 flub (431 3.2 [22] -2200 fl. 4.0 [22] 3500 HV
116 xhex [47] 32 [ZSj -2500 4.8 [31] 2580 H K 0 05
II 7 Fcub [47j 3.21 [31] 2760 48 [38] 3nowoon
118 ?.hex [52] 3.2 [35] 2300 :I 3 0 4 . 0 [43] 350M000
119 Pub [52] 3 21 [38] -2227 48 [45] 3500 HV 0.05
I20 ?.hex [76] 32 [39] 2827 (per) 48 [57] 2600 HV
I21 Fcub [76] 3.21 [43] 2760 4 [671 2600 HV
122 hex 198) 3 22 I451 7200 4 1701 2500 HK 0 I
123 xhex [I121 32 [57] 2986 (d) 48 [77] 3nw3500 HV
I24 3.15 [67] 2760 (dl 48 [97] 20W6000 HV 0 05
I25 3.21 [70] 21x0 48 [98] 2585 HK
I26 3.22 [77] 2760 3544 [I081 3000 HV 0.2
127 32 [XR] 2760 200&60no
I28 3 22 (971 2700 2580 HK 0.1
I2Y 3.22 [98] 2500 -4000 H V
I30 25-34 [lox] 3000
131 32 [I061 2500 H V 30
132 ~4on-3nonHV 0. I
133 2400-3500 H V
I34 TrC fcc [I] 044 [I1 5 “I 3780 29 1790

I35 cub-BI [6] 0 4454 [h] 14 65 [7] 3877 2.91 1490
136 fcc [XI 0 4456 [lo] 14.5 (XI 1101 38x0 3.65 I800
137 cub-BI [lo] 0 4455 [Ill 14.5 [I41 3983 5.5 inoo H V
138 cub-BI [Ill 0 4455 (251 14.48 [25] 3980 2.91 2500
I39 cub-BI [25] 0 4456 1321 14.5 [38] 3919 29 1547-1952
I40 cub-Bl (301 0.4456 [52] 14 3 (391 4000 2 85 1790 HV
141 cub-BI [46] 0.4456 [74] 14 5 [57] 3880 2 85 I550 HK 0 05
142 cub-BI [52] 0.44 [98] 13 9 (741 dono 3 04 5 34 1800 H V o 05
143 fcc [57] 0.445 [I091 1448 (771 3985 2 85 1600H V
144 fuc [74] 0.44547 [I121 145 [MI 3780 56 [77] loon H V 134j
I45 kc [98] 14.3 [90] 3880 291 [80] I599 HV
146 cub-BI [I091 I4 5 [98] 3985 56 (971 I790
I47 cub [I121 139 [I061 2985 56 1981 I599
148 13.9 [I081 3983 2 85 1871 179nHVO2
149 14.5 [in71 3880 2.85 [I061 I550 HV
I50 2500 HV
151 Ta2C-P hex [ I , 1121 0 31110.495 PI [I] 3400 (d) [I] 6 2,4 8 [28] 40 [6] 71.2 Ill I700
152 hex [lo] o.3103/0 49378 [6] I 5 05 (XI [6] 3500 (per) [73] 188.4 I61 1000- I720
153 hex [Ill 0.3102/0 494 104.3 1000 HK
154 hcp [I41 0.3106:0.4945 203.1
I55 hex (251 0.3091:n.493 203.1
I56 hex [52] 0.31037/0 49394
157 0.31042:O 4941
I58 Ta,C-Z hex [6] 0.3 1046’0.49444 u121
159 TIC fcc [I] n 433 [I] 4.93 [I] 3067 I41 7.42 [I] 2093 [I] 52 [41 I838 4 141 3200 HV [1.13.35.44] 12 [7]
I60 cub-BI [3] 0.429~~33 (61 4.93 141 3150 [?I 7.4 [31 17-235 [R] 68 171 1796 3.22 [6] 3200 HV [58.68.95] II 14 [25]
[9Ll 0092 [LLI 8p9Z 1161 zoz
SZOO AH 0095 ILSl o w 1061 IOZ
AH motmooz I6El 0185 ILL1 ooz
B H 099Z lor1 triz [PLl 66 I
AH b60Z [SZl OW [LSI 861
SO0 AH 008Z (9011 ZZP L90l1 09 [Ell OE87 [6E1 [6011 SIVO 161
AH 0562 1161 0 [ELI 6'001 IL6l 65 1011 8P9Z [SZl [PLI 91VO 961
000E-OSPZ ILL1 Z'P I9Pl 6'001 [LLI 65 I9011 Z'L [XI OERZ [Oll [zsl Z81P'O 561
AH 0085 b L I ZEW
' YZ [Ill 9'101 65 [LLI r'L [LI OLLZ [Ll [SZI 91P'O P6 I
0065 [LSI ZZ'P I91 6101 09 [LSI Z'L L91 OSLZ 1011 991P'O E61
WH O9LZQPEZ loll PE'P [sl 1'501 951 [Sl ZP [or] SSY [Sl OLLZ [91 riwo 561
0562 "1 87 [I1 Z'Lll OS1-09 [91 61'P [a1 5'9 "1 om "1 zvo 3A 161
SO0 AH 00LE-0085 06 I
z o AH oozr 681
AH 8865 881
5500 AH 007s E6b 181
SO0 AH OOOS 6P 9R I
SOOAHSESI 80P 581
SO0 AH OOZZ oLir X P P81
DOZS-00PE 1561 090s E6'P EX1
SO0 AH OOZS-OWE [9011 PL L [861 L~OE E6'P 281
s o 0 AH ommoor [861 9'8-8 [La] 190s 6P 181
1.0AH wzr L90E 6'P 081
AH ooomrs osir E6 P 61 1
006r LPlE E6'P 811
WH OLPZ 19011 89 oLor Ebb LLl
AH 000f (Z61 55 wt f6'P 91 I
OS AH mzt [861 ZS 091s r6b SLl
1'0 B H OLPZ [L61 ZS osir Z6f PLI
008Z [LXI 5s LPit 6P ELI
AH 0085 ZS WIE t o ZLI
SO0 AH OOZE 89 wit r6P ILl
omz zs [9011 IZ 190s 6P OLI
SO 0 AH OOLfHM8Z 1191 55 1191 ft 190s P6'P 691
SO 0 WH 0085 [LSl 89 ILSl oz 190s E6b 891
E'O AH 008Z-0091 [OPI ZS [Ed 62 OblE E6P 191
00Sf-0085 [8E1 OL ILEl IZ oLor E6b 99 I
AH o o x r - 0 ~ 6 7 [zrl SP [srl lwpz om 16P 891
0067 ZS HOE (X) 16P 09 I
OOZE 5'89 L90E 16') $91
AH 0062 5s 0x1s 80 P Z91
5 AH 0085 [91 L6tZ ZS L6l zr-iz OLlf E6'P 191
[iYI OtLi YIP0
[ESI (PlOOLi YSLP Oi8PhhiO
[Sil OELi 9liLPO'hOXlS 0
ILl LSXi I LP 0;Xhi 0
191 0XRi*iLi SSLp'O~iOOtO
(51 ooxi i L P 01662 0
X ~ L P0'10oi'n
5 0 AH OROZ
AHOOli
AH O S f i
Si0 0 AH OOCZ [XOll EY 51
OOSi "011 8 51
AH 0Wi [YO11 SI Ll
SO 0 AH 000Z-00f I [YO11 96 Y [861 ZL'SI
000Z-00i1 [X61 iL [Lhl ZL'SI
AH 0011 [Lhl ii [Oh] L'SI
1lZl [Sh] iL [xi L SI
AH 000; [LR] S6 Y [LLl iL'SI
AH OOPZ lox1 ii'L [PLI EY S I
YH onRi LLLl iL [LYI L'SI
AH ORLl 1911 26 Y [LSI L'SI
AH OROZ [Pi1 LOZ Y-Lil Y [is1 L'SI
AH 00Oi-oOY1 [hY] i'L- [SPI LSl
"El OOPi [LYI 96 Y [hfl L'SI
[rrl IOYH0081 [LS! 96'9 iRil 1E l
[ril A H nooi 00x1 [SP] 56 Y lit1 LSI
a o AH o o ~ i IOPI 56 Y [SCI LLSl
so o YH O S E ~ [Xi1 S'L ktl i'jl mi o Z Y O ~ Zn
AHO81i [iil ir~ [Ell OOYi [Sil L'SI 16i n
SXOZ [ti1 LO L [OII YLLi [PI1 L 51 8t8i 0 Y06i 0
161 OOti [El Li L [61 YLLZ [Oll ixlRS1 iixz n ~ 0 6 io
181 AH OWZ OOOi [PI1 L 1x1 OtLi 161 iLSl lit1 LlRZ 0 Y06i 0
Its1 x-c [YL'SP'OPI 110Ii [nil 69 Y [Ll LYXZ lLl L'SI [Sil ltXi 0 OOhi 0
[jtl idly; l i t '(11 'il 00 IZ I61 iL 191 SXLZ [YI L $1 6iRi 0 YO62 0
lSi1 x-S It1 AH O O U [PI L [PI YLLZ bl i L 51 L t R i 0 Y06i 0
[Ll 8 ILS 'I1 nxoi 111 iL [I1 OOYi [I1 L 51 X i 0 6i 0
L6SP 0 Y06i 0
LLSP o i o h i n
LLSP 0 ZOhi 0
[ti1 SYIi [YI S'P LhLSP 0 lpO6i 0
it05 0 SPLS 0 LLSP 0
SSP 0 YO0 I 6P1 1
ios o PLS n LSP o
[YO11 OLLi IXOlI LL5

,AH 9605 [L61 XPYi [LO11 X'S
AH 0062 [061 OlRi 19011 PS
Table 1. Continued
No. Symbol Crystal Lattice Denskty Melting Linear thermal Thermal Electrical Enthalpiy Young Mxro Oxidation
Structure parameters point expansion, a conductivity, X resistivity modulus hardness resistance
(nm) (gem-') ("c) (10.' K-I) (W m-' K-') ncm) (kJ mol-') (los N miK2) (10 N mm-') (100°C)
250 fa 1251 2795 1731 2000-2500 (621

251 1900 HV P51
252 zrc fcc [I] 0.47 6.5 [I] 3535 Ill 6.73 [1,6] 20.93 [I] 63-156 (61 263.8 [I] 38 [I] 2560 [L,571 12 171
253 cub81 [6,76] 0.46894.476 6.51 (xj 161 3400 IS] 6.93 [XI 20.5 [XI 42 11 181 7 151 4 191 2600 HV [6, 81 11-14 [25]
254 fcc 181 0 4698 6.51 [?I 3532 [6] 7.0-74 191 I2 191 42 [9] 196.8 1111 4.07 [lo] 2600 [7, 10,32.45]
255 cub-BI [lo] 0.4698 6 59 (x) [lo] 3530 [?I 6.7 (101 20 52 (251 75 (2Sl 202.1 (461 3 88 (321 2500 191
256 cub-Bl (11,251 04685 6.9 I251 3530 [XI 8.3 I301 20 52 1321 43 1321 196.8 1731 3.9 (381 260C-2900 1131
257 cub-Bl 130.521 04698 66 [38] 3445 [9,45. 77, 97, 6.7 (32,481 19 (381 50 (381 184.6 (981 4 [45] 2560HV 113. 17.981
9x1
258 tet 1461 0.4694 6.73 [39] 3420 1101 57 1451 209 1981 42 1571 3. .32& [74] 2600 HV 0.05 (251
4 02
259 fcc [571 0.4693 6.73 [74,97] 3530 [39. 57,901 7 a 7 . 4 1771 20.5 (106] 42 1661 4 (771 2925HV 139,901
260 fcc 1741 0.47 6.63 (97,981 3443 I301 7-7.4 197) 42 (771 4 (971 2700 1591
261 fcc (981 0 467 6 63 [9,45,771 3420 [32.38.461 7-7.4 I981 42 [97.981 4 (981 lOo(t2800 174
262 mb-BI IlOS] 0.4693 6.73 11081 3530 [13.25, 1061 6.73 [I061 42 [I061 3 48 [I061
263 cub (1121 651
264 6.7
Nitrides
265 AIN hex [6] 0.311/0.498 3.05 >2200 [5, 106) 6 [38] 10 1381 10" 1381 288.9 [S] 3.15 1431 I200
266 hex 1431 0.31 IjO 494 3 09 1900 [6] 5.4 (431 Il&170[43] 10" 1571 290 (43) -3 1.571 1230HV
267 hex 1521 0.31 10/0.4975 3.25 2230(d) [7] 48 [471 30 [57] 10" [66] 3186 1461 3 [74] 25OOHV
268 hex (741 0.311\0.4979 3.05 2300(dj 1381 7 1571 165 [I081 10" [771 318 I651 3 437 1771 1230HK 1
269 hex (I121 0.311 14/0.49792 3.26 2397 1391 57 (771 10" (971 318 2 1731 35 138,971 1200
270 3.26 2227(d) [43] 5.7 [97] 0~~
1 0 ~ ~ - 1(ioq 1200 HV
27 1 3 26 2400 [47, 571 5.3 [lox]
272 3.26 2517 1731
273 3 26 2250 (d) (77,971
274 3.09
275 3.2
276 3.25
277 BN hex 111 0.251/0.669 2.25 3000 [I] 3.8 (381 284.7 [I] 3 x lo" (261 252.5 (51 0.9 [I] 4400HV
2?8 hex 130) 0 2504/0.6661 2.25 3000 [5] -2.8/41.5 [30] 25 138) 10" 138) 119.3 [6] 09 (381 4700
279 hex (391 0.2504/0.666 I 2 25 3000 [30] 1431 25 1431 10'' 1661 225 1431 3 1431 4700HV
280 hex (431 0 362 2.27 3000 (d) 1381 4.8 [XI- Is0-2Ol[81- 10" 191 252.5 1461 6.6 [9] 5000
2x1 hex 1461 0.36158 2.34 2997 (391 2 . 5 4 7 [9] 42 [9] 10'' (141 252.5 [73] 5.9 (141 4700 HK 0 05
282 hex 1521 3 48 2727 [43] 2.5-4.7 [35] 41.9 [35] lozo 1261 6.8 1361 210(t3600
283 hex [78] 3.48 2300(d) (71 [43] 200 [36] 10'8 [?7] 7.2 [36] 9OOC-9500 HV (mo)
284 cub [Y] 2.2 2730 [9] 3.5 1571 42 [43] 10"-10!~ [IOX] 8 1431 7000-8000 HV (PO)
285 cub (141 3.4b3.49 (mo)(351 2730 1141 14.0 (IOX] 15-33 [I081 6.8 (571 4050HK
286 cub 1351 3.3-3.4 (p0)[35] 2730 1251 20 [I061 6.6 (771 4700HK
287 cub [36] 3 48 1361 3027 6.6 1971 4000 HK 1571
288 cub [39] 34 1361 3000 4700 HK 0.1 1581
289 cub [43] 3 45 1391 2730 6006-8000 H K 0 I [70]
290 cub 1571 3 48 [43.88] 3000 200&4000 [761
291 cub [58, 761 345 [57] 2300 5000 H V [771
292 cub [77, 781 3 48 [70.77] 8000 H V 0.2 [881
293 cub [I121 2 34 [I061
294 22 [I071
295 18 [lox]
296 CrN f c c j l ] 0415 [I1 61 Ill I050 23 [lo] II 72 {lo] 640 I571 118-124 [61 4 [60] 1806-2100 171 7-7.5 [I251
291 cub-Bl 13.61 04149 [3] 5 39-7 75 161 1450 (d) 23 [32] 1 I 72 1321 640 1661 123.1 1461 3.236 1741 1100 [10.32.591
298 cub [I121 0414 161
. I
77 I500 23 1601 640 (771 I24 8 [lo?] 4 [77] 2006-2200HV005 [23]
299 cub-Bl [lo] 04149 [I01 6 I !x! I500 23 [971 640 [97] 123 I [73] 4 1971 2606-2900 HKOO5 [45]
300 cub-BI [ZS] 0.4148 [25] 6.1 I500 (d) 2.3 [XI] 118 I [Ill] 33 [XI] 1090 [571
30 I cub-Bl [46] 04149 1321 5Y loso 23 [I101 2 65-2.9 [I041 1x06-2200 H V 0 05 [hO]
302 cub.Bl 1521 04148 [52I 6 I !x) I050 23 [Ill] 4 11101 1965~2295HV 1641
303 fcc [57] 0.414 174. 1091 6I I500 7 S(XS0C- [ I l l ] 4 illlj 1890HV001 j75j
304 fcc [60] 0.4149 5.9 -1040C) 2.45 [I201 IIOOHV 1771
305 fcc 174. 751 0416 6 12 2300 HV [791
306 cub-Bl 176. 1091 0415 6 12 13284140 HVOOOOO5 [Xh]
307 fca [XI] 04148 59 70rMI00 H U 0 I [I031
308 cub [l04. 1101 0414 680-780 HU 0 I [I041
309 - 100&1700 H V 0 0 I5 [II51
310 1300 H V 0.05 [I181
31 I 810HV [1201
312 175P2650 H V 0 25 [I251
313 2600-2900HK 0 0 5 11251
314 Cr:N hex [25] 0 4760,’O 4438 59 - I500 94 308 [Ill] 3 138 1741 2250 H V 0 01 1751 8 6 3 1103 161
315 hcp [6. 52.751 0 4759 0 4438 59 1590 9 41 252 [Ill] 27 [Ill] 2500HV 1791 114 3 [46. 731
316 hex 13.74.761 0 48 I I 0 4484 58 lLCU2330 HK 1851 1277P [lo?]
