Chawla

CERAMIC MATRIX COMPOSITES
SECOND EDITION
CERAMIC MATRIX COMPOSITES
SECOND EDITION
by
K. K. Chawla
University of Alabama at Birmingham
Birmingham, AL
U.S.A.
SPRINGER SCIENCE+BUSINESS MEDIA, LLC

Library of Congress Cataloging-in-Publication Data
Chawla, Krishan Kumar, 1942-

Ceramic matrix composites 1 by K.K. Chawla. - 2nd ed.
p.cm.
ISBN 978-1-4020-7262-8 ISBN 978-1-4615-1029-1 (eBook)
DOI 10.1007/978-1-4615-1029-1
1. Ceramic matrix composites. 1. Title.
TA418.9.C6 C42 2002

620.1 '4- -dc21
2002190808
Copyright© 2003 by Springer Science+Business Media New York

Originally published by Kluwer Academic Publishers in 2003
Softcover reprint ofthe hardcover 2nd edition 2003
Ali rights reserved. No part of this work may be reproduced, stored in a retrieval
system, or transmitted in any form or by any means, electronic, mechanical,
photocopying, microfilming, recording, or otherwise, without the written permission
from the Publisher, with the exception of any material supplied specifically for the
purpose of being entered and executed on a computer system, for exclusive use by
the purchaser of the work.
Pennission for books published in Europe: oennissions@wkap.nl

Permission for books published in the United States of America: permissions@wkap.com
Printed on acid-free paper.

Lead me from darkness to light
Brihadaranyaka Upanishad
Ch.II.3
To
Kanika, Anita, Nikhilesh, and Nivedita

for putting up with him and it, once more
Contents
Preface to the Second Edition

Preface to the First Edition
1. Introduction 1
2. Ceramic Matrix Materials 11
3. Ceramic Reinforcements 47
4. Processing 109
5. Interface 139
6. Micromechanics: Elastic, Thermal,
and Physical Properties 169
7. Mechanical Behavior: Load Transfer;
Monotonic, Cyclic, and Fatigue Behavior 205
8. Thermal Stresses 263
9. Interface Mechanics 291
10. Macromechanics 355
11. Applications 395
Appendixes 425
Indexes 431
PREFACE TO THE SECOND EDITION
Charles Dickens starts his A Tale of Two Cities with the following words:
"It was the best of times, it was the worst oftimes, it was the age of wisdom, it
was the age of foolishness, it was the epoch of belief, it was the epoch of
incredibility, it was the season of Light, it was the season of Darkness, it was
the spring of hope, it was the winter of despair ... " These words provide a good
characterization of the state of affairs in the field of Ceramic Matrix
Composites (CMCs) during the last decade ofthe twentieth century and the first
decade ofthe twenty first century. New continuous ceramic fibers have become
available and new innovative techniques of processing CMCs have been
developed. The Achilles' heel of monolithic ceramics is their extreme
brittleness. Since the publication of the first edition of this book, substantial
progress has been made in processing of continuous ceramic fibers and their
composites. A reinforcing phase (particle, short fiber or continuous fiber) can
improve the toughness of these materials without sacrificing the well known
advantages of ceramic materials, viz., enhanced wear resistance, hardness,
corrosion resistance, and high-temperature capability. A range of ceramic
matrix composites (CMCs) has been developed that combine a matrix material
with a reinforcing phase of different composition (such as alumina and silicon
carbide) or the same composition (alumina/alumina or silicon carbide/silicon
carbide). Major applications of ceramic matrix composites include cutting tools
and wear resistant parts. Other minor but significant applications include hot
gas filters, high pressure and heat exchangers, and gas turbine seals. On the
other hand, persistent problems include lack of toughness and poor oxidation
resistance at high temperatures. High cost (both raw materials and processing)
continues to be a major obstacle to widespread application ofCMCs. Another
major obstacle is the lack of reliability in current materials. Some niche
applications involving the use of CMCs have appeared but the ceramic engine
remains a distant goal. All this points to a belief! have held for a long time.
That belief is that, just as in the economics, in research also there is no such
thing as a free lunch. Invariably, it is a road that involves a painstaking and
persistent effort to study various problems involved and find solutions to them.
Along the way, there will be advances, some incremental, some more than that.
Eventually, there will be results to show. Research and development work in
the areas of CMCs will continue to be of paramount importance for the
foreseeable future. Notable progress has been made yet more work needs to be
done because the challenges are indeed daunting. My effort at bringing out the
second edition of Ceramic Matrix Composites is a step in that direction.
In this revised second edition, I have updated the information in regard to the
material systems for ceramic matrix composites, the reinforcements and
matrices, processing, properties, and applications of CMCs. I have brought it
up to date on the work done in the decade since the first edition came out. I
have added new sections and new figures in all the chapters. Based on inputs
I received from various faculty members and students, I have tried to clarify
concepts where needed. I am grateful to all these people who talked to me
personally or sent email messages in this regard. I am grateful to my program
managers at the US Office of Naval Research (primarily, Dr. S. Fishman and
later Dr. A.K. Vasudevan) who thought it fit to fund my research work in this
area. Other people who have been helpful, in tangible or intangible ways, to
make this second edition see the light of the day include: (in alphabetical order)
are: A. Boccaccini, K. Carlisle, N. Chawla, Z. Chen, C. Coffin, G. Gladysz, G.
Gouadec, M. Koopman, R. Kulkarni, B.A. MacDonald, D. Mendoza, A.
Mortensen, K.L.Murty, B.R. Patterson, B.V. Patel, J.J. Petrovic, J.M. Rigsbee,
S. Scarritt, H. Schneider, N. Stoloff, Y.K. Vohra, and A. Tobin. Finally I am in
an everlasting debt of gratitude to my family members, Nivi, Nikhil, Anita, and
Kanika for their forbearance. Anita and Kanika cheerfully bore all the burden
of producing a camera-ready typescript.
K.K. Chawla
Birmingham, AL
August, 2002
PREFACE TO THE FIRST EDITION
Materials science and engineering (MS&F,) is by its very nature an

interdisciplinary activity. Researchers from a wide variety of disciplines,
metallurgy, ceramics, physics, chemistry, mechanics, electrical and electronic
engineering, etc. can and do participate in the MS&E activities. The need and
desirability of such an interdisciplinary effort is understandable inasmuch as
advanced or high-performance materials are critical for any of the modem
industries. It is almost axiomatic that progress in any field (energy, building
materials, transportation, electronics, aerospace, electric power, consumer
products, etc.) depends on the availability of suitable materials having specific
characteristics. In this regard, let me quote from another work of mine:
It is a truism that technological development depends on advances in the field

of materials. One does not have to be an expert to realize that a most advanced
turbine or aircraft design is of no use if adequate materials to bear the service
loads and conditions are not available. Whatever the field may be, the final
limitation on advancement depends on materials (Chawla, 1987).
It is pertinent to quote from some other sources about a fundamental change

that is occurring in the materials field:
A fundamental reversal in the relationship between human beings and materials
[has occurred]. Its economic consequences are likely to be profound.
Historically humans have adapted such natural materials as stone, wood, clay,
vegetable fiber and animal tissue to economic uses. The smelting of metals and
the production of glass represented a refinement in this relationship. Yet it is
only recently that advances in the theoretical understanding ofthe structure of
physical and biological matter, in experimental technique and in processing
technology have made it possible to start with a need and then develop a
material to meet... (Clark and Flemings, 1986).
And:
... the classical model of materials application has been inverted. We once
sought applications for materials. We now have applications driving the
creation of materials. We now design materials for what we need (Press, 1990).
Designing materials for specific applications is, indeed, the underlying

philosophy of composite materials. The materials marketplace is increasingly
becoming a highly competitive arena where substitution of traditional materials
by engineered materials is the norm; quality and value added to the material as
xii PREFACE
well as the energy cost are critical in the final cost. A study by the US National
Research Council (1989) has emphasized in no uncertain terms the importance
of synthesis, processing, characterization and performance of materials for
success in the international marketplace. Add to this the ever-important public
demand for a clean and healthy environment, and one can easily realize how
important the whole material life cycle of a given component is. I wish to
emphasize the item of environmental impact of the new materials and
processes, including the recyclability of materials. Ultimately, the engineered
materials must last longer, reduce material waste, be more energy efficient. I
believe that composite materials can contribute to a safe and healthy
environment.
Although metals, ceramics, and polymers make the three legs of what might be
called the tripod ofMS&E, increasingly it is becoming evident that the lines of
demarcation between traditional disciplines such as metallurgy, ceramics and
polymers are getting quite blurry. Consider the following. Ceramic materials are
being made from polymeric precursors, metals are being produced with a glassy
rather than crystalline structure while semi-crystalline polymers are finding
commercial applications. This intermingling of materials is most evident in the
field of composite materials where one has the ultimate objective of tailoring
a material having a specific set of characteristics starting from components
having different characteristics. Tying together process and microstructural
control to the desired performance goals in the final component or product is
the ultimate goal. In fact it is now recognized on all hands that in order to meet
the diverse and exacting demands, materials must be engineered at every step.
In this regard, mother nature has an abundance of lessons for us. Materials in
nature are tailored over a very large spectrum of length scale, from atomic or
molecular level to micro to macroscopic dimensions. Some very interesting
examples of nature's work can be found in collagen-based composite materials
such as skin, cartilage, bone, sea shells, etc. Nature has designed these
composites for multifunctional applications requiring sometimes flexibility and
strength and at other times resistance to various environments.
The theme ofthis book is: processing, structure, properties and performance of
ceramic matrix composites. My definition of the ceramic matrix is rather broad
for the purposes of this book. It includes inorganic silica-based glasses,
crystalline ceramics, glass-ceramics, intermetallics, and that very special
material, in elemental form, called carbon. All of these have an implicit
unifying thread in that they are fairly high-temperature structural materials.
This, I believe, is the first dedicated text on the subject of ceramic matrix
composites. There are, of course, many conference proceedings or multi-author
books available on the subject, for example, Warren, 1991 and Mazdiyasni,
1990. In this book, I have excluded cement and similar building materials,
mainly because they are not high-temperature structural materials. For anyone
interested in the topic of fiber-reinforced cement-based composites, there are
books available on the subject (Balaguru and Shah, 1992; Benkur and Mindess,
1990).
xiii PREFACE
The plan of the book is as follows. After an introductory chapter, we first

examine ceramic matrix materials (Chapter 2) and the processing,
microstructure and properties of reinforcement (Chapter 3). Inasmuch as rather
dramatic developments have occurred in the area of ceramic fibrous
reinforcements, Chapter 3 is rather large. This is followed by processing of
ceramic matrix composites (Chapter 4). We examine the interface region in
CMCS, in some detail, in Chapter 5. Properties of ceramic matrix composites
are then examined in detail in Chapters 6 through 10. Chapter 6 describes the
micromechanical aspects of elastic, physical, and thermal properties. Chapter
7 gives a description of the mechanical behavior of composites: monotonic,
fatigue, and creep. Chapter 8 gives a thorough description of thermal stresses
in composites. The important subject of interface mechanics and the various
mechanisms that can be exploited to obtain improved toughess in a ceramic
matrix composite are discussed in Chapter 9. This is followed by a discussion
of laminated composites in Chapter 10. Finally, I discuss the various
applications and performance-related topics in Chapter 11. I have tried to
derive every important relationship not requiring very complex mathematical
treatments.
I have aimed this book at the final-year undergraduate and first-year graduate
students in materials science and engineering as well as the practicing engineer
or scientist. I have used portions of the material contained in this text for a
senior undergraduate course, a first-year graduate course, and in short courses
for engineers from the industry. The respohse was very heartening. I hope that
the book will also obtain the same response from a much broader audience.
Except for the very basic ideas of materials science and engineering, I have
assumed very little prior knowledge of any special kind on the part of the reader
that is required to follow the material. The book may thus appear to be rather
pedantic, at places, to the more experienced reader. I apologize for that. There
are plenty of references and suggested reading material for the reader who
wants to dig deeper. I have provided problems at the end of each chapter in the
hope that by solving them, the reader will add to her/his understanding of the
material in the text.
There remains the pleasant task of acknowledging a number of people without

whose valuable input, tangible and intangible, this work would not have been
possible. In alphabetical order, they are: A. Choudhury, M.K. Ferber, F.
Gerstie, J.R. Hellmann, J.C. Hurt, O.T. Inal, M.G. Jenkins, B.A. MacDonald,
T.A. Michalske, J.M. Rigsbee, P.K. Rohatgi, S. Suresh, K. Upadhya, and A.K.
Vasudevan. An immense debt of gratitude is owed to S.G. Fishman and R.
Pohanka, my program managers at the US Office of Naval Research, for their
understanding, patience, and encouragement over the years. I should also like
to thank my graduate students and post-doctoral research associates whose work
has contributed to my understanding of the subject matter. The ones who stand
out in this group are: J.-S. Ha, R. Venkatesh, R.U. Vaidya, and Z.R. Xu.
Portions of the text were read and commented upon by N. Chawla, V. Gupta,
E. Kroshe, T.A. Parthasarathy, R.U. Vaidya, and Z.R. Xu. I am truly thankful
xiv PREFACE
to these people for important feedback. I am also very thankful to the following
for their generous hospitality during my sabbatical in 1992-93: J. Cohen, Y.-P.
Chung, I. Daniel, K.T. Faber, M.E. Fine, S.P. Shah and J.R. Weertman (all at
Northwestern University, Evanston, IL), B. Ilschner and F .. Rezai-Aria (Ecole
Polytechnique, Lausanne, Switzerland).
Thanks are also due to my family members, Nivi, Nikhil and Kanika, for
understanding my compulsion to undertake such time-consuming ventures. That
is the intangible part. They also rendered more tangible help in sorting things
out, preparation of figures, copy-editing and indexing. As always, I can never
thank enough my parents, Manohar Lai and Sumitra Chawla, for their
selflessness, constant encouragement and inspiration. I have always considered
the book-writing assignments that I have undertaken as educational in nature.
This one was no exception. In this regard, I wish to record my appreciation of
N. Hancox and M. Dunn for inviting me to undertake this work.
REFERENCES
Chawla, K.K. (1987) Composite Materials, Springer-Verlag, New York, p. 1.

Clark, J.P. and Flemings, M.C. (Oct. 1986) Sci. Am., p. 51.
Press, F. (1990) Met. Trans., 21A, 1337.
National Research Council (1989) Materials Science and Engineeringfor the 1990s,
Nat. Academy Press, Washington, DC.
Warren, R. (ed.) (1991) Ceramic Matrix Composites, Blackie and Son, Glasgow.
Mazdiyasni, K.S. (ed.) (1990) Fiber Reinforced Ceramic Composites, Noyes Pub., Park
Ridge, NJ.
Balaguru P. and Shah, S.P. (1992) Fiber-Reinforced Cement Composites, McGraw-Hili,
New York.
Benkur, A. and Mindess, S. (1990) Fiber-Reinforced Cementitious Composites,
Elsevier Applied Science, London.
ABOUT THE AUTHOR
Professor Chawla obtained his B.S. from Banaras

Hindu University and his M.S. and Ph.D. degrees
from the University of Illinois at
Urbana-Champaign. He has taught and/or done
research at (in alphabetical order) Ecole
Polytechnique Federale de Lausanne (Switzerland),
Instituto Militar de Engenharia (Brazil), Laval
University (Canada), New Mexico Tech (USA)
Northwestern University (USA), University of
Alabama at Birmingham (USA), and University of
Illinois at Urbana-Champaign (USA).
He has published extensively in the areas of processing, microstructure, and

mechanical behavior of materials, in general, and composite materials, in
particular. Professor Chawla is author or coauthor of about 200 papers. His
other books include the following: Composite Materials: Science &
Engineering, 2nd Ed., Springer-Verlag, New York, 1998; Fibrous Materials,
Cambridge University Press, Cambridge, 1998; Mechanical Metallurgy,
Prentice-Hall, Englewood Cliffs, NJ, 1984 (co-author); Mechanical Behavior
of Materials, Prentice-Hall, Englewood Cliffs, NJ, 1998 (co-author);
Principios de Metallurgia Mecanica, Editora Edgard Blucher, Sao Paulo,
Brazil, 1981(co-author)
He has organized international symposia and edited books in the area do

composites and fibers. He is on the editorial boards of various journals. During
1989-1990, Professor Chawla served as a Program Director for metals and
ceramics in the Division of Materials Research, National Science Foundation,
Washington, DC. Among his other honors are: Eshbach Society Distinguished
Visiting Scholar Award (Northwestern University); US Dept. ofEnergy Faculty
Fellowat Oak Ridge National Lab.; Distinguished Researcher Award (New
Mexico Tech); Distinguished Alumnus Award (Banaras Hindu University). He
was the chair of Materials & Mechanical Eng. Department (UAB) during 1998-
200 I.Professor Chawla is a Fellow ofASM international.
A NOTE TO THE READER
In this text, I have followed the standard American usage. This will be
immediately evident in the spellings of certain words such as fiber, center,
behavior, etc. rather than fibre, centre, behaviour, respectively. In general,
compound words are not hyphenated. Prefixes such as co, pre, semi are closed
up with the word they modify. For example, pushout rather than push-out,
preexisting rather than pre-existing, etc. However, I have left the spellings
unchanged in the journal title in a reference or a book title in a citation. While
I have bowed to the American usage in the language of the text, I have
rigorously followed the Systeme International (SI) units. This stems from my
belief that sooner or later the American scientists, engineers, and industry will
be using SI units. The widespread use of SI units in the US has suffered a long
delay. The scientific merits of the SI units are overwhelming. If not the
scientific merits of the SI units, then the force of economic necessity will do the
job. By using SI units, I am hereby contributing my mite in that direction. I am,
however, fully cognizant that one should be able to convert from one system of
units to another. Hence, detailed information on this topic is given in Appendix
B.
Cfiapter 1
INTRODUCTION
A composite material is a material that has a chemically and/or physically

distinct phases distributed within a continuous phase. The composite generally
has characteristics better than or different from those of either component. The
matrix phase is the continuous phase, while the distributed phase, commonly
called the reinforcement phase, can be in the form of particles, whiskers or
short fibers, continuous fibers or sheet. Figure 1.1 shows the types of
composites based on the form of reinforcement. Oftentimes it is convenient to
classify different types of composites as per the matrix material characteristics,
viz., polymer matrix composites (PMCs), metal matrix composites (MMCs),
and ceramic matrix composites (CMCs). The reinforcement in any matrix
~
~ Lammate
Fig. 1.1 Types of composites based on the form of reinforcement.

2 INTRODUCTION
can be polymeric, metallic, or ceramic. Polymeric matrix composites containing

reinforcement fibers such as carbon, glass, or, aramid are quite commonly used
as engineering materials. Metals containing ceramic particles, whiskers, or
fibers (short or long) are also gaining in importance. The ceramic matrix
composites are the newest entrants in the composites field.
The choice of individual components, in regard to their synthesis and chemical

composition, that make up a composite, the processing required to produce that
composite, and the mechanical behavior of the composite are factors of obvious
importance to the ultimate performance of the composite. However, an
additional and a very important factor that enters in the evaluation of the
performance of a composite is the presence of an interface region between
matrix and reinforcement, which can occupy a rather large area. We discuss the
importance of interface in chapter 5 and elsewhere in the book. Suffice here to
say that a control of the interface characteristics is a key to designing
composites having an optimum set of properties. Figure 1.2 shows how the
ultimate performance of a composite is linked to these factors.
The subject matter of this book is ceramic matrix composites. The term
ceramics covers a wide variety of non-metallic, inorganic materials, which are
frequently processed at high temperatures. In view of the advances made in the
last quarter of the twentieth century, it is convenient to categorize the ceramic
materials into the following two classes:
(i) Traditional or conventional ceramics: These are generally in a monolithic

form. These include bricks, pottery, tiles, and a variety of art objects.
(ii) Advanced or high performance ceramics: These represent an improved class
of ceramic materials where, frequently, some sophisticated chemical processing
route is used to obtain them. Generally, their characteristics are a sensitive
function of the high quality and purity of the raw materials used. Examples of
these high performance ceramics include oxides, nitrides, and carbides of
silicon, aluminum, titanium, and zirconium.
Table 1.1 lists some important high performance ceramic materials. These
advanced ceramics show unusually high resistance to heat, chemicals, and wear.
These very characteristics, however, make them difficult to fabricate in a simple
and economical way. It is this class of high performance ceramics that are
commonly used in making ceramic matrix composites. Monolithic high
performance ceramics combine some very desirable characteristics, e.g., high
strength and hardness, excellent high temperature capability, chemical
inertness, wear resistance, and low density. They are, however, not very good
3 INTRODUCTION
Processing
Performance
of
Composite
Fig. 1.2 Performance of a composite is linked to some important factors : composition

of components, their mechanical behavior, processing, and the characteristics of the
interface between matrix and reinforcement.
under tensile and impact loading, and unlike metals, they do not show any
plasticity and are prone to catastrophic failure under mechanical or thermal
loading (thermal shock). We characterize this difference in the behavior of
metals and ceramics by saying that metals are forgiving while ceramics are not
forgiving. The forgiving nature of metals has its source in the high mobility of
dislocations in them, which allows them to deform plastically before fracture.
Table 1.1 Properties of some important ceramic materials
Material Young's Poisson's Coefficient of

Modulus Ratio !hermal Expansion
E (GPa) v a (]{).6 K- J)
SiC 420 0.22 4

AIP3 380 0.25 8
Cordierite 130 0.25 2
(2AI 20 3'MgO'5Si0 2)
Mullite 215 0.25 4
(3A120 3·2Si02)
Sodalime glass 70 0.23 9
4 INTRODUCTION
Plastic deformation being an energy absorbing process, the fracture process in

metals involves extensive energy dissipation. The absence of such an energy
dissipating phenomenon in ceramics causes them to fail in a catastrophic
fashion, i.e., makes them unforgiving. Understandably, some of the major effort
of the materials community in the field of structural materials over the last
quarter of the twentieth century has been directed toward incorporating a
variety of energy dissipating phenomena in the fracture process of ceramics,
i.e., imparting them a damage-tolerant behavior. Despite significant advances
in toughness enhancement of monolithic ceramics, it is my view that ceramic
matrix composites represent a quantum leap in that direction. In this book, we
shall explore this important topic in some detail.
1.1 CERAMICS VIS-A- VIS OTHER MATERIALS
It is instructive to compare the use or service temperature of different types of

materials. It is the maximum temperature at which a material can be used for an
extended period. Figure 1.3 compares the approximate service temperature
ranges of some important polymers, metals, and ceramics. Clearly, ceramics are
the only class of materials viable at very high temperatures. As mentioned
above, ceramics also have higher hardness, strength, and elastic modulus than
metals and polymers. They also have lower density, thermal expansion
coefficient, and thermal and electrical conductivity. In particular, the
characteristics of low density and low thermal expansion of ceramics assume
a great deal of importance in most all applications. Density and thermal
expansion coefficients of some materials of interest are shown in Figs. 1.4 and
1.5, respectively. The greatest drawback of ceramics vis a vis metals is their
extremely low fracture toughness, which in practice means that these materials
have a very low tolerance of crack-like defects. The symbol KrC' having the
units ofMPa,(m, called fracture toughness, is used to denote this characteristic
of materials. Table 1.2 gives some typical values. Note the characteristically
low values of fracture toughness of ceramics. One can try to minimize the
quantity of structural defects in ceramics and also ensure that they are smaller
than a certain critical size by careful control of the raw materials and
processing. However, any preexisting defects can grow to a critical size in
service because of mechanical and thermal loading, environment, etc.; and,
consequently, a catastrophic fracture ensues. Therefore, a critical need exists
for increasing the toughness of ceramics. Also of importance is the need to
improve the in-service reliability of ceramics. This has to do with the flaw size
and flaw distribution. Beyond a shadow of doubt, minimizing the pernicious
problem of catastrophic fracture in ceramics is a very important goal. Let us
5 INTRODUCTION
Polymers Metals Ceramics
Fig. 1.3 Service temperature limit (indicative) of polymers, metals, and ceramics.
examine this concept of fracture toughness and see how Mother nature might
have some interesting lessons for us in this area. As Hillig (1985) points out,
strong winds (50 -80 km/h) can snap trees and poles like matchsticks. A birch
tree, however, can get away with the loss of only a few leaves. The reason for
this is that its trunk is strong and resilient. Under the action of a gale, the tops
of young birch tree can bend as much as 45 0 • This high degree of resilience or
compliance in a birch tree results from the appropriate ply architecture that the
nature provides. This resilience is what we call toughness in a more scientific
parlance. The low toughness of ceramic materials also manifests itself in
another form, viz., strength of ceramics has a very high degree of sensitivity to
the presence and size of flaws such as cracks, voids, inclusions, etc.
Improving the toughness of ceramic materials is thus a major objective. One of

the important approaches to attain this goal is via ceramic matrix composites.
As should be clear from the discussion above, high performance ceramics must
have superior structural and/or mechanical characteristics because they find
applications in some very demanding environments, e.g., rocket nozzles, heat
exchangers, automobile engines, and cutting tools. Yet another important factor
6 INTRODUCTION
2°Jt---------------jkw
10 --------------- ~o
---------------- Cu
8 ________________ Fe
6 -- _____________ FeAI, NiAl
---------------- TiB2
---------------- Ti
4 ________________ AhOJ
________________ SiC
---------------- AI
2 ---------------- S~N4
--------------- C
- - - - - - - - - - - - - - - - Polymers
_ _ _ _ _ _ 0 '--_ _ _ _ _..... _ _ _ _ _ __
Fig. 1.4 Density of some materials.
is the cost of ceramics. Although, we shall not discuss this point in this book,
there is no gainsaying the importance of the cost factor. The great challenge is
to produce consistent and reliable ceramic components having superior
properties but without an excessive cost, i.e., they should be competitive on a
cost/performance basis with the materials they seek to replace. In this regard,
the use of some new processing methods (e.g., sol-gel, injection molding, hot
isostatic pressing, etc.) is very promising in our efforts to have an economical,
high volume production of less brittle ceramic materials.
There have been also major developments in the area of reinforcement

materials, continuous fibers, discontinuous fibers, whiskers, and particles.
Continuous ceramic fibers have become available commercially. Among the
reinforcements of ceramic matrices, one can include carbides, borides, nitrides
and oxides. They have high melting points, low density, high modulus, and high
strength. Similarly, there have been seminal advances in the processing of
ceramics and ceramic composites. We provide detailed accounts of these
developments in this book. Specific examples of reinforcements include carbon,
silicon carbide, titanium diboride, silicon nitride, and alumina. Common
ceramic matrix materials include various glasses, glass-ceramics, and ceramics
such as carbon, silicon carbide, silicon nitride, aluminides, and oxides.
7 INTRODUCTION
24 -------
-------------------------------------------
_____________________________ Sn Al
__________________________________________ ~n
20
16 ----------------------NiAl, Cu
::::::::::::Nf--------------~g()
12
------------------------------------- ~
------------------------------ Soda glass AW"I
8 -----------------------------------------------
---------- Ti '~3
------------.--------- TiEl2
:::::::::::: ~~::V::::::: _~1!!~!~ ___________ SiC

4 _________________________ Elorosilicate glass
----------------------------.Si3N4
_________________ Lithium alummosilicate glass-ceramic
____________________ C (parallel to the fiber axis)
o
-4
Fig. 1.5 Coefficient of thermal expansion of some materials.
Incorporation of fibers, whiskers or particles in a ceramic matrix can result in

a tough ceramic material. This happens because incorporation ofreinforcements
(fiber, whisker, or particle) introduces energy dissipating phenomena such as
debonding at the fiber/matrix interface, crack deflection, fiber bridging, and

fiber pullout, etc. If such phenomena can be made to operate in ceramic
materials, then we shall have introduced some toughening mechanisms and the
ceramic material would be expected to show a damage-tolerant behavior. In this
regard, proper control of the characteristics of the interface region is of obvious
8 INTRODUCTION
importance. Yet another point that the reader should note is that while the ratio
of the modulus of the reinforcement and the polymer or metal matrix is
generally between 10 - 100, this ratio for a CMC is rather low, and can
frequently be equal to one or even less. The high modulus ratio in PMCs and
MMCs allows an efficient load transfer from the matrix to the fiber via a strong
interface. However, unlike in PMCs and MMCs, the low modulus ratio in a
CMC means that the reinforcement and the matrix are not much different in
their load bearing capacity, i.e., a simple increase in strength of a ceramic
material is rarely the objective. The only exception to this is the case when in
a CMC we have a low modulus matrix material, such as a glass matrix.
Table 1.2 Typical fracture toughness (K Ic) values for some materials.
Material
Polymers
Polyethylene 1-2
Nylon 3
Epoxy, Polyester 0.5
Metals
100-300
Pure metals
(copper, nickel, aluminum)
Aluminum alloys 20-50

Titanium alloys 50-110
Low carbon steel 50
Cast iron 4-10
Ceramics
Sodalime glass 0.5-1
Magnesium oxide 3
Alumina 1-3
Silicon oxide 2-4
Silicon nitride 3-5
1.2 POTENTIALITIES OF CERAMIC MATRIX COMPOSITES
As I have said above, CMCs represent the latest entry in the field of
composites. They also represent a most formidable set of problems and very
9 INTRODUCTION
exciting potential applications. At the risk of repeating myself, I would like to

point out that the raison d'etre for most of the effort in the field of composite
materials is that they provide some unique combination of mechanical, thermal,
electrical, and other characteristics. This is clearly because there are many
situations where the monolithic or conventional materials just do not meet the
requirements, i.e., many important technological applications are limited by the
availability of materials. For example, very substantial gains in performance
can be attained in advanced aerospace propulsion and power systems if
structural ceramics could be employed.
Potentially, ceramic components can be introduced into automotive engines,

where the most important resultant gain would be a reduction in fuel
consumption. Other non-aerospace applications ofCMCs include dies and tool
bits, medical implants, and land-based power and transport engines. Heat
exchangers, in particular, represent an area of application where ceramic
materials can be cost-effective because of their high temperature capability.
One can use the waste heat from the furnace exhaust, for example, to preheat
the inlet combustion air and thus save fuel consumption. Such heat exchangers
can be used in industrial furnaces, gas turbines, fluidized bed combustion units.
Ceramics can also result in fuel efficiency in heat engines because of higher
operating temperatures, more compact design, and reduction or elimination of
cooling systems (Katz, 1985). The temperature requirement in such applications
is not as high as in aerospace applications. The extremes in the high
temperature requirements of aerospace materials applications are epitomized
by the reusable space vehicle which can withstand temperatures as high as 1500
°c or higher. Some applications in the field of aerospace, in particular, requires
material capable of withstanding temperatures in excess of 2000 °c. For
example, the concept of a conventional plane taking off from a conventional
runway and going into an orbit in space represents an enormous materials
challenge. Lightweight materials with high strength at high temperatures
represent a sine qua non condition. Military applications include armor,
radomes, etc.
Other structural part applications, besides the heat exchangers mentioned

above, include automotive engine components, turbochargers, etc. Wear parts
such as seals, nozzles, pads, liners, grinding wheels, brakes, etc. are some of the
common places where ceramics find applications along with the traditional
tungsten carbide or specialty steels. Carbon fiber reinforced carbon composites
are being used in aircraft brakes (more about it in Chapter 11). As inserts for
some metal turning and milling operations, ceramics have proved to be better
than the tungsten carbide based tool inserts. This is because the ceramics retain
their high hardness and strength to temperatures in excess of 1000 °c. The only
10 INTRODUCTION
characteristic in which the tungsten carbide/cobalt tool inserts excel the ceramic
tool inserts is toughness. Bearings made of advanced ceramics have been used
in missiles. The primary material in this area is silicon nitride. Materials
characteristics such as low density, high temperature capability, corrosion
resistance, and the ability to operate with little lubrication are desirable in
bearing applications.
REFERENCES
Hillig, W.B. (1985) in Tailoring Multiphase and Composite Ceramics, Plenum Press,
New York, p. 697.
Katz, R.N. (1985) Mater. Sci. & Eng., 71, 227.
SUGGESTED READING
Green, D.J. (1998) An Introduction to the Mechanical Properties 0/ Ceramics,

Cambridge University Press, Cambridge.
Kingery, W.D., H.K. Bowen, and D.R. Uhlmann (1976) Introduction to Ceramics, 2nd
ed., John Wiley, New York.
Meyers, M.A. and K.K. Chawla (1999)Mechanical Behavior 0/ Materials, Prentice
Hall, Upper Saddle River, NJ.
Scholze, H. (1991) Glass, Springer-Verlag, New York.
Strnad, Z. (1 986)-Glass Ceramic Materials, Elsevier, Amsterdam.
Wachtman, J.B. (1996) Mechanical Properties o/Ceramics, John Wiley, New York.
Cliapter 2
CERAMIC MATRIX MATERIALS
In this chapter we briefly review the basic characteristics of ceramics

(crystalline and noncrystalline) such as structure and properties, followed by a
description of some conventional and nonconventional processing techniques
for glass and ceramics. Finally, we examine the characteristics of some
important ceramic matrix materials, to wit, glass, silicon carbide, silicon nitride,
alumina, glass-ceramics, sialons, intermetallics, and carbon. Clearly, this will
be but a cursory review of a very large field, but it should provide the reader
with some relevant information on important ceramic matrix materials. For
more details, the reader should refer to some of the Suggested Reading at the
end of the chapter.
2. 1 BONDING AND CRYSTALLINE STRUCTURE
Ceramics exist as both crystalline and noncrystalline compounds. Glasses, a

subclass of ceramic materials, are noncrystalline, while ceramics other than
glasses have a crystalline structure. Sometimes some elemental materials such
carbon and boron can also be included among ceramics. Metals, on the other
hand, consist of individual elements rather than compounds. In crystalline
ceramics, stoichiometry dictates the ratio of one element to another in a
compound, e.g., iron oxide (FeO), silicon carbide (SiC), silicon nitride (Si 3N4 ),
etc. Nonstoichiometric ceramic compounds, however, occur frequently. We
generally designate these as Fe 1_x 0, where x is less than one. Table 2.1 lists
some important ceramic materials, some of which are good candidates for
ceramic matrix materials in composites. Table 2.2 gives physical and
mechanical parameters of some ceramic materials. The reader should take these
values as indicative rather than absolute.
Ceramics have mostly ionic bonding and some covalent bonding. In ionic
bonding there occurs a transfer of electrons between atoms that make the
compound. Generally, positively charged ions balance the negatively charged
12 CERAMIC MATRIX MATERIALS
Table 2.1 Some important ceramic materials.
Single Oxides Alumina (AI 20 3)

Zirconia (Zr0 2)
Titania (Ti0 2)
Magnesium oxide (MgO)
Silica (Si02)
Mullite (3AI2 0 302Si02)
Spinel (MgOo AIP3)
Carbides Silicon carbide (SiC)
Boron carbide (B4C)
Titanium carbide (TiC)
Nitrides Boron nitride (BN)
Silicon nitride (Si3N4)
Intermetallics Nickel aluminide (NiAI, Ni3AI)
Titanium aluminide (TiAI, Ti3AI)
Molybdenum disilicide (MoSi 2)
Elemental Carbon (C)
Boron (B)
ions to give an electrically neutral17 compound, for example, NaCl, where Na+
ions balance the Cl- ions. In covalent bonding, the electrons are shared between
atoms. The characteristic high strength as well as brittleness of ceramic
materials can be traced to these types of bonding which make the PeierIs-
Nabarro potential very high, i.e., inherent resistance to dislocation motion in the
lattice is very high. Thus, unlike metals, a stress concentration at a crack tip in
a crystalline ceramic cannot be relieved by plastic deformation, at least not at
low and moderate temperatures. This has led to attempts at toughening ceramics
by means other than large scale dislocation motion.
2.2 COMMON CRYSTAL STRUCTURES IN CERAMICS
Most the ceramic crystal structures are variants of face centered cubic (FCC)
or hexagonal close-packed (HCP). Generally, metallic cations are smaller than
the nonmetallic anions. Thus, in crystalline ceramics, the metallic cations
occupy interstitial positions in a lattice made of nonmetallic ions.
Common crystal structures in ceramics are:

Table 2.2 Physical and mechanical characteristics of some ceramic materials.
Coefficient
Young's a/Thermal Fracture
Density, Melting Modulus, Expansion, Toughness,
p, Point, E, a, K/c,
glcm 3 °C GPa 1O-6/K MPam'h
Al20 3 3.9 2050 380 7-8 2-4

SiC 3.2 420 4.5 2.2 - 3.4
Si3N4 3.1 310 3.1 2.5 - 3.5
MgO 3.6 2850 210 3.6
Mullite 3.2 1850 140 5.3 3.5 - 3.9
Borosilicate 2.3 60 - 70 3.5 0.5 - 2
Glass
Soda-lime 2.5 60 - 70 8.9 0.5 - 2
Glass
1) Simple cubic: This cubic structure is also called cesium chloride (CsCI)
structure. Examples are CsCI, CsBr, CsLIt is not as common as other structures
listed below.
2) Close-packed cubic: This is a variant of the Face Centered Cubic (FCC)
structure and is also called sodium chloride (NaCl) structure. Examples ofthis
structure include CaO, MgO, MnO, NiO, FeO, BaO, etc. Oxygen ions occupy
the face centered positions while the metal ions occupy the interstices.
3) Hexagonal close-packed: Examples of this structure include AIP3' Fe20 3,
Cr20 3 , etc.
Figure 2.1 shows some of these common crystal structures in ceramics.
2.3 NONCRYSTALLINE CERAMICS
Noncrystalline or amorphous (i.e., without form) ceramics are not really solids
but supercooled liquids. Now, liquids flow under their own mass. However,
they can become very viscous at low temperatures. Very viscous liquids (for
Simple cubic (CsCI) type Face-centered cubic (NaCI type) Hexagonal close-packed
Fig. 2.1 Common crystal structures in ceramics.
example, honey in the winter time) behave solidlike although they maintain a
disordered structure characteristic of a liquid, i.e., they do not undergo a
transformation to a crystalline structure. Thus, noncrystalline ceramics such as
glasses may, in many respects, behave like solids but structurally they are
liquids.
As we mentioned above, glasses are, microstructurally, not solids, but very

viscous, supercooled liquids. They can also be regarded as inorganic
thermoplastics inasmuch as they soften when heated. It should be pointed out
that organic thermoplastics can have a large fraction of crystalline phase. The
major advantage of any silica-based glass is its ease of fabrication, which
allows processing by melt infiltration and compression molding to be used.
Glass has a low elastic modulus, generally less than that of other common
reinforcements. It also has a low strain-to-failure and toughness. Table 2.3 gives
some important varieties of glass and their characteristics.
When a liquid is cooled, molecular or atomic rearrangement occurs such that

a closer packing of atoms results. Figure 2.2 shows a plot of specific volume vs.
temperature for a glassy and a crystalline material. In the case of a crystalline
material, there occurs a precipitous drop in volume (as well as change in other
properties) at the melting point, Tm while in the case of a glassy material a
gradual change occurs in the slope of the curve at the glass transition
temperature, Tg • At temperatures below the glass transition temperature, Tg, the
supercooled glass becomes rigid and no further rearrangement of atoms or
molecules occurs. Thus, glass is an amorphous or noncrystalline solid with a
super cooled
liquid
'"
E
o"
>
-="
·u
Q)
c.
Vl
T,
Temperature ---~
Fig. 2.2 Specific volume vs. temperature for a glassy and a crystalline material.
frozen-in structure of a liquid. However, unlike true liquids but like solids, it
shows resistance to shear forces and shows a Hookean behavior, i.e., strain
produced is linearly proportional to the applied stress at low temperatures.
Silica-based glass is a generic term representing an interesting and versatile

class of materials. Glasses of various compositions can be obtained and they
can show a range of properties. Structurally, however, all silica-based glasses
have the same basic building block: a tetrahedron made up of four large oxygen
ions with a silicon ion at the center of the tetrahedron, Fig. 2.3(a). The
composition of a single tetrahedron is Si04 . Each oxygen ion is, however,
shared by two tetrahedra, giving the bulk composition of Si02 • It is this basic
building block that is repeated in three dimensions in silica and silica-based
materials. Different repeat patterns can result in different structures for the
same composition. Thus, quartz is a crystalline form of pure silica while
ordinary glass, which is amorphous, has a random network of silica tetrahedra.
Figure 2.3 (b) shows such a network of silica tetrahedra.
As pointed out above, glasses are, in terms of their microstructure, not solids
but very viscous, supercooled liquids. They can be regarded as inorganic
thermoplastics inasmuch as they soften when heated. One should point out that
organic thermoplastics can have a large fraction of crystalline phase. The major
advantage of any silica-based glass is its ease of fabrication, which allows
processes such as melt infiltration and compression molding to be used. Glasses
have low elastic modulus, a low strain-to-failure and toughness. Table 2.3 gives
some important varieties of glass and their salient characteristics.
0 ........ 0 ' -
• ........ Si..
(a)
(b)
Fig. 2.3 (a) A tetrahedron made up of four large oxygen ions with a silicon ion at the
center; (b) a network of silica tetrahedra.
2.4 GLASS-CERAMICS
Glass-ceramics represent a special class of ceramics. They are polycrystalline

materials having fine ceramic crystallites in a glassy matrix. The amount of
crystalline phase can be as large 95-98% by volume. The grain size of the
crystalline phase is generally very fine, less than 1 ~m. Nucleating agents such
Table 2.3 Some important types of glasses.
Glass Softening point Density, p, Toughness, KJv

°C gcm- 3 MPavm
Fused Quartz 1580 2.6

(Silica)
96% Silica (Vycor) 1500 2.5
Soda-lime 700 2.4 0.7
(Window glass)
Borosilicate 825 2.3 0.8
(Pyrex)
as TiO z or ZrOz are introduced during the melting operation in the glass to give
a controlled crystallization that results in a glass-ceramic. Two types of
nucleating agents can be used:
(i) Agents having low solubility in glass (e.g., Pt, Cu, Au, Ag.)
(ii) Agents having high solubility in glass (e.g., Ti0 2, Zr0 2 , pps; 1-20 wt.%)
The process of making a glass-ceramic involves heating the appropriate glass

composition to the molten stage, forming the article or the component,
annealing, and then performing the appropriate heat treatment to convert to
glass-ceramic, Fig. 2.4. On heat treating the amorphous glass, small crystallites
(about 1 nm) nucleate and grow until they hit adjacent particles. On further heat
treating, angular crystallites form.
Titanium dioxide is commonly used as a nucleating agent in lithium

aluminosilicate, Li20-AI2 0 3-SiOz; more commonly known as LAS. Ti0 2
precipitates nucleate in the glass matrix when heated for 1.5 hours at 780°C.
The temperature is then raised to 950°C when the glass matrix crystallizes, the
crystallization process starting at the nucleating Ti0 2 precipitates. The
transformation of glass into a crystalline (96-98%) phase occurs with little
change in shape.
Some of the important glass-ceramic materials are described below:
The important crystalline phases in this system are B - spodumene (Li 20' Al z0 3 '
4Si02), B-eucryptite (LiP ·A120 3 ·2SiOz) or solid solutions of one or both of
1
I
!
Time •
Fig. 2.4 Schematic of processing of a glass-ceramic.
these. The crystallite size is very fine: 30 - 40 nm. Thermal shock resistant
glass-ceramics having characteristically low coefficient of thermal expansion
are frequently made from this system. Lithium aluminosilicate (LAS) with near
zero coefficient of thermal expansion is used for optical and telescopic mirrors.
Corningware, a line of thermal shock resistant cookware, is based on ~
- spudomene (Li0 2 - AlP3 - nSi0 2 or LAS) glass-ceramic. Other trade names
for such glass-ceramic materials include Zerodur and Ceran.
Many stable crystalline phases are possible in this system: cristobalite,

tridymite, cordierite (2MgO·2AIPz·SSiO z), enstatite (MgO·Si0 2 ), and mullite
(2Si0 2·3AIP3)·
These are generally hard, abrasion resistant, and transparent to microwave

frequencies at high temperatures. Radar antennae and radomes for aircraft are
common applications.
These are called fluormicas because they contain fluorine and they have a
layered silicate, "mica", as the main crystalline phase. The easily cleavable
mica phase makes for easy machinability of this glass-ceramic.
In this glass-ceramic, wollastomite (CaO·Si0 2) and anorthite (CaO·

AI20 3 ·2Si02 ) are the main crystalline phases.
2.4.1 Mechanical Properties of Glass-Ceramics
Glass-ceramics can be regarded as composites ofglass and crystalline ceramics.

Thus, important parameters affecting their mechanical properties are:
(i) Size and amount of the crystalline phase.

(ii) Strength of the interface between the glassy and ceramic phases.
(iii) Elastic modulus mismatch between the glassy and ceramic phases.
(iv) Thermal expansion mismatch between the glassy and ceramic phases.
There are certain desirable features in glass-ceramics from a mechanical

property viewpoint. For example, one would like to have no voids or porosity
and a fine and uniform crystal size in the final product. A fine grain size is
sought for strength and toughness while a uniform grain size is desirable
because any exaggerated grain in the midst of a generally small grain size
constitutes a serious flaw in a ceramic material. A comparison of strength and
modulus of glass-ceramics and some other common materials is presented in
Table 2.4.
Table 2.4 Mechanical properties of glass-ceramics compared to some other

materials.
Material Flexural strength Young's modulus

MPa GPa
Glass 55-70 70
Glass-Ceramics 70-350 80-140
AI2 0 3 200-50 280-380
Steel 300-1400 210
2.5 CONVENTIONAL PROCESSING OF CERAMICS AND GLASSES
There are various ways of producing glasses and ceramics. In conventional

processing of ceramics, excluding glasses, the following steps are involved:
(i) The starting material, generally in powder form, is prepared.

(ii) The powder is formed into a desired shape, with the help of a binder
material. This is called a green body.
(iii) The green body is sintered at a high temperature to obtain a sufficiently
dense material. Quite frequently, a combination of high temperature and
pressure (hot pressing) is used.
(iv) The sintered compact is finished into the final shape.
Melting and casting or drawing of a molten ceramic into a desired shape is not
very common. The conventional processing of glass, however, frequently
involves the melting of raw materials, followed by casting. In the case of glass
fibers, the molten glass is drawn through electrically heated platinum bushings
(more about this in Chapter 3). Some of the important conventional methods of
processing ceramics are described below.
2.5.1 Sintering of Ceramic Powders
Sintering is a common technique for consolidating ceramic powders. This

densification technique essentially involves atomic diffusion processes. The
green compact has some moisture and organic binder material which is burned
out at low temperatures. This is followed by heating to a temperature range
where atomic diffusion can lead to mass transfer from the particles into open
spaces in the green compact. Inevitably, this technique results in a tremendous
amount of shrinkage. A 10-20% shrinkage with its attendant problems of
cracking, distortion, and/or residual stresses is not an uncommon. Such
problems can be avoided, to some extent, in reaction-formed ceramics, e.g.,
reaction-bonded silicon nitride (RBSN), see Sec. 2.5.5.
2.5.2 Injection Molding of Ceramic Powders
Injection molding is a very attractive, mass production technique. Under

conditions where the ceramic powder can be incorporated in an organic vehicle,
one can have a variety of fast production methods, e.g., extrusion, blow
molding, and injection molding. These techniques are carry overs from the
polymers field. One important difference in the case of ceramics is that one
must remove the organic binder material before a fully sintered body with a
near theoretical density can be obtained.
2.5.3 Hot Isostatic Pressing (HIP)
Hot isostatic pressing (IllP) involves an application of uniform pressure at an

elevated temperature, Fig. 2.5. The pressure is applied via an inert gas (nitrogen
or argon) in a pressure vessel. Resistance heaters within the pressure vessel are
generally used for heating. Typical pressures can range between 100 and 300
MPa while temperatures can be as high as 2000 °C. Under these conditions the
argon gas has the viscosity of water and heat transfer occurs by a combination
of convection and radiation. The time period of a lllPping cycle may vary
between 1 hand 8 h. The three variables, temperature, pressure, and time are
adjustable parameters which can be optimized for a given material. Modern
lllPping units are equipped with microprocessors which allow one to program
the operating cycle and monitor the control of the three main processing
variables. The major advantage of lllP compared to pressureless sintering is
that the amount of additives can be reduced to a minimal level and a close to
theoretical density is obtained. In conventional sintering, the additives can
amount to 5 - 15 wt. % of ceramic. Clearly, such a high level of additives
Evacuate
_ _ and seal
Material
loading ............-......~
Container Loaded, sealed and

evacuated container
w
m ~ c::tJ
~. m
Stripping of
/ Hot-I '0 "all, .,,,,'n,
Consolidated
Part container
Fig. 2.5 Hot isostatic pressing (HIP) involves the simultaneous application of uniform
pressure and elevated temperature.
cannot but influence the material properties in a negative manner.
2.5.4 Chemical Vapor Deposition (CVD)
Chemical vapor deposition is another technique that can be used to make

ceramics. It is often used to make thin film electronic devices. The CVD
process can give additive free ceramics but has the disadvantage that it is a very
slow process and complex shapes cannot be produced.
2.5.5 Reaction-Forming
In reaction forming, also called reaction-bonded ceramic processing, a densified

ceramic body is obtained via a chemical reaction (Washburn and Coblenz,
1988). Commonly, a loose compact of solid powder (e.g., silicon) is reacted
with a liquid or gas (e.g., nitrogen) to give the desired ceramic. Thus, reaction
forming involves mass transport between a solid and a vapor or a liquid phase.
The reactants combine to produce a bonded body, hence the name reaction
forming or reaction bonding. A classic example is reaction bonded silicon
nitride or RBSN (Moulson, 1979). One main advantage of reaction bonded
ceramic processing is the small amount of dimensional change that occurs in
reaction bonding vis it vis 10-20% linear shrinkage commonly observed in
sintering. By far the biggest drawback of this process is the rather large amount
of porosity (10-15%). Such large porosity levels, inevitably, result in poor
mechanical properties. In addition, some unreacted components may be present.
2.6 NONCONVENTIONAL PROCESSING OF CERAMICS AND GLASSES
In this section, we discuss methods of obtaining ceramics and glasses via

chemical routes, microwave sintering, and self-propagating high temperature
synthesis (SHS), all of which can be grouped under nonconventional processing
techniques.
2.6.1 Ceramics via Chemical Routes
A direct melting process can result in a fully dense ceramic, but the
temperatures involved are extremely high. Also, quite frequently, the melt
viscosity at reasonable processing temperatures is too high to allow proper
homogenization. At times, the oxides show large differences in volatility and
a direct melting process can result in large losses of some of the constituents.
Thus, making ceramics and glasses via a chemical route assumes great
importance in such cases. A chemical process of making a ceramic or glass can
lead to a high degree of homogeneity on a molecular scale and consequently a
high purity product can be obtained. Among such novel processing techniques
developed mostly in the last quarter of the twentieth century, one of the most
innovative approach to ceramic and glass processing is the sol-gel technique.
We give a brief description of this process. The reader is referred to the list of
suggested reading at the end of the chapter for more detailed information on the
subject of sol-gel per se.
The sol-gel route of making any glass or ceramic involves the formation ofthe
appropriate glass or ceramic structure by chemical polymerization of suitable
compounds in the liquid state (sol) at low temperatures, followed by chemical
reactions such as hydrolysis or condensation at temperatures much lower than
those used in powder processing or direct melting. A sol is a colloidal
suspension in which the individual particles are so small that no sedimentation
occurs. The particle size in a sol generally varies between 1 - 100 nm. Thus,
the sol has a free-flowing form, which is a characteristic of the liquid state. It
can also be obtained by mixing a metal containing precursor (e.g., a metal
alkoxide), a solvent (e.g., an alcohol), a catalyst (e.g., an acid), and water.
Hydrolysis and polycondensation reactions make the sol viscosity increase until
a gelled state is obtained. This gel is like a wet solid. A gel is a suspension in
which the liquid medium is viscous enough to behave more or less like a solid.
A "precursor" material in the form of a gel is obtained from the sol. The "wet"
gel consists of a network of particles, alcohol, and water. The wet gel is dried
to remove any unwanted residues (water, organic compounds, etc.). A large
amount of shrinkage accompanies this step. The porous gel is then converted
into glass or ceramic by heating at temperatures much lower than those used in
direct melting processes or by conventional powder processes. In principle, the
desired glass or ceramic can be obtained in an appropriate form: powder, film,
fiber, etc. The gel is the starting material for conversion into glass or ceramic.
Figure 2.6a shows the sol-gel process flow diagram while Fig. 2.6b shows the
microstructural changes that occur during the process.
The sol-gel process really is not new. In 1846, a French chemist, Jacques
Ebelman reportedly formed a transparent solid that could be drawn into
fibers.The chemistry of the process became clearer in the 1920's. Essentially,
tetrachlorosilane (SiCI4) when reacted with ethyl alcohol gives the alkoxide
tetraethylorthosilicate, also called tetraethoxysilane [Si(OC2H s)4]' This
alkoxide, commonly known by the acronym TEOS, serves as the starting
material for making an inorganic polymer containing Si and 0 via the sol-gel
(.)
'®
liquid
. til (jel Aged gel
Residual
porosity
D ried gel P:lt l i: l ll~ ( jlass

cJcm•• licd or
gcl
(b)
Fig. 2.6 (a) The sol-gel process flow diagram; (b) microstructural changes occurring
during the sol-gel process.
route, i.e., the silica glass or ceramic, or even glass-ceramic. Most sol-gel
processing involves the formation of metal oxides (ceramics or glasses) from
metal alkoxides. A metal alkoxide has the chemical formula ofM(OR)n' where
M is a metal or metalloid, R is an alkyl group such as CH3, C2H s, etc., and n is
the valence of the metal atom. Hydrolysis of metal alkoxides provides a low
temperature route to produce ceramics. Most metal alkoxides react with water
to yield hydrous metal oxides which give the metal oxide on heating. Two or
more metal alkoxides can be reacted to produce a mixed-metal oxide which
yields binary or ternary oxides on hydrolysis.
The metal alkoxide route has been exploited largely to obtain oxide ceramic
systems. Among nonoxide ceramics, the notable examples are polycarbosilane
and aminosilane routes for SiC and Si3N4 production (Hench and Ulrich, 1984).
Methods of Obtaining Silica-Based Gels
There are two ways of obtaining silica-based gels:
(a) Destabilization of silica sol (pure or with metal ions added to aqueous
solutions of salts) to obtain a homogeneous gel. When a sol is destabilized, the
resultant product can be precipitates, unaggregated particles, or a homogeneous
gel. It is the last one that we desire.
(b) Hydrolysis and polycondensation of organometallic compounds generally
metal alkoxides) dissolved in alcohols and a small amount of water.
(aJ Destabilization of Silica Sol
Destabilization of a silica sol can be done by either increasing the temperature

or the addition of electrolytes. Temperature increase reduces the quantity of
intermicellar liquid by evaporation and increases the thermal agitation. This
increases particle collisions and their linking by condensation of surface
hydroxyls occurs. Electrolyte addition modifies the pH of the sol and reduces
electric repulsion between particles. Addition of acid diminishes the pH to 5 - 6
and induces gel formation by aggregation. The sol is then converted into gel
progressively with microgel aggregates slowly invading the whole volume of
sol. When about half of the silica has been converted to gel, a rapid increase in
viscosity is noted.
Aging
Partial coalescence of particles takes place during aging treatment and a

strengthening of the network occurs. At the neck joining the particles there is
a negative radius curvature. Thus, local solubility at the neck is less than near
the particle surface. Therefore, transport and deposition of silica occur
preferentially to the neck region and neck thickening results. This results in a
strengthening ofthe particulate network, Fig. 2.7 (Zarzcyki et aI., 1982).
The difference between sol (a colloidal suspension) and gel (a

three-dimensional network of particles) and the phenomenon of precipitation
(formation of separate aggregates or flocculation) is shown schematically in
Fig. 2.8. Sol-gel transition occurs only if there are no active forces which
promote coagulation into aggregates of higher silica concentration than the
original sol. Metal cations, particularly polyvalent ones, tend to cause
precipitation rather than gelling.
(b) Hydrolysis and Polycondensation of Organometallics (Metal Alkoxides)
This method involves three steps (Zarzycki, 1984; Sakka, 1982):
Fig. 2.7 Strengthening of the particulate network by neck thickening.