317 hex [I041 04759 0.4438 6 51 IIOOHUOI [1031
318 hen [110.112] 02750447 1570HV005 [1111
319 0 481 I3 0.44841
320 HI% Ec [8] 0 452 13 8 3310 6.9 II 3 [XI 26 [71 369 4 1121 33341 [291 170CL2000
321 cub-Bl [lo] 0 4526 13 8 ( x ) ilbl 3300 69 2177 I101 I5 (321 369 [I71 4 64 1401 2000 H V
322 cub-BI 1251 0.451 I I 97 [74] 2700 6.9 21 77 1321 33 1571 369 3 [461 4.64 1451 1600
123 fcc 1521 0 4513 13 84 1981 3387 69 56. 5 11061 300 1651 3.8 1741 1500-3500 H V
324 fcc [57] 04518 1197 [XI] 3300 6.6 369.3 [73] 4.64 1871 150&8590 H V
325 fcc [74] 0 452 138 [I061 1387 66 3.8 1811 1600HVOI
326 cub-BI 1761 0 4525 13 8 [IO7] 3928 66 2700
327 fcc [YR] 0 451 3928 6.9 1640
328 fcc [Sl] 0 452 3330 6.9 1700
329 cub [II?] 0 45257 3300 6.9 2000
330 3300 160!-35W
33 I 3928 2700 H V
332 3200 2030
333 3305 1300 H U 0 I
334
Table 1. Continued
No. Symbol Crqstal structure Lattice Density Melting Linear thermal Thermal Electrical tnthalpiy Young Micro Oxidation
parameters point expansion, cz conductivity. X resistivity modulus hardness resistance
(nm) (gem-') K-’) (W m? K-’) Ikm) (kl mol-’) (10’ N rnm-’) ( I 0 N mm-’) ( I 00°C)
~ ~~ ~ ~~~
335 Mo?N-y fcc [6, 101 0.41394.4160 [lo] 9.46 [6] 700(d) 1571 17.88 [6] 70 3-81.6 161 630 (571
336 cub [I121 0.4128 1101 527 (731 69.5 (461
337 fcc I251 0.41554.4160 [25l 69 5 1731
338 fcc 1521 0.4163 1521
339 Icc [57,76] 04163 Ill21
340 Mo2N-B tet [6] 0.4184.42110.8-0.820 [6] 93 [6]
341 let 176, 112) 0.421010.8060
342 NhN-6 fcc [I, 81 0.44 83 [I] 2300(d) [I] 10.1 161 2 9 N . 6 1 [6] -200 1571 237.8 [I] 1378-1422
343 cub-Bl [6] 0.439 8 . 2 6 8 4 [6] 2300 18) 10 1 18) 3 74 [8] 237.8 161 1400 HV
344 cub41 [25] 0.4392 8.4 125) 2050 1251 1400
345 fcc [57] 0.4375 2630(d) 1571 1400
346 NbN-E hex (61 02951,’1.1271 8.4 [7] 2570 (71 10 I [lo] 3.77 [lo] 60 1321 21 1.9 151 4 814 [741 I4on
341 hex [lo] 0.2958/1.1272 7.3 (x) [lo] 2300 191 10 I [32) 3.77 1321 58 [66] 2366 [I21 4.8 1771 1400
348 hex 114.461 0.2958/1.1272 8 1741 2204 110. 32,971 10.1 1771 58 (771 234 1461 4.8 1971 1400 HV
349 hex (74,761 0.295211.125 8.43 1771 2630 1461 58 1971 236.6 (731
350 hex-Bl 1521 0 298610.5548 843 I971 2204(d) i771
351 NbN hex [I121 0.2980:1.1270 84 1106, 1081 2573 [I061 10.1 [lO6] 200 [I061
352 73 [I071
353 Nh,N hex [I] 021,0.496 8.3 [I] 2430 121 3.26 [6] 4 27-8.71 161 255.8 [I] 2120
354 hcp [6. 521 0.3052,’0.4964 8.08-8 62 [6] 2400 [73] 255.8 [6] 1620-1 820
355 hex [25] o 3058~04961 8.33 [52] 253.3 [12.73]
356 oriho (461 0.305610.4948 8 31 (x) [52] 248 7 1461
357 SIN, hex [6] ~0.78~0.56 3.44 [1.106] 1900 (11 2.4 [I] 20-24 [Y] 10” [Y] 750.5 [6] 21 1410 HV
358 hex (221 ?: 0.75810.5623 3.44 [6] 1900 [61 2.5 [9] 17-30 1221 10’8 1141 760 [22] 2.06 1700
359 a$ hex 143) p: 0.7610 30 3.44 (71 1900 I71 3 1141 I2 1381 in“-io18 [22] 630 (431 1’3.20 1700 H K 0.05
360 hex 1761 1:0 77541/0 56217 3.19 [Y] 1900 I91 2.7-3 5 1221 1&15 1431 10” 138) 745 2 [46] p. 2.90 1700 HK
361 z.P.hex 11121 p: 0.76044,O 29075 3 I8 1141 1900(d) 1141 28 1381 30 [57] lo6 [57] 2.2 a 28OCL4000
362 3.2 1221 1900(d) [221 2.7-3.3 1431 32 1671 lo1* 1661 1.63 0 8: 2000-3000
363 3.2 [38] 1900(d) (381 4.2 (471 10-13 [I081 >lo’’ (671 -3 2300 HV 0.05
364 344 139) I897 1391 24 1571 1018 1771 3.1 3340 HV
365 3.19 [43] 1897(d) 1431 3.3 [67l 1018 1971 3 2 0 W 5 0 0 HK
366 2.75-2.95 [47] 1900 (571 3 [TO] 10”-10~’ [lox] 2. I 1800 HV
361 I 8-3 I 147) 1900 [771 -1.5 (761 2. I cl.1700HKOI
368 3.44 [SZ] 1900 [W 2.5 [77] 1.7-3.0 25WiW0 HV
369 3.15(x) [52] 1900 [I061 2.5 [97] 3300
370 3.25 1571 33 (1081 1$00 HV 30
371 32 167) 2200-3200 HV 0. I
372 3 21 1701 5@3900
373 3.19 (71.971 I720 HV
374 3.2 1881 1400 HV 0.2
375 34 [lo?] 80&2000 HV
376 2.4-3.44 [lox]
377 T d N 0 52 0.29 [I1 [11 3000 111 36 8 58 [6l 128 161 225.7 5 756 1741 3240 5-8 [25]
378 0 5185,10.2908 [6] 13.613 8 [6] 3090 [5,25.88] 5 96 [S] 135 [25] 243 1 1060
379 0.5185/0 2908 [lo] 14 35 (x) [6] 3360 [l] 3.6 8 79 [lo] 128 [32] 25 I 1500-3000
380 0.495/0.305 (251 16.3 (7.741 2090 [8] 36 8 79 [32] I28 1571 252 5 1300 HV
381 0 518510 2908 [32] 14 34(x) (lO,52[ 3093 [lo] 13&250 [84] 252 5 1000
3x2 0 5 18510 2908 1521 14.1 [25,107] 3093 1321 1060
383 0.519210.2908 [74] 13.6 [52] 2950 (461 1150-3000
384 0 51Y18\0 29081 [Il>] 13.8 (881 W30(d) (571 3230 HV 0.2
385 163 [lO6] 3360 [I061
386 I6 3 [I081
387 Ta2N hex [21 0 3010 493 [2] 1 8 [I] 3000 121 10 05 [6] 263 161 270 9 3wo
388 hcp [6.10l 0 30476 0 49187 [6] I5 46 [6] 2700 [46] 19&250 [84] 270 9 I220
389 ortho 146) 0.3048/0.4919 [lo] 15.46 1521 2727 1131 270 5
390 hcp [14,52] 0.3048 0 4919 1521 15 86(x) [52] 273
391 hex [I121 0 3044510 49141 [In]
392 TIN fcc [I] 0 423 [I] 5 21 [I] 2950 [I] 9 35 29.31 Ill 25 [7l 336 6 25 [I] 2450 HV
393 cub-BI [3) 0.4238 [3] 5 21 161 2950 [?I 94 28 9 [XI 25 191 303 I 59 191 180~2800
394 cub-BI [6] 0.423 161 5 21 171 3220 [6] 9.35 30 19) 21 7 [ZS] 336 2 2.51136 [I?] 2400 HV
395 rcc 1x1 0 424 [lo] 54 [9] 2950 [7] 9.4 19.26 [lo] 25-30 [??I 338 I 2 56 (321 2500-2800
396 cub-BI [lo] 0 423 [25] 5.39 (x) [lo] 2930 181 9.35 19.26 1321 75-500 [27] 336 2 56 1371 2000
397 fcc 1231 0.424 [2Y] 5 21 1251 2950 [9) 9.35 29.2 [37] 25 1321 336 6 2.6 [38] 2000-2500 HV 0.05
398 cub-BI [25] 0 424 1321 5 21 [37] 2949 [lo] 9 35 38 [38] 30 [38l 337 59 1401 1-6000 HV 0 001
399 fcc [41] 0 4245 93 30 1571 25 [40l 333 6 1451 2000 HV 0 015
400 cub-BI [46] 0 423 94 19 [9l] 25 (571 338.1 25 [53] 220&2800 HV 0 05
40 I cub-BI [52] 0 424 54 (451 2930-2950 (291 93 70 [96] 17 161) 336.6 59 1601 1770 HK 0 I
402 Pc [57] 0 4241 4 73 1521 2949 132) 93 25 [66] 2 56 [61] 2300-2600
403 fcc [60] 0 424 5 43 1531 2950 [37] 9 35 25 [77] 256 1621 1170HK0.1
404 fcc [74] 0 42 5 21 [57] 2950 1381 9.35 25 [87] 2 51-6.40 [741 2450 HV 30
405 cub-Bl 1761 0 42 5.3 [60] 3205 (391 8.3 25 I911 59 [77] 1994 HV
406 fcc [loll 0 422 5.21 [74] 2900 [40] 8.3 25 [97] 2 56 [SO] 2100
401 fcc 19x1 0.424173 5.4 [77] 7.950 (451 9 35 25 (981 2 . 7 9 4 5 1 [82] 2200 HV
408 fcc ISl] 52 1881 2950 [46] 94 I7 [92] 59 [87] 3300 HK 0.05
409 cub-BI [I091 5 43 [90] 2950 I571 93 22 (1061 2.66640 [XY] 2300
410 cub [I121 5 21 [91] 2950 [60] 94 5.9 [91] 160&2000 HV
41 I 54 [93) 2950 [61] 94 5.9 [93] 2000-2700 HV
412 52 1961 2950 [62] 8 339.4 2.6 [95] 1900-2400
413 5.4 [97] 2949 1761 9 35 59 [Y7l 2450 HV0.I
414 5.4 [98] 2950 1771 83 59 [98] I050 HV 0.1
415 5 43 [92] 2950 [88] 9 35 25 [92] 200&2400 HV 0 05
416 5.21 [XI] 2900 187) 9 35 4 1811 170C-2800HV 0 05
417 5.21 [I061 3205 [90] 15OC-2700 HV 0 05
418 54 [I071 2930 [91] 1725-1825 HV 0.02
419 5.22 [lox] 2950 (931 1775-1935 HV
420 2950 [Y8] 1800 HV
421 2950 [92] 2160 HV
Table 1. Continued
Structure parameters point expansion, c( conductivity, X resistivity modulus hardness resistance
(nm) (g cm-’) (“C) K-‘) (W m-’ K-’) ncm) (kJ mo1-l) (10’ N mm?) (10 N mm-’) (100°C)
422 2947 [I061 1666-3098 HV

423 153k2730 HV 0.01
424 2450 HV 0.2
425 2100 HV
426 1994 HV
427 2300 HV 0.03
428 2500-2800 HV 0.05
429 2200 HV 0.05
430 180k2800 HV 0.05
43 1 500-1200HU 0.1
432 20W3350 HK
433 750-1200 H U 0.1
1640 HV 0.05
2225-2290 HK 0.01
22W2400 HV 0.01
434 Ti,N-E let [6] 0.494/0.303 [I011 3 0 W W O HV

435 tet [41,112] 0.49452/0.30342 [I121 240k2700 HV 0.01
436 VN fcc [1,8] 0.41 [I] 6.13 [7.74] 2050 (d) [!I 8.1 [XI 11.3 [6] 85-100 [6] 147.8 [S] 4.6 [77] 1520
437 cub-BI [6.25] 0.4137 [6] 5.62 [6] 2050 [5,6,7,8,25,73]8.1 [lo] 11.3 [XI [32] 217.3 [6] 4.6 [97] 140CL1600
438 cub-BI [46,52] 0.414 [3,10,32] 6.04 (x) 161
.. 2177 [10,32] 8.1 [32] 11.3 [lo] [57] 217.3 [46] 1500 HV
439 fcc [57,74] 0.4126 PI 6 [lo] 2350 (461 9.2 [77] 11.3 [32] [66] 217.3 [73] 1500
440 cubdl [3,10] 0.4169 [52] 6.04 [25,52] 2350 (d) [571 9.2 [97] [771 1520
441 cub-BI [76,109] 0.4139 [74] 6.11 177.971 2177 I77.971 8.1 [I061 85 [97l 1560 HV
442 cub [I121 0.413 [lo91 6.13 ilOSl . 2050 SO61 200 [I061
443 0.413916 [I121 6.13
444 6.1
445 WN-6 hcp [6] 0.2893/0.2826 2 5 W 0 0 0 HV 0.05 [23l

446 hex [10,52] 0.2893i0.2826
447 hex [I121 0.2893/0.2826
448 W,N fcc [6,52,57] 0.4126 [10,52,74] 12 IOl&180 [83]

449 cub [10,83] 0.4118 ~251
450 fcc [25,74] 0.4128 [831
45 I cub [I121 0.4126 [I121
452 ZrN fcc [I] 0.46 [I] 6.93 [61 3000 [Il 6 [I] 16.75 [I] 13.6 [6] 365.5 [I] 5.1 [9l 2000 111 1 2 [ 7 1
453 cub-BI [6] 0.458 [3] 7.09 [7] 2980 [5.7,8] 7.9 [8] 10.9 [8] 21 [7] 336.2 [5] 5.1 I451 1300-2000 [7] 11-14 [25]
454 fcc [8] 0.459 [6] 7.32 [9] 2980 [25.38] 7.2 [9] 28 [9] 21 [9] 365.5 [I21 3.93-1.60 [74] 1900 HV [XI
455 cub-BI [lo] 0.4577 [lo] 7.3 (x) [lo] 2980 [39,90] 7.24 [lo] 20.52 [lo] 13.6 [25] 368.4 [46] 5.1 [77] 1600 19.591
456 cub-BI [25] 0.456 [25] 6.93 [25] 2982 [6.9.10] 7.24 [32] 20.52 [32] 21.1 1321 366 [65] 5.1 [93] I500 [10.32]
457 cub-BI 1461 0.4575 7.3 [38] 2982 [32,45] 6.5 1381 19 [38] 20 1381 365.5 1731 5 I [97] I510 HK 0.1
458 cub-BI [52] 0.4577 7.09 1391 2982 [77.93,97] 7.24 1451 21 1571 1520 HV
459 fcc 1571 0.4562 7.32 1451 2930-2982 1291 7.2 [77] 21 1661 2300
4Ml fcc 1741 0.4578 7.09 i74;106] 2950 (463 7.2 1931 21 1771 1520
461 cub-BI [76] 0.463 7.32 [77] 2955 157) 7.2 1971 21 1971 1250HV 0.1
462 cub-BI [I091 0.457756 7.09 [90] 2980 [I061 7.24 11061 14 [I061 1500-2600
463 cub [I121 7.32 [93] I600HV
464 7.32 [97] 2500 HV 0.05
465 7.09 [in81 2130 HV0.05
466 7.4 [in71
Borides
467 AIB? hex [52] 0.3@86/0.3252 3.17(x) (521 1975 1461 67 [46]
468 hex [I 121 0.30054;0.325276
469 AIB,, tet 1521 O.lZ58jO.lO20 2.57 1521 2070 (per) [5] 10” 1661 201 [46] 4.3 1771 a: 2400 HK 1581
470 tet [I121 0.1030/0.1433 2.58 1771 2070(per) 1461 2 x 10’2 1771 2600 HV 1771
47I 1.01611.428
472 mono 1521 1.884
i.704~1.100~ 2150(d) [77]
473 0.8522~1.100/0.7393
474 CrB ortho [6] 0.296910.7858/0.~932 6.05 [6] I550 [6] 1251 -75.4 1461 1 6 1 8 [25]
475 ortho [25] 0.296910.785810.2932 6.05 [25,52] 1550 [25] 75.4 1731
476 ortho 1521 0.296910.785810.2932 6.11(x) 1521 I550 1731
477 ortho [I121 0.29663/0.78666,
0.29322
478 CrBl hex [I] 0.279I0.307 5.6 [1.106] 2200 [I] 11.1 [I] 56 [7] 94.6 [5] 2.15 [I] 2250 (1.571
479 hex [6] 0.296910.3066 5.6 [6,107] 2170 [5] 11.1 1571 21 [25] 129.8 [6] 2.15 1571 1700 161
480 hex [25] 0.296910.3066 5.6 [7] 18OC-2300 [6] 10.5 1771 56 1571 -94.2 1461 5.4 1771 2200 [71
48I hex I521 0.296910.3066 5.6 1251 2150 [7] 10.5 1971 I8 [66] 94.2 [73] 18on~~o.i [251
482 hex I571 0.29730/0.30709 5.6 1571 1850 [25.106] 7.1 [I061 18 1771 1800 WI
483 hex [I121 5.58 1771 2200 [571 I8 1971 2250 HV 177.971
484 5.58 191 1850-1900 1731 21 [I061 2100 [591
485 2188 177.971
486 Cr$, tet [I] 6.1 [I] 1900 [I] 125.6 [I]
487 tet (61 6.12 [6] 1960 1251
488 let 1521 6.14 1251
489 tet 11121
490 FeB ortho [52] 6.3 [52] 1650 [5,46,73] 71.2 5.46.73 14M1190 171
49I ortho [I091 1550 1611
492 ortho [I 121
493 Fe,B let I521 -7.0 1521 1389 [46,61.73] 71.2 46.73
494 let [I091
495 tet [I121
Table 1. Continued
Structure parameters point expansion, u. conductivity, X resistivity modulus hardness resistance
(nm) (g cm-’) (“C) K-’) (W m-’ K-’I ncm) (kl mol-’) (lo5 N mm-*l (10 N mm-’) (100°C)
496 HfB cub-Bl [52] 0.462 [521

497 ortho [I121 0.6517/0.3218/0.4919 [I121
498 HIB? hex 125) 0.3141/0.3470 1251 11.01 [7] 3200 5.3 [571 430 1981 I0 1251 336 6 [5] 2800 171 11-17 1251
499 hex 1521 0.3141/0.3470 1521 11.2 (25,1071 3380 5.7 19x1 15.8 1571 336.2 [46] 2900 [571
500 hex [57] 0.314245/0.347602 [I121 10.5 [52,108] -3060 5.3 [I061 12 (1061 336.2 (731 2900 [59
%I hex [I121 11.2(x) 1521 3370 2850 PSI
502 11 1571 3250 400M6000 HVO.O1 [ill
503 10.96 [I061 2147
504 3240
505 LaB6 cub [IOY] 0.415 IW 4.76 [YO] 2530 28 (1061 Z770 HV Pol
506 cub [I121 0.41569 11121 4.7 [l06,lO7l 2200
507 2.61 I1081
508 LaB, cub [52] 0.4156 1521 4.73 [7] 2530 5.84 130) 17 171 1400-2500 I71
509 cub [55] 0.4145 1551 4.76 1391 2200 6.4 [77] 15 1771 2770 HV 1-19]
510 0.4156 1551 4.73 [77] 2770 2530 HV 1771
511 MOB-6 tet 1251 0.31 10/1.695 1251 8.3 [ZS] -2500 50 [25] 1570 HV 0.1 1251
512 let 1521 0.3110/1.695 ~521 8.77 (x) [52] 1930 50 (571 1570 P31
513 te1 [I121 0.3105/1.697 1521 8.3 [52] 2350 25w 1571
514 0.3108/1.697 [I121 8
515 MOB-p ortho [25,52] 0.316!0.861/0.308 [25] 8.5 [571 2500 [591
516 ortho [57] 0.316/0.861/0.30X (521
517 ortho [I121 0.316/0.845/0.308 [I 121
518 MOB> hex [I] 0.3/0.31 [I1 7.8 [I] 2100 45 [25] 96.3 [I] l38DHV D.l (27 11-14 125)
519 hex [25, 521 0.305/0.3113 [251 7.12 [7] 2300 30 [57l 3000 [571
520 hCP 1571 0.306/0.310 1521 8 [25] 2250
52 I hex [I 121 0.304/0.307 ill21
522 Mo,B tet [I] 0.55/0.47 Ill 9.2 [I] 2140 40 [25] 106.8 [I] 2500 111
523 tet (251 0.5543/0.4735 125.521 9.1 [25] 1850 1660HVO.l ~ 5 1
524 te1 152, 1091 0.5547/0.4739 [I121 9. I [521 166C 1331
525 let (1121 0.554/0.474 11091 9.31 (x) [521
526 Mo,Br hex [I] 0.3/2.1 [I1 7.2 [I] 2300 8.6 [77] 18 1771 209.3 [I] 6.7 (771 3200 “I
527 hex 1251 0.3011/2.093 1251 7.48 (x) [52] 2140 8.6 [97] 18 [97] 6.7 [97] 2350 HV 1771
528 hex [52] 0.3011/2.093 1521 7.45 (77,971 2140 5 IW 25 [ID61 2350 HV P71
529 hex [ I I?] 0.30 I I74/2.09369 [I 121 7.12 [I061 2100
530 7.8 “071
531 Mo,B2 tet [I] 0.60/0.31 Ill 9 [I] 2240 175.8 [I] 2300 Ill
532 NhB ortho [6.25] 0 3298 0.X724 0 3166 [h. 251 I67 5 [6] 2200 [61
513 ortho [52] 032Y2!0.8713.03165 [5?)
534 ortho (1121 0 32973 0 87229 /I121
031883
535 NbBz hex [I] 031’033 “I 68 (I] 3000 7 1’96 [6] 1675 [I] 32 1151 150.7 (I] -2.6 [6] 2600 [1.6.57.59]11-14 1251
536 her [a] 0.3099,0.3271 I61 72 [6] 3000 [77] 16 75-35 2 [6] I? 1571 175.8 IS] 63 [77] 130&2600 [71
537 hen 1251 0.3086 0 3306 [28] [7. 571 3036 [97] 16 75 [?5] I? 1771 251 2 [46] 63 [97] 1800-1900 [281
538 hex [82l 0 3096,0.3306 (521 66 (35,521 -2900 h? (1061 16 7 (1061 I? (971 251 2 [731 2600 H V 1771
539 hex [57] 0.311133~032743 [Il?] 7.21 in) [52] 3036 32 [I061
540 hex [I121 6.98 [77.97] 3036
54 I 697 1106, 1081 2900
542 SiBh ortho [I121 1.4470 I835009946 [I121 243 [I] 1950 83 10’ [hb] 3.3 [77] 1910 [11
541 247 16. 1081 1947 10’ [77] 33 [97] 245&?800HV [39]
544 243 [3Y] 1900 54 10’ [Y7] 2300 H K 0 I 15x1
545 243 (701 1900 2400-2800 H V 0 I [70]
546 2.43 1771 m m HV (77. 971
547 TaB, hex [I] 0.31 0 33 [1,87] 3150 51 [I] 21 35 [6l 68 1251 209 3 [5] 2 62 161 2200 [1.57] 11-14 [25]
548 hex [6] 0 3085 0 3249 I2 I (x) [h] 3100 7 88 [6] 10 89 [25] ?I 1571 193 8 [6] 68 1771 2615 HK 0 I [61
549 hex [25] 0 3078/0.3265 I2 58 171 3040 82 1571 109 [I061 14 1771 209 3 [46] 2500 [7. 591
550 hex [52] 0 309803 0.322660 11.7 [25] -3000 82 [771 68 [I061 209 3 [73] 210&2400 [281
551 hex 1571 11.7 [521 3100 54 [I061 2500 H V [471
552 hex [I121 1?.6(x) (521 3150 2000-2700 [761
553 12.58 [771 3037
554 II I5 [106. 1081 3000
555 I2 6 [I071 2lOOHV [771
556 Td!B let [I] 058048 I52 [I] 1920 2200 “I

557 !el [b. 251 0 5778’0 4864 1518(x) [bl 2200 161
558 tet [52] 0 577810 4846 2430HV 1471
559 tet 11091 0 578 0 486
560 let [I I?] 0.5783:O 4866
561 Ta,B? tet [I] 0.62’0 33 2770 [11

562 tet 161 0 6184101284 2770 [61
563 tet 11121 (I 6184 0 3286
564 Ta,B, ortho [6. 251 0 329 I 4010 313 (6.25. 521 13 5 [6. 521 3350 1571
565 ortho [52] 032914010313 [I121 I36O(x) 16.521
566 ortho [I121
567 TiBl hex [I] 0 310 32 [I] 4.5 [1.107] 2900 6 39 [I] 2596 [I] 9 [71 1507 [I] 3.7 [I] 3480 [I] 11-17 [25]
568 hex 16) 0 302810 3228 [6] 4.38 [h. 1061 2900 78 191 26 13 16) 7 191 324 I 151 56 [9] 300&3400 171 13 [43l
869 hex [25] 0.3026’0 3213 [25] 4 38 [71 2920 4,8419.83 [30] 26-39 [9[ 7 [31] 280 [43] 5.6 [31] 250&3500 191
570 hex [30] 0.3028’0 3228 [52] 45 [9. 1081 3230 7.8 [31] 26.13 [25] 10 [38] 324.1 [46] 37 [38] 3400HVO05 1251
571 hex 1311
.. 0.303034 0 322953 [74] 44 [25] 3225 7.4 1381 27 1381 7 1401 305 1481 4.8 1401 3000 H V 131.97.981
572 hex [43] 03 [YX] 45 [31] 3225
Table 1. Continued
No. Symbol Crystal structure Lattice Density Melting Linear thermal Thermal Electrical Enthalpiy Young Micro Oxidation
parameters point expansion, a conductivity, X resistivity modulus hardness resistance
(nm) (g cm-3) ("C) (10.' K-I) (W m-' K-') (lo-' Qcm) (kJ mol-') (10' N mm-') ( I 0 N rnm-,) (100T)
573 hex 1521 0 303034,0.322953 4.5 3225 [77,97,98] 8 [431 24-56 [48] 9 1511 324.1 1731 4.8 I451 3400 1331
574 hex [57] 0.3018-0.3023/ ... 4.5 2980 [39,40.76] 7.8 1451 26 9 1981 I (66) 150.7 1981 5.6 1481 3300 HV (39,90,77]
575 hex (741 .../0.3194 322 4.52 2980 [87,90] 6.618.6 (481 26 I (1061 7 (771 3.74 1571 3370 140. 761
576 hex 1981 4.5 2980 1401 6.39 [571 7 [871 4.8 [I61 3000 I451
577 hex 11121 4.52 3197 [43] 8 1761 7 1971 5.6 171 2900-3700 HV 1471
578 4.38 3325 1451 7.8 1771 7 1981 4.8 1x71 25W3000 1481
579 4.5 2790 [I061 8 1871 15, 5 [I061 5.6 1971 3480 157
580 4.5 7.8 1971 5.6 [98] 3300 1591
581 4.5 7.8 1981 3370 HV (871
582 4.4 5.9 IW 34x0 HV 0.2 1881
583 164631 IS HV 0.05 (1231
584 IR43HV0.05 [I191
585 2 2 W 9 0 0 HVO.O1 [I271
586 Ti,B tet [I] 0.61'0.46 2200 171 2500 111

587 let p51 0.611/0.456
588 tet 1521 0.61010.453
589 VB, her [I] 0.310.31 48 2400 L5,1 5.3 203 9 (731 5.1 1771 2080 [I, 571 13 I71
590 hex 161 0.2998/0.3057 4.56 2450 161 5.3 5.1 1971 2077 (61 8-14 1251
89 I hex (251 0.2'99810.3057 5.1 2100 [25] 7.6 13@2100 171
592 hex 1521 0.2998jO.3057 5. I 2400 (571 76 2100 i5Yl
593 hex [57] 0.29976110.305620 4.61 2747 177,971 6.1 2150HV 1771
594 hex 11121 5 10 (x) 2120 HV 1971
595 5
596 5.05
597 4.92
598 WB let [I] 031L7 15.5 2860 [I. 251 3750 111
599 te1 1251 0.311511.693 15.3 2685-2920 161
.. 3750 1571
600 6: tet i 5 i 1121 0.3115/1.693 15.3 i52j 2860(d) 1571
601 p: ortho 1251 0.319/0.840/0.307 1251 16.0 (x) [52]
602 ortho 1521 0.319/0 840/0.307 152) 15 5 1571
603 onho (57) 0.311655/1.69101 [I121 15.7 [I071
604 10.77 I1081
605 W,B tet [I] 0.56047 Ill 16.5 [I] 2770 [I] 4.7 [I061 21-43 [I061 2350 [I] 8-14[25]
606 tet 1251 0 556410.4740 1251 16 125,521 2670-2780 161
607 tet 1521 0.5564/0.4740 1521 16.72 (x) [52l 2710 1251
608 tet (1091 O.S56!0.474 IW 10.77 (1061 2900 [I061
609 tet [ I 121 0 556810.474 [I 121
610 WB: 1513 [I) 2920 [7] 2600 171
611 W?B< hex 03’1.3Y 13 I [I] 2300 [I,571 78 !I 1571 77 1771 2700
612 hex U.?YX? I S87 II (521 29x0 PI 78 19 1771 7.7 [97] 2700
613 hcp 0 2982 1.387 13 1 IX) [52] 2200-2980 161 IY [9i] 2700HV
614 hex O:YX~?O~~S 13.03 [J7,9?) 2365 177. 97
615 ZrB fcc 0 J7 65 [I] 3000 1633 3600

616 cub-61 0 468 57 I521 3600
617 cub-Bl 0.465 6.7 in) LQI
618 ZrB, hen [I1 0 32)O35 61 [I, 38. 571 3000 (1.30. 1061 6.83 3.5 2200
619 ha IM 0.3170 o 3531 5 64 [6. 71 3040 [S. 39.901 6 I2 3.5 2600
620 hex 0.316910 3530 6.17 ps. 521 2992 “4 6.8 3.5 22Ml HV 0 05
62 I hex 0 1170~0.7533 6.17 1391 3250 171 59 54 2252 HV
622 hex 0 32 -6 (471 2990 [25. 38. 57.881 6 83 54 2200 HV
623 hex 03l6870~0353002 6 09 ( X I 1521 3200 (461 59 54 965 H V 0 I
624 hex 6 II 177. 971 3245 [77,97. 981 59 2300
625 hex 6.11 1981 5.9 2300-3000
626 6 PXl 55 2300 H V
627 6 17 I901 2200 H V 0 2
628 61 IW
629 6 0x5 (1081
630 5 64 lW
631 ZrB,? cub 0 14 3 65 I 2 I4 1251 502.4 [I] 2500
632 fcc 0 7408 3.7 2500
633 cub 0 7408 3.7
634 cuh o 7408 3.63 (1)
Silicides
635 CrSi cub I11 0.462 5 38 [6] 1550 [I] 53.2 1461 1000
636 cub 0 4619 1550 (61 54.8 [73] 95&1050
637 cub 0 462 1475 [63]
638 cuh 0.46074 4629 1457 (pa) 1731
639 cub 0 462
640 CrSi? hex “I 0 44?:0 655 491 [61 1630 161 100.5 [I] 1100
64 I hex 0.443 l/0.6364 4.4 1251 1451 146) 80 1461 xw110
h42 hex 0.442210.635I 4.91 [SZl 1570 I251 80.1 1731 I IS0 HV 0.1
443 hex o 4427’0 6375 4.4 (1071 1457 1731
644 hex 0 4420,O6349 5.5 IW
635 hex [II?] 0.44281 0.63691
646 Cr1Si cu h 0.455 6.52 (61 1710 [I] 105 5 [I] 90&980
447 cub 0.45W 6.52 1251 1770 151 105.5 IS] 1005
648 cuh 0.4555 6.45 1521 1750 [61 92.1 146)
649 cu h 0 455W.4564 1770 146. 63. 73) 105.5 (731
650 cub 0 4558
Table 1. Continued
~~~~ ~
No Symbol Crystal Lattice Density Melting Linear thermal Thermal Electrical Enthalpiy Young Micro Oxidation
structure parameters point expansion, 01 conductivity, X resistivity modulus hardness resistance
(nm) (g cm+) (“C) K-‘) (W m-’ K-’) ncm) (kJ rno1-l) (10’ N mm-*J (10 N mm-’) (IOO’C)
651 Cr3Si2 5.6 [6] 151&1610 (61 1420 [7] 10Sc-I2W I61
652 4.7 171 I550 1000 I71
653 Cr,Si tet [I] 0.919/0465 111 5.6 111 1560 [I] 1280 “I
654 Cr& tet [6] 0 9170/0.4636 1251 5.9 161 1647 1461 211.4 1461
655 tet 1521 0.9 I70,’0.4636 1521 5.6 1521 I647 [73] 223.2 1731
656 5.9(xJ 1521
657 LaSi2 tet 1301 0.4281/1.375 11121 -1520 [30] 10.78/17.05 (301
658 tet 11121 > 1500 163)
659 MoSi2 tet [I] 0.32/0.786 [I] 6.3 [1,107] 2050 [I,7,22.30] 8.4 [I] 221.9 [I] 21 [71 108.9 [I] 3.84 [I] 1290 III 17171
660 tet 1221 0.32/0.65 122) 6.31 17, 1081 2030 [25,39. 73,901 8 1221 50 122) 20 1221 131.9 1461 4.4 [22) 1300 17,221 > 17 I251
66 I tet [25.28] 1).320/0.786 1251 6.2 1221 2020 1501 5.32115.55 1301 21.5 125) 131.5 [SO] 1290 H V 0 I [251 17 I1001
662 tet [30.52] 0.3205/0.7848 [28] 6.12 1251 2190 (per) 1631 131.8 1731 1200 H V (39. YO1
663 tet 1461 0.3203:0.7886 1521 6 [3Y, 901 900HV0.1 1581
664 tet [I091 0.320/0.786 IIW
665 tet [I121 0.32047/0.78449 11121
666 MolSi cub [I] 0.489 Ill 8.8 111 2150 [I] 100.5 [I] 1310
667 cub 1251 0489 1251 8.4 [25] 2050 [25] 116.4 [5] 1310HV0.1
668 cub 1461 0.489 1521 84 (521 2025 1731 116.4 (461
669 cub 152, 1121 0.489 [I121 8.97 1521 116.5 I731
670 Mo,Si, tet [I.1121 0.96410.49 III 7.8 Ill 2100 [I] 280.5 [I] 1170
67 1 cub 1461 0.96210.490 1521 2180 [SO] 310.2 146. SO]
672 tet 1521 0.96483/0.49135 11121 2190 1731 309.8 1731
673 NbSi, hex [I] 0.4810.66 [I1 5.5 Ill 1950 [I] 8.4 161 6.3 1251 50.2 [I] 700
674 hex 161 0.4785/0.6576 161 251 5.69(x) 161 1950 [7] 35.6-67.8 [6] 66h1320 H V
675 hex 1251 0.4797/0.6590 1281 5.7 171 1950 1251 138.2 (461 lOSOHV0.1
676 hex [28, 521 0.479506589 1521 5.29 1251 1930 (521 138.2 (731
677 hex [I121 0.47971/0.6592 [lli] 5.37 [losl
678 5.6 [I071
679 Nh& hex [I] 0.36j0.50 111 7.8 111 1950 [I] 550
680 hex 161 0.35910.446 161 8.01 161
681 Nh& 01: tet 161 3:0.6570/1.1884 [6] a: 7.09 (x) 161 2480 151 0: 7.3/4.6 161 452.2 151
682 p: tet [6] p l.Ml0/0.5070 [6] B: 7.?O(xJ [6] 2480 [63] 452.2 (731
683 3:tet 152,721 a: 0.658311.1884
684 p: tet 152,711 p: l.OOl0/0.5070
685 tet ]I121 I1 6569Rll.lX87
686 TaSi2 hex 161 0 4773'0.6552 92 [7] 2200 8 Y18.8 161 38 1507 1410 161 11 171
687 hex [25] 0 4773,0.6552 Y.O?(x) (61 2400 46 1193 I200 17) 11-14 1251
688 hex (28.521 0 4783106567 883 (251 2200 8.5 1192 1560 HV 0 I 1251
689 hex 146) 0.4781,O 6564 9.14 152. 1081 I560 1331
690 her (I 121 0 47835110.656980 8.4 (A) (52]
69 I 9.1 [107]
692 Ta,Si iei [I] 061,050 13 5 [I] 2450 I24 83.7 I500 111
693 let [a] 0615705039 13 54 (xi (6. 521 126 4
694 tet 1521 06157,05039 I? 4 125 6
695 tet [ll?] 06157~05039
696 T a , S let (61 TI o 9880:~5060 2500 1631 T2 5 5'8 0 (61 I08 335 4
697 let [6] TZ.0 6516 I 1873 2500 1731 334 9
698 let (461 TI~O9820,05010 334 9
699 tet 1521 T206513,I 1864
700 tei Ill21 0 6516, I 1873
701 TiB! ortho (251 0823604773'08523 (h.251 4 39 [90] 1520 18 1344 892 HV (35. 901 II [?I
702 ortho 1281 0 8263'0 4800'0 8553 12x1 4 17. 1081 1470 123 I34 3 700 171 8-1 I 125)
703 urtho (521 0 8253,O 4783'0.8540 1521 4 39 125. 391 1540 134.4 870 HV 0 I
704 ortho [IOO] 0 8267 0 4800,O 8551 11001 44 [I071 -1527 870 HV I
705 ortho [IIZ] 0 82687 0 85534 407 (1001 -Is50
0 47983
706 1480
707 TI,% her [I] 0 75 0 52 43 [I] ?I20 7.03 15 61 1301 577 8 986
708 hex (251 074650516? 4 32 [loo] 2120 9.2 [I001 579 986 HV 0 I
709 hex 130.521 0 7465 0 5162 2130 579 9 968HV I
710 hex (I001 0?44805141 -2150 579. I
711 hex [IIZ] 07444,05143 579 5
712 VSI! hex [I] 0.46!0 64 45 Ill

L 1
I650 95 95 960
713 hex 161 04562,06359 4 3 4 4.42 [h] 1680 I25 6 89G960
714 hex [25] 0 457310.6374 57 171 1759 125 6 IZO&II?OHV
715 hex 128. 521 0 4571/0 6372 471 (25. 1071 1090 HV 0 I
716 hex [I121 0 45723010.63730 442 jl08l
717 VISi cub [I] 047 55 [I] 2060 97 154 9 I500

718 cub (61 0 4712 5 33-5 67 161 2030 (di 1507 1410-1560
719 cub (25. 521 0 4721 567 [52] 1935 I50 7
720 cuh [I121 047253
721 V5Si, tei 1521 TI 094?8104750 2150 (5.631 462 2

722 lei [ll?] 0.94276'047555 2010 1461 462 6
723 WSil let [I] 03210788 .. 9.5 Ill 2165 11. 391 6.5 [??I 45 1221 125 92 I 1090
724 tet (61 0.32 0 78 (221 9.4 [7: 1081 2170 [7] 10 92 9 1100
Table 1. Continued
parameters point expansion, u conductivity, h resistivity modulus hardness resistance
(nm) (g cm-9 ("C) K-') (W m-' K ' ) ncm) (kJ mol-I) (10' N mm-') (10 N m K 2 ) (100°C)
725 tet I221 0.3212/0.7880 1251 9.8 1221 203W2W I221 33.4 1251 92.8 [73] I200 [221
126 let [25,28] 0.321310.7829 1281 9.3 125, 107) 2150 [25.46] 1090HV0.1 1251
727 let 146,521 0.321 1/0.7868 1521 2160 1731 1074 HV 1391
128 tet [IIZ] 0.3211/0.7829 [IIZ] 2210 1901
729 W& tet [I] 0.561,10.496 Ill 14.56 [I] 2320 [I,731 125.6 [I] 770 Ill
730 tet 161 0.96410.497 (521 13.06 (100] 2350 (99,631 134.4 151
73 I tct [5?1 0.960110.4972 Ill21 135.2 146)
732 tet [I121 134.6 1131
733 ZrSi hex 146,521 0.7005/1.2772 1521 5.65 (x) [52] 2090 (461 141.8 1461
734 onho [SZ] 0.6982/0.3786/0.5302 [52] 2107(per) (731 154.9 1731
735 ortho [ I 121 0.6981!0.3785/0.5301 [I121
736 ZrSi, ortho [6] 0.372/1.41610.367 16,251 4.87 [6] 1700 [25] 9.7 [loo] 161 [25] 159.4 (IOO] 2.348 [IMI] 1030HV0.1 (251 8-11 [ZS]
737 ortho [25,28] 0.3698/1.4761/0.3665 (281 4.88 125, 1081 1520 [I001 151.1 1631 942 HV 1 IW
738 ortho [52, 1001 0.3721/1.468/0.3683 152, 1001 4.83 [lo01 1517(per) 1731 159.5 173)
739 ortho [I121 0.36958/1.4751/0.36654 [I121 4.9 11071
140 Zr2Si tct [I] 0.66/0.54 Ill 6 [I] 2110 11,461 309.8 [I] 1230 111
74 I tet [46, 521 0.661210.5294 1521 6.22 152) 2107(per) I731 339.1 [46]
742 bl [I121 0.66oS/0.5298 [I121 208.5 (731
743 Zr5Si, hen [46] 0.7886/0.5551 [52] 2150 1461 614.2 1461
744 hex (52, 1121 0.7885j0.5558 [I121 576.1 I731
Oxides
745 Al?O,-a hex 0.5127 - a = 55'16.7' 1521 3.99 2043 8 [31 30. I [XI 1580.I 151 4 [7] 2100HV0.05 PI > 17 1251
746 hex 0.512 1571 3.98 2045 8.6 [a] 27-36 (Y] 1590 (431 4 (91 2200 HV 141 18 1431
741 rhom 0.513 --u = 55.3' I1091 3.9 2030 7.2-8.6 I91 4.2-16.7 1161 1678.5 3.6 1161 ZIOOHV [8. 77,87,98] 20 1701
748 rhom 0.554410.9024 I1121 3.99 2047 7-9 [I61 25 I381 1690 3.5 I251 IXW?XK! I91
149 rhom 0.47588i/1.2992 ju~j 3.8-3.9 2015 9.5110 (301 27 1431 1676.4 4. I 1381 250W3000 HV 0.05
750 hex 3.98 2050 8.3 1381 2s [57] 4 1401 ?Ow HK 0.1 1331
751 hex 4 2050 9 1401 27 [h3] 4 [.I31 2800 133. 691
752 rhomihex 3.8 2050 8 1431 28 167) 4 [A51 1800 1341
753 hex 399 2041 7.7 [ q 30.1 19x1 X.? IS31 28WHV
754 hcx/rhom 3.96 (x) 2300 6.8 1531 35 (1061 2.5 (531 2100
755 3.98 2047 ?: 8.5 1561 27-36 [I211 5.2 ph] 2020HK
756 3.96 2040 7.2 .rj7i. 4 (h?] 18WHV
757 3.97 2050 8.4 1621 4 16-31 2100
758 3.95 2050 8 3.7 (h71 ZOSOHV
759 3.99 2054 8.5 37 [hY] 1200-1600HV
760 3.96 2015 9 4 1701 250&3000 HV
761 3.98 [77] 2050 4 I761 2000
762 3.9 188) 2300 4 1771 ZIOOHVOI
763 398 (981 2047 3.8 [82] 20W2300 HV
764 3.98 (92) 23M) 4 [871 2080 H K 0 I
765 4 “5] 2050 38 [95] 2000-2500 HV
766 3.99 [lO6] 2047 4 1981 1200-1600 HV 0 1
767 3.9 [I071 2046 2.5 [92] 2300
768 3.9 [I211 2045 294 [I061 20w2500
769 2045 400 (121) 855-975 HV 0 02
770 2000 HV 30
771 15OLL2200 HV 30
772 2000-3000 HV 0 I
173 23OO HV 0 2
774 2000 HV
775 2600-2800 HV
776 A120,-? cub (svtnellj 1510HK
[521
777 fcc 0.3958 1521 2200
778 cub 0.79 lll2l
779 cub-El o 774.8 iiosi
2450 9 1431 264 (431 lo?’ (661 569 3 [?5] 1230-1490 HV
780 Be0 hex 0 269910.4401 [52l 3 (251 IS] [S]
781 hex 3 03 L39.771 2530 [?5] 9 1771 10’’ (771 580 [43] 3.9 [43] IS00 HV
782 301 1431 2570 1391 608.8 (461 3.9 [77]
783 3.01 1521 2567 [43] 599.1 PI
784 3 00(xj 1521 2580 [45]
785 2550 1771
186 Cr02 tet 0.441,0?91 1521 4.8 (521 582 8

787 tet 04421’0 2916 [I121 490(x) [52] 582
788 CrO, ortho 0.573/0.852/0.474 [6] 2.81 [6] 170-198 [6] 579 9
789 ortho 0 574310.8557/0.4789 1521 2.7 1521 185 [46] 589.9
790 ortho 0 57494,’O 8556’0.4796 [I121 2.82 (x) [52] 578.6
791 Cr,O, rhom 0.536-%=55’ [6l 5.21 [6] 2440 [7] 6.7 [53] 10” [6] 1130.4 1000 HV 1341
792 rhom 0.5361 -rr=55’ I521 5.2 [7] 2300 [39] 6.7 [60] 1.3 x lo9 [I061 1140.5 2915 HV 1531
793 hex 0.495876/1.35942 [I121 5.21 [39] 2400 [46] 5.6 [I061 1130.4 2300 HV PI
794 5.22 1521 2343 [53] 12W1700HV 1531
795 5.25 (x) [52] 2343 [a] 12W1700HV0.1 IS81
796 5.41 1531 2708 1106) 2000 1591
797 5.4 1601 1200-1700 HV0.05 [60]
798 5.21 [I061 I100 HV 0.02 [64]
799 Cr,O cub 0 4544 1521
800 Cr,O, tet 0 87210.750 1521

801 tet 0.6145 0.755 [II?]
802 Hf02 mono 0 512 1521 9.7 [7] -2900 15.471 10 [43l 3 [43l 5 x 10’’ [I061 1053.4 900 [591
Table 1. Continued
No Symbol Crystal structure Lattice Density Melting Linear thermal Thermal Electrical Enthalpiy Young Micro Oxidation
parameters point expansion, a conductivity, X resistivity modulus hardness resistance
(nm) (g cm-7 (“C) K-’) (W m-’ K - ’ ) ncm) (kl mol-’) (10’ N mm-’) (10 N nim-’) (100°C)
803 mono [46] 0.51 [I091 9 68 [43] 2790 [7] 6.5 1771 1010 [43] 780 H V 1771
804 cub [IW] 0 5281~0.51819~0.5115[I121 10.2 I771 2897 [43] 1113.7 1461
805 mono [I121 9.63 [I061 2810 [46] 11138 [731
806 96 [I071 2790 [I061
807 MkQ cub [43] 0.4208 1521 3.6 [7] 2850 [5] [43] 36 [43] 10” [66] 568 6 [5] 3.2 [43] 745 H V
808 cub-El [461 0.42 [IOY] 3.5 [25, 1071 2800 [7.25.39. 531 [771 36 [I061 10” [77] 570 I431 3.2 [771 520HV
809 cub-El [52] 04213 [I121 3 65 [391 2837 (431 II 2 [I061 10” [I061 601.6 [46] 2.4 [I061 -400HV
810 cub-Bl [IOU] 3.58 [43. 531 2825 [46. 731 601.6 [73] 520 H V 0. I
811 cub [I121 3.77 1771 2827 [77] 750 H V
812 3 58 (106. 1081 3073 [lob]
813 Nb!O, mono [I121 2893/03827;1.758 [I121 4.5 [7] 1460-1520 [61 1.76 [6] 1905.8 161 740
814 447 [I061 1510 171 1903 3 [73] 700
815 4.6 [I071 IS12 (731
816 1785 [I061
817 S10 2.1 17, 1071 1750 [71 98.4 [461

818 21 [I051 1610 [46]
819 2.13 [106,108] 1705 [I051
820 1975 [I061
821 SiO, quartr- quartr 233 [6] 1703-1729 [6] 04 [I061 I38 [I061 10;’ [hh] 911 [46] 0.5-10 [82] 113&1260 161
822 trigonal [I091 0.4093,’O 5393 [6] 2.2 [7.105] 1713 [7] 05-075 [I081 12-1.4 [I081 10.’ [I061 911.5 [73] 1.114 [I051 1200 1591
823 0 421.’0.539 [I091 162.7 [47] 1722 [46] H IIIS] 0 654.75 [I081 IOW1?00H V IS?]