•
• • tf4r
•
• "II
•
• • ~
• • •
•
Sol Gel Precipitate

Fig. 2.8 Difference between a sol, a gel and a precipitate.
(i) Mix the appropriate metal alkoxides, water-soluble salts if possible, and
ethanol solutions to yield the desired cation ratio.
(ii) Hydrolyze the above mixture with a water based acid solution
which can be diluted with alcohol. This polymerizes the mixture and a gel is
formed as per the following reaction
where M is the metal and R is an alkyl group such as (C2HS). At this stage one
can choose the conditions to get the bulk, powder, or fiber form of the precursor
material.
(iii) Heat the gel slowly to obtain the oxide (glass or ceramic) through the
reaction
M(OH)n --+ MO n12 + 1/2 n HP
During the process of gel drying, the solvents (water and any residual organic
material) will be eliminated. Concomitantly, sintering of the porous oxide takes
place. Very controlled rate of drying is required because of the large shrinkage
that accompanies the drying process. At an intermediate stage, a solid with a
very large amount of porosity on a micro scale is obtained. This is called
xerogel. The resultant glass or ceramic is virtually the same density, thermal
expansion, refractive index, and mechanical characteristics as the material of
the same composition made by direct melting.
2.6.2. Microwave Sintering
Microwave energy provides another new avenue for processing ceramics and
CMCs. In conventional sintering, heat is provided by an external source. In the
microwave process, just like in the simple home microwave oven, the heat is
generated internally in the material. Volumetric heating occurs because of
internal frictional losses. Microwaves generate electromagnetic waves in the
frequency range of 0.3 to 300 GHz (wavelength range from I m to I mm), and
they are coherent and polarized.
Many insulating ceramic materials such as A1 20 3, MgO, Si02, and many glasses
absorb microwaves at above a certain critical temperature, Terit • Large sections
can be heated uniformly, because of lower thermal gradients.
Microwave processing can be more economical than conventional heating and

sintering, because heating rates by microwave radiation can be as much as 50
times greater than the conventional heating process. This translates to a higher
throughput rate. In microwave processing, thermal energy is coupled to the
component being heated, and heat losses are low, so less power is needed.
The rapid heating rate causes ceramics processed by microwave radiation to

have finer and more uniform grain size than conventionally processed ceramics.
2.6.3 Self-Propagating High Temperature Synthesis (SHS)
This technique can be used to produce a variety of refractory materials (Munir

and Anselmi-Tamburini, 1989). A classic example of SHS is the well known
thermite reaction. The thermite reaction is a highly exothermic reaction that
occurs when a mixture of finely reaction is a highly exothermic reaction that
occurs when a mixture of finely divided aluminum powder and ferric oxide is
ignited by a fuse. Temperatures as high as 2500 °C may be reached. This
technique involves synthesis of compounds without an external source of
energy. Certain exothermic reactions can be exploited to synthesize ceramic
compounds which are which are difficult to fabricate by conventional
techniques. For example, one can mix titanium powder and carbonblack, cold
press it, and ignite the compact at the top in a cold-walled vessel. A combustion
wave will pass through the compact, giving titanium carbide. Figure 2.9 shows
a schematic ofthe formation of SiC by SHS technique.
SHS products are rather porous. There are two sources for this porosity: (i) the
original mix of reactant powders has about 30-50% porosity, and (ii) there is a
Product
SiC
Reactants
Si+C
Fig. 2.9 A schematic of SiC fonnation by self-propagating high-temperature synthesis

(SHS).
large difference in the molar volume of the reactants and the products. Yet
another source of porosity may be due to any adsorbed gases at the extremely
high temperatures generated. A modification of the technique to obtain
synthesis together with densification involves the application of high pressure
during the combustion or immediately after the completion ofthe reaction when
the product temperature is still quite high. Hot pressing, rolling , and shock
waves are some of the techniques to apply the pressure.
Among the salient features of SHS are:
1. High combustion temperature (up to 4000 0C).

2. Simple, low-cost equipment.
3. Good control of chemical composition.
4. Different shapes and forms can be obtained.
Many ceramics such as borides, carbides, nitrides, silicides, and sialons as well
as composites such as SiCw + Al20 3 have been synthesized by means of SHS.
Synthesis can be followed by explosive or dynamic compaction of ceramics.
The reader should consult some of the references listed under Suggested
Reading for greater details.
2.7 PROCESSING AND PROPERTIES OF SOME CERAMIC MATRIX MATERIALS
Now we briefly describe the processing and properties of some important

ceramics that are used or can be used as matrix materials in ceramic

composites.
The characteristics such as high hardness, high strength, high modulus, wear
resistance, etc. are achievable only in a crystalline ceramic in a fully dense
condition. Generally, a fully dense state is easier to achieve in oxide ceramics
than in non oxide ceramics. Typically, a nonoxide ceramic such as reaction
bonded silicon carbide (RBSC) is highly porous and thus very weak. Even in
hot pressing of these ceramics, one must use additives such as magnesia or
yttria to attain the theoretical density. These additives generally have
deleterious effects on the creep and oxidation resistance ofthe ceramic.
2.7.1 Silicon Carbide
Silicon carbide is a very hard and abrasive material. It has excellent resistance
to erosion and to chemical attack in reducing environments. In oxidizing
environments, any free silicon present in a silicon carbide compact will be
oxidized readily. Silicon carbide itself can also be oxidized at very high
temperatures.
Silicon carbide does not occur in nature, although silicon and carbon are two
of the most abundant elements on earth. Carborundrum is a trade name for
silicon carbide. It is a combination of the Latin words carbon and corundum.
Silicon carbide can be obtained in two forms: a-SiC (hexagonal) and P-SiC
(cubic). The hexagonal a-silicon carbide has many polytypes which differ in the
stacking sequence. Typical physical and mechanical properties of silicon
carbide are given in Table 2.5.
There are four important routes to obtain silicon carbide:
(i) Hot pressing

(ii) Sintering
(iii) Reaction sintering
(iv) Chemical vapor deposition (CVD)
A brief description of these methods follows.
(i) Hot pressing
Silicon carbide is very hard to sinter and will not deform under usual conditions
of hot pressing. Additives such as MgO, B, C, or Al to either a or ~-SiC must
be used. Typically, at 1900-2000 °C one uses a pressure of 35 MPa. Hot

processing temperatures are thus very high. Also, hot pressed SiC must be
shaped by diamond-tipped tools, which is expensive.
(ii) Sintering
Sintered silicon carbide does not melt but decomposes into a silicon-rich vapor
and a carbon-rich residue. It is also very creep resistant at high temperatures.
Such unusual characteristics make conventional sintering of silicon carbide
difficult. Prochaska (1974) used sub micrometer P-SiC powder with a low
oxygen content was used, with 0.5% Band 1% C as additives and succeeded
in pressureless sintering SiC to full density between 1950-21 00 °C in vacuum
or inert atmosphere.
Sintered SiC has the advantage that it can be shaped in the "green" state, i.e.,
when it is soft. The following points are very important in pressureless sintering
of SiC:
1. Use of a sub micrometer powder is important because this gives a large
surface area, short diffusion distances, and large thermodynamic driving force
for densification.
2. Chemistry of the starting powders is also very important. Carbon removes the
SiOz layer from the SiC particles. Addition of boron is thought to accelerate
volume and grain boundary diffusion. It is important to avoid a nonuniform
distribution of the C additive and an exaggerated grain growth of a-SiC.
(iii) Reaction Sintered SiC (RBSC)
A mixture of SiC powder, graphite, and a plasticizer is made into a plastic body.
This is pressed, extruded, and injection molded to obtain a green body. The
plasticizer is burned off by pyrolysis.
Silicon metal (liquid or vapor) is then infiltrated into the body. This reacts with
the carbon powder to form SiC in situ. Any excess silicon (2-12%) goes to fill
any voids, giving a reasonably nonporous body. Reasonable strength to melting
point of silicon (1400 0c) or beyond has been obtained, depending on the
amount of free Si retained.
Properties of monolithic silicon carbide made by different processes can vary

considerably. Typical values are provided in Table 2.5 (Leatherman and Katz,
1989).
2.7.2 Silicon Nitride
There are two forms of silicon nitride, a and ~, and both are hexagonal, with the
c-axis of a-Si3N4 being about twice that of ~-Si3N4 (Messier and Croft, 1982).
Contamination ofSi3N 4 with oxygen is a perennial problem. A brief description
of the various processing techniques is given below.
Sintered Silicon Nitride
Pressureless sintering of silicon nitride has obvious attractions. The

temperatures required (:::: 1700 -1800 0c), however, are in the range where
thermal decomposition of silicon nitride can be a problem. One can prevent the
decomposition by either maintaining a nitrogen overpressure or by using
sintering aids and conditions such that the process can be done in nitrogen at
atmospheric pressure.
Sintered silicon nitride has strength levels between those of hot-pressed silicon
nitride and reaction-bonded silicon nitride.
Table 2.5 Typical properties of silicon carbide made by different processes.
Bend Strength (4-point), Coeff of Thermal

MPa Young's thermal conduct-
R.T JOOO"C 1375"C modulus, expansion ivity, k
E, a,
GPa 10-6 K- 1 Wm-1K- 1
Hot
Pressed 690 620 330 317 3.0 30-15
(MgO
additive)
Sintered
(Yz03 655 585 275 236 3.2 28-12
additive)
Reaction-
bonded 210 345 380 165 2.8 6-3
(2.45
glcm3)
Hot-Pressed Silicon Nitride (HPSN)

This process involves heating a-Si3N4 and some sintering aid under pressure.
Typically, the pressure is about 30 MPa, the temperature range is 1600-1800
DC, and the time involved is a few hours. MgO is a common sintering aid. It is
thought that the MgO combines with the Si02 (a common contaminant on the
Si3N4 powder surface) to form magnesium silicate liquid phases. This liquid
penetrates between Si3N4 grains, providing a high diffusivity path, and becomes
a glassy phase on cooling from the pressing temperature. a-Si3N 4 dissolves in
the liquid silicate and elongated B-Si3N4 grains precipitate out from this. After
the a ... Btransformation is complete, the microstructure ofHPSN is comprised
of B-Si3N4 bonded via a thin layer of the glassy phase at the boundary. This
glassy phase is the cause for poor strength at temperatures above 1200 DC.
Other possible sintering aids are Ce02, Y203, and Zr02 • Typically, 98% of
theoretical density can be obtained.
As with any hot-pressing process, the process is expensive and complex shapes
are difficult to form. Among the parameters that control the properties of the
final Si3N4 product are:
• Purity and phase composition of the initial powder.

• Additive and its amount
• Hot-pressing variables time, temperature, pressure.
MgO is by far the best known densification additive used for Si3N 4. Typically,
because of the magnesium silicate grain boundary phase, creep resistance starts
falling off between 1200 and 1350 DC.
Reaction-Bonded Silicon Nitride
Silicon powder is compacted into a desired shape, say, by slip casting, and then
heated in a nitrogenous atmosphere (pure N2 or N2 + H 2) in the temperature
range of 1100-1400 °C. The following chemical reactions may occur in the
process of nitridation of Si:
3Si(s) + 2N2(g) --+ Si3Nis)

3Si(g) + 2N2(g) --+ Si3N4(s)
SiCs) + Si02 --+ 2SiO(g)
The last reaction describes any loss of Si during the process.
Then reaction-bonded silicon nitride is porous (> 10%), is less oxidation

resistant than hot-pressed silicon nitride at intermediate temperatures, and has
a strength < 400 MPa, usually 250 MPa.
HIP of Silicon Nitride
ConventionallllP involves cladding of Si3N4 green body by glass. Uniformly

dense and precise shape control are possible, but interaction between glass and
Si3N4 as well as surface damage when the cladding is removed can result. A two
step nitrogen overpressure process is a better alternative. In the first step, the
silicon nitride body is sintered to closed porosity under 1- 10 atmospheres of
nitrogen. In the second step, a pressure of 20- 100 atmospheres of nitrogen is
applied. This process eliminates the need for cladding.
Chemical Vapor Deposition (CVD)
Films of silicon nitride, made by chemical vapor deposition, are used as

insulating films and diffusion masks. For this purpose, silanes or silicon halides
are reacted with ammonia at temperatures in the range of 800-1100°C. The
following chemical reactions occur:
3SiHlg)(silane) + 4NH3(g) ---+ Si3Nls) + 12H2 (g)

3SiClig) + 4NH3(g) ---+ Si3Nis) + 12HCI(g)
3SiHCllg)(trichlorosilane) + 4NH3(g) ---+ Si3Nis) + 9HCI(g) + 3Hlg)
A fully dense product can result, but complex shapes are difficult to obtain.
Typical properties of silicon nitride made by different modes are given in Table
2.6 (Messier and Croft, 1982).
Table 2.6 Properties of silicon nitride made by different methods
Sintered
Property RBSN HPSN HPSN Si3N 4
(MgO) (Y,OJ (y,o,)
Density, Q (glcm 3 ) 2.8 3.2 3.4 3.2

Young's Modulus, E 210 300 310 275
(GPa)
Poisson's Ratio, 'U 0.22 0.25 0.27 0.23
Shear Modulus, G 86 120 122 113
(GPa)
Flexural Strength, 288 760 920 665
(4-Point) MPa
2.7.3 Alumina
There is only one thermodynamically stable phase, viz., a-AI2 0 J (hexagonal).

a-Alumina has a hexagonal structure with aluminum ions at the octahedral
interstitial sites. Each aluminum ion is surrounded by six equidistant oxygen
ions. Figure 2.10 shows the hexagonal close-packed structure of a-alumina. A
and B layers contain oxygen ions while C layers contain aluminum ions and
vacant sites. The C layers are only two-thirds full for charge neutrality. Figure
2.11 shows the two slip systems, basal and prismatic, in a hexagonal structure.
In alumina, at temperatures below 1000 °C, slip can occur on these two slip
planes. A fine-grained alumina structure is generally desirable at low
temperatures because at low-temperatures, both strength and toughness increase
with a decreasing grain size.
Alpha - alumina toughened by zirconia particles has gained commercial

importnace. Zirconia can have either a tetragonal or monoclinic form. The
tetragonal-t monoclinic transformation of zirconia particles is accompanied by
a volume change which results in a toughening of alumina. Zirconia toughened
alumina is produced by mixing alumina and zirconia particles, and densifying
by sintering around 1500 C or by hot pressing. Hot isostatic pressing can also
0
be done. Zirconia particles are generally located at the grain boundaries.
Typical properties of monolithic alumina are given in Table 2.2.
0 Oxygen
A • AI
':; Vacant
C1
B
C2
C3
B
C1
Fig. 2.10 Hexagonal close-packed structure of a-alumina.

'L a
Prismatic Plane --H..,
Basal Plane
Fig. 2.11 Two slip systems in a hexagonal structure.
2.7.4 Mullite
Mullite is a solid solution of alumina and silica in the compositional range of

71-75% alumina. Commonly, mullite is represented by formula, 3 AlP3.2
Si02.It has excellent strength and creep resistance as well as low thermal
expansion and conductivity. Table 2.2 gives the typical properties of mullite.
For detailed information about mullite and mullite matrix composites, the
reader should consult Somiya et aI., 1990.
2.7.5 Boron Nitride
Boron and nitrogen can form the following BN compounds which are
isostructural polymorphs of carbon (Lipp et aI., 1989).
a-BN: This has a hexagonal, layered structure similar to graphite with a

theoretical density of2.27 g cm-3.
P-BN: This variety has a cubic, diamond-like structure and is extremely hard
like diamond. Its theoretical density is 3.48 g cm-3.

y-BN: This is also hexagonal, but with a higher density of3.48 g cm-3.
The hexagonal variety of boron nitride shows natural lubricity because of easy
cleavage. The easy cleaage follows from its lamellar structure. It is frequently
used as a mold release agent during hot pressing of ceramics such as silicon
nitride. Boron nitride is applied generally as a paint, paste, or aerosol. Because
of its similarity with graphite, hexagonal BN is sometimes referred to as white
graphite. It should be mentioned, however, that graphite is a good electrical
conductor while BN is an insulator. Unlike graphite, BN also shows an
excellent resistance to oxidation. Boron nitride in a powder form is also used
as an additive for silicon nitride and alumina.
Finally, an important form of boron nitride should be mentioned, viz., pyrolytic

boron nitride. It is manufactured by reacting ammonia and a boron halogenate
at about 2000 °C and depositing the BN vapor on a graphite substrate or
mandrel. The characteristic feature of pyrolytic boron nitride is the high degree
of crystal orientation with the hexagonal basal plane parallel to the mold
surface and the c - direction perpendicular to the substrate.
2.7.6 Boron Carbide
Boron carbide is another low density, high melting point, high hardness ceramic
(Thevenot, 1990). Boron carbide powder is produced by reacting briquettes of
B 20 3 and C mixture in an electric arc furnace. The overall reaction is:
Boron carbide powder can be sintered to a dense material by pressureless

sintering, hot pressing, or hot isostatic pressing. Table 2.7 gives a summary of
the properties of boron carbide.
2.7.7 Sialons
The term sialons was originally coined to indicate a solid solution of Al 20 3 in

Si3N 4. Sialon is an acronym used to denote phases in the Si-AI-O-N systems that
are made up of (Si,Al)(O,N)4 tetrahedra. However, other metal oxides, such as
MgO, BeO, and Y203 can also form solid solutions in ~-Si3N4' Ceramic alloying
is used to obtain sialons (Jack, 1986). Silicon is replaced by aluminum and
nitrogen by oxygen in W-sialons. In a'-sialons, one replaces Si with Al in a-
Si3N 4. Variations in the amounts of ionic and covalent bonding lead to
variations in properties. Sialons can be vitreous as well as crystalline.

Increasing nitrogen results in increased viscosity, density, and hardness, i.e.,
refractory and stronger glasses. Because the solid solutions produce a distorted
~-Si3N4 lattice, these are called W-sialons.
There are two main groups:
(i) W-sialon plus glass: This has a high room temperature strength (3-point bend
strength", 1000 MPa) but poor creep properties above 1000 °C because of the
glassy phase.
(ii) W-sialon plus Y AG: This has a ceramic oxide rather than a glassy phase as
the grain boundary phase and, consequently, has excellent creep and oxidation
resistance.
Sialons find applications in cutting tools for metals, seals, bearings and wear
components, and in molten metal handling.
2.7.8 Carbon
Carbon is used as a matrix with carbon fibers. Carbon can resist very high
temperatures but only in the absence of oxygen or air. Commonly carbon matrix
is obtained by chemical vapor deposition (CVD) or using on organic precursor
(a resin or pitch) to infiltrate a fibrous preform, followed by pyrolysis of the
precursor to convert it into carbon.
Carbon can exist in a variety offorms: amorphous, graphitic or hexagonal form,

diamond cubic form, and some newer forms called fullerenes, etc. Carbon
matrix obtained via CVD has commonly a disordered structure consisting of a
mixture of graphitic and amorphous regions. Carbon matrix obtained via
pyrolysis pitch or resin is amorphous.
Typical properties of carbon matrix include a low coefficient of thermal

expansion « 2 10- 6 K- 1), a density of about 1.5 g cm·3 , a relatively low young's
modulus of30 -35 GPa.
2.7.9 Intermetallics
The demand for materials that are strong, stiff, and ductile at high temperatures
has led to resurgence of interest in intermetallics. Westbrook defines
intermetallics as all compound phases (ordered or disordered) of two or more
metals (Westbrook, 1960). The field of intermetallics is, indeed, very vast.
Pearson's Handbook of Crystallographic Data for Intermetallic Phases,

published in 1985, consists of three volumes and contains data on thousands of
intermetallic compounds (Villers and Calvert, 1985). This should give the
reader some idea of the scope of the field. This, however, does not imply that
there exists an equally vast database on mechanical and physical properties of
the intermetallics. On the contrary, information on these is woefully inadequate.
The most promising intermetallics for high temperature load-bearing

applications such as turbine engine components include the aluminides of
nickel, titanium, and iron. However, they are limited to use temperatures well
below 1600 DC. Many silicide intermetallics have the potential of meeting
structural and oxidation requirements to 1600 DC.
For service temperatures up to 1600 DC, besides high strength, creep resistance,
fracture toughness, one needs characteristics such as oxidation resistance and
microstructural stability. Meschter and Schwartz (1989) surveyed some silicide
intermetallics for high temperature applications. Based on criteria such as
Table 2.7 Typical properties of boron carbide (Thevenot, 1990).
Property Hot- Sintered B 4C

pressed
B4C B4C B4C
CI wt%q (Jwt%q
Total carbon content", wt % 21.7 22.5 24.8
Porosity, % <0.5 <2 <2
Bulk density, g/cml 2.51 2.44 2.46
Mean grain size, 11m 5 8 7
Flexural strength (four-point 480±40 351±40 353±30

bend), M N/m2
Young's modulus, GPa 441 390 372
Shear modulus, GPa 188 166 158
Poisson's ratio 0.17 0.17 0.17
Fracture toughness, K Ic 3.6±0.3 3.3±0.2 3.2±0.2

(SENB)), MPa m 112
"The amount of free carbon can be calculated approximately from Cfree =1.28 Ctotal-28.
availability, phase changes in the temperature range of interest, and oxidation

resistance, TisSi3 and MoSi2 seem to be the most promising materials; TisSi3 has
the lowest density and MoSi2 has superior oxidation resistance. Figure 2.12
shows a plot of melting point vs. density for intermetallics having 0.8 Tm =
1600 °C. Here we are assuming that intermetallics retain strength u to 80% Tm
(Tm in K). This puts a lower limit on Tm equal to 206rC for a service
temperature of 1600 °C. Also selected in this graph is an upper limit on density,
viz., the density of nickel. There are seven silicides plus MoSi 2 (2020 ± 20°C)
that fall in this window.
Many of these compounds show brittle behavior at low temperatures « 800°C)

and plastic behavior at high temperatures (> 800°C). Consequently, these
materials are brittle at room temperature and they creep at high temperatures.
It would appear that making composites with these by using particle, whisker,
or fiber reinforcement can lead to toughening at low temperatures and creep
resistance at high temperatures. We shall discuss the processing, structure, and
properties ofintermetallic matrix composites in other chapters. Table 2.8 gives
a summary of the typical properties of some intermetallics.
3,200...----..,....---------------------,
~
• -Si Compounds ~
3,000 p" EEl -Be Compounds ~
<!J) - AI Compounds
..
~ - Re Compounds
r
2,800 ~
O-Others
E 2,600
~ ~
SiC
~
Z EEl
.
'"
(j;
E 2,400
Zr 5 Si. ---t\
- (\ jZrSi:
:
~
Cl zrJSi2~.J M0 5 Si J ( •
-E
CD
2,200
.. : ... ~ •• 0
o
:2
.... T~5. ~I.'..~ ·~.~~~~OSi·;······~·· .•.. ·.. 6··· 0.. ······· ...········ .. ·· .. ··· .. ·T m· .. • •• ~
........ .
2,000
EEl •• : • • : e...
~ ~ : (fJ..J0 0
o
1,800
: .: <0
1,600
EEl, ~~~OO<!J)i <!J)0 8 0
0
0 o
2 4 6 8 10 12 14 16 18 20 22
Density (g/cmJ)
Fig. 2.12 A plot of melting point vs. density for intermetaIIics having 0.8Tm = 1600 °C
(after Meschter and Schwartz, 1989).
Table 2.8 Properties of some intermetallic compounds
Intermetallic (:~stal~tructure Melting Density, Young's

(:ompound Point, 0(: glcm 3 Modulus,
GPa
B2 1250-14 5.6 263
FeAl (Ordered BCC) 00
Fe3Al D03 1540 6.7
NiAl B2 1640 5.9 206

(Ordered FCC)
Ni3A1 LI2 1390 7.5 33.7

(Ordered FCC)
TiAl LIo 1460 3.9 94

(Ordered tetr.)
Ti3A1 D0 19 1600 4.2 210

(Ordered RCP)
MoSi2 Tetragonal 2030 6.3
2.8 CREEP IN CERAMICS
In most ceramic materials, the number of operative slip systems is limited

because of their ionic/covalent bonding.
A typical creep curve is shown in Fig. 2.13. The curve is divided into three
stages: primary (stage I), secondary or steady state (stage II), and tertiary (stage
III) in which an accelerated damage leads to fracture.
The creep rate, e, in the steady state can be expressed by the following general
relationship:
e = (ADGb/kT) (b/dg)m (a/Gt
where A is a dimensionless constant, D is the diffusivity, G is the shear
modulus, b is the Burgers vector, k is the Boltzmann's constant, T is the
temperature in kelvin, dg is is the grain size, a is the applied stress, and m and
1
3
v-; ,
...L
I
I
2
Time (log scale) -

Fig. 2.13 A typical creep curve.
n are the exponents. Different values of m and n correspond to different

creep mechanisms (Langdon, 1975).
In polycrystalline ceramics, creep can occur by void growth (at grain

boundaries or within grain). This type of creep is called diffusional creep,
because the mechanism responsible forthis is the operation of vacancy currents.
This is commonly made worse by the presence of any glassy phase at the
boundaries. Commonly, under creep conditions, cavitation is observed in the
intergranular glassy phase. At higher stress levels, power law creep occurs. As
the creep mechanism changes from diffusional to power law, the damage rate
increases significantly. At a homologous temperature (the ratio of temperature
in kelvin to the melting point in kelvin, T/T m ) equal to or greater than 0.6,
creep in ceramics can occur by dislocation climb.
2.9 FRACTURE OF CERAMICS
As we have said repeatedly, ceramics have high strength but low fracture
toughness. The low fracture toughness has its origin in the extreme sensitivity
of ceramics to the presence of flaws in them.
2.9.1 Flaw Types in Ceramics

Various flaw types can occur in ceramics. We can categorize these into three
broad types:
(i) Processing induced flaws: Inclusions, pores, isolated large grains,

laminations induced during pressing, machining induced nicks, and thermal
stresses.
(ii) Design induced flaws: Sharp corners, holes, etc.
(iii) Service induced flaws: Environmental degradation, thermal stresses,
impact, and wear.
Porosity is one of the more common defects in ceramics. There exists a variety
of porosities in ceramics. Closed porosity is a serious problem in ceramic
bodies. Apparent porosity measures the interconnected porosity in a body, and
determines the permeability of the ceramic by gases and fluids. Careful
processing can eliminate many flaws.
2.9.2 Strength and Toughness of Ceramics
Flaws (internal or surface) in ceramics reduce their strength. Fracture stress of

an elastic material (e.g., a ceramic) containing an internal crack of length 2a is
given by the Griffith relationship:
where E is the Young's modulus and y is the surface energy of the crack surface.
In terms oflinear elastic fracture mechanics, we describe the situation existing
a =
f
~ 2E1
a
at the tip of a crack in terms of a parameter called stress intensity factor, K. We

can write for stresses in the vicinity of a crack tip, at a point (r,9) in Fig. 2.14,
as
where fij(9) is a function of the angle, 9. Fracture occurs when the stress
intensity factor K attains a critical value, K1c • K]c> also known as the plane strain
fracture toughness, is a material parameter. An equivalent approach to the stress
intensity approach is based on the energy viewpoint. According to this
approach, fracture occurs when the crack extension force, G, reaches a critical
value, Gle • For ceramic materials, Gle=2,),. It can also be shown that
(plane stress)
and
EG
(plane strain)
(1 -V)2
where v is the Poisson ratio. These expressions show that the K and G
approaches are equivalent.
2.9.3 Ceramic Matrix Requirements
Although, no one ceramic is going to meet all the requirements, one can make
a list of some desirable characteristics of a ceramic matrix. That is, what one
would like to see in an ideal matrix material. In a real matrix, one can only hope
for a large number of these characteristics. An ideal ceramic matrix material
should:
• infiltrate a bundle of fibers, whiskers, or particulate preform.

• form a mechanical or frictional bond with the reinforcement.
• have no chemical reaction with the fiber reinforcement during fabrication or
service.
• not damage the fiber physically.
• have good resistance to creep, fatigue, and impact.
• have a high toughness.
• should be chemically stable, i.e., it should be impermeable to moisture,
resistant to oxidation, should not hydrate or volatilize, etc.
In air or oxygen, oxide ceramics are inherently more stable than nonoxide
ceramics. Alumina (AI 20 3 ) is one of the most stable ceramics in a variety of
environments, although its melting point (2050 0c) is not one of the highest.
For most crystalline ceramics, at ambient temperatures, the finer the grain size,
the smaller will be the flaw size, and the higher the strength. At high
temperatures, the strength of ceramics is mainly determined by the
characteristics of the glassy phase that is frequently present at the grain
boundaries.
x
Crack
Fig. 2.14 A crack tip and a point (r, 9) near the crack tip.
During high temperature sintering, densification occurs by diffusional processes

involving migration of vacancies from pores to sinks such as grain boundaries,
other internal boundaries, and free surface. Hot pressing is resorted to obtain
high density and a fine grain size.
REFERENCES
Hench, L. and D.R. Ulrich (1984) (eds.), Ultrastructure Processing of Ceramics,

Glasses and Ceramics, Wiley, New York.
Jack, K.H. (1986) in Non-Oxide Technical and Engineering Ceramics, Elsevier App.
Sci., London, p. l.
Langdon, T.G. (1975) in Deformation ofCeramic Materials, Plenum Press, New York,
p. 10l.
Leathennan, G.L. and R.N. Katz (1989) Superalloys, Supercomposites and
Superceramics, Academic Press, Boston, p. 673.
Lipp, A., K.A. Schwetz, and K. Hunold (1989) J. Eur. Ceramic Soc., 5
Meschter, P.J. and D.S. Schwartz (Nov. 1989) "Silicide-Matrix Materials for High
Temp. Applications", JaM, 52.
Messier, D.R. and W.J. Croft (1982) in Preparation and Properties of Solid State
Materials, vol. 7 ,Marcel Dekker, New York, New York, p. 13l.
Mouison, A.J. (1979) J. Mater. Sci., 14, 1017.
Munir, Z.A. al\d U. Anselmi-Tamburini (1989) Mater. Sci. Reports, 3, 277.
Prochaska, S. (1974) in Ceramics For High Performance Applications, J.J. Burke, A.E.
Gorum, and R.N. Katz (eds.), Brook Hill, Chestnut Hill, MA., pp. 239-252.
Sakka, S. (1982) in Treatise on Materials Sci. and Tech., vol. 22, (M. Tomozawa and
R.H. Doremus, eds), Academic Press, New York, pp. 129-167.
Somiya, S., R.F. Davis, and J.A. Pask (1990) (eds.), Mullite and Mullite Matrix
Composites, Amer. Ceram. Soc., Westerville, Ohio.
Thevenot, F. (1990) J. Euro. Cer. Soc., 6, 205.
Villers, P. and L.D. Calvert (1985) (eds.), Pearson's Handbook o/Crystallographic
Data/or Intermetallic Phases, vol. 1,2,3, ASM, Metals Park, OH.
Washburn, M.E. and W.S. Coblenz (1988) Amer Cer. Soc. Bull., 67, 356.
Westbrook, J .H. (1960) (ed.), Mechanical Properties o/Intermetallic Compounds, John
Wiley, New York, p. 1.
Zarzycki, J. (1984) in Glass: Science and Technology, vol. 2, Academic Press, New
York, pp. 209-249.
Zarzycki, J., M. Prassas, and J. Phalippou (1982) J. Mater. Sci., 17,3371.
SUGGESTED READING
Barsoum, M. (1997) Fundamentals o/Ceramics, McGraw-Hill, New York.

C.J. Brinker and G.W. Scherer (1990) The Sol-Gel Science, Academic Press, New
York.
Chiang, Y.-M., D. P. Birnie and W. D. Kingery (1997) Physical Ceramics, John Wiley,
New York.
Green, D.J. (1998) An Introduction to the Mechanical Properties of Ceramics,
Cambridge University Press, Cambridge.
Jones, R.W. (1989) Fundamental Principles o/Sol-Gel Technology, The Institute of
Metals, London.
Kingery, W.D., H.K. Bowen, and D.R. Uhlmann (1976) Introduction to Ceramics, 2nd
ed., John Wiley, New York.
Reed, J.S.(1988), Introduction to the Principles o/Ceramic Processing, John Wiley,
New York.
Scholze, H. (1991) Glass, Springer-Verlag, New York.
Strnad, Z. (1986) Glass Ceramic Materials, Elsevier, Amsterdam.
Wachtman, J.B. (1996) Mechanical Properties o/Ceramics, John Wiley, New York.
Cliapter 3
CERAMIC REINFORCEMENTS
I
Ceramic reinforcements can be produced in the form of continuous fiber, short

fiber, whisker, or particle. Continuous ceramic fibers are very attractive for
reinforcing ceramic materials. They combine rather high strength and elastic
modulus with high temperature capability and a general freedom from
environmental attack, making them attractive as reinforcements in high
temperature structural materials. Continuous fibers are, however, more
expensive than particulate reinforcements. It is convenientto divide the ceramic
reinforcements into oxide and nonoxide categories. Table 3.1 lists some
important ceramic reinforcement materials available in different forms.
There are three types of ceramic fiber fabrication methods: chemical vapor
deposition, polymer pyrolysis, and sol-gel process. The last two routes involve
rather novel techniques of obtaining ceramics from organometallic polymers.
Undoubtedly the great breakthrough in the area of ceramic fibers has been the
Table 3.1 Some important ceramic reinforcements
Particle SiC, TiC, AlP3
Discontinuous Fibers
(a) Whiskers SiC, TiB2' Al20 3
(b) Short Fibers Glass, A120 3, SiC, (AIP3+SiOJ,
vapor grown carbon fibers
Continuous Fibers
Oxide AIP3' (AI20 3+Si0 2), Zr0 2,
silica-based glasses
Nonoxide B, C, SiC, Si3N4 , BN
48 CERAMIC REINFORCEMENTS
concept of pyrolyzing polymeric precursor fibers, under controlled conditions,

containing desired species to produce high temperature ceramic fibers. For
example, a polycarbosilane precursor will give silicon carbide. The concept is
really quite broad and I, for one, include the polymer pyrolysis route to produce
a variety of carbon fibers in this category. In this case, a suitable carbon based
polymer (PAN, Pitch, etc.) is subjected to controlled heating to produce carbon
fibers. The pyrolysis route of producing ceramic fibers has been used with
polymers containing Si, C, N, and B with the end products being SiC, Si 3N 4,
B4 C, and BN, respectively.
We describe below some general features of reinforcements and then provide

a more detailed description of processing, microstructure, and properties of
some important ceramic reinforcements.
3.1 DEFINITIONS
One can transform practically any material (polymers, metals, or ceramics) into
a fibrous form (Chawla, 1998). Nature makes extensive use of materials in a
fibrous form. Historically and traditionally, however, fibers (synthetic as well
as natural) have been used mostly in the textile industry. Thus, some common
terms used with fibers have their origin in textile technology terminology. We
examine some of the important terms below.
Fibers are the fundamental unit in making textile yarns and fabrics. There are
many naturalfibers, organic and inorganic. Organic natural fibers include silk,
wool, cotton, jute, sisal while asbestos and basalt are example of natural
inorganic fibers. Synthetic fibers include polymeric, metallic, and ceramic
fibers. Polymeric fibers such as nylon, polyethyleneterephthalate (PET),
polyethylene (PE), polyacrylonitrile (PAN), polytetrafluoroethylene (PTFE),
aramid" etc. are well established fibers. Metallic filaments such as steel,
aluminum, copper, tungsten, molybdenum, gold, silver, etc have been available
for a long time. Among ceramic and glass fibers, glass fiber for reinforcement
polymeric materials has been available since the 1940s; ceramic fibers such as
carbon, silicon carbide, alumina, etc. became available in the 1960s and
onward.
To begin with, we define some commonly used terms in the field of

reinforcements. Some of these definitions are merely operational definitions
while other terms are well entrenched in the textile industry. The aspect ratio
of a fiber is the ratio of its length to diameter (or thickness). A fiber can be
defined as an elongated material having a more or less uniform diameter or

thickness of less than 250 J.l.m and an aspect ratio of more than 100. Note that
it is not only an operational definition but also a purely geometrical one that
applies to any material. Let us now define some other terms related to
reinforcem ent.
FILAMENT: a continuous fiber (aspect ratio approaching infinity).

MONOFILAMENT: a continuous fiber of diameter greater than 100 J.l.m.
RIBBON: a fiber of rectangular cross-section with width to thickness ratio
greater than 4.
PARTICLE: a more or less equiaxial form, i.e., its aspect ratio is about 1.
STAPLE FIBER: fiber in the form of short, discrete lengths (10- 100 mm long).
ROVING: A bundle of yams or tows of continuous filaments (twisted or
untwisted).
Tow: A bundle of continuous (twisted or untwisted) fibers.
WIRE: a metallic filament.
WARP: The lengthwise yam in a woven fabric.
WEFT: The transverse yam in a woven fabric (also called FILL).
WHISKER: a tiny, whisker-like fiber, a few mm long, a few J.l.m in diameter, that
is a single crystal. The single crystal condition is a material requirement.
WOVEN FABRIC: A flat, drapable sheet made by interlacing yams or tows.
YARN: A generic term for a bundle of untwisted or twisted continuous fibers.
Finally, the following yam related terms are relevant:
Yarn Structural Parameters. A fiber yam is a multifilament assembly,

frequently employed in making woven forms. There are certain yam structural
parameters that are important in determining its properties.
Linear Density. This term gives us an idea ofthe yam size. Denier, commonly
used in the U.S, is the weight in grams of a yam 9000 m long. Tex, commonly
used in Europe, is the weight in grams of a yam 1000 long. Strictly speaking,
these linear density terms can be applied to single fibers as well. The basic idea
is that the larger the linear density (denier or tex), the heavier the yam. For
example, an individual fiber may be 1.5 den, while the fiber yam, may be
available in 200 den, 400 den, or all the way to 1500 den. Clearly, as the denier
(or tex) increases, the number of individual fibers in the yam increases.
Number of Fibers. This gives the number of individual filaments in the yam.
This parameter is really fixed by the number of holes in the spinneret.
Twist in the Yarn. Two types of twists can be given to the yarn: a
counterclockwise twist or S twist, and a clockwise twist or Z twist. We can
make a ply yarn by using reverse twist directions which serves to balance out
any residual stresses. We can also twist two or more plies together to make a
cord.
Commonly, the yarn designation on a fiber provides information such as name,

linear density, number of fibers, and fiber type. In the case of a woven fabric,
the warp yarns are the lengthwise fibers, while the fill or weft yarns are in the
transverse direction. Figure 3.1 shows these warp and fill (or weft) directions
in two types of weaves: plain and satin. A plain weave has warp ends going
alternately over and under a fill yarn (or weft) yarn. It is a firm construction,
that is, very little slippage occurs. The strength of the fabric is uniform in the
two orthogonal directions but there is a lot of porosity in plain weave. In the
satin weave, warp ends weave over a number of fill yarns and then under one
fill yarn. For example, a five-harness satin, shown in Fig. 3.1, has warp ends
going over four fill yarns and then under one. A satin weave is more pliable or
drape able than a plain weave and can take complex shapes and contours. It is
also less porous than a plain weave.
Fibers have some unique features which stem mainly from their small cross-
section and large aspect ratio:
• high degree of flexibility

• higher strength than the bulk material of the same composition.
The long length also makes it imperative, in most cases, to incorporate them in
some medium, called matrix, to hold them together, i.e., to make a fiber
reinforced composite. It should be emphasized that this, by no means, is the sole
purpose of matrix in a composite.
3.2 FIBER FLEXIBILITY
Flexibility is an important attribute of fine fibers. This is a very important

concept and worth discussing in some detail. A high degree of flexibility is
really a characteristic of a material having a small diameter and a low modulus
(Dresher, 1969). Flexibility of fibers permits a variety of techniques to be
employed for making composites with these fibers. It turns out that flexibility
of a given material is a function of its elastic modulus, E and the moment of
inertia of its cross-section, I. The elastic modulus of a material is quite
Warp
Fill or Weft
Plain Weave Satin Weave

Fig. 3.1 Warp and fill (or weft) directions in two types of weave: plain and satin.
independent of its form or size and is generally a material constant for a given
chemical composition and fully dense material. Thus, for a given composition
and density, the flexibility of a material is determined by its shape, or more
precisely by its diameter. The product of the bending moment (M) and the
radius of curvature (R) is called flexural rigidity. We can use the in verse of this
flexural rigidity as a measure of flexibility. From elementary strength of
materials, we have the following relationship for a beam bent to a radius R:
MIl =EIR
where E is Young's modulus of the material from which the beam is made and
the other terms are defined above. For a beam or a fiber of diameter, d, we have
I = d4/64. Substituting this in the above expression and rearranging the above
expression, we can write
MR = EI = E1td4/64
or,
IIMR = 641E1td4 (3.1)
where d is the equivalent diameter and I is the moment of inertia of the beam
(fiber). Equation (3.1) indicates that (lIMR), a measure of flexibility, is a very
sensitive function of diameter, d. Given a sufficiently small diameter, it is
possible to produce, in principle, a fiber as flexible as any from a polymer, a
metal, or a ceramic. In other words, one can make very flexible fibers out of a
ceramic such as silicon carbide or alumina provided one can make a fine
enough diameter. Producing a fine diameter ceramic fiber, however, is a
problem in ceramic processing.
3.3 CARBON FIBERS
Carbon is a very versatile element. It is very light, and has a theoretical density
of2.27 g cm-l, and exists in a variety of forms. Two most important forms of
carbon are diamond and graphite. Among the relatively new forms of carbon
one should mention the Buckminster Fullerene, also known as Buckyball, and
the drawn out version of Buckyballs, called variously as nanotubes or
nanowires. It is the graphitic form of carbon that is important in understanding
the carbon fibers. The hexagonal structure of carbon in the graphitic form is
shown in Fig. 3.2.
Carbon fiber is a generic name representing a family of fibers. Over the years,
it has become one of the most important reinforcement fibers in all kinds of
composites. It is unfortunate fact that the terms carbon and graphite are used
interchangeably in commercial practice as well as in some scientific literature.
Rigorously speaking, graphite fibers are a form of carbon fiber that is obtained
after heating to a temperature greater than 2400 °C. This process, called
graphitization, results in a highly oriented, layered crystallographic structure,
which, in turn, leads to significantly different chemical and physical properties
from non-graphitic forms of carbon. An extreme case of the graphite structure
is, of course, a graphite single crystal. Such a single crystal has a hexagonal
symmetry and consequently anisotropic characteristics. In particular, its elastic
properties will be transversely isotropic in the layer plane. A hexagonal
symmetry requires 5 independent elastic constants. The Young's modulus E(9)
as a function of angle e (i.e., the angle between the c-axis and the stress axis)
for a hexagonal crystal is given, in terms of the compliances S, by the following
expression:
Fig. 3.2 Hexagonal structure of graphite.

Table 3.2 gives the compliance (S) and stiffness (C) values ofa graphite single
crystal (Kelly, 1981). If we plug in these values in Eq. (3.3), we can obtain a
curve of the calculated tensile Young's modulus E as a function ofe (angular
displacement from the "a" axis) for a single crystal of graphite, Fig. 3.3. This
figure shows clearly the highly anisotropic nature of a graphite single crystal.
It should be clear to the reader that the more aligned the basal planes in a
carbon fiber, i.e., the more graphitic the structure, the higher the modulus in the
direction of the axis.
3.3.1 Fabrication of Carbon Fibers
Carbon fibers are fabricated by controlled pyrolysis of an organic fiber

precursor (Ezekiel and Spain, 1967; Watt and Johnson, 1969; Johnson and
Tyson, 1969 Watt, 1970; Diefendorf and Tokarsky, 1975; Singer, 1979).
Depending on the precursor and processing, one can get a variety of carbon
fibers with different strength and modulus. Important types are High Strength
(HS), High Modulus (HM), Intermediate Modulus (1M), Super High Modulus
(SHM), etc.
1200
a
w
0
a.
c.9
II)
::l
~
1000
800
1111
Basal Planes
Lc
::l
"'CJ 600
0
~
II)
~o> 400
C
::l
?- 200
0
0 20 40 60 80
Orientation, e degree
Fig. 3.3 Calculated tensile Young's modulus E as a function of e (angular displacement
from the "a" axis) for a single crystal of graphite.
Polyacrylonitrile (PAN) precursor fibers are more expensive than rayon.

Nevertheless, PAN is more commonly used because the carbon fiber yield is
about double from PAN than that from rayon. Pitch based carbon fibers are also
important for two reasons: i) pitch is potentially a very cheap raw material for
making carbon fibers and ii) the carbon fibers obtained from pitch can have
much superior properties than those obtained from other precursors.
Table 3.2 Compliance (S) and stiffness (C) values for a graphite single crystal (after
Kelly, 1981)
S, GPa-1 C,GPa
SIl 0.00098
= C II = 1060
S33 0.0275
= C33 = 36.5
S44 0.25
= C44 =4
SI2 = -0.00016 C I2 = 180
SI3 = -0.00033 C 13 = 15
There are certain essential steps common to all processes of carbon fiber
manufacture:
(a) Fiberization, i.e., extrusion of a polymer melt or solution into a precursor

fiber.
(b) Stabilization (oxidation or thermosetting) is done at relatively low
temperatures (200-450 0c), usually in air. This renders the precursor infusible
during the subsequent high temperature processing.
(c) Carbonization is carried out in an inert atmosphere (pure N2 generally) at
1000-2000 oCr At the end of this step the fiber has 85-99% carbon content.
(d) Graphitization (optional) is done in Ar or N2 at a temperature greater than
2500 oCr This step increases the carbon content to more than 99 % and imparts
a very high degree of preferred orientation to the fiber.
3.3.2 PAN-Based Carbon Fibers
The flexible polymeric chain structure of the PAN precursor fiber is shown in
Fig. 3.4a while the same polymer after stretching but without stabilization is
shown in Fig. 3.4b. On load removal, the structure shown in Fig. 3.4b collapses
to that shown in Fig. 3.4a. The more rigid, ladder structure obtained after
stabilization treatment is shown in Fig. 3.4c. A flow diagram showing the
f
(a)
(b) (e)
Fig. 3.4 Structural changes in PAN (a) precursor fiber (b) stretched precursor without
stabilization (c) after stabilization.
various processing steps for PAN-based carbon fiber is shown in Fig. 3.5. The
PAN precursor has a flexible polymer chain structure, but it has an all carbon
backbone chain, Fig. 3.6a, containing polar nitrile groups. During the
stabilization treatment, the PAN precursor fiber is heated to 200 - 220°C under
tension. During this treatment, the nitrile groups react to for a ladder structure,
which is a rigid and thermally stable structure, Fig. 3.6b. Also, when PAN is
Fig. 3.5 Flow diagram for the fabrication of PAN-based carbon fiber.
heated in air at 220 0 C, the absorbed oxygen cross links the chains, the fibers turn
black, and a stable ladder structure is formed. This treatment, done under
tension, helps in maintaining the orientation of the ring structure. During the
carbonization treatment between 1000 - 1500 °C, well developed hexagonal
networks of carbon form and a considerable amount of gases are evolved. This
gas evolution is partly responsible for some crack formation in the carbon fiber,
resulting in a lower tensile strength.
Commercially produced carbon fibers invariably have a size, i.e., a protective

surface coating. There are two main reasons for this: (i) ease of hand ling, and (ii)
improved adhesion with the polymeric matrix. The uncoated carbon fibers pick
up surface charge easily when they come in contact with rubbing surfaces such
as rollers, pulleys, guides, spools, etc. Handling such fibers (winding, weaving,
or braiding) can cause fiber breakages and tiny fragments of carbon fibers can
become airborne and short electrical machinery. In view of the fact that the
major market for carbon fibers is in polymer composites, it is understandable
that commonly a low molecular weight epoxy based coating (without the
hardener) is used. An oxidative surface treatment is also given
which produces a rough, etched surface having reactive oxide sites that aid in
mechanical and chemical bonding with a polymeric matrix. Clearly, before using
in a ceramic matrix, one must burn off any size on carbon fibers.
CN--CH2
/
CN--CH
'/" CH 2
CN--CH
'/" CH 2
'/" CH2
(a)
o 0
CH, C
I CH, C
I CH,
~/~/~/~/
CH CH CH CH
I
C
I
C
I
C
I
C
/ ~N/ ~N/ ~N/ ~
(b)
Fig. 3.6 (a) The carbon backbone chain structure of PAN (b) the ladder structure of
PAN after stabilization.
3.3.3 Pitch-Based Carbon Fibers
Pitches form an important and low cost raw material for producing carbon fibers.
There are three common sources of pitch:
• petroleum asphalt
• coal tar
• polyvinyl chloride (PVC)
Pitches are thermoplastic in nature and are difficult to carbonize without being
first stabilized against fusion during pyrolysis. A schematic of the process of
making carbon fibers from a pitch is shown in Fig. 3.7. It involves the following
steps:
• extrude or melt spin into a fibrous form

• stabilize between 250°C and 400 °C
• carbonize
• graphitize
Spinnability of the pitch and its conversion into a nonfusible state are the most
important steps (Diefendorf and Tokarsky, 1975). These characteristics depend
on the chemical composition and molecular weight distribution of the pitch. The
molecular weight (MW) of the pitch controls the viscosity and the melting range,
i.e., MW controls the temperature and speed of spinning. The pitch composition
depends on its source. In fact, there is tremendous variability in the composition
of a pitch because it can be a mixture of hundreds of different species, varying
with the source of crude oil and the process conditions in the refinery. Indeed,
this is one of the main problems in the manufacture of pitch-based carbon fibers.
Generally, the pitches are made of different organic compounds containing
groups of condensed benzene-ring systems separated by and carrying alkyl
chains. Petroleum and coal-tar pitches may also contain large amounts of sulfur.
PVC pitch is also a mixture of several different organic compounds - polynuclear
aromatic compounds. These compounds have 3 or 4 aromatic nuclei, i.e.,
consisting of 3 or 4 rings per molecule. One suggested formula is C62 H52 which
is obtained by thermal degradation of PVC at 400°C under nitrogen for 30
minutes, the yield being 20%. The melting range is 150-200 °C and the
molecular weight is between 700-800.
Suitability of pitch for conversion to carbon fiber depends on a number of

factors. According to Singer (1979), the pitch should have a high carbon content
(> 90%), high aromatic content (> 50%), low impurities and molecular weight
Refine
Adjust Mol. Wt and
Chemical Composition
Thermal Degradation oxidation
stress
graphitization
Fig.3.7 Schematic of the process of making carbon fiber from a pitch.
and molecular weight distribution, viscosity and rheological characteristics

consistent with the ease of spinning, followed by an easy conversion to the
infusible state.
3.3.4 Spinning and Rheology of Mesophase Pitch
Although melt spinning is commercially popular, centrifugal spinning and jet

spinning can also be used. Mesophase pitch, a thermoplastic, has a nematic
liquid crystal structure, i.e., its molecules are rigid rodlike. The two-phase
precursor pitch is agitated prior to spinning to form a homogeneous mixture and
spun into filaments in the temperature range where the viscosity is in 1-20 Pa.s
range. Fibers can be spun at speeds of3-100 m min- l with diameters of 10-20
Jlm and having the same composition as the mesophase pitch. The as-spun
mesophase fibers are anisotropic because of the nematic liquid crystal structure.
These fibers have large, elongated, anisotropic domains (about 4Jlm diameter)
aligned along the fiber axis and are thermoplastic in nature. These anisotropic
domains can easily be viewed under polarized light. Fibers drawn from isotropic
pitches or the ones drawn from rayon ar acrylic (PAN) precursor do not show
such anisotropic domains. Oxidation treatment is given next to stabilize
thermally against internal relaxation and rendering the fibers infusible. The rest
of the process is essentially similar to the PAN-based carbon fiber fabrication,
except that stress is applied during graphitization.
3.3.5 Structure and Properties of Carbon Fibers
Table 3.3 presents property data for three different types of PAN-based carbon
fibers, while Table 3.4 gives the properties of mesophase pitch-based carbon
fibers. Notice the high density and high modulus of pitch-based fibers compared
to PAN-based fibers. The very high temperature treatment of graphitization (to
a temperature as high as 3000 0c) increases the degree of order in carbon fibers.
This is accompanied by a large increase in the elastic modulus ofthe fiber. The
tensile strength of PAN-based fibers when subjected to high temperature
treatment, however, falls, Fig. 3.8 (Watt 1970). This is attributed to the presence
of discrete flaws on the fiber surface and within it. Most of the volumetric
defects in carbon fibers originate from:
(i) inorganic inclusions

(ii) organic inclusions
(iii) irregular voids from rapid coagulation
(iv) cylindrical voids precipitated by dissolved gases.
Table 3.3 Properties of PAN-based carbon fiber (strand data) (after Riggs, 1985)
Characteristic High Super High High

Strengtlf Strength b Modulus c
Filament diameter, !lm 5.5-8.0 5.4-7.0 8.4
Density, g/cmJ 1.75-1.80 1.78-1.81 1.96
Carbon content, wt% 92-95 99-99+ 99+
Tensile strength, MPa 3100-4500 2400-2550 1865
Tensile modulus, GPa 225-260 360-395 520
Strain at fracture, % 1.3-1.8 0.6-0.7 0.38
Electrical resistivity, !lOrn 15-18 9-10 6.5
Thermal conductivity, W(mKY' 8.1-9.3 64-70 -120
• Thomel T-300, T-500, T-600, T-700; Celion 3000, 6000,1200; AS2, AS4, AS6,
IM6
b Thomel T-50, Celion G-50, HMS
c Celion GY-70
Table 3.4 Properties of mesophase pitch-based carbon fiber (after Singer, 1981)
Property Thorne/ Thorne/ Thorne/

P555 P755 PlOO
Filament diameter, 11m 10 10 10
Density, glcm3 2.02 2.06 2.15
Carbon content, wt. % 99 99 99+
Tensile strength, MPa 1895 2070 2240
Tensile modulus, GPa 380 517 690
Strain at fracture, % 0.5 0.4 0.3
Electrical resistivity, l1ilm 7.5 4.6 2.5
Thermal conductivity, W(mK)"1 110 185 515
These defects get transformed during the high temperature treatment into diverse
imperfections. Basal-plane cracks called Mrozowski cracks are perhaps the most
important flaw type that limit the tensile strength of carbon fibers. These occur
as a result of anisotropic thermal contractions within the ribbon structure on
cooling from high temperature treatment (> 1500 DC). These cracks are generally
aligned along the fiber axis. Their presence lowers the tensile strength of the
fiber by providing easy crack nucleation sites. The fiber elastic modulus,
however, is unaffected because the elastic strains involved in the modulus
400
o 0
1200 1400 1600 1800 2000 2200 2400 2600
Heat trea1ment temperature, °c
Fig. 3.8 Elastic modulus and tensile strength of carbon fiber as a function of high
temperature treatment (after Watt, 1970).
measurement are too small. Surface flaws can also limit the tensile strength of
the carbonized fibers. Oxidation treatments tend to remove the surface defects
and thus increase the strength levels ofthe fiber.
3.3.6 Structure of Carbon Fibers
Many researchers have studied the structure of carbon fibers; see, for example,
Perret and Ruland, 1970; Bennett and Johnson, 1978, 1979; Inal et aI., 1980;
Reynolds and Sharp, 1974; Fourdeux,et aI., 1971; Riggs, 1985; Singer, 1981;
Bennett et aI., 1983. A scanning electron micrograph of pitch-based carbon
fibers is shown in Fig. 3 .9a while an atomic force microscope picture of the same
fibers is shown in Fig. 3.9b. Note the surface striations and the roughness on a
microscopic scale. When viewed in a transmission electron microscope, the
heterogeneous microstructure of carbon fibers becomes clear. In particular, the
rather pronounced irregularity in the packing of graphitic lamellae from the fiber
surface inward can be seen. The basal planes are much better aligned in the near
surface region of the fiber. In very general terms, the graphitic ribbons are
oriented more or less parallel to the fiber axis with random interlinking oflayers,
longitudinally and laterally. Figure 3.1 Oa shows a two-dimensional
representation of this lamellar structure. Deurberger and Oberlin (1991) studied
the structure of PAN-based carbon fibers obtained from several stabilized
precursors and carbonized in under various conditions and temperatures. An
example of the structure ofa carbon fiber from their work is shown in Fig. 3.10b.
This is a high resolution electron micrograph of a longitudinal section of a fiber
carbonized at 1330 °C. The micrograph shows 002 lattice fringes which
represent the graphitic planes of the carbon fiber. The lamellar structure of the
carbon fiber is clear. A schematic of the three-dimensional structure of PAN-
based carbon fiber is shown in Fig. 3.11 (Fourdeux et aI., 1971). Note the
distorted carbon layers and the rather irregular space filling. The degree of
alignment of the basal planes increases with the final heat treatment temperature.
3.4 OXIDE FIBERS
Ceramic oxide fibers, continuous and discontinuous, have been commercially

available since 1970s. We describe below important aspects of processing and
microstructure of these fibers.
3.4.1 Alumina Type Oxide Fibers
Alumina can have y, 0, 11, and a allotropic forms. a-alumina is the

Fig. 3.9 (a) SEM micrograph of pitch-based carbon fibers (b) An atomic force
micrograph (AFM) ofa pitch-based carbon fiber (courtesy ofK. Upadhya).
thermodynamically stable form. In practice, it is very difficult to obtain the

precise time and temperature conditions to proceed from y to a. Among other
problems, internal porosity trapped by grain boundary migration at high
processing temperatures is difficult . to eliminate completely. Among the
desirable characteristics one would like to have in any ceramic fiber for load-
bearing structural applications are:
• high theoretical density, i.e., low porosity

• small grain size for low temperature applications
• large grain size for high temperature applications
• high purity
(a'
(b)
Fig. 3.1 0 (a) Two-dimensional representation of the lamellar structure of carbon (b) high
resolution electron micrograph of a carbon fiber showing lattice fringes (courtesy of A.
Oberlin).
1--------- ... 8 Jim - - -- -"'"'1
Fig. 3.11 Schematic of the three dimensional structure of PAN-based carbon fiber (after
Fourdeux et ai., 1971).
One should note that a low firing or sintering temperature will give a smaller
grain size but at the same time lead to an unacceptable level of residual porosity.
Should one go to higher processing temperatures, one can eliminate porosity but
excessive grain growth will result because of the high temperatures involved.
One can avoid this dilemma by introducing a second phase that restricts grain
boundary mobility while the porosity is removed at high temperatures. In
principle, it is possible to choose the type and amount of the second phase that
inhibits the grain growth at the service temperature. One way to use oxides of
silicon, phosphorus, boron, or zirconium as grain growth inhibitors but they
generally lower the working temperature. Among the examples are: a-alumina
plus 15-20% zr02 , 8-alumina plus about 4% of Si02 , and a-alumina plus 0.4%
Fe20 3 and 0.25% Si02 • The reader should note that for high creep resistance, one
requires a large grain size because grain boundary related phenomena such as
grain boundary sliding can lead to a rather large creep strain in a fine-grained
material.
Many companies have the capabilities of producing polycrystalline alumina

fibers of different types. Du Pont produced, in 1980s, two alumina type
continuous filaments: a polycrystalline a-alumina yarn called FP, and an a-
alumina + 15-20% zr02 yarn, called PRD-166. Sumitomo Chemical Co.
produces a fiber that can have a range of composition: 70 -100% Al20 3 and
30-0% Si02 • A short fiber, called staple fiber, of8-alumina(96%) (trade name
Saffil) is available commercially. Single crystal continuous aluminum oxide or

sapphire fibers can be produced by drawing from molten alumina. A fiber
produced by this method, called Saphikon, has a hexagonal structure with its c-
axis parallel to the fiber axis, i.e., the basal plane (0001) is perpendicular to the
fiber axis. The diameter is rather large, between 75- 250 flm. We describe below
the salient features of some of these fabrication methods and the properties of
fibers obtained.
(a) a-Alumina Fiber
Continuous a-alumina fiber and an (a-alumina + zirconia) fiber were made by

spinning of a viscous solution by researchers at Du Pont. Although, Du Pont
does not produce these fibers any more, the fabrication of these fibers
represented an important step in the processing of alumina type fibers. The
fabrication procedure for the a-alumina fiber, called FP fiber, involved dry
spinning of an aqueous slurry mix of alumina particles and additives (Dhingra,
1980). The dry spun yarn was subjected to a two-step firing. The first step oflow
firing controls the shrinkage, while in the second step, called flame firing,
improves the density of a-AIP3. A thin silica coating, applied to the surface of
this fiber, can serve to heal the surface flaws, giving about 50% higher tensile
strength than the uncoated fibers. Figure 3.12, a TEM micrograph, shows the
alumina grains in the FP fiber. A modification of the FP alumina, called PRD-
166 fiber, was also made by DuPont (Romine, 1987). PRD-166 was an a-
alumina fiber, about 20 flm in diameter, containing 15 - 20 weight % yttria
stabilized zirconia particles. PRD-166 fiber had a rough surface and an average
grain size of about 0.5 flm. The zirconia particles were about 0.1 flm and located
mostly at grain boundary triple points. Their function was to inhibit grain growth
in alumina. Figure 3.13a shows the microstructure of the PRD-166 fiber as seen
in a scanning electron microscope while Fig. 3.13b shows the microstructure of
the PRD-166 fiber as seen in a transmission electron microscope (Nourbakhsh
et aI., 1989).
Minnesota Mining and Manufacturing Co. (3M Co.) has developed a series of
alumina +silica fibers via sol-gel route. In particular, a pure a-alumina fiber has
been commercialized under the trade name Nextel 610. The sol-gel process of
making fibers involves the following steps common to all sol-gel processing:
(a) formulate sol

(b) concentrate to form a viscous gel
(c) spin the precursor fiber
(d) calcine to obtain the oxide fiber
E
::i..
C)
.....
(a)
(b)
Fig. 3.13 (a) Microstructure ofan (a-alumina + zirconia) or PRD-166 fiber as seen in a
scanning electron microscope. Zirconia particles (white) are located mostly at the grain
boundary triple points. (b) Microstructure of PRD-166 fiber as seen in a transmission
electron microscope.
Specifically, in the case ofthe 3M process of making AIP3 fiber, the following
steps are involved:
• Use an organic basic salt solution as a precursor.