824 2 1-2.2 (471 1373-1473 [73]
825 CriStoballtK: cristobalite 2.2 11061 1713 [IOS]
826 t ~ t I61 0.70,0.69 [6] 2.6 [I071 1983 [I061
827 cub [IOU] 0.704 [IOY] 2.18 [I081
828 Ta,O,-u tet [6] 0.3808 0.3567 [52] 8.37-9.48 [6] 1880 (51 2 161 1953 I IS] 66&1030 1991
829 tKt [52] 75 [7] 1887 [h]
830 8.53 (x) 1521 1880 [71
83 I 82 [99]
832 TdlOr-p ortho 161 0 619 0 367 0 389 [6] 8 18-891 [6] 1785 [6] 2 161 10” [h] 2045 3 [h]
833 ortho (461 0 6192 4 4019 0 I898 (521 8.30(xJ [52] 1900 (461 2047 3 [46]
834 ortho [5?] ?047 3 [73]
835 Pd!Ol ortho [II?] 0 6198 4 029 0 3888 [I121 83 [I051 1880 11061 [ 1061
836 7.53 (1061 1880 (IOS]
837 8.7 [I071
838 Tho, fcc [6] 05859 [6] 10.05 (x) [6] 3250 [5] 10 1431 10 1431 1016 [6h] I I71 I [5] I 3X [h] 950 H V
839 cub 143) 05595 [52] 9.7 [7] 2997-3250 [a] 9.3 [77] 10 [I061 10’‘ 1771 11711 (431 14 I251
840 fcc [46] 0.521 Y7 1251 2990 2 x 10" [I061 1227 5 1461 24 [431
841 fcc [52] 0.5597 10 [43] 3050 1227 2 1731 24 [771
842 cub [ I091 10 [??I 3217 148 [I061
843 cub [II?] YY [I061 3370
844 10 [I071 3220
R45 3300
846 3323
847 Ti0 cub-Bl 16.461 0417 4 88 161 1750 [6. 7. 32. 1051 7 6 [I061 II [I061 [32] 520 [??I 1100
848 Cub [II?] 0418 49 [7. 1051 2020 I461 1800
R4Y cub-BI [52,76] 0424 493 [I061 1700 11061 I OIIO
850 cub-Bl [IOY] 04177
851 Ti0: tet [4h] 0 4543:O 2959 4 19 [6l 1900 8 [Y] I.2 x lo"' [I061 945 4 I461 2 05-2 80 [Y] 767 I000 HK
852 let 1521 0.449,'o 2959 4I [7] 1855 5 [lox] 945 4 I731 2 05 1771 7oo I Ino
853 cub-BI [25] 045933'0?9592 425 [Yl 1867 0 8-20 [X?] 1~IOOHV
854 tet IW 2.40-2.49 1411 IXbO 205 [98] 600 1000 HV
855 tel [II?] 4 16 [53] 1840 065 I 4 9 11241 600- 1000 HV I1 I
856 425 [77.98] 1867 I100 HV
857 42 [88] 1860 600 7on
8% 424 [I061 1860 1000 HV 1) 2
859 42 [105.107] 1913 II50 HV
860 3.02-3.92 [I241
861 TI~O, rhom 1461 0 5454 - x = 59 5' 4.6 [5?, 1051 2130 IO'Y-IO:' [lox] 1433 I [I] YXO HV
862 rhom [52] 0 5139,l 3659 3.58 1x1 [52] 1842 IS?I 9 [46.73]
863 rhom [I121 4.05 [I081 1760
864 ThoI mnno 1461 0.9828'0.3776,0.9898 [II21 1780 2461 1461

865 mnno 11121 24608 [73]
866 WO? mnno 16.46, 521 0 5560,O 4884,O 5546 [52] I? I [6] 1500-1600 161 (1-1. 5 ) x [hi 587 [61
10"
867 tet (521 0 4870,0.2776 [52] 11.05 1521 1724 (per) I461 590 I I461
868 tet [IOY] 0 55610.555 11091 1082(x) [52] 1724(d) [?)I 590 I I731
869 mnnn 11121 0.55754~048995.' [I121
0 55608
870 wo, 111 [6] 0.382 0.748)O 728 [6] 7 16 [hl 1473 161 0.6 [I061 843 4 1461
87 I mono [46. 521 0.3835/0.7517/0 7285 [52] 7 16 [7. 1061 1473 17. 1061 843 5 1731
872 tet 1521 0 525010 39 I 5 [52] 1472 [46]
873 Ill 11121 07309107522:07678 [I121
874 Y?O; bcc 1461 1.0601 [521 5 [7. 1051 2465 151 8 [iXl 14 1381 1906 7 1461 I8 [38]
875 b€C I521 1.06 [I091 4.5 1381 2415 [7] 81 [lox] XI? [I081 IY067 1731
a76 cub [IOY] 1.04041 [I121 5.03 [I081 2450 [38]
877 cub [I 121 4 84 [I061 2704 [46]
878 2420 [73]
879 2410 [I051
880 2683 [I061
Table 1. Continued
parameters point expansion. c( conductivity. X resistivity modulus hardness resistance
(nm) (B cm-7 ("C) (lo-' K - ' ) (W in-' K - ' ) ncm) (kJ mol-') (10' N mm ') (10 N m m - * ) (100°C)
881 Zr02 cub 1391 0.511 1521 5.6 17, 1051 2150 [S] 7.5- 10.5 [9] 0.7-2.4 [9] 10l6 (91 1035 IS] 1.63 [h] 1200 PI > 11 [ZS]
882 cub 1521 0.521/0.5209,'0.5315 161 5.7 19, 881 2681 [7] I0 1381 2 1381 10" 1381 1040 [43] 1.8 [9] I200 HV 139.771 17 1431
883 mono (39,431 0.522Oj0.5211;0.53XI [52] 5.8 (251 2611 (9,43,53,77] 9-10.5 (431 2 (431 loi6 (661 1101.5 (461 1.7 1251 ISOOHV 1531
884 mono 146,521 0,508410.5165 1521 5.56 138,431 2690 125, 391 10 1611 2.Y 1671 10" [77] 1098.2 1731 2.4 [38l IOOOHV 1581
885 tef 1521 0.508 [I091 5.56 (mo)[39] 2680 138,461 10 1771 1.7 [I061 2.05 1431 1500 HK 0.1 1581
886 cub [I091 6.21 (cub) 1391 2700 I881 17.2 [I061 2.1 [67] 1600 I591
887 mono 16, 1121 0.51463iO.52135iO.5311 [IlZ] 6.1 153, 1081 2681 11061 1.9 [771 1300 HV 30 1671
888 tet [I121 0.364/0.521 [IIZ] 6.01 1671 2700 [IOS] 400-1000 1761
889 5.76 1771 I100 HV 0.2 1881
890 5.6 [I061 150&1650 11161
References 99 1
References
I . R. Kieffer and F. Benesovsky, Hartstoffe. Springer, Vienna, 1968.
2. H. Holleck, Binare und ternare Carbide und Nitride der Ubergansmetalle und ihre Phasenbezie-
hungen. Gebr. Borntrager Verlag, Berlin, Stuttgart, 1984.
3. U. Konig, Untersuchung von kathodenzerstaubten binaren und ternaren Hartstoffschichten
zur VerschleiBminderung von Hartmetallen. Tribologie Reibung, Verschle$, Schmierung.
-
Band 9, Springer, Berlin 1985, pp. 282, 287, 296, 306, 31 I , 316.
4. H. Kolleck, Die Bedeutung von Phasengleichgewichten bei der Entwicklung VerschleiBfester
Werkstoffe. Chem. 2. 1982, 106, 216.
5. M. Riihle, Zum technischen Stand der Dispersionshartung - Teil I . Metall. 1982,36, 128 1-1285.
6. Gmelin: Handbuch der anorganischen Chemie. Verlag Chemie: Weinheim 1960. Angaben in den
jermeiligen Bande der Elemente (z.B. Cr, Ti,
7. R. A. Haefer, OberJachen- und Diinnschichttechnologie Teil I: Beschichtung von Oberjfachen.
8. H. Frey and G. Kienel, Diinnschichttechnologie. VDI-Verlag, Diisseldorf 1987, p. 121.
9. B. Rother and J. Vetter, Plasma-Beschichtungsverfahren und Hartstoflschichten. Deutscher
Verlag fur Grundstoffindustrie, Leipzig, 1992, p. 171-172.
10. L. E. Toth, Transition Metal Carbides and Nitrides. Academic, New York, 1971, pp. 5-7, 80-
81,92,95-97 and 148-150.
11. E. K. Storms, The Refractory Carbides, Academic, New York, 1967.
12. H. L. Schick, Thermodynamics of Certain Refractory Compounds. Vols. 1 and 2, Academic,
New York, 1966.
13. E. Hornbogen, Werkstofe: Aufbau und Eigenschaften von Keramik, Metallen. Polymer- und
Verbundwerkstoflen. Springer, Berlin, 1987, p. 286.
14. H.-J. Bargel and G. Schulze, Werkstoffe.CDI-Verlag, Diisseldorf, 1988, pp. 293-294.
15. V. Demarne and E. Bergmann, Verwendung von Hartstoff- und Hartmetallschichten fur
Schneidwerkzeuge. In I . Int. PVD-Tagung an der TH Darmstudr am 15.116. Murz 1983, Vol.
20, ISBN 3-88607-027-1, THD Schriftenreihe Wissenschaft und Technik 20, 1983, pp. 269-277.
16. K. H. Kloos et a/. Herstellung und Eigenschaften gesputterter Aluminiumoxidschichten. In 2.
Int. PVD-Tagung an der TH Darmstadt am 11.112. Marz 1986, Vol. 30, THD Schriftenreihe
Wissenschaft und Technik, 1986, p. 100.
17. H. Jehn et a/. Beeinflussung der Schichteigenschaften mangetrongesputterter Nitridschichten
durch Substrattemperatur und Sputteratmosphare. In 2. Int. PVD-Tagung an der T H Darm-
stadt am 11,112. Murz 1986, Vol. 30, THD Schriftenreihe Wissenschaft und Technik, 1986,
p. 138.
18. J.-E. Sundgren et a/. Structure and Properties of HfN Coatings grown by Reactive Sputtering.
In 2. Int. PVD-Tagung an der T H Darmstadt am 11,112. Marz 1986, Vol. 30, THD Schriften-
reihe Wissenschaft und Technik, 1986, p. 185.
19. P. Johannsen, Beschichtete HSS- und Hartmetall-Werkzeuge in der GroBserienfertigung. In 2.
Int. PVD-Tagung an der TH Darmstadt am 11.112. Marz 1986, Vol. 30, THD Schriftenreihe
20. U. Kopacz and H. Jehn, Harte- und Haftfestigkeitsmessungen an Nitridschichten auf
Schnellarbeitsstahl. In 2. Int. PVD-Tagung an der T H Darmstadt am 11.112. Marz 1986,
Vol. 30, THD Schriftenreihe Wissenschaft und Technik, 1986, pp. 217-218.
21. H. R. Stock and P. Mayr, Hartstofheschichtung mit dem Plasma-CVD-Verfahren. In 2. Int.
PVD-Tagung an der T H Darmstadt am 11.112. Marz 1986, Vol. 30, THD Schriftenreihe
22. A. Petzold, Physikalische Chemie der Silicate. Deutscher Verlag fur Grundstoffindustrie,
Leipzig, 1991, p. 182.
23. D. Kammermeier, Charakterisierung von binaren und ternaren Hartstoffschichten anhand
von Simulations- und Zerspanungsuntersuchungen. Fortschritt-Ber. VDI Reihe 2 Nr. 271,
VDI-Verlag, Diisseldorf, 1992, p. 2.
24. R. F. Bunshah, Deposition Technologies for Films and Coatings. Noyes Publications, New
Jersey, 1982, p. 149.
25. R. Kieffer and P. Schwarzkopf, Hartstoffe und Hartmetalle. Springer, Vienna, 1953.
26. S. u. Z. Marinkovic et al. Boron Nitride Coatings. Kernforschungsanlage Julich GmbH, 1989.
27. S. u. Z. Marinkovic et al. Titanium Nitride Coatings. Kernforschungsanlage Jiilich GmbH, 1988.
28. B. Lonnberg, Synthesis, Structure and Properties of some technologically important Carbides,
Borides and Silicides. Acta Univ. Uppsala, 1988, pp. 17-21, 26 and 30.
29. T. Wolf, Herstellung und Charakterisierung von TiN, ZrN und HjN. Dissertation Universitat
Karlsruhe, 1982.
30. H. Bleckmann, Rontgenographisclze Bestirnmung der therrnischen Ausdehnung an verschiedenen
metallischen und nichtmetallischen Hartstoffen. Dissertation RWTH Aachen, 1971.
3I . G. Hilz, Zum EinfluJ innerer Grenzflachen auf Aufiau und Eigenschaften mehrphasiger Hart-
stoffschichten. Kernforschungszentrum Karlsruhe GmbH, KfK 5022, 1992.
32. G. D. Brundiers, Herstellung, Aujbau und Eigenschaften von Hafniumverbindungen irn System
Hf-C-N-0. Kernforschungszentrum Karlsruhe GmbH, KfK 2161, 1975, p. 7.
33. P. Schwarzkopf and R. Kieffer, Cemented Carbides. Macmillan, New York, 1960, pp. 138 and
214.
34. A. R. Lansdown and A. L. Price, Materials to Resist Wear. Pergamon Press, Oxford, 1986,
pp. 7 and 99.
35. R. Reinhold and S. Becker, Superharte Schneidwerkstoffe der Spanungstechnik. VEB Verlag
Technik, Berlin, 1982, p. 17.
36. W.-E. Borys, Vergleichsuntersuchung zum Einsatz hochharter polykristalliner Schneidstoffe
beim Frusen. Dissertation Universitat Hannover, 1984, p. 3.
37. W. Konig and H. U. Schemmel, Untersuchung moderner Schneid.ytoffe - Beanspruchungs-
gerechte Anwendung sowie VerschlezJlursachen. Forschungsbericht des Landes Nordrhein-
Westfalen, Nr. 2472, 1975, p. 54.
38. F. Loffler, Eigenschaften von keramischen Hartstoffschichten auf Silicium- und Aluminiumbasis.
Fortschritt-Ber. VDI Reihe 5 Nr. 180, VDI-Verlag, Dusseldorf, 1990, p. 23.
39. R.-J. Peters, Beschichten mit Hartstoffen. VDI Technologiezentrum Physikalische Technologien.
VDI-Verlag, Dusseldorf, 1992, pp. 2-3.
40. H . Freller and H. P. Lorenz, Kriterien fur die anwendungsbezogene Auswahl von Hart-
stoffschichten. VDI-Verlag, Dusseldorf, 1992, p. 22.
41. C. Ribeiro, Silbernes TiN durch Magnetronsputtern und seine Anwendungsbeispiele. VDI-
Verlag, Dusseldorf, 1992, p. 124.
42. E. Erturk and H.-J. Heuvel, Neue Schichtsysteme mit dem Arc-PVD-Verfahren. VDI-Verlag,
Dusseldorf, 1992, p. 254.
43. F. Thummler et al. Fortschritte mit neuen Werkstoffen: Keramik fur den Maschinenbau.
Keram. Z. 1988,40, 158-159.
44. K. Dettling, VerschleiDfeste Werkstoffe fur Extrusionswerkzeuge. Keram. Z . 1988, 40, 802.
45. R. Riesenberg et al. Mechanische Eigenschaften von Schichten. Wissenschaftliche Schriften-
reihe TU Chemnitz, Heft 9/1990, p. 53.
46. 0. Kubaschewski et al. Metallurgical Thermochemistry. Pergamon,. Oxford, 1979, pp. 268ff.
47. C. E. Morosanu, Thin films by chemical vapour deposition. Thin Films Sci. Technol. 1990,7,
pp. 225,229, 381, 391 and 424.
48. R. Telle, Boride eine neue Hartstoffgeneration? Chemie in unserer Zeit 1988, 22, 95.
-
49. P. Hedenqvist, Evaluation of vapour-deposited coatings for improved wear resistance. Acta
Univ. Uppsala, 1991, p. 13.
50. T. C. Chou and T. G. Nieh, Anisotropic grain growth of d-NiMo. Thin SolidFilms 1991,219,61.
51. G. Berg; E. Broszeit, C. Friedrich, W. Herr and K. H. Kloos, Grundlageneigenschaften und
VerschleiBverhalten von HfB,- und Hf(B,N)-Schichtsystemen. Mat.-wiss. u. Werkstoftech.
1994.25, 175-179.
52. W. B. Pearson, Lattice Spacings and Structures of Metals and Alloys. Pergamon, London,
1958. DD. 219ff.
53. H. Simon and M. Thoma, Angewandte Oberjlachentechnik f u r metallische Werkstoffe.Hanser,
Munchen, 1985, pp. 103, 108 and 131.
54. G. Berg, C. Friedrich, E. Broszeit and K.-H. Kloos, Comparison of fundamental properties of
r.f. sputtered Ti,N-and Hf,N coatings on steel substrates. Surf. Coat. Technol. 1995, 74-75,
135-1 42.
Rejerences 993
55. H. Bolmgren, Synthesis and structural studies of some refractory borides. Acta Univ. Uppsala
1992, p. 19.
56. N. P. Cheremisinoff, Handbook qf Ceramics and Composites. Dekker, New York, 1990, p. 24.
51. W. Schedler, Hartmetali fur den Praktiker. VDI-Verlag, Dusseldorf, 1988, pp. 3, 8-9, 12 and
207.
58. K.-H. Habig, VerschlegundHurte von Werkstofen. Hanser, Munchen, 1980, pp. 125, 164,211
and 263-268.
59. H.-D. Steffens and W. Brandl. Moderne Beschichtungsverfahren. Lehrstuhl fur Werkstofftech-
nologie der Universitat Dortmund, 1992, pp. 242 and 261.
60. J. H. Mittendorf, Einsatz von Werkzeugheschichtungen,furnusgenlahlte Umjormverfahren. Dis-
sertation RWTH Aachen D82, 1992, p. 123.
61. H. Benninghoff, Industriereife CVD-Verfahren. Metalloberfliiclie 1976, 30, 476.
62. E. Erben et a/. CVD ein modernes und vielseitiges Beschichtungsverfahren. Metall 1981, 35,
-
1256.
63. M. Ruhle, Zum technischen Stand der Dispersionshartung Teil 2. Metall 1985, 39, 520, 522
-
and 528.
64. K.-H. Habig, VerschleiB-Schutzschichten - Entwicklungstendenzen zur Optimierung von
Eigenschaften und Verfahren. Metall 1985, 39, 915.
65. H. Holleck, Neue Entwicklungen bei PVD-Hartstoffbeschichtungen. Metal1 1989, 43, 622.
66. S . Hofmann, Charakterisierung keramischer Werkstoffe mit der Auger-Elektronenspek-
troskopie: Moglichkeiten und Grenzen. Material~~,i~~senschaft und Werkstofftechnik 1990, 21,
99.
67. H. Kolaska and K . Dreyer, Hartmetalle, Cermets und Keramiken als verschleiBbestandige
Werkstoffe. Metall 1991, 45, 227 and 232.
68. W. Wahl, VerschleiBfeste metallische GuBwerkstoffe. VerschleiBfeste Werkstoffe. Vortruge der
VDI-Taguiig Stuttgart 1973, VDI-Berichte Nr. 194 (1973), p. 82.
69. G . Schumacher, Hartmetall fur den VerschleiBschutz. Vortruge der VDI-Tagung Stuttgnrt
1973, VDI-Berichte Nr. 194 (1973) p. 113.
70. E. Gugel, Nichtoxidkeramische Werkstoffe fur die VerschleiBtechnik. Vortruge der VDI-
Tagung Stuttgart 1973, VDI-Berichte Nr. 194 (1973), p. 139.
71. B. Finnern, Konstitution, Verfahrenstechnik und Anwendung verschleiBfester Diffusions-
schichten. Vortriige der VDI-Tagung Stuttgart 1973, VDI-Berichte Nr. 194 (1973), p. 219.
72. K.-H. Kloos, WmkrtqfJkzmdeIV, Vorlesungsumdruck TH Darmstadt, 1994, pp. 2.4.2.5f.
13. I. Barin and 0. Knacke, Thermocliemical Properties of Inorganic Substances. Springer, Berlin,
1971 and 1973.
74. H. Jehn, G. Reiners et a/. Charakterisierung dunner Schichten. DIN-Fachbericht 39. Beuth,
Berlin, 1993, pp. 134, 155 and 171.
75. K. Kashigawi et a/. Chromium nitride films by radio-frequency reactice ion-plating. Vuc. Sci.
Technol. 1986, A4, 210 and 214.
76. J.-E. Sundgren et ul. A review of the present state of art in hard coatings grown from the Vapor
phase. Vac. Sci. Technol. 1986, A4, 2260ff.
77. H. Holleck, Material selection for hard coatings. Vac. Sci. Techno/. 1986, A4, 2663.
78. W. Gissler et a/. Mixed phase nanocrystalline boron nitride films: preparation and character-
ization. Thin Solid Films 199I , 199, 1 15.
79. R. A. Andrievski et ul. Structure and microhardness of TIN compositional and alloyed films.
Thin Solid Films 1991, 205, 173-174.
80. N. Zhang and V. Wang, Dislocations and hardness of hard coatings. Thin Solid Films 1992,
214, 5.
81. C. Friedrich, G. Berg, E. Broszeit and K.-H. Kloos, X-ray diffractometry analysis of r.f.
sputtered hard coatings based on nitrides of Ti, Cr, Hf. Surf. Coat. Technol. 1995, 74-75,
279-285.
82. E. Vancoille et a/. Mechanical properties of heat treated and worn PVD TiN, (Ti, Al)N, (Ti,
Nb)N and Ti(C, N) coatings as measured by nanoindentation. Thin Solid Films 1993,224, 170
and 175.
83. S. D. Marcus and R. F. Foster, Characterization of low pressure chemically vapor-deposited
tungsten nitride films. Thin Solid Films 1993, 236, 331 and 332.
84. Sun, Xin et a/. Properties of reactively sputter-deposited Ta-N thin films. Thin Solid Films
1993,236, 350.
85. W. Heinke, A. Leyland, A. Matthews, G. Berg, C. Friedrich and E. Broszeit, Evaluation of
PVD nitride coatings, using impact, scratch and Rockwell C adhesion tests. Thin Solid
Films 1995, 270,431438.
86. A. Schroer, Synthese und chemische, mikrostrukturelle und mechanische Charakterisierung von
Hartstojschichten ayf Stiihlen. Dissertation Heidelberg, 1992.
87. J. Berger Beschichten mit Hartstoffen. VDI-Verlag, Diisseldorf, 1992.
88. E. Dorre, Nichtmetallische Hartstoffe, in Abrasion und Erosion, H. Uetz (Ed.), Hanser,
Miinchen, 1986,433437.
89. C. Biselli and L. Chollet, Young’s modulus of TIN and TIC coatings, in Mechanics of Coatings.
Tribology Series 17 D. Dowson, M. Godet and C. M. Tayloer (Eds), Proccedings of the 16th
Leeds-Lyon Symposium on Tribology, Lyon, Amsterdam, Elsevier, 1990.
90. P. Mayr and H. Vetters, Gefiige, Oberflachenzustand und technologische Eigenschaften von
Hartstoffschichten. H T M 1993, 48, 281-287.
91. R. Wild, Hartstoffschichten leisten Widerstand. Metalloberflache 1990, 44, 355-361.
92. R. SaB, and P. Thienel, Oberflachenbehandlungen von SpritzgieRwerkzeugen. Metallober-
,fliiche 1991, 45, 313-322.
93. R. Rochotzki and J. Vetter, Festkorperreibung von PVD-Hartstoffschichten: TIN,, ZrN,,
TIC,, TiC,/i-C. Neue Hiitte 1989, 34, 272-271.
94. R. S. Bonetti and M. Tobler, Amorphe diamantartige Kohlenstoffschichten. Metallobery%iche
1990,44, 209-211.
95. K. Keller and F. Koch, CVD-Beschichtung von FlieRwerkzeugen. VDZ-Zeitschrift.1989, 131,
42-50.
96. M. Fieber and W. Konig, Kontaktbedingungen beim Einsatz beschichteter Werkzeuge. Tribo-
logie + Schmierungstechnik 1995,42, 135-140.
97. G. Kienel (Ed.): Vakuumbeschichtung 4 und 5, Anwendungen Teil 1 und 2. VDI-Verlag,
Diisseldorf, 1993.
98. B. W. Matthes, Abscheidung, anwendungsbezogene Prufung und Optimierung von PVD-
Hartstofieschichtungen des Systems Ti-B-N/Al2O3aus der Hochfrequenz-Kathodenzerstau-
bung. Dissertation, TH Darmstadt, 1993.
99. P. J. Martin et al. Properties of thin films of tantalum oxide deposited by ion-assisted deposi-
tion. Thin Solid Films 1994, 238, 183-184.
100. R. Rosenkranz, Strukturen. physikalische und mechanische Eigenschaften der hochsch-
melzenden Ti5Si3-,TiSi2- und ZrSiz-Phasen. Fortschritt-Ber. VDI Reihe 5 Nr. 291, VDI-
Verlag, Diisseldorf, 1993.
101. E. D’Anna et a/. Synthesis of thin films of semiconductor and refractory metal nitrides by laser
irradiation of solid samples in ambient gas. Thin Solid Films 1992, 218, 224-225.
102. J. L. Davis et al. Thin metal films on polyether imide. Thin Solid Films 1992, 220, 221.
103. G. Berg, C. Friedrich, E. Broszeit and C. Berger, Development of chromium nitride coatings
substituting titanum nitride. Surf. Coat. Technol. 1996, 86-87, 184191.
104. C. Friedrich, G. Berg, E. Broszeit and C. Berger, Measurement of the hardness of hard
coatings using a force indentation function. Thin Solid Films 1996, 290-291, 216-220.
105. A. G. Balzers, Coating materials and sources selector guide. Company information. Balzers
AG, FL-9496 Balzers, Fiirstentum Liechtenstein.
106. A. G. Leybold, Berechnungsgrundlagen fur Diinnschicht-Technik. Company information 01 -
017.9.70.61.02.21/KT 5.
107. Cerac Inc.: Advanced special inorganic materials. Company information 1995. Cerac Inc., PO
Box 1178, Milwaukee, WI 53201-1178, USA.
108. Goodfellow: Firmenkatalog zu Metallen, Keramiken und anderen Materialien. Edition 19951
96. Deutsche Niederlassung: Goodfellow GmbH, Postfach 1343, D 61213 Bad Nauheim.
109. K. Schafer and C. Synowitz, Taschenbuchfur Chemiker und Physiker. Bd. 3,3. Aufl. Springer,
Berlin, 1970.
110. W. Herr, B. Matthes, E. Broszeit, M. Meyer and R. Suchentrunk, Influence of substrate
material and deposition parameters on the structure, residual stress and adhesion of sputtered
Cr,Ny hard coatings. SUTJCoat. Technol.,1993, 60, 428433.
References 995
1 1 I. M. Atzor, Aspekte des Magnetronsputterns zur Herstellung verschleifl- und korrosionsbestiin-

diger Schichten auf Chrombasis. VDI-Fortschrittsbericht Reihe 5 , Nr. 156, VDI, Diisseldorf,
1989.
1 12. International Center for Diffraction Data Ponder Diflraction File, ASTM, Swarthmore, PA
19081, USA, CD-ROM 1992.
113. S. Luridiana and A. Miotello, Spectrographic study of oxide growth of arc evaporated TiN
and ZrN coatings during hot air oxidation. Thin Solid Films 1996, 29k291, 289-293.
114. P. J. Kelly, 0. A. Abu-Zeid, R. D. Arnell and J. Tong, The deposition of oxide coatings by
reactive unbalanced magnetron sputtering. Sur$ Coat. Techno/. 1996, 86-87, 28-32.
1 IS. C. Gautier. H. Moussaoui, F. Elstner and J. Machet, Comparative study of mechanical and
structural properties of CrN films deposited by d.c. magnetron sputtering and vacuum arc
evaporation. Surf. Coat. Technol. 1996, 8 6 8 7 , 254262.
116. M. S. Wong, W. J. Chia, P. Yashar, J. M. Schneider, W. D. Sproul and S. A. Barnett, High-
rate reactive d.c. magnetron sputtering of ZrO, coatings. Sur-: Coat. Techno/. 1996, 86-87,
381-387.
117. M. H. Staia, E. S. Puchi, D. B. Lewis, J. Cawley and D. Morel, Microstructural characteriza-
tion of chemically vapor deposited TIN coatings. Surb Coat. Technol. 1996, 86-87, 432&l37.
118. A. Igartua, J. Lauricirca, A. Aranzabe, T. Leyendecker, 0. Lemmer, G. Erkens, M. Week and
G . Hanrath, Application of low temperature PVD coatings in rolling bearings: tribological
tests and experiences with spindle bearing systems. Surf: Coat. Techno/. 1996,86-87, 460-466.
119. S. Heck, T. Emmerich, I. Munder and J. Steinebrunner, Tribological behaviour of Ti-Al-B-
N-based PVD coatings. Surf. Coat. Technol. 1996, 86-87, 467471.
120. K.-D. Bouzakis, N. Vidakis, T. Leyendecker, 0. Lemmer, H.-G. Fuss and G. Erkens, Deter-
mination of Fatigue behaviour of thin hard coatings using the impact test and FEM simulation.
Surf. Coat. Technol. 1996, 86-87,467A7 1.
121. F. Fietzke, K. Goedicke and W. Hempel, The deposition of hard crystalline A120, layers by
means of bipolar pulsed magnetron sputtering. Szrb Coat. Technol. 1996, 86-87, 657-663.
122. G . DErrico, R. Chiara and E. Guglielmi, PVD coatings of cermet inserts for milling applica-
tions. Surf. Coat. Technol. 1996, 86-87, 735-738.
123. J. Steinebrunner, T. Emmerich, S. Heck, I. Munder and R. Steinbuch, A novel impact tester
operating at elevated temperatures for characterising hard coatings. Surf: Coat. Techno/. 1996,
86-87, 748-752.
124. 0. Anderson, C. R. Ottermann, R. Kuschnereit, P. Hess and K. Bange, Density and Youngs
modulus of thin TiOz films. Fresenius J A n d Chem 1997, 358, 3 15-3 18.
125. B. Navinsek, P. Panjan and I. Milosev, Industrial applications of CrN (PVD) coatings, depop-
ited at high and low temperatures. Surf. Coat. Techno/. 1997, 97, 182-191.
126. J.-R. Park, Y. K. Song, K.-T. Rie and A. Gebauer, Hard coating by plasma-assited CVD on
plasma nitrided stellite. Surf: Coat. Techno/. 1998, 98, 1329-1335.
127. E. Kelesoglu and C. Mitterer, Structure and properties of TiBz based coatings prepared by
unbalanced D C magnetron sputtering. Surf. Coaf. Techno/. 1998, 98, 148-1489.
Index
a-modification, silicon nitrides 753 - silicon carbides 695

a-SiAION structure 755 ff, 768 ff alumina addition, titanium carbide SHS 362
ah initio pseudopotential approach 256 f alumina-aluminum DMO composites 3 14
Aboudi model 72 alumina-based ceramics 184, 192
abrasive applications, diamond materials 528 ff - tool applications 6 4 8 4 8 2
abrasive wear aluminothermic reduction 875
- silicon nitrides 784 aluminum
- tungsten carbide-cobalt hardmetal 962 - cemented borides 896
abrasives 5 ff - diamond synthesis 498
- silicon carbides 688,736 aluminum borates 164
absorption bands, CVD diamond 577 aluminum boride type structures 805 f
absorption coefficient, diamond-like carbon 638 aluminum interlayers, diamond-like carbon
acetylene 629 films 641
Acheson process 683,688 f aluminum kryolite melt 880
acid stability, boron carbides 855 aluminum lithium alloys 289
acidic solutions aluminum-magnesium systems, directed metal
- CVDdiamond 415 oxidation 295
- silicon carbides 735 aluminum-magnesium-zinc system, directed
activated sintering 877 metal oxidation 292
activation energy, self-propagating synthesis 357 aluminum-nitride-silicon carbide system, self-
active corrosion 141 ff propagating synthesis 354
additives aluminum nitrides 715
- boroncarbides 843 aluminum oxides 787
- cemented ternary borides 925 aluminum titanate 3 18
- diamond synthesis 498 AMBORITE milling 559
adhesion ammonia, carbide synthesis 2 10
- diamond-like carbon films 640 f amorphous carbon XLIX, 272
- silicon carbides 722 f - films 6 2 3 4 4 7
adiabatic combustion temperature 323 amorphous composites, nanocrystalline 112
agglomerates, alumina-based ceramics 654 amorphous covalent ceramics (ACC) 446 ff
aggregates amorphous hard materials 36 ff
- diamond 380 amorphous structures, carbon nitrides 261
- directed metal oxidation 307 amorphous zone, silicon carbides 690
aggressive environment 584 anatase structures, titanium oxides 32
alkali halide gas 43 anisotropic growth, silicon nitrides 763
alkali stability, boron carbides 855 annealing, nanostructures 130
alkaline earth nitrosilicates 29 anthraxolite 376
alkoxides antireflection coatings, CVD diamond 589
- polymer-ceramic transformations 447 anvils, diamond synthesis 490,565
- thin film processing 463 apatite XL
allenes LI applications 477-995
allotropes XLIV, 27 1-285,485 - alumina-based ceramics 6 4 8 4 8 2
- diamond synthesis 390 f - borides 802,933
allylhydropolycarbosilanes 463 - cemented borides 927 ff
alumina 3, 292 - CVD diamond 410,573422
- bulkcomponents 69 - diamond materials 479-572
- directed metal oxidation 29 1 f - diamondboron nitrides 527
- elasticmoduli 71 - diamond-like carbon films 640 ff
998 Index
- directed metal oxidation 3 16 binders

- polycrystalline ultrahard materials 548 ff - bulkcomponents 68
- silicon carbides 736 ff - carbides 12
- silicon nitrides 751 ff, 782 ff, 792 - cemented ternary borides 9 19
- single crystal diamond 559 ff - polycrystalline diamond 516
- TM carbideshitrides 202 ff, 238 ff - self-propagating synthesis 357
arc evaporation, diamond-like carbon films 627 - titanium boride-iron composites 910
Archimedian three connected nets 275 - tungsten carbidexobalt hardmetal 952
argon 43 - tungsten carbides 165
argon sputtering binding energies, boron carbide-silicon carbide
- boroncarbides 8.56 ceramics 859
- ncNiCr 111 birefringence, silicon carbides 7 I9
armor, directed metal oxidation 3 17 black body radiation 45
armor modules 94,97 black composites, alumina 669
Arrhenius behavior block sawing, diamond abrasives 543
- self-propagating synthesis 323 blocking, corrosion 146
- silicon carbides 450 boehmite
aspect ratio - alumina-based ceramics 653.661
- silicon carbides 695 - Vickers hardness 184
- siliconnitrides 767 Boltzmann constant 45,93, 322
ASTM, silicon carbides 736 bonded tools, diamond/boron nitrides 528
AT60 composite 674 bonding 3 f, 198
atomic force microscope (AFM) 54,405,440 - borides 803
atomic hydrogen, diamond synthesis 507 - boron nitrides 420 f
attack modes, corrosion 141 f - carbon allotropes 274
attrition milled powder mixtures 916 - carbon nitrides 258
Auger electron spectroscopy (AES), 2 11 - diamond-like carbon 630 ff
augmented-plane wave (APW) calculations 15 - TM carbides/nitrides 204 ff
austenitic phases, carbides 16 borane-dimethylamine 60
automated pressing, silicon carbides 704 boride-based hard materials 802-945
boride-zirconia composites 888 ff
borides
P-modification, silicon nitrides 753 - bulkcomponents 69
P-phases, silicon carbon nitrides 265 - crystal structures 8 f
ballas 512 - data collection 965-995
ballistic properties 94,97 ff boron
band gaps, borides 803 - diamond synthesis 500
barriers, coulombic 94 - silicon carbide-based materials 683
BC(8/C4/T4/T8) series, carbon allotropes boron-based hard materials LVIII
272 ff boron-boron interactions, borides 808
bell jar reactor 398 boron carbide-based cermets 895 ff
belt devices, diamond synthesis 490 boron carbide-based composites 857 ff
bending strength boron carbide ceramics 837 ff
- alumina-based ceramics 649 boron carbide process 875
- boron carbides 852,865 boron carbide-silicon carbide ceramics 857
- silicon nitrides 772,778 boron carbide-transition metal diboride
bending tests 73 ceramics 861
benzene boron carbides XL
- carbide synthesis 208 - ballistic properties 94
- diamond-like carbon films 627,633 - corrosion 161 ff, 176
Berkovich model 86,195 - crystal structures 8 f
Berman-Simon line 374 - silicon carbides 7 14
beryllium oxides 30 boron-carbon-aluminum system 820
bias voltage, diamond-like carbon films 624 boron-carbon-metal system 819
bias-induced heteroepitaxy 404 f boronxarbon-silicon system 8 19,822
binary carbides/nitrides 202 ff boronxarbon system 8 13
binary systems, borides 81 3 ff boron-carbon-titanium system 823 ff
Index 999
boron carbonitrides LXII, 526 calibration, diamond anvil cell 43

boron doping CALPHAD (calculation of phase diagrams) 213
- carbon allotropes 276 carbide conversion 876
- diamond-like carbon 638 carbide nitride-boride-silicide composite-based
boron halides 855 hard materials 888
boron nitrate 406 carbide-reinforced composite ceramics,
boron nitride/silicon carbide double layers 309 alumina-based 669 ff
boron nitrides XL-LXXI, 6f, 253 ff carbides
- alumina-based ceramics 648 - bulkcomponents 69
- bulkcomponents 69 - corrosion 155ff
- corrosion 171 ff - data collection 965-995
- crystallization 5 10 ff - superstoichiometric 127
- cubic polycrystalline 479 f - transition metal diboride cermets 916
- CVD diamond 597 ff - transition metals 12 ff
- high pressure melting 49 carbon
- nanostructures 104, 125 f - corrosion 154ff
- self-propagating synthesis 340 - diamond-like XLIX
- vaporphase deposition 42M45 - hydrothermal synthesis 374
boron-rich nitrides 526 - self-propagating synthesis 338
boron-silicon system 815,821 - silicon carbide-based materials 683
boron suboxides XL, LXV, 526 carbon allotropes XLIV, 485
boron-transition metal systems 830 - diamond synthesis 390 f
boro-silica glass 860 - doping 271-285
borothermic reduction 876 carbon-based hard materials XLIV ff
boundary conditions, corrosion 142 carbon black
boundary sliding, nanocomposites 121 - boroncarbides 843
Bragg angle - directed metal oxidation 293
- crystal stuctures 36 - titania-aluminum system 364
- nanocomposites 119 carbon black reduction 875
breakdown carbon-boron system 813
- corrosion 147 carbon clusters 470
- spinel layers 300 f carbon content, tungsten carbide-cobalt
Bridge method 853 hardmetal 952
Bridgmen anvils 490 carbon dioxide, diamond synthesis 490
Brillouin zone 276 carbon dioxide laser heating technique 4 1 4 5
Brine11 hardness XLI, 85, 107 carbon filaments, silicon carbides 697
brittle failure 67 carbon films, diamond-like 6 2 3 6 4 7
brittleness 20, 74 ff carbon-h ydrogen-oxygen system, hydrothermal
bromine reactions 855 synthesis 376 ff
bronzes 895 carbon phase diagram 485
brookite structures 32 carbonado 5 12
brown corundum 658 carbonitrides LIII, 202-270,523
bubble formation, silicon carbides corrosion 160 - corrosion 173ff
bulk components, mechanical properties 68 ff - nanostructures 104 ff
bulk modulus XLII carborundum 7,683
- carbon allotropes 277 carbothermic reduction
- carbon nitrides 263 - boronoxides 838
- polycrystalline boron nitride 519 - metaloxides 875
- silicon carbon nitrides 265 - silica 691
bulk properties, TM carbides/nitrides 203 - silicon carbides 7 15
Burgersvector 93, 110, 115 carburization rate, diamond synthesis 394
carrier gas hot extraction 21 1
cascade arc plasma jet 624
C,, see: fullerenes cast iron machining 660, 667,670
calcite XL cast self-propagating high temperature
calcium silicate, melting temperatures 51 synthesis 342
calcium sulfate coatings 307 catalysts, diamond synthesis 498 f
cathodic arc deposition 422 - titanium boride-yttria stabilized zirconia
caustic alkalis 155 composites 893
caustic oxidizing media 415 coagulation, alumina-based ceramics 654
cavities 92 coarse grain, CVD diamond 580
cemented borides 895 ff coarsening see: Ostwald ripening
cemented carbides 12,238 f coarser powders, polycrystalline diamond 5 13
cemented carbonitrides 238 f coated abrasives 736
cemented diamond compositions 5 12 coatings 66,70 f
cemented ternary borides 9 19 ff - boroncarbides 838
centrifugal compaction 654 - boronnitrides 429
ceramic alumina 3 - diamond-like carbon films 640
ceramic bodies processing 446-476 - directed metal oxidation 307
ceramically bonded silicon carbides 700,72 1 - nanocomposites 121 f
cermets 202,345 - self-propagating synthesis 366
charge carrier density - TM carbides/nitrides 242 f
- boronnitrides 438 cobalt
- carbon nitrides 260, 266 - diamond synthesis 498
charge coupled devices (CCD) 45 - hardness 948
chemical analysis, TM carbides/nitrides 210 f - transition metal diboride cermets 915
chemical bonding cobalt hard metals 238
- borides 803 cobalt-iron-system 488
- carbides 15 cobalt-tungsten-carbon system 5 16
- nitrides 25 coesite-stishovite phase boundary 55
chemical compositions see: compositions cold compaction 5 13
chemical properties cold isostatic pressing 65 1
- boron carbides 851 f, 855 f colloid size range growth, diamond 504
- silicon carbides 720 color
chemical vapor deposition (CVD) - boronnitride 510
- boronnitrides 421 - diamond synthesis 498
- carbide synthesis 210 - silicon carbides 690,7 19 f
- diamond films 390-419 - TM carbides/nitrides 224
- hydrothermal synthesis 374,386 combustion analysis
- nanostructures 116 f - diamond-like carbon 63 1
- silicon carbides 697,707 - TM carbides/nitrides 21 1
- thin films 463 combustion tubes
- transition metal borides 876 compact nitridation, silicon powders 749
- transition metal carbides/nitrides 24 1 compact tension test 84
chemical vapor deposition (CVD) diamond compaction homogeneity, alumina-based
XLVII, 484 ceramics 654
- applications 573-622 complex kinetics, corrosion 148 ff
chemical vapor infiltration (CVI) 3 10,717 composite ceramics, whisker-reinforced 669
chevron-nothed threepoint bend beam 84 composites
chlorides 208 - aluminum-magnesium 300
chlorination 374 - boron carbide-based 857 ff
chlorine reactions 855 - nanocrystalline 112
chromium - silicon carbides 7 10
- diamond synthesis 498 - transition metal carbides 880
- transition metal diboride cermets 915 compositions
chromium binder 923 f - cemented ternary borides 923
chromium carbides 17,208 - titanium boride-iron composites 901
chromium diborides 83 1 compounds
chromium-nitrogen system 219 - borides 809
cluster fragmentation, cemented ternary - isoelectronic 6 f
borides 922 - transition metals 203
clusters compressive strength
- alumina-based ceramics 653 - polycrystalline boron nitride 519
- carbon allotropes 272 - silicon carbides 7 18
Index 1001
compressive stress, sapphire 186 - titanium boride-titanium carbide system 880

concrete machining 540 ff crack growth rate, silicon nitrides 772 f
conducting target sputtering 430 crack propagation 67 f, 75 ff, 86
conduction band, nanostructures 132 cracking, directed metal oxidation 291,298
conductivity creep behavior 92 ff
- alumina-based ceramics 667 - alumina-based ceramics 667
- boroncarbides 854 - directed metal oxidation 313
- boron nitrides 421 - nanocomposites 121
- CVD diamond 412,582 - silicon nitrides 75 1,778
- data collection 965-995 creep feed grinding 537
- directed metal oxidation 289.3 16 crosslinking, polymer-ceramic transformations
- polycrystalline boron nitride 519 447
- silicon carbides 721 crushing
- silicon nitrides 75 1 - boroncarbides 838
- TM carbides/nitrides 203,225 - polycrystalline diamond 513
CONFLAT vacuum flanges 606 crystal structures 1-285
continuous fiber-reinforced silicon carbide matrix - borides 804
composites (CMCs) 717 ff, 739 - boroncarbides 854
controlled atmosphere, thin film processing 464 - boron nitrides 420 f
conversion process - carbon allotropes 277
- self-propagating synthesis 328 f - carbonitrides 258
- thinfilms 464 - cemented ternary borides 9 I9
copper - data collection 965-995
- diamond synthesis 488 - polycrystalline boron nitride 520
- directed metal oxidation 291,305 - silicon carbides 685 f
cordierite, silicon carbides 695 - silicon nitrides 753 ff
core-mantle composite model 110 - TM carbides/nitrides 204 ff
core drills 549 - zirconium/titanium borides 879
core-rim structures, transition metal diboride crystallite size effect, hydrothermal synthesis
cermets 918 316.382
cores, silicon carbides 688 crystallites
comer sharing - alumina-based ceramics 655,661
- carbides 19 - boride-zirconia composites 889
- nitrides 26 - diamond synthesis 400,503
- titaniumoxides 33 - silicon nitrides 759
corrosion 14&182 crystallization, diamond 485
- directed metal oxidation 3 16 cubic boron nitride XL-LXXI
- silicon nitrides 749,786 ff see also: boron nitride
corundum XL cubic diamond
- alumina-based ceramics 648 f, 653,658 - hydrothermal synthesis 374
- crystal stuctures 30 - synthesis 509
- silicon carbide based materials 683 cubic structures
- single phase sintered 670 - borides 809
Coulomb barriers 94 - boron nitrides 6,42@445
Coulombic forces, carbon/silicon-nitirides 254 - cobalt 949
covalent bonding XL, 253 - diamond Sf
- borides 803 - silicon carbides 685
- boron carbides 841 - silicon carbon nitrides 266
- boron nitrides 420 f - TM carbides/nitrides 205
- carbides 197 Cubitron 664
- hydrothermal synthesis 374 curing 448
crack branching 113 cutting edge displacement (CED) 674,679
crack bridging 96 cutting edges 658,666
- silicon carbides 695,710 f cutting tools
crack deflection - alumina-based ceramics 648 ff
- boride-zirconia composites 890 - carbonitrides 202f
- silicon carbides 7 10 f - cemented borides 927
1002 Index
- CVDdiamond 411,611ff density functional based tight-binding

- material selection 68 f (DF-TB) 273,276
- polycrystalline boron nitride 5 19 density of states (DOS)
- polycrystalline diamond 550 - carbon allotropes 279 f
- silicon carbides 736 - TM carbides/nitrides 206
- silicon nitrides 784 deposited layers, TM carbides/nitrides 241
- single crystal diamond 563 deposition mechanisms 66
- temperature effects 199 deposition methods
- TM carbides/nitrides 241 - boronnitrides 421 f
cyclohexadiene 273 see also individual types (CVD, PVD, etc.)