• Drive out (decompose and volatilize) the organics without causing cracking,
blistering, or other defects.
• Fire at 1400 °C under carefully controlled conditions.
• A low temperature straightening treatment.
Figure 3.14 shows the 3M process schematically. A fine-grained a-AIP3 fiber

is obtained by seeding the high temperature a-alumina with a very fine hydrous
colloidal iron oxide (Wilson, 1990). The fine iron oxide improves the nucleation
rate of a-AIP3' with the result that a high density, ultrafine, homogeneous a-
AIP3 fiber is obtained. The rational for seeding with iron oxide as follows.
Basic salts of aluminum decompose into transition aluminum oxide spinels such
as TJ-AI20 3above 400°C. These transition cubic spinels convert to hexagonal a-
Al 20 3 on heating between 1000 to 1200 °C. The problem is that the nucleation
rate of pure a-A120 3 is too low and results in large grains. Also, during the
transformation to a phase, large shrinkage results in a rather large porosity
(Kumagai and Messing, 1985; Suwa et aI., 1985).
Seeding of alumina with fine particles would appear to be a solution. a-FeP3 is

isostructural with a-AIP3' only 5.5% lattice mismatch (Wilson, 1990). Hydrous
colloidal iron oxide sol appears can be an efficient nucleating agent. According
to Wilson (1990), without the seeding of iron oxide, the TJ-alumina to a-alumina
transformation occurs at about 1100 °C. With 1% Fe20 3, the transformation
temperature was decreased to 1010 °C, while with 4% Fe20 3, the transformation
temperature came down to 977 °C. Concomitantly, the grain size was refined.
Nextel610 fiber has 0.4 - 0.7% FeP3' Besides Fe20 3, about 0.5 wt. % Si02 is
added to reduce the final grain size, although Si02 inhibits the transformation to
the a phase. The Si02 addition also reduces grain growth during soaking at 1400
°C.
Many other alumina or alumina-silica type fibers are available, most of which
are made by the sol-gel process. Sumitomo Chemical company produces a fiber
that is a mixture of alumina and silica. The flow diagram of this process is
shown in Fig. 3.15. Starting from an organoaluminum (polyaluminoxanes or a
mixture of polyaluminoxanes and one or more kinds of Si containing
compounds), a precursor fiber is obtained by dry spinning. This precursor fiber
is calcined to produce the final fiber. The fiber structure consists of fine
Sol Reservoir
Filter
Pump
Spinneret
Pyrolysis Furnace
Fig. 3.14 3M process of making AlP3 fiber.
crystallites of spinel. Si02 serves to stabilize the spinel structure and prevents it
from transforming to a-AIP3 (Chawla, 1998). Mention has been made of the
Nextel series of fiber produced by the 3M Company. These are mainly Al 20 3+
Si02 and some B 20 3. The composition and properties of these fibers are given
in Table 3.5. The sol-gel manufacturing process used by 3M Co. has metal
alkoxides as the starting materials. Metal alkoxides are M(OR)n type compounds
where M is the metal and n is the metal valence, and R is an organic compound.
Selection of an appropriate organic group is very important. It should provide
sufficient stability and volatility to the alkoxide so M - OR bonds are broken and
MO- R is obtained to give the desired oxide ceramics.
Hydrolysis of metal alkoxides results in sols which are spun and gelled. The
gelled fiber is then densified at relatively low temperatures. The high surface
free energy available in the pores of the gelled fiber allows for densification at
a relatively low temperature. The sol-gel process provides a close control over
solution composition rheology of fiber diameter. The disadvantage is that rather
large dimensional changes must be accommodated and fiber integrity conserved.
Table 3.5 Properties of some oxide fibers·
Tensile Young's
Fiber Type Composition, Diameter, Density, Strength, Modulus,
wt.% Jjm g cm- 3 MPa GPa
Nextel312 AI20 3-62.5, Si0 2- 10-12 2.70 1700 150

24.5, B 20 3-13
Nextel440 AI20 3-70, Si0 2-28, 10-12 3.05 2000 190

BPr 2
Nextel550 AIP3-73, Si02-27 10-12 3.03 2000 193
Nextel610 A1 2Or 99+ 10-12 3.9 3100 370
Nextel AlP3-89, Zr0 2-1O, 10-12 4.10 2550 358
650 Y20 3-1
Nextel Ah03-85, Si02-15 10-12 3.40 2100 260

720
Saffil AI 20 3-96, Si0 2-4 3 2.3 1000 100
Saphikon Single Crystal AlP3 70-250 3.8 3100 380
Sumitomo A1 20 r 85, Si0 2-15 9 3.2 2600 250
·Manufacturer's data
In the process used by 3M Co. for making the Nextel oxide fibers, aluminum
acetate [Al(OH)2 (OOCCH3) ' 1I3H3B03], e.g., Niaprooj, from Niacet Corp., is
the starting material (Sowman, 1988). Aluminum acetate having an A120/B 20 3
ratio of3 to 1 becomes spinnable after water removal from an aqueous solution.
In the fabrication of3M continuous fibers, a 37.5 % solution of basic aluminum
acetate in water is concentrated in a rotating flask partially immersed in a water
bath at 32 - 3 6 °C. After concentration to an equivalent A120 3 content of28.5 %,
a viscous solution with viscosity, 11, between 100-150 Pa s is obtained. This is
extruded through a spinneret having 130 holes of 100 flm diameter under a
pressure of 800-1 000 kPa and a shiny, colorless fibers are obtained on firing to
1000 °C. The microstructure shows cube-shaped and lath-shaped crystals. The
boria addition lowers the temperature required for mullite formation and retards
the transformation of alumina to a-A120 3. One needs boria in an amount
equivalent to or greater than a 9A120 3:2B 20 3 ratio in A1203-B203-Si02
compositions to prevent the formation of crystalline alumina.
A sol-gel method is also used to produce silica-stabilized alumina (Saffil) and

calcia-stabilized zirconia fibers (Birchall et aI., 1985). The saffil fiber is a 0-

A1 20 3, short staple fiber that has about 4% Si02 and a very fine diameter (3 J..lm).
The aqueous phase contains an oxide sol and an organic polymer. The sol is
extruded as filaments into a coagulating (or precipitating) bath in which the
extruded shape gels. The gelled fiber is then dried and calcined to produce the
final oxide fiber. For alumina, aluminum oxychloride [AI 2(OH)sCI] is mixed
with a medium molecular weight polymer such as 2 wt. % polyvinyl alcohol.
This solution is slowly evaporated in a rotary evaporator until a viscosity of
about 80 Pa.s is attained. It is then extruded through a spinneret, the fibers are
wound on a drum and fired to about 800°C. The organic material is burned away
and a fine-grained alumina fiber having 5- 10% porosity and a diameter of 3 - 5
J..lm is obtained. The fibers produced at this stage are suitable for filter purposes
because of their high porosity. By heating them to 1400- 1500 °C, which causes
a 3-4% of linear shrinkage, one obtains a refractory alumina fiber suitable for
reinforcement purposes.
Yet another continuous, polycrystalline, a-alumina fiber, trade name Almax,

is prepared by dry spinning a viscous slurry consisting of an aluminum salt, a
fine powder of intermediate alumina, and an organic binder to produce the
precursor fiber; this is followed by prefiring (calcining) and firing (sintering) the
precursor fiber to produce an alumina fiber. Table 3.6 gives typical properties
the Almax fiber. Note that the density is less than the theoretical density of
alumina.
Polycrystalline oxide fibers of yttrium-aluminum garnet, Y3AIs0 12 (YAG)

composition and a composite fiber consisting of a-alumina and YAG were
prepared by sol-gel method (Towata et aI., 2001). The process involved the use
of a-alumina or YAG seed particles. a-alumina seed particles accelerated the
phase transformation from a-alumina to a-alumina. The YAG seed particles
influenced the multi-step transformation of yttrium and aluminum oxide.
A technique called Edge-defined Film-fed Growth (EFG) has been used to make
continuous, monocrystalline sapphire (AI 20 3) fiber(LaBelle and Mlavsky, 1967;
LaBelle, 1971; Pollack, 1972; Hurley and Pollack, 1972; Gasson and
Cockayne,1970). LaBelle and Mlavsky (1967) were the first ones to grow
sapphire (AIP3) single crystal fibers using a modified Czochralski puller and
radio frequency heating. In 1971, these authors devised a growth method, called
the edge- defined, film-fed growth (EFG) method. Figure 3.16 shows a
schematic of the method. Growth rates as high as 200 mm/min have been
attained. The die material must be stable at the melting point of alumina;
molybdenum die is used. A sapphire seed crystal is used. A capillary supplies a
IOrganoalumino compound I
•
Alkyl aluminum
or
AIkoxy aluminum
(AlR3)
t1-ot
Polymerization
AIR, + H20 -
.___ Organic solvent + Si-containing

compound (alkyl silicate)
IDry spinning I
•
Precursor fiber
(organoaluminum polymer
and alkyl silicate)
I Calcination I
•
Fig. 3.15 Flow diagram of the Sumitomo process for making a mixture of alumina and
silica fiber.
constant liquid level at the crystal interface. Molten alumina wets both,
molybdenum and alumina. The crystal grows from a molten film between the
growing crystal and the die. The crystal shape is defined by the external shape
of the die rather than the internal shape. Perhaps, the most important item about
the Saphikon fiber is its single crystal nature, i.e., there are no grain boundaries.
Thus, under creep conditions involving grain boundary related phenomena such
cavitation and grain boundary sliding will not be operative and one can expect
Table 3.6 Properties of Almax fiber
Chemical Composition AIP3 > 99.5 wt%

Crystal Structure a-AIP3
Color white
Diameter lOflm
Number of filaments 1000 filaments/yam
Density 3.6 g cm-3
Tensile strength 1.80Pa
Elastic modulus 3200Pa
a creep resistant fiber. However, single crystal alumina can undergo dislocation
creep on basal plane. It would thus appear that if one can orient the basal plane
such that basal slip does not occur then, one can obtain a creep resistant fiber_
Such will be the case for a single crystal fiber with its c-axis parallel to the fiber
aXIs.
A laser-heatedfloating zone method can be to make a variety of ceramic fibers.

Gasson and Cockayne (1970) used laser heating for crystal growth of A1 20 3,
YP3' MgA1204' and Naz03' This method has been used to grow single crystal
fibers of AIP3' Y P3' TiC, TiB2' mullite, AI 20/YAG eutectic (Haggerty, 1972;
Sayir and Farmer, 1995, Sayir et aI., 1995). A CO2 laser is focused on the molten
zone and a source rod is brought into the focused laser beam. A seed crystal,
dipped into the molten zone, is used to control the orientation. Crystal growth
starts by moving the source and seed rods simultaneously. Mass conservation
dictates that the diameter is reduced as the square root of the feed rate/pull rate
ratio. Mention should be made of a containerless melting technique has been
used to grow continuous fibers of alumina and YAG-composition (Y3AIs0 12 )
directly from the melt. Containerless melting eliminates heterogeneous
nucleation by container surfaces. A continuous - wave CO2 laser beam is used
for heating (Weber et aI., 1998). Specimens are levitated in a gas jet and
stabilized with an acoustic positioning device. The levitated samples are heated
and melted with a CO2 laser beam.
3.5 NONOXIDE FIBERS
Continuous nonoxide ceramic fibers are also available commercially. Silicon

carbide fiber must be regarded as a major development in the field of ceramic
Temperalllte
M ell ing
Point
Molybde num Crucible
Fig. 3.16 Schematic of edge-defined, film-fed, growth process of making a single crystal
alumina fiber.
reinforcements during the last quarter of the twentieth century. In particular, a

process, developed by the late Professor Yajima in Japan, involving a controlled
pyrolysis of a polycarbosilane precursor to yield a flexible fine diameter fiber
must be considered to be the harbinger of the making ceramic fibers from
polymeric precursors. In this section we describe the processing, microstructure,
and properties of silicon carbide and some other nonoxide fibers.
3.5.1 Silicon Carbide Fibers
We can easily classify the fabrication methods of silicon carbide as conventional

and non-conventional. The former category would include the chemical vapor
deposition while the latter would include controlled pyrolysis of polymeric
precursors. Yet another important type of SiC available for reinforcement
purposes is SiC whiskers. We give a brief description of these.
(a) Silicon Carbide Fibers made by CVD
Silicon carbide fibers can be made by chemical vapor deposition on a substrate

heated to about 1300 °C (DeBolt et al., 1974). The substrate can be tungsten or
carbon. The reactive gaseous mixture contains hydrogen and alkyl silanes.
Typically, a gaseous mixture consisting of 70% hydrogen and 30% silanes is
introduced at the reactor top, where the tungsten substrate (-13 11m diameter)
also enters the reactor. A detailed schematic of the process used to make such a
fiber, trade name sigma fiber, is shown in Fig. 3.17. Mercury seals are used at
both ends as contact electrodes for the filament. The substrate is heated by
combined direct current (250 rnA) and very high frequency (VHF ~ 60 MHz) to
obtain an optimum temperature profile. To obtain a 100 flm to SiC
monofilament, it generally takes about 20 s in the reactor. The filament is wound
on a spool at the bottom ofthe reactor. The exhaust gases (95% ofthe original
mixture + HCI) are passed around a condenser to recover the unused silanes. An
efficient reclamation of the unused silanes is very important for a cost effective
production process. Such CVD processes result in composite monofilaments
which have built in residual stresses. The process is, of course, very expensive.
Methyltrichlorosilane is an ideal raw material for this process as it contains one
silicon and one carbon atom, i.e., a stoichiometric SiC will be deposited. The
chemical reaction is:
An optimum amount of hydrogen is required. If the hydrogen is less than

sufficient, chlorosilanes will not be reduced to Si and free carbon will be present
in the mixture. If too much hydrogen is present, an excess Si will be present in
the end product. The final monofilament (100 -150 flm) consists of a sheath of
mainly P-SiC with some a-SiC on the tungsten core. The {Ill} planes in SiC
deposit are parallel to the fiber axis.
A series of surface modified silicon carbide fibers, called SCS fibers, has been
used in a variety of ceramic matrix composites. These special fibers have a
complex through the thickness gradient structure. SCS-6, for example, is a thick
fiber (diameter = 142 flm) and is produced by chemical vapor deposition of
silicon and carbon containing compounds onto a pyrolytic graphite coated
carbon core. The pyrolytic graphite coating is applied to a carbon monofilament
to give a substrate of3 7 flm. This is then coated with SiC by CVD to give a final
monofilament of 142 Ilm diameter. The surface modification of the SCS fibers
consists of the following. The bulk of the 1 flm thick surface coating consists of
C doped Si. Figure 3.18 shows schematically the cross-sections of the two SCS
type silicon carbide fibers and their characteristic surface compositional
gradients. In SCS-6 fiber, zone I at and near the surface is a carbon rich zone. In
zone II, Si content decreases, and in zone ill, the Si content increases back to the
stoichiometric SiC composition. Thus, the SCS-6 silicon carbide fiber has a
surface graded outward to be carbon rich. Mann et al. (1999) verified the
dramatic changes in the mechanical characteristics such as Young's modulus and
hardness of SCS-6 fiber as function of fiber radius using nanoindentation
techniques. Besides SCS-6 there are two other varieties of CVD silicon carbide
fibers on a carbon core that are available commercially. Typical properties of
WFilament
@ Hydrogen
supply
Exhaust 1 - -.....-.---11 Flow I

1 meters
r
IScrubber I
--
l
l LI Silane
~I vaporizer
r l
l
IRefrigeration ~---,
r
I Silane l~_--,
supply I
IDistillation I
Photo-
Gases/ • optical
for - - diameter
recuperation~ sensor
Waste
by-products ~iC/w
Fig. 3.17 A schematic of the process of making silicon carbide monofilament by CVD.
(The trade name is sigma fiber and the substrate is tungsten.)
SCS-type silicon carbide fibers are given in Table 3.7. Another SiC fiber made
by CVD is called the sigma fiber; this has core of tungsten filament. Such fibers
made by CVD on a heated substrate are a kind of composite fibers, and can show
unusual behavior. Lara-Curzio and Sternstein(1993) examined the behavior of
such composite fibers when subjected to thermomechanicalloading. Their main
conclusion was that one must take into account the strains experienced by the
substrate during the CVD process of making such composite fibers. Not doing
Table 3.7 Properties ofSCS-type fibers'
Proe.erD!. SCS-6 SCS-9A SCS-Ultra

Filament diameter, !lm 140 78 140
Density, g/cm3 3.0 2.8 3.0
Tensile strength, GPa 3.45 3.45 5.86
Tensile modulus, GPa 380 307 415
Coefficent of thermal expansion, 4.1 4.3 4.1
1O- 6 K- 1
• Source: Specialty Materials Inc.
so can lead to a severe underestimate of the residual stresses in the fiber. In

particular, in the case of an SCS-6 fiber, they identified the presence of large
residual radial stresses at various interfaces (carbon substrate/pyrolytic graphite,
pyrolytic graphite/SiC) with structural defects in the fiber. According to these
authors, the origin of rather large residual stresses was in the high temperature
of CVD and the large anisotropy in the pyrolytic graphite layer. They also
observed an anomalous thermal expansion behavior in the SCS-6 fiber in the
temperature interval 20-1500 °C. The SCS-6 fiber showed a thermal hysteresis
phenomenon, a sudden shrinkage at 1375 °C during heating and a comparable
expansion during cooling at 1307 °C. Interestingly, they observed that the
temperatures at which these transitions occurred as well as the size of the
hysteresis loop depended on the magnitude ofheatinglcooling rate. These results
were explained by the authors in terms of a thermomechanical composite sphere
assembly model (see chapter 6) based on the melting of a collection of silicon
particles dispersed in an SiC matrix. The silicon particles are believed to be
present in the SiC sheath of the fiber as non-stoichiometric products from the
CVD process.
(b) Nonoxide Fibers Via Polymers
As pointed out above, the SiC fibers obtained via CVD are very thick and not
very flexible. Work on alternate routes of obtaining fine, continuous, and
flexible fiber had been in progress for sometime when in the mid-nineteen
seventies the late Prof. Yajima and his colleagues (1976, 1980) in Japan
developed a process of making such a fiber by controlled pyrolysis of polymeric
precursor. This method of using silicon based polymers to produce a family of
ceramic fibers having good mechanical properties, good thermal stability, and
oxidation resistance has an enormous potential. Figure 3 .19 shows a general flow
SCS-6 (- 140 !lID) SCS-9 (75 !-lm)
Pyrolytic graphite
coated carbon core ~
- --
""-
Inner zone: ""-~
D
--- carbon-rich ~ - SiC
Outer zone:
Stoichiometric ~ - SiC
~
~_
/
/
/
Carbon-rich surface /
coating (0-4!Am) .-/
SCS-6 fiber
u 1.0
::;:..
C".f.l
~-SiC
Distance from
the surface, !-tm
Composition of the surface region

Fig. 3.18 Schematic of two SCS-type silicon carbide fibers and their characteristic
surface compositional gradients.
diagram for making ceramic fibers from a polymeric precursor. Note the
similarity with the process of making a carbon fiber from a polymeric precursor.
The various steps involved in this polymer route can be listed as follows (Wax,
1985):
(a) polymer characterization (yield, molecular weight, purity, etc.)
(b) melt spin polymer into a precursor fiber
(c) cure the precursor fiber to crosslink the molecular chains, making it infusible
during the subsequent pyrolysis.
(d) pyrolyze the precursor fiber under controlled conditions to obtain the ceramic
fiber.
Polymeric Precursor
Melt or so lution
spinni ng
"
Precursor Fiber I
Curin 9
"
Cured or Stabilized
Fiber
Controlled Pyrolysis
"
Ceramic or Glass
Fiber
Fig. 3.19 Flow diagram for making ceramic fibers from a polymeric precursor.
Specifically, the Yaj ima process of making SiC fiber from a polymeric precursor
fiber involves the following steps and is shown schematically in Fig. 3.20.
Polycarbosilane, a high molecular weight polymer, containing Si and C is
synthesized. This involves the use of a commercially available material, viz.,
dimethylchlorosilane. Solid polydimethylsilane is obtained by dechlorination of
dimethylchlorosilane by reacting it with sodium. Polycarbosilane is obtained by
thermal decomposition and polymerization of polydimethyl silane. This is carried
out under high pressure in an autoclave at 470°C in an argon atmosphere for
8-14 h. A vacuum distillation treatment at up to 280°C follows. The average
molecular weight ofthe resulting polymer is about 1500. This is melt spun from
a 500 hole nozzle at about 350°C under N2 gas to obtain the so called pre-
ceramic continuous, precursor fiber. The precursor fiber is quite weak (tensile
strength ~ 10 MPa). This is converted to inorganic SiC by curing in air, heating
to about 1000 °C in N2 gas, followed by heating to 1300 °C in N2 under stretch.
During the pyrolysis, first stage of conversion occurs at around 550°C when
cross-linking of polymer chains occurs. Above this temperature, the sidechains
containing hydrogen and methyl groups decompose. Fiber density and
mechanical properties improve sharply. The conversion to SiC is occurs above
about 850°C.
CH, / Q
Si
CH3/ '0
Dichlorodimethylsi1ane
I Dechlorination
I
t-B
with Na (to NaCl)
. CH3 n
Polydimethy)silane
Polymerization at
I 470 OC in autoc:Iaft
CH3 H
t~-?1CH3 H n
Polyalrbosilane
Melt spinning at
I 3SOOC(N2l
Polycarbosi1ane
fiber
Curing 190 OC in air
r or Rf in ozone
Polycarbosi1ane fibers
with molecular c:ross-
linking by oxygen to
awid subsequent
melting
Pyrolysis heating to
I 1300 OC or less on in
vacuum (1000 0CIh)
SiC fiber
Amorphous or micro-
CI')'Stalline rt-SiC
Fig. 3.20 The Yajirna process of making fme diameter SiC from a polycarbosilane.
(c) Structure and Properties of Nicalon Fiber
TheNicalon multifilament fiber (1 0-20 ~m diameter) as commerciaUyproduced

consists of a mixture of P-SiC, free carbon, and Si02 • The properties ofNicalon
start degrading above about 600°C because of the thermodynamic instability of
composition and microstructure. Ceramic grade "Nicalon" fibers, carrying the

designation NLM, Hi, and Hi-S having low oxygen content are also available.
The structure ofNicalon fiber has been studied by many researchers. Figure 3.21
shows a high resolution transmission electron micrograph ofNicalon type SiC
produced in laboratory, indicating the amorphous nature of the SiC produced by
the Yajima process. The commercial variety of Nicalon has an amorphous
structure while another, a non-commercial variety, showed a microcrystalline
structure (SiC grain radius of 1.7 nm) (Simon and Bunsell, 1984). The
microstructural analysis shows that both the fibers contain, in addition to SiC,
Si02 and free carbon. A model of Nicalon fiber consisting of ~-SiC crystals
mixed with some free carbon with a composition of SiCPy , with x+y being
equal to 4 is shown in Fig. 3.22 (Laffon et aI., 1989). The density of the fiber is
about 2.6 glcm 3 which is low compared to that of pure ~-SiC which is not
surprising in view of the fact that the composition is a mixture of SiC, Si02 and
C.
The properties of Nicalon fiber are summarized in Table 3.8. A quick

comparison ofNicalon SiC fiber with CVD SiC fiber shows that the CVD fiber
is superior in properties mainly because it is almost 100% ~-SiC while the
Nicalon fiber is a mixture of SiC, Si02 , and free carbon. In particular, it should
be mentioned that CVD SiC fiber shows a superior creep resistance vis a vis
Nicalon fiber (DiCarlo 1985).
(d) Other SiC Type Fibers
There are many silicon carbide type fibers, with elemental compositions ofSi-C,
Si-N-C-O, Si-B-N, Si-C-O, and Si-Ti-C-O, that are commercially available.
Almost all of them are made from polymeric precursors. A multifilament silicon
carbide fiber, called Tyranno (Yamamura et aI., 1988) is made by pyrolysis of
poly (titano carbosilanes) and it contains between 1.5 - 4 wt. % titanium. A textile
grade silicon carbide fiber made, called Syiramic, has, according to the
manufacturer, a nanocrystalline structure (crystallite size 0.5 Ilm), a density of
3.0 g cm- 3, a tensile strength of 3.15 GPa and an elastic modulus of 405, GPa.
(e) Use of Raman Spectroscopy to characterize SiC type fibers
Raman spectroscopy is a powerful non-destructive tool for characterizing fibers

and to study the deformation behavior of organic and inorganic fibers (Young
and Day, 1989; Shadier and Galiotis., 1995; Gouadec and Colomban, 2001). It
is a very sensitive tool for compositional and structural analysis, with a high
Fig. 3.21 High resolution transmission electron micrograph of Nicalon type SiC
produced in a laboratory, indicating the amorphous nature of the SiC produced by the
Yajima process (courtesy ofK. Okamura).
Table 3.8. Typical properties ofNicalon SiC fiber
Ceramic HVR' grade LVR" grade

grade
Density, g cm-3 2.55 2.32 2.45-2.55
Tensile strength, MPa 2960 2930 2960
Young's Modulus, 192 186 192
GPa
Strain to failure 1.5 1.6 1.5
Coefficient of 4
thennal expansion,
10-6 K- 1
Volume resistivity, n 103 >10 6 05-5.0
cm
, low dielectric fiber (high volume resistivity)
•• higher conductivity fiber (low volume resistivity)
Fig. 3.22 A "'OdeI ofNicaJon liber - • cotnposition ofSiCkOy (aJ!e, Latfoo et al.• 1989).
degree of sensitivity to the nature of the chemical bond. Characteristic Raman

spectra can be obtained from these fibers, even when the fibers are embedded in
a sufficiently transparent matrix to depths of 5 - 30 flm. The basis of the
technique is that under an applied stress, specific Raman bands shift from their
normal positions. Under tension, the peaks of the Raman bands shift to lower
frequencies. The magnitude of frequency (or wavenumber) shift is a function of
the material, Raman band under consideration, and the Young's modulus of the
material. The shift in Raman bands results from changes in force constants
because the bonding is anharmonic in nature.
Many of the "SiC" type fibers contain free carbon. One can analyze this by the
shift (S") in the 1580 cm- 1 Raman band of carbon as a function of applied strain.
This band is characteristic of aromatic!graphitic forms of carbon. A compressive
strain results in a positive S" while a tensile strain results in a negative S". Figure
3.23 shows the shift in Raman wavenumber per percent of applied strain as a
function of inverse Young's modulus square root for a series ofNicalon fibers,
Tyranno fiber, PAN and pitch-based carbon fibers and two organic fibers,
aramid and PBZT (Gouadec and Colomban, 2001). Hi and Hi-S grades of
Nicalon do not show any shift because carbon does not form a continuous
amorphous network but is in the form of isolated nanoprecipitates.
IS o Aramid
mpBZf
"'Pilch
~PAN
10 OTyranno
• NLM-NlCalon
+ Hi-Nicalon
A Hi-S-NlCalon
,-.
S
~
;:;-
·s
~ 0
~ o l70
+
1000 x E-11l
-5
(GPa11l)
-10
•
-15
Fig. 3.23 Shift in Raman wavenumber per percent of applied strain as a function of
inverse Young's modulus square root for a series ofNicalon fibers, Tyranno fiber, PAN
and pitch-based carbon fibers and two organic fibers, aramid and PBZT (Gouadec and
Colomban, 2001).
~ 100
--- SiC
-o-Carbon
J
.S 80
1 60
~
•e 40
0
20
Z 0
0 10 20 30 40 50 60 70
0
400 ' ,
40
~ 300 30 ~
=- =-
~ 200 20 ~
~
10 =
0 0
0 10 20 30 40 50 60 70
Distance from fiber center (Jim)
Fig. 3.24 Correlation between the Raman spectra along the radius of an SCS- 6 fiber
(Gouadecand Columban, 2001) and Young's modulus and hardness (Mann eta!., 1999).
Yet another example of the versatility of Raman spectroscopy is given in Fig.

3.24. In this figure we have combined the results from Gouadec and Colomban,
2001 and Mann et al. (1999) on SCS-6 fiber. The correlation between the Raman
spectra taken along the radius of an SCS-6 fiber and hardness of modulus results
aswe proceed from the isotropic carbon core, pyrolytic carbon coating through
SiC and outer carbon rich layer is excellent.
3.6 WHISKERS
Whiskers are monocrystalline, short fibers with extremely high strength. This
high strength, approaching the theoretical strength, comes about because of the
absence of crystalline imperfections such as dislocations. Being mono crystalline,
there are no grain boundaries either. Typically, whiskers have a diameter of a
few /lm and a length ofa few mm. Thus, their aspect ratio (length/diameter) can
vary between 50 and 10000. Whiskers, however, do not have uniform
dimensions or properties. This is perhaps their greatest disadvantage, i.e., the
spread in properties is extremely large. Handling and alignment of whiskers in
a matrix to produce a composite are other problems.
Whiskers are normally obtained by vapor phase growth. Early in the nineteen
seventies, a new process was developed, starting from rice hulls, to produce SiC
particles and whiskers (Milewski et aI., 1974; Lee and Cutler, 1975). The SiC
particles produced by this process are very fine in size. Rice hulls are a waste by-
product of rice milling. For each 100 kg of rice milled, about 20 kg of rice hull
is produced. Rice hulls contain cellulose, silica, and other organic and inorganic
materials. Silica from soil is dissolved and transported in the plant as monosilicic
acid. This is deposited in the cellulosic structure by liquid evaporation. It turns
out that most of silica ends up in hull. It is the intimate mixture of silica within
the cellulose that gives the near ideal amounts of silica and carbon for silica
carbide production. Raw rice hulls are heated in the absence of oxygen at about
700°C to drive out the volatile compounds. This is called coking. Coked rice
hulls, containing about equal amounts of Si02 and free C, are heated in inert or
reducing atmosphere (flowing N2 or NH3 gas) at a temperature between 1500 and
1600 °C for about 1 h to form silicon carbide as per the following reaction
3C + Si02 -+ SiC + 2CO
Figure 3.25 shows a schematic ofthe process. When the above reaction is over,
the residue is heated to 800°C to remove any free C. Generally, both particles
and whiskers are produced together with some excess free carbon. A wet process
is used to separate the particles and the whiskers. Typically, the average aspect
ratio of the as-produced whiskers is 75.
The vapor-liquid-solid (VLS) process of growing whiskers has been known for
sometime (Lindemanis, 1983; Milewski et aI., 1985; Petrovic et aI., 1985). The
acronym VLS stands for vapor feed gases, liquid catalyst, and solid crystalline
whiskers. The catalyst forms a liquid solution interface with the growing
crystalline phase while elements are fed from the vapor phase through the liquid-
vapor interface. Whisker growth takes place by precipitation from the
supersaturated liquid at the solid liquid interface. The catalyst must take in
solution the atomic species of the whisker to be grown. For SiC whiskers,
transition metals and iron alloys meet this requirement. Silicon and carbon are
supplied in the form ofSiO and CH4 gases, respectively. The SiO gas is obtained
by carbothermal reduction ofSi02. Figure 3.26 shows the process chemistry of
the VLS whisker process. Generally a range of whisker morphologies is
obtained. The tensile strength values have been reported to range from 1.7 to
23.7 GPa in 40 tests (Milewski et aI., 1985; Petrovic et aI., 1985). Whiskers
lengths were about 10 mm and the equivalent circular diameter averaged 5.9 /lm.
Their average tensile strength and modulus were 8.4 GPa and 581 GPa,
respectively. The VLS process is extremely slow.
Rice hull
Carbon tube
reactor
Shred and disperse
Whisker/carbon separation
Whiskerlhull relict separation
Carbon oxidation
I SiC whiskers I
Fig. 3.25 Process of making silicon carbide whiskers from rice hulls.
3.6.1 Silicon Carbide in a Particulate Form
SiC in particulate form has been available quite cheaply and abundantly for
abrasive, refractories, and chemical uses. In this conventional process silica in
CO(g)
SiO(g) + ~ = Si + CO(g)
Generator
SiD(g) + 2C = SiC(s) + CO(g)
Fig. 3.26 Chemistry of the Vapor-Liquid-Solid (VLS) process of making SiC whiskers.
the form of sand and carbon in the form of coke are made to react at 2400 °C in
an electric furnace. The SiC produced in the form of large granules is
subsequently comminuted to the desired size.
3.7 EFFECT OF TEMPERATURE ON STRENGTH OF CERAMIC FIBERS
One needs to examine characteristics such as high temperature strength and

stiffness at high temperature, creep behavior and oxidation resistance of fibers
meant to be used at high temperatures under load and in air. We provide a brief
review of these characteristics of various ceramic fibers.
Figure 3.27 shows schematically strength as a function of temperature for some

ceramic fibers. Carbon fiber has an excellent high temperature in inert
atmosphere. In air, it starts oxidizing as temperatures as low as 450°C. The
simple chemical reaction for the oxidation can be represented by
We discuss the problem of low oxidation resistance of carbon in some detail in

chapter 11 where we discuss the carbon/carbon composites. Silicon-based

nonoxide fibers, represented by SiC and SixNy , show a reasonable oxidation
resistance for controlled composition fibers. Silicon carbide starts oxidizing
above 1300 °C.
Pysher et al. (1989) measured high temperature strengths of various oxide and
nonoxide ceramic fibers. High temperature strength of the SiC type fibers was
limited by oxidation and internal void formation while in the case of oxide fibers
intergranular glassy phase led to softening. Thermal stability of ceramic fibers
derived from polymeric precursors is of special interest, mainly because they
frequently have some undesirable phases present in them. Polycarbosilane-
derived SiC type fibers, such as Nicalon or Tyranno, involve a thermal oxidation
curing process as described above and can contain as much as 10 mass% oxygen.
Such fibers decompose at temperatures above 1200 °C in a nitrogen or argon
atmosphere with SiO and CO gas evolution:
SiCxOy (s) --). SiC (s) + SiO (g) + CO (g)
Polycrystalline Oxide
Carbon Single Crystal Oxide
Single Crystal
Oxide
b '~-+-_Polycrys!alline
Oxide
T-
Strength drops at Intermediate
Poor OxidatIOn Resistance Temperatures for Polycrystanine
C+o.- Co. Oxides
T-
Resonable Oxidation
Resistance
Fig. 3.27 Schematic of strength vs. temperature for some ceramic fibers.
This reaction is accompanied by a change in the morphology and crystal

structure of the fiber and a loss of tensile strength (Lipowitz et aI., 1990). With
a view to control the oxygen content of such fibers, Okamura et al. (1992)
adopted a curing process using radiation-chemical reactions and produced
oxygen-controlled SiC fibers. For example, by radiation curing in an oxygen
atmosphere, they could make SiC fibers with a gradient in oxygen content. By
electron-radiation curing polycarbosilane in vacuum or helium, SiC fibers with
less oxygen were prepared. These low oxygen fibers showed high strength and
Young's modulus to temperatures as high as 1500 °C and, at a given temprature,
the strength and modulus increased with decreasing oxygen content. Figure 3.28
summarizes their results on tensile strength and modulus as a function of
temperature and oxygen content.
Mah et al. (1984) studied the degradation ofNicalon fiber (SiCxO y ) after heat
treatment in different environments. The strength of this fiber at temperatures
greater than 1200 °C degraded because of CO evaporation from the fiber as well
as P-SiC grain growth. Another unusual feature observed with ceramic fibers
made via pyrolysis of polymeric precursors, especially with compositions Si-C-O
and Si-N-C-O, is that they have lower densities than the theoretical values. The
theoretical density, Pt, value can be calculated by using the relationship
Pt=PYi
where p is the density, V is the volume fraction, the subscript i indicates the ith
phase, and summation over all the phases present is implied. Lipowitz et al.
(1990) used x-ray scattering techniques to show that porosity present is such
fibers was due to globular pores of nm size and that the pore fractions ranged
from 5% to 25%. According to these authors, nanochannels form during the
early states of pyrolysis when rather large volumes of gases are given out. In the
later stages of pyrolysis, of gases are given out. In the later stages of pyrolysis,
during densification these nanochannels suffer a viscous collapse and nanopores
are formed. The volume fraction of nanopores decreases with increasing
pyrolysis temperature. The reader should note that a higher density and a lower
void fraction will lead to a higher elastic modulus of these ceramic fibers.
Oxides are thermodynamically stable in air, giving them excellent oxidation

resistance. The strength, however, drops off at intermediate temperatures.
Alumina starts creeping at 1300 °C or so, unless it is in the form of a single
crystal. We have already described above (see Sec. 3.6) the single crystal
alumina fiber, Saphikon, made by fiber pulling from the melt. The single crystal
structure of the Saphikon provides much better creep resistance than
polycrystalline alumina fiber. This is especially true for the c- orientation
I I
3.0r- Oxygen
content
(mass%)
• <0.5
0=4.0
.. " 1.9
t:.. = 10.B
100
O~ __~__~I__~__~IJ 1 I 1
1400 1600 1800 1400 1600 1800
Temperature, K Temperature, K
Fig. 3.28 Tensile strength and Young's modulus as a function of temperature for
polycarbosilane derived SiC fiber with different oxygen amounts (after Okamura et aI.,
1992).
fiber with the basal plane is perpendicular to the fiber axis. It is easy to see that
in this orientation the basal slip during creep will not occur and the fiber will
show good creep resistance. It turns out, however, that ifthe fiber orientation is
off by more than 3 degrees, easy basal slip comes into play and some of the
creep resistance will be lost. Single crystal YAG (yttrium aluminum garnet) is
another oxide fiber that shows excellent creep resistance. YAG has a complex
cubic structure and is less anisotropic than the hexagonal sapphire. Single crystal
YAG fiber can be grown by either laser floating process or edge-defined film-fed
growth (EFG) method. The topic of creep of CMCs is further discussed in
chapter 7.
3.8 COMPARISON OF FIBERS
Two most important characteristics of any reinforcement material are its strength
and Young's modulus. First of all, we note that all these high performance fibers
have very low density values. Given the general low density of these fibers, the
best of these fibers group together in the top right hand corner of Fig. 3.29. The
reader will also recognize that the elements comprising these fibers pertain to the
0
a..
(!) 4
-
I
..c:
0'1 3
c
-
(J)
"-
en
2
(J) Glass I:~::::::::: ::::::::::~:~I
(/) Mullite
c
~ g~-Pitch.
IsotropIc
oL-~~--~~--~----~--------~
o 100 200 300 400 500
Young's Modu Ius, GPo
Fig. 3.29 Strength vs. Young's modulus for some ceramic fibers.
first two rows ofthe periodic table. Also to be noted is the fact that, irrespective
of whether in compound or elemental form, they are mostly covalently bonded
which is the strongest bond. Generally, such light strong, and stiff materials are
very desirable in most applications, but particularly so in aerospace field, land
transportation, energy related industry, housing and civil construction, etc. Fiber
flexibility is associated with the Young's modulus and the diameter (see Sec.
3.1). In the general area of high modulus (i.e., high E) fibers, the diameter
becomes the dominant parameter controlling the flexibility. For a given E, the
smaller the diameter the more flexible it is. Fiber flexibility is a very desirable
characteristic if one wants to bend, wind, and weave a fiber in order to make a
complex shaped final product.
Some of these fibers have quite anisotropic characteristics. The strength,

modulus, and thermal characteristics can vary with direction, especially along
the fiber axis and transverse to it the axis. In particular, thermal expansion
coefficient of carbon is quite different in the radial and longitudinal directions.
This would also be true of any single crystal fiber or whisker, e.g., alumina
single crystal fiber which has a hexagonal structure. In this respect,
polycrystalline fibers such as SiC and Al 20 3 fibers are reasonably isotropic.
Ceramic matrix composites are meant to be used at very high temperatures,

indeed. An important problem that arises at these very high temperatures (> 1500
0c) is that of fiber and matrix oxidation. Carbon fiber, for example, does not
have a great oxidation resistance at temperatures above 400°C. Silicon carbide
type ceramic fibers are the only suitable candidates for reinforcement at very
high temperatures (> 1200-1300 0c) and in air. It would appear oxide fibers
would be the likely candidates, because of their inherent stability in air, for
applications at temperatures higher than 1300 °C.
Another important characteristic ofthese high performance fibers in their rather

low values of strain to fracture, generally < 2 - 3 %. This means that in a CMC,
the reinforcement and the matrix may not be much different in terms of strain to
fracture. Also, the ratio of the modulus reinforcement and the matrix may be 2- 3
or as low as 1. This is a very different situation from that encountered in PMCs
and MMCs. It points out that in CMCs, the interface has a much more crucial
role to play.
3.9 STATISTICAL ANALYSIS OF FIBER STRENGTH
Fracture of brittle materials, in general, involves statistical considerations. Most

reinforcements for ceramic materials are ceramic materials which fail in a brittle
manner. Ceramic materials, fibrous or otherwise, will have randomly distributed
defects on their surfaces or in the interior. The presence of these defects results
in a rather large scatter in the experimentally determined strength values of
ceramic materials. In this statistical analysis of fiber strength, we follow Rosen's
(1965) treatment. Wagner (1989) gives a more detailed treatment.
We can regard a fiber of a given length to be made up of a series of chain links.

When such a fiber is loaded, the link or the segment containing the longest
defect will fail first and cause the fiber fracture. The longer the fiber, the higher
is the probability of a link having a critical flaw size required for failure. In other
words, one would expect that the mean strength of a short fiber length will be
greater than the mean strength of a long fiber length. The fiber failure occurs
when the weakest link fails. This is called the weakest-link assumption. It turns
out that such a "weak-chain" material is well described by a statistical
distribution known as the Weibull distribution, named after the person who first
proposed it. The basic assumption is that a fiber has a distribution of flaws (on
the surface and/or in the interior). The Weibull distribution assumes that all
segments or chain links have the same type of flaw but of different lengths. The
Weibull distribution gives the failure probability, F(O"), ofa fiber at an applied
stress 0" by the following expression:
F( 0") = 1 - exp [ - acr~ ]
where a and pare statistical parameters. F( 0") is called the cumulative frequency
distribution function. The parameter Pis referred to as the Weibull modulus. It
is a measure of the variability in the strength of the fiber, the higher the value of
Weibull modulus, the higher the uniformity of strength values. The probability
density function for a brittle fiber oflength L that obeys Weibull distribution can
be obtained by differentiating the above expression and takes the following form
.1(0") = Lap O"~ -) exp ( - L a O"~) (3.3)
We define the kth moment, Mk> of a statistical distribution, as
(3.4)
For a fiber showing the strength distribution given by Eq. (3.3), we can obtain
an expression for the mean strength as follows:
0" = M) = f 0".1(0") dO" (3.5)

o
= f exp( - L a O"~) . 0" • Lap O"~ -) dO"

o
= f Lap O"~ -) . 0" . exp ( - L a O"~) dO"

o
= Jexp( - L a (J~) .
~ (L a (J~)13
I · d(L a (J~) (3.6)
o (L 0,)13
Let La(J~ :; x. Then Eq. (3.6) becomes
~-I
cr = j exp( - x)· (x) ~ I dx (3.7)
o (La)13
Now, letting ~+ 11~ = n, we can convert Eq. (3.7) to the following simple
form:
Jexp( -x)· x
1 ~
cr = (L a) -13 n - 1 dx
(3.8)
o
The integral portion of Eq. (3.7) is called the gamma function of order nand
denoted by [(n). Thus, the mean strength is given by
1
cr = (L a) -13 . [(n) (3.9)
Some specific values of[(n) are given below
[(0) = 00
[(112) =.[7t
[(1) = 1
[(2) = 1
[(3) = 2!
Substituting the value of n, we have the mean strength given by
-- 1
a = Ml 1 r [ 1 + f3 ]
= (L a) P. (3.10)
The standard deviation s of this distribution is given by
1
S = [M2 -Mn2 (3.11)
S2 = 1-12 = M2 - (J2
The expression for standard deviation, s, for the Weibull distribution under
consideration can be evaluated as follows.
s = {P; = VM2-M; = [M2 - M; ]1/2
(3.12)
Let us evaluate the first integral in this expression. The second integral is given
by Eq. (3.1). Substituting for f( 0'), we get for the mean strength
ILa~O'(P-l). 0'2 • exp( -L aO'P) dO'

o
(L a (13)2/13
= Jexp( - LaoP),
00
. d(Lao P)
o (La)21 p
Denoting LaoJl by x, we have the integral given by

~-l
/3
Joexp( -x), (x)(Lafp
00
dx
n
The expression for s, Eq. (3.12), then becomes
s = [a'L npa'-'exp( -L n 0') do - ( (Ln)-~ r( i) I +
The coefficient of variation, J,l, for this distribution is given by

s
fl
Note that fl is a function only of the parameter P; in fact, it is an inverse function

of the coefficient of variation fl.
For fibers characterized by Weibull distribution and having P> 1, we have the
following general relationships:
For glass fibers, fl '" 0.1, P= 11
For alumina, carbon, boron, silicon carbide type fibers, fl = 0.2-0.4, P=2.7- 5.8.
Consider the mean fiber strength (eq. 3.10) again
cr = (aL r I1~ r ( 1 + *)
For a unit length of fiber (L = 1)
where
For p> 1, we have 0.88 ~ k ~ 1.0. This means that we can regard the quantity
a- 1/6 as the reference level strength.
We can find the statistical mode a:, the most probable strength value, as
follows:
j(cr) = Lapcr(~-l) exp(-Lacr~)
At cr = cr" (mode value), df(cr)/dcr = O. Thus,

dj{o) = La~ (~-1)0(13- 2 ) exp(-Lao13)

do
For large~,
~ -1
- - ==>
~
and
-.!.
0* = (aL) P
Thus, the statistical parameters a and p can be obtained from experimental

values of 0 and Jl.
3.9.1 Experimental Determination of Wei bull Parameters for a Fiber
A statistical analysis of the fiber tensile strength values determined on a series

of fiber samples can be easily made by a two-parameter Weibull distribution.
The Weibull analysis gives us the cumulative probability offailureF(a) of the
fiber at a stress 0', as
F(a) = 1 - exp( - aa~) (3.13)
where Pis the Weibull modulus and a is a scale parameter. Pis a measure of the
scatter in the strength data. Rearranging Eq. 3.12, we get
In[ln( I )] = plna + Ina (3.14)

1 -F(a)
From Eq. 3.14, we can easily obtain a and Pgraphically. One generally arranges
the tensile strength values of the single filaments in an ascending order and
assigns a probability of failure using an estimator given by
FrO') = _1_·_ (3.15)

\; (1+N)
where F(a j ) is the probability of failure corresponding to the ilhe strength values
and N is the total number of fibers tested. Substituting Eq. 3.15 in Eq. 3.14, we
have
In[ln( N + 1 .)] = plna; + Ina

N+ 1-1 (3.16)
Eq. 3.16 says then that a plot of In [(N + 1)/(N + 1 -i)] vs. a j on a log-log graph
will be a straight line ifthe tensile strength data follows a Weibull distribution.
The intercept on the y-axis will then be a and the slope will be p. Figure 3.30
shows such a double log plot for some Nextel fibers of (gage length = 25 mm)
(Wilson and Visser, 2001). An example of cumulative probability of failure of
Nexte1312 fiber as function of fiber strength is shown in Fig. 3.31 (Kerr et ai.,
2002). The Weibull mean tensile strength, cr, standard deviation, s, and
coefficient of variation (CV) can be obtained as follows:
(3.17)
(3.18)
where
and the coefficient of variation, CV, is given by
CV = 100 ( ~l
2,---------------------------~----------------~
Nextel610: Me . . = 3.40 GPa,m = 10.1
Nextel6SO: Mean = 2.60 GPa,rn= 6.8
Nextel720: Mean = 1.98 CPa, m = 7.6
-1
~
.-<
-2
.5
.5
-3
~
Nutel650
"
Nu:td 610
-4
" :>C
-5
" >:
-6
0.2 0.4 0.6 0.8 1.2 1.4
In Tensile Strength, GPa
Fig. 3.30 Weibull distribution of some Nextel fibers (after Wilson and Visser, 2001).
-e
...... •
0.8
••
u..
~
.a ••
•
••
'iii
....0
II. 0.6
•
~ •••
:aC\'I
..0
0.4 •
~
a.
0.2
0
500 1000 1500 2000 2500 3000
T ensile Strength (MPa)
Fig. 3.31 Cumulative probability of failure of Nextel 312 fiber (after Kerr et aI., 2002).
REFERENCES
Bennett, S.C. and D. J. Johnson (1978) in 5th Int. Carbon and Graphite Con!, Soc.
Chern. Ind., London, p. 377.Bennett, S.C. and D. J. Johnson (1979) Carbon, 17,25.
Bennett, S.C., D.J. Johnson, and W. Johnson (1983) J Mater. Sci., 18,3337.
Birchall, J.D., J.A.A. Bradbury, and J. Dinwoodie (1985) in Strong Fibres, Handbook
o/Composites, vol. 1, North-Holland, Amsterdam, p. 115.
Chawla, KK (1998) Fibrous Materials, Cambridge University Press, Cambridge.
DeBolt, H.E., V. 1. Krukonis, and F. E. Wawner (1974) in Silicon Carbide 1973, Univ.
of S. Carolina Press, Columbia, SC, p. 168.
Deurbergue, A. and A. Oberlin, (1991) Carbon, 29,691.
Dhingra, A.K (1980) Phil. Trans. Roy. Soc., London, A294, 41l.
DiCarlo, lA. (June 1985) J 0/ Metals, 37, 44.
Diefendorf, R.J. and E. Tokarsky (1975) Polymer Eng. & Sci., 15, 150.
Dresher, W.H. (April 1969) Journal 0/ Metals, 21, 17.
Ezekiel, H.N. and R. G. Spain (1967) J Polymer Sci. C., 19, 27l.
Fourdeux, A., R. Perret, and W. Ruland (1971) in Carbon Fibres: Their Composites and
Applications, The Plastics Inst., London, p .57.
Gasson, D.G. and B. Cockayne (1970) J Mater. Sci., 5,100.
Gouadec, G. and P. Colomban (2001) 1 Euro. Ceram. Soc., 21, 1249.
Haggerty, J.S. (May 1972) NASA-CR-120948.

Hurley, G.F. and J.T.A. Pollack (1972) Met. Trans., 7, 397.
Inal, O.T., N. Leca, and L. Keller (1980) Phys. Stat. Sol., 62, 681.
Johnson, D.J. and C. N. Tyson (1969) Brit. J. App. Phys., 2, 787.
Kelly, B.T. (1981) Physics o/Graphite, App. Sci. Pub., London.
Kumagai, M. and G.L. Messing (1985) J. Am. Ceram. Soc., 68, 500.
Kerr, M., J.J. Williams, N. Chawla, and K.K. Chawla (2002) in Proc. of MRS
symposium, vol 702 ,Mater. Res. Soc., Warrendale, PA, p. 223.
LaBelle, H.E. and A.I. Mlavsky (1967) Nature, 216, 574.
LaBelle, RE. (1971) Mater. Res. Bull., 6, 581.
Laffon C., A.M. Flank, P. Lagarde et al. (1989) J. Mater. Science, 24, 1503.
Lee, J.-G. and I.B. Cutler (1975) Am. Ceram. Soc. Bull., 54, 195.
Lara-Curzio, E. and S. Stemstein (1993) Composites Sci. & Tech., 46, 265.
Lindemanis, A. (1983) in Emergent Process Methods/or High Technology Ceramics,
Plenum Press, New York.
Lipowitz, J., J.A. Rabe, and L.K. Frevel (1990) J. Mater. Sci., 25, 2118.
Mah, T., N.L. Hecht, D.E., McCullum, J.R., Hoenigman, H.M. Kim, A.P. Katz, andH.A.
Lipsitt (1984) J. Mater. Sci., 19, 1191.
Mann, A.B., M. Balooch, J.H. Kinney, and T.P. Weihs (1999)J. Amer. Ceram. Soc., 82,
111.
Milewski, J.V., F.D. Gac, J.J. Petrovic, and S.R. Skaggs (1985) J. Mater. Science, 20,
1160.
Milewski, J.V. J. L. Sandstrom, and W. S. Brown (1974) in Silicon Carbide-1973,
University of S. Carolina Press, Columbia, SC, p. 634.
Nourbakhsh, S., F.L. Liang, and H. Margolin (1989) J. Materials Sci. Letters, 8, 1252.
Okamura, K. and T. Seguchi (1992) J. Inorganic and Organometallic Polymers, 2, 171.
Perret, R. and W. Ruland (1970) J. App. Cryst., 3, 525.
Petrovic, J.J., J.V. Milewski. D.L. Rohr, and F.D. Gac (1985) J. Mater. Sci., 20, 1167.
Pollack, J.T.A. (1972)J. Mater. Sci., 7, 787.
Pysher, D.J., K.C. Goretta, R.S. Hodder, Jr., and R.H. Tressler (1989) J. Amer. Ceram.
Soc., 72, 284.
Reynolds, W.N. and J.V. Sharp (1974) Carbon, 12, 103.
Riggs, J.P. (1985) in Encyclopedia o/Polymer Science & Engineering, 2nd ed., vol. 2,
John Wiley & Sons, New York, p. 640.
Romine, J.e. (1987) Cer. Eng. Sci. Proc., 8, 755.
Rosen, B.W. (1965) in Fiber Composite Materials, ASM, Metals Park, OH, 58.
Sayir, A. and S.C. Farmer (1995) in Ceramic Matrix Composites, MRS proceedings, vol.
365, Maer. Res. Soc., Pittsburgh, p. 11.
Sayir, A., S.C. Farmer, P.O. Dickerson, and H.M. Yun (1995) in Ceramic Matrix
Composites, MRS proceedings, vol. 365, Maer. Res. Soc., Pittsburgh, p. 21.
Schadler, L. and C. Galiotis (1995) Inti. Mater. Rev., 40, 116.
Simon, G. and A. R. Bunsell (1984) J. Mater. Sci., 19,3649.
Singer, L. (1979) in Ultra-High Modulus Polymers, Applied Sci. Pub., Essex, England,
p.251.
Singer, L. (1981) Fuel, 60, 839-841.

Sowman, H.G. (1988) in Sol-Gel Technology, Noyes Pub., Park Ridge, NJ, p. 162.
Suwa, Y., R. Roy, and S. Komarneni (1985) J. Am. Ceram. Soc., 68, C-238.
Towata, A., H.J. Hwang, M. Yasuoka, M. Sando, and K. Niihara (2001) Composites A,
32A,1127.
Wagner, H.D. (1989) in Application of Fracture Mechanics to Composite Materials,
Elsevier, Amsterdam, p. 39.
Watt, W. (1970) Proc. Roy. Soc., A319, 5.
Watt, W. and W. Johnson (1969) App. Polymer Symposium, 9, 215.
Wax, 8.G. (1985) Amer. Cer. Soc. Bull., 64 (185) 1096.
Weber, J. K. R., J. J. Felten, B. Cho, and P. C. Nordine (1998) Nature, 393, 769.
Wilson, D.M. and L.R. Visser (2001) Composites A, 32A, 1143.
Yajima, S., K. Okamura, J. Hayashi, and M. Omori (1976) J. Amer. Ceram. Soc., 59,
324.
Yajima, 8. (1980) Phil. Trans., R. Soc., London, A294, 419.
Yamamura, T., T. Ishirkawa, M. Shibuya, T. Hiasyuki, and K. Okamura (1988) J. Mater.
Sci., 23, 2589.
Young, R.J. and R.J. Day (1989) Brit. Polymer J., 21, 17.
SUGGESTED READING
Bunsell, A.R. (ed.) (1988) Fibre Reinforcements for Composite Materials, Elsevier
Amsterdam.
Chawla, K.K. (1998) Fibrous Materials, Cambridge University Press, Cambridge.
Elices, M and J. LLorca (eds.) (2002) Fracture in Fibers, Elsevier, Oxford.
Peebles, L.H. (1995) Carbon Fibers, CRC Press, Boca Raton, FL.
Cliapter4
PROCESSING OF CERAMIC MATRIX

COMPOSITES
In this chapter we describe some of the important processing techniques for

fabricating ceramic matrix composites. Among the items that one should take
into account for choosing reinforcement and matrix materials are:
• melting point
• volatility
• density
• elastic modulus
• coefficient of thermal expansion
• creep characteristics
• strength
• fracture toughness
• compatibility between fiber and matrix
•• chemical compatibility
•• thermal compatibility (should be able to withstand high
temperature excursions)
•• compatibility with the environment, internal as well as external.
The external compatibility mainly involves oxidation and evaporation
characteristics.
Most of these characteristics will have an important bearing on the processing

route chosen as well as the final properties and performance of the resultant
composite.
Ceramic matrix composites( CMCs) can be made either by conventional powder

processing techniques used for making polycrystaIIine ceramics or by some new
and rather unconventional techniques. Some of the techniques, novel or not, are
no more than variants on the processing of monolithic ceramics. What is
important to realize is that the processing of CMC should be considered an
integral part of the whole process of designing a CMC component. For
example, any damage to the reinforcement because of processing will result in
a less than desirable performance in the final product. Fiber or whisker
orientation in the ceramic matrix is an integral part of the fabrication process
108 PROCESSING OF CERAMIC MATRIX COMPOSITES
and the reinforcement orientation has a most important bearing on the

mechanical response of a CMC to an applied load.
4.1 COLD PRESSING AND SINTERING
Cold pressing of a matrix powder and fiber mixture followed by sintering is a

natural extension from conventional processing of ceramics. Shrinkage is
common problem associated with sintering of most ceramics. This problem is
exacerbated when a glass or ceramic matrix is combined with a reinforcement
material. Thus, after sintering, the matrix generally shrinks considerably and the
resulting composite exhibits a significant amount of cracking. One of the
reasons for high shrinkage after sintering is that fibers and whiskers, i.e.,
reinforcements with high aspect ratio (length/diameter), can form a network that
may inhibit the sintering process. Depending on the difference in thermal
expansion coefficients of the reinforcement and matrix, a hydrostatic tensile
stress also develops in the matrix on cooling, which will counter the driving
force (surface energy minimization) for sintering (Kellet and Lange, 1989; Raj
and Bordia, 1989). Thus the densification rate of the matrix will, in general, be
retarded in the presence of reinforcement (Prewo, 1986; De Jonghe et aI., 1986;
Rahaman and De Jonghe, 1987; Sacks et aI., 1987; Bordia and Raj, 1988).
4.2 HOT PRESSING
Hot pressing is frequently used in a combination of steps or in a single step in

the consolidation stage of CMCs. Hot pressing is an attractive technique
because the simultaneous application of pressure and high temperature can
significantly accelerate the rate of densification, resulting in a pore-free and
fine grained compact. An example of a common hot-pressed composite is SiC
whisker reinforced AIP3' used in cutting tool applications.
A common variant of conventional hot pressing is the slurry infiltration

process. It is perhaps the most important technique used to produce continuous
fiber reinforced glass and glass-ceramic composites (Sambell et aI., 1974;
Prewo and Brennan, 1980; Brennan and Prewo, 1982; Phillips, 1983; Cornie et
aI., 1986). The slurry infiltration process involves two main stages: (i)
incorporation of the reinforcing phase into a "slurry" of the unconsolidated
matrix and (ii) matrix consolidation by hot pressing.
Figure 4.1 (a) shows a schematic of the slurry infiltration process. The first stage
involves some degree of fiber alignment, in addition to incorporation of the
reinforcing phase in the matrix slurry. The slurry typically consists of the
matrix powder, a carrier liquid (water or alcohol), and an organic binder. The
organic binder is burned out prior to consolidation. Wetting agents may be

added to ease the infiltration of the fiber tow or preform. The fiber tow or fiber
preform is impregnated with the matrix slurry by passing it through a slurry
tank. The impregnated fiber tow or preform sheets are similar to the prepregs
used in fabrication of polymer matrix composites (PMCs). The impregnated
tow or prepreg is wound on a drum and dried. This is followed by cutting and
stacking of the prepregs, and consolidation by hot pressing. The process has the
advantage that, just as in PMCs, the prepregs can be arranged in a variety of
stacking of sequences, e.g., unidirectional, cross-plied (0 /90 /0 /90 etc.), or
0 0 0 0
,
angle-plied (+9/-9/+9/-9, etc.) Figure 4.1(b) shows an optical micrograph of

a transverse section of a unidirectional alumina fiber/glass matrix composite
(some residual porosity can be seen in this micrograph) while Fig. 4.1(c) shows
the pressure and temperature schedule used during hot pressing of this
composite.
As mentioned above, the slurry infiltration process is well suited for glass or
glass-ceramic matrix composites, mainly because the processing temperatures
for these materials are lower than those used for crystalline matrix materials.
The hot pressing process does have the limitation of not being able to produce
complex shapes. Application of a very high pressure during hot pressing can
also easily damage the fibers and decrease the strength of the composite. The
fibers may also be damaged by mechanical contact with refractory particles of
a crystalline ceramic by or from reaction with the matrix at very high
processing temperatures. The matrix should have as little porosity as possible
in the final product as porosity in a structural ceramic material is highly
undesirable. To this end, it is important to completely remove the fugitive
binder and use a matrix powder particle smaller than the fiber diameter. The hot
pressing operational parameters are also important. Precise control within a
narrow working temperature range, minimization of the processing time, and
utilization of a pressure low enough to avoid fiber damage are important factors
in this final consolidation part of the process. Fiber damage and any
fiber/matrix interfacial reaction, along with its detrimental effect on the bond
strength, are difficult to avoid in hot pressing. As with any processing
technique, the fibers should suffer little or no damage during handling.
In summary, the slurry infiltration process generally results in a composite with

fairly uniform fiber distribution, low porosity, and relatively high strength. The
main disadvantage of this process is that one is restricted to relatively low
melting or low softening point matrix materials.
Whisker reinforced CMCs are generally made by mixing the whiskers with a
ceramic powder slurry, dried, and hot-pressed. Sometimes hot isostatic pressing
rather than uniaxial hot pressing is used. Whisker agglomeration in a green
body is a major problem. Mechanical stirring and adjustment of pH level of the
Stack of matrix
impregnated
fiber tapes
Fibers Pressure
Binder bumouot
Composite sod'c
Hot pressing
800-92S·C Graphite die
(a)
(b)
Fig. 4.1. Slurry infiltration process followed by hot pressing: (a) Schematic of process,
(b) Optical micrograph of a transverse section of a unidirectional alumina fiber/glass
matrix composite (some porosity can be seen in this micrograph), and (c) pressure and
temperature schedule used during hot pressing of this composite (see on p. 110).
- - - Temperature
6
100 ------ Pressure
5
800
600
r-----------,
400
,I
I 2
200
,,
I
II
OL---~~--+---~~--~~O
o 50 100 15 200
Time, min
(c)
suspension (matrix powder/whiskers in water) can be of help in this regard.

Addition of whiskers to a slurry results in very high viscosity. Also, whiskers
with large aspect ratios (>50) tend to form bundles and clumps (Liu et aI.,
1991). Obtaining well-separated and deagglomerated whiskers is of great
importance for reasonably high density composites. Use of organic dispersants
(Barclay et aI., 1987) and techniques such as agitation mixing assisted by an
ultrasonic probe, and deflocculation by proper pH control (Yang and Stevens,
1990) can be usefully employed.
Most whisker reinforced composites are made at temperatures in the 1500-1900

°C range and pressures in the 20-40 MPa range (Homeny et aI., 1987; Shalek
et ai. 1986). Figure 4.2 shows a scanning electron micrograph of a hybrid
composite, consisting of SiC fibers (Nicalon) and whiskers in a glass-ceramic
matrix (Chawla et aI., 2001).
4.3 REACTION BONDING PROCESSES
Reaction bonding processes similar to the ones used for monolithic ceramics
can also be used to make ceramic matrix composites. These have been used
mostly with silicon carbide or silicon nitride matrices. Among the advantages
of this method are:
• Little or no matrix shrinkage occurs during densification.

Glas ceramic matrix

with whiskers
Nicalon fiber
Fig. 4.2 SEM micrograph of a hybrid composite, consisting ofNicalon fibers and SiC
whiskers in a glass-ceramic matrix (Chawla et aI., 2001).
• Large volume fractions of whiskers or fiber can be used.

• Multidirectional, continuous fiber preforms can also be used.
• The reaction bonding temperatures for most systems are generally
lower than the sintering temperatures, so that fiber degradation can be
avoided.
One great disadvantage of this process is that high porosity is hard to avoid.
A hybrid process involving a combination of hot pressing with the reaction

bonding technique can also be used (Bhatt, 1986; Bhatt, 1990). Figure 4.3(a)
shows the flow diagram for this process (after Bhatt, 1990), while Fig. 4.3 (b)
shows a micrograph of a composite (SCS-6 fiber/ Si 3N 4) made by this process
(Chawla, 1997). Silicon cloth is prepared by attrition milling a mixture of
silicon powder, a polymer binder, and an organic solvent to obtain a "dough"
of proper consistency. This dough is then rolled to make a silicon cloth of
desired thickness. Fiber mats are made by filament winding of silicon carbide
with a fugitive binder. The fiber mats and silicon cloth are stacked in an
alternate sequence, subjected to binder burnout, and hot pressed in a
molybdenum die in a nitrogen or vacuum environment. The temperature and
pressure are adjusted to produce a handleable preform. At this stage, the silicon
matrix is converted to silicon nitride by transferring the composite to a nitriding
Fiber mat Silicon tape

conta.ining containing
fugitive SiC powder
polymer and fugitive
binder polymer binder
SiC/Si
Preform
Heat to remove binder
Hot-press to consolidate
Nitride to convert silicon

to silicon nitride
Composite
(a)
Fig. 4.3. (a) Flow diagram of the reaction bonding process for processing SCS-6 fiber/
Si3N4 composites (after Bhatt, 1990) and (b) Microstructure of SCS-6 fiber/ Si3N4
composite showing uniform fiber distribution and small amounts of residual porosity
around the periphery of the large diameter fibers (courtesy ofR.T. Bhatt).
E
:t
E
::t
o·
o
It)
(b)
Fig 4.3 - contd
furnace between 1100 and 1400 °C. Typically, the silicon nitride matrix has
about 30% porosity, which is not unexpected in reaction bonded silicon nitride.
Note also the matrix density variations around fibers in Fig. 4.3(b).
Reaction bonding processing has been applied to alumina (Claussen et aI.,

1989, 1994; Kristofferson et aI., 1993) and mullite (Wu and Claussen, 1994;
matrix fabrication. In the case of an oxide matrix, reaction bonding uses direct
oxidation of starting powders to create the matrix. The process has a shorter
processing time than forming matrix by infiltration techniques. For the case of
alumina, a starting powder consisting of AI-A1203 or a slurry infiltrated
compact is heat treated in an oxidizing atmosphere. The aluminum powder
oxidizes to A1203, and in the process a 28% volume expansion occurs which
partially compensates for the sintering shrinkage. Oxidation is usually
completed below 1100°C, where sintering does not take place. Mullite is
obtainable, the distance by adding SiC to the starting powder, which is then
oxidized during processing to form the Si02 component. An alternative
technique of producing mullite involves the use of an AI-Si alloy as a starting
material. Because problems with shrinkage during densification are avoided,
reaction bonding is an attractive and fast method of producing fiber reinforced,
ceramic composites. Reaction bonding processing has also been used profitably
to make alumina
4.4 INFILTRATION
Infiltration of a preform made of a reinforcement can be done with a matrix

material in solid, liquid, or gaseous form.
Liquid Infiltration: This technique is very similar to liquid polymer or liquid

metal infiltration, Fig. 4.4. Proper control of the fluidity of liquid matrix is, of
course, the key to this technique. It yields a high density matrix, i.e., no pores
in the matrix. Almost any reinforcement geometry can be used to produce a
virtually flaw-free composite. The temperatures involved, however, are much
higher than those encountered in polymer or metal processing. Processing at
such high temperatures can lead to deleterious chemical reactions between the
reinforcement and the matrix. Thermal expansion mismatch between the
reinforcement and the matrix, the rather large temperature interval between the
processing temperature and room temperature, and the low strain to failure of
ceramics can add up to a formidable set of problems in producing a crack-free
CMC. Viscosities of ceramic melts are generally very high which makes the
infiltration of preforms rather difficult. Wettability of the reinforcement by the
molten ceramic is another item to be considered. Hillig (1988) has discussed the
melt infiltration processing of ceramic matrix composites in regard to chemical
reactivity, melt viscosity, and wetting of the reinforcement by the melt. A
Piston
o Melt
o
o
o
o
Prefonn
o h~~mm-::l
o
o
oo 0Or-Heating
coils
o
o 0
o 0
o ~ __________ ~ 0
Fig. 4.4. Schematic of liquid infiltration processing. The technique is very similar to
liquid polymer or liquid metal infiltration.
preform made of reinforcement in any form (for example, fiber, whisker, or

particle) having a network of pores can be infiltrated by a ceramic melt by using
capillary pressure. Application of pressure or processing in vacuum can aid in
the infiltration process. Assuming that the preform consists of a bundle of
regularly spaced, parallel channels, one can use Poissuelles' equation to obtain
the infiltration height, h:
h I yrtcosEl
~ 2n
where r is the radius of the cylindrical channel, t is the time, y is the surface
energy of the infiltrant, e is the contact angle, and 11 is the viscosity. Note that
the penetration height is proportional to the square root of time and inversely
proportional to the viscosity of the melt. Penetration will be easier if the contact
angle, e is low (i.e., better wettability), and the surface energy (y) and the pore
radius (r) are large. However, if the radius, r, ofthe channel is made too large,
the capillarity effect will be lost.
We may summarize the advantages and disadvantages of different melt

infiltration techniques as follows. The advantages are:
(a) The matrix is formed in a single processing step.

(b) A homogeneous matrix can be obtained.
The disadvantages of infiltration techniques are:

(a) High melting points of ceramics mean a greater likelihood of reaction
between the melt and the reinforcement.
(b) Ceramics have higher melt viscosities than metals; therefore, infiltration of
preforms is relatively difficult.
(c) Matrix is likely to crack because of the differential shrinkage between the
matrix and the reinforcement on solidification. This can be minimized by
choosing components with nearly equal coefficients of thermal expansion.
4.5 DIRECTED OXIDATION OR THE LANXIDE™ PROCESS
A version of liquid infiltration is the directed oxidation process, or the

Lanxide™ process l (Urquhart, 1991). A schematic of the directed metal
oxidation process called DIMOX™ is shown in Fig. 4.5. The first step in this
process is to make a preform. In the case of a fibrous composite, filament
winding or a fabric lay-up may be used to make a preform. A barrier to stop
growth of the matrix material is placed on the preform surfaces. In this method,
a molten metal is subjected to directed oxidation, i.e., the desired reaction
product forms on the surface ofthe molten metal and grows outward. The metal
is supplied continuously at the reaction front by a wicking action through
channels in the oxidation product. For example, molten aluminum in air will get
oxidized to aluminum oxide. If one wants to form aluminum nitride, then
molten aluminum is reacted with nitrogen. The reaction can be represented as
follows:
Al + air --+ Al20 3

Al + N2 --+AIN
The end product in this process is a three dimensional, interconnected network

of a ceramic material plus about 5 - 30% of unreacted metal. When filler
particles are placed next to the molten metal surface, the ceramic network forms
around these particles. As we said above, a fabric made of a continuous fiber
can also be used. The fabric is coated with a proprietary coating to protect the
fiber from highly reducing aluminum and to provide a weak interface, which is
desirable for enhanced toughness. Some aluminum (6-7 wt. %) remains at the
end of the process. This must be removed if the composite is to be used at
temperatures above the melting point of aluminum (660°C). On the other hand,
the presence of a residual metal can be exploited to provide some fracture
toughness in these composites.
Growth barrier
Preform fabricatioJl
Composite {
( MaMx.~ ...",
~ CMCpllrf
Fig. 4.5. Schematic of the directed metal oxidation process (DIMOX™). (Courtesy of
Lanxide Corp.)
Proper control of the reaction kinetics is of great importance in this process.

The process is potentially a low-cost process because near-net shapes are
possible. Also, good mechanical properties (strength, toughness, etc.) have been
reported (Urquhart, 1991).
The main disadvantages of this process are:
(a) It is difficult to control the chemistry and produce an all-ceramic matrix by

this method. There is always some residual metal which is not easy to remove
completely.
(b) It is difficult to envision the use of such techniques for large, complex parts,
such as those required, say, for aerospace applications.
4.6 IN-SITU CHEMICAL REACTION TECHNIQUES
In-situ chemical reaction techniques to produce CMCs are extensions ofthose

used to produce monolithic ceramic bodies. We describe below some of the
more important techniques, viz., chemical vapor deposition (CVD) or chemical
vapor infiltration (CVI) and different types of reaction bonding techniques.
4.6.1 Chemical Vapor Deposition (CVD) and Chemical Vapor Infiltration

(CVI) .
When Chemical Vapor Deposition (CVD) technique is used to infiltration

rather large amounts of matrix material in fibrous preforms, it is called
Chemical Vapor Infiltration (CVI). Common ceramic matrix materials used are
SiC, Si3N4 , and HfC. The CVI method has been successfully employed by
several researchers to impregnate fibrous preforms (Fitzer and Hegen, 1979;
Fitzer and Schlichting, 1980; Fitzer and Gadow, 1986; Stinton et aI., 1986;
Burkland et aI., 1988). The preforms can consist of yarns, woven fabrics, or
filament wound three-dimensional shapes. Figure 4.6 shows a filament wound
Nicalon tube and a braided Nextel tube before CVI and after CVI.
Chemical vapor infiltration (CVI) has been used extensively for processing near
net shape CMCs. The first attempts at using CVI as a processing technique
were in densifyingporous graphite bodies with carbon (Bickerdike et aI., 1962).
In fact, about half of the commercially available carbon-carbon composites
today are made by CVI (Lowden et aI., 1993). CVI can be thought
of as a bulk form of chemical vapor deposition (CVD), which is widely used in
depositing thin coatings. The process involves deposition of the solid matrix
over an open volume, porous fibrous preform by the reaction and
decomposition of gases. An example of a CVI reaction is the deposition of
titanium diboride, which has a melting temperature of 3225 °C but can be
deposited at 900°C via CVI:
The HCe by-product is very common in such reactions. The solid materials are
deposited from gaseous reactants onto a heated substrate. A typical CVD or
CVI process would require a reactor with the following parts:
1. A vapor feed system.

2. A CVD reactor for heating the substrate and feeding the gaseous reactants.
3. An effluent system where exhaust gases are handled.
Figure 4.7 shows such a reactor in its simplest form. One can synthesize a
variety of ceramic matrixes such as oxides, glasses, ceramics, intermetallics by
CVD. There are two main variations of the CVI process. Isothermal chemical
vapor infiltration (ICVI) relies on diffusion for deposition (Naslain et aI., 1983;
Naslain, 1992). The preform is maintained at a uniform temperature while the
reactant gases are allowed to flow through the furnace and deposit the solid
species. To obtain a uniform matrix around the fibers, deposition is conducted
at low pressures and reactant concentrations. When the CVI process is carried
out isothermally, however, surface pores tend to close first, restricting the gas
flow to the interior of the preform. This phenomenon, sometimes referred to as
FILAMENT WOUND
NICALON
TUBE BRAIDED
NEXTEL
TUBE
CVD/CVI SIC
FILAMENT WOUND
N ICALON TUBE CVD/CVI SiC
BRAIDED NEXTEL
TUBE
Fig. 4.6. Filament wound Nicalon tube and a braided Nextel tube before and after being processed by Chemical Vapor Infiltration
(CVI). (Courtesy of Thermo Electron Corp.)
(') "!rJ
00000000
N
i o;Q'
.... 01:>0.
-
(') 1-
~ -..!
_.. Q
~
< CI}
~ (') ""0
o"'1 ::r
(1)
~
o '"
0
s·= ""...... ~
,..,. 'i:l
t:n ~ ~{{~~{{{{~*~~{~~~{{{~
q(')a. ~:r~~~~~~~~~~~~~~l\ ~
\r'..r-,.r-,.r-...r-n."\r',.r.,r-TIo.n.,..n.1"Ion.,.r-,..n.~n.'\r'n.n.ri", C':l
~. a, tr:I
g ~ $$~}$~~~~~~~$$~f:§$~ en
en
,.....(') ~~~~~~$$~~j:~~;:::~~~
"'n.n.ri",..n.n.n."n.ri"...r-n.n.ri".....n.,.r-n.n.n.n.~n.fton.,.r-n.n...,r;
~ \o.."'\r'~n.n.n.ri"ri"n.n.~,..n. n.'I'\on...........
..... ri"T\n.n.n.ri"'l"tr.n.,.r,
"'n.n...r'...r-n.n.TIo.n.ri",.r-,ro..,r-n.n.,.rn.n.n.n.n.n.,r-n.n.ri\r'n.~
tl ~
'Tr'ri",.r,r-,.r~ri"ri"n.~n.n.,.r,.rn.n.n.n.,.r-n.n.l'It.'"
~i
........
"'" (') "'n,'I"Io...:"ri"TIon."r-,r-"r-,.r,.r1'lon.,.r,.r,..,1"II..,r-,.,."'"r-"r-,.r","; ~
'n.n.TIo.n.~,.r-,{"n.n.,.r""n.n.n.,.r..r'n.n.,{"n.n.n.n.n.n.ri"n."';
§ e,
Q.< :J:J ~~~~~~~~*~~~~~~
\r',.r-,.r:...,n.n.n.,{",.rn.n.,.r..... n.,..,n.,.r,.r,.r".n.~,r,
cr~ ~,.r,.rn.n.n.n.,.r,.r,.r,.rn.~ri"ri",r-,..,n.~~~
"",g ':...'.~n.n.,.rn.n,{",.r,.r,.rn.n.,."'.,.r,.rn.n.,.r,.r,.rn.n.n.1"II.n.,
00-1 'n.1"lo.n.n.,.r.....n.,.r,.r,.r.,{"ri".....n."r-,r-.....n.,.r'l"lo""TIo.n.y\,n..".n.,.r,
c;> .... m>
e ~~-
OO Z
::;] :e ~
n
d ::s -I 'O"Ij ~
(1) ::n "'I ......
Q.q 00'
~
g. ~. 0'8
"'I~ c:l
(1) 0
S'" ><
= ::s C':l
~rn ~
-<
< .... 000000
~ "'" ~
en
"'1
'8'0a
tr:I
s· (')
::ng (1)
-...,
en
- rn n~
J:t,.....
a~
0"'-'
::s tn·
\: o~.
~~
""(1Q
"il 0
(")st-
~~
""'e,
canning, necessitates multiple cycles of impregnation, surface machining, and

reinfiltration to obtain an adequate density. One can avoid some of these
problems by using a forced gas flow and a temperature gradient approach to
chemical vapor infiltration (Stinton et aI., 1986; Caputo et aI., 1987). Forced
chemical vapor infiltration (FCVI) uses a combination of thermal gradients and
forced reactant flow to overcome the problems of slow diffusion and
permeability obtained in ICVI. This can eliminate, to some extent, the need for
multiple cycles. Thus, FCVI processes typically yield much shorter infiltration
times, while still obtaining uniform densification of the matrix and low residual
porosity. As a comparison, a 3 mm part infiltrated by ICVI could take several
weeks, while the same part infiltrated by FCVI would only take several hours.
As is true with all CVI processes, with increasing densification a point of
diminishing returns occurs, such that after a certain time the incremental
increase in density is not proportional to the time required for deposition.
The deposition process in CVI is governed primarily by mass and heat transfer,
as well as chemical kinetics (Carlsson, 1990; Gokoglu, 1992). A graphite holder
in contact with a water-cooled metallic gas distributor holds the fibrous
preform. The bottom and side surfaces thus stay cool while the top of the
fibrous preform is exposed to the hot zone, creating a steep thermal gradient.
The reactant gaseous mixture passes unreacted through the fibrous preform
because of the low temperature. When these gases reach the hot zone, they
decompose and deposit on and between the fibers to form the matrix.
As the matrix material gets deposited in the hot portion of the preform, the
preform density increases; and the hot zone moves progressively from the top
of the preform toward the bottom. When the composite is formed completely
at the top and is no longer permeable, the gases flow radially through the
preform, exiting from the vented retaining ring. To control deposition, the rate
of deposition must be maximized while minimizing density gradients.
Deposition reaction and mass transport are competing factors, so very rapid
deposition results in the exterior of the preform being well infiltrated, while
severe density gradients and a large amount of porosity are present within the
preform. Very slow deposition rates, on the other hand, require large times and
are not economically feasible. A balance between the two factors is required for
optimum infiltration.
Commonly, the process involves a decomposition of a chemical compound in

the vapor form to yield the desired ceramic matrix on and in between the fibers
in a preform. For example, methyltrichlorosilane (CH 3 SiCI3 ), the starting
material to obtain SiC, is decomposed at between 1200-1400 K:
The vapors of SiC deposit as solid phases on and between the fibers in a
freestanding preform to form the matrix. The CVI process is very slow because
it involves diffusion ofthe reactant species to the fibrous substrate, followed by
outflow of the gaseous reactant products. The CVI process of making a ceramic
matrix is, indeed, a kind of low stress and low temperature CVD process, and
thus avoids some of the problem associated with high temperature ceramic
processing. Using CVI, when processing CFCMCs, one can deposit the
interfacial coating on the fibers as well as the matrix in situ. For example, for
Nicalon/SiC composites with a carbon interface, the carbon layer is deposited
first, and then the SiC matrix is infiltrated without changing the preform
conditions. The fibrous preforms are stacked layer by layer between perforated
plates, through which the gases pass during infiltration. The carbon coating is
typically deposited by means of a hydrocarbon gas at around 1000°C and
reduced pressure, to protect the fibers.
The graphitic coating on the fibers has a characteristic aligned structure of the
basal planes. These basal planes are parallel to the fiber direction, but
perpendicular to the incoming crack front, so deflection of cracks at the weakly
bonded basal planes takes place instead of fracturing the fibers. The softer c-
axis of the graphite is also aligned in the perpendicular direction, to
accommodate the thermal residual stresses that arise from processing. The
matrix consists of a nucleation zone in a small region at the coating/matrix
interface. After this, long columnar grains are seen which are perpendicular to
the surface of the fiber. The preferred orientation is such that the (111) planes
are aligned parallel to the fibers. The grains are composed predominantly ofb-
SiC with a cubic structure with small disordered regions of a-SiC. For CVI
composites reinforced with woven fiber fabrics, the nature of the porosity is
trimodal. Macroporosity is found between fiber bundles and between layers of
fabric, with pore sizes less than 100 mm. Microporosity occurs between fibers
in the fiber bundle, and the pore size is usually on the order of 10 mm. Lowden
et al. (1993), found that 70% of the pore volume was in the form of
microporosity within the fiber bundle, 25% between the cloth layers, and 5%
as holes between layers of the fabric.
This variant of CVI which combines forced gas flow and temperature gradient
avoids some of the problems mentioned earlier. Under these modified
conditions, 70-90% dense SiC and Si3N4 matrixes can be impregnated in SiC
and Si3N4 fibrous preforms in less than a day. Under conditions of plain CVI,
it would take several weeks to achieve such densities, i.e., one can reduce the
processing time from several days to less than 24 hours. One can also avoid
using binders in this process with their attendant problems of incomplete
removal. The use of a graphite holder simplifies the fabrication of the preform
and the application of a moderate pressure to the preform can result in a higher
than normal fiber volume fraction in the final product. The final obtainable
density in a ceramic body is limited by the fact that closed porosity starts at
about 93-94% oftheoretical density. It is difficult to impregnate past this point.
Advantages of a CVI technique or any variant thereof include:
(a) Good mechanical properties at high temperatures.

(b) Large, complex shapes can be produced in a near-net shape.
(c) Considerable flexibility in the fibers and matrices which can be used (oxide
and non oxide).
(d) It is a pressureless process and relatively low temperatures are used, for
example, compared to the temperatures involved in hot-pressing.
Among the disadvantages, one should mention:
(a) The process is slow and expensive.
4.6.2 Reactive Consolidation or Liquid Phase Sintering
The term reaction bonding is used rather loosely in literature. Some researchers
use it to encompass all processes not involving hot pressing. We shall restrict
the term to processes involving chemical reaction(s) between components to
produce the desired end product. Reactive consolidation or liquid phase
sintering are the other more descriptive terms of such processes. Some
commercial success has been obtained in making silicon carbide based-
composites by reaction bonding. Siliconized silicon carbide is the name give to
a composite of SiC grains in a silicon matrix. Commercially, such composites
are available under different designations (e.g., K- T, Refel, and NC-435) from
different producers.
Polymers are used to bond the preforms containing carbon in the form of carbon
and silicon carbide, followed by pyrolysis of the polymer to give more carbon
(Forrest et aI., 1972; Willermet et aI., 1978). One such CMC, called Refel
(Forrest et aI., 1972), is formulated to give a minimum of free silicon, which in
practice is limited to about 10 vol. %. The infiltration is done under reduced
pressure at temperatures up to 1700 °C and for hours. Hillig et al. (1975) used
a low modulus carbon fiber instead of carbon powder in the preform. Molten
silicon reacts with carbon fibers to form SiC. The original geometry of the
carbon fibers is retained. A big advantage of SiC/Si composite is that the
constituents are in chemical eqUilibrium and they have closely matched thermal
expansion coefficients. Carbon fiber in the form of cloth, tow, felt, or matte is
used as a precursor. A preform is made of carbon fiber and infiltrated with
liquid silicon. Silicon reacts with carbon fibers to form SiC fibers in a Si
matrix. Typical composition of the resultant composite is Si (30 - 50 %) + SiC
fiber. Silicon matrix limits the use temperature to about 1400 OC.
In another version of this process, a liquid phase forms as a result of an

exothermic reaction between elemental powders. A good example is that from
the field of intermetallics, e.g., nickel aluminides. The following steps are
involved:
(a) Mix nickel and aluminum in stoichiometric proportions

(b) Cold isostatic press to 70% theoretical density to obtain a green body.
(c) Vacuum encapsulate the green body in a 304 stainless steel can.
(d) Subject the canned material to reactive hot isostatic pressing.
4.7 SOL-GEL
Sol-gel techniques, which have been used for making conventional ceramic
materials, can also be used to make ceramic matrix materials in the interstices
of a fibrous preform. We have described the sol-gel technique in chapter 2.
Very briefly, a solution containing metal compounds, e.g., a metal alkoxide,
acetate, or halide is reacted to form a sol. The sol is converted to a gel, which
in turn is subjected to controlled heating to produce the desired end-product: a
glass, a glass-ceramic, or a ceramic. Characteristically, the gel to ceramic
conversion temperature is much lower than that required in a conventional
melting or sintering process. A schematic of a typical sol-gel process for
processing CMCs is given in Fig. 4.8. Some of the advantages of these
techniques for making composites are the same as the ones for monolithic
ceramics, viz., lower processing temperatures, greater compositional
homogeneity in single phase matrices, potential for producing unique
multiphase matrix materials, etc. Specifically, in regard to composite material
fabrication, the sol-gel technique allows processing via liquids oflow viscosity
such as the ones derived from alkoxides. Covalent ceramics, for example, can
be produced by pyrolysis of polymeric precursors at temperatures as low as
1400 °C and with yields greater than those in CVD processes. Among the
disadvantages of sol-gel are high shrinkage and low yield compared to slurry
techniques. The fiber network provides a very high surface area for the matrix
to gel on. Consequently, the shrinkage during the drying step, frequently, to
results in a large density of cracks in the matrix. Generally, repeated
impregnations are required to produce a substantially dense matrix. It is easy
to see that many of the polymer handling and processing techniques can be used
for sol-gel as well. Impregnation of fibrous preforms in vacuum and filament
winding are two important techniques. In filament winding, fiber tows or
rovings are passed through a tank containing the sol and the impregnated tow
is wound on a mandrel to a desired shape and thickness. The sol is converted
to gel and the structure is removed from the mandrel. A final heat treatment
then converts the gel to a ceramic or glass matrix.
The sol-gel technique can also be used to prepare prepregs by the slurry
o
o .....t - - --
o
o
~t---- Hea ting Coils
o
HcOl. to Con\Iertlhc Gel
into Gla s o rCeramic
Fig. 4.8 Schematic of sol-gel process.
infiltration method. The sol in the slurry acts as a binder and coats fibers and
glass particles. The binder burnout step is thus eliminated because the binder,
being of the same composition of the matrix, becomes part of the glass matrix.
4.8 POLYMER INFILTRATION AND PYROLYSIS (PIP)
Polymeric precursors can also be used to form a ceramic matrix in a composite.

Because of the generally high cost of processing CMCs, polymer infiltration
and pyrolysis (PIP) is an attractive processing route because of its relatively low
cost, while maintaining small amounts of residual porosity and minimal
degradation of the fibers (Greil, 1995; French, 1996; Hurwitz et aI., 1989).
Moreover, this approach allows near net-shape molding and fabrication
technology that is able to produce nearly fully-dense composites (Sato et aI.,
1995; Gonon et aI., 1995). In PIP, the fibers are infiltrated with an organic
polymer, which is heated to fairly high temperatures and pyrolyzed to form a
ceramic matrix. Due to the relatively low yield during the conversion from
polymer to ceramic, multiple infiltrations are required to obtain an acceptable
density of the composite.
Polymeric precursors for ceramic matrices allow one to use conventional

polymer composite fabrication technology that is readily available, and take
advantage of processes used to make polymer matrix composites (French, 1996,
Hurwitz et aI., 1989). These include complex shape forming and fabrication.
Furthermore, by processing and pyrolyzing at lower temperatures (compared to
sintering and hot-pressing, for example) one can avoid fiber degradation and the
formation of unwanted reaction products at the fiber/matrix interface. French
(1996) lists some desirable characteristics in a preceramic polymer:
• High ceramic yield from polymer precursor.

• Precursor that yields a ceramic with low free carbon content (which
will oxidize at high temperatures).
• Controllable molecular weight, which allows for solvent solubility and
control over viscosity for fabrication purposes.
• Low temperature crosslinking of the polymer which allows resin to
harden and maintain its dimensions during the pyrolysis process.
• Low cost and toxicity.
Most preceramic polymer precursors are formed from chloro-organosilicon

compounds to form poly(silanes), poly(carbosilanes), poly(silazanes),
poly(borosilanes), poly(silsesquioxanes), and poly(carbosiloxanes) (Greil,
1995). The synthesis reaction involves the dechlorination of the chlorinated
silane monomers. Since a lot of the chlorosilane monomers are formed as
byproducts in the silicone industry, they are inexpensive and readily available.
The monomers can be further controlled by an appropriate amount of
branching, which controls important properties such as the viscosity of the
precursor as well as the amount of ceramic yield.
All silicon based polymer precursors lead to an amorphous ceramic matrix,

where silicon atoms are tetrahedrally arranged with non-silicon atoms. This
arrangement is similar to that found in amorphous silica (French, 1996). High
temperature treatments typically lead to crystallization and slight densification
of the matrix, which results in shrinkage. At high temperatures, the amorphous
ceramic begins to form small domains of crystalline phase, which are more
thermodynamically stable (Lipowitz et aI., 1990). Si-C matrices derived from
polycarbosilane begin to crystallize at 1100-1200 °C while Si-C-O
(polysiloxanes) and Si-N-C (polysilazanes) remain amorphous to 1300-1400
°C.
Typically, one chooses the range of the molecular mass of the polymer,
followed by shaping of the product (Greil, 1995). The polymer is then
crosslinked and finally pyrolyzed in an inert or reactive atmosphere (e.g., NH3)
at temperatures between 1000-1400°C. The pyrolysis involves the following
three steps. In the first step, between 550-880 °C, an amorphous hydrogenated
compound of the type Si(C.ObNcBd) is formed. The second step involves
nucleation of crystalline precipitates such as SiC, Si3N4 , and Si02 at
temperatures between 1200 and 1600°C. Grain coarsening may also result from
consumption of any residual amorphous phase and reduction in the amount of
oxygen due to vaporization ofSiO and CO. Porosity is typically of the order of
5-20 vol.% with pore sizes of the order of 1 to 50 nm. It should be noted that the
average pore size and volume fraction of pores decreases with increasing
pyrolysis temperature, since the amount of densification (and shrinkage)
becomes irreversible at temperatures above the maximum pyrolysis
temperature.
The main disadvantage of PIP is the low yield of ceramic during the polymer-
to-ceramic transformation and the resulting shrinkage, which typically causes
cracking in the matrix during fabrication (Greil, 1995; French, 1996; Hurwitz
et aI., 1989; Lundberg et aI., 1990; Sirieix et aI., 1990; Sato et al. 1995; Gonon
et aI., 1995). Because of the shrinkage and weight loss during pyrolysis,
residual porosity after a single impregnation is of the order of 20-30%. To
reduce the amount of residual porosity, multiple impregnations are needed.
Reimpregnation is typically conducted with a very low viscosity prepolymer,
so that the slurry may wet and infiltrate the micropores existing in the preform.
Usually, reimpregnation is done by immersing the part in the liquid polymer in
a vacuum bag, while higher viscosity polymers require pressure impregnation.
Typically, the amount of porosity will reduce from 35 % to less than 10 % after
about 5 impregnations.
Significant gas evolution also occurs during pyrolysis (French, 1996). Thus, it
is advisable to allow these volatile gases to slowly diffuse out of the matrix,
especially for thicker parts. Typically, pyrolysis cycles ramp to 800-1400 °C
over periods of 1-2 days, to avoid delamination. Recall that pyrolysis must be
done at a temperature below the crystallization temperature of the matrix (or
large volume changes will occur) and below the degradation temperature of the
reinforcing fibers. The pyrolysis atmosphere is most commonly argon and
nitrogen, although in ammonia a pure amorphous silicon nitride with low
amounts of free carbon, can be obtained (Burns and Chandra, 1989). Such an
atmosphere may also lead to the formation of nitrides from the reaction of filler
particles (Erny et aI., 1993). With the formation of the ceramic, the gaseous by-
product reaction can be written as (Greil, 1995):
P (s, Q) ~ C (s) + G (g)
where P is the polymer, C is the ceramic, and G is the gaseous by-product. With
the loss of the volatile gaseous products, the ceramic yield, a is the ratio of
ceramic formed and the initial amount of polymer:
a = m(C) = 1- meG)
m(P) m(P)
Following Greil (1995), we can define another term, density ratio as

~ = density of the ceramic product, p(C)/ density of the polymer precursor, pep)
P(P)
{3= P(C)
Two extreme cases of polymer-ceramic conversion can be considered. If the

volume is not constrained, then diffusional flow will cause the pores to be filled
but a high amount of shrinkage will take place. The maximum volume change
that occurs during conversion can be written as:
'V = a~ - 1
If the volume is constrained (Le., 'V = 0) then shrinkage does not occur, but a
large amount of residual porosity is present. The maximum amount of porosity
can be written as:
1t =1- a~
When the product a~ is smaller than unity, for example in organosilicon

polymer precursors, it is almost impossible to obtain fully dense ceramic
products from a polymer precursor without shrinkage. For example, for
pyrolysis of filler-free poly(silazane) to form bulk Si3N 4, either a large amount
of porosity (>8%) or a large amount of shrinkage (20%) took place (Riedel et
aI., 1992).
Fitzer and Gadow (1986) used repeated infiltration and in situ thermal
decomposition of porous reaction-bonded ceramics; for example infiltration of
porous preforms of silicon carbide and silicon nitride with silazanes and
polycarbosilanes, to obtain Si3NiSiC composites. The following steps are taken
in processing the composites:
(i) Porous SiC or Si3N4 fibrous preform with some binder phase is prepared.
(ii) Fibrous preform is evacuated in an autoclave.
(iii) Preform is infiltrated with molten precursors, silazanes or polycarbosilanes,
at high temperature (780 K) and the argon or nitrogen pressure is slowly
increased from 2 to 40 MPa. The high temperature results in a transformation
of the oligomer silane to polycarbosilane and simultaneous polymerization at
high pressures.
(iv) Infiltrated samples are cooled and treated with solvents.
(v) Samples are placed in an autoclave and the organosilicon polymer matrix
is thermally decomposed in an inert atmosphere at a high pressure and at
temperature in the 800-1300 K range.
(vi) Steps (ii) through (v) are repeated to attain an adequate density. To produce
an optimum matrix crystal structure, the material is annealed in the l300- 1800
K range.
Polymer-derived ceramic matrix composites, similar to carbon/carbon

composites, typically have a cracked matrix from processing as well as a
number of small voids or pores. The large amount of shrinkage and cracking in
the matrix can be contained, to some extent, by the additions of particulate
fillers to the matrix, which, when added to the polymer reduce shrinkage and
stiffen the matrix material in the composite (Greil, 1995). Figure 4.9(a) shows
a schematic of filler-free versus active filler pyrolysis. Figure 4.9(b) shows the
microstructure of a PIP CMC, indicating some residual porosity and a clear
"filler network" in the matrix of the composite (Chawla et ai., 1998). Particulate
or whisker ceramics used as fillers in the polymeric matrix can serve a variety
of purposes (French, 1996):
• Reduce and disrupt the formation of matrix cracks that form during
shrinkage of the polymer.
• Enhance ceramic yield by forming reaction products during pyrolysis.
• Strengthen and toughen the weak amorphous matrix and increase the
interlaminar shear strength ofthe composite.
The filler must be submicrometer in size in order to penetrate the tow bundle
and the filler's coefficient of thermal expansion must match that of the
polymeric matrix. It should be noted that the filler must not be used in very high
fractions and the slurry should not be forced into the reinforcing fibers since
abrasion of the fiber fabric may take place. This is especially true with hard,
angular fillers or ceramic whiskers. Typically, the volume fraction of filler is
15-25% of the matrix volume fraction. High filler loading may result in an
increase in interply spacing and lower volume fraction of fibers. When an
"active" filler phase is added to the polymer, it reacts with solid or gaseous
decomposition products to form new carbide phases (Greil, 1995):
P (s, Q) + T (s) ---+ C (s) + M (s) +G (g)
where T is the active filler, and M is the carbide phase formed. The other
symbols have the same meaning as described in the previous expression. The
maximum volume change of a precursor containing an active filler, \/f*, can be
expressed as:
where V T' is the critical volume fraction of filler that determined maximum
particle packing density of the reacted filler phase in the pyrolyzed product, a ™
and WM describe mass change of the filler phase and density ratio. For a™ and
~TM > I, volume expansion of the filler phase may compensate for polymer
shrinkage during pyrolysis.
If one assumes isotropic dimensional changes, the linear shrinkage e is related

to the volume shrinkage \If by:
Thus, by controlling the amount of filler, the degree of shrinkage can be

controlled.
Fiber architecture may have an impact in regard to polymer infiltration and

pyrolysis (PIP). One of the key factors is wetting of the fiber bundles. During
pyrolysis, the precursor shrinks around the fibers, so cracks are introduced. For
example, two dimensional woven fabrics seem to have less propensity in
developing interlaminar cracks than do cross-ply or unidirectional architectures.
Satin weaves are preferred versus plain weaves because more uniform cracking
is achieved and large cracks between weave crossover points are avoided
(Hurwitz, 1992). Due to the looser nature of the satin weave (it is more
drapable), better wetting and densification may take place, although the loose
nature of the weave also makes it more difficult to handle.
4.9 SELF-PROPAGATING HIGH TEMPERATURE SYNTHESIS (SHS)
This technique can be used to produce a variety of refractory materials. The

main disadvantage is that SHS products are very porous, because of the fairly
large porosity present in the original mix of reactants and the large volume
change that results when the reactants transform to the products. Any adsorbed
gases at the elevated temperatures used during this process can also add to the
porosity of the final product. Synthesis concomitant with densification can
improve the situation to some extent. This will involve application of high
pressure during the combustion or immediately after the completion of the
combustion reaction when the product temperature is still quite high. Hot
pressing, rolling, and shock waves are some of the techniques used to apply the
necessary pressure.
The SHS technique involves synthesis ofcompounds without an external source

ofenergy. One exploits exothermic reactions to synthesize ceramic compounds,
which are difficult to fabricate by conventional techniques. For example, one
can mix titanium powder and carbon black, cold press the mixture, and ignite
Filler-Free Active Filler

Pyrolysis Pyrolysis
• •• •
•• •
• •
Filler Network
(a)
(b)
Fig. 4.9 (a) Schematic oftiller-free versus active filler pyrolysis and (b) microstructure
of a woven Nicalon/Si-C-O-N matrix composite with a SiC filler, showing small
amounts of residual porosity and a clear filler free-network in the matrix of the
composite (Chawla et al., 1998).
the compact at the top in a cold-walled vessel. A combustion wave will pass
through the compact, giving titanium carbide.
Among the salient features of SHS are:
(a) High combustion temperature (up to 4000 0c)

(b) Simple, low-cost equipment
(c) Good control of chemical composition
(d) Different shapes and forms can be obtained
Many ceramics such as borides, carbides, nitrides, silicides, and sialons as well
as composites such as SiCw + Al2 0 3 have been synthesized by means of SHS.
The SHS process gives a weakly bonded compact. Therefore, the process is
generally followed by breaking the compact, milling, and consolidation by
some technique such as HIPing. Explosive or dynamic compaction can result
in a relatively dense product. A good example of an SHS process to make
composites is the proprietary process of Martin Marietta Corp., USA, called the
XD™ Process, wherein exothermic reactions are used to produce multiphase
alloy powders. These are hot pressed at 1450 °C to full density. Reinforcement
in the form of particles, whiskers, and platelets can be added to the master alloy
to make a composite. A good example is that of TiB2 particles, about 1 !lm
diameter, distributed in intermetallic matrixes such as TiAI, TiAI + Ti3AI, NiAI,
etc.
4.10 ELECTROPHORETIC DEPOSITION
The phenomenon of electrophoresis has been known since the beginning of the
nineteenth century, but its applications in are of processing of ceramics and
ceramic composites is relatively recent. Electrophoretic deposition (EPD)
should not be confused with electroplating. In electroplating, ions are the
moving species and they undergo ion reduction on deposition. In EPD, on the
other hand, solid particles migrate with no charge reduction on deposition.
Also, the deposition rate in EPD is» 1 mmlmin while in electroplating it» 0.1
mmlmin. Electrophoretic deposition is a relatively simple and inexpensive
technique, which can be profitably exploited for infiltration of tightly woven
fibre preforms (Illston et aI., 1993). EPD makes use of nanoscale ceramic
particles in a stable non-agglomerated form (such as in a sol or colloidal
suspension) and exploits their net surface electrostatic charge characteristics
while in suspension. On application of an electric field the particles will
migrate toward and deposit on an electrode. If the deposition electrode is
replaced by a conducting fibrous preform, the suspended particles will be
attracted into and deposited within it, providing an appropriate means of
effectively infiltrating densely packed fibrous bundles. A schematic diagram of
the basic EPD cell is shown in Fig. 4.10a while an SEM micrograph of a
Nicalon fiber mat infiltrated with a mixed sol of mullite composition using
electrophoretic deposition is shown in Fig. 4.1 Ob (Kaya et aI., 2000). Note the
high-level of ceramic matrix infiltration achieved by this technique.
The movement of ceramic sol particles in an aqueous suspension within an
electric field is governed by the field strength, and the pH, ionic strength and
viscosity of the solution (Illston et aI., 1993). The electrophoretic mobility of
charged particles in a suspension is given by the following equation called the
Smoluchowski equation (Brown and Salt, 1965):
electrophoretic mobility = VIE = ES/41tT]
where V is the velocity, E is the field strength, e is the dielectric constant, z is

the zeta potential, and h is the viscosity. The zeta potential is a parameter for
characterizing a suspension. It can be determined by measuring particle velocity
in an electric field. According to the above equation, a suitable suspension for
EPD should have: a high particle surface charge, a high dielectric constant of
the liquid phase, and a low viscosity. In addition, a low conductivity of the
suspending medium to minimize solvent transport would be desirable
(Boccaccini et aI., 2001).
(+) Digital ["lance H

(+)
Boehmite sol
Fig. 4.10 (a) Schematic of a basic electrophoretic deposition (EPD) cell (after Kaya et
aI., 2000) (b) Transverse section of a Nicalon fiber mat infiltrated with a mixed sol of
mullite composition by electrophoretic deposition (Boccaccini, et aI., 2001).
(b)
Fig 4.10 - contd
4.11 SUMMARY
In this chapter, we have described a number of processing techniques for

ceramic matrix composites, fiber reinforced as well as particle reinforced. Some
of the techniques are mere extensions of the conventional powder-based
processing techniques used for monolithic ceramics. Others are novel
techniques devised specifically for processing of ceramic matrix composites.
Among these the most important and promising are summarized in Table4.1.
Table 4.1 Summary ofCMC fabrication processes and examples of typical composite
systems fabricated by these processes.
Process Examples
Slurry infiltration SiC/Glass-Ceramic, CarboniGlass-
(ply stacking and hot pressing) ceramic, C/glass, Mullite/Glass
Powder processing and hot pressing SiCIAIP3' AIP/AlP3
Gas-liquid metal reaction SiC/AI 20 3, SiC/SiC
(Lanxide)
Sol-gel C/glass, Mullite/Mullite
(infiltration and sintering/hot pressing)
Chemical vapor infiltration SiC/SiC, C/SiC
(infiltration of a woven preform)
Polymer conversion C/C, C/SiC, SiC/Si-C-N
(infiltration and pyrolysis)
l36 PROCESSING OF CERAMIC MATRIX COMPOSITES
REFERENCES
Barclay, S.J., J.R. Fox, and H.K. Bowen (1987) J. Mater Sci., 22, 4403.
Bhatt R. T. (1986), National Astronautics and Space Administration, NASA TN-88814.
Bhatt R.T. (1990) J. Mater. Sci., 25, 3401.
Bickerdike, R.L., A.R.G. Brown, G. Hughes, and H. Ranson (1962) Proc. o/the Fifth
Conference on Carbon, S. Mrosowski, M.C. Studebaker, and P.L. Walker (eds.),
Pergamon Press, Buffalo, NY, p. 575.
Boccaccini, A.R., C. Kaya, and K. K. Chawla (2001) Composites Part A, 32,997.
Bordia, R.K.. and R. R~ (1988) J. Am. Ceram. Soc., 71, 302.
Brown, D.R. and F.W. Salt (1965)J. App. Chem., 15,40.
Brennan, J.J. and K.M. Prewo (1982) J. Mater. Sci., 17,2371.
Burkland, C.V., W.E. Bustamante, R. Klacka. and J.-M Yang (1988) in Whisker- and
Fiber-Toughened Ceramics, ASM IntI., Materials Park, Ohio, p. 225.
Bums, G.T. and G. Chandra (1989) J. Am. Ceram. Soc., 72,334.
Caputo, A.J., D.P. Stinton and R.A. Lowden.(1987) Am. Ceram. Soc. Bull., 66,1987,
368.
Carlsson, J.O. (1990) Thin Solid Films, 168, 19.
Chawla, K.K. (1987) Composite Materia/s, 2 nd ed.,Springer-Verlag, New York.
Chawla N., K.K. Chawla, M. Koopman, B. Patel, C.C. Coffm, and J.I. Eldridge (2001)
Compo Sci. Tech., 61, 1923.
Chawla, N., Y.K. Tur, J.W. Holmes, J.R. Barber, andA. Szweda(1988)J. Am. Ceram.
Soc., 81, 1221.
Chawla, N. (1997), Metall. & Mater. Trans. A, 28A, 2423.
Claussen, N., T. Le, and S. Wu (1989) J. Eur. Ceram. Soc., 5, 29.
Claussen, N., S. Wu, and D. Holtz (1994) J. Eur. Ceram. Soc. ,14, 209.
Cornie, J.A. , Y.-M. Chiang, D.R. Uhlmann, A. Mortensen, and J.M. Collins (1986)
Am. Ceram. Soc. Bull., 65, 293.
De Jonghe, L.C., M.N. Rahaman, C.H. Hseuh (1986) Acta Met., 39, 1467.
Erny, T., M. Seibold, O. Jarchow, and P. Greil (1993) J. Am. Ceram. Soc. 76,207.
Fitzer, E. and D. Hegen (1979) Angew. Chem., 91, 316.
Fitzer, E. and J. Schlichting (1980) Z. Werkstofftech., 11, 330.
Fitzer, E. and R. Gadow (1986) Am. Ceram. Soc. Bull., 65, 326.
Forrest, C. W., P. Kennedy, andJ. V. Shennan (1972) Special Ceramics, British Ceramic
Research Association, Stoke-on-Trent, U.K., vol. 5, p.99.
French, J.E. (1996) in Handbook o/Continuous Fiber Ceramic Composites, American
Ceramic Society, Westerville, OH,), p. 269.
Gokoglu, S.A.(1992) Mater. Res. Soc. Symp. Proc. vol. 250, p.17.
Gonon, M.F., G. Fantozzi, M. Murat, and J.P. Disson (1995) J. Eur. Ceram. Soc. 15,
185.
Greil, P. (1995) J. Am. Ceram. Soc., 78, 835.
Homeny, J., W.L. Vaughn, and M.K. Ferber (1987) A mer. Cer. Soc. Bull., 67, 333.
Hurwitz, F.I., J.Z. Gyekenyesi, and P.J. Conroy (1989) Ceram. Eng. Sci. Proc., 10,750.
Hurwitz, F.J. (1992) NASA Tech. Memo, 105754.
Illston, T.J, C.B. Ponton, P.M. Marquis, E.G. Butler (1993) Third Euroceramics, vol.
l37 PROCESSING OF CERAMIC MATRIX COMPOSITES
1, P. Duran and J.F. Fernandez (eds.), Faenza Editirice Iberica, Madrid, pp. 419-424.
Kaya,C., A.R Boccaccini., and K.K. Chawla (2000) J. Am. Ceram Soc., 20, 1189.
Kristofferson, A., A. Warren, J. Brandt, and R. Lundberg (1993) in Proc. Int. Cont
HTCMC-I, (ed. R. Naslain et al.),Woodhead Pub., Cambridge, UK,., p. 151.
Lipowitz, J., lA. Rabe, L.K. Frevel, and RL.Miller (1990) J. Mater. Sci., 25, 2118.
Liu, H.Y., N. Claussen, M.J. Hoffmann, and G. Petzow (1991)J. Eur.Ceram. Soc. 7,
41.
Lowden, RA, D.P. Stinton, and T.M. Besmann (1993) in Handbook o/Continuous
Fiber Ceramic Matrix Composites, Amer. Ceram. Soc., Westerrville, 0 H, p. 205.
Lundberg, R., R. Pompe, and R. Carlsson (1990) Compo Sci. Tech. 37, 165.
Naslain, R et al. (1983) Euro-CVD-Four, The Centre, Eindhoven, p. 293.
Naslain, R. (1992) in Ceramic Matrix Composites, R. Warren, ed., Chapman and Hall,
London, p. 199.
Phillips, D.C. (1983) in Fabrication o/Composites, North-Holland, Amsterdam, p. 373.
Prewo, K.M.(1982)J. Mater. Sci., 17,3549.
Prewo, K.M.(1986) in Tailoring Multiphase and Composite Ceramics, Materials
Science Research, Plenum Press, New York, vol. 20, p. 529.
Prewo, K.M. and J.l Brennan (1980) J. Mater. Sci., 15,463.
Prewo K.M., J.J. Brennan, and G.K. Layden (1986) Am. Ceram. Soc. Bull., 65, 305.
Rahaman, M.N. and L.C. De Jonghe (1987) J. Am. Ceram. Soc., 70, C-348.
Raj, R.and R.K Bordia (1989) Acta Met., 32, 1003.
Riedel, R., G. Passing, H. Schonfelder, and R.J. Brook (1992) Nature, 355, 355.
Sacks, M.D., H.W. Lee, and O.E. Rojas (1987) J. Am. Ceram. Soc., 70, C-348.
Sambell, R.A.J., D.C. Phillips, and D.H. Bowen (1974) in Carbon Fibres: Their Place
in Modern Technology, The Plastics Institute, London, p. 16/9.
Sato, K., H. Morozumi, A Tezuka, O. Funayama, and T. Isoda (1995), in High
Temperature Ceramic-Matrix Composites II, American Ceramic Society, Westerville,
OH,p.199.
Shalek, P.D., J.J. Petrovic, G.F. Hurley, F.D. Gac (1986) Am. Ceram. Soc. Bu/l., 65,
351.
Sirieix, F., P. Goursat, A. Lecomte, and A Dauger (1990) Camp. Sci. Tech., 37, 7.
Stinton, D.P., A.J. Caputo, and R.A. Lowden (1986) Am. Ceram. Soc. Bull., 65, 347.
Stinton, D.P., AJ. Caputo, R.A Lowden, and T.M. Besmann (1986) Ceram. Eng. Sci.
Proc., 7, 983.
Urquhart, AW. (1991) Mater. Sci. Eng., A144, 75.
Wu, S. and N. Claussen (1994) J. Amer. Ceram. Soc., 77,2898.
Yang, M. and R. Stevens (1990) J. Mater. Sci., 25, 4658.
SUGGESTED READING
Krenkel, W., R Naslain, and H. Schneider (eds.) (2001) High Temperature Ceramic
Matrix Composites, Wiley-VCH, Weinheim, Germany.
Kroke, E., Y.-L. Li, C. Konetschny, E. Lecomte, C. Fasel, and R. Riedel (2000)
Silazane derived ceramics and related materials, Mater. Sci. Eng., R26, 97.
Narula, C.K. (1995) Ceramic Precursor Technology and its Applications, Marcel
Dekker, New York.
National Materials Advisory Board (1991) High Temperature Metal and Ceramic
Matrix Compositesfor Oxidizing Atmosphere Applications, NMAB-376, Washington,
DC.
Cliapter 5
INTERFACE
The interface region has a great deal of importance in determining the ultimate
properties of a given composite. There are two main reasons for this importance
of the interface region:
(i) the interface occupies a very large area per unit volume in composites (see
Sec. 5.1 below), and
(ii) in general, the matrix and reinforcement form a system that is not in
thermodynamic equilibrium. This can lead to a series of interaction at the
interface.
In this chapter, we describe the general features of an interface in a ceramic

matrix composite and some specific tests that have been developed to determine
the mechanical characteristics of interface in CMCs. For details, regarding the
microstructure of specific CMC systems as well as interface engineering to
obtain an enhanced toughness, the reader should turn to chapters 7 and 9.
We can define an interface between any two phases (for example, between a
reinforcement and a matrix) as a surface across which there occurs a
discontinuity in one or more material parameters or properties, such as lattice
parameter, density, elastic modulus, thermal expansion coefficient, strength,
fracture toughness, etc. This definition of an interface treats it as a
bidimensional entity over which the discontinuity occurs rather abruptly. It
would, however, be more realistic to treat it as an interface zone or region
having a certain thickness across which the discontinuity in property occurs
gradually.
5.1 INTERFACIAL AREA IN COMPOSITES
Let us examine in some detail the extent of interfacial area in composites.