Czochralski growth 190 deposition temperature, diamond-like
carbon 633
design, superhard materials 109 ff
damage detector materials, CVD diamond 584
- creep 94f DIAFILM 575 ff, 583
- intergranular 199 diamond XL-LXXI, 5 ff
- silicon nitrides 780 f - alumina-based ceramics 648 f
- tolerance 67 - carbon allotropes 271 ff
data collection, hard material properties - classifications 498 f
965-995 - corrosion 154ff
DBC5O milling 561 - hydrothermal synthesis 374-389
De Hoff diagram 840 diamond abrasives, synthesis 491
Debye temperature 879 diamond anvil cell (DAC) 508
decomposition reactions, CO, laser heating - laser heating 41-65
54 ff diamond films 390-419
deep grinding 537 diamond incorporation 367
defects diamond materials 479-572
- borides 813 diamond single crystals 379
- boroncarbides 842 diamond synthesis 481 ff
- carbides 12 - carbonnitrides 257
- ceramic tool materials 65 1,662 diamond-like carbon films (DLC) 623-647
- silicon nitrides 771 f diboride systems 831
definded cutting edge 666 dielectric loss, CVD diamond 601
deformation XLI dielectric properties, CVD diamond 583
- indentation test 87 dies, diamond synthesis 490
- modes 76,94 f, 185 differential thermal analysis (DTA) 36, 151,446
- polycrystalline diamond 514 diffusion, directed metal oxidation 291
dense materials, self-propagating synthesis diffusion-controlled growth, silicon nitrides 763
342 ff diffusion-cooled CO, laser 596
dense shapes, silicon carbides 699 diffusion layers, TM carbides/nitrides 246 f
densification diffusional techniques, carbide synthesis 2 10
- boroncarbides 844 diffusivity 229 ff
- silicon nitrides 755 ff dilution effect, self-propagating synthesis 335
- titanium boride-iron composites 906 f diodes 606
- transition metal borides 876 ff direct coagulation casting (DCC) 654
density XLII direct current plasma jet deposition XLVIII
- alumina-based ceramics 654 directed metal oxidation (DMO) 92,289-321
- boroncarbides 854 dislocations 183, 188, 196
- CVDdiamond 601 - boroncarbides 842
- data collection 965-995 - creep 94
- diamond-like carbon 632 - CVD diamond 412
- polycrystalline boron nitride 519 - nanocomposites 121
- silicon carbides 695,718, 721 - transition metal diboride cennets 918
- titanium boride-iron composites 910 displacements
- TM carbides/nitrides 203 - control 77
- zirconium/titanium borides 879 - Voigt bounds 71
density effect, self-propagating synthesis 339 dissolution, nanostructures 130
Index- 1003
dissolution temperatures, diamond 380 - diamond-like carbon 637 f

distortions, CVD diamond 589,595 - silicon carbides 719
doping electrochemical applications, CVD diamond 415
- alumina-based ceramics 661 electro-corundum 658
- boroncarbides 852 electrolysis, transition metal borides 876
- diamond-like carbon 638 electron cyclotron heating (ECH) 598
- directed metal oxidation 305 electron cyclotron resonance (ECR) 422
- sp’ bonded carbon allotropes 271-285 electron energy loss spectroscopy (EELS) 63 1
double bonds, carbon allotropes 278 electron mobility, CVD diamond 414
double layers, directed metal oxidation 309 electron paramagnetic resonance (EPR) 379
double torsion test 84 electron probe microanalysis (EPMA) 2 11
dressers, CVD diamond 61 1,616 f electron spectroscopy for chemical analysis
dressing tools 562 (ESCA) 725
drilling electron transport, directed metal oxidation 290
- carbonitrides 202 f electronic applications, CVD diamond 413
- polycrystalline diamond 554 electronic properties, carbon allotropes 279
- polycrystalline ultrahard materials 549 electronic speckle pattern interferometry
dry grinding (ESPI) 593
- abrasives 532, 538 electroplated tools 529,532
- silicon carbides 704 Elektroschmelzwerk Kempten (ESK)
dry machining 6 14 process 688f
dry powder processing 65 1 element synthesis, silicon carbides 692
dry pressing 704 energy-dispersive analysis of X-rays (EDX) 119
drysawing 547 energy-volume relationship, carbon nitrides 264
ductile binders 68 engineering requirements 70
ductile-to-brittle transition enstatite 488
- self-propagating synthesis 356 enthalpy
- titanium carbide 235 - data collection 965-995
duplex layer, directed metal oxidation 292,298 - self-propagating synthesis 323
duplex techniques, titanium nitride coatings 342 enthalpy increase, diamond synthesis 505
epitaxial growth 291
equilibrium volumes XLII
q-carbides 18 erosion tests 8 1, 89 f
earth’s lower mantle materials 5 1 f Eshelby treatment 31 1
edge sharing etching
- nitrides 27 - diamond synthesis 404
- titaniumoxides 33 - silicon nitrides 770
edge toughness 722 - titanium boride-yttria stabilized zirconia
Einstein equation 130 composites 893
EkasicT 722ff eutectic concentrations, titanium boride-iron
elastic modulus 71 ff, 81 ff composites 909
- directed metal oxidation 3 10 eutectic reactions, borides 823
- SiAlON 460f evaporation
- silicon carbides 718 f - carbide synthesis 210
- siliconaluminocarbosilanes 454,457 f - diamond-like carbon films 627
elastic properties, TM carbides/nitrides 23 1 f even/odd phases, carbon nitrides 259
elastic recoil detection (ERD) 63 1 evolutionary selection, diamond synthesis 401
elastic recoil detection analysis (ERDA) 436 experimentals
electric discharge grinding 613 - CO, laser heating 44 f
electric resistivity - corrosion 1SO ff
- boroncarbides 854 - diamond anvil cell 43 f
- data collection 965-995 - self-propagating synthesis 33 1 ff
- zirconium/titanium borides 879 - SiAlON 460f
electrical conductivity 2 2 5 3 16 - silicon carbide-water system 383
electrical properties - siliconaluminocarbosilanes 452
- boron nitrides 441 - thin film processing 463
- CVDdiamond 413 - ncTiN/aSi,N, deposition 118
1004 Index
extended X-ray absorption fine structure fractional parameters, carbon/silicon-

spectroscopy (EXAFS) 264,446 nitirides 255
external corrosion 141 f fracture corrosion stress 158
extinction coefficient, boron nitrides 440 fracture strength 81 f, 601
extrusion, silicon carbides 704 fracture stress
- CVDdiamond 580
- nanostructures 105, 109
failure 67,94 ff fracture toughness 67 f, 74,83 f
- cemented borides 929 - alumina-based ceramics 649,659,663
fatigue, mechanical 95 f - borides 928,934
fatigue strength, silicon nitrides 75 1 - boron carbides 853 f, 865,870
feldspar XL - CVDdiamond 588
Fermi level - directed metal oxidation 289
- borides 808 - polycrystalline boron nitride 5 19
- carbon allotropes 274 - SiAlON 460f
- diamond-like carbon 637 - silicon nitrides 75 1,771
- nanostructures 132 - titanium boride composites 890,912
- TM carbideshitrides 206 fragmentation, directed metal oxidation 29 1
femtic binders, cemented borides 929 Frank-Kasper polyhedron 18
ferrous material machining 557 Frenkel excitations 638
fiber-reinforced directed metal oxidation fretting wear 656
composites 314 friction coefficients
fiber-reinforced plastics 61 I - boronnitrides 440
fibers - CVD diamond 618
- directed metal oxidation 307 f - diamond-like carbon 640 f
- silicon carbides 697,7 18 - nanocomposites 127
Fick diffusion 142 ff - silicon carbides 722 f, 732
field-activated self-propagating high temperature - silicon nitrides 783
synthesis 348 ff full-width-at-half-maximum (FWHM)
figures-of-merit, CVD diamond 413 - CVDdiamond 575
film adhesion 427 - diamond synthesis 405
fine grain, CVD diamond 580 - silicon carbide-water system 386
finish machining 613 f fullerenes
flaking 70 - carbon allotropes XLVI, LIII, 27 1 ff
flat faces, diamond synthesis 401 - complexes 253ff
flatness - hydrothermal synthesis 374
- CVD diamond windows 595 furan coke 505
- diamond submount 610 fused corundum 658
flaws 88,184 fused grinding materials 648 ff
- alumina-based ceramics 653,662
- creep 95
- nanostructures 111 f y-alumina 291
flexural strength 82 y-point sampling, carbon allotropes 276
- boron carbide ceramics 870 garnets, phase diagrams 56
- polyaluminocarbosilanes 456 gas corrosion
- silicon carbides 714, 718, 721 - boron carbides 163, 172
- titanium boride-iron composites 913 - silicon carbides 159
fluorescence spectroscopy 48 - silicon nitrides 167
fluorite XL gas fusion methods 735
flute grinding high-speed steel 541 gas phase diagnostics 395
Forschungszentrum Karlsruhe (FZK), CVD gas pressure 624
diamond 598 gas pressure sintered silicon nitrides
forsterite 488 (GPSN) 752,758
Fourier heat transfer 322,350 gas temperatures, diamond synthesis 392
Fourier transform infrared spectroscopy gaseous nitridation 290
(FT-IR) 36,437,464 gasification, graphite 307
- borides 811 gaskets, diamond synthesis 490
Index 1005
GefStoffV 736 graphyne 272

gelcasting 654 gray cast iron 669
gemstons ruby 30 green bodies, polyaluminocarbosilanes
General Electric process 692 452,458
generalized gradient approxiamtion (GGA) 255 green density
geometrical distortions, CVD diamond 595 - alumina-based ceramics 660
germanium 488 - self-propagating synthesis 360
gettering 498 - silicon carbides 700
GG25 cast iron 669 Griffith model 74 f, l l 0 f
Gibbs-Thomson equation 840 grinding
Gibbs free energy - alumina-based ceramics 648 f, 6.58 ff
- diamond synthesis 407 - boron nitride abrasives 530,538
- titanium boride-iron composites 902 - CVD diamond 613
Gibbs free enthalpy, nanocomposites 120, 124 f - diamond abrasives 530 ff
girdle devices 490 - dislocations 194, 199
glas-reinforced plastics 553 - hardness XLI
glassy phases 92 - silicon carbides 704,736
glow discharge deposition grits
- diamond-like carbon films 623 - alumina-based ceramics 658, 665
- nanocomposites 117 f - borides 802
grades group (IV-VI) elements 830
- CVDdiamond 574ff group IV carbide system 21 3
- silicon carbides 687 group IV nitride system 216
grain boundaries group IVB carbonitrides 202-252
- boroncarbides 841 group IVB transition metal carbide layers 241
- corrosion 141 group V carbide system 2 13
- deformation 183-20 1 group VB nitride system 21 7
- junctions 91 group VIB transition metal carbide system 21 3
- magnesia 292 group VIII elements
- nanostructures 110, 130 - borides 830
- silicon nitrides 75 1 - diamond synthesis 498
grain boundaries interlocking 886 growth
grain size 183-20 1 - boron nitrides 433 ff
- alumina-based ceramics 654 - CVDdiamond 580
- boron carbides 838,871 - diamond synthesis 392,407 ff, 492,496
- silicon nitrides 75 I - diamond-like carbon films 625
- tungsten carbides 950 f - directed metal oxidation 300 ff
grains - patterns 147 ff
- alumina-based ceramics 664 - silicon nitrides 760
- CVD diamond 575,580 gypsum XL
- polycrystalline diamond 514 gyrotron tube windows 583,597 f
granite block sawing 543
granules, alumina-based ceramics 65 1
graphite XLIV ff, 253 ff H3, carbon allotropes 273 f
- borides 813 hafnium nitrides 23 1
- boroncarbides 838 halides 876
- cemented borides 895 Hall-Petch relation 93
- diamond films 390 ff - nanostructures 110 ff
- diamond synthesis 486 ff - TM carbides/nitrides 235
graphite4iamond transformation 407 - tungsten carbide-cobalt hardmetal 948 ff
graphite reduction, transition metal borides 875 halogenated hydrocarbons 377
graphitic components, directed metal hard materials XLI-LXXI
oxidation 293 - structures 1-285
graphitic structures 259 hard tool ceramics 650
graphitization hardened steel machining
- diamond corrosion 155 - alumina-based ceramics 667,672 ff
- polycrystalline diamond 514 - CVD diamond 618
1006 Index
hardness XLI-LXXI, 79,84 ff high resolution transmission electron microscopy

- alumina-based ceramics 648-682 (HR-TEM) 119
- boride-zirconia composites 891 high speed finish machining 6 13
- borides 802 high speed steel (HSS) 541,555
- boron carbides 851,854,865 high temperature expansion, TM nitrides 230
- boron nitrides 421,440 high temperature hardness 934
- cemented ternary borides 925 high temperature properties
- cobalt 948 - silicon nitrides 777 ff
- CVDdiamond 410 - polyaluminocarbosilanes 457 f
- diamond-like carbon 632,639 high temperature strength 3 13
- polycrystalline boron nitride 5 19 highest occupied molecular orbitals (HOMO),
- SiAlON 460f carbon allotropes 279 ff
- silicon nitrides 75 1 ff, 775 f highly ordered pyrolytic graphite (HOPG)
- titanium boride-iron composites 910 416
- tungsten carbide-cobalt hardmetal 946-964 holemobility 414
- tungsten carbides 947 hole transport 290
- ultrahard materials 521 homogeneity
- zirconium/titanium borides 879 - alumina-based ceramics 654
Hashin-Shtrikman model 72 - corrosion 141f
health 73.5 Hookelaw 71,81
heat resistance 607 f - nanostructures 104, 108, 129
heat sinks 318 Hopkinson bar test 97
heated filament-assisted chemical vapor host lattice, TM carbides/nitrides 205
deposisiton (Hl-CVD) XLVII hot compaction 5 13 f
Hertzian contact damage 783 hot corrosion 151
Herzian indentation test 84,98 - boroncarbides 163
heteroepitaxial films 400,404 - boronnitrides 172
heterostructures, nanocrystalline 114 ff - directed metal oxidation 3 16
hexagonal-cubic transition 433 f - silicon carbides 157 f
hexagonal graphite - silicon nitrides 167 f
- diamond synthesis 508 hot filament chemical vapor deposition
- hydrothermal synthesis 374 (HF-CVD) 39 1 ff
hexagonal structures hot isostatic pressing (HIP)
- boron nitrides LVIII, 6,420 ff - boron carbides 847 f, 861
- cobalt 949 - self-propagating synthesis 345
- diamond 5 ff - silicon carbides 707,721
- silicon carbides 7,685 - silicon nitrides 166,749 ff, 778 f
- titanium borides 816 hot pressing
- TM carbides/nitrides 205 - boron carbides 847 f, 861
high frequency glow discharge 117 f - silicon carbides 705,721
high power gyrotron tubes 593 - silicon nitrides 749 ff
high power infrared lasers 589 - transition metal borides 877
high pressure decomposition 54 ff hot target sputtering boron nitrides 431
high pressure diamond synthesis 487 f hybridization XLIV
high pressure high temperature phase diagrams - diamond-like carbon 630
54 ff - nanostructures 104
high pressure high temperature (HPHT) synthesis - silicon carbides 686
- boronnitrides 420 hydroboric acid layers 855
- diamond XLVII,508 hydrocarbons
- hydrothermal 374,380 - CO, laser heating 60
high pressure laminates (HPL) 613 - diamond synthesis XLVIII, 489
high pressure microwave source reactor - hydrothermal synthesis 377
(HPMS) 398 hydrochloric acid 3 16
high pressure modifications, silicon nitrides 753 hydrogen, diamond synthesis 390 f, 395, 507
high pressure process, diamond synthesis 504 hydrogen-free amorphous carbon (ta-C) 627 f
high pressure stability, CO, laser heating 54 ff hydrogenated amorphous carbon (a-C:H)
high purity materials 838 623,630 ff
Index 1007
hydrothermal corrosion ion plating 624 f

- boroncarbides 162 ion sputtering 111
- silicon carbides 156 ionic bonding, boron nitrides 420
- silicon nitrides 166 ionic diffusion 291
hydrothermal synthesis, diamond 374-389 iron, diamond synthesis 498
iron binder 923
iron catalysis 41 1
icosahedral structures iron-nickel-system, diamond synthesis 488
- borides 809, 815 iron oxides/silicates, melting temperatures 5 1
- boroncarbides 8 isoelectronic compounds 6 f
ignition, boron carbides 838 isostatic pressing 704
ignition temperature 322,359 isothermal microwave plasma chemical vapor
ilmenite structures 57 deposition 397
immersion tests 150 isotopic labelling 409
impurities isotropic compounds 809
- CVDdiamond 412
- diamond synthesis 501
incubation Jagodzinski notation 7 f, 13
- corrosion 150 Japan Atomic Energy Research Institute
- directed metal oxidation 295 (JAERI) 602
indentation crack length method 910
indentation hardness 107
indentation size 183-201 kimberlite 377,488
indentation tests 80 f, 86 kinetics
- hardness XLI - corrosion 142ff
- SiAlON 460f - titanium boride-iron composites 906
inductive plasma emission (ICP) spectroscopy kinked faces 401
735 Kjeldahl analysis 21 1
industrial applications Knoop hardness XL, 86, 193
- diamondboron nitrides 527 - boronnitrides 440
- TM carbidestnitrides 238 ff - data collection 965-995
inelastic deformation 183 ff - nanostructures 107
inert atmosphere pyrolysis 452 f - polycrystalline boron nitride 5 19
infrared lasers 589 - SiAlON 460f
infrared seekers 583 - ultrahard materials 521
initiation process, directed metal oxidation Koehler model 1 14
289,295 Koks model 110
injection molding
- boroncarbides 846
- silicon carbides 704 laminate wood flooring manufacturing 613
injection molding machine parts 932 Langmuir probe 438
interface controlled growth, silicon nitrides 763 Lanxideprocess 917
interfacial energy, diamond synthesis 505 laser ablation
intergranular damage 199 - carbon nitrides 257
intergranular microcracks 91 - diamond-like carbon films 627
interlaminar strength 71 8 laser diode arrays (LDA) 606 ff
interlocking grain boundaries 886 laserheating 41 65
intermediate phases, titanium borides 8 16 laser-induced damage threshold (LIDT) 594
internal corrosion 141 f laser irradiation, carbide synthesis 210
international thermonuclear experimental reactor laser synthesis, silicon carbides 693
(ITER) 598 lasers, CVD diamond 589
interstitial compounds 15 lattice defects 185, 188
interstitials, TM carbides/nitrides 205 lattice dislocations 93
ion-ion core repulsion, carbon allotropes 276 lattice parameters
ion beam-assisted deposition (TBAD) 523,627 - boronnitrides 420f
ion energy distribution, diamond synthesis 406 - carbon allotropes 274 f, 277
ion irradiation, thin film processing 467 f - silicon nitrides 7.53
- TM carbidednitrides 203 magnesia see: magnesium oxides
- zirconium/titanium borides 879 magnesiothermic reduction 875
- data collection 965-995 magnesium
layers 241 - diamond synthesis 488
- corrosion 141 - directed metal oxidation 289 ff, 304
Lee-Gurland equation 955 magnesium oxides 779,787 f
Lehoczky model I 14 - crystal stuctures 31
Lely process 708 - directed metal oxidation 292
lifetime diagrams, cemented borides 929 ff magnesium oxides/silicates, melting
light covalent ceramics 522 temperatures 5 1
light emitting diodes 737 magnetron sputtering
line spread function (LSF) 578 - boron nitrides 430,435 f
linear rate constant, corrosion 144 - carbonnitrides 524
liquid media corrosion main group element nitrides 24 f
- boron carbides 161 main group element oxide ceramics 30 f
- boron nitirides 17 1 MAK (maximal zulassige Konzentration) value,
- silicon carbides 156 silicon carbides 736
- silicon nitrides 166 manganese 498
liquid phase sintering (LPS) manufacturing
- boroncarbides 845 - alumina-based ceramics 650
- cemented ternary borides 920 - polycrystalline diamond 5 13
- polycrystalline diamond 5 13,516 martensitic phases, carbides 16
- silicon carbides 705,709, 721,738 mass selected ion beam deposition (MSIB)
- silicon nitrides 756 - boronnitrides 422
- titanium boride-iron composites 899 - diamond-like carbon 632
liquid polymer infiltration (LPI) 718 f mass spectroscopy (MS) 36,446
liquid shaping, alumina-based ceramics 65 1 mass transport, directed metal oxidation 290
liquid silicon infiltration (LSI) 719 f material grades, CVD diamond 574 f
lithium 289 material properties 1-285
lithium aluminosilicate (LAS) 3 15 - boron carbides 851 ff
lithium doping - boronnitrides 421
- carbon allotropes 27 I - cemented ternary borides 923
- directed metal oxidation 304 - CVD diamond 410,574 ff