140 INTERFACE
Consider a fiber reinforced composite. It can readily be shown that, for a given
fiber volume fraction, the total area of the fiber/matrix interface increases as the
fiber diameter decreases. Consider a composite bar containing N continuous,
unidirectionally aligned fibers, each of diameter d. Let the length, width, and
height of the composite be~, w, and h, respectively. Then, for the fiber volume
fraction, we can write
volume of fibers (5.1)

vj =
volume of composite ~wh
The area of the fiber/matrix interface, lA, can be expressed as

fA =N 1t d ~ (5.2)
From Eqs. (5.1) and (5.2), we get
fA = (4 ~ w h)· (Vj ) ( ~) (5.3)
Equation (5.3) says that the fiber/matrix interfacial area, for a given fiber
volume fraction, varies inversely with the fiber diameter. Let us take the volume
of the composite to be 1 m3, and in order to simplify Eq. (5.3), let the fiber
volume fraction be 0.25. Then, Eq. (5.3) reduces to
(5.4)
d
where IA is now the interfacial area per unit volume in a composite containing
25% fibers by volume. Figure 5.1 shows a plot of the interfacial area as a
function of lid, where we have indicated some of the commercial fibers. As one
goes from the extremely tiny whiskers to rather large diameter CVD type fibers
such as SiC, B, etc., the interfacial area decreases as lid.
We can do a similar calculation for a particle reinforced composite. Assume N

spherical ceramic particles embedded in a spherical shell ofthe matrix. Let the
total volume of the composite be 1 m3 • Then, we can write
141 INTERFACE
10'
10 1
10°
10- 1
10-'
10- 4
10° 10 1 10' 103
a,l'm
Fig. 5.1 Interfacial area, IA (in m for a fiber volume fraction of 0.25 as a function of
2)
fiber diameter, d (in ~m). For a given fiber volume fraction, the finer the fiber, the
higher the interfacial area.
volume of particles
volume of composite 1
Area of the particle/matrix interface per unit volume,

6 V
fA = N 1t d2 = -_P
d
Taking Vp= 0.25, we have
1.5
(5.5)
d
Equations (5.4) and (5.5) show similar inverse dependence of interfacial area
on the diameter of the fiber or the particle.
142 INTERFACE
The type of bonding at the interface, which controls the interface strength, is of
obvious importance in ceramic matrix composites. In ceramic matrix
composites made by melt-infiltration, there is another aspect that assumes
importance, viz., wettability. We briefly discuss these concepts first and then
discuss the types of interface bonding, methods of interface testing, etc.
5.2 CRYSTALLOGRAPHIC NATURE OF THE INTERFACE
In crystallographic terms, one can describe an interface between two phases as

coherent, semi-coherent, or incoherent. A coherent interface implies a one-to-
one correspondence between lattice planes on the two sides of the interface. A
coherent interface thus has some coherency strains associated with it because
of the straining of the lattice planes in the two phases to provide the continuity
at the interface. An incoherent interface, on the other hand, consists of such
severe atomic disorder that no matching of lattice planes occurs across the
boundary, i.e., no continuity oflattice planes is maintained across the interface.
This eliminates coherency strains but the energy associated with the boundary
increases because of severe atomic disorder. A semi-coherent interface is one
that does not have a very large lattice mismatch between the phases, and the
small mismatch is accommodated by the introduction of dislocations at the
interface.
Crystallographically, most of the interfaces that one encounters in fiber,

whisker, or particle reinforced composites are incoherent. There are, however,
some exceptions to this. For example, in a nickel aluminide containing alumina
particles made by XD™ process (see Chap. 4), Wang and Arsenault (1991)
observed that the interface was semi-coherent, see Fig. 5.2. In such an interface,
the lattice mismatch between the matrix (NiAl) and the reinforcement (AI 20 3)
is accommodated by dislocations as indicated by arrows in Fig. 5.2.
5.3 WETTABILITY
Wettability can be defined as the ability of a liquid to spread on a solid surface.

Consider the general case of a liquid drop resting on a solid substrate as shwon
in Fig. 5.3. The three forces are the three specific energies or surface tensions
(i.e., energy per unit area): Ysv' the energy of the solid/vapor interface; Ys~' the
energy of the liquid/solid interface; and Y~v' the energy of the liquid/vapor
interface. When we put a liquid drop on a solid substrate, we replace a portion
of the solid/vapor interface by a liquid/solid and a liquid/vapor interface.
Fig. 5.2 A semi-coherent interface in an NiAIJAlP3 composite made by the XD™ process. (Courtesy ofR.I. Aresenault).
144 INTERFACE
Ysv
YIv
Fig. 5.3 Schematic of partial wetting.
Thermodynamically, spreading of the liquid will occur only ifthis results in a

decrease in the free energy of the system, i.e.,
Ys! + Y!v < Ysv
From the equilibrium of forces in the horizontal direction in Fig. 5.3, we have;
Ys! + Ylv cose = Ysv
where e is the contact angle. Or,
(5.6 )
For e = 0°, we have perfect wetting, while for e = 180°, we have no wetting.
For 0 °< e < 180 0 , there will be partial wetting. The contact angle for a given
system can vary with temperature, hold time, presence of any adsorbed gases,
145 INTERFACE
roughness ofthe substrate, etc.
It is important to realize that wettability and bonding are not synonymous terms.
Wettability describes the extent of intimate contact between a liquid and a
solid. It does not necessarily mean a strong bond at the interface. One can have
excellent wettability and a van der Waals type bonding, i.e., a weak or low
energy bond. A low contact angle, meaning good wettability, is a necessary but
not sufficient condition for strong bonding.
5.4 TYPES OF BONDING
There can be two main types of bonding at an interface:
1. Mechanical bonding
2. Chemical bonding
A brief description of these follows.
5.4.1 Mechanical Bonding
Mechanical keying between two surfaces can lead to bonding between them
and can be quite important in ceramic matrix composites. Figure 5.4a shows an
ideal, smooth interface while Fig. 5.4b shows a real interface which is rough,
which will allow for interlocking. Imagine, for example, a situation where the
matrix in a composite radially shrinks more than the fiber on cooling from a
high temperature. This would lead to a gripping of the fiber by the matrix even
in the absence of any chemical bonding, Fig. 5.5. The matrix penetrating the
crevices on the fiber surface, by liquid or viscous flow or high temperature
diffusion, can also lead to some mechanical bonding. In Fig. 5.5, we have a
radial gripping stress, 0"" which can be related to the interfacial shear strength,
'ti> as
where ~ is the coefficient of friction, generally, between 0.1-0.6. In general,

mechanical bond is a low energy bond vis a vis a chemical bond.
5.4.2 Chemical Bonding
In chemical bonding atomic or molecular transport, by diffusional processes,

146 INTERFACE
Ideal Real
fiber fiber
.~: ~~y
;; matrix
::x.
(a) (b)
Fig. 5.4 (a) An ideal, smooth interface; (b) a real, rough interface.
T·1
~--ar
Matrix Fiber
Fig. 5.5 A frictional or mechanical interface with a transverse gripping force (Le., in
the radial) direction.
147 INTERFACE
is involved. Solid solution and/or chemical compound formation may occur at

the interface, resulting in a reinforcement/matrix interfacial reaction zone with
a certain thickness. For diffusion controlled growth in an infinite diffusion
couple with planar interface, we have the following important relationship:
where x is the thickness of the reaction zone, D is the diffusivity, and t is the
time. The diffusivity, D, depends on the temperature in an exponential manner
D = A exp( - i\Q/kT)
where A is a pre-exponential constant, i\Q is the activation energy for the rate
controlling process, k is the Boltzmann's constant, and T is the temperature in
kelvin.
For a composite containing cylindrical fibers of a small diameter, the diffusion

distance is very small. However, to a first approximation, we can write
where K is a pseudo-diffusivity constant and has the dimensions of diffusivity

(m2 S-I). The reader should bear in mind that this approximate relationship can
be expected to work for composites where the reaction thickness is small
compared to the interfiber spacing. Under these conditions, one can use an
Arrhenius type relationship, K = A exp( - i\QlkT), where A is a pre-exponential
constant. A plot ofln K vs. liT can then be used to obtain the activation energy,
i\Q, for a fiber/matrix reaction in a given temperature range.
5.5 ROLE OF INTERFACE IN CMCs
The fiber/matrix interface affects the behavior of composites. Specifically, in

CMCs, interfacial bonding affects the fracture behavior of the composite. A
strong interfacial bond will allow an oncoming crack go unimpeded through the
interface and the composite will fail in a brittle manner. The interaction of a
crack in the matrix with a weak interfacial bond, on the other hand, is likely to
lead to debonding at the interface, followed by crack deflection, crack bridging,
fiber fracture, and finally fiber pullout. All these additional energy absorbing
phenomena lead to an enhanced fracture toughness and a non-catastrophic
148 INTERFACE
failure mode.
Ifwe examine the key issues involved in developing ceramic matrix composites
that show damage-tolerant or enhanced toughness characteristics, it soon
becomes clear that by far the most important area in this regard is the
reinforcement/matrix interface. In particular, fiber/matrix deb onding, crack
deflection and fiber pullout can contribute to toughness enhancement. The
sequence of failure of a CMC as a function of the interfacial bond strength is
depicted in Fig. 5.6. If the interfacial bond between the fiber and ceramic matrix
is very strong, then a crack starting at any point in the composite will rapidly
run through the composite cross section and result in a planar fracture, Fig.
5.6a. In other words, the presence of a strong fiber/matrix bond does not allow
any extra expenditure of energy in the fracture process, i.e., the fracture process
in a CMC having a strongly bonded reinforcement/matrix interface will be a
low energy fracture process, not unlike that encountered in monolithic
ceramics. In fact, ifthe modulus of the fiber is not very high compared to that
of the matrix, then in a strongly bonded CMC, virtually, no gain in toughness
is obtained. However, if the interface is weak, then it can act as a crack arrestor,
by deflecting the propagating crack in the matrix in a direction perpendicular
to its original direction, Fig. 5.6b.
r r
Fibe
"'.... ~
>
)
;:;;=
, ,
/~ ,
Matrix
~
Strong
~
Weak
interfac interface
(a) (b)
Fig. 5.6 Failure of a CMC as a function of interfacial bond strength: (a) strong
interfacial bond; (b) weak interfacial bond.
149 INTERFACE
The model of crack deflection or the formation of secondary crack at a weak

interface was first proposed by Cook and Gordon (1964). Let us consider a
fiber/matrix interface, perpendicular to the main advancing crack. Cook and
Gordon estimated the strength of the interface necessary to cause a diversion
of the crack from its original direction when both, fiber and matrix, have the
same elastic constants. At the tip of any crack, a triaxial state of stress (plane
strain) or a biaxial stress (plane stress) is present. Figure 5.7 shows
schematically the stress distribution at a crack tip. The main applied stress
component, cry, has a very high value at the crack tip, and decreases sharply
with distance from the crack tip. The stress component acting normal to the
interface, crx , is zero at the crack tip; it rises to a maximum value at a small
distance from the crack tip and then falls off in a manner similar to cry. Now, it
is easy to visualize that if the interface tensile strength is less than the maximum
value of crx , then the interface will fail in front of the crack tip. According to the
estimates of Cook and Gordon, an interfacial strength of 1/5 or less than that of
the main stress component, cry, will cause the opening of the interface in front
of the crack tip. One can also analyze the interaction between a crack and an
interface in terms of a fracture energy parameter instead of strength (He and
Hutchinson. 1989). One makes a chart of G/G r vs. cr, where Gj is the mixed-
mode interfacial fracture energy of the interface, and Gr is the mode I fracture
energy of the fiber, and cr is the measure of elastic anisotropy. We discuss it
detail in chapter 9. Suffice to point out here that such a plot shows the
conditions under which the crack will deflect along the interface or propagate
through the interface into the fiber. For the special case of zero elastic
mismatch, i.e., for cr = 0, the model predicts debonding at the fiber/matrix
interface for G/Gr less than about 0.25. Conversely, for G/Gf greater than 0.25,
the crack will propagate across the fiber. More sophisticated analyses of
interface mechanics have been proposed and will be discussed in Chapter 9.
Frequently, one would like to avoid completely or reduce the extent of a

chemical reaction at the reinforcement/matrix interface. A common way of
accomplishing this is to deposit a thin coating on the reinforcement prior to
incorporating it into a matrix to form a composite. The coating should be
thermodynamically stable with one or both the components of the composite.
In CMCs, one resorts to this type of interface engineering to change the
character of interfacial bonding from chemical to mechanical (Maheshwari et
aI., 1989). The objective is to control the characteristics of the interface.
Mechanical characteristics of the interface such as strength and toughness
control the overall performance of a composite. We describe some of the
techniques to determine these characteristics of the interface in Sec. 5.6.
150 INTERFACE
5.5.1 Thermal Stresses
Thermal stresses arise in composite materials because of the mismatch between

the coefficient of expansion of the components making up the composite.
Mismatch in the elastic constants of the components exacerbates the problem.
It is easy to see that in fiber reinforced composites, if the matrix thermal
expansion coefficient is higher than that of the fiber then on cooling from high
temperature radial compression will result, i.e., the matrix will grip the fiber.
Such a radial gripping of the fiber by the matrix will increase the strength of the
interface, thus, increasing the tendency toward brittleness in the composite.
This is because phenomena such as fiber debonding and pullout, required for
enhanced toughness, become difficult to operate if the fiber and matrix are
strongly bonded. Just to give an example, significant radial clamping stresses
on silicon carbide whisker can be expected in an alumina matrix because of the
y Interface
)-x
y
~----------------------.x
Distance from notch
Fig. 5.7 Schematic of stress distribution at a crack tip (after Cook and Gordon,
1964).
151 INTERFACE
thermal stresses due to the thermal mismatch between SiC and alumina
(Angelini et aI., 1987). We treat the subject of thermal stresses in composites
in chapter 8.
5.6 MEASUREMENT OF INTERFACE BOND STRENGTH
A quantitative measure of the strength of the interfacial bond between matrix

and reinforcement is of great importance. In this regard, many methods have
been devised. We describe below some of the important ones.
5.6.1 Curved Neck Specimen Test
This technique (Broutman, 1969; Chamis, 1974) is commonly used with

polymer matrix fiber composite. A special mold is used to prepare a curved
neck specimen of the composite containing a single fiber along its central axis,
Fig. 5.8. The specimen is compressed and the fiber/matrix debonding is
observed visually. The curved neck shape of the specimen enhances the
transverse tensile stress at the fiber/matrix interface. The transverse tensile
stress leading to interface debonding results from the fact that the matrix and
the fiber have different Poisson's ratios. Different Poisson's ratios of the fiber
and the matrix mean that if we apply a compressive load tothe composite,
different amounts of expansion in the fiber and matrix will occur in the
transverse direction. If the matrix Poisson ratio, vrn , is greater than that of the
fiber, vr, then on compression, there will result a transverse tensile stress at the
center ofthe neck and perpendicular to the interface whose magnitude is given
by (Broutman, 1969)
(5.7)
where a is the net section compressive stress (i.e. load/minimum area), E is the
Young's modulus, v is the Poisson's ratio, and the subscripts f and m denote
fiber and matrix, respectively. One can measure the net section stress, a,
corresponding to interface debonding and compute the interfacial tensile
strength from the above expression. There are some important points that
should be taken into account before using this test. One needs a special mold
to prepare the specimen and a very precise alignment of the fiber along the axis
152 INTERFACE
Fig. 5.8 A curved neck specimen of a composite containing a single fiber along its
central axis.
is a must. Finally, a visual examination is required to determine the point of

interfacial debonding. This limits the technique to transparent matrix materials
such as glass. Acoustic emission detection technique may be used to avoid
visual examination.
5.6.2 Bend Tests
Bend tests are easy to perform but they do not give a true measure of the
interfacial strength. We describe below some variants bend tests.
(aJ Transverse Bend Test
A three-point bend test configuration with fibers aligned perpendicular to the

specimen length is called a transverse bend test. There are two possible
arrangements of fibers as shown in Figs. 5.9 a and b. Under either one of these
153 INTERFACE
P/2 P/2
j-,
)
(a)
P/2 Pf2
(b)
Fig. '5.9 Transverse three-point bend tests: (a) fibers aligned perpendicular to the
loading direction; (b) fibers aligned parallel to the loading direction. Qdenotes the fiber
direction. In both cases, the fiber/matrix interface in the outermost layer will be put in
tension.
configurations, the fracture will occur on the outermost surface of the specimen
which is under a tensile stress and thus will put the fiber/matrix interface under
tension, which will be a measure of tensile of the fiber/matrix interface. The
transverse strength is given by
154 INTERFACE
3PS
cr =-- (5.8)
where P is the applied load, S is the load span, b is the specimen width, and h
is the specimen height.
(b) Longitudinal Bend Test or Short Beam Shear Test
This test is also known as the InterLaminar Shear Strength (ILSS) test. In this
test, the fibers are aligned parallel to the length ofthe three-point bend bar, Fig.
5.10. In such a test, the maximum shear stress, 't, occurs at the midplane and is
given by
3P
't = - - (5.9)
4bh
The maximum tensile stress occurs at the outermost surface and is given by Eq.
5.8. Dividing Eq. 5.9 by Eq. 5.10, we get
't h
(5.10)
cr 2S
Equation 5.10 says that if we make the load span, S, very small, we can
maximize the shear stress, 't, such that the specimen fails under shear with a
crack running along the midplane.
The reader should bear in mind that the interpretation of this test is not
straightforward. Clearly, the test becomes invalid if the fibers fail in tension
before shear induced failure occurs. The test will also be invalid if shear and
tensile failure occur simultaneously. It is advisable to make an examination of
the fracture surface after the test and make sure that the crack is along the
interface and not through the matrix.
(c) Iosipescu shear test
This test was originally developed for use with isotropic materials (Iosipescu,
1967). It is a special type of bend test that can be sued to measure interfacial
155 INTERFACE
....-- - - - S - - - - - . I
PI2 P/2
~,
Fig. 5.10 ILSS (InterLaminar Shear Strength) test. 1 denotes the fiber direction.
Fig 5.11 The Iosipescu shear test specimen

156 INTERFACE
shear strength parallel and perpendicular to the fiber directions. In this test a
double-edged notched specimen is subjected to two opposing force couples.
This is a special type of four-point bend test in which the rollers are offset, as
shown in Fig. 5.11., to accentuate the shear deformation. A state of almost pure
and constant shear stress is obtained across the section between the notches by
selecting a proper notch angle and notch depth (90 0 and 22% of full width).
The average shear stress is given by
P
't =- (5.11)
bh
where P is the load, b is the breadth and h is the height of specimen. The most
attractive point of this test that a large region of uniform shear is obtained vis
a vis other tests. However, there can be a substantial stress concentration near
the notch tip in orthotropic materials (not so in isotropic materials) such as fiber
reinforced composites. The stress concentration is a function of the fiber
volume fraction and the fiber orientation.
5.6.3 Fiber Pullout and Pushout Tests
Single fiber pullout and pushout tests have been devised to measure interfacial
characteristics. Such tests give load vs. displacement curves, with a peak load
corresponding to fiber/matrix debonding and a frictional load corresponding to
the fiber pullout from the matrix. The mechanics and interpretation of these
tests is rather involved and knowledge of the underlying assumptions is
important in order to get a useful information from such tests. Some of these
aspects related to interfaces are discussed in Chapter 9. The main simplification
that is frequently made in the analysis of such tests involves averaging the load
values over the entire interfacial surface area to get the interface debond
strength and/or frictional strength. Analytical and finite element analyses show
that the shear stress is a maximum close to the fiber end and falls rapidly within
a distance of a few fiber diameters. Thus, one would expect the interface
debonding to start near the fiber end and progressively propagate along the
embedded length. We describe the salient features ofthese tests.
(aJ Single Fiber Pullout Tests
Single fiber pullout tests are frequently used in PMCs and to a lesser extent in
MMCs. These tests can provide useful information about the interface strength
157 INTERFACE
in model composite systems. They are not very helpful in the case of
commercially available composites. One must also carefully avoid any fiber
misalignment and introduction of bending moments. The mechanics of the
single fiber pullout test is rather complicated (Cham is, 1974; Penn and Lee,
1989; Marshall et aI., 1992; Kerans and Parthasarthy, 1991). In all the variants
of such test, the fiber is pulled out of the matrix in a tensile testing machine and
a load vs. displacement record is obtained.
Figure 5.12a shows the experimental setup for such a test. A portion of fiber,
length Q, is embedded in a matrix and a pulling tensile force is applied as
shown. If one measures the stress required to pull the fiber out of the matrix as
a function of the embedded fiber length, the plot shown in Fig. 5.12b is
obtained. The stress required to pull the fiber out without breaking it increases
linearly with the embedded fiber length, up to a critical length, Qc ' For an
embedded fiber length, Q~Qc, the fiber will fracture under the action of the
tensile stress, a, acting on the fiber. The tensile stress, a, on the fiber results in
a shear stress, 't, at the fiber/matrix interface. A simple force balance along the
fiber length gives
For e < ec, the fiber is pulled out and the interfacial shear strength is given by
ar
't = - (5.12)
2e
One measures the load required to debond as a function of the embedded fiber
length. Then, we can write
P = 2nre't (5.13)
and the interfacial shear strength, 't, can be calculated from the slope of the P
vs. Q plot. For e > ec, fiber failure rather than pullout occurs. The interfacial
shear strength is a function of the coefficient of friction, Jl, and any normal
compressive stress at the interface, i.e., 'tj = Jla... A common source of radial
compressive stress is the shrinkage of the matrix during cooling from the
processing temperature. The most doubtful assumption in this analysis is that
158 INTERFACE
Matrix
Pulling force
Fiber
(a)
I
Fiber --"+01.. 1----- Fiber fracture
pullout 1
Embedded length, 1 -
(b)
Fig. 5.12 (a) Experimental setup for fiber pullout test; (b) stress required to pull fiber
out vs. embedded fiber length.
the shear stress acting along the fiber/matrix interface is a constant.
The fabrication of the single fiber pullout test sample is often the most difficult
part; it entails embedding a part of the single fiber in the matrix. A variation of
this simple method is to embed both ends of the single fiber in the matrix
material, leaving the center region of the fiber uncovered.
The peak load corresponds to the initial debonding of the interface. This is
followed by frictional sliding at the interface, and finally by the fiber pullout
from the matrix, during which a steady decrease in the load with displacement
is observed. The steady decrease in the load is attributed to the decreasing area
of the interface as the fiber is pulled out. Thus, the test simulates the fiber
159 INTERFACE
pullout that may occur in the actual composite, and more importantly, provides
the bond strength and frictional stress values.
The effect of different Poisson contractions of fiber and matrix can lead to a
radial tensile stress at the interface. The radial tensile stress will no doubt aid
the fiber/matrix debonding process. The effect of Poisson contraction together
with the problem that the imposed shear stress is not constant along the
interface complicates the analysis of fiber pullout test. Fiber pullout tests have
been used to determine the interfacial shear strength in a variety of fiber
reinforced ceramic or glass matrix composites.
(b) Pushout or Indentation Tests
Interfacial frictional sliding is an important parameter, especially in fiber

reinforced ceramic matrix composites. Takaku and Arridge (1973), among
others (Gao et aI., 1988; Shetty, 1988; Bright et aI., 1991), have analyzed the
interfacial sliding friction phenomenon using a shear-lag approach. We discuss
the topic of shear-lag analysis in some detail in chapter 7. Many researchers
(For example, Marshall, 1984; Doerner and Nix, 1986; Ferber et aI., 1993;
Mandell et aI., 1986; Mandell et aI., 1987; Marshall and Oliver, 1987; Cranmer,
1991; Chawla et aI., 2000) have used the technique of pressing an indenter on
a fiber cross-section in a composite to measure the interfacial bond strength in
a fiber/reinforced composite. An instrumented indentation test, the apparatus
for which is sometimes called a nanoindenter, has been developed that allows
extremely small forces and displacements to be measured. A nanoindenter is
essentially a computer-controlled depth-sensing indentation system. Indentation
instruments have been in use for hardness measurement for more than a
century, but "depth sensing" instruments having high resolution became
available in the 1980s (Doerner and Nix, 1986; Weihs and Nix, 1991).
Consequently, very small volumes of a material can be studied and a very local
characterization of microstructural variations is possible by mechanical means.
Hence, frequently the name mechanical properties microprobe is given to such
an instrument (cf electron microprobe). Both pointed and rounded indenters
can be used to displace a fiber aligned perpendicular to the composite surface.
By measuring the applied force and the displacement, interfacial stress can be
obtained. One generally loads several fibers in a polished cross section of
composite system. Most commercially available nanoindenter instruments are
capable of accurately applying very small loads (mN) via a Berkovich
pyramidal diamond indenter having the same depth-area ratio as a Vickers
diamond tip indenter. A nanoindenter records the total penetration of an
160 INTERFACE
indenter into the sample. The schematic of the indenting mechanism of one
such instrument is shown in Fig. 5.13a while the indentation system and the
corresponding indenter movement and corresponding voltage output are shown
in Fig. 5.13b. The indenter is a triangular pyramid and its position is determined
(a)
....__ Porce Coil
Load Voltage
Moving Plate
Displacement Voltage
(b)
Fig. 5.13 (a) Schematic ofa nanoindentation machine; (b) indentation system and the
corresponding indent movement and voltage.
161 INTERFACE
by a capacitance displacement gage. The capacitance gage can detect

displacement changes less than one nm while the applied force can be detected
to less that 1 IlN. The indenter can be moved toward the sample or away from
the sample by means of a magnetic coil assembly.
In the pushout test, one pushes a fiber out, by means of an indenter, in a CMC
sample with the fibers aligned perpendicular to the viewing surface. Such a
fiber pushout test can give the frictional shear stress, 't, acting at the
fiber/matrix interface. The pushout test uses a thin specimen (1- 3 mm). In the
pushout test, a three-region curve is obtained. An example of such curve
obtained from a fiber pushout test done on a Nicalon fiber reinforced glass-
ceramic matrix composite is shown in Fig. 5.14. (Chawla et aI., 2001). In the
first region, the indenter is in the contact with the fiber and the fiber sliding
length ~ is less then the specimen thickness t. This is followed by a horizontal
region in which fiber sliding length is greater than or equal to the sample
thickness. In the third region, the indenter comes in contact with the matrix. In
the first region, we can determine the interfacial shear stress, 't from Eq. (5.18).
The specimen thickness should be much greater that the fiber diameter for this
relationship to be valid. In the horizontal region, the interfacial shear stress is
given by:
't = F/21t r t
where t is the specimen thickness. In the third region of Fig. 5.14, the value of
the interfacial shear stress cannot be determined because the indenter comes in
contact with the matrix.
There are certain assumptions implicit in such an indentation test
1. Any elastic depression of the matrix adjacent to the fiber is negligible.

2. There are no surface stress concentrations.
3. There is no change in the fiber diameter due to the Poisson expansion during
compression of the fiber.
4. There are no residual stresses.
These assumptions are not unreasonable if the specimen thickness is large

compared to the fiber diameter.
It has been observed that the fiber will slide along the interface over a distance
162 INTERFACE
20
15
'.,.,"
n.
~ 10
~
Indenter hits
~ matrix
en
5
+-
0
-1 0 2 3 4 5 6
Displacement (u ml
Fig. 5.14 An example of a force vs. displacement curve obtainable from a pushout test
done on a Nicalon fiber reinforced glass-ceramic matrix composite. (Chawla et aI.,
2001).
that is dependent on the load applied by the indenter. In this model, the load on
the indenter is assumed to be balanced by the frictional stress at the interface,
and the effect of radial expansion during indentation is neglected. The fiber is
elastically compressed by the indenter load over the debonded length, which is
assumed to be dependent on the interfacial friction.
Indenter technique have advantages as well as disadvantages with regard to the

accuracy of the values and the ease of performance of the test. Different
indenter geometries are available. Figures 5.15 shows some of these. The thin-
slice technique, having the least complicated geometry, reduces the
approximations in the analysis and errors in the experiment. The preparation of
the specimen for the test is, however, difficult and often impossible for some
fiber/matrix combinations.
Some other variations on these techniques have been proposed. For example
Wang et al. (Wang et aI., 1992) used a dimpler, commonly used for specimen
preparation for transmission electron microscopy, to grind and polish a CMC
specimen to the desired thickness. Their reasoning for adopting this variation
was that in the conventional thin slice fiber pushout configuration, bending may
occur under indentation loading. Such bending of the specimen will result in an
inaccurate measurement of the interfacial strength. Their specimen preparation
technique results in thin, damage-free areas suitable for fiber pushout tests. A
nanoindenter instrument can then be used on such a sample to obtain the
163 INTERFACE
Load
Fig. 5.15 Various indenter/specimen geometries. m andfdenote matrix and fiber,

respectively
interface shear strength. Another technique, called thin slice compression test,
involves taking a slice of an aligned fiber composite cut normally to the fibers,
put between two plates and subject it to a compression test. The top plate is of
a soft metal such as aluminum while the bottom plate is made of rigid materials,
for example, silicon nitride. The slice of composite is loaded to a desired peak
stress and then unloaded. At the top surface the elastic mismatch is maximum
and as the interface deb onding and sliding occurs, the top plate retains a
negative image of the fiber protrusion (Shafry et aI., 1989; Kagawa and Honda,
1991).
A comparison of different fiber pushout techniques is presented in Table 5.1

(Parthasarthy et aI., 1989).
164 INTERFACE
Table 5.1 Various techniques of measuring interfacial shear strength (after

Parthasarathy et aI., 1989)
TechnifJ.ue Advantage Limitation

Matrix crack Real composite, Inferred "strength" not well
spacing directly related to defmed
composite behavior
2 Indentation Real composite Small diameter, coreless fibers
3 Micro- Real composite Identification of the debonding
debonding event
4 Fiber pushout in Real composite Specimen preparation
a thin slice of
composite
5 Single fiber Simulation Tedious specimen preparation
pullout
5.6.4 Interfacial Shear Strength via Matrix Crack Spacing
Consider a unidirectional continuous fiber-reinforced composite. If the failure

strain of the matrix is less than that of the fiber, then on subjecting such a
composite to tensile force, the matrix will develop multiple cracks with spacing
between X and 2X, Fig. 5.16. The distance X, according to the ACK model
(Aveston et aI., 1971), is given by
where cr'm is the stress in the matrix at which cracking begins, r is the fiber
radius, 't' is the shear stress at the interface, and Vr and V m are the volume
fractions of the fiber and the matrix, respectively. In terms of the composite
stress, cr e, the spacing X is given by
where Ee and Em are the elastic moduli of the composite and matrix,
respectively.
Ifwe subject a unidirectional, continuous fiber reinforced composite to a tensile

165 INTERFACE
Matrix crack
Fig. 5.16 Multiple cracking in matrix with crack spacing betweenXand 2X.
test and obtain uniformly spaced, multiple matrix cracking, then we can
estimate the interfacial fiber/matrix shear strength from these two equations.
In practice, interfacial shear strength obtained from matrix crack spacing
measurements may not always agree with the experimental values obtained by
direct measurements such as a fiber pushout tests. This may be, in part, due to
any variation in various parameters such as processing conditions, interfacial
roughness, etc.
One may, however, question the validity of these equations because the ACK
model assumes no bonding at the interface. This is particularly true in
composite systems that show extensive chemical bonding between fiber and
matrix or if there is large degree of radial gripping of the fiber by the matrix
because of residual thermal stress or roughness induced clamping by the matrix.
Later, Aveston and Kelly (1973) modified the ACK model to include partially
bonded and fully bonded interfaces. The principle difference between the AK
model and the ACK model is that the shear stress along the interface is uniform
in the latter. The two models are in reasonable agreement except when
fiber/matrix bond shear strength is much greater than the matrix failure stress.
5.6.5 Laser Spallation Technique
Gupta et al. (1990, 1992) devised a laser spallation technique to determine the
166 INTERFACE
tensile strength of a planar interface between a coating (thickness> 0.5 11m) and
a substrate. Figure 5.17 shows their experimental set up. A collimated laser
pulse impinges on a thin film sandwiched between the substrate and a confining
plate. This plate is made of fused quartz which is transparent to Nd:YAG laser
(wavelength = 1.06 11m). An aluminum film works as the laser-absorbing film.
Absorption of the laser energy in the gold or aluminum film causes a sudden
expansion of the film. The expansion of the constrained film produces a
compressive shock wave in the substrate and moves toward the
coating/substrate interface. When the compression pulse hits the interface, part
of it is transmitted into the coating. This compressive pulse is reflected into a
tensile pulse at the free surface of the coating. If this tensile pulse is of a
sufficient magnitude, it will remove the coating from the substrate. Gupta et al.
later used a laser Doppler displacement interferometer to record the time rate
of change displacement of the coating free surface as the compressive pulse is
reflected. By means of a sophisticated Tektronix SeD 1000 digitizer, they could
obtain a time resolution of about 0.5 ns for recording displacement fringes. This
information is then related to the stress pulse history at the interface. A direct
recording of the stress pulse makes their technique useful for interface systems
involving ductile components. We describe some results obtained by this
technique in chapter 9.
Substrate:
Energy Si crystal, pilCh-55
absorbing ribbon, etc. SiC
film (Au Or AI)
Confining fused Pressure

quail, plate wave
Pulsing laser
Surface
Wavelength = 1.06 J1ffi acceleral ions
Maximum energy = 850 mJ measured
onnal pul e duration = 8 n
Fig. 5.17 Schematic of the laser spallation technique (after Gupta et aI., 1990).
167 INTERFACE
REFERENCES
Angelini, P., W. Mader, and P.F. Becher (1987) in Mat. Res. Soc. Symp., vol. 78,
Materials Research Society, Warrendale, PA, p. 24l.
Aveston, J., G.A. Cooper, and A. Kelly (1971) in The Properties ofFibre Composites,
IPC Science and Technology Press, Survey, England, 15.
Aveston, J. and A. Kelly (1973) J Mater. Sci., 8, 129.
Bright, I.D., S. Danchaivijit, and D.K. Shetty (1991) J Amer. Cer. Soc., 74, 115.
Broutman, L.J. (1969) in Interfaces in Composites, ASTM STP #452,34.
Chamis, C.C. (1974) in Composite Materials, Vol. 6, Academic Press, New York, 32.
Chawla, N., K.K. Chawla, M. Koopman, B. Patel, C. Coffm, and J.I. Eldridge (2001)
Composite Sci. & Tech., 61, 1923.
Cook, I. and I.E. Gordon (1964) Proc. Roy. Soc. London, A228, 508.
Cranmer, D.C. (1991) in Ceramic and Metal Matrix Composites, Pergamon Press, New
York, p. 157.
Doerner, M.F. and W.D. Nix (1986) J Mater. Res., 1, 60l.
Ferber, M.K., A.A. Wereszczsak, L. Riester, R.A. Lowden, and K.K. Chawla, (1993)
Ceram. Eng. Sci. Proc., 13,168.
Gao, Y.C., Y.W. Mai, and B. Cotterell (1988) SAMPE, 39, 550.
Gupta, V., A.S. Argon, J.A. Cornie, and D.M. Parks (1990) Mater. Sci. Eng., A126,
105.
Gupta, V., A.S. Argon, J.A. Cornie, D.M. Parks (1992) J Mech. Phys. Solids, 4, 14l.
He, M.Y. and Hutchinson, J.W. (1989) J App. Mech., 56,270.
Iosipescu N. (1967) J Mater., 2, 537.
Kagawa, Y. and K. Honda (1991) Ceram. Eng. Sci. Proc., 12, 1127.
Kerans, R.J. and T.A. Parthasarathy (1991) J Amer. Cer. Soc., 74, 1585.
Maheshwari, A., K.K. Chawla, T.A. Michalske (1989) Mater. Sci. & Eng., AI07, 267.
Mandell, J.F., D.H. Grande, T.H. Tsiang, and F.I. McGarry (1986) in Composite
Materials: Testing & Design, ASTM STP 327, ASTM, Philadelphia, 87.
Mandell, J.F., K.C.C. Hong, and D.H. Grande (1987) Ceram. Eng. Sci. Proc. 8, 937.
Marshall, D.B. (1984) J Amer. Ceram. Soc., 67, C259.
Marshall, D.B. and W.C. Oliver (1987) J Amer. Ceram. Soc., 70, 542.
Marshall, D.B., M.C. Shaw, and W.L. Morris (1992) Acta Met., 40, 443.
Parthasarathy, T.A, N.J. Pagano, and R.I. Kerans (1989) Ceram. Eng. Sci. Proc., 10,
872.
Penn, L.S. and S.M. Lee (1989) J Compo Tech. & Res., 11,23.
Shafiy, N., D.G. Brandon, and M. Ferasaki (1989) Euro-Ceramics, 3,.453
Shetty, D.K. (1988)J Amer. Cer. Soc., 71, C-107.
Takaku, A. and A.G.C. Arridge (1973) J Phys. D: Appl. Phys., 6, 2038.
Wang, L. and R.J. Arsenault (1991) Met. Trans. A, 22A, 3013.
Wang, S.-W., A. Khan, R. Sands, and A.K. Vasudevan (1992) J of Materials Sci.
Letters, 11, 7739.
Weihs, T.P. and W.D. Nix (1991)J A mer. Ceram. Soc., 74, 524.
Cliapter6
MICROMECHANICS: ELASTIC, THERMAL,

AND PHYSICAL PROPERTIES
In this chapter, we examine the elastic and physical properties of ceramic

matrix composites. In particular, we give a micromechanical description ofthe
different elastic constants, thermal expansion coefficients, thermal conductivity,
density, etc. in terms of the same constants or of the individual components and
the geometric arrangement of the components.
6.1 ELASTIC CONSTANTS OF A FIBER REINFORCED COMPOSITE
Hooke's law relating a second rank stress tensor, O"ij' and a second rank strain
tensor, ekb can be written as
(6.1)
where C ijkl is a fourth-rank elastic stiffness tensor and the indices iJ,k take
values of 1,2, and 3. The stress and strain are second rank tensors; they have
32 = 9 components each. The stiffness tensor, Cijkl has 3 4 = 81 components. It
turns out that the stress, strain and the stiffness tensors are symmetric, i.e., O"ij
= O"jj, e kl = elk and Cjjkl = Cklij := Cjikl := Cijlk • These result, in the most general case,
in a reduction of the maximum number of independent elastic constants from
81 to 21. Equation (6.1) can be simplified and rewritten in a contracted notation
more appropriate for matrix operations as
(6.2)
In a similar manner we can write Hooke's law in terms of an elastic compliance

matrix, Sij' as
(6.3)
170 MICROMECHANICS: ELASTIC, THERMAL, AND PHYSICAL PROPERTIES
Both Cij and Sij are 6 x 6 symmetric matrices, i.e., Cij = C ji and Sij = Sji' Also, the
stiffness and the compliance matrices are inverse of each other, i.e.,
[C] [S] = [I]

where [I] is the identity matrix. The total number of independent constants
required to completely describe the elastic behavior of a material decreases with
increasing crystal symmetry elements present. Table 6.1 presents a summary.
For an isotropic material (e.g., a statistically homogeneous polycrystal), there

are only two independent elastic constants, while for cubic symmetry, one
needs three constants. The most general situation is that of the triclinic system
(no crystal symmetry elements present) which requires 21 independent elastic
constants.
We can write Hooke's law in an extended form for cubic and hexagonal
symmetry as follows:
Cubic
0'1 C II C I2 C I2 0 0 0 £1
0'2 C II C I2 0 0 0 £2
0'3 C II 0 0 0 £3
0'4 C44 0 0 £4
O's C44 0 £s
0'6 C44 £6
Hexagonal
0'1 C II C I2 C I3 0 0 0 £1
0'2 CII C I3 0 0 0 £2
0'3 C 33 0 0 0 £3
0'4 C44 0 0 £4
0'5 C44 0 £s
0'6 (C II -C I2 )/2 £6
Table 6.1 Independent elastic constants for crystals of different symmetry
Symmetry No. of Constants

Constants
Orthorhombic 9 CII' CIZ, C13' Czz,CZ3 ,C33 ,C44,C55 , C66

Trigonal 6 CII,CIZ,CI3,CI4,C33,C44
Tetragonal 6 CII,CIZ,C13,C33,C44,C66
Hexagonal 5 CII,CIZ,C13,C33,C44
Cubic 3 Cll,CIZ,C44
Isotropic 2 Cll,C IZ
For an isotropic material, the number of independent constants is reduced from

3 in a cubic system to 2 because of the following relationship:
In engineering usage, for an isotropic material, any two of the following four
engineering elastic constants will suffice; Young's modulus (E), Poisson's ratio
(v), shear modulus (G), and bulk modulus (K). This is so because of the
following relationships that exist among these four constants:
E = 3K(1 - 2v)
G = E/2(1 +v)
K = E/3(1 - 2v)
and v = (E/2G) - 1
Thus, one may measure all the four constants given above for an isotropic
material, but only two are independent.
Figure 6.1 shows cross sections of some fiber reinforced composites with fibers
arranged in a square (or rectangular), a hexagonal, or a random array. A
unidirectional fiber reinforced composite with fibers arranged in a hexagonal
or random manner, in the transverse section, is said to be transversely isotropic,
i.e., if we consider the fibers to be along direction 1, then there is no preferred
direction in the 2 - 3 plane. Such an arrangement of fibers gives us the same
number of independent elastic constants as the hexagonal crystal, i.e., one needs
five elastic constants to fully describe the elastic behavior of such a composite.
A square or rectangular arrangement of fibers in the transverse section, Fig. 6.1,
gives an orthotropic symmetry, i.e., three mutually perpendicular planes of
symmetry. Such an orthotropic symmetry requires nine independent elastic
--Ji-I:e I: ••
!r------.- i i •
•
•
I I
I I
. I eie
I I
I I
~b~
·0
(a)
ee
• e
••• e
•• e.• e (b)
•• e
• e
• e •
(c)
• •
Fig. 6.1 Various fiber arrays in a matrix: (a) rectangular (a*b) or square (a = b); (b)
hexagonal; (c) random. Hexagonal and random arrays are transversely isotropic.
constants of a composite, knowing the elastic constants of the individual

components.
6.1.1 Strength of Materials Approach
We can get some quick estimates of elastic constants of a composite by using

a strength of materials approach. Two simple cases are isostrain and isostress
models for Young's moduli of a unidirectionally aligned, fiber reinforced
composite. We also derive expressions for the principal shear modulus and
principal Poisson's ratio.
(a) Longitudinal Young's modulus
Ifwe apply the isostrain condition to a unidirectional, fiber reinforced omposite

loaded along the fiber direction, we get the longitudinal Young's modulus, Eel
or Ell' of the composite, Fig. 6.2. The isostrain condition (or action in parallel)
says that the strains in the fiber, matrix, and composite are identical, and so we
can write
.;.. x
Fig. 6.2 A unidirectional fiber reinforced composite loaded in tension parallel to the
fibers (isostrain or action in parallel).
(6.4)
where E is the strain, M is the change in length, eis the original length, and the
subscripts f, m, and cl indicate fiber, matrix, and composite in the longitudinal
direction, respectively.
For components behaving elastically, from Hooke's law we can write the
uniaxial stress acting on the fiber and the matrix as
where a is the stress, E is the Young's modulus, and the subscripts have the
meanings given above.
The applied load on the composite, Pc is partitioned between the fiber and the
matrix, i.e.,
Pc=Pr+Pm
or a el Ae = ar Ar + am Am
= Er Eel Ar + Em Eel Am
= (Er Ar + Em Am) Eel
ael = Eel Eel = (ErAIAe + Em Am/AJ Eel
From this we get the following expressions:
(6.5)
(6.6)
where V is the fiber volume fraction. The variation of the longitudinal Young's
modulus, Eel (or Ell) as a function of fiber volume fraction, Vr, as per Eq. (6.5)
is shown in Fig. 6.3.
(b) Transverse Young's modulus
The transverse modulus, Eet or E 22 , can be estimated by using an isostress

condition, i.e., the fiber, matrix, and composite experience the same stress (Fig.
6.4). Thus,
(6.7)
where the subscript ct denotes the composite in the transverse direction and the
other subscripts have the significance given earlier.
We can write for the total displacement, .M, of the composite in the thickness
direction
~te = ~tm + ~tr
400
350
300
~
o;l
250 Ed
c.:>
'"
;:l
"3
200
""
0
::;s 150
E"
100
50
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Vr
Fig. 6.3 Variation of longitudinal (Eel) and transverse modulus (Eet ) with fiber volume
fraction (Vr)'
where ~1m and ~4 are the displacements in the matrix and fiber, respectively.
Dividing throughout by te, the gage length, we get the strain in the transverse
direction
or
(6.8)
Using Hooke's Law, we can rewrite Eq. (6.8) as
oelEet = OetV mlEm + OetVIEf

or llEel = V mlEm + V IEf = llE22 (6.9)
The variation of the transverse Young's modulus, Eel (or E 22 ) as a function of

fiber volume fraction, Vb as per Eq. (6.9) is shown in Fig. 6.3.
(c) Poisson's ratio
Consider again a composite containing unidirectionally aligned fibers and

loaded in tension parallel to the fibers, Fig. 6.2. The composite will extend el
longitudinally (direction I) and contract e2 transversely (direction 2). The
contraction in direction 2 due to all the fibers will be -elVf Vf , where Vf is the
fiber volume fraction and Vf is the Poisson's ratio of the fiber. Similarly, the
contraction due to the matrix will be -eiVm vm • The total contraction of the
composite in direction 2 is then given by
Defining the principal Poisson's ratio of the composite as V I2 = -eie l , we get
(6.10)
(d) Shear modulus
In the case of longitudinal or principal shear modulus, both the fiber and matrix
are subjected to the same shear stress, Fig. 6.5. Shear strains in the matrix and
fiber are given by
Fig. 6.4 A unidirectional fiber reinforced composite loaded in tension perpendicular to

the fibers (isostress or action in stress).
r---- -----~----
L L_ _ _ "--------{
t
Fig. 6.5 A unidirectional fiber reinforced composite loaded in shear parallel to the
fibers.
'Ym = -r/Gm and 'Yr = -r/Gr

where -r is the shear stress and G is the shear modulus. The total shear
displacement in the composite, d, can be written as
d=yt (6.11)
where 'Y is the average shear of the composite and t is t9tal thickness of the
composite. We can write the total shear displacement in terms of the shear
displacements of the components as
or
(6.12)
From Eqs. (6.11) and (6.12) we get
(6.13)
Writing shear strains as shear stress divided by shear modulus in Eq. (6.13), we
get
or
(6.14)
where G I2 is the principal shear modulus of the composite.
6.1.2 Micromechanical Approach
Let us recapitulate the main differences between an anisotropic and an isotropic

body. An anisotropic body is one in which the properties depend on the
direction in which they are measured. In the most general case of anisotropy,
i.e., a body having no elements of symmetry, one needs 21 independent elastic
constants to completely define its elastic behavior. The triclinic system is such
a system. An isotropic body, on the other hand, has no preferential direction,
i.e., a given property has the same value in all directions in such a body. Such
a body needs only 2 independent elastic constants. As a general rule, any

symmetry elements present will reduce the number of independent elastic
constants, the isotropy condition being the simplest of all. In an isotropic
material, when a tensile stress is applied in the z direction, a tensile strain ez
results in that direction. In addition to this, because of the Poisson's ratio effect,
two equal compressive strains (ex=ey) result in the x and y directions. In an
anisotropic body, the two transverse strain components will not be equal, i.e.,
ex #ey • A composite containing uniaxially aligned fibers has a plane of symmetry
perpendicular to the fiber direction, i.e., material on one side of the plane will
be the mirror image of the material on the other side. Such a composite will
have 13 independent elastic constants. Generally, additional symmetry elements
are present, which simplifies situation. A square array of fibers, for example,
has symmetry planes parallel to the fibers as well as perpendicular to them.
Such a composite is an orthotropic material, i.e., it has three mutually
perpendicular planes of symmetry and has 9 independent elastic constants.
Hexagonal and random arrays of aligned fibers simplify the situation further.
Such fiber composites are transversely isotropic and have five independent
elastic constants.
In an anisotropic material, there are two Poisson's ratios. The principal or major
Poisson ratio (V I2 ) gives the transverse strain caused by an axially applied stress
while the minor Poisson ratio (V21 ) gives the axial strain caused by a
transversely applied stress. These two Poisson's ratios in an anisotropic material
are shown in Fig. 6.6. In an isotropic material, V 12 and V 21 are equal and we have
only one Poisson's ratio. In a transversely isotropic fibrous composite, V12 and
V 21 are not equal, but the two quantities are related: v 12 E22 = V 21 E II • The elastic
constants for a transversely isotropic composite in terms ofthe elastic constants
ofthe two components are summarized in Table 6.2 (Hash in and Rosen, 1964).
In this case, we have the 2-3 plane to be isotropic, i.e., the properties in
directions 2 and 3 are identical. We treat the matrix as an isotropic material and
treat the fiber as an anisotropic material. Thus, E and v are the two constants
required for the matrix while five constants (En, Ea , Gn , Ga , and Vf) are
required for the fiber. Oftentimes; composite structures are fabricated by
stacking thin sheets of unidirectional composites called plies in an appropriate
orientation sequence dictated by elasticity theory. It is of interest to know the
properties of a ply; for example, its elastic constants and strength. In particular,
it is useful to be able to predict the characteristics of a composite lamina if we
know the individual component characteristics and their geometrical
disposition. Later in the macromechanical analysis we treat a ply as a
homogeneous but thin orthotropic material. In the case of thin sheets or plies,
it is not unreasonable to ignore the elastic constants in the thickness direction,
leaving 4 independent elastic constants, viz., Ell> E 22 , V l2 and G 12 , i.e., one less
than the number for a thick but transversely isotropic material. The missing
constant is G 23 , the transverse shear modulus in the 2-3 plane normal to the fiber
axis.
We provide below a brief survey of the various micromechanical techniques

used for predicting the elastic constants. In the Self-Consistent Field methods,
one makes approximations of phase geometries. Commonly, the phase
geometry is represented by one single fiber embedded in a matrix cylinder. This
outer cylinder is embedded in an unbounded homogeneous material whose
properties are taken to be the average properties of the composite material. The
radii of the cylinders are dictated by the fiber volume fraction. A uniform load,
applied at infinity, introduces a uniform strain field in the fiber. Elastic
constants are then obtained from this strain field. The results obtained are
independent of fiber arrangements in the matrix and, in general, are reliable at
low fiber volume fractions (Vd, reasonable at intermediate Vf> and unreliable
at high Vf'
-----Ld---- -
V12
t
IT]
Fig. 6.6 The two Poisson's ratios in an anisotropic material: the principal one (v 12) gives
the transverse strain caused by an axially applied stress and the secondary one (v21 )
gives the axial strain caused by a transversely applied stress.
180 MICROMECHANlCS: ELASTIC, THERMAL, AND PHYSICAL PROPERTIES
Table 6.2 Elastic moduli and stress-strain relationships for a transversely isotropic
fibrous composite (after Rashin and Rosen, 1964).
Elastic moduli
1
1 Cll-E '2
= -(--)
2 K23
4K23 V2
<p=I+--
E
Stress-Strain Relationshi s
Exact methods deal with specific geometries, e.g., fibers arranged in a

hexagonal, square, or rectangular array in a matrix. The elasticity problem then
is solved with mathematical techniques involving infinite series, complex
variables, or numerical analysis. Energy methods involve variational calculus
and can be used to obtain upper and lower bounds on a property. The property
is determined exactly only if the upper and lower bounds coincide. Generally,
the upper and lower bounds are not close.
For particulate composites, Hashin (1962) proposed a composite sphere

assembly model wherein the composite is thought to be made up of units
consisting of a spherical particle and its surrounding matrix shell. In each
spherical unit, the volume fractions of the particle and matrix are the same, but
the spherical units can be of any size. Figure 6.7 shows such an assembly.
Fig. 6.7 A composite made up of an assemblage of spherical particles surrounded by

their respective matrix shells. The volume fractions of phases are maintained constant
in each unit; individual units can be of any size.
Hashin analyzed this model by variational methods and obtained a closed-form

solution for the bulk modulus and close upper and lower bounds for the
effective shear modulus. The bulk modulus, K, is given by
Kp
Km+iG (V K + Vmm
3mpp
K )
K=
VK+VK+iG
p m m p 3 m
where K, G, and V indicate the bulk modulus, shear modulus, and volume
fraction, respectively, and the subscripts p and m refer to the particle and the
matrix, respectively. Hashin and Shtrikman (1963) and Rosen (1973) have also
analyzed macroscopically isotropic, particulate composite with an arbitrary
internal phase geometry and only phase volume fractions being specified. The
bounds of Hashin and Shtrikman on the shear modulus, G and the bulk
modulus, K are given below:
(1- Vp )
Kp+--------~3-V----
(K -K
m p
+ 3K
p
+~Gp )
where ~ > ~ and Gp > Gm • For K,11 < Kp and Gp < Gm the inequalities will be
reversed. Treating the particulate composite as an isotropic material, we can
obtain the bounds on the Young's modulus of the composite, Ee , by using the
following relationship:
9KG
E =
(3K +G)
For 0.5 < E/E m < 3, the bounds are close enough to give us a value within
~10% of the true modulus.
In the case of longitudinal elastic modulus of a unidirectionally aligned fiber

composite, however, the bounds are close enough to give us a good indication
of the elastic constant. Hill (1964, 1965) derived bounds for the elastic
constants that are analogous to those due to Hashin and Rosen (1964) and
Rosen (1973). He put rigorous bounds on the longitudinal Young's modulus, E,
in terms of the bulk modulus in plane strain (~), Poisson's ratio (v), and shear
modulus (G) of the two phases without making any restrictions on fiber form
or packing geometry. The bulk modulus in plane strain, ~, is the modulus for
lateral dilation with zero longitudinal strain and is given by
~ = E/[2(1-2v)(1+V)]
The bounds on the longitudinal modulus, Ell, are
4VrV rn(vr -vrnYI(VI~m + V m/kpf + 1/Grn)

:;; Ell - ErVf - EmVm
:;; 4VrV m(vr-vrn)2/(VI~rn + Vm/kpf + lIGc) (6.15)
We can readily verify from Eq. (6.15) that the deviations from the rule of
mixtures for Ell are quite small. If we substitute some values for practical
composites such as carbon or silicon carbide fibers in an alumina matrix we
shall find that the deviations in Ell from the rule of mixtures are less than 2%.
Notice that the deviation from the rule-of-mixtures value comes in from the (vf
- vrn)2 factor. Also, for Vf = vrn , we have Ell given precisely by the rule of
mixtures.
For the Poisson's ratio of a unidirectionally aligned fiber composite, Hill (1965)
also showed that
and
(6.16)
If Vf < Vrn and E f » Em' then, Vl2 will be less than that predicted by the rule of
mixtures (= vrYr+vrnV rn). It is easy to see that the bounds on V12 are not as close
as the ones on Ell. This is because (ve - vrn) appears in the case ofvl2 (Eq. 6.16)
while (ve - vrn)2 appears in the case of Ell (Eq. 6.15). If(vf -vrn) is very small, the
bounds will be close enough to allow us to write
(6.17)
We can summarize the results of Hashin et al. as follows. For a transversely

isotropic composite with fibers along direction 1 and the 2-3 plane being the
transverse (isotropic) plane, the equations of five independent moduli (based
on the results of Hashin and Rosen, 1964 and Hill, 1965) are given below:
Plane-strain bulk modulus, k23 :
where ~ and kr are the plane-strain bulk moduli of the matrix and fiber
respectively. The plain-strain bulk modulus is defined as:
1
k = K+-G
3
In-plane shear modulus, G12 :
Longitudinal Young's modulus, Ell:
For most practical purposes the last term is negligible in the above expression.
Longitudinal Poisson's ratio, v J2 :
Transverse plane shear modulus, G23 :
In this case, the upper and lower bounds are not coincident.
Lower bound:
Upper bound:
where
Y+~m
a =--
y-l
~m = - - -
3 -4Vm
Finally, we provide in Table 6.3 a summary of the elastic constants for a

transversely isotropic, unidirectionally reinforced fiber composite (Chamis,
1983). Here we treat the fiber to be anisotropic and the matrix isotropic. Thus,
in Table 6.3, when computing the moduli in the transverse direction, we take
the relevant transverse modulus value of the fiber.
6.1.3 Halpin-Tsai-Kardos Equations
Halpin, Tsai, and Kardos (Halpin and Tsai, 1967; Kardos 1971; Halpin and
Kardos, 1976) devised an empirical approach to obtain some generalized
equations which readily give quite satisfactory results for unidirectional
composites compared to the more complicated micromechanical equations.
These equations contain adjustable fit parameters and work quite well with low
fiber volume fractions. They can also provide useful estimates of properties of
composites containing discontinuous fibers or whiskers oriented in the loading
direction. The adjustable parameters must be obtained from experimental data
or must conform to some analytical solution. One uses a single equation of the
form
(6.18)
Table 6.3 Bounds for transversely isotropic fiber reinforced composites (after Chamis,
1983).
Longitudinal modulus
Transverse modulus
Shear modulus
Shear modulus
Poisson's ratio
Poisson's ratio J!,

v = ~-1
23 20"
and
(6.19)
where p represents composite moduli, e.g., Ell, E22 , G 12 or G 23 ; Pr and Pm are the
corresponding matrix and fiber moduli, respectively; Vr is the fiber volume
fraction; and ~ is a measure of reinforcement which depends on boundary
conditions (fiber geometry, fiber distribution, and loading conditions). The term
~ is a fit parameter that is used to make Eq. 6.13 conform to the experimental
data. The function 1'\ in Eq. (6.19) is constructed in such a waytbatp=Pm when
V r = 0 and p = Pf when Vr = 1. Furthermore, the form of 1'\ is such that
and
These two extremes provide the bounds, not necessarily tight, on the composite
properties. Thus, values of ~ between 0 and 1 will give an expression for p
between these extremes. Some typical values of ~ are given in Table 6.4. For
example, for the transverse modulus, E22 , we can write from Eqs. (6.18) and
(6.19)
and
11 = (E/Em - 1)/(E/Em + ~)
Nielsen (1974) has modified the Halpin-Tsai-Kardos expressions to include the

maximum packing factor, Fmax.
where
11 = (P/Pm - 1)/(P/Pm + ~)
The maximum packing factor, Fmax is equal to 0.907 for hexagonal packing and
0.785 for square packing. A suggested value ofFmax for random packing is 0.82.
Table 6.4 Typical values of ~ for different moduli
Modulus (
En 2(@/d)
E22 0.5
0 12 1.0
- 0 21 0.5
K 0
6.2 COEFFICIENT OF THERMAL EXPANSION
Thermal energy is ever present and is responsible for the atomic or molecular
vibration about a mean position in any material. As the temperature of the
materials is increased, the amplitude of thermal energy induced vibrations is
increased and the interatomic or intermolecular spacing increases, i.e., an
expansion of the body occurs. Most materials show such an expansion with
increasing temperature.
In general, the thermal expansion of a material is greater in the liquid state than
in the crystalline state, with the transition occurring at the melting point. In the
case of a glassy material, such a transition occurs at what is called the glass
transition temperature. Over a certain range of temperature, one can relate the
temperature interval and thermal strain by a coefficient, called the coefficient
of thermal expansion. In the case of a linear strain, the linear thermal expansion
coefficient, a, is a second-rank symmetric tensor, and is related to the strain
tensor, E, by the following relationship:
(6.20)
where /). T is the temperature change. The thermal expansion coefficient, a,

generally does not have a constant value over a very large range of temperature.
Thus, we can define aij in a more general way by taking into account this
variation with temperature as follows
We define a volumetric coefficient of thermal expansion, p, involving the

volumetric strain; this is given by
(6.21)
where V is the volume and T is the temperature. For small strains, it can be
easily shown that
The volumetric expansion coefficient, p, is equal to the sum of the diagonal

terms of the strain tensor, i.e.,
(6.22)
or
(6.23)
As we said above, only over some specified range of temperature can the
coefficient of thermal expansion be treated as a constant. Consider a
temperature range ll.T over which a is a constant. Then we can write Eq. (6.20)
in an extended form as
<lz2 <lz3 ll.T (6.24)
Or, using the contracted notation, we can write
61 a1
62 <lz
63 Cl3
ll.T (6.25)
64 Cl4
6s as
66 a6
The number of independent components of aj as well as the different nonzero

components of aj for different crystal symmetries are given in Table 6.5.
Ifwe know the coefficient of thermal expansion along the fiber direction and
perpendicular to it, then we can find the coefficient of thermal expansion in any
arbitrary direction [hkl] as follows. Let the arbitrary direction [hkl] have
Table 6.5 Independent components of Ili
Symmetry No. of Components

Components
TricIinic 6 Ill' 1l2, 1l3, 1l4,ll s,116

Monoclinic 4 Il l ,1l2,1l3,lls
Orthorhombic 3 Ill' 1l2' 113
Hexagonal/Tetragonal 2 III= 112 = Il~
= Ilil
113
Cubic III = 112 = 113 = Il
direction cosines n l, n 2, and n3, then we can write for the linear coefficient of
thermal expansion, (1hkl' in that direction as
(6.26)
In a transversely isotropic fibrous composites (i.e., hexagonal symmetry), we

have (1, = (12 = (1~, perpendicular to the fiber axis and (13 = 1l1I' parallel to the
fiber axis. Then, remembering that n12+ n22 + n3 2 = 1, we can recast Eq. (6.26)
as
(6.27a)
(6.27b)
where e is the angle between direction [hkl] and the fiber axis.
6.2.1 Expressions for Coefficients of Thermal Expansion of Composites
Various equations have been proposed for obtaining the coefficients thermal
expansion of a composite, knowing the material constants of the components
and their geometric arrangements (Turner 1946; Kerner 1956; Schapery 1968).
Different equations predict very different values of expansion coefficient for
a given composite. Almost all expressions, however, predict expansion
coefficient values different from that given by a simple rule of mixtures (= arVr
+ am Vm)' This is because these equations take into account the important fact
that the presence of a reinforcement, with an expansion coefficient different
from that of the matrix, introduces a mechanical constraint in the system matrix.
A fiber will cause a greater constraint on the matrix than a particle. One can
regard a particulate composite as a homogeneous material in a statistical sense,
i.e., assuming a uniform distribution of the particles in the matrix, Fig. 6.8. Let
us denote the volume fractions of the two phases making a particulate
composite by V 1 and V2 (= 1- V I)' Various researchers have derived bounds and
given expressions for the coefficients of expansion and other transport
properties such as thermal conductivity. Kerner (1956) developed the following
expression for the volumetric expansion coefficient of a composite consisting
of spherical particles dispersed in a matrix:
(6.28)
where subscripts c, m, and p denote the composite, matrix, and reinforcement,

respectively; Pis the volumetric expansion coefficient and K denotes the bulk
modulus. Kerner's expression does not differ significantly from the rule of
mixtures because the constraint provided on the matrix by particulate
reinforcement is much less than that provided by fibers. The coefficient
of linear thermal expansion according to Turner (1946) is given by:
am VmKm + ap VpKp
(6.29)
VpKp + VmKm
where the symbols have the significance given above. Turner's expression,
generally, gives an expansion coefficient much lower than the rule-of-mixtures
value.
Fig. 6.S Unifonnly distributed particles in a matrix giving an isotropic composite.

Unidirectionally aligned fibrous composites have two (or sometimes three)

thermal expansion coefficients: ad in the longitudinal direction and act in the
transverse direction. Fibers generally have a lower expansion coefficient than
that of the matrix, and thus the former mechanically constrain the latter. This
results in an a c1 smaller than act for the composite. At low fiber volume
fractions, it is not unusual to find the transverse expansion of a fibrous
composite, act, greater than that of the matrix in isolation. The long stiff fibers
prevent the matrix from expanding in the longitudinal direction and as a result
the matrix is forced to expand more than usual in the transverse direction. It
should be pointed out that in the case of some CMCs, this situation can be
reversed. For example, in the case of alumina fibers(a = 8 x 10- 6 K- 1) in a low
expansion glass or ceramic matrix, it is the matrix that will constrain the fibers,
i.e., the situation in this case is reverse of the one commonly encountered.
We give below some expressions for the coefficients of thermal expansion of

unidirectionally reinforced fiber composites. All of these analyses involve the
following assumptions:
(a) the bonding between the fiber and matrix is perfect and mechanical in
nature, i.e., no chemical interaction is allowed.
(b) the fibers are continuous and perfectly aligned.
(c) the properties of the constituents do not change with temperature.
Schapery (1968) used energy methods to derive the following expressions for
expansion coefficient of a fibrous composite, assuming Poisson's ratios of the
components are not very different. The longitudinal coefficient of expansion for
the composite is
(6.30)
and the transverse coefficient of expansion is
(6.31 )
and
(6.32)
where a c1 is given by Eq. (6.30) above. For low fiber volume fractions, V r < 0.2
or 0.3, act can be approximated by

act ~ (1 + vm)amVm + a/Vr (6.33)
Figure 6.9 shows the variation of acl and act for carbon fibers in a glass matrix.
The following numerical values for the properties of matrix (indicated by
subscript 1) and fibers (indicated by subscript 2) were used to used to construct
this figure:
Borosilicate glass matrix: am = 3.5 X 10- 6 K- I , Em = 60 GPa, vm = 0.2

PAN-based carbon fiber: UrI = 0.1 X 10-6 K- I, af2 = 7.0 x 10- 6 K-I,
En = 390 GPa, Vf = 0.2
Note the marked anisotropy in the coefficient of thermal expansion for this
system. This anisotropy is accentuated by the inherent anisotropy of expansion
in the carbon fiber. The reader should note that at Vr equal to 1, we have the
two values of the coefficient of expansion of carbon fiber itself. Alumina fiber
in a glass type matrix can have the situation where the fiber has a higher
coefficient of expansion than the matrix'. Polycrystalline alumina fiber is
isotropic in expansion. Figure 6.10 shows a plot of the longitudinal and
transverse coefficient of thermal expansion of the composite: alumina fiber (Ur
= 8.5 X 10-6 K- I , E f = 380 GPa) in a glass matrix (am = 4.75 x 10- 6 K- I , Em = 60
GPa).
7r-~~~~~~~~~-'
6
5
I
~ 4
to
I
o 3
t:J 2
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Vf
Fig. 6.9 Variation of acl and act for carbon fibers in a glass matrix. Note that carbon
fiber itself (Vf =1) is anisotropic in a.
The anisotropy in expansion can be reduced somewhat if the composite

contains randomly oriented short fibers or whiskers in three dimensions. We
can write for its isotropic thermal expansion coefficient as
a '" (6.34)
3
where act and act are given by Eqs. (6.30) and (6.31).
It should be pointed out the length of fiber as well as fiber orientation have an
effect on the coefficients of thermal expansion It would appear that the fiber
length has a more sensitive effect on the expansion characteristics because the
constraint on the matrix is highest when the fibers are continuous. An extreme
case of a low constraint is one of a matrix containing spherical particles, i.e.,
9.5
9.0
8.5
8.0
i 7.5
~
'"I 7.0
.....
0
~~
E-<
6.5
U
6.0
5.5
5.0
4.5
4.0
10.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Vr
Fig. 6.10 Variation of longitudinal and transverse coefficient of thermal expansion for
an alumina fiber/glass matrix composite. Alumina fiber is isotropic in expansion.
aspect ratio = 1. Continuous fibers with an aspect ratio of infinity represent the
other extreme in constraint. The effect of the fiber length of the thermal
expansion can be incorporated into Schapery's equation as (Marom and
Weinberg, 1975; Vaidya and Chawla, 1994)
where the value of k is given by
k=-
e (e<e c' O<k<O.5)
2e c
ec
k=l--e (e>e c ,O<k<l)
2
where e is the length and Qc is the critical length of the fiber (for more about
critical fiber length, see chapter 7).
Chamis (1984) used a simple force balance to derive expressions for a. t and~.
Rosen and Hashin (1970) used a plane strain model to derive the following
expressions for a. t and a.2 :
where a:= acYr + amVm' K.: is the bulk modulus of the composite, Ee is the
elastic modulus of the composites, and k. is the plane strain transverse bulk
modulus of the composite.
6.3 THERMAL CONDUCTIVITY
The heat flow in a material is proportional to the temperature gradient, and the
constant of proportionality is called the thermal conductivity. Thus, in the most
general form, using indicial notation, we can write
where qj is the heat flux along X; axis, dT/dxj is the temperature gradient across
a surface that is perpendicular to the x.; axis, and kij is the thermal conductivity.
As should be evident from the two indexes, thermal conductivity is also a
second-rank tensor. Although, kjj not a symmetric tensor in the most general
case, it is a symmetric tensor for most crystal systems. For an isotropic material,
kij reduces to a scalar, i.e., a number, k. For an orthotropic material, we have
three constants along the three principal axes, viz., kll' k22' and k33 • For a
transversely isotropic material such as a unidirectionally reinforced fibrous
composite, there will two constants: thermal conductivity in the axial direction,
k.b and that in the transverse direction, keto The thermal conductivity in the axial
direction, k.1 can be predicted by a rule-of-mixtures type expression (Behrens
1968)
(6.35)
where krl is the thermal conductivity of the fiber in the axial direction and ~
is that of the isotropic matrix, and V r and Vm are the volume fractions of the
fiber and the matrix, respectively.
In a transverse direction, the thermal conductivity of a unidirectionally aligned

fiber composite (Le., transversely isotropic) can be approximated by the action-
in-series model discussed above. This would give
Figure 6.11 shows the variation in longitudinal and transverse thermal

conductivity as a function of fiber volume fraction for the carbon fiber/glass
matrix composite. In this plot we have taken the thermal conductivity of the
glass matrix to be 1.7 Wm-1K- 1 and that of carbon fiber to be 70 and 10
Wm-1K- 1 for transverse and longitudinal directions, respectively. However,
more sophisticated expressions have been derived for the transverse thermal
conductivity. In real composites, it is likely that the thermal contact at the
fiber/matrix interface will be less than perfect because of thermal mismatch,
etc. Hasselman and Johnson (1987) analyzed the thermal conductivity of
composites where the interface region may have a different thermal
conductance than the matrix or the fiber. They arrived at the following
expression for thermal conductivity in the transverse direction:
k ct = k2 = I<a. [(klI<a. - 1 - klah i) Vr + (1 + klI<a. + k/ahi)]/[1 - k~ + k/ah i) Vr

+ (1 + klI<a. + k/ahi)] (6.36)
where I<a. is the matrix thermal conductivity, k f is the transverse thermal

conductivity of the fibers, Vr is the fiber volume fraction, a is the fiber radius,
and hi is the thermal conductance of the interface region. It can be
'~ 80~--~--~--~--~--~--~---r---r--~---'
'E 70
3_ 60
>.
."§; 50
g 40
"0
§ 30
u
_ 20
c
~ 100 r~~;:==~======~==:c:::C:::I:::i::~:::J
f=. t
o 0.1 0.2 0.3 0.40.50.60.7 0.80.9 1.0
Vf
Fig. 6.11 Longitudinal and transverse thermal conductivity as a function of fiber volume
fraction for a carbon fiber/glass matrix composite.
seen that the effect of interfacial conductivity is governed by the magnitude of

nondimensional parameter, klahi. Ffor perfect thermal contact hi will have a
value of infinity and it will be equal to zero for a pore. Bhatt et al. (1992) have
studied the important role of interface in controlling the effective thermal
conductivity of composites. In particular, one can use the measurement of
thermal conductivity as a nondestructive tool to determine the integrity of the
fiber/matrix interface.
We can also use Halpin-Tsai-Kardos equations to obtain the following

expression for the transverse thermal conductivity of a composite containing
unidirectionally aligned fibers:
and
11 = [(kf2/~) -1] /[(kf2/~) + 1]
where we have taken l; equal to I in the Halpin-Tsai-Kardos equation.
We can find the thermal conductivity kx and ky in two arbitrary but mutually
perpendicular directions, x and y if we know the thermal conductivity in the
principal directions 1 and 2" respectively, by the following equations:
kx = k( cos2 e + k2 sin2 e
~ = k( sin2 e + k2 cos2 e
kxy = (k2 - k() sine cose
where e is the angle measured from the x-axis to the I-axis and k"y can be
considered to be a thermal coupling coefficient.
6.4 DENSITY
Density is one physical property for which the rule of mixtures works
absolutely and irrespective of the geometry of the phases involved. The total
mass of a composite is the sum total of the masses of the reinforcement and the
matrix, that is,
m c = m r +m m (6.37)
where m is the mass and the subscripts c, r, and m indicate composite,

reinforcement, and matrix, respectively. The reinforcement can be a particle,

a whisker, or a fiber. Note that Eq. (6.37) is valid even in the presence of any
voids in the composite. The volume of the composite, however, must include
the volume of voids Vv. Thus, we can write for the volume of a composite,
(6.38)
Dividing Eq. (6.37) by me and Eq. (6.38) by Vc and denoting the mass and
volume fractions by~, M m, and V" V m' V v, respectively, we can write
and
The density of the composite Pc= mjvc is then given by
P = me = m, +mm P,v, +pmvm

c ve ve Vc
Or
We can also derive an expression for Pc in terms of mass fractions. Thus,
mr mm
-+-+vv
P, Pm
Pc = - - - - -
M
_r
M v
+ --2!!. + _v
Mr Mm Vv
-+-+-
Pr Pm PcVc
Mr Mrn Vv (6.39)
-+-+-
Pr Pm Pc
We can use Eq. (6.39) to obtain indirectly the volume fraction of voids in a
composite. Rewriting Eq. (6.39), we obtain
From this, we can obtain the volume fraction of voids as
(6.40)
6.5 ELECTRICAL PROPERTIES
Many electrical properties of composites can be predicted by simple

expressions. Although such characteristics fall outside the purview ofthis text,
we shall discuss one property, viz., dielectric constant. Dielectric materials are
used to increase the capacitance of electrical capacitors. They are also used as
electrical insulators. The dielectric constant of a material can vary with
temperature and frequency. The dielectric constant of a material, also known
as the relative permittivity, E" can be defined as the ratio of the permittivity of
the material (e) to that of vacuum (eo):
(6.41)
Consider a dielectric of relative permittivity, E" inserted in a parallel plate

capacitor, Fig. 6.12. The capacity C of this capacitor is given by
A
C = -s (6.42)
d
where A is the plate area, d is the plate spacing, and e is the permittivity of
vacuum.
Dielectric
Positive Plate
~~----{-U
+ +
Negative Plate
Fig. 6.12 Dielectric of relative pennittivty, sr' inserted in a parallel plate capacitor.
A
C = -B
d 0 Br = C0 Br
where Co is the capacity without the dielectric material in between the plates.
Using the action-in-series, we get
1 VI v2
= -+-
Bes BI B2
Using the action-in-parallel, we get
The reader is referred to a review article by McLachlan et al. (1990) on

electrical resistivity of composites.
REFERENCES
Behrens, E., (1968) J. Composite Materials, 2, 2.

Bhatt, H., K. Y Donaldson, D.P.H Hasselman, and R.T. Bhatt (1992) J. Mater. Sci., 27,
6653.
Chamis, C.c. (1983) NASA Tech. Memo. 83320.
Chamis, C.C. (1984) Sampe Quarterly, 15, 14.
Halpin, J.C. and J.L Kardos (1976) Polymer Eng. & Science, 16,344.
Halpin, J.C. and S.W. Tsai (1967) Environmental Factors Estimation in Composite
Materials Design, AFML TR 67-423.
Hashin, Z. (1962)J. App. Mech., 29,143.
Hashin, Z. and B.W. Rosen (1964) J. App. Mech., 31, 223
Hashin, Z. and S.A. Shtrikman (1963) J. Mech. Phys. Solids, 11, 127.
Hasselman, D.P.H. and L.F. Johnson (1987) J. Composite Materials, 27,508.
Hill, R. (1964)J. Mech. Phys. Solids, 12,199.
Hill, R.(1965) J. Mech. Phys. Solids., 13, 189.
Kardos, J.L. (1971) CRC Critical Reviews in Solid State Sci., 3, 419.
Kerner, E.H.(1956) Proc. Phys. Soc., 808.
Marom, G. and Weinberg, A.(l975)J. Mater. Sci., 10,1005.

McLachlan, D.S., M. Blaszkiewicz, and R.E. Newnham (1990) J. Amer. Ceram. Soc.,
73,2187.
Nielsen, L.E. (1974) Mech. Prop. o/Polymers and Composites, vol. 2, Marcel Dekker,
New York.
Rosen, B.W (1973) Composites, 4, 16.
Rosen, B.W. and Z. Rashin, (1970) IntI. Jour. Eng. Sci., 8, 157.
Schapery, R.A.(1968) J. Composite Materials, 2, 380.
Turner, P.S.(1946) J. Res. Nat. Bur. Standards, 37, 239.
Vaidya R.U.and K.K. Chawla (1994) Composites Science and Technology, 50, 13.
SUGGESTED READING
Daniel, I.M. and O. Ishai (1994) Engineering Mechanics 0/ Composite Materials,

Oxford University Press, New York.
Rashin, Z. (1983) Analysis of Composite Materials - A Survey, J. App. Mech., 50 481.
Nye, J.F. (1969) Physical Properties o/Crystals, Oxford Univ. Press, London, p. 131.
Cliapter 7
MECHANICAL BEHAVIOR: LOAD

TRANSFER; MONOTONIC, CYCLIC, AND
CREEP BEHAVIOR
In this chapter, we discuss various aspects of mechanical behavior of ceramic

matrix composites: mechanics of load transfer from the matrix to the fiber,
behavior ofCMCs under monotonic and cyclic (fatigue) loading, and creep
conditions.
7.1 LOAD TRANSFER FROMMATRIX TO FIBER - SHEAR-LAG MODEL
Besides holding the fibers together, the matrix has the important function of
transmitting the applied load to the fibers. Many researchers (Cox, 1952;
McCartney, 1989; Gresczuk, 1969; Lawrence, 1972; Lee and Daniel, 1992)
have contributed to this subject of load transfer from the matrix to the fiber.
Let us focus our attention on a fiber embedded in a matrix. Figure 7 .1a shows
the situation prior to the application of an external load, where we have drawn
imaginary lines running through the fiber/matrix interface in a continuous
manner in the unstressed state. Let us load this composite axially. We assume
that no direct loading of the fibers occurs. Then, the fiber and the matrix will
experience locally different axial displacements because of the different elastic
moduli ofthe components. Under such circumstances, our imaginary vertical
lines in the unstressed state will become distorted as shown in Fig. 7.1 b. The
resultant displacement in the fiber, u, and in the matrix, v, are shown in Fig.
7.2. The larger the difference between the elastic modulus of the fiber and the
matrix, the larger will be the difference in their axial displacements. Different
axial displacements in the fiber and the matrix mean shear strains are produced
in the matrix on planes parallel to the fiber axis and in a direction parallel to
the fiber axis, i.e., displacements and stresses do not vary with angle about the
206 MECHANICAL BEHAVIOR
(a) Before Deformation
(b) After Deformation

Fig. 7.1 A fiber embedded in a matrix: (a) unstressed state and (b) stressed state.
Imaginary vertical lines become distorted on application of load.
fiber axis. Thus, the transfer of the applied load to the fiber occurs by means
of these shear strains in the matrix. Frequently, this model has been referred
to in the literature as the shear-lag model.
It is instructive to examine the stress distribution along the fiber/matrix

interface. In a brittle matrix (polymer or ceramic) composite, both the matrix
and fiber are likely to remain elastic until they fracture. In the case of a metal
matrix composite, one can easily imagine a stage before fracture where the
matrix deforms plastically while the fiber deforms elastically. Consider a
single fiber oflength ~ embedded in a matrix. Let us subject this composite to
an axial strain, e. We make the following assumptions in the shear-lag model:
(a) There exists a perfect bond between fiber and matrix, i.e., there is no
sliding between the two.
(b) Poisson's ratios of fiber and matrix are equal, which implies an absence of
transverse stresses when the load is applied along the fiber direction.
(c) All load transfer occurs via shear in the matrix and at the fiber/matrix
interface. This shear stress is assumed to be constant. There is no direct
loading of the fibers.
(d) Stress concentrations at the fiber ends are ignored.
I
" 2R
I
1
Fig. 7.2 A single fiber embedded in a matrix and the assembly subjected to an axial
strain, e, and the resultant displacements in the fiber (u) and the matrix (v).
Let the displacement of a point at a distance x from one extremity of the fiber
be u, in the presence of a fiber, and v, in the absence of a fiber, Fig. 7.2. The
load transfer from matrix to fiber occurs because of the difference in these two
axial displacements, u and v. For the transfer of load from matrix to fiber, we
can write the following expression:
dP/dx = B (u - v) (7.1)
where P is the load on the fiber, and B is a constant that depends on geometric
arrangement of fibers, matrix type, and moduli of fiber and matrix.
Differentiating Eq. (7.1), we get
d2P/dx2 = B (du/dx - dv/dx) (7.2)
The axial strain in the matrix away from the fiber is e = dv/dx while that in the
fiber, using the Hooke's law, is du/dx = PIEr Ar, where P is the axial load on
the fiber, Ar is the cross-sectional area of the fiber, and Er is the fiber modulus.
Thus, we can rewrite Eq. (7.2) as
A solution to this linear differential equation is
P = Er Ar e + S'cosh ~x + T'sinh ~x
O'r= P/A r = Ere + S cosh ~x + T sinh ~x
where Sand T are arbitrary constants to be determined from the appropriate

boundary conditions and ~ = (B/Ar Eryl2.
We use the following boundary condition to evaluate the constants Sand T:
Of = 0 at x = 0 and x = e
This boundary condition follows from the fact that there is no direct loading
of the fiber from the two ends. The condition P = 0 at x = 0 gives
o = e Ef + S
S = -eEf
while the condition P = 0 at x = e gives
o = eEf+Scosh~Q+Tsinh~Q = eEf-eE(cosh~e+Tsinh~Q
T = e E [cosh ~ Q- 1] = e E tanh ~
f sinh~Q/2 f 2
Using the "halfangle" trigonometric formulas, we get the following expression

for the tensile stress on the fiber
o = E e [1 _cosh ~(Q/2 - x) ]
f f cosh ~(Q/2)
(7.4)
Q
for 0 < x < -
2
The maximum possible value of strain in the fiber is the imposed strain, e, and,
thus, the maximum stress is eEr' Thus, shear-lag analysis predicts that if the
fiber is long enough, then the stress in the fiber will increase from the two ends
to a maximum value, Of = Ere. Thus, the average stress in the fiber of length Q
is given by
- = +
~
fQ crdx = E e [1 -
I
tanh(~Q/2)
~e/2
(7.5)
o
We can obtain the variation of shear stress, 't, along the fiber/matrix interface
by considering the equilibrium of forces acting over an element dx of fiber,
radius, rr, and acting in the fiber direction, Fig. 7.3. We can write from Fig. 7.3
(7.6)
Now P, the tensile load on the fiber, is equal to nr/crr. Substituting this in Eq.
(7.6), we get
(7.7)
From Eqs. (7.4) and (7.7), we obtain
't =
EIrfe A
p •
~
sinh ( i-x)
2
(7.8)
2 cosh~
2
Figure 7.4 shows the variation of crr and 't with distance x along the fiber. The
maximum shear stress in Eq. (7.8) can be either the shear yield stress of the
matrix or the shear strength of the fiber/matrix interface; whichever is smaller
should be used in Eq. (7.8). In ceramic matrix composites, the shear strength
of fiber/matrix interface would be expected to be lower than the matrix failure
stress in shear, there being no realistic possibility of yielding in shear of the
ceramic matrix. Thus, in CMCs, one can safely take 'tmax to be the shear
strength of the fiber/matrix interface, 'ti.
Now we may determine the constant B. As mentioned earlier in this section,

the value ofB depends on the fiber packing geometry. Consider Fig. 7.4 again
and let the fiber length, e, be much greater than the fiber radius, rr. Imagine the
fibers to be embedded in tubes of radius R. Then 2R will be the center to
center fiber spacing. Let us also denote the shear stress in the fiber direction
L---L
p p + t.. p
<If---+--'
" L
.. dx - - - - - - 4.....1
Fig. 7.3 Equilibrium offorces acting over an element dx of fiber (radius, rr).
at a distance r from the axis as -r(r). Then, at the fiber surface (r = rf)' we can
write
dP
dx
B = _ 21trf -r(rf )
(7.9)
(u -v)
Let w be the real displacement in the matrix. Then, at the fiber/matrix

interface, no sliding being permitted, w = u. At a distance R from the center
of a fiber, i.e., at the surface of an imaginary tube containing this fiber, the
matrix displacement is unaffected by the fiber presence, and we have w = v.
Considering now the equilibrium of forces acting on the matrix volume
between rf and R, we can write
(7.10)
(7.11 )
For the shear stress in the matrix, we can write from Hooke's law, -r(r) = Gm'Y,
where G mis the matrix shear modulus and 'Y is shear strain given by
dw = "C(r) = "C(r)rj
y= dr G G r (7.12)
III m
Then, integrating from r to R, we get
(7.13)
But, by definition
~w = (v-u) = -(u-v) (7.14)
T,(rj)rj
(7.15)
u-v
From Eqs. (7.9) and (7.15), one gets
B = (7.16)
and from Eq. (7.13), one can obtain an expression for the load transfer
parameter, ~,
(7.17)
The term (Rlrf ) in this expression brings in the fiber packing factor. For a
square array of fibers, In (Rlrf) = 112 In (7tlVr), while for a hexagonal packing,
In (Rlrf ) = 112 In (21t!V3Vf). We can define the packing term, in a more general
way, in the following manner: In (Rlrf) = 112 In (<f>maxNf)' where <f>max is the
maximum packing factor. Substituting this in Eq. (7.17), we get
41tG m
The greater the value of the quantity (GmlEf ), the greater is the value of Band
the more rapid is the load increase in the fiber from either end.
It follows from the shear-lag analysis that in order to load a fiber to its
maximum strength, the shear strength given by Eq. (7.8) must be large. In a
metal matrix composite, the relevant shear strength is the matrix yield stress
in shear, because the metallic matrix will deform plastically in shear rather
easily in response to the high shear stresses developed. Of course, if the
fiber/matrix interface shear strength is less than the matrix yield stress in
shear, then the interface will fail first. We expect this to be the likely situation
in ceramic and polymer matrix composites, because in these brittle matrix
composites, frictional sliding at the interface is more likely than plastic flow
of the matrix. Therefore, in the case ofCMCs, the limiting shear stress will be
the interface strength in shear, "Cj • If the differential shrinkage between the
matrix and the fiber during processing or any other source of mechanical
interlocking at the interface results in a radial pressure p on the fibers, then "Cj
will be equal to fJP, where fJ is the coefficient of sliding friction between fiber
and matrix. Considering once again the equilibrium offorces over an element
#12 of the fiber
1td 2 Q
a-- 't.1td-
'j 4 I
2
or
Q aj
-
d 2't;
Here, we consider #12 and not #because the fiber is being loaded from both
ends. Given a sufficiently long fiber and a strong enough interface, it should
be possible to load the fiber to its breaking stress, afu, by means of load
transfer through the matrix. Let (#Id)c be the minimum fiber length to diameter
ratio necessary to accomplish this. We call the ratio (Q/d), the aspect ratio of
a fiber, and (lJd)e, the critical aspect ratio necessary to attain the fracture
strength stress of the fiber, (fji,. Thus, we can write
(7.18)
For a given fiber diameter, d, we can think of a critical fiber length, 1,., thus,
(7.19)
Over a length 1,12 from either end, the load in the fiber builds up. Strain builds
in a likewise manner. Beyond f/2 from the two ends, (i.e., in the middle
portion of fiber) the local displacements in matrix and fiber are the same, and
the fiber carries the major load while the matrix carries only a minor portion
of the applied load. It is instructive to decipher the implications ofEq. (7.19).
First of all, it tells us that the fiber length fmust be equal or greater than ec for
the fiber to be loaded to its maximum stress, (fji/. Secondly, the load transfer
length of a fiber in a matrix is independent of the total length of the fiber, i.e.,
the ratio, f/f, decreases as the fiber length increases, Fig. 7.5. An examination
of Fig. 7.5 shows that even for (fld) > (lJd)c the average stress in the fiber will
be less than the maximum stress to which it is loaded in its central region. In
fact, we can write for the average fiber stress as follows:
+fa;Jx
Q
af = = +[ap-ec)+Pa}c] = +[a}-ec(af-pa)]
o
or
af (1 -l.=f)
Q/Q
c
(7.20)
atrix
y._.f})-i._--
2rr ;
p p+t..P
(-j ~i_b~_<
:: ~'t :: \
:/J2 : :/J2: \
;..--.: ;..--.:
0:
Of max :
't :
O L
. ~ _ _--_~
Fig. 7.4 Variation of tensile stress, 0, and shear stress, 't, along the fiber/matrix
interface.
( lId) < ( lid )c ( lid ) = ( lid )c (lId) > ( lid )c
l-~- -~-~--~
Fig. 7.5 Load transfer as a function of fiber length. Fiber length !must be equal to or
greater than "" for the fiber to be loaded to its maximum stress, 0fu' ~ is the critical
length of fiber for load transfer.
where par is the average stress in the fiber over a portion ~/2 of its length at
both the ends. Thus, one may regard pas a load transfer function.
Many researchers have proposed modified shear lag analyses for a

unidirectionally reinforced ceramic matrix composite under longitudinal
tensile loading. For example, Lee and Daniel (1992) obtained closed form
solutions for stress distributions in the matrix and fiber as a function of applied
stress and constituent properties. They assumed that the brittle matrix, with a
lower strain to fracture than the fiber, failed by transverse cracking under
longitudinal tensile load. They obtained closed form expressions for the matrix
crack density, debond length, reduced axial stiffness of the damaged composite
as a function of applied stress and constituent properties. The shear lag
parameter in this case also turns out to be a function of geometry and moduli
of the constituents.
7.2 MONOTONIC BEHAVIOR
. The term monotonic behavior refers to the behavior of a material under an

applied stress that increases in one direction, i.e., not a cyclic loading
condition. We first provide a brief description of some important features of
mechanical testing of ceramics and then discuss the monotonic mechanical
behavior of some CMCs.
7.2.1 Mechanical Testing of Ceramics
Flexure tests are commonly used to study the mechanical behavior of ceramics
at low and high temperatures. This is because the technique is easy to use and
inexpensive. However, the interpretation of flexure test data is not straight
forward. This is especially so in fiber reinforced ceramics, mainly because
they are highly anisotropic. These considerations have led to the development
of tensile testing of ceramics and ceramic matrix composites (Kandil and
Dyson, 1988; Fett et aI., 1988; Carroll and Tressler, 1988, 1989; Carroll et aI.,
1989; Wiederhorn and Hockey, 1991; Wiederhorn et aI., 1988; Davidge and
Briggs, 1989; Ohji, 1988; Liu and Brinkman, 1985; Seshadri and Chia, 1987;
Vaidyanathan et aI., 1988; Jenkins, et aI., 1990). Actually, as Davidge and
Briggs (1989) point out, flexure tests are easy to perform but difficult to
interpret while tensile tests are difficult to do but easy to interpret.
(a) Bend or Flexure Test
Bend test is one of the easiest to do and is very commonly used, especially in
the evaluation of brittle materials. There are two basic governing equations for
a simple beam elastically stressed in bending:
M
---
E
(7.21)
J R
and
M 0"
- - (7.22)
J Y
where M is the applied bending moment, I is the second moment of area of the
beam section about the neutral plane or the moment of inertia, E is the Young's
modulus of elasticity of the material, R is the radius of curvature of the bent
beam, and 0" is the tensile or compressive stress on a plane distance y from the
neutral plane.
For a beam of a uniform, circular cross section, the moment of inertia I of the
beam is given by
rrd 4
J=- (7.23)
64
where d is the diameter of the circular beam. For a beam of a uniform,

rectangular cross section, we have
bh 3
J=- (7.24)
12
where b is the beam width and h is the height ofthe beam.
Bending takes place in the direction of the depth, i.e., hand yare measured in
the same direction. Also, for a beam with symmetrical section with respect to
the neutral plane, replacing hl2 (or d/2) for yin Eq. (7.2) gives the stress at the
beam surface.
Figure 7.6 shows the elastic normal stress distribution through the thickness
of a bent beam. The stress and strain vary linearly with distance across the
section, with the neutral plane representing the zero level. The material on the
outside or above the neutral plane of the bent beam is stressed in tension while
that on the inside or below the neutral plane is stressed in compression. In the
elastic regime, the stress and strain are related by
cr=EE (7.25)
From equations (7.21), (7.22) and (7.25) we can obtain the following simple
relationship valid in the elastic regime,
(7.26)
Thus, the strain E obtained in a beam bent to a radius of curvature R varies

linearly with distance y from the neutral plane across the beam thickness.
We have given the expressions for three-point and four-point bend strength in
chapter 5, which we recall here. The maximum tensile stress in a three-point
bend test occurs on the outermost surface under the middle loading point and
is given by
r---1-----~--c--~--~
I
"- • '" Ne~ltral

aXIs
Fig. 7.6 Normal stress distribution through the thickness in an elastically bent beam.
The stress and strain vary linearly with distance across the section, with the neutral
plane representing the zero level.
3PS
0" " ' - - (7.27)
where P is the applied load, S is the load span, b is the specimen width, and h
is the specimen height.
In such a test, the maximum shear stress, 't, occurs at the midplane and is given
by
3P
't " ' - - (7.28)
4bh
The ratio of the maximum tensile stress to shear stress is thus given by
't h
(7.29)
°" 2S
It is easy to see that the beam under test will fail in tension ifthe span to depth
ratio, Slh, is made very large. Typically, for S/h values greater than 20, failure
in tension will occur, while for Sih less than 5, failure in shear will result. The
test will also be invalid if shear and tensile failure occur simultaneously. It is
advisable to make an examination of the fracture surface after the test and
make sure that the fracture has taken place in the presumed failure mode. For
an elastic beam bent in four-point, the bending moment in the central position
is given by
20"max I EI
M= (7.30)
112 R
where I is the moment of inertia, tl2 is the distance from the neutral axis to the
outer surface, O"max is the maximum normal stress on a transverse section, and
R is the radius of curvature.
Figure 7.7 shows the variation in stress with specimen length for different
tests. Actually, what we plot Fig. 7.7 (a and b) is the variation of bending
moment, M, but for a given set of specimen dimensions, this corresponds to
the stress variatio~ as well. In a three-point bend test, Fig. 7.7a, the maximum
stress occurs along a plane in the center of the specimen span length. In a four-
point bend test, Fig. 7.7b, a constant stress acts over the inner span length of
the specimen. Four-point bending is also called pure bending since there are
no transverse shear stresses on the cross-sections of the beam between the two
loading points, i.e., the inner span length. In a tensile test, the whole gage
length of the specimen is subjected to a uniform stress, see Fig. 7.7c. We
discuss the tensile test situation below at some length because of the increasing
importance of tensile testing for ceramics.
A schematic stress-strain curve obtained in a bend test done under

displacement control is shown in Fig. 7.8 a while an actual curve obtained in
a three-point bend test for a BN coated Nextel480 fiber (a mullite type fiber
with some boria in it) in a glass matrix is shown in Fig. 7.8b. From the stress
corresponding to first appearance of cracks in the matrix, 0'0 onward, the
stress-strain curve becomes nonlinear until the ultimate stress, O'u is reached.
At O'u' fiber fracture occurs. Straining past this point results in fiber pullout.
Note the noncatastrophic failure mode in this case. Note that the curve
pertaining to the uncoated Nextel480/glass matrix showed a catastrophic
t t f
I
t
t t t
·:~;'
.. ..
____ (f~
··· .. ..
(a)
(b)
-c-:-::=J-- .. ..
j r-----------~..
SJ----
.
(e)
Fig. 7.7 Variation in stress with specimen length: (a) three-point bend test; maximum
stress occurs along a plane in the center of the span length; (b) four-point bend test;
a constant stress acts over the inner span length; (c) tensile test; the whole gage length
of the specimen is subjected to a uniform stress.
failure. This is because in the uncoated fiber reinforced composite interfacial

debonding and sliding are absent. There was also an increase in 0"0 of the
composite vis it vis that of the unreinforced matrix. This type of stress-strain
behavior has been likened by some researchers to the yield point phenomenon
observed in steels (Majumdar et aI., 1988). It would appear that the matrix
cracking initiation stress, 0"0 can be a "reliable" design parameter. In the case
of the fiber reinforced composite with proper interface characteristics, unlike
the monolithic glass or ceramic, after the initial matrix cracking stress, 0"0' the
composite does not fail abruptly. Instead, it shows a region of stable crack
growth. In this region various energy absorbing mechanisms such as
fiber/matrix debonding, crack deflection at the fiber/matrix interface, and fiber
pullout occur. These phenomena are conspicuous by their absence in
monolithic ceramics. Thus, use of continuous fibers in a ceramic matrix can
result in an increase in the strain-to-fracture and noncatastrophic failure. We
shall come back to this theme in chapter 9, and, in particular, on ways to
engineer the fiber/matrix interface to obtain such damage tolerant
characteristics.
(b) Tensile testing
Tensile testing is widely used to obtain mechanical characteristics of nonbrittle

solids such as metals. Brittle materials such as ceramics and glasses are
frequently tested in flexure (3-point or 4-point). Although flexure tests are
easy to do, results generally do not reflect randomly occurring defects in
highly stresses parts, e.g. in a turbocharger rotor (Wiederhorn and Hockey,
1991; Wiederhorn et aI., 1988). In a bend test, only a small material volume
of the test piece experiences the maximum load. In a tensile test, on the other
hand, the whole gage length ofthe specimen is subjected to a uniform stress.
This is the reason that the strength values obtained from a bend test are
considerably higher than those from a tensile test. In a tensile test, we get a
uniform stress field and relatively large effective volume oftest material.
Yet another complication that occurs in bending tests of fiber reinforced

composites has to do with the shifting of the neutral axis in flexure (Steif and
Trojnacki, 1994). The reader should recall that in a simple beam of a
monolithic material, the entire beam fails when the applied moment reaches
Fiber bundle failure
Fiber pullout
1
Strain -->
(a)
300
Nextel 480/BN/giass
250
«I 200
Q.
::::E
i.. 150
tiS 100
50
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Apparent deflection, mm
(b)
Fig. 7.8 (a) A schematic stress-strain curve ofa continuous fiber reinforced ceramic
matrix composite in a three-point bend test done under displacement control. The
curve is linear elastic up to <Jo when the first matrix crack appears. From <Jo onward,
the curve becomes nonlinear until the ultimate stress, <Ju is reached. At <Ju, fiber
fracture occurs. Straining past this point results in fiber pullout. (b) A three-point bend
test curve for a BN coated Nextel 480 fiber in a glass matrix. For the sake of
comparison, a similar curve of the uncoated Nextel 480 fiber/glass matrix is also
shown. Note the noncatastrophic failure mode in the case of BN coated fiber
composite. There is also an increase in <Jo of the composite vis a vis that of the
unreinforced matrix.
a critical value, which corresponds to maximum stress on the tensile side of

the beam. In a fiber reinforced CMC, the attainment of the critical bending
moment does not necessarily correspond to failure. The beam continues to
sustain load after reaching the critical bending moment and the stress
distribution continues to be linear through the thickness of the beam. For
equilibrium, the tensile forces must be balanced by compressive forces. The
compressive strength is much higher than tensile strength, the force balance
is maintained by a shifting of the neutral axis by some vertical distance. This
shift of the neutral axis continues to occur until the final failure of the beam.
The ratio of shear strength in flexure to tensile strength in flexure varies as a
function of the aspect ratio (length to transverse dimension) of the beam. As
the aspect ratio increases, the ratio of crfT. increases. This means that length of
the beam should be large enough to induce failure in tension. We make use of
the converse phenomenon in short beam or interlaminar shear test (see chapter
5) where shear stress is maximized.
Tensile tests are important for evaluating inherent flaw distributions and the
resultant strength distributions (Davidge and Briggs, 1989; Dyson et aI., 1989;
Ohji, 1988; Liu and Brinkman, 1985). One major problem in tensile testing
of ceramic and ceramic composites is the bending moments that can be easily
introduced during specimen alignment, gripping, and testing. Such accidental
bending moments are referred to as parasitic bending moments. Various grip
designs have been developed (Ohji, 1988; Liu and Brinkman, 1985) to take
care ofthis problem of parasitic bending moments. One such development, due
to Liu and Brinkman (1985), involves a hydraulic, self-aligning grip system
that results in a condition of near-zero bending moment during a tensile test.
The specimen grip is self-aligning and has a built-in flex coupling consisting
of eight tiny hydraulic pistons. These pistons reduce bending in the specimen
to near-zero, less that 1% bending in the gage length. With proper attention to
the gripping of sample, tensile tests on ceramics can be performed at
temperatures up to 1300 °C without much problem (Ohji,1988).
(c) Tensile Stress-Strain Behavior
If the fiber and the matrix have the same failure strain, then we can write for
the ultimate tensile strength of a composite (creu) as
where crfu and crmu are the ultimate tensile strengths of the fiber and the matrix,
respectively. This condition of the fiber and the matrix having an identical
223 MECHANICAL BERAVIOR
strain to failure being highly unlikely, we consider below two realistic

situations that are possible: (a) fiber failure strain less than that of the matrix
and (b) fiber failure strain greater than that ofthe matrix. Figure 7.9 shows the
stress-strain curves of the fiber and matrix for these two situations. The first
case is similar to that commonly encountered in most metal matrix and some
polymer matrix composites. It is also possible in some glass and glass-ceramic
matrix composites. Figure 7.9a also shows the variation of strength of the
composite as a function of the fiber volume fraction, Vf . This figure serves to
define two quantities, a minimum volume fraction, Vmin and a critical volume
fraction, Verit, of fibers. Vmin corresponds to the point of equality between the
strength of the composite and the strength of matrix times the volume fraction
of the matrix in the composite. Verit is the fiber volume fraction at which the
composite strength equals the strength of 100% matrix. For Vf > Vmin' we are
likely to have single fracture in the composite. In the second case, shown in
Fig. 7.9b, the fiber failure strain is greater than that of the matrix. Again,
depending on the relative strain-to-failure of the fiber and the matrix as well
as the relative amounts of the components in the composite, we can have a
single crack failure or a multiple crack failure in the composite, Fig. 7.9b.
Since the fiber failure strain is greater than that of the matrix, the matrix will
fail first. When that happens, then extra load is thrown onto the fibers. Two
things can happen at this point: If the amount of fibers is less than a critical
volume fraction, V erit, they will not be able to sustain this extra load and break.
Thus, at or below V erit, the matrix failure results in a failure of the composite.
The condition for mUltiple cracking in the matrix can be written as
where or' is the stress on the fiber corresponding to the failure strain of the
matrix. The condition of equality in the above expression serves to define a
critical fiber volume fraction, V erit • For fiber volume fractions greater than Verit,
a series of cracks will form in the matrix and it will fracture into thin slices.
Assuming that the fiber strain to failure is greater than that of the matrix, a
schematic tensile stress-strain curve of a unidirectional Nicalon fiber
reinforced SiC composite is shown in Fig. 7.10. The curve is linear elastic up
to 0 0 when the first matrix crack appears. On continued loading, additional
load is thrown on to the fibers. Assuming that the fibers can withstand this
additional load and that fiber/matrix interface remains intact, then additional
load will be transferred back to the matrix as per shear lag analysis described
above. This will result in, more or less, parallel cracks in the matrix, which
will be perpendicular to the tensile axis. Eventually, the specimen will be
traversed by parallel cracks, perpendicular to the fiber axis, and distance x and
Single fracture
Efu Emu
Strain - v,-
(a)
elu>emu
Uiu Single: Multiple matrix
fracturel fracture elu
I
u;
T T
- - -- --
gj
--
CIl
CIl
--
~
--
~
CI) CI)
----\
emu u",=u;V,+umu (1-V,)
Umu emu 0 V...