load control 76, 187 ff - data collection 965-995
local density approximation (LDA) 255 - diamond-like carbon 632
localized corrosion 141 f - ncMnN/Si,N, composites 119 ff
long range order (LRO) 16 - polycrystalline boron nitride 5 19
longer wavelength infrared (LWIR) band 584 - silicon carbides 719 ff
lonsdaleite XLII, 5 - silicon nitrides 753 ff, 771 ff
- carbon allotropes 273 - titanium boride-iron composites 909
- diamond synthesis 5 10 - TM carbides/nitrides 224 ff
- hydrothermal synthesis 375 - transition metal borides 878
loss tangent, CVD diamond 588,600 materials 477-995
low angle scattering 578 - corrosion 154ff
low pressure solid-state source (LPSSS) mean free path 949 f
- diamond synthesis 490,508 meandering 113
- hydrothermal synthesis 374 see also: cracking
low pressure sustained growth, diamond 504 measurement, corrosion 150 ff
lubrication, silicon carbides 722 f measurement models 193 f
mechanical grade, CVD diamond 575
mechanical properties-microstructure
machining relation 6 6 1 0 3
- alumina-based ceramics 649 ff, 660 mechanical properties
- diamond abrasives 540 ff - borides 802
- laminate wood flooring 613 - boron carbide ceramics 865
magic angle spinning nuclear magnetic resonance - boronnitrides 440
(MAS-NMR) 36 - cemented borides 923,928
lndex 1009
- data collection 965-995 - diamond synthesis 395,406

- diamond-like carbon 639 f methane hydrogen plasma sustained growth 504
- directed metal oxidation 3 10 methyl radicals 408
- self-propagating synthesis 327 ff methylhydroxylsiloxanes 463
- silicon carbides 720 Meyer hardness 85
- silicon nitrides 771 ff microanalysis, physical 21 1
- polyaluminocarbosilanes 456 microcracking
- titanium boride-yttria stabilized zirconia - boride-zirconia composites 890
composites 892 - boron carbide ceramics 873
- transition metal borides 878 - coatings 70
medium density fireboard (MDF) 553 - nanostructures 110, 113
melonophlogite 274 microcrystalline diamonds 379
melting, directed metal oxidation 295 microhardness XL, 84, 194
melting points 3 f - carbides 215
- borides 802,814 - data collection 965-995
- boroncarbides 854 - light covalent ceramics 521
- cemented borides 897 - TM carbides/nitrides 203,234 f
- creep 94 - tungsten carbides 947
- data collection 965-995 see also: Knoop hardness
- diamond synthesis 498 micropores, alumina-based ceramics 653
- high pressure 49 microstructure-mechanical properties relation
- polycrystalline boron nitride 520 66- 103
- self-propagating synthesis 331,342 microstructures
- silicon carbides 685 - cx-SiAlON materials 768 ff
- TM carbides 202 ff, 224 - alumina-based ceramics 656
- TM nitrides 224 - p-silicon nitrides 758
- zirconium/titanium borides 879 - boroncarbides 844
melts corrosion - boron carbide-silicon carbide ceramics 857
- boron carbides 163 - cemented carbonitrides 240
- boronnitrides 172 - diamond-like carbon 630 ff
metal-bonded tools 529,534 - directed metal oxidation 302 ff
metal borides, cemented 895 ff - Ekasic W 733
metal-boron interactions 808 - nanocrystalline 128
metal carbides - silicon carbides 706
- cemented borides 895 - silicon nitrides 749
- diamond-like carbon films 629 - titanium boride-iron composites 909 f
- transition metal borides 876 - tungsten carbide-cobalt hardmetal 946
metal containing amorphous hydrocarbon microwave-plasma chemical vapor deposition
629 f, 634 f (MWP-CVD) 391,397
metal hydrides 875 migration
metal matrix composites (MMC) - directed metal oxidation 304
- cemented borides 895 - high pressure melting 49
- CVDdiamond 611 - self-propagating synthesis 343
- polycrystalline diamond 553 military applications, CVD diamond 584
- silicon carbide whiskers 696 milling
metal oxidation, directed 289-321 - cast iron 670
metal oxide silicon field-effect transistor - single crystal diamond 559
(MOSFET) 415 mining bits, polycrystalline diamond 556
metallic binders, cemented ternary borides 919 modeling procedures
metallic impurities, silicon carbides 735 - carbon/silicon-nitirides 254 ff
metallic melts 855 - heat resistance 607
metallic nanoclusters 132 modifications, silicon nitrides 753
metals corrosion modulation transfer function (MTF) 578
- silicon carbides 156 moduli
- silicon nitrides 167 - carbon nitrides 263 ff
metastable processes, diamond synthesis 504 - elastic 71 ff, 81 ff
methane 60,208 - polycrystalline boron nitride 5 I9
- ultrahard materials 521 nickel 915
- zirconium/titanium borides 879 nickel binders 923 f
modulus of rupture (MOR) see: rupture modulus nickel-carbon phase diagram 491
Mohs hardness XL, 80, 107 nickel chromides 111
molding nickel doping 305
- boroncarbides 846 nickel-manganese system 488
- silicon carbides 704 nickel silicides 325
molding tools 932 niobium
molybdenum - diamond synthesis 498
- diamond-like carbon 634 - diamond-like carbon 634
- directed metal oxidation 291 niobium-nitrogen system 218
- transition metal diboride cermets 915 nitridation, aluminum 290
molybdenum boride-type structures 807 nitrides
molybdenum-iron-boron system 832 - bulkcomponents 69
molybdenum-nitrogen system 219 - corrosion 166 ff
molybdenum silicides 20 f, 330 - crystal structures 23 ff
monolithic ceramics 97 - data collection 965-995
monolithic components processing 450 - superlattices 114
morphology nitridosilicates 29
- boron nitride films 423 f nitrogen
- diamond synthesis 495 - diamond anvil cell 43
Mottlaw 637 - diamond synthesis 498 ff
mounting techniques, CVD diamond 605 - silicon carbide based materials 683
mullite nitrogen doping
- directed metal oxidation 293 - carbon allotropes 271,276
- silicon carbides 695 - diamond-like carbon 638
multi-anvil devices, diamond synthesis 49 1 nitrogen pressure
multiblade sawing 544 - nitrides 212
multilayer coatings 242 - silicon nitrides 753
multiphase hard materials 888 nitrogen-rich boron nitrides LXII
multiphase systems, transition metal diboride noble gases 43
cermets 916 nodule spacing, directed metal oxidation 305
multiphonon absorptions bands, CVD diamond nomenclature, crystal structures 7 f
577 noncutting applications
multiple cracking 29 1 - polycrystalline diamond 555
- single crystal diamond 564
nondestructive evaluation (NDE) 91
nanoceramics 709 non-equilibrium thermodynamics, diamond
nanocrystalline composites 116 ff synthesis 507
nanocrystalline diamond 379 nongraphitic carbons 505
nanocrystalline materials 110 f non-isothermal plasma deposition 400
nanoindentation 108 notch-beam method 910
nanostructured superhard materials 1 0 4 139 novolac type resins 843
nanotubes 271 nuclear magnetic resonance (NMR) 63 1
natural occurrence, silicon carbides 684 nuclear reaction analysis (NRA) 63 1
natural polycrystalline diamond 5 12 nucleation
Nd-YAG laser 41 - boron nitrides 433 ff
near cubic structures 266 - CVDdiamond 580
neck formation - diamond synthesis 400 ff, 492
- boroncarbides 849 - magnesia 292
- cemented ternary borides 922 - siliconnitrides 760
negative electron affinity (NEA) 415 numerical simulations, heat resistance 607
Newkirk process 289 ff
Nicalon
- polymer-ceramic transformations 448 occupational health 736
- reinforced lithium alumiosilicate 3 15 occurrence, silicon carbides 684
- silicon carbides 695 octahedral structures 205
Index 1011
odd/even phases, carbon nitrides 259 passive infrared windows 584

Ohm’s law 71 9 Pauling rules 33
onions, PDA diamonds 533 f
~ carbon 374 Pearson symbol 4,203
- diamond synthesis 507 Peierls-Nabarro stress 110
optical applications, CVD diamond Peierls barrier 359
575,583 ff, 589 Peierls stress 72
optical properties penetration depth, ballistic 79
- boronnitrides 440 periclase 302
- CVD diamond 575 f, 589 periodic bond chains (PBC) 401
- diamond-like carbon 637 f permittivity 598
- silicon carbides 719 perovskite oxides 59
orbitals 279, 808 perovskite structures 27
ordered defects 12 Petchlaw 92
organic compounds 60 ff petroleum coke 838
oriented globular growth 512 phase diagrams
ornamental layers 243 - borides 812 ff
orthorhombic structures - carbonitride systems 221 f
- borides 10 - CO, laser heating 54 ff
- cemented ternary borides 920 - silicon carbide- aluminum nitride 716
- titanium borides 8 16 - silicon carbides 685
Ostwdd-Volmer rules 489 - transition metal-nitrogen systems 216 ff
Ostwald ripening - transition metal carbides 15
- nanocomposites 123, 129 f - tungsten-carbon-cobalt system 239
- titanium boride-iron composites 91 1 phase equilibria, carbide systems 213 ff
Ostwald rule 686 phase properties, polycrystalline boron
output windows, gyrotron tubes 583 nitride 520
oversaturation, silicon nitrides 761 ff phase reactions, cemented ternary borides 920
oxidation phase segregation, nanocomposites 1 16, 125
- boron carbides 163,855,858 phase transitions, laser heating 4 1 4 5
- directed 289-321 phases, silicon nitrides 755
oxidation resistance 965-995 phenolic resins 842
oxide ceramics phosphorus doping 27 1
- bulkcomponents 69 physical boundary conditions, corrosion 142 f
- crystal structures 30 ff physical microanalysis, TM carbides/nitrides
oxides 965-995 211 ff
oxyacetylene 391 physical properties 66 f
oxycarbide composites 674 - boron carbides 851 ff
oxygen acetylene torch XLVIII - boron nitrides 421,440
oxygen attack, corrosion 141 - cemented ternary borides 923
oxygen contamination, titanium boride-iron - CVD diamond 410 ff, 574
composites 900 - data collection 965-995
oxygen pressures 159 - diamond-like carbon 632,637 ff
oxynitride solution 760 ff - polycrystalline boron nitride 5 19
- silicon carbides 695,719 ff
- silicon nitrides 753 ff, 771 ff
Palmqvist cracks 87 - titanium boride-iron composites 909
parabolic rate constant, corrosion 145 - transition metal borides 878 f
paracyanogen 104 - transition metal carbides/nitrides 203, 224 ff
Paris law 95 physical vapor deposition (PVD)
ParthC-Yvon rule 13 - boronnitrides 421
partial pressure, directed metal oxidation 291 - carbide synthesis 210
particle size, self-propagating synthesis 338 - silicon carbides 708
passivation - thin films 463
- directed metal oxidation 291 - TM carbides/nitrides 241
- spinels 295f n-n bonding 274
passive corrosion 141 f, 145 f pileup models 110
1012 Index
pitting 158 f, 786 Poole-Frenkel excitations 638

Planck formula 45 pore sizes, polycrystalline diamond 5 13
plasma-assisted chemical vapor deposisiton porosity 84,91
(PA-CVD) XLVII - boroncarbides 841
plasma decomposition carbonitrides LV - residual 184
plasma deposition 624 f - self-propagating synthesis 339,346
plasma-enhanced chemical vapor deposition - polyaluminocarbosilanes 455
(PE-CVD) 523 - tungsten carbide-cobalt hardmetal 953
plasma etching potentials, cemented borides 927
- diamond synthesis 404 powder processing
- silicon nitrides 770 - alumina-based ceramics 65 1,665 f
- titanium boride-yttria stabilized zirconia - carbides 12
composites 893 - directed metal oxidation 293
- titanium boride-iron composites 912 - silicon carbides 691
plasma vapor deposition 116 f precursors
plastic deformation see: deformation - diamond XLVIII
plastic indentation 108 - diamond-like carbon films 624
platelets - polymer-ceramic transformations 447,453 f
- Ekasic W 733 preforming, directed metal oxidation 307
- silicon carbides 696 preheating, self-propagating synthesis 366
platelets alignement 92 preparation
point defects - boron carbides 837 f
- boroncarbides 842 - carbonitrides LVI
- diamond svnthesis 501 - diamond synthesis 391 ff
poisoning agents, directed metal oxidation - diamond-like carbon films 623
290,307 - SiAlON 460
Poisson ratio 72 - polyaluminocarbosilanes 452 ff
- boron carbides 854,865 - transition metal borides 875
- CVDdiamond 588 - transition metal carbides/nitrides 207 f
- nanostructures 108 pressing 704
- polycrystalline boron nitride 5 19 pressure, directed metal oxidation 291
- TM carbides/nitrides 232 pressure conditions, diamond anvil cell 43
- zirconium/titanium borides 879 pressure effects, self-propagating synthesis 347
polishing, diamond abrasives 544 pressure-temperature conditions, diamond
polyaluminocarbosilanes (PAIC) 45 1 ff films 390
polybenzene 274,278 pressure-temperature diagram, diamond
polyborosilizanes 174 synthesis 493
polycarbosilane (PCS) pressure-temperature regime, CO, laser 42
- boroncarbides 845 pressureless sintering
- polymer-ceramic transformations 446 f - boroncarbides 841
- silicon carbides 698 - transition metal borides 876
- thin film processing 463 process-related properties, alumina-based
pol ycrystalline boron nitrides ceramics 654
421,512,518 ff, 611 processing 287-476
polycrystalline diamond 482 f, 512 ff, 611 processing defects 91
polycrystalline grits 648 processing zones, temperature effects 199
polycrystalline silicon carbides 685 production
polycrystalline superlattices 114 - silicon carbide nanoceramics 709
polycrystalline tungsten carbides 948 - silicon carbides 709,717
polyhedron structure, carbides 18 production routes
polymer to ceramic transformations 446-476 - alumina-based ceramics 650
polymeric networks, carbon allotropes 272 - boroncarbides 838
polymer-pyrolysis derived (PP) fibers 697 f - silicon carbides 688 ff
polyphase ceramic composites 289 ff protective scales, corrosion 141
polysilanes 446 f pseudo-hot-isostatic pressing 345 f
polysiloxanes 446 f pseudopotential approach, carbon/silicon
polytitanocarbosilane (PTC) 447 nitrides 256
Index 1013
purity rebound test 81

- boroncarbides 838 recession rate, corrosion 152
- diamond synthesis 501 reciprocating grinding 538
- silicon carbides 690 recrystallization
pyrolysis - boroncarbides 842
- amorphous materials 36 - silicon carbides 700, 721
- boroncarbides 843 REFEL process 290,294
pyrope glass 57 refractive index
pyrophyllite gaskets 490 - boron nitrides 421,440
- CVD diamond 588,599
- diamond-like carbon 633,639
QQC deposition 490,508 - silicon carbides 7 19
quality control, silicon carbides 734 ff refractories, directed metal oxidation 289
quantum tunneling 29 1 refractory carbides 3
quartz 377 refractory grade silicon carbides 688
Quasam 274 refractory materials, self-propagating
quaternary systems synthesis 334
- borides 818 refractory metal wires 392
- carbonitrides 222 ff refractory metals 634
- diborides 831 refractory nitrides 1 13
reinforced dies 490
reinforced grades, alumina-based ceramics
R8, carbon allotropes 273 f 649,668
radiation effects, CVD diamond 60.5 relative stability, carbon/silicon nitrides
radiofrequency diode sputtering 430 255, 263,268
radiofrequency glow discharge deposition 623 residual porosity 91, 184
rain drop impact, CVD diamond 584,588 residual stress 66
Raman spectra resin bond polishing head 545
- CVD diamond 412,575 resin-bonded tools, diamondboron nitrides 529
- diamond anvil cell 48 resins, boron carbides 842 f
- diamondfilms 394 resistance curves 78,89
- diamond XLVIII resistivity
- silicon carbide-water system 383 - boron nitrides 421
- thin films 469 ff - data collection 965-995
- tungsten carbides 165 - TM carbides/nitrides 203
Ramsdell notation 7 f - zirconium/titanium borides 879
random aggregates 5 12 Reuss-Voigt average 3 10
rare earth hexaborides 1 I Reuss model 72
rare earth nitridosilicates 29 rhenium filaments 394
reactants rhombohedral graphite 509
- diamond synthesis 408 rhombohedral graphitic structures, carbon/silicon
- self-propagating synthesis 342,366 nitrides 256
reaction-bonded aluminum oxide (RBAO) rhombohedral structures XLIV
290,293 f - boron nitrides LVIII, 420 f
reaction-bonded mullite (RBM) 293 - corundum 30
reaction-bonded silicon carbides 700,72 1 - silicon carbides 7,685
reaction-bonded silicon nitride (RBSN) rigidity 104
290 ff, 749-801 roadway sawing 546
reaction-boronizing sintering 920 Rockwell hardness X I , 85
reaction control, scale formation 145 - titanium carbide reinforced alumina 674 f
reaction-sintered silicon carbide (RSSC) - tungsten carbide-cobalt hardmetal 946
290,293 room temperature
reactive atmosphere pyrolysis 460 f - grain boundaries 195 ff
reactive plasma techniques 210 - silicon nitrides 771 f
reactor, microwave-plasma chemical vapor - Youngmodulus 234
deposition 39 1 ruby fluorescence scale 43
rebound hardness 107 rule-of-mixtures, elastic moduli 72
1014 Index
rupture modulus 72 self-sustaining high-temperature synthesis

- directed metal oxidation 3 10 (SHS) 207
- polyaluminocarbosilanes 456 semi-empirical approaches, carbon/silicon
rupture strength 926 nitirides 254 f
Rutherford backscattering (RBS) 21 1,395,46,4 semiconductor properties
rutile structures - bluediamond 500
- nitrides 26 - CVDdiamond 414
- titaniumoxides 32 - silicon carbides 7 I9
SH1 composites, titanium carbide reinforced
alumina 673 ff
o-bonds shaping techniques, silicon carbides 699,704 f
- carbon allotropes 27 1 shear modulus
- diamond-like carbon 630 - boron carbide ceramics 865
salt baths 210 - nanostructures 115
salt melts corrosion - titanium carbide 235
- silicon carbides 157 - zirconium/titanium borides 879
- silicon nitrides 167 shear stress 105
sample preparation, diamond anvil cell 42 shock wave production XLVlI
sand erosion 588 shock wave tests
sapphire 30, 186 ff, 597 ff - hydrothermal synthesis 374
saturation, silicon nitrides 761 ff - sapphire 186
sawing 542 - methane 60
SC24, carbon allotropes 275 short range order (SRO), carbides 16
scales shrinkage
- corrosion 141 - boron carbides 841
- directed metal oxidation 305 f - cemented ternary borides 92 1
scanning electron microscopy (SEM) 54 - polycrystalline diamond 5 14
- thin film processing 464 - silicon carbides 705
- cemented carbonitrides 240 - polyaluminocarbosilanes 455
- cemented ternary borides 926 - titanium boride-iron composites 906 f
- diamond 379,393,404 Si3N4,nanostructures 104
- silicon nitrides 777 f SiAlON see: silicon aluminum oxynitride
- titanium boride-iron composites 900 Sicalon 716
- titanium carbide-boron powder blend 862 see also: siliconcarbides
- tungsten borides 885 silica carbides, thin film processing 463
scanning tunneling microscopy (STM) 636 silica formers, corrosion 142
Schemer equation silica reactor tube 397
- crystal stuctures 36 ff silicides 20 ff, 965-995
- nanocomposites 119 silicon, diamond synthesis 404
Schrodinger equation 276 silicon aluminum oxynitride, pyrolysis 460 ff
scratch hardness XLI, 80, 89, 107 silicon aluminum oxynitride-bonded silicon
secondary ion mass spectrometry (SIMS) 21 1 carbides 700
- boronnitrides 434 silicon based materials LXV
- diamond synthesis 396 silicon borides 815
- thin film processing 464 silicon carbide-titanium carbide-titanium diboride
secondary neutron mass spectrometry composites 883 f
(SNMS) 211 silicon carbide A1,OC 716
Seebeck coefficient 854 silicon carbide-based hard materials 683-748
Seebeck elements 737 silicon carbide-carbon systems 717 f
seeding, alumina-based ceramics 652,661 silicon carbide metal matrix composites 614
selection criteria, diamond synthesis 401 silicon carbide-silicon carbide systems 717 f
self-propagating high-temperature synthesis silicon carbides
(SHS) 322-373 - alumina-based ceramics 648,668
self-propagating synthesis, boron - boron carbide ceramics 857
carbides 839 - corrosion 141, 156ff
self-scavenging reactions, diamond - crystal structures 6 f