Strain --+ V,--+
(b)
Fig. 7.9 Stress-strain curves and strength of composite vs. fiber volume fraction
showing conditions for single and multiple cracking, in a unidirectionally aligned
fiber composite: (a) matrix strain to failure greater than that of the fiber, and (b)
matrix strain to failure less than that of the fiber.
2x apart. The distance x, assuming a constant interfacial shear stress, 'tj is

given by (Aveston et aI., 1971)
(d) Steady State Cracking
When a crack in CMC is small in length, the stress required for its growth
depends on its length. When the crack becomes sufficiently long and when it
is bridged by fibers, we obtain a state of steady state cracking. Under this
condition the crack grows at a constant stress. At some distance away from the
leading edge of the crack, an equilibrium crack opening develops and the load
is fully supported by the fibers. The stress intensity associated with matrix
cracking then becomes independent of crack length. According to Marshall et
aI. (1985), a steady state cracking will exist when the crack length exceeds a
critical value, called the characteristic crack length. A CMC showing such
steady state cracking will have very desirable damage tolerant characteristics
and significant nonlinearity before reaching the maximum stress value.
(e) Modulus, Strength, and Toughness
We have repeatedly said that a mere increase in elastic modulus and strength
is generally not the objective in ceramic matrix composites because the
monolithic ceramics have very high strength. However, such an increase in
elastic modulus and strength can be a welcome attribute for low modulus
matrix materials, for example, glasses, glass-ceramics, and some crystalline
ceramics such as MgO, mullite, etc. The major driving force, however, for the
use of continuous fibers in ceramic or glass matrix materials is to toughen
them. This idea of using continuous fibers in a brittle matrix to obtain
enhanced toughness can be traced to the 1960s (Bowen, 1968; Evans, et aI.,
1961; Tattersall and Tappin, 1966). Various researchers have used different
measures of toughness. One convenient term in this regard is the work of
fracture. Work of fracture can be defined as the amount of energy required to
produce a unit area of fracture. The concept of work of fracture can be used
as a qualitative measure of toughness. It should be pointed out, however, that
the concept of work of fracture has value only as a comparative tool; it is not
a design parameter.
Bowen (1968) has cited a 1961 report by Evans et aI. (1961) wherein it was
pointed out that a practical way to obtain an enhanced toughness, i.e., an
increased work offracture in ceramics, may be via obtaining multiple cracking
such that no single crack could propagate unhindered and cause failure.
Tattersall and Tappin (1966) analyzed the work of fracture and observed that,
for a wide variety of materials, the work of fracture increased as the toughness
increased.
Most of the early work on fiber reinforced ceramic composites involved

carbon fibers in glass or glass-ceramic matrix composites. Some of the early
work involved the use of ductile, metallic wires as well as the exploitation of
phase transformations to obtain a composite structure. With the commercial
availability of fine-diameter Nicalon fiber yarn and large-diameter
monofilament silicon carbide fiber, silicon carbide/glass and silicon
carbide/glass-ceramic systems were studied. Both these fibers provide carbon
at the fiber surface, resulting in a weak interface with most ceramic matrix
materials. In particular, lithium aluminosilcate (LAS) glass-ceramic matrix
containing silicon carbide fibers were found to provide very superior high
temperature capability and toughness. Almost parallel to this glass-ceramic
matrix work, researchers were examining silicon carbide/silicon carbide and
silicon carbide/silicon nitride systems and carbon/carbon systems. All of these
nonoxide systems have problems in oxidizing atmospheres at high
temperatures. This led to some work on oxide fiber/oxide matrix composites,
which are inherently stable in oxidizing environments. We provide a summary
of some of this work.
Crivelli-Visconti and Cooper (1969) showed that reinforcement of vitreous

silica by carbon fiber could substantially increase the work offracture vis a vis
the unreinforced matrix material, i.e., one could use long fibers to increase the
toughness of a brittle ceramic. They reported flexural strengths as high as 380
MPa for carbon fiber reinforced vitreous silica. There was no evidence of any
decrease in strength up to 800°C. More importantly, however, they reported
that the work of fracture (WOF) obtained from the area under the force-
displacement curve was 11 kJm- 2 compared with 4 Jm- 2 for the matrix alone.
Other results of various ceramic matrices reinforced with carbon fibers were
reported soon after. Pyrex glass reinforced with (high stiffness and high
strength) carbon fibers was investigated by Phillips and coworkers (Phillips,
1972; Sambell et aI., 1972; Phillips et aI., 1972; Phillips, 1974). Strengths
measured in 4-point-bend tests gave values of 459 MPa for composite
reinforced with high stiffness and 575 MPa for composites reinforced with
high strength fibers. Interlaminar shear strength values of71 and 32 MPa were
obtained for composites reinforced with high stiffness and high strength
carbon fibers, respectively.
Zwissler et al. (1977) studied the mechanical behavior of a composite made by

hot pressing ofwustite and stainless steel wires. The fracture strength as well
as the critical stress intensity factor for crack initiation increased linearly with
the volume fraction of stainless steel wire. The Young's modulus in the
longitudinal direction also increased as a function of the steel fiber volume
fraction. The improvements in the work of fracture and the critical stress
intensity for crack initiation were attributed to crack bridging action of the
wires. The bridging action exerts a closure force on the crack and reduces the
applied stress intensity at the crack tip.
Various glass matrix compositions reinforced with carbon fibers have been
shown to have strength and modulus increase with fiber volume fraction, in
accord with the rule-of-mixtures type behavior (Prewo et aI., 1986; Dawson
et aI., 1987). An example of such a linear increase in the strength of the
composite as a function of fiber volume fraction was observed by Dawson et
al. (1987) in Nicalon fiber reinforced borosilicate glass, Fig. 7.11. A linear
increase in strength of unidirectionally reinforced carbon fiber/glass
composites with fiber volume fraction Ve up to about 55% was observed
(Phillips et aI., 1972), Fig. 7.12. Beyond 55% Vf> increased matrix porosity led
to a drop in strength. The Young's modulus also increased linearly with Ve,
Fig. 7.13, but at higher Vr it deviated from linearity owing to matrix porosity
and possible fiber misalignment (Phillips et aI., 1972). The dashed line in Fig.
7.13 corresponds to the rule-of-mixtures prediction. The experimental values
are consistently below the rule-of-mixtures prediction line because of the
porosity in the glass matrix. Phillips et al. (1972) made a correction for the
effect of matrix porosity on its Young's modulus by using the following
expression
where Em is the modulus of the matrix with pores, Emo is the modulus of the
matrix with zero porosity, and P is the porosity in the matrix. Application of
this correction to the measured modulus values gives a better agreement with
the rule-of-mixtures prediction.
A composite consisting ofNicalon fiber yarn reinforced glass-ceramic matrix

(e.g., lithium aluminosilicate - LAS) has been shown to be a very promising
system. A glass-ceramic matrix has the advantage of easy processing in the
glassy state, high theoretical density and virtually no mechanical damage to the
fibers, followed by conversion to a crystalline ceramic. The flexural strength
as a function of temperature in argon for unidirectional Nicalon reinforced
lithium alumino silicate matrix materials showed an increase in strength before

dropping off due to softening of the matrix (Prewo et aI., 1986). This peak in
strength has been attributed to the ability of soft glass or glass-ceramic matrix
to distribute the stresses more uniformly. Eventually, at very high
temperatures, the matrix becomes too soft to sustain any load.
300r---------------------~--~~--~
250
~200
~
~ 150
(jj
100
50
0.1 0.2 0.3 0.4 0.5 0.6

Strain (%)
Fig 7.10 A schematic tensile stress-strain curve of a fiber reinforced composite.
3000
2500 Rule of Mixtures ./
~
./
0 ./
~ 2000
..
./
./
./
... .-
=5 1500 ./
01 I'
c ./
Q)
L 1000
-+--'
(f) .~.,..
500 ./
./
./
B.o 1.0
Fig. 7.11 Linear increase in strength of the composite as a function of fiber volume
fraction in Nicalon fiber-reinforced borosilicate glass (after Dawson et aI., 1987).
60 80 100
Fig. 7.12 Linear increase in strength of a unidirectionallyreinforced carbon fiber/glass

composites with fiber volume fraction Vrup to about 55% [Davidge and Briggs, 1989].
Beyond 55% Vr, high matrix porosity caused a less increase than expected from the
linear line.
Lee and Daniel (1992) used a modified shear lag analysis to predict stress-
strain and stress vs. matrix crack density curves and obtained a reasonable
agreement with experimental results for silicon carbide/calcium aluminosilcate
glass-ceramic composite subjected to monotonic loading.
Silicon carbide whisker reinforced alumina composites show a good

combination of mechanical and thermal properties. Work by Becher and
coworkers (Wei and Becher, 1985; Becher and Tiegs, 1988; Becher et aI.,
1988) and others (Homeny et aI., 1987; Porter et aI., 1987) has shown these
composites can have substantially improved strength, fracture toughness,
thermal shock resistance, and high temperature creep resistance over that of
monolithic alumina. Figure 7.14 gives an example of strength improvement in
the silicon carbide whisker/alumina composites as a function of the whisker
volume fraction and test temperature. Similar results have been obtained with
silicon carbide whisker reinforced silicon nitride composites (Shih et aI.,
1992). A combination of reaction bonding and hot pressing was used to obtain
fully dense SiCjSi3N 4 composites with improved flexural strength compared
to the unreinforced silicon nitride. Not unexpectedly, these authors observed
Rule of Mixtures
'\v/
300 -
.
/
/
/-'0
p
200 -
/.
/-6
If •
/~. • Measured
100 - /
//
o Calculated,
Corrected for
Matrix Porosity
I I I
o 10 20 30 40 50 60 70 80 90 100
Vr. %
Fig. 7.13 Linear increase in Young's modulus with percentage fiber volume in a
unidirectional, continuous carbon fiber/glass composite (after Davidge and Briggs,
1989).
that the stability of SiC whiskers as well as fracture toughness depended on the
amount of additives used in the processing.
Theoretical models to explain the experimental observations in terms of a

variety of deformation mechanisms have been proposed by various
researchers, for example, Wei and Becher, 1985; Becher and Tiegs, 1988;
Becher et aI., 1988), Aveston et aI., 1971; and Kerans and Parthasarathy, 1991.
They attributed the increase in toughness or work of fracture to crack bridging
and fiber pullout. Crack bridging results in closure stresses on the crack faces
while the frictional energy is dissipated as the loosely gripped fibers are pulled
out of the matrix. We shall discuss this topic of toughness in some detail in
chapter 9.
7.3 FATIGUE
There is some confusion in the literature about the terminology of fatigue. In

this book, we define fatigue as the phenomenon of mechanical property
degradation leading to failure of a material or a component under cyclic
loading. The operative word in this definition is cyclic. This definition thus
0
a... 1000
~
I
I- 800
(9
z
W
0:
I- 600
(j)
W C 10 vol %
0:
=>
l-
400
e. 5 vol % I ±~~
u I i l
c::r
0:
LL
200 ---""'1
0 400 800 1200
TEST TEMPERATURE ( 0 C)
Fig. 7.14 Strength increase in silicon carbide whisker/alumina composites as a
function of the whisker volume fraction and test temperature.
excludes the so-called phenomenon of static fatigue which is sometimes used

to describe stress corrosion cracking in glasses and ceramics in the presence
of moisture. The term dynamic fatigue is frequently used to describe strength
as a function ofloading rate. Perhaps one way avoid all this confusion is to use
the term cyclic fatigue to describe the effects of cyclic loading (mechanical,
thermal, or electrical). Understanding cyclic fatigue behavior of composites
of all kinds is of great importance, because without such an understanding it
would be virtually impossible to gain acceptance of the design engineers.
Many high volume applications of composite materials involve cyclic loading
situations, e.g., automobile components. It would be a fair admission that this
understanding of the fatigue behavior of composites has lagged that of other
aspects such as the elastic stiffness or strength. There are many reasons for this
situation. Conventional wisdom had it that cyclic fatigue was unimportant so
far as ceramics were concerned. It turns out that the problem of cyclic fatigue
in ceramics and ceramic matrix composites is an important one, as we show
below. Engineers and researchers began to appreciate the importance of cyclic
fatigue in ceramics and ceramic matrix composites only in 1970s. The fracture
resistance of CMCs under cyclic conditions needs to be evaluated for design
in a variety of potential structural applications. For example, it is not unusual
to have a design requirement for a ceramic component in an automotive gas
turbine to withstand over 30,000 cycles offatigue (low cycle fatigue) (Helms
and Haley, 1989). Yet another problem in this regard is that the application of
conventional approaches to analyze the fatigue behavior of composites, for
example, the stress vs. cycles (S-N) curves or the application oflinear elastic
fracture mechanics (LEFM), is not straightforward. This problem is, of course,
common to all composites. The main reasons for this are the inherent
heterogeneity and anisotropic nature of the composites, especially in fiber
reinforced composites. These inherent characteristics lead to damage
mechanisms in composites that are very different from those encountered in
conventional, homogeneous, or monolithic materials.
7.3.1 Conventional Approaches
(aJ S-N Curves
A very popular conventional fatigue testing technique used with metals

involves determination of the so-called S-N curves, where S is the stress
amplitude and N is the number of cycles to failure, Fig. 7.15. In general, for
ferrous metals, one obtains a fatigue limit or endurance limit, curve A in Fig.
7.15. For stress levels below this endurance limit, theoretically, the material
can be cycled indefinitely. If, however, such an endurance limit does not exist,
for example for nonferrous metals (curve B in Fig. 7.15), then one can
arbitrarily define a certain number of cycles, say 106, as the cutoff value. We
define some useful parameters for our discussion of the fatigue phenomenon
in Fig. 7.16:
cyclic stress range, .10" = O"max - O"min

cyclic stress amplitude, O"a = (O"max - O"min) /2
mean stress, O"m = (O"max + O"min) / 2
stress ratio, R = O"min / O"max
The fatigue behavior of a material can be conveniently divided into two stages:
crack initiation and propagation. Quite frequently, a rule-of-thumb approach
is used, viz., increase the monotonic strength of the material which
concomitantly results in an increase in its cyclic strength. This rule-of-thumb
assumes that the ratio of fatigue strength/tensile strength is fairly constant. It
is generally true that the maximum efficiency in terms of stiffness and strength
gains in fiber reinforced composites occurs when the fibers are continuous,
uniaxially aligned and the properties are measured parallel to the fiber
direction. As we go off-angle, the role ofthe matrix becomes more important.
One major drawback of studies offatigue behavior of a material using this S-N
~
~
I-
~
lIE
c.
::l L
.
\oJ
II:
I-
tor. to· 101
CYCLES TO FAILURE (N, I
Fig. 7.15 S-N curve from conventional fatigue testing. S is the stress amplitude and
N is the number of cycles to failure. Curve A has an endurance or fatigue limit SL
while curve B does not.
approach is that no distinction can be made between the crack initiation phase
and the crack propagation phase.
(b) Fatigue Crack Propagation Tests
Such tests are generally conducted in an electrohydraulic closed-loop testing

machine on notched samples. The results are presented as
log (da/dN) vs. log ~K
where da/dN is the crack growth per cycle and ~K is the alternating stress
intensity factor.
CTmOK
U)
U)
I.&J
...a:
U) 0 TIME
O"min
Fig. 7.16 Some important parameters in the study of cyclic fatigue.

Crack growth rate, da/dN, is related to the cyclic stress intensity factor range,
L\K, according to the power law relationship (Paris, Gomez, and Anderson,
1961; Paris and Erdogan, 1963):
daldN = A(L\K)m
where A and m depend on the material and test conditions. The applied cyclic
stress intensity range is given by
L\K = YL\aJ"a
where Y is a geometric factor, L\a is the cyclic stress range, and a is the crack
length.
The major problem in this kind of test is to make sure that there is one and
only one dominant crack that is propagating. It was mentioned above that the
complexities in composites lead to the presence of many modes of damage,
such as matrix cracking, fiber/matrix debonding, crack deflection along the
interface leading to multidirectional cracking, ply delamination in the case of
a laminated composite, fiber fracture, etc. Some of these modes appear rather
early in the fatigue life of composites. It thus becomes important to have a tool
to characterize such a complex state of damage. Progressive loss of stiffness
has been found to be a very important and useful characteristic that can be
used to study the fatigue of polymeric laminate composites. Schematics of the
different types of damage zones formed in an isotropic material (e.g., a metal,
or a ceramic, or a polymer) and a fiber reinforced composite, which is an
anisotropic material, are shown in Figs. 7.17 and 7.IS, respectively. In the case
of an isotropic material, a single crack propagates in a direction perpendicular
to the cyclic loading axis (mode I loading), see Fig. 7.17. In a fiber reinforced
composite, on the other hand, a variety of subcritical damage mechanisms lead
to a highly diffuse damage zone as shown in Fig. 7.1S (a,b,c).
A comparison of damage accumulation as a function of cycles in a composite

and in a monolithic material is shown in Fig. 7.19. In a monolithic material,
with a single dominant crack propagating in a self-similar manner, the damage
increases at a slow rate in the beginning and then accelerates as the number
of cycles or cycle ratio increases. In a fiber reinforced composite, especially
in a laminate consisting of plies containing fibers at 0° and 90°, the damage
increases at a fast rate in the beginning when the cracking in the matrix in 90°
plies occurs quickly in the early stages of cycling. The damage then tapers off
and we get a plateau region before final failure. Two possible routes of damage
Fig. 7.17 Damage zone in a conventional, homogeneous, monolithic material

(isotropic).
in the composite are shown in Fig. 7.19. We come back to this problem later
in this section. First we describe some results obtained with CMCs using
conventional approaches to fatigue.
Prewo and Brennan (1980) and Brennan and Prewo (1982) studied the tensile
fatigue behavior of Nicalon-type silicon carbide fiber reinforced lithium
aluminosilicate (LAS) glass-ceramic composite. They used two different types
::::
.... ::::
.:.: ~:~:
:.:. :::: ::::
:::: =:!: ;~~~
::~: ::::
:.:.
.:.: :jj
~~~
~ 1~ ~ ~
~~ m~ -0:
Fig. 7.18 Diffuse damage zone in a fiber reinforced composite (anisotropic): (a) fiber
breakage and local debonding, (b) matrix cracking, deflection of the principal crack
along a weak fiber/matrix interface and (c) deflection of the principal crack.
Q)
.~ Composites
til
~
~ Metals
Q ({f---In-iti-ati-on---+--p~ropagation
Initial
imperfections
Life
(cycles or time)
Fig. 7.19 Comparison of damage accumulation in composite laminates and in

monolithic materials.
of LAS as the matrix material. Differences in processing led to a strong

interfacial bond in Nicalon/LAS I and a weak interfacial bond in NicaloniLAS
II. The composite with LAS I matrix showed a linear tensile stress-strain curve
to failure while the composite with LAS II showed a markedly nonlinear
behavior due to extensive matrix cracking prior to ultimate failure. It was
observed that the level of tensile stress at which the inelastic behavior
(proportional limit) of composite began had an important bearing on the
fatigue behavior of the CMC. The residual tensile strength and elastic modulus
of the LAS I composite after fatigue was the same as that of the as-fabricated
composite. In the LAS II composite, cycling below the proportional limit
produced the same result. However, on cycling to stress levels higher than the
proportional limit, a second linear stress-strain region having a modulus less
than the initial modulus was observed. Presumably, this change in behavior
was due to matrix microcracking at stresses above the proportional limit.
Wang et al. (1991) investigated the behavior of a carbon fiber reinforced

silicon carbide composite, cross-plied, i.e., fibers arranged in layers at 0 and
90°, under cyclic loading. They used tension-tension loading of smooth and
notched samples and compression-compression loading. Damage in pulsating
tension consisted of cumulative microcracking and spalling. Tensile cyclic
creep was observed in notched specimens. Smooth specimens showed less
creep and fewer surface cracks and spalling than the notched samples. The
authors attributed this to fewer imperfections in the smooth specimen before
cycling. Pulsating compression fatigue also caused creep and the damage
involved edge delaminations. This work shows that, as expected, cyclic
loading causes cumulative damage and thus will be expected to degrade the
inelastic characteristics of the composite.
As we said above, it was generally thought that the phenomenon of cyclic

fatigue was unimportant in ceramics. Work done by Suresh and coworkers
(Suresh, 1991; Suresh et aI., 1988; Han and Suresh, 1989) on fatigue crack
growth in a variety of brittle solids in compression, tension, and tension-
compression fatigue shows that mechanical fatigue effects, i.e., due to cyclic
loading, occur at room temperature in brittle solids as well. Suresh et al.
(1988) suggest that a variety ofmechanisms such as microcracking, dislocation
plasticity, stress- or strain-induced phase transformations, interfacial slip, and
creep cavitation, etc. can promote an inelastic constitutive response in brittle
solids of all kinds under compressive cycling. Particularly, in CMCs, the
deformation mechanisms operating near the crack-tip differ significantly
under static and cyclic loading. Under pulsating compression, nucleation and
growth of stable fatigue cracks occurs even at room temperature. For example,
Suresh et al. observed that cyclic compressive loading induced mode I fatigue
crack growth in SiC whiskers/Si 3N4 matrix composites. They also observed
whisker pullout and breakage after fatigue cycling. Such behavior is generally
not observed under monotonic loading. This mode I fatigue crack growth
under far-field cyclic compression occurs because a residual zone of tensile
stress is generated at the crack tip on unloading.
We have mentioned that one of the main problems in high temperature

cracking of polycrystalline ceramics is the presence of intergranular glassy
phases. Sintering and other processing aids can form glassy phases at the
boundaries which can result in rather conspicuous sub critical crack growth.
Such subcritical crack growth can become very important in ceramic matrix
composites because whiskers such silicon carbide can undergo oxidation.
Han and Suresh (1989) examined the tensile cyclic fatigue crack growth in
a silicon carbide whisker (33 voI.%)/alumina composite at 1400 °C and
above. The composite showed subcritical fatigue crack growth at stress
intensity values far below the fracture toughness. The fatigue behavior was
characterized by the cyclic stress intensity factor, stress ratio, and frequency.
Examination of the region near crack tip by optical and transmission electron
microscopy showed that the nucleation and growth offlaws at interface was
the main damage mechanism. Figure 7.20 shows the microcrack zone ahead
of the main tensile crack in this composite after unloading from a peak stress
at 1500 DC. Han and Suresh observed oxidation of silicon carbide whiskers
to a silica type glassy phase in the crack tip region at 1400 DC, in air. The
alumina matrix can react with the main oxidation product, viz., silica to form
aluminosilicates, SiC-rich or stoichiometric mullite, etc. Viscous flow of
glass can result in interfacial debonding, followed by the nucleation, growth,
and coalescence of cavities. Figure 7.21 shows the formation pockets of
Fig. 7.20 A fatigue test temperature at 1500 °C resulted in a significant size of the damage zone at the tensile crack tip. (Courtesy
of S. Suresh)
239 MECHANICAL BEHAVlOR
glass at the whisker tips (white regions marked by arrows) in the SiCjAlP3
composite subjected to a constant load fracture at 1400 °C in air. More
information about such oxidation of SiC whiskers is provided in the next
section on creep. What we wish to emphasize here is that there is a
difference in deformation and failure mechanisms under static and cyclic
loadings. When the composite shown in Fig. 7.21 was subjected to cyclic
loading, such large scale oxidation of whiskers was not observed. Instead,
cyclic loading resulted in multiple cracking of whiskers and the formation
of glass at these crack sites, see Fig. 7.22. A careful examination of Fig. 7.22
shows the meniscus of the molten glassy phase that has flowed inside the
cracked whisker.
An interesting phenomenon that comes into play in high cycle fatigue of

CMCs is frictional heating. Repeated friction and sliding between fiber and
matrix can lead to a temperature rise (Holmes and Cho, 1992; Shuler et aI.,
1993). Figure 7.23 shows the temperature rise as a function of number of
cycles in Nicalon fiber/calcium aluminosilicate (CAS) matrix composites,
cycled at 200 Hz, under dry and lubricated conditions (Serensen and
Holmes, 1995). Oil was used as a lubricant. Under lubrication we have
reduced interfacial friction, which results in larger a number of cycles being
needed to attain the same temperature rise as under dry conditions.
7.3.3 Novel Approaches
Incorporation of fibers can be expected to improve the fatigue strength of

fiber reinforced composites in the fiber direction. Fibers, aligned along the
stress axis and in large volume fractions, will show high monotonic strength
values which are translated into high cyclic or dynamic fatigue strength
values as well. Typically, the S-N curves of such composites are very flat.
This, however, does not work very well for highly anisotropic fibrous
composites, especially for laminated composites made by stacking plies of
different orientations. Here, we can learn some useful lessons from some of
the work done in the area of fatigue of polymer matrix composites. One
technique that has been used in PMCs to study the evolution damage under
cyclic conditions is that of stiffness reduction of a composite as a function
of fatigue cycles. This is nothing but what has been termed damage
mechanics by some researchers. What one does in this approach is to use the
loss of stiffness as a measure of damage in the composite. Consider again
Fig. 7.19 which compares, schematically, damage accumulation as a
function of cycles in a composite and in a monolithic, homogeneous material
under constant stress amplitude fatigue. Because of a multiplicity of
Fig. 7.21 Formation of glass pockets at the whisker tips (white regions marked by
arrows) in the SiCjAlz03 composite subjected to a constant load fracture at 1400
°C in air. (Courtesy of S. Suresh).
Fig. 7.22 When the composite shown in Fig. 7.21 was subjected to cyclic loading,
large scale oxidation of whiskers was not observed. Instead, multiple cracking of
whiskers and the formation of glass at these crack sites occurred. Note the meniscus
of the molten glassy phase that has flowed inside the crackedwhisker. (Courtesy of
S. Suresh).
Nt (cycles to failure)
Fig. 7.23 Temperature rise as a function of number of cycles in Nicalon

fiber/calcium aluminosilicate (CAS) matrix composites, cycled at 200 Hz, under dry
and lubricated conditions (after SjiSrensen and Holmes, 1995).
cracking modes possible in composites, it is convenient to talk in terms of a

damage parameter rather than the length of a dominant crack. In Fig. 7.19 we
plot damage ratio against cycle ratio. The damage ratio is the current damage
normalized with respect to the damage at final failure. The cycle ratio,
similarly, is the number of cycles at a given instant divided by the number of
cycles to failure. In a homogeneous material, the term damage simply
represents the crack length, and not surprisingly it increases monotonically
with cycling. In the case of a fiber composite, we do not have a simple and
unambiguous manifestation of damage, such as a crack length. Instead,
damage implies the crack density. Note that, unlike in homogeneous
materials, the damage in a composite accelerates at first and then decelerates
with cycling. In Fig. 7.19, we show two possible routes for damage
accumulation in CMCs. This distinctive behavior is very important. As was
pointed out above, a fiber reinforced composite, especially in the laminated
form, can sustain a variety of subcritical damage (cracking of matrix,
fiber/matrix decohesion, fiber fracture, ply cracking, delamination, etc.). For
example, in a [0/90] laminate, the 90 degree ply will crack first. Such
cracking of a ply will result in a relaxation of stress in that ply and with
continued cycling no further cracking occurs in that ply. Ply cracking could
involve cracking in the matrix and maybe along the fiber/matrix interface.
Other damage accumulating mechanisms include the growth of existing
cracks into interfaces leading to ply delamination. The delamination of a ply
results in a reduction of stress concentration on the neighboring plies. As can
be seen, the subcritical damage can accumulate rather rapidly on cycling.
The various types of subcritical damage mentioned above result in a reduction

of the load-carrying capacity of the composite, which in tum manifests itself
as a reduction of stiffness and strength of the composite (Hahn and Kim,
1976; Highsmith and Reifsnider, 1982; Talreja, 1985; Ogin et aI., 1985;
Johnson, 1989). Work with polymer- and metal-matrix composites shows that
this change in stiffness values is a good indicator of the extent of damage in

these composites. Figure 7.24 shows an example of stiffness loss as a function
of number of cycles from the work of Ramakrishnan and Jayaraman (1993)
wherein they studied the cyclic fatigue behavior of Nicalon fiberlborocilicate
glass-ceramic matrix composites. The matrix, having a lower strain-to-failure
than the fibers, starts cracking first. Thereafter, we have two possible routes.
In the case of a weak interface, we observe fiber/matrix debonding followed
by fiber pullout, etc. In the case of a strong interface, matrix cracking leads
to failure and a brittle failure of the composite.
The modulus reduction rate (-lIEo)(dE/dN) at a given value of ElEo, is

tangent to this curve at that value of ElEo where E is the current modulus, Eo
is the initial modulus of the un cracked material, and N is the number of
cycles. Karandikar and Chou (1992) used this approach with unidirectionally
reinforced Nicalon fiberlcalcium aluminosilicate (CAS) composites and
obtained correlations between crack density and stiffness reduction.
7.4 CREEP
Creep is time-dependent strain, which becomes important at high

temperatures. The basic governing equation for creep can written in the
following form
'i!E.'"
1'01
....
::
IS
.,.,
0.9
"
~
tll
0.8 O!---!-.50~--:'1OO!:-:---:1~50~~2~OO:::--"::"25~O~~300
N (cycles)
Fig. 7.24 Schematic of the progression of cyclic fatigue damage in under two
different interfacial conditions (after Ramakrishnan and layaraman,1993).
_ = A( cr/G)n (b/d)ffi exp (-~QIkT)

where e is the creep strain rate, cr is the applied strain, n is an exponent, G

is the shear modulus, b is the Burgers vector, d is the grain size, AQ is the
activation energy for creep, k is the Boltzmann's constant, and T is the
temperature in kelvin. Creep sets a limit on the maximum temperature at
which a material can be used. In general, this limit increases with the
melting point of a material. An approximate limit can be estimated to lie
about half of the melting temperature in kelvin.
Creep of ceramic materials has been investigated quite extensively, for

example, see the review papers by Routbort et al. (1990) and Lin and Becher
(1990). The stress exponent, n, typically varies between 3 and 5 in the
dislocation climb regime and between I and 2 when diffusional mechanisms
are operating. Pure dislocation creep occurs commonly in single crystal
ceramics while in phenomena such as grain boundary sliding, vacancy
motion in grains and in the grain boundaries, and dislocation can cause creep
in polycrystalline ceramics. The applied stress, grain size, porosity, and
impurity content are important variables.
A problem that is common to perhaps all compoistes has to do with the

mismatch in creep strength of the fiber and the matrix. Assuming a well
bonded interface, then the axial strains in the fiber and matrix will be equal.
Under a constant load creep condition, a redistribution of axial stresses will
occur. Say the matrix has a lower creep strength than the fiber, the stress in
the matrix will decrease and that in the fiber will increase until the creep
rates of the two become equal.
The basic equation given above describing the creep behavior is independent
of the sign of the applied stress, i.e., it is immaterial whether tensile and
compressive stress is applied. Wiederhorn and Hockey (1991) have analyzed
the creep behavior ofCMCs, both particle and whisker reinforced. In two-
phase ceramics, creep rate in tension is faster than in compression for
identical stress and temperature conditions. At first sight, this would appear
to be attributable to the ease of cavitation and microcracking during tension
than in compression because tensile stresses assist cavitation while
compressive stresses tend to close the cavities and microcracks. However,
the results of Wiederhorn and Hockey's (1992) on creep of siliconized
silicon carbide in tension and compression at 1300°C were unusual. Their
siliconized SiC composite consisted of SiC (about 67% by volume) in a
matrix of silicon. It showed a highly asymmetric creep behavior between
tension and compression, much more creep in tension than in compression.
As we said above, such asymmetry in creep behavior has been observed in
many two-phase ceramics. In single phase ceramics the asymmetry vanishes.
The interesting observation of Wiederorn and Hockey was that in their
siliconized SiC composite cavitation did not occur at low tensile stresses so
the cavitation alone could not explain the creep asymmetry in their material.
They used a model from the area of soil mechanics to explain their results.
As per this model, sand with 38% voids shows dilatant behavior (expands
by about 3%) during the initial stages of axial strain. Their siliconized SiC
had 33% silicon by volume, which Wiederhom and Hockey treated as voids.
In short, silicon flows but the movement of packets of SiC grains determines
the creep rate; such movement of packets of grains is more difficult in
compression than in tension.
Figure 7.25 shows the creep behavior, in four-point bending, of silicon

carbide whisker (20 vol. %) reinforced alumina and unreinforced alumina
(Lin and Becher, 1990). The figure shows the creep data in the form of creep
strain rate as a function of stress at 1200, 1300, and 1400 °C. The creep rate
of the composite was significantly reduced compared to the unreinforced
alumina. For the creep tests done at 1200 and 1300 °C, the stress exponent,
n in the relationship E A((jt, is 2 for the composite, not much different from
the value of 2.3 for the unreinforced alumina, indicating thereby that the
creep rate controlling process is similar in these two materials at these two
temperatures. This improvement is thought to be due to a retardation of grain
boundary sliding by SiC whiskers present at the grain boundaries. The creep
245 MECHANICAL BElIAVIOR
creep curve at 1400 °C for the composites shows a marked change in the n
value at a stress level of about 125 MPa indicating a change in the rate
controlling process. A stress exponent value of about 2 is generally thought
to be due to grain boundary sliding. The higher stress exponent and the
higher creep rate at 1400 °C was attributed by the authors to extensive
cavitation.
Although continuous ceramic fibers can lead to substantial toughening of

ceramics at room temperature, most of these fibers are not sufficiently creep
resistant. In fact, creep rates of many fibers are much higher than those of
the corresponding monolithic ceramics because of extremely small grain
10- 5
1200°C
10- 6
I
(J)
~ 1300°C
10- 7
4-Point
Flexure in air
10- 9
50 100 200 300 400
.a(MPa)
Fig. 7.25 Creep strain rate as a function of stress at 1200, 1300, and 1400 °C. For
tests at 1200 and 1300 °C, creep resistance is improved due to a retardation of grain
boundary sliding by SiCw present at the grain boundaries. The creep curve at 1400
°C for the composites shows a marked change in n at about 125 MPa indicating a
change in the rate controlling process (after Lin and Becher, 1990).
size, presence of glassy phases, etc. (Routbort et aI, 1990; Lin and Becher,
1990; Pysher et aI., 1989; Bender et aI., 1991). In the case of creep of
ceramic matrix composites, one needs to consider the intrinsic creep
resistance of the fiber, matrix, and interface region. We have described the
creep characteristics of various ceramic fibers in chapter 3. Oxide fibers are
fine-grained and generally contain some glassy phase. Nonoxide fibers are
also fine grained, multiphasic (with some glassy phase) and susceptible to
oxidation (Lessing and Gordon, 1977). Nonoxide fiber/nonoxide matrix
composites, such as SiC fiber/SiC and SiC fiber/Si3N4' generally show good
low temperature strength, but oxidation resistance is the major limitation.
Strength ofNicalon type SiC fiber is very sensitive to temperature above
1200 °C and its environment (Mah et aI., 1984). Nonoxide fiber/oxide matrix
composites or oxide fiber/nonoxide matrix composites, such as carbon/glass,
SiC/glass, SiC/alumina and AI20/SiC, generally do not possess high
oxidation resistance because the permeability constant for the diffusion of
oxygen is high, resulting in rapid oxygen permeation through the oxide
matrix (Mah et aI., 1984; Okamura, 1987; Hermes and Kerans, 1989). Prewo
et aI. (1986) observed that the glass matrix did not prevent the degradation
of carbon fiber caused by oxidation. When hot pressed SCS-6 fiber/Si3N4
matrix composite was subjected to creep, a complex glassy layer (Si02 -
Y 20 3-MgO) formed on the surface (Holmes, 1991). Figure 7.26 shows an
example of glass layer on sample subjected to creep at 1350 °C and at 110
MPa. The glass layer formed by oxidation of silicon nitride and sintering
aids. Holmes also observed that the extent of fiber pullout decreased as the
applied stress increased in a creep test. Rather pronounced separation along
the fiber/matrix interface was observed after low stress (70 MPa) creep.
Figure 7.27 shows an example of such debonding. Severe degradation at
temperatures above 1000 °C has been observed in many nonoxide systems,
for example, in a CVI-processed silicon carbide fiber reinforced SiC matrix
due to the evolution of CO, SiO, and other gases (Lin and Yang, 1991).
Creep behavior of laminated ceramic composites has also been studied.

Figure 7.28 compares the steady state creep rate of monolithic silicon
nitride matrix [0] and [0/90] cross-plied SCS-6 fiber/Si3N4 composites as a
function of applied stress at 1200 °C (Yang and Chen, 1992). Clearly, the
creep resistance of the composite is superior to that of the monolithic silicon
nitride, while the creep resistance of the unidirectional composite is superior
247 MECHANICAL BERAVIOR
Fig. 7.26 FOlmation ofa glass layer on sample of SiCI Si3N4 composite to subjected
to creep at 1350 °C and at 110 MPa. The glass layer forms by oxidation of silicon
nitride and sintering aids (courtesy of lW. Holmes).
7.27 Separation along the SCS-6 fiberlSi 3N4 matrix interface after low stress (70
MPa) creep (Courtesy of J.W, Holmes).
to that of the cross-plied composite because the fibers in the 90° direction
contribute less to creep resistance than the fibers in the 0° direction.
From the above discussion of high temperature behavior of non oxide

composites (even when only one component is a nonoxide), it would appear
that in situations where stability in air at high temperatures is a prime
objective, oxide fiber/oxide matrix composites should be most promising
because of their inherent stability in air. Among such systems of composites
are the following. Alumina-based and mullite fibers are the most widely
used oxide fibers, while glass, glass-ceramics, alumina, and mullite are the
most widely used oxide matrices. There are two categories of oxide/oxide
composites: an oxide matrix reinforced with uncoated oxide fibers and an
oxide matrix reinforced with coated oxide fibers.
In the first category of oxide/oxide composites, strength and modulus ofthe

composite are generally improved compared to the unreinforced oxide. The
toughness characteristics ofthese composites are not substantially changed
because of the strong chemical bonding at the fiber/matrix interface.
Interface tailoring via fiber coating (the second category) is employed
-2.0
g m
-2.6
-3.2
·W o
bIl -3.8 ~o
.3 0 o
[0/90J
o
-4.4 o 00 [OJ
o
-5.0
-s·hs 2.75
Log 0'
Fig. 7.28 Steady state creep rate of monolithic silicon nitride matrix, [0] and [0/90]
cross-plied SCS-6 fiber/Si3N4 composites as a function of applied stress at 1200 °C
(Yang and Chen, 1992).Creep resistance of the composite is superior to that of the
monolithic silicon nitride, while the creep resistance ofthe unidirectional composite
is superior to that of the cross-plied composite.
extensively in order to achieve the desired properties of the composites; we

discuss this topic in chapter 9.
7.4.1 General Remarks on High Temperature Capability of CMCs
Creep resistance of alumina and mullite can be improved significantly by

incorporation of SiC whiskers, Fig. 7.29 (Nixon et aI., 1987; Chokshi and
Porter, 1985). In this regard, the purity and stoichiometry of silicon carbide
becomes very important. Figure 7.30 shows an example of rice-hull derived
silicon carbide whisker/alumina composite made by hot pressing and then
exposed to 1200 °C. This figure shows that these silicon carbide whiskers
get oxidized at and near the surface. Figure 7.31 shows the same sample as
in Fig. 7.30 after etching. The etchant removed the oxidized part of the
whisker, leavingthe remnants of SiCw with a reduced cross section and
irregular shape. It is known that rice-hull derived silicon carbide whiskers
as well as Nicalon type continuous silicon carbide fibers that are not the
purest form of silicon carbide tend to oxidize when exposed to air at
temperatures as low as 1000 °C. Under such circumstances, oxide
composites, i.e., oxide matrix containing and oxide reinforcement, would be
promising. Applications ofoxide/oxide composites at elevated temperatures,
10-3
10-4
.....
-
I
UJ 1Q"5
ai
e! 1Q-6
0-
Q)
~ 10-7
U
10-8
10-9
10 100
Stress, MPa
Fig. 7.29 Creep resistance of alumina and mullite can be improved significantly by
incorporation of SiC whiskers.
however, are limited by plastic flow under creep conditions. It has been
estimated that there are at least 38 oxides with their melting temperature
above 2000 °C (Routbort et aI., 1990). Commercially available oxide fibers
include polycrystalline alumina, mullite, and single crystal filament of
sapphire. Polycrystalline fibers show large strength reduction at 1200 °C and
Fig.7.30 Effect of exposure in air at 1200 °C on a rice-hull derived silicon carbide

whisker/alumina composite made by hot pressing. Silicon carbide whiskers get
oxidized at and near the surface. (Courtesy of A. Wereszczak).
Fig. 7.31 The same sample as in Fig. 7.30 after etching. The etchant removed the
oxidized part of the whisker, leaving the remnants of SiCw with a reduced cross
section and irregular shape. (Courtesy of A. Wereszczak).
almost a complete loss of strength at 1300 °C. Single crystal oxide fibers
such as alumina and YAG can withstand temperatures above 1300 °C.
Most ceramic oxides have rather poor creep resistance. This susceptibility
of oxides to creep can be limiting factor on the lifetime of oxide/oxide
composites at high temperatures. One can estimate the allowable creep rate
for an application by using a Monkman-Grant type curve (Parthasarathy et
aI., 1991), in which we plot the logarithm of creep rate as a function ofthe
logarithm of rupture time. If we assume a failure strain of 1% for ceramics,
then we can estimate the allowable creep rate for a given lifetime. Let us say
that we wish to have creep rates of 10- 8 S-1 or less for extended service
applications. Figure 7.32 shows creep data for some oxide matrix materials
(Warshaw and Norton, 1962; Penty and Hasselman, 1972; Evans, 1970;
Seltzer and Talty, 1975; Lessing et aI., 1975; Pysher and Tressler, 1992).
All data are plotted for an applied stress of 100 MPa. For a creep rate of
1O- 8s- 1 at a temperature above 1400 °C, it would appear that there is no
oxide matrix material capable of providing adequate creep strength. To
achieve an acceptable creep rate for extended service applications, a good
creep resistant oxide fiber would be required to reinforce a less creep
resistant matrix. Figure 7.33 shows the creep resistance data for some oxide
fibers (Jakus and TUlluri, 1989; Johnson et aI., 1987; Garboriaud, 1981;
Firestone and Heuer, 1976; Corman, 1991; Bullock et aI., 1977). Note that
polycrystalline oxide fibers possess limited creep resistance, while single
crystal fibers, sapphire and yttria-alumina-gamet (YAG), can maintain good
creep resistance above 1400 °C. The creep performance of a composite
system may be estimated as follows (Monkman and Grant, 1956). Assuming
the creep of fiber and matrix phases can be described by a common power
law expression, we can write
where cr denotes the stress, c denotes the strain, cr fo ' cfo' crmo ' and Cmo are
constants obtained from empirical stress-strain rate relationships, and
subscripts, f and m refer to the fiber and matrix, respectively. The strain rate,
cc' in the composite at a stress, crc, is weighted according to the volume
fraction of the two phases, V f and V m' assuming a fully bonded interface and
both fiber and matrix being creep at the same rate,
1Q-3
YAG.3"",
10-4
AI2O,.7"",
/
~ 10-5
1i ~ Mullite. 1.2 "'" ~
e!
~
f1O-S
()
10-7 /
5~
Y-PSZ. 41 "'"
MUlllt••
10-"
4.5 5.0 5.5 6.0 6.'5
Temperature.104/K
Fig. 7.32 Creep rate as function of inverse temperature for some oxide matrix
materials for an applied stress of 100 MPa. For a creep rate of 1O- 8/s at a
temperature above 1400 °C, it would appear that there is no oxide matrix material
capable of providing adequate creep strength.
Examples of two different oxide matrices reinforced by 30 vol% Y AG single

crystal are given in Table 7.l. The failure stress values, cr c ' for these two
composites at a creep rate at 10- 8s- 1 and at a temperature of 1500 °C are
predicted to be 34 and 84 MPa, respectively. This difference in the critical
stress for the same reinforcement with different matrices indicates that a
very creep resistant reinforcement and a good creep resistant matrix are
required to achieve desirable creep performance of the oxide/oxide
composites at high temperatures.
It is worth pointing out that properties of a single crystal are a function of its
orientation. This is true of single crystal fiber as well. Figure 7.34 compares
the creep rate as a function of inverse temperature for Y AG and alumina
single crystal fibers. Note that the YAG single crystal exhibits creep rate
dependence on crystal orientation than alumina single crystal. Rather large
differences occur in the flow stress for the three slip systems of an alumina
single crystal. In particular, an off-axis loading can significantly affect the
fiber and composite properties. It can be seen from Fig. 7.34, shown above,
that the flow stress of such a material at 00 orientation can be more than 10
times greater than that at an off-axis orientation at high temperature. As we
pointed out above, Y AG, a cubic compound, is less orientation dependent
than alumina.
Table 7.1 Examples of creep in oxide/oxide CMCs (assuming a fully bonded

interface).
Temperature: 1500 °C, Vr::30%

Creep rate: 10- 8 sol
Case 1
Fiber: YAG Single crystal, f= 116 (110)*
Matrix: Alumina Polycrystalline, m = 111.6 *
Stress = 34 MPa
Case 2
Fiber YAG Single crystal, f= 116 (110)*

Matrix: Alumina Single crystal, m = 114.5 (0001)*
Stress = 84 MPa
*Corman (1991) **Chokshi and Porter (1985)
10"- 100 MPa
10-5
.-
m 10-6
.sf
~
C.
Q) 10-7
~
()
10-8 Sapphire (0001)
I
10-9
4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0

Temperature. 104/K
Fig. 7.33 Polycrystalline oxide fibers have a lower creep resistance than single
crystal fibers, sapphire and yttria-alumina-garnet (YAG).
Temperature. °C
1800 1500 1250 1000
\ Sapphire (42°)
\
YAG (100) \
\
\
Sapphire (T 0001) \
I' \
\
\
VAG (110)
&YAG (111) 100 MPa

10-'
4.5 5.5 6.5 7.5
104 jTemperature. K
Fig. 7.34 A comparison of creep rate as a function of inverse temperature for YAG
and alumina single crystal fibers. Note that the YAG single crystal exhibits a lower
creep rate dependence on crystal orientation than alumina single crystal.
Although easy interface sliding is desirable for toughness, too weak an

interface in a composite is not desirable when considering its creep
resistance. It has been predicted that low values of fiber/matrix bonding
increase the creep rate by a factor of two or so for short fiber- (whisker-)
reinforced ceramics (Wang and Liu, 1989). In this regard, we should point
out some interesting results on eutectic composites. Eutectic composites
such as AI20 3N 3AIsO'2' AI20/GdAI03, and AI20/Er3AIsO'2are made by
unidirectional solidification from the melt of appropriate composition. The
main characteristic is of these composites is the insitu formation of a strong
interface. These composites maintain their tensile strength close to the
melting point and increases with decreasing characteristic dimension of the
microstructure but low fracture toughness and thermal shock resistance.
Their microstructure consists of a three-dimensional network structure
containing single crystal YAG phase distributed in a matrix of single crystal
alumina. Figure 7.35 shows the creep curves in tension at 1773 K under
different stresses (Hirano et aI., 2001). Note primary and secondary stages.
The steady state creep curves for the single and polycrystal samples in argon
showed a power-law relationship, see Fig. 7.36 (Hirano et aI., 2001). The
3r---~----~--~~--~----r----'
T = 1773K in air
0=140MPa
0=150 MPa
OL-__ ~ ____ ~ __ ~ ____ ~ __ ~~ __ ~
o 234 5 6
Time. S (10 4 )
Fig. 7.35 Tensile creep curves at 1773 K under different stresses for AI20/Y3AlsOI2
eutectic composites (after Hirano et aI., 2001). Note primary and secondary stages.
10.2 r--...-_..........,....,......,r---...-..............................
AbOJIYAG Sintered Composites
10') (polycrystalline) ....
Compression in argon ....... ".04 Compression
1873K ............. ' in argon
•• .&: ..........
Tension
1(t6
,~JI
10.1
AbOJIYAG Eutectic Comwsites

11,mK
I~·~_~~~~~~_~(~sin~e~c~~~
10 100 1000
Stress, MPa
Fig. 7.36 Steady state creep curves for the single and polycrystal samples of
Al203N3AlsOl2 eutectic in argon showed a power-law relationship (After Hirano
et aI., 2001). The creep resistance is better in compression than in tension and for
the single crystal than for the polycrystaI. (After Hirano et aI., 2001)
tensile creep rates are faster than in compression, although the environments
are also different. Note also that the single crystal eutectic composite has
better creep resistance than the polycrystalline eutectic. Finally, we present
a comparison of tensile creep rate at 1200 °C for 100 h of some CMCs with
a nickel-based oxide-dispersioned superallOY, MA 754, see Fig. 7.37
(Holmes and Wu, 1994). The creep rate ofCMCs is lower than that ofMA
754. But does this mean that the CMCs can replace the superalloy in ajet
engine? Not quite. Most of the results such as those shown in Fig. 7.37 they
are obtained over short time periods and generally in inert atmospheres.
Then there is problem of compatibility between requirements for toughness
such as ease of interface sliding with the requirements for enhanced creep
reisistnace. The message of the results compiled in Fig. 7.37 is simply that
the CMCs do have a great potential as high temperature materials.
Time for 1% £
10-6 0.00032 years
Inconel MA 754 1200 0 C (2.7 h)
Stephens & Nix
NicaloniCAS
(40v/o), IOppm O.
Abbe &
Chermant ~
.....
I
10-8 0.032 years
UJ (270 h)
aJ~ \ #0
~ •
#
"'"
fr
e
U
SCS-6IHPSN
(3Ov/o), air
1 year
3.2 years
10- 10 (27000 h)
10 years
SCS-6/RBSN
(24v/o)
10 ppm 0z
0th -60MPa
10- 12 320 years
30 100 300
Stress, MPa
Fig. 7.37 Comparison of tensile creep rate of some CMCs with a nickel-based
oxide-dispersion hardened superalloy, MA 754 at 1200 °C for 100 h. (After Holmes
and Wu, 1995).
REFERENCES
Aveston, J., G.A. Cooper, and A. Kelly (1971) in Properties ofFibre Composites,
IPC Science and Technology Press, Guildford, p.l5.
Becher, P.F., C.-H. Hsueh, P. Angelini, and T.N. Tiegs. (1988) J. Amer. Ceram.
Soc., 71,1050.
Becher, P.F. and T.N. Tiegs, (1988) Adv. Ceram. Mater., 3, 148.
Bowen, D.H.(l968) Fibre Sci. & Tech., 1, 85.
Bender, B.A., J.S. Wallace, and DJ. Schrodt (1991) J. Mater. Sci., 12,970.
Brennan JJ and KM. Prewo (1982) J. Mater. Sci., 17, 1201.
Bullock, E., M. McLean, and D.E. Miles,(1977) Acta. Metall., 25, 333.
Carroll, D.F. and RE Tressler (1988) J. Amer. Ceram. Soc., 71, 472.
Carroll, D.F. and R.E. Tressler (1989) J. A mer. Ceram. Soc., 72, 49.
Carroll, D.F., S.M. Wiederhom, and D.E. Roberts (1989) J. Amer. Ceram. Soc., 72,
1610.
Chokshi, A.H. and lR. Porter (1985) J. Amer. Ceram. Soc., 68, C144.
Corman, G.S. (1991) Ceram. Eng. Sci. Proc., 12,1745.
Cox, H.L. (1952) Brit. J. App. Phys., 3,72.
Crivelli-Visconti, I. and G. Cooper (1969) Nature, 221, 754.
Davidge, R.W. and A. Briggs (1989) l Mater. Sci., 24, 2815.
Dawson, D.M., RF. Preston, and A. Purser (1987) Ceram. Eng. Sci. Proc., 8, 815.
Dyson, B.F., R.D. Lohr, and R. Morrel (eds.) (1989) Mechanical Testing of
Engineering Ceramics at High Temperatures, Elsevier Applied Sci., London
Evans, C.C.,J.E. Gordon, D.M. Marsh, and N.J. Parrott (1961) Tube Investments
Research Laboratories, Report No. 133, as cited in Bowen, 1968.
Evans, P.E. (1970) J. Am. Ceram Soc. 53, 365.
Fett, T., K Keller, and D. Munz (1988) J. Mater. Sci., 23, 467.
Firestone, R.F. and A.H. Heuer, (1976) J. Amer. Ceram. Soc., 59,24.
Garboriaud, R.l (1981) Phil. Mag., 44A, 561.
Gresczuk, L.B. (1969) in Interfaces in Composites, ASTM STP 452, Amer. Soc.
for Testing and Materials, Philadelphia, p. 42.
Hahn, H.T. and R.Y Kim (1976) J. Composite Materials, 10, 156.
Han, L.x. and S. Suresh (1989) J. Amer. Ceram. Soc., 72, 1233.
Helms, H.E. and PJ. Haley (1989) in Ceramic Materials and Components for
Engines, V.l Tennery (ed.), Amer. Ceram. Soc., Westerville, Ohio, p. 1347.
Hermes, E.E. and R.J. Kerans. (1989) in Mat. Res. Soc. Symposium Proceedings,
MRS volume 125, Materials Res. Soc., Warrendale, PA, p. 93.
Highsmith, A.L. and KL. Reifsnider (1982) in Damage in Composite Materials,
ASTM STP 775, Amer. Soc. Testing & Mater., Philadelphia, p. 103.
Hirano, K, Y. Harade, and T. Suzuki (2001) in High Temperature Ceramic Matrix
Composites, W. Krenkel, R Naslain, and H. Schneider (eds.), Wiley-VCH,
Weinheim, Germany, p. 709
Holmes, J.W. (1991)J. Mater. Sci., 26,1808.
Holmes, lW. and C Cho (1992) J. Amer. Ceram. Soc., 75, 929.
Holmes, J. W. and X. Wu. (1994) in Elevated Temperature Mechanical Behavior of

Ceramic Matrix Composites, S.V. Nair and K. Jakus (eds.), Butterworth-
Heinneman, Boston.
Homeny, J., W.L Vaughn, and M.K. Ferber (1987) Amer. Ceram. Soc. Bull., 66 333.
Jakus, K. and Tulluri, V. (1989) Ceram. Eng. Sci. Proc., 10, 1338.
Jenkins, M.G., M.K Ferber, and RL. Martin. (1990) Ceram. Eng. Sci. Proc., 11,
1346.
Johnson, D.D., A.R. Holtz, andM.F. Grether (1987) Ceram. Eng. Sci. Proc., 8, 744.
Johnson, W.S. (1989) in Metal Matrix Composites: Testing, Analysis, and Failure
Modes, ASTM STP 1032, Amer. Soc. Testing & Mater., Philadelphia, p. 194.
Kandil, F.A. and B.F. Dyson (1988) IntI. J. High Temp. Ceramics, 4, 243.
Karandikar, P.G., and T.-W. Chou (1992) Ceram. Eng. Sci. Proc., 13, 882.
Kerans, RJ. and T.A. Parthasarathy (1991) J. Amer. Ceram. Soc., 74, 1585.
Lawrence, P. (1972) J. Mater. Sci., 7, 1.
Lee, J.-W. and I.M. Daniel (1992) in Composite Materials: Testing and Design,
ASTM STP 1120, Amer. Soc. for Testing and Materials, Philadelphia, p. 204.
Lessing, P.A. and RS. Gordon (1977) J. Mater. Sci., 12,2291.
Lessing, P.A., R.S. Gordon, and K.S. Mazdiyasni. (1975) J. Amer. Ceram. Soc., 58,
149.
Lin, H.-T. and P.F. Becher. (1990) J. Amer. Ceram. Soc., 73, 1378.
Lin, W. and J.-M. Yang (1991) J. Mater. Sci., 26, 4116.
Liu, K.C. and C.R Brinkman,(1985) in Proc.: 23rd Automotive Technology
Development Contractors' Coordination Meeting, Dearborn, MI, October 21-25, p.
165.
Mah, T., N.L Hecht, D.E. McCullum, J.R. Hoenigman, H.M. Kim, A.P. Katz, and
H.A. Lipsitt (1984) J. Mater. Sci., 19, 1191.
Majumdar, B.S., G.M. Newaz, and A.R. Rosenfield (1988) in Proc.: 7th Inti. Con!
Fracture, Pergamon Press, New York, p. 2805.
Marshall, D.B., B.N. Cox, and A.G. Evans (1985) Acta Met., 33, 2013.
McCartney, L.N. (1989) Proc. Roy. Soc. Lon., A425, 215.
Monkman, F.C. and N.J. Grant (1956) Am. Soc. Test. Mater. Proc., 56, 593.
Nixon, RD., S. Chevacharoenkul, and RF. Davis (1987) in Creep ofHot-Pressed
SiC Whisker Reinforced Mullite, Proceeding of the IntI. Conf. on Mullite, Tokyo,
Japan, Nov. 9-10, p. 579.
Ogin, S.L., P.A. Smith, and P.W.R Beaumont (1985) Composites Sci. Tech., 22,
23.
Ohji, T. (1988) Int. J. High Technol. Ceram., 4, 211.
Okamura, K. (1987) Composites, 18, 107.
Paris, P.C. and F. Erdogan. (1963) J. Basic Eng., Trans. ASME, 85 528.
Paris, P.C, M.P. Gomez, and W.P. Anderson (1961) Trend in Engineering, 13,9.
Parthasarathy, T.A., T. Mah, andK. Keller (1991) Ceram. Eng. Sci. Proc., 12,1767.
Penty, RA. and D.P.H. Hasselman (1972) Mater. Res. Bull., 7, 1117.
Phillips, D.C. (1972) J. Mater. Sci., 9, 1175.
Phillips, D.C. (1974) J. Mater. Sci., 9, 1847.
Phillips, D.C., RA.J. Sambell, and D.H. Bowen (1972) J. Mater. Sci., 7, 1454.
Porter, J.R., F.F. Lange, andA.H. Chokshi (1987) Amer. Ceram. Soc. Bull., 66, 343.
Prewo, K.M. and J.J. Brennan (1980) J. Mater. Sci., 15,463.
Prewo, K.M., J.J. Brennan,and G.K. Layden (1986) Amer. Ceram. Soc. Bull., 65,
305.
Pysher, D.J., K.C. Goretta, R.S. Hodder, Jr.,and R.E Tressler (1989) J. A mer.
Ceram. Soc., 72, 284.
Pysher, D.J. and R.E. Tressler (1992) J. Mater. Sci., 27, 423.
Ramakrishnan V. and N. Jayaraman (1993) J. Mater. Sci., 28, 5580.
Routbort, J.L., K.C. Goretta, A. Dominguez-Rodriguez, and A.R de Arellano-
Lopez (1990) J. Hard Materials, 1,221.
Sambell, R.A.J., D.H. Brown, and D.C. Phillips, J. Mater. Sci., 7 (1972) 663.
Seltzer, M.S. and P.K. Talty (1975) J. Amer. Ceram. Soc., 58, 124.
Seshadri, S.G. andK-Y. Chia(1987)J. Am. Ceram. Soc., 70, C242.
Shuler, S.F., J.W. Holmes, X.Wu,. and D. Roach (1993) J. Amer. Ceram. Soc., 76,
2327.
Shih, CJ., J.-M. Yang, and A. Ezis (1992) Composites Sci. & Tech., 43, l3.
S0rensen B. and J.W. Holmes (1995) Scripta Met. et Mater., 32,1393.
Steif, P. and A. Trojnacki.(1994) J. A mer. Ceram. Soc., 77, 221.
Suresh, S. (1991) J. Hard Materials, 2, 29.
Suresh, S., L.X. Han, and J.J. Petrovic (1988) J. A mer. Ceram. Soc., 71, c158.
Talreja, R. (1985) Fatigue of Composite Materials, Technical University of
Denmark, Lyngby, Denmark.
Tattersall, H.G.T. and G. Tappin (1966)J. Mater. Sci. 1,296-301.
Vaidyanathan, R, J. Sankar, and V.S. Avva(1988) Ceram. Eng. Sci. Proc., 9, l383.
Venkatesh, Rand K.K. Chawla (1992) J. Mater. Sci. Lett., 11,650.
Wang, Y.R. and D.S. Liu (1989) Ceram. Eng. Sci. Proc., 10, 1154.
Wang, Z., C. Laird, Z. Hashin, B.W. Rosen, andC.F. Yen (l991)J. Mater. Sci., 26,
5335.
Warshaw, S.l. and F.H. Norton (1962) J. A mer. Ceram. Soc., 45,479.
Wei, G.C. and P.F. Becher (1985) Amer. Ceram. Soc. Bull., 64, 333.
Wiederhom, S.M. and BJ. Hockey (1991) Ceramics IntI., 17,243
Wiederhom, S.M., W. Liu, D.F. Carroll, and T.-J. Chuang (1988)J. Amer. Ceram.
Soc., 12,602.
Yang, J.-M. and Chen, S.T. (1992) Adv. Composites Lett., 1,27.
Zwissler, J.G., M.E. Fine, and G.W. Groves (1977) J. Amer. Cemm. Soc., 60, 390.
SUGGESTED READING
Chawla, K.K. (1998) Composite Materials: Science & Engineering, 2 nd ed.,

Springer-Verlag, New York.
Jenkins, M.G. (1995) Standards And Codes for Ceramic Matrix Compsoites, Adv.
Compsoite Maeriais, 8, 55.
Krenkel, W., R. Naslain, and H. Schneider (eds.) (2001) High Temperature

Ceramic Matrix Composites, Wiley-VCH, Weinheim, Germany.
Nair S.V. and K. Jakus (eds.) (1994) Elevated Temperature Mechanical Behavior
o/Ceramic Matrix Composites, Butterworth-Heinneman, Boston.
Quinn, G.D. and R. Morrell (1991) Design Data for Engineering Ceramics: A
Review of the Flexural Test, J. Amer. Ceram. Soc., 74, 2037.
Suresh, S. (1998) Fatigue o/Materials, 2nd ed., Cambridge Univ. Press, Cambridge,
UK.
Cliapter8
THERMAL STRESSES
In this chapter, we discuss in some detail the important subject of thermal

stresses in composites. This is of importance in all composites, PMCs, MMCs,
and CMCs. The analytical expressions obtained in Sec. 8.2 and 8.3 have, of
course, general validity for a variety of composites. We shall then apply the
results obtained in these sections to CMCs and derive some important
guidelines to obtain enhanced fracture toughness in these materials. Finally, we
dwell on the subject of thermal shock and thermal fatigue in CMCs.
8.1 IMPORTANCE OF THERMAL STRESSES IN COMPOSITES
Thermal stresses are very important in all composite materials, but particularly
so in ceramic matrix composites because, in general, a ceramic matrix has a
very low strain to fracture. Ceramics, in general, also have a lower coefficient
ofthermal expansion than polymers and metals. Figure 8.1 shows linear thermal
expansion as a function of temperature for some polymers, metals, and
ceramics. Note the rather low coefficient of expansion of glass-ceramic (LAS)
and fused silica. Thermal stresses are internal stresses that arise when there
exists a constraint on free dimensional change of a body (Chawla, 1973a,
1973b). In the absence of this constraint, the body can experience free thermal
strains without any accompanying thermal stresses. The constraint can have its
origin in:
(a) a temperature gradient,

(b) crystal structure anisotropy (noncubic structure, for example),
(c) phase transformations resulting in a volume change, and
(d) a composite material made of dissimilar materials (i.e., materials having
different coefficients of thermal expansion).
In general, the problem of thermal gradient can be a serious one in ceramic

264 THERMAL STRESSES
2.0
1.6
~
1.2
"
0
'"
"0.
0
x
w
'0 O.B
E
Q)
.c
f- Silicon Carbide
~
Q)
"
:.:J
0.4
tt.~~~======::1111~:""--- Fused
o II Silica
LAS Glass-Ceramic (LiAISi 20 S )
-0.2 L - - _ - ' - _ - L _ - - ' - _ - l ._ _L-_.L.-_-L-_-.L..:-:--_-'
Temperature, 0 C
8.1 Linear thermal expansion as a function of temperature for some polymers, metals,
and ceramics.
material. A thermal gradient /). T is inversely related to thermal diffusivity, a, of

a material. The thermal diffusivity, a, in turn, is related to the thermal
conductivity, specific heat, and density of the material:
where Cp is the specific heat, p is the density, and k is the thermal conductivity.
Metals generally have high thermal diffusivity; therefore, any thermal gradients
that might develop are dissipated rather quickly. Ceramics, on the other hand,
have low thermal diffusivity and, thus, poor thermal shock resistance. It is easy
to see that the presence oftemperature gradients in CMCs can make an already
complex situation, because of the inherent heterogeneity of a composite, more

complex. It is worth emphasizing that in composite materials even a uniform
temperature change (i.e., no temperature gradient) will result in thermal stresses
due to the ever present mismatch in the coefficients of thermal expansion of the
reinforcement and the matrix (Lla = ar - am). Thermal strain, in the absence of
any temperature gradients, is given by LlaLl T, where Ll T is the amplitude of
temperature change. During cooling from the generally high processing
temperatures used with CMCs, a large magnitude ofthermal stresses can result
because of the thermal mismatch between the reinforcement (particle, short
fiber or long fiber) and the matrix. In the case of a particulate or fibrous
composite, an interfacial pressure can develop during cooling that is similar to
that obtained on embedding a spherical ball of radius, r + Llr, in a spherical hole
of radius, r, or a cylinder of radius, r + Llr, in a cylindrical hole of radius, r.
Specifically, thermal stresses generated depend on the reinforcement volume
fraction, reinforcement geometry, thermal mismatch, the temperature interval
(Tfinal - T initial)' and modulus ratio, E/Em' where the subscript r denotes the
reinforcement which can be a particle, a whisker or a long fiber and m denotes
the matrix. Generally, am> a., i.e., on cooling from T to To (T > To), the matrix
would tend to contract more than the reinforcement, putting the reinforcement
in compression and the matrix in tension. In extreme cases, the axial
compressive stress in a fiber can be large enough to cause buckling of the fibers
and lead to the generation of interfacial shear stresses.
The mismatch in thermal expansion (or contraction) between the components

of a composite material is difficult to eliminate. One can, however, try to
exploit this inherent characteristic of a composite material to obtain some
desirable characteristics in the composite. One such way is to choose the
components such that the residual stress pattern existing at the end of the
processing is one that gives the desired final characteristics in the composite.
Specifically, in ceramic matrix composites, one would like to enhance crack
deflection at the fiber/matrix interface. Such crack deflection at the interface is
a prerequisite for enhanced toughness in the composite. The type of interfacial
bonding (chemical or mechanical) as well as the nature and magnitUde of
thermal stresses existing at the end of processing of a composite will be of great
importance in this regard. Since the choice of the matrix and reinforcement in
a composite is dictated by application, one frequently incorporates a fiber
coating, appropriate for a given composite system, to optimize the stress
distribution and bonding at the fiber/matrix interface. In ceramic matrix
composites, suitable interfacial coatings are chosen such that the bonding is
primarily mechanical in nature. In a mechanical bond, the degree of interfacial
roughness will then become an important parameter.
Thermal stresses arising from a thermal mismatch between the two components
of a composite are a function of a number of variables:
a = F (E, ~a., ~T, a, b, V) (8.1)
where E represents the elastic moduli, ~a. is the thermal mismatch, ~T is the
temperature change, a and b the radii of the components, and V represents the
volume fractions of the components. The specific geometric distribution of one
component in the other one can have very important effects. We derive below
the analytical expressions for thermal stress components in two types of
composites: a central particle surrounded by its associated spherical shell of
matrix and a central fiber surrounded by its cylindrical shell of matrix.
8.2 THERMAL STRESSES IN PARTICULATE COMPOSITES
Consider a particulate composite consisting of small ceramic particles

distributed in a ceramic matrix. Ifwe regard this composite as an assembly of
elastic spheres of uniform size embedded in an infinite elastic continuum, then
it can be shown from the theory of elasticity (Timoshenko and Goodier, 1951;
Brooksbank and Andrews, 1970) that an axially symmetrical stress distribution
will result around each particle. Figure 8.2 shows a schematic of such a particle
reinforced composite. Let us say that each particle has a radius a while the
surrounding matrix sphere has a radius b. This spherical symmetry problem
calls for the use of spherical coordinates, r, e, and <p as indicated in Fig. 8.3. We
have the following stress, strain, and displacement components:
ar> aa = a'll
cr> ca = c'll
ur = u, independent of e or <p
The equilibrium equation is
dar 2
+ -(a - aa) = 0 (8.2)
dr r r
267 THE~STRESSES
Fig. 8.2 A particle reinforced composite consisting of a particle of radius, a embedded

in a matrix of radius, b.
~--------~--- y
,,
, " : p'
,
'
6..
"
x
Fig. 8.3 Spherical coordinate system.
while the strain-displacement relationships are
E =-
du
r dr
u
(8.3)
r
Substituting Eq. (8.3) in Eq. (8.2), we get the governing differential equation
for our problem:
(8.4)
The solution to this differential equation is
u = Ar + -
c
r2
We can now apply the following boundary conditions:
(i) stress vanishes at the free surface (i.e., at r = b)

(ii) the radial stress at the interface (r = a) is the interfacial pressure, P, i.e.,
cr,(a) = -p
When we apply these boundary conditions, we find that the stresses in the
ceramic particle are
cr'p = p = constant = a6p (8.5)
while the stresses in the matrix are

269 THE~STRESSES
,(Jrm
P
- Vp [a'-r3 - V
p 1 (8.6)
p
(Jem -
- Vp [ 2 r3 a'
1- + V
p 1 (8.7)
p
( 8.8)
whereVp is the particle volume fraction, a is the particle radius, b is the matrix
radius, and other symbols have the significance given earlier. Figure 8Aa shows
the three-dimensional stress distribution in a particulate composite. The particle
is under a uniform pressure, P, while the matrix has radial and tangential stress
components. The radial and tangential components in the matrix vary with
distance as shown in Fig. 8Ab. The radial component goes to zero at the free
surface, r = b, as per our boundary conditions. The tangential component has
a nonzero value at the free surface.
8.3 THERMAL STRESSES IN FIBER REINFORCED COMPOSITES
This important problem has been analyzed by a number of researchers

(Poritsky, 1934; Hull and Berger, 1934; Chawla and Metzger, 1972; Scherer,
1986; Herrmann and Wang, 1991; Hsueh et aI., 1988). In the case of a
unidirectionally reinforced fibrous composite, it is convenient to use polar
coordinates because ofthe inherent axial symmetry. Figure 8.5a shows the basic
270 THE~STRESSES
b
MATRIX
(a)
(b)
Fig. 8.4 (a) Three-dimensional stress distribution in a particulate composite. The

particle is under a uniform pressure, P, while the matrix has unequal radial and
tangential stress components, (b) Radial and tangential components in the matrix
varying with distance.
fiber/matrix unit consisting of a central fiber (radius, a) surrounded by its sleeve

of matrix (radius, b). The matrix radius, b, will depend on the volume fraction
of the matrix. Figure 8.5b shows the three-element unit consisting of the central
fiber surrounded by two concentric sleeves of the coating and the matrix
material. Such a simple axisymmetric model can be used to obtain the three
dimensional state of thermal stress up to a moderate fiber volume fraction. The
limit of fiber volume fraction depends on when the stresses from one unit
composite shown in Fig. 8.5 start to overlap with the stresses emanating from
other such units. Hsueh et al. (1985) have used a three-cylinder model for very
low « 5%) fiber volume fractions. If we assume that the stress fields of various
units do not overlap as long as the interfiber spacing is more than one fiber
diameter, then the following thermal analysis will be valid for fiber volume
fractions between 20 and 25 (Chawla, 1998).
The axial symmetry means that we can treat the problem in terms of the
principal stresses that are independent of e. We derive expressions for the
thermal stresses in a three-element cylindrical composite. The two-element,
uncoated fiber composite will then be a special case, wherein the coating is the
same material as the matrix. For the axisymmetric case under consideration, the
radial and circumferential stresses will have expressions of the following form
(Poritsky, 1934; Hull and Berger, 1934; Chawla and Metzger, 1972):
B.
air =A.----! (8.9a)
, r2
B.
a,lI = A.+-'
, r2
(8.9b)
The subscript i represents fiber (t), coating (c), and matrix (m) and the constants
Ai and Bi are determined by the boundary conditions. Also note that air + a ie =
2~.
For the case of a coated fiber with a finite matrix region we can write the
following expressions for tangential and axial strains in the fiber (r = 0 to a),
coating (r = a to b), and matrix (r = b to c):
272 THE~STRESSES
Matrix
:c:
," ,
~ , ,
b' :
I , ,
, ,
h h
I ,
(a) (b)
Fig. 8.5 (a) Basic fiber/matrix unit consisting of a central fiber (radius, a) surrounded
by its sleeve of matrix (radius, b), (b) three-element unit consisting of the central fiber
surrounded by two concentric sleeves of the coating and the matrix material.
E;z = ~.[cr;z-v;(cr;r+cr;6))+(l;LlT
I
(S.10b)
~. [cr;z - v;2A;] + (l;LlT
I
Note that in the case ofthe fiber, Br = 0, otherwise the radial stress will become
infinite at r = O. This leads to the special condition that crre = crfr = Af
8.3.1 Boundary Conditions
Let the radial stresses at the fiber/coating and coating/matrix interfaces be 0".
and O"b' respectively. At r = c, the free surface, O"mr (= O"c) is zero, because the
free surface does not support a normal stress. Using this boundary condition,
we can write
at r = c o = Am
Am
1 1
ame =Bm -+- (S.11a)
c 2 r2
amr =Bm (S.11b)

c2 r2
a me = Am [1 + ~l
r
2
Additional relationships between Ai and Bi parameters can be obtained by

applying the interfacial boundary conditions.
At r = a, the stress continuity at the interface requires that
This gives
B
A = A __C
f c
a2
274 THE~STRESSES
The condition of displacement continuity gives
The tangential strain equality leads to the following expression
_1
E [( A c + Bc)_
c a
2
vc[( A c - Be)
2
a
II
+ cr cz + uc!l. T
The axial strain equality condition gives
E [crcr -vc(2A c)] + uc!l.T

_1
c
At r == b, applying the stress continuity boundary condition, we have

Using the displacement continuity for the tangential strain, we get
(In,-ac )I1T
A _
= _c +B
_
EC
/b'1.
c_
V
- _c (A
EC C C
B c2 +b2
-B /b2) _ ~
E 2b2
m
l C
j
- v
~O"
EC cz
0" V
v
+ ~ + -.!!!...
E 111 E 111
I-B [22b2b 211
m
C-
C
and for the axial strain, we obtain
E [0"C-v
_1
C
ZCC
2A ] + a C I1T E [0" mzm
= _1 -v 2 Bm2
m C
I+ a m I1T
0" - ---.!!!!..
(a -a )I1T = ...5.. 0" - _c 2A c _ ( A __
v __ B -.-!!!.. -,-c_~
(2 b 2)
C
)2V
m C Ec Em Ec C b2 E m (2
C
-b2)
We now have the following unknowns:
We also note that a force balance in the axial direction gives
(8.13)
After combining the equations obtained from the application of the boundary
conditions, we can write
(8.14)
(8.15)
(8.16)
(8.17)
where WI through W 4 and KI through KI6 are defined as follows:

-2v
= __f
K6
a 2E
'/
1
K7 - -
Ec
Ec c
c 2-+b+
-1( l - v ) +1- ( -
Em
2
v
2_b 2 m
1
C
K -
14 -
2
b2
VIII (
Em
b2 1
c 2 -b 2
K 15
Ec
1
K 16 -
Em
Equations 8.14 through 8.17 can be solved for Ae, Be' O"rz' O"m" and O"w This
thermoelastic solution can provide information about the magnitude of the
elastic stresses involved and if the elastic range will be exceeded or not. In the
case of a metallic matrix, it is likely to deform plastically in response to these
thermal stresses (Chawla, 1973a; Chawla, 1973b; Arsenault and Fisher, 1983;
Christman and Suresh, 1988). In the case of a ceramic matrix, cracking can
occur to relieve the high level of thermal stresses. Figure 8.6 shows the
distribution of the radial, tangential, and axial stresses in a silicon carbide
fiber/alumina matrix composite for a temperature change of - 1 °C for two fiber
volume fractions, 5 and 35 %. In this figure we have used the data for a CVD
SiC fiber (see chapter 3) and a fully dense alumina (see chapter 2). We can
draw some important inferences from Figure 8.6:
(i) Axial stress is uniform in fiber and matrix, although its magnitude is
different in the two and depends on the respective elastic constants.
(ii) In the fiber, i.e., the central component f, O"rf and O"er are equal in magnitude
and sense. In the matrix, i.e., the sleeve m, O"nn and O"am vary as [1 - (b2/r2)] and
[1 + (b 2/r2)], respectively.
(iii) When either the temperature difference or the expansion coefficient
difference goes to zero, the thermal stresses vanish, as expected.
Figure 8.6 also shows that the change in the axial stress magnitude with a
change in fiber volume fraction is much more than in radial and tangential
components. Also to be noted is the compressive radial stress at the fiber/matrix
interface in this system, a situation not very favorable for interfacial debonding.
If we incorporate a coating between the fiber and the matrix, the stress
distribution will change. This is commonly done to prevent a strong chemical
bond between the matrix and the fiber and thus provide a somewhat weak
mechanical bond which is conducive to fiber pullout, a desirable feature for
enhanced toughness. An example involving a coating is the alumina fiber/silica-
'0.70
0.56 SiC/AI 2 0 3
0.42 fl T= -1°C
Vf =0.05
10.28
vi '0.14'
<Il a~m
....~
C/) 0.00
co
E -0.14
Q;
..c -0.28:
I-
-0.42
-0.56
-0,70 '--_ _...._ _....I.._ _...I._ _....&_ _ ~
0,000 62.60 125.20 187.80 250.40 313.00

(a) Distance From Fiber Center, IJm I'
1.00 _--oop---__ --...,..--.....,~- ......

0.80 SiC/AI 2 0 3
0.60 flT= -1°C
0.40 Vf =0.35
vi
<Il
0.20 ,Matrix
Fiber
....~
C/)
0.00
co
E -0.20
Q;
..c
I- -0.40
-0.60
-0.80
-1.00
0.00 23.60 47.20 70.80 94.40 118.00
Distance From Fiber Center, IJm
Fig. 8.6 Distribution of thermal stresses in a CVD silicon carbide fiber/ alumina matrix
composite for a flT = -loC, (a) V r = 5% and (b) Vr = 35%.
based glass matrix composite. Commonly, a carbon coating is used such that
one has an alumina fiber/carbon coating/glass matrix composite. Alumina and
silica-based glass form a strong chemical bond; the carbon coating prevents that
and provides a weak mechanical bond. It can be shown that the radial stress in
alumina/glass and alumina/carbon/glass is tensile. Even though the radial stress
is tensile in alumina/glass, the strong chemical bonding between silica-based
glass and alumina inhibits debonding. In alumina/carbon/glass composites, the
tensile radial stress combined with weak mechanical bonding due to the carbon
coating aids in debonding. This has been verified by a number of researchers
(Fitzer and Schubert, 1979; Brun and Singh, 1988; Jurewicz et aI., 1989;
Feldman, 1986). We discuss this subject at some length in chapter 9.
8.4 THERMAL STRESSES IN LAMINATED CMCs
The expression for coefficient of thermal expansion for laminated ceramic

matrix composites follow essentially the same as the ones given by Schapery
for the fiber reinforced CMCs (Gladysz and Chawla, 2001). They studied the
A120/Sn02 and AI20/CaTi03 laminates, among others. An interesting
observation regarding the Alp/CaTi03 laminate was that the higher 11
component, viz., CaTi03 showed regularly spaced transverse cracks which
developed during processing due to the large thermal stresses caused by the
large CTE mismatch between Al 20 3 and CaTi03 and are similar to those
reported by Cai et ai. (1997) in AlzO/ZrOz laminates (ZrO z is the higher 11
component in this system).
8.S MEASUREMENT OF RESIDUAL STRESSES
Residual stresses are elastic stresses that remain in a body after all external
loads have been removed. It is important to realize that, for a body in
equilibrium, residual stresses must form a balancedforce system in a body. The
magnitudes of tensile and compressive stress components, however, need not
be equal. The most important source of residual stresses in CMCs, as discussed
above, is the thermal mismatch between the components. These can be relieved,
to some extent, in a CMC by micro cracking of the matrix. That, however, may
not always be the case. For example, in a three-dimensional carbon/carbon
composite, Feldman (1986) observed warping in thin slices of the composite.
As we pointed out in chapter 3, carbon fiber itself is highly anisotropic in
thermal expansion. The high processing temperatures involved make the
residual stress problems particularly severe. Feldman cut thin slices of different
thicknesses from a three-dimensional carbon/carbon composite and observed

that thin slices of the material, with asymmetric cross section, showed a residual
curvature.
The most important effect of the thermal expansion mismatch between the
reinforcement and matrix is that residual stresses build up in the composite
when the composite is subject to any temperature change, say, during cooling
from the processing temperature to room temperature, or during service. The
residual stresses have significant effects on both interface sliding stress and
matrix cracking stress (Evans and Marshall, 1989) as well as ultimate strength
(Hutchinson and Jensen, 1990). The residual axial stress will directly
superimpose on the applied load in the fiber direction, and large tensile residual
stress will induce cracks in either fiber or matrix. The radial tensile forces at the
interface will encourage fiber debonding and slippage, which in turn result in
high toughness, high work of fracture, and a fibrous appearance on the fracture
surface. The interfacial sliding resistance can be expressed as (Hutchinson and
Jensen, 1990)
where 't is the shear stress acting at the interface, 0"2 is the stress normal to the
interface, Jl is the friction coefficient and 'to is the sliding resistance when 0"2 is
positive (tensile).
Large mismatch in coefficients of thermal expansion fiber and matrix are

undesirable. Of course, if the ratio of coefficients of thermal expansion of the
fiber, a f to that ofthe matrix, am is equal to unity, it will completely eliminate
thermal strains in the composites. This, however, is unlikely to be achieved in
most CMCs. On the other hand, a ceramic fiber is likely to have a greater strain
to failure than the matrix. Also, a tensile stress normal to the interface is
desirable for ease of debonding. Thus, arUm is desirable in most cases.
It should be clear to the reader that a quantitative idea of residual stresses in a

body made of a ceramic composite is very important. The subject of
measurement of residual stresses in a body or a component is, indeed, very vast.
In this section, we just wish to provide a brief introduction to the subject and
let the interested reader consult references (Predecki et at, 1988; Majumdar et
aI., 1988a; Majumdar et aI., 1988b; DiGregrio et aI., 1992; Barrera and Dutta,
1993) for details.
282 THE~STRESSES
A number of researchers have used different experimental techniques to

measure residual stresses. X-ray and neutron diffraction techniques can be used
with crystalline materials. Predecki et al. (1988) and Majumdar et al. (1988a,
1988b), among others, used x-ray diffraction and neutron diffraction,
respectively, to analyze the state of residual stress in silicon carbide
whiskers/alumina composites. In both these techniques, the crystal lattice is
used as a strain gage. The x-ray diffraction technique can measure average
strains in the near surface regions, about 50 !lm thick. Neutrons can penetrate
a material deeper than x-rays, therefore neutron diffraction technique can
provide strains in the bulk of the composite. The basic principle is that the
presence of a residual stress will change the interplanar spacings of the lattice
planes and thus cause a shift in the diffraction peaks. In both these techniques,
however, the strains are measured directly and the stresses are obtained through
appropriate elastic constants. Thus, in both cases, diffraction of an incident
beam of neutrons or X-rays occurs according to the Bragg's law of diffraction:
2d hkl sin9 = A
where dhkl is the interplanar spacing corresponding to a Bragg refection of (hkl)

planes, 29 is the diffraction angle, and A is the wavelength ofthe neutron or X-
ray beam. If dhkl represents the lattice spacing of the strained lattice due to
residual stress and do represents the spacing of the unstrained lattice planes,
then the residual strain can be written as
Raman spectroscopy can be used to measure residual stress directly to about 1

!lm of the surface (DiGregrio et aI., 1992). We discussed the use of Raman
spectrosocopy in chapter 3. Characteristic Raman spectra can be obtained from
fibers, even when the fibers are embedded in a matrix. For a sufficiently
transparent matrix, the technique can be used to depths of 5 - 30 !lm. The basis
ofthe technique is that under an applied stress, specific Raman bands shift from
their normal positions.
8.5 THERMAL FATIGUE AND THERMAL SHOCK
As we have pointed out, in any composite material, there exists a very

fundamental physical incompatibility between the fiber and the matrix, to wit,
the difference in their thermal expansion (or contraction) coefficients. This
incompatibility leads to thermal stresses whenever there is a temperature
change. It should be emphasized that thennal stresses in composites will arise

even if the temperature change is uniform throughout the volume of the
composite. Such thermal stresses can be introduced in composites during
cooling from high fabrication temperatures or during any temperature
excursions (inadvertent or by design) during service. Turbine blades, for
example, are very much likely to be susceptible to thermal fatigue.
Sudden changes in environmental temperature can cause large stress gradients

in materials having a low thermal conductivity (or thennal diffusivity). Such
exposure of a material to transient thennal conditions is called thennal shock.
Brittle materials such as ceramics and glasses fail catastrophically if the thennal
shock causes its tensile strength to be exceeded. Enhanced toughness in CMCs
makes them attractive for high temperature applications where thermal shock
conditions might prevail. One common technique involves subjecting the
composite to a water quench, a simple and inexpensive process, and
detennining the maximum temperature interval that the material can withstand
without fracturing. Before presenting some results on the behavior of CMCs
under thennal cycling and thermal shock conditions, we define some thennal
shock parameters.
8.5.1 Thermal shock parameters
The generalized Hooke's law relating stress and strain components for an
isotropic solid, in Cartesian (x, y, z) coordinates, can be written as
1 1
-[0" -v (0" +0" )], E._ -[0" -v (0" +0")]
E Y Z x E Z x Y
where s is the strain, 0" is the stress, E is the Young's modulus, and v is the
Poisson's ratio.
If a large plate has a thennal gradient across its thickness and it is cooled from
both surfaces, then we have a state of biaxial stress at any point. The two stress
components are given by
aEl!.T
(I-v)
where a. is the coefficient of thermal expansion and f). T is the change in

temperature. For rapid cooling from a temperature To we can assume that only
the surface temperature change to T., then the maximum stress will occur at the
surface and we can write
uEf).T
°max
(I-v)
In a general way, important information about stress intensification can be

derived from the thermal stresses associated with the nonlinear temperature
distribution in a body. These stresses can be described for convective heat
transfer conditions by using a non-dimensional parameter called the Biot's
modulus, /3. Biot's modulus is a dimensionless heat transfer number that is the
ratio of the product of surface heat transfer coefficient (h) and a sample
dimension (ro) to the thermal conductivity (k). Thus,
hr"
k
The surface heat transfer coefficient, h, depends on the quenched material and
the surroundings. One can incorporate Biot' s modulus into a function,f@} such
that ATc
(J (I-v)
max 1{P)
uE
where 0max is the maximum surface stress that develops in a quenched body, A Tc
is the critical temperature difference that will initiate cracks in the sample,
generally at the surface, and the parameter /(/3) is given by:
4.67 51
j{P) = 1.5 + - - 0.5 exp(--)
p p
For a severe quench, /3 is very large, which makes/@} approximately equal to
1.5.
A common thermal shock measure is the maximum difference in temperature

(..11) jump in surface temperature that a brittle material can sustain without
cracking or catastrophic failure. The difference in temperature (..11) corresponds
to the critical temperature at which fracture will occur. There are several
equations that have been derived to express the thermal shock resistance of
ceramics. For very rapid cooling rates,
R = I1Tc
where O"f is the failure stress of the material.
Another thermal shock parameter, designated R', is defined as follows:
R'
where we have introduced the thermal conductivity, k, of the material. For slow
cooling rate, the parameter f(fJ) is about 1I0.31~, which gives the critical
temperature difference as
R'
Thermal shock resistance, as seen in the above expressions, is a function of

several material properties such as Young's modulus, thermal conductivity,
coefficient ofthermal expansion, and fracture strength. High values of thermal
conductivity and strength with low values of modulus and thermal expansion
would be recommended to avoid fracture initiation by thermal shock. Other
parameters that are significant include the heat transfer coefficient, specimen
size, porosity, and duration of thermal shock. All of these factors are important
to increase the thermal shock resistance of a ceramic. But due to the brittle
nature of ceramics, one must also consider the effect of crack propagation in
ceramics. This is taken into account by defining a yet another parameter,
designated as R" that includes fracture surface energy (y) or fracture
/I
toughness (K,c):
2
R /11/ = Ey K 1c
Ceramics containing cracks show a decrease in strength when subjected to a

temperature change. The dominating cracks are found at the surface where
tensile stresses are generated, as compared to the compressive stress inside the
ceramic.
8.5.2 Thermal shock and thermal fatigue of CMCs
In view of the fact that CMCs are likely to find major applications at high
temperatures, it is of interest to study their behavior under conditions of
isothermal exposure as well as under conditions of thermal cycling. We should
mention that more realistic testing for thermal shock resistance involves rather
expensive equipment such as burner rigs or other hot gas testing facilities and
combustion chambers (Eckel et aI., 1997). Many researchers have contributed
in this area (Blissett et aI., 1997; Boccaccini et aI., I997a, I997b; Bouchetou
et aI, 1993; Bhatt, 1992; Bhatt and Palczer, 1997; Case et aI., 1993; Chawla et
aI., 2001; Kagawa et aI., 1993; Kim et aI., 1990; Lamicq et aI., 1986; Sinha et
aI., 1994, Tiegs and Becher, 1987; Wang et aI., 1996; Wang and Chou, 1991;
Wang and Singh, 1994; Wetherhold and Zawada, 1991). We provide a summary
of the results. Wetherhold and Zawada (1991) studied the behavior of ceramic
grade Nicalon fiber in an aluminosilicate glass matrix under isothermal and
thermal cycling. In the 650 - 700°C, isothermally exposed as well as thermally
cycled samples showed rapid oxidation and loss in strength. Oxidation behavior
overshadowed any thermal cycling effect for these test conditions. The
embrittlement was attributed to oxygen infiltration from the surface, which
destroyed the weak carbon-rich interface in this composite. At 800°C, however,
less embrittlement was observed and the fiber toughening effect remained. This
decreased embrittlement at higher temperatures was attributed to smoothening
of the sample surface by glass flow and slow oxygen infiltration. Thermal shock
resistance of glass matrix containing aluminum titanate particles was improved
as evidenced by an increase in ~ Tc from about 100 to 400 K for 30% aluminum
titanate (Boccaccini and Pfeifer, 1999). Boccaccini et al. (1997a, 1997b)
studied the thermal shock behavior ofNicalon fiber (NL 202) reinforced Duran
glass (a borosilicate glass) matrix composite, a system that has almost zero
thermal mismatch between the fiber and the matrix. Thermal shock quenching
in water (22°C) from 620 °C led to matrix microcracking. The evolution ofthis
damage was studied by measuring the change in Young's modulus and internal
friction; a decrease in the modulus and a simultaneous increase in internal
friction occurred. The change in internal friction was more sensitive to
microstructural changes, especially in the early stages of damage development,
than the change in Young's modulus. Crack healing occurred in the glass matrix
composites when cycled above the glass transition temperature of the matrix.
Lamicq et al. (1986) used a water quench technique to test the thermal shock
resistance of SiC fiber/SiC composite materials made by chemical vapor
infiltration. The composite showed a loss in flexural strength gradually over the
range of i1T = 300-750° C. Increasing i1T to 1200° C resulted in no further
decrease in flexural strength. Bhatt and Phillips (1990) also used water
quenching on a reaction bonded SCS-6 SiC fiber/Si3N4 composite and observed
a decrease in its flexural strength at i1T greater than 600°C: however, tensile
strength remained unchanged! They attributed this fact to a loss in interply
integrity alone, with shear delamination occurring in flexural tests but not in
tensile tests.
Kagawa et al. (1993) incorporated Nicalon fibers into two different matrix
materials-borosilicate glass and lithium alumino silicate (LAS) - for which the
fiber/matrix interfacial bond is very weak. Unidirectional composites were
made by hot pressing. Water quench thermal shocks with i1T values of250 to
1000 ° C produced multiple cracking of the matrix perpendicular to the fiber
axis for i1T greater than 600° C. This cracking occurred only in the outermost
surface of the material. Flexural strength decreased by less than 5% at i1T = 800
°C. Neither elastic modulus nor flexural strength were affected at i1T < 600°C.
Also, the depth of crack penetration was independent of any increase in i1T, and
only the increase in crack density at the surface was responsible for the
decrease in elastic modulus. For the LAS matrix composite, however, the cracks
formed parallel to the fiber axis, but again they formed only at the surface.
Flexural strength remained virtually unaffected by thermal shock. Elastic
modulus, however, decreased as a result of matrix microcracking for i1T > 600
°C. At 800°C, a phase change occurred in the LAS matrix, which could be a
possible source of microcracking.
Changes in stiffness were used to analyze the thermal shock behavior of a

hybrid composite consisting ofNicalon fiber reinforced hybrid composite with
a matrix of barium magnesium aluminosilicate (BMAS) glass with silicon
carbide whiskers (Chawla et aI., 2001). Thermal shock from 850°C to room
temperature, however, showed a reduction in modulus and damage to the
microstructure. Fiber pushout tests, performed to evaluate changes in
288 THE~STRESSES
fiber/matrix interface strength after thermal cycling. Thermal shock

experiments showed a substantial decrease in modulus when quenched from
850°C and extensive cracking due to the high cooling rates associated with
quenching. Close to 50% decrease in modulus was observed after thermal shock
from 850 °C to room temperature. This was attributed to extensive fiber
cracking and a limited amount of cracking in the matrix. The specimens cycles
at 750°C to room temperature, however, showed only a limited amount of
cracking and no decrease in modulus. Fiber pushout measurements, on the other
hand, showed a similar decrease in interfacial shear strength, similar to that
observed in 850°C. Since the crack spacing is greater than fiber pushout
specimen thickness, a decrease in modulus caused by fiber fracture is unlikely
to affect interfacial strengths measured by the fiber pushout technique.
The following points of difference between the performance of monolithic

ceramics and that of CMCs under thermal shock are worth noting (Wang and
Singh, 1994):
• Fiber reinforcement changes both the crack initiation and the crack
propagation behavior of a ceramic.
• The mismatch of properties between the fibers and matrix (e. g. the
Young's modulus, coefficient of thermal expansion, thermal
conductivity, or symmetry of properties) causes additional thermal
stresses within the material.
• Interfaces between the fibers and matrix provide important interactions.
Wang and Chou (1991) summarize the following conclusions from their work
on SiC fibers in a borosilicate glass matrix:
• The thermal shock resistance parameter, R, cannot be expressed simply

for CMCs as it is for monolithic ceramics, and it is therefore most
sensible to adopt /). To as a relative measure of thermal shock resistance.
• The following material properties affect thermal shock behavior in the

order of greatest to least importance: coefficient of thermal expansion,
Young's modulus, and thermal conductivity.
• As the fiber orientation changes from 45 0 to either 00 or 900, the failure

mechanism of CMCs under thermal stress evolves from delamination
to matrix microcracking.
• Ultimate failure of CMCs does not necessarily occur from being

subjected to temperature differences greater than Ll T e , and these
materials may maintain their load-carrying ability to a significant
degree.
REFERENCES
Arsenault R.J. and RM. Fisher(1983) Scripta Met., 17,67.

Barrera, E.V. and I. Dutta (1993) Residual Stresses in Composites, The Mineral, Metal
and Materials Society, Warrendale, PA.
Bhatt, RT. (1992) J. Amer. Ceram. Soc., 75, 406.
Bhatt, R T. and RE. Phillips (1990) J. Materials Science, 25, 340.
Bhatt, R. T. and A. R. Palczer (1997) J. Materials Science, 32,103.
Blissett, M.J., P.A smith, and J.A. Yeomans (1997) J. Materials Science, 32, 317.
Boccaccini, AR. and K. Pfeiffer (1999)Glass sci. Technol. Glastech Ber. 72,32.
Boccaccini, A.R, C.B. Ponton, and KK Chawla (I 997b) Mat. Sci. Eng. A, 241A, 142.
Boccaccini,A.R, D.H. Pearce, J. Janczak, W. Beier, and C.B. Ponton (1997a) Mater.
Sci Technology, 13, 852.
Bouchetou, M. L., M. Huger, and C. Gault (1993) J. de Physique Ill, 3, 1883.
Brooksbank, D. and KW Andrews (1970) J. Iron and Steel Inst., 208, 582.
Brun, M.J. and R.N. Singh (1988) Adv. Ceram. Mater., 3, 505.
Cai, P.Z., D.J. Green, G.L. Messing (1997)J. Amer. Ceram. Soc., 80, 1929.
Case, E. D., Y. Kim, and W. J. Lee (1993) in Thermal Shock Behavior and Thermal
Fatigue of Advanced Ceramics, G. A. Schneider and G. Petzow (eds.), Kluwer
Academic Publishers, Dordrecht, the Netherlands, p. 393.
Chawla, KK(1973a) Phil. Mag., 28, 401.
Chawla, K.K. (1973b) Metallography, 6, 55.
Chawla, KK (1998) Composite Materials: Science & Engineering, 2 nd edition,
Springer-Verlag, New York, p. 328.
Chawla, KK and M. Metzger (1972) J. Mater. Sci., 7, 34.
Chawla, N., KK Chawla, M. Koopman, B. Patel, C.C. Coffin, and J.1. Eldridge,
(2001) Compo Sci. Tech., 61, 1923.
Christman, T. and S. Suresh (1988) Acta Met., 36, 1691.
DiGregrio, 1.F., T.E. Furtak, and J.1. Petrovic (1992) J. Appl. Phys., 71, 3524.
Eckel, AJ., J.z. Guekenyesi, T.P. Herbell, and E.R. Generazio (1991) NASA Tech.
Memo. 103777
Evans, A.G. and D.B. Marshall (1989) Acta Met., 37, 2567.
Feldman, L.A. (1986) J. Mater. Sci. Letter, 5, 1266.
G. M. Gladysz and KK Chawla(200 1) Composites A, 32 173.
Herrmann, K.P. and Y.Q. Wang (1991) in Inelastic Deformation of Composite
Materials, Springer-Verlag, New York, p. 445.
Hsueh, C.S., P.F. Becher, and P. Angelini (1988) J. Amer. Ceram. Soc., 71, 929.
Hull, AW. and E.E. Burger (1934) Physics, 5, 384.
Hutchinson, J.W. and H.M. Jensen (1990) Mech. Mater., 9,139.

Jurewicz, A.J.G., RJ. Kerans, and J. Wright (1989) Ceram. Eng. Sci. Proc., 10,925.
Kim, Y., W.J. Lee, andE.D. Case (1990) in Metal and Ceramic Matrix Composites, R.
B. Bhagat, A. H. Clauer, P. Kumar and A. M. Ritter (eds.), The Minerals, Metals, and
Materials Society, Warrendale, PA, p. 479.
Kagawa, Y., N. Kurosawa, and T. Kishi (1993)J. Mater. Sci., 28, 735.
Lamicq, P.J., G.A. Bernhart, M.M. Dauchier, and J.G. Macem (1986) Bull. ofAmer.
Ceram. Soc., 65, 336.
Lee, W. J. and E.D. Case (1989) Material Science and Eng. A1l9, 113.
Majumdar, S., M.A. Bertinetti, and S.R. MacEwen (1988a) J. Am. Ceram. Soc., 73,
858.
Majumdar, S., D. Kupperman, and J. Singh, (1988b)J. Amer. Ceram. Soc., 71858.
Noyan, I.C. and J.B. Cohen (1987) Residual Stress, Springer-Verlag, New York.
Poritsky, H. (1934) Physics, 5, 406.
Predecki, P., A. Abuhasan, and C.S. Barrett (1988) Advances in X-Ray Analysis 31,
231.
Scherer, G. (1986) Relaxation in Glass & CompOSites, John Wiley, New York.
Sinha, A., K. Kokini, and K.J. Bowman (1994) Material Science and Engineering,
AI88,317.
Tiegs, T.N. and P.F. Becher (1987) J. Amer. Ceram. Soc., 70, C109.
Timoshenko S. and J.N. Goodier (1951) Theory ofElasticity, McGraw-Hill, New York,
1951, p. 416.
Fitzer, E. and P. Schubert(1979) Rev. Int. Hautes Temperatures Refractaires, 16, 147.
Wang, H.V. and R.N. Singh (1994) Int. Mater. Review, 39, 228.
Wang, H.V., R.N. Singh, and R.A. Lowden (1996)J. A mer. Ceram. Soc., 79,1783.
Wang, Y.R. and T.-W. Chou (1991) J. Mater. Sci., 26, 296l.
Webb, J.E., R.N. Singh, and R.A Lowden (1996) J. Amer. Ceram. Soc.,79, 2857.
Wetherhold, R.C. and L.P Zawada (1991) in Fractography of Glasses & Ceramics,
V.D. Frechete and J.R. Varner (eds.), Ceramic Transactions, vol. 17, Amer. Ceram.
Soc., Westerville, OH, p. 391.
SUGGESTED READING
Noyan, I.C. and J.B Cohen (1987) Residual Stress, Springer-Verlag, New York.
Schneider, G.A. and G. Petzow (eds.) (1993) Thermal Shock Behavior and Thermal
Fatigue ofAdvanced Ceramics, Kluwer Acad. Pub., Dordrecht, The Netherlands.
Cliapter 9
INTERFACE MECHANICS AND

TOUGHNESS
In this chapter, we discuss the role of interface in CMCs with a special

emphasis on mechanics of the fiber/matrix interface as related to that somewhat
elusive property called toughness. We discussed the importance and general
features of interface in composites in Chapter 5. Now we look at the subject of
tailoring this interface in CMCs with a view to obtain an enhanced fracture
toughness in these materials. We give some examples of the interface structure
in some CMCs, briefly review the topic of toughness in different types of
CMCs and then we analyze in some detail the toughening mechanisms in fiber
reinforced ceramic matrix composites because of their immense promise.
9.1 INTERFACE INCMCs
Once the matrix and the reinforcement of a composite are chosen, it is the set
of characteristics of the interface region that determines the final properties of
the composite. In this regard a thorough microstructural characterization of the
interface region assumes a great deal of importance. Many sophisticated
techniques are available to characterize the interface region. We give examples
from some important CMCs that bring home two points: firstly, there are tools
available that allow us to make an in-depth characterization and secondly, the
interface region in real composites can be immensely complex and variable.
The chemical composition of the components, the crystallographic as well as
chemical nature of the interface region, the effect of processing conditions, etc.
can affect the properties and performance of the composite to a very large
extent.
Disordered and discontinuous layers have frequently been observed the

reinforcement/matrix interfaces in various CMCs. The technique of high
resolution electron microscopy (HREM) imaging can provide a considerable
detail of the interface structure, chemical and crystallographic. Figure 9.1
292 INTERFACE MECHANICS AND TOUGHNESS
Fig. 9.1 An example of the highly disordered interface region in a rice-hull derived
silicon carbide whisker/alumina composite made by hot pressing (courtesy of A.
Weresczczak). Note also the stacking faults in the whiskers.
shows an example of the interface region in a rice hull derived silicon carbide
whisker/alumina composite made by hot pressing. One can see the stacking
faults in the SiC whisker and the disordered region at the interface. Das
Chowdhury et al. (1 992a, 1992b) observed discontinuous and continuous
interfacial layers in silicon carbide whisker reinforced silicon nitride
composites by high resolution electron imaging and analytical microscopy.
Examples of their observations and analyses of interfaces in these composites
are shown in Fig. 9.2 and 9.3. Figure 9.2 shows a bright field TEM micrograph
of the interface in a SiCwf'Si3N4 doped with Y203 and A1 20 3. The circled region
was analyzed by high resolution electron microscopy, see Fig. 9.3b and d. A
high resolution electron microscopy image of the interface, Figure 9.3a, shows
an amorphous region at the interface. The structural width of the interface
region is 1.5-2 nm. The arrow indicates the scan direction. Figure 9.3b shows
the variation of oxygen concentration across the interface as a function of
distance. A sharp change in oxygen concentration occurs at the interface. The
reader will also notice rather large difference in chemical and structural widths
of the interface. Figure 9.3c shows a high resolution electron image of another
interface region ofthe same composite while Fig. 9.3d shows the oxygen profile
across the interface. In this case no abrupt change in oxygen concentration at
the interface was detected. One also notes the high oxygen concentration in the
silicon nitride matrix and the large difference between the chemical and
structural width of the interface.
Fig. 9.2 A bright field TEM micrograph of the interface in a SiCjSi3N 4 doped with
Y203 and A120 3. The circled region was analyzed by high resolution electron
microscopy (courtesy ofK. Das Chowdhury).
(a)
1.2
O-K
1.0
0.8
;: 0.6
x
E
~
0
U 0.4,
0.2
0.0
0 200 400 600 800 1000 1200
Distance A
(b)
Fig 9.3 (a) HREM image ofthe interface showing an amorphous region atthe interface.
The structural width ofthe interface region is 1.5 - 2 nm. The arrow indicates the scan
direction. (b) Variation of oxygen concentration across the interface as a function of
distance. Note the sharp change in oxygen concentration at the interface. (c) HREM
image of another interface region of the same composite (d) oxygen profile across the
interface in (c); in this case no abrupt change in oxygen concentration at the interface
was detected. Note the high oxygen concentration in the silicon nitride matrix and the
large difference between the chemical and structural width of the interface. (Courtesy
ofK. Das Chowdhury).
295 INTERF ACE MECHANICS AND TOUGHNESS
(c)
0.5
O-K
0.4
'" S
>< 0.3
~::>
0
U 0.2 SiC Silicon Nitride
~ interface
0.1
0 200 400 600 800 1000 1200
Distance A
(d)
Fig 9.3 - contd
A frequent and important observation is the presence of interfacial glassy

phase(s), e.g., a glassy phase at the interface between a silicon carbide whisker
and silicon nitride matrix (Schoenlein et aI., 1988). We discuss some aspects
of interface control to obtain desired properties in Sec. 9.3.
9.2 TOUGHENING MECHANISMS IN VARIOUS CMCs
Extensive efforts have been devoted to the subject of toughening ceramics by

making composites. We provide a summary of some of these approaches to
enhance toughness of ceramics via the composite route; following which we

analyze in some detail the toughening mechanisms in fiber reinforced CMCs.
(i) PARTICLE TOUGHENING. Interaction between particles that do not undergo

phase transformation and a crack front can result in toughening from the
following sources: crack bowing between particles (Lange, 1970; Evans, 1972),
crack deflection at the particle (Faber and Evans, 1983) and crack bridging by
ductile particles (Budiansky, et aI., 1988). When a crack bows between
particles, it causes the stress intensity factor along the bowed segment to
decrease while resulting in a corresponding increase in stress intensity at the
particle. A particle can make a propagating crack t urn away from the direction
normal to the applied stress and deflect. This interaction increases the crack
surface area and results in increased energy absorption. Crack deflection is
more effective than crack bowing in toughening materials. Crack bridging can
happen while matrix is incorporated by ductile particles. The particles can
deform elastically and/or plastically up to failure. Crack bridging is more often
seen at short-fiber- or whisky-reinforced composites. Thermal residual stress
caused by the mismatch in coefficient of thermal expansion(CTE) of the matrix
and the particles can lead to changes in the fracture mode in the matrix and,
thus, can increase or decrease the toughness of the materials. Thermal stresses
can serve to toughen the material if the matrix is put under compressive
stresses. Thermal stresses can also weaken the material if the matrix is put
under tensile stress. Compressive thermal stresses in the matrix are generated
when the CTE of the particle exceeds that of the matrix. Under such conditions,
an increase in toughness of the composites occurs. Examples include: TiCp
ISiC (Janney, 1987), TiB 2 p/SiC (Taya et aI., 1990) and Alp3p/glass/(Todd et
aI., 1999). Earlier, Mendelson and Fine (1974) observed an increase in the
fracture initiation energy and the fracture strength of wustite (FexO) by two
methods: (a) by coherent precipitation of Fe30 4 (13 v/o) in wustite. The
difference in cleavage habits between the precipitate and the matrix and pinning
of the crack-front led to an increase in both the fracture-initiation and fracture
strength, and (b) by exploiting the eutectoid decomposition ofwustite into a-Fe
and Fe30 4 • a-Fe formed a continuous network at the prior wustite boundaries,
contributed plastic work during the fracture process, and, thus, prevented
catastrophic fracture. A linear increase in the fracture-initiation energy as a
function of the volume fraction of a-Fe was observed. A similar increase was
observed in the work of fracture as determined from the area under the load-
deflection curve (Mendelson and Fine, 1974). Toughening of glasses by
incorporating metallic particles which undergo plastic deformation is also
possible (Krstic et aI., 1981).
(ii) TRANSFORMATION TOUGHENING. This involves a phase transformation of

the second phase particles at the crack tip with a shear and a volumetric or
dilational component, thus reducing the tensile stress concentration at the crack
tip. In particulate composites such as alumina containing partially stabilized
zirconia, the volume change associated with the phase transformation in
zirconia particles is exploited to obtain enhanced toughness. In a partially
stabilized zirconia (ZrOz + Y Z03), the crack tip stress field can cause a stress
induced martensitic transformation in ZrOz from tetragonal phase (t) to
monoclinic phase (m):
ZrOz (t) ~ ZrOz (m)
The transformation at the crack tip results in dilation of the transformed zone
around a crack. This increase in volume is opposed by the surrounding
untransformed material leading to compressive stresses which tend to close the
crack. This makes crack propagation difficult leading to an increased toughness.
Figure 9.4 shows this transformation toughening mechanism. Transformation
in the wake of a crack can result in a closure force that tends to resist the crack
opening displacement. Crack deflection at zirconia particles can also contribute
to toughness. A good survey of transformation toughening in particulate CMCs
Fig. 9.4 Transformation toughening mechanism.

can be found in Green et al. (1989). Kriven (1995) have exploited the phase
transformation concept in fiber reinforced CMCs containing a fiber coating
which undergoes a phase transformation involving a volumetric change; we
discuss this below.
It should be noticed that, in a practical composite, more than one type of

toughening mechanism can happen, and the effect of the simultaneous action
of two or more mechanisms needs to be considered. The simplest assumption
would be that the toughening effects act independently and would simply add.
However, the possibility of greater toughening through interaction (synergism)
than that resulting from simple addition should not be ignored.
It can be seen from the above analysis that in all these mechanisms, control of
the crack propagation, i.e., crack path, plays a key role in the improvement of
crack growth resistance or fracture toughness of CMCs. There has been some
work to quantify the crack path, and relate it to the mechanical behavior of the
materials (Bhargava and Patterson, 1997; Wu et aI., 2000, 2001). They analyzed
the influence of crack path on fracture toughness of particle reinforced CMCs
such as Al 20 3 particles in a cubic zirconia matrix, silicon carbide or alumina
particles in a glass matrix. They defined a stereo logical parameter, called
degree of contact, R, to characterize interactions between crack and reinforced
particle. As one would expect, they found that the thermal and elastic misfits
between the matrix and the particulate reinforcement, the loading and/or
unloading rate, and the test environment all significantly influenced the crack
path. The crack propagation velocity was directly related to the variation in
loading rate. The amount of particle-crack interaction significantly influenced
the fracture toughness of ceramic composites. For a composite in which the
matrix had higher thermal expansion coefficient (n) and lower Young's
modulus (E) than the particle, increasing interaction between particles and
cracks decreased fracture toughness. For a composite in which the matrix had
lower n and lower E than the particles, decreasing interaction between particles
and cracks decreased the fracture toughness. This work showed that, excluding
the environmental effect, crack-particle interactions played a key role in
influencing the fracture behavior of the composites. This degree of contact
parameter provides a simple and useful tool for analysis of the influence of
variables on particle-crack interaction and influence of such interaction on
fracture behavior of ceramic composites.
(iii) WHISKER OR FIBER REINFORCEMENT. Toughening by whiskers or long

fibers is the main theme of this chapter. Almost everything that we have said
about the particle toughening mechanisms above can be brought to play in fiber
reinforced CMCs as well. Whisker or fiber reinforcements can bring into play
a series of energy absorbing mechanisms in the fracture process of CMCs and,
thus, make them damage tolerant.
A summary of the various toughening mechanisms that can be made to operate

in ceramic matri~.composites is given in Table 9.1.
Table 9.1 Ceramic matrix composite toughening mechanisms.
Mechanism Requirement
1. Compressive prestressing U f > urn' will result in an axial compressive

of the matrix prestressing of the matrix after fabrication.
2. Crack impeding Fracture toughness of the second phase (fibers or

particles) is greater than that of matrix locally.
Crack is either arrested or bows out (line tension
effect).
3. Crack deflection Weak fiber/matrix interfaces deflect the

propagating crack away from the principal
direction.
4. Fiber (or whisker) Failure along fiber/matrix interface leads to fiber
pullout pullout on further straining.
5. Phase transformation The crack tip stress field in the matrix can cause
toughening the second phase particles (fibers) at the crack tip
to undergo a phase transformation causing
expansion (il V> 0). The volume expansion can
squeeze the crack shut.
9.2.1 Toughening mechanisms in fiber reinforced CMCs
If we incorporate continuous fibers such as C , SiC, or Al 20 3 to a glass or

.ceramic matrix, we can obtain a stress-strain curve in tension of the type shown
in Fig. 9.5a while in flexure we get a stress-strain curve as shown in Fig. 9.5b.
These curves have the following salient features:
• An initial elastic behavior up to apI, the proportional limit

• At a certain stress microcracks appear in the brittle matrix.
• The cracks bypass the fibers and leave them bridging the cracks .
• Under continued loading, we have regularly spaced cracks in the matrix,
bridged by the fibers.

• Fiber fracture and fiber pullout, followed by failure of the composite. A
damage tolerant behavior showing a noncatastrophic failure occurs in a
composite consisting of two brittle components.
The final failure of the composite is not the result of the passage of a single
crack, i.e., a self-similar crack propagation does not occur. Thus, it is difficult
to define an unambiguous fracture toughness value, such as a K1c value
(Marshall et aI., 1985). Under some circumstances, matrix cracking stress, 0"0
can be independent of the preexisting flaw size and is a material property. 0"0
is a function of matrix fracture toughness, interfacial shear strength, fiber
volume fraction, fiber radius, and the elastic constants of the fiber and matrix.
For fully bridged cracks, the 0"0 value can be independent of crack length
(Marshall et aI., 1985).
Although, the appearance of the first matrix crack does not signify a complete
failure of the composite, it does result in a reduced slope of the stress-strain
curve, i.e., a decreased modulus. It also implies an easy access path for any
aggressive environmental species that might be present. Matrix microcracking
will cause a reduction of strength and modulus and might lead to internal
oxidation and spalling as well. Thus, in practical terms, any microcrack
toughening is not likely to be very high and the reliability of such composites
containing microcracks also will not be very high.
Fiber bundle failure

I.Y.",,,,.r~"·"·• . .
fiberpu!loul I'.
Strain, !> Strain ----.
(a) (b)
Fig. 9.5 Typical stress-strain curves obtained with continuous fiber reinforced ceramic
matrix composites. (a) in tension (b) in flexure.
In fiber reinforced ceramic matrix composites, the local response of the

fiber/matrix interface during fracture is of great importance. The structure and
behavior of the interface region is the key to effective control of damage in
composites and enhancement of toughness. To this end, it becomes necessary
to identify the basic failure mechanisms or origin of fracture toughness for a
given fiber/matrix system, and then explore means for exploiting and/or
controlling these mechanisms. There have been many theoretical and
experimental studies of fracture in fiber composites and of toughening
mechanisms. When a crack moves through a brittle matrix containing
unidirectional fibers, the following failure mechanisms may be expected to
operate:
• Matrix fracture
• Debonding at the crack tip, followed by crack deflection
• Debonding in the crack wake, followed by crack deflection
• Frictional sliding between the fiber and matrix
• Fiber failure
• Fiber pullout
It has been amply demonstrated (Donald and McMillan, 1976; Ramualdi and
Basto, 1963; Majumdar, 1970; Sambell et aI., 1972; Aveston et aI., 1973;
Aveston, 1973; Vaidya et aI., 1992) that reinforcement with continuous fibers
such as carbon, alumina, silicon carbide, mullite fibers in cement, glass, and
glass-ceramic matrix can result toughening. It is not necessary for all these
failure mechanisms to operate simultaneously for a given fiber/matrix system
and often in many composite systems only one or two of these toughness
contributions will dominate the total fracture toughness. Let us examine these
mechanisms in some detail.
9.2.2 Criteria for Debonding and Crack Deflection
It would appear that the debonding (sometimes the term delamination is also
used, but, we prefer debonding) of the fiber/matrix interface is a prerequisite
for phenomena such as crack deflection, crack bridging by fibers, and fiber
pullout. Cook and Gordon (1964) analyzed the crack deflection at an interface
between materials of identical elastic constants, i.e., the same material joined
at an interface, see Sec 5.5 and Figs. 5.6 and 5.7 in chapter 5.
Two materials that meet at an interface are more than likely to have different
elastic constants. The modulus mismatch causes shearing of the crack surfaces.
This results in a mixed-mode stress state in the vicinity of a crack tip at an
interface involving both the tensile and shear components. This, in turn, results
in a mixed-mode fracture, which can occur at the crack tip or in the wake of the
crack. Figure 9.6 shows this crack front and crack wake debonding in a fiber
reinforced composite. What this means in practical terms is that instead of a
simple, one parameter description by the critical stress intensity factor Kle , one
needs a more complex formalism of fracture mechanics to describe the
situation. The parameter K under such a situation becomes scale sensitive, but
the critical strain energy release rate, Gle , is not a scale sensitive parameter. G
is a function of the phase angle, 'P, which is a function of the normal and shear
loading. One needs to specify G and 'P to analyze the debonding at the
interface. G and 'P are related, through the so-called Dundurs parameters, a and
~, as defined below:
lGp -v2) - G2(l-V I)j

a
rG$l-v2) + G2(l-v$]
1 [G$l-2v2) - Gi l - 2v$]
(9.1)
2 [G$l-v 2) + G2(l-v$]
The parameter a can also be defined as
where
- b
E=--
1 - v2
The expression for the phase angle, 'P, in terms ofthe elastic coefficients of the
two media, radius r from the crack tip and the displacements, u and v, at the
crack tip is as follows:
\jI = q> - (Inrl21t) In[(l-~)I(l +~)]
Such expressions have been used by many researchers (He and Hutchinson,
198~; Evans and Marshall, 1989; Ruhle and Evans, 1988; Gupta, 1991; Gupta
et aI., 1993) to analyze the conditions for fiber/matrix debonding in terms of the
energy requirements. One result of this body of work is shown in Fig. 9.7 in
terms of a chart of G/Gf vs. a, where Gi is the mixed-mode interfacial fracture
energy of the interface, Gr is the mode I fracture energy of the fiber, and a is the
measure of elastic anisotropy as defined above. For opening mode or mode I,
'P = 0° while for mode II, 'P = 90°. The plot in Fig. 9.7 shows the conditions
under which the crack will deflect along the interface or propagate through the
interface into the fiber. For all values of G/G f below the dashed line area,
interface debonding is predicted. For the special case of zero elastic mismatch,
i.e., for a = 0, the fiber/matrix interface will debond for G/Gf less than about
0.25. Conversely, for G/G r greater than 0.25, the crack will propagate across
the fiber. In general, for the elastic mismatch, a greater than zero, the minimum
interfacial toughness required for interface debonding increases, i.e., high
modulus fiber tend to favor debonding. One shortcoming of this analysis is that
it treats the fiber and matrix as isotropic materials. This is not always true,
especially for the fiber.
Il) r III
t r III
1
r1
~ -
t I
'-------
I ,
r ,)
-r
-
frl l ll/
.--,
crad, \\ a "~ crack
~
d¢bondi n" ~ dcbon ling
Fig. 9.6 Crack front and crack wake debonding in a fiber reinforced composite.
1.0
<!>
.....
0.5
o ~--~----~--~--~
-1.0 -0.5 o 0.5 1.0
Elastic mismatch, a
67.5
..,..'" 45.0
~
22.5
-1.0 -0.5 0 0.5 1.0
Elastic mismatch, a
Fig. 9.7 Charst of G/Gr vs. elastic mismatch, a and phase angle, 'I' as function of a
(after Evans and Marshall, 1989).
Gupta et al. (1993) derived strength and energy criteria for crack deflection at
a fiber/matrix interface for several composite systems, taking due account ofthe
anisotropic nature of the fiber. Their laser spallation technique was described
in chapter 5. This technique allows one to measure the tensile strength