synthesis 408 - directed metal oxidation 308 ff
- fibers 446 f snlvcnt\. diainond synthc\i\ 398 f
- hydrotherinal treatment 3x2 ff Soret effect, self-propagating rynthesis 341
- wlf-propugating synthehis 35 I +sp' bonding 274
- temperature effect\ I98 \p' bonded carbon allotropes 271-285
- thin film proce\\ing 3h3 sp' hybridization 104
silicon carbon nitrides 204 ff' space groups 4 ff
- corrosioii I73 f - borides 805, XI5
silicon carbon nitrogen slstem\ 253-270 - carbon nitrides 158
silicon containing hystems. polymer-ceramic - ccnicntcd ternary borides 9 I9
transformations 446 f - silicon nitrides 753
vlicon nitride ba\ctl hard materials 739--801 spacing, directed metal oxidation 30.5
silicon nitrides XL. 253-270 .;palling
- bulk component\ 69 - coatings 70
- corrosion I60 ff - corrmion 147
- cry\ial structures 23 sphalerite 5tructure 173. 5 10
- nanostructures I13 ff spin propagation 325, 336
silicon oxides 55 \pincl\
silicon sputtering 856 - crystal stuctures 31
polyaluminocarho~ilanes 452 ff - directed metal oxidation 292. 295 ff. 305 f
silicothcrinic reduction 875 - silicon nitrides 753
bingle crystal data I 8 8 spruy drying 65 1
single crystal diamond, applications 559 ff \pringh;ick. polyaluininocarbosilanes 452
sinsle layers. self-propagating \ynthesis 366 sputtering
single-phase ceramics X7X - boron carbides 856
single-phase sintered corundum 670 - boron nitrides 429 ff
single-point ignition. boron carbides 838 - carbon nitrides 524
sintered grinding materials 638 f. 658 ff - diamond-like carbon film\ 624 f
sintered pohder derived (SP)fibers 697 f - hydrotherinal synthesis 374
\intcrecl reaction-honded \ilicon nitride - ncNiCr 1 1 1
(SRBSN) 749-801 - thin films 463
sintered silicon carbides 700 ff. 721 ST12. carbon allotropes 273 f
sintered silicon nitrides (SSN) 749-801 stability
sin tering - alumina-based ceramics 667
- alumina ceramic\ I84 - cwbidcs 212
- alumina-based ceramic\ 6-5 1 - carbon nitrides 255, 263. 268
- borides X I S . 828 - CO, laser heating 54 ff
- boron carbides 839 ff - diamond 380
- cemented ternary boridcs 920 ff - perovjkite oxides 59
- polycrystalline diamond 5 12,s I 6 - silicon nitridcs 255. 268
- titanium boridc-iron compo\ites 899 stable processes. diamond synthesis SO4
- transition metal borides 876 f stacking faults 412
slide rings 737 stacking sequences XLV
sliding tests 785 - borides 805
sliding wcar 723.729 ff - boron nitrides 420
slip casting 700. 704. 739 - diamond 5 , 510
slip dislocations 95 - nitrides 2X
\mall angle diffraction (SAD) pattern. silicon - silicidcs 20
carbide-water system 385 - silicon carbides 7,686
small angle X-ray scattering (SAXS) 635 - TM carbidesinitrides 205
sodium 289 1 \tar of FZK 602
softening. nanostructures 1 I1 static properties, carbon allotropes 278
w - p e l derived corundum 66 I steel, elastic moduli 71
sol-gel technology 652 f steel machining 660,667 ff
solid-state laser3 31 \tepped faces, diamond synthesis 40 I
solid-state reactions. titanitnn horide-iron stiffiie5s 66
composites 009 JCB tr/.so: Young modulus
soliti-\tate sintering 703. 709. 72 I stishovite structures XL. 33. 5.5 526
1016 Zndex
Stone-Wales transformations 282 superlattices 114

stone machining 540 ff superplastic silicon carbide sintered body 710
strain energy release rate (SERR) 76 supersaturation, diamond synthesis 492
strength superstoichiometric carbides 127
- alumina-based ceramics 649 surface adsorption, diamond synthesis 493
- boron carbides 852,865 surface chemistry, polycrystalline diamond 5 16
- cemented ternary borides 926 surface damage 67
- CVDdiamond 580ff surface graphitization, diamond corrosion 155
- directed metal oxidation 31 1 surface states 183-201
- mechanical 66 f, 72 SYNDITE 025 553
- polycrystalline boron nitride 519 synthesis 287476
- silicon carbides 7 18 - boron carbides 838 f
- siliconnitrides 751, 771 - boronnitrides LX
- titanium boride-zirconia composites 890 - diamond XLVII, 504 ff
- titanium boride-iron composites 913 - polycrystalline diamond 5 12 f
strength-grain size relation, diboride - silicon carbides 692
composites 887 - TM carbides/nitrides 207 f
stress
- CVD diamond 610
- residual 66f talc gaskets 490
- silicon nitrides 776 talcum XL
stress intensity factor (SIF) 72,77,95 tantalum 498
stress rupture 94 tantalum carbides 12
stress-strain characteristics, nanostructures 129 tantalum filaments, diamond synthesis 394
stress testing 186 ff tantalum-nickel-boron system 832
Stribeck test data 729 ff tantalum-nitrogen system 219
strong tool ceramics 650 target materials, diamond-like carbon films 629
structural application, silicon nitrides 792 Tauc relation 638
structural chemistry, hard materials 3 4 0 technical applications
structures 1-285 - borides 813 f
- borides 804 - cemented ternary borides 919 f
- boroncarbides 854 - TM carbidednitrides 202 ff
- carbon allotropes 277 technical demands, alumina-based ceramics
- carbon nitrides 261 660,667
- cemented ternary borides 919 technical grades, silicon carbides 687
- cobalt 949 technical scale production, boron carbides 838
- diamond-like carbon 632 technological properties 965-995
- silicon carbides 685 ff temperature strength 92
- silicon carbon nitrides 265 temperatures 183-201
- TM carbides/nitrides 204 ff - boroncarbides 841
- zirconium/titanium borides 879 - CO, laser heating 45.48
see also: crystal structures - CVDdiamond 582
subgranular residual porosity 91 - diamond-like carbon 633
sublattice structures, borides 811 - dielectric constants 604
submicrometer cutting 666 - self-propagating synthesis 325 f
submount heat resistance 607 - silicon carbides 688
suboxides, boron LXV, 526 - silicon nitrides 765
substrate based CVD fibers 697 - tungsten carbide-cobalt hardmetal 953
substrate bias voltage 624 tensile fracture, silicon carbides 7 10
SuperFZK 601 tensile strength
superabrasive tools 479 ff - CVDdiamond 588
supercritical water 379 - silicon carbides 697,718
superequilibrium atomic hydrogen 507 ternary composites, borides 888
superhard materials 253-270 ternary systems
- nanostructured 104-139 - borides 818 ff
Superior Graphite Company, silicon - boron-metals 824 ff, 836
carbides 692 - phase diagrams 56 f
Index 1017
ternary transition metal carbides 213 - directed metal oxidation 295

ternary transition metal nitrides 221 thermorefractive effect, CVD diamond 589
Tersoff potential 255,261 thin films processing 446,463 ff
testing 80ff three dimensional carbon phases L
tetragonal structures, cemented ternary tight-binding
borides 920 - carbon allotropes 273,276
tetragonal zirconia 888 - carbon/silicon nitrides 255 f
tetrahedral amorphous carbon 428 time dependence, self-propagating synthesis 33 I
textural changes, melting temperatures 49,54 titanium
textured films 400 - diamond synthesis 498
thermal barriers, directed metal oxidation 3 18 - diamond-like carbon 634
thermal conductivity - directed metal oxidation 305
- alumina-based ceramics 667 titanium boride-iron composites 897
- boroncarbides 854 titanium boride-yttria-stabilized zirconia
- boron nitrides 421 composites 892
- CVD diamond 412,582 titanium boride-zirconium boride system 879
- data collection 965-995 titanium borides, physical properties 879
- directed metal oxidation 289 titanium-boron-nitrogen films 125
- silicon carbides 721 titanium-boron system 333,816
- TM carbides/nitrides 225 titanium carbide-boron system 825
thermal effects, CVD diamond 589 titanium carbide-reinforced alumina 673
thermal emission, silicon carbides 46 titanium carbides XL, 12
thermal expansion coefficient (CTE) - alumina-based ceramics 669
- alumina 69 - corrosion 174
- boroncarbides 854 - self-propagating synthesis 324,356
- CVD diamond 588,610 titanium-carbon-nitrogen system 222
- data collection 965-995 titanium carbonitrides, corrosion 174
- polycrystalline boron nitride 5 19 titanium-chromium mixed borides 833
- silicon carbides 718,721 titanium diboride-boron carbide system 880
- silicon nitrides 753 titanium diboride-nickel, cobalt composites 914
- TM carbides/nitrides 203,228 titanium diboride-titanium carbide system 880
- zirconium/titanium borides 879 titanium diboride-transition metal boride
thermal grade, CVD diamond 575 composites 883
thermal management, laser diode arrays 606 ff titanium diboride-based cermets 897 ff
thermal migration, self-propagating titanium diborides XL, 83 1
synthesis 343 - boron carbide ceramics 861
thermal properties, silicon carbides 720 - corrosion 176
thermal resistance, CVD diamond 607 f titanium dicarbides, silicon carbides 71 1
thermal shock titanium interlayers, diamond-like carbon
- CVDdiamond 587 films 641
- directed metal oxidation 3 13 titanium-iron-boron system 827
thermal shock resistance titanium-mol ybdenumxarbon-nitrogen
- silicon carbides 714 system 222
- silicon nitrides 749 titanium-nickel-boron system 827
thermal spikes, boron nitrides 427 titanium nitrides
thermal stress, CVD diamond 610 - corrosion 174
thermite process 838 - diamond synthesis 500 f
thermochemistry, diamond synthesis 407 ff titanium-nitrogen system 216,323
thermodynamic stability, diamond 380,504 titanium oxides, crystal stuctures 32
thermodynamics titanium silicides 330
- hydrothermal synthesis 378 titanium-tungsten-boron system 836
- self-propagating synthesis 349 titanium-tungsten-carbon-nitrogen system 222
- silicon carbide-water system 382 tools
- TM carbides/ni trides 2 12 ff - applications 648482
thermoelastic effect, CVD diamond 589 - bonded 528
thermogravimetric analysis (TGA) 36,446 - superabrasive 479 f
- corrosion 151 topaz XL
1018 Index
toroidal anvils 490 truing tools 562

toughness tungsten
- directed metal oxidation 31 1 - diamond-like carbon 634
- tungsten carbide-cobalt hardmetal 962 - transition metal diboride cermets 915
toxicology, silicon carbides 736 tungsten carbide-cobalt hardmetal 946-964
transformation processes, diamond tungsten carbide-nickel, bulk components 68
synthesis 509 tungsten carbides 12, 512
transformation toughening 92 - bulkcomponents 68
transgranular fracture, silicon nitrides 774 - cobaltmetals 238
transgranular microcracks 91 - corrosion 165
transition carbide systems 213 - diamond-like carbon films 629
transition metal alloys 488,498 - elasticmoduli 71
transition metal boride ceramics 874 ff - grain size effects 197
transition metal borides, crystal structures 9 - hardness 947
transition metal-boron system 8 17 - self-propagating synthesis 324
transition metal carbides 202-252 - thermal expansion 228
- corrosion 174 - Youngmodulus 233
- crystal structures 12 ff tungsten diborides 83 1
transition metal diboride cermets 915 ff tungsten filaments
transition metal diboride-silicon carbide - diamond synthesis 392
composites 881 ff - silicon carbides 697
transition metal diborides 71 1,861 ff tungsten nitrde-aSi,N, system 124
transition metal doping 305 tungsten silicides 20
transition metal nitrides 15, 202-252 turbostatic layers, boron nitride films 423
transition metal nitrides twins 183, 188, 196
- corrosion 174 - boroncarbides 849
- crystal structures 25 f type I/II CVD diamond 574
- nanostructures 113 ff type Ib/IIab diamonds 498
- wetting 131 type IIa single crystal diamonds 606
transition metal-nitrogen systems 216 f Tyranno fibers 383,697
transition metal oxides 32 f
transition metal silicides 20 ff
transition pressures, carbon/silicon nitrides 255 ultrahard materials 479 ff, 521 ff
transition temperatures, silicon carbides 159 ultrahigh pressure anvils 565
transmission electron microscopy (TEM) ultraviolet spectroscopy (UPS) 21 1
185,446,464 undefined cutting edges 658
- CVD diamond 576 f unit cell
- diamond XLVIII - borides 804
- diamond-like carbon 634 - carbonnitrides 258
- silicon nitrides 763 - crystal structures 8 ,2 0
- tungsten borides 808
transport properties, borides 803
transverse rupture strength vacancies 130
- cemented ternary borides 926 vacancy complexes 272
- directed metal oxidation 3 10 vacancy ordering 205
TRIBO 2000/-1, silicon carbides 729 ff vacuum arc evaporation 627
tribological properties vacuum hot gas extraction 21 1
- boronnitrides 440 vacuum sintering, boron carbides 847
- diamond-like carbon films 639,642 ff valence bonds 633
- silicon carbides 723 ff valence electron concentration (VEC) 206,234
trigonal structures 7 valence electrons 8 10
- prismatic 205 vanadium 634,924
triple layers vanadium borides 808
- directed metal oxidation 309 vanadium carbides 68,324
- self-propagating synthesis 366 vanadium nitrogen system 217
triple point, boron nitride 5 11 vapor-liquid-solid process, silicon carbide
tripling, tungsten silicides 21 whiskers 693
Index 1019
vapor-solid process, silicon carbide wetting

whiskers 693 f - borides 803
vapor phase deposition 42&445,692 - carbonitrides 202
vapor phase formation/condensation, silicon - cemented borides 895
carbide whiskers 693 - diamond-like carbon films 644
vapor phase reactions, titanium boride-iron - directed metal oxidation 307
composites 909 - self-propagating synthesis 347
vented tube furnace 838 - titanium boride-iron composites 900
Vickers hardness XLI, 86, 183-201 whiskers 92
- alumina-based ceramics 649,655 - alumina-based ceramics 649,668
- CVDdiamond 588 - silicon carbide-boron carbide ceramics 857
- data collection 965-995 - silicon carbides 693,736
- nanostructures 105 white corundum 658
- SiAlON 460f wide band gap 441
- titanium boride-titanium carbide system 880 Williams expansion 77
- tungsten carbide-cobalt hardmetal 947 wire sawing 542
viscosity, boron carbides 846 wood flooring manufacturing 613
vitrified bonded tools 529,532 work hardening model, nanostructures 110
voids working steels 658
- carbides 13 wurtzite XLIV
- nanostructures 110, 113 - berylliumoxide 31
- polycrystalline diamond 5 13 - boron nitrides LVIII, 6
- TM carbides/nitrides 205 - silicon carbides 685,715
Voigt bounds 71 Wykoff parameters 255
volatile chlorides 208
Volmer equation 488
X-ray diffraction (XRD) 9 1,446
- boron carbides 163
wafers, silicon carbides 708 - diamond 380,394
Wagner model 159 - diamond-like carbon 634
Warren-Averbach analysis 119 - self-propagating synthesis 353
water vapor corrosion 162 - silicon carbides 686
wear parts X-ray spectroscopy (XPS) 36,211
- cemented borides 932
- CVDdiamond 611,617f
wear properties 89 f yield stress XLI
- boronnitrides 440 Young modulus 7 1,80 f, 93
- CVD diamond 4 10 - boron carbides 85 I , 854, 865
- diamond-like carbon films 643 - boron nitrides 421
- directed metal oxidation 3 14,3 17 - cemented borides 928
- silicon carbides 722 f - CVD diamond 581 f, 588
- transition metal borides 878 - data collection 965-995
- transition metal carbideshitrides 242 - diamond-like carbon 639
wear resistance - diboride composites 886
- alumina-based ceramics 655,658,667 - directed metal oxidation 31 1
- silicon nitrides 75 I , 782 ff - nanostructures 110
- tungsten carbide-cobalt hardmetal 962 - polycrystalline boron nitride 519
Weibull modulus 73 , s I - silicon carbides 697,721
- CVD diamond 580 f, 590 - polyaluminocarbosilanes 4.58
- silicon nitrides 772,783 - titanium boride-iron composites 910
weight gain - TM carbides/nitrides 203,232
- boroncarbides 856 - zirconium/titanium borides 879
- directed metal oxidation 297 yttria-stabilized zirconia 892
weight loss yttrium-aluminum-garnet (YAG) 41
- silicon nitrides 789 f yttrium borides 805
- polyaluminocarbosilanes 455 yttrium oxides 787
wet grinding 532
Zeldovich criterion 336 zirconium
zeta potential 846 - diamond synthesis 498
Zhadanov notation 7 f - directed metal oxidation 305
zinc zirconium borides 879
- diamond synthesis 488 zirconium-boron system 8 17
- directed metal oxidation 289 - self-propagating synthesis 333
zinc sulfide window 584 zirconium-carbon-nitrogen system 222
zincblende structures XLIV zirconium corundum 658
- boron nitrides LVIII, 6,420 ff, 510 zirconium dioxide 35
- carbon nitrides LVI, 258 zirconium oxide/aluminumoxide/siliconoxide
- ultrahard materials 521 layers 309
zirconia 92 ZnSe windows 589 f

Handbook of Ceramic Hard Materials - 2000 - Riedel

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Handbook of Ceramic Hard Materials - 2000 - Riedel

Uploaded by

Copyright:

Available Formats

Ralf Riedel (Editor)

Structure and Properties of Ceramics

J. Bill, F. Wakai, F. Aldinger

Weinheim . New York . Chichester . Brisbane . Singapore . Toronto

Library of Congress Card No. Applied for

Deutschc Bibliothek Cataloguing-in-Publication Data:

C WILEY-VCH Verlag GmbH, D-69469 Weinheiin (Federal Republic of Germany), 2000

One of the clearest hierarchies in materials science and engineering is provided by

List of Contributors XXVII

Introduction: Novel Ultrahard Materials

Part I Structures and Properties

1 Structural Chemistry of Hard Materials

2 Phase Transitions and Material Synthesis using the C02-Laser

3 Mechanical Properties and their Relation to Microstructure

3.1.2 Bulk Components 68

4 Nanostructured Superhard Materials

5 Corrosion of Hard Materials

5.2 Corrosive Media 140

6 Interrelations Between the Influences of Indentation Size, Surface

7 Transition Metal Carbides, Nitrides, and Carbonitrides

8 New Superhard Materials: Carbon and Silicon Nitrides

8.3.4 Relative Stability 263

9 Effective Doping in Novel sp2 Bonded Carbon Allotropes

Part I1 Synthesis and Processing

1 Directed Metal Oxidation

1.7 Corrosion of Directed Metal Oxidation Composites 3 16

2 Self-propagating High-Temperature Synthesis of Hard Materials

3 Hydrothermal Synthesis of Diamond

4 Chemical Vapor Deposition of Diamond Films

4.3.2 Reactive Species in Diamond Chemical Vapor Deposition, the Role

5 Vapor Phase Deposition of Cubic Boron Nitride Films

6 Polymer to Ceramic Transformation: Processing of Ceramic Bodies

6.4 Results 453

Part 111 Materials and Applications

1 Diamond Materials and their Applications

1.3.4 Polycrystalline Cubic Boron Nitride 5 18

2 Applications of Diamond Synthesized by Chemical Vapor Deposition

2.6.3 Chemical Vapor Deposited Diamond Dressers 616

3 Diamond-like Carbon Films

4 Ceramics Based on Alumina: Increasing the Hardness for Tool

4.4.2 Carbide Reinforced Composite Ceramics Based on A1203 669

5 Silicon Carbide Based Hard Materials

6 Silicon Nitride Based Hard Materials

8 The Hardness of Tungsten Carbidecobalt Hardmetal 946

8.4.1 Relationship between Hardness and Toughness 962

9 Data Collection of Properties of Hard Materials

a indent size, half length of diagonal m

radius of radial crack m

strain energy release rate

adiabatic combustion temperature K or "C

3PB three-point bend

EDAX energy-dispersive analysis of X-rays

LAS lithium aluminosilicate

X-ray absorption near edge structure

* This is a revised version of the paper published earlier [ 1521.

Minerals/ Formula Mohs Microhardness,' Microhardness

Silicon carbide* Sic 26 20-38

where d is the diagonal of the square-based diamond pyramid of the Vickers

Material Crystal Vo Density B, Ecoh Refs

In Table 2, some experimental and calculated structural properties such as the

Carbon-based Hard Materials

- Direct Current (DC) Plasma Jet Deposition [52,53],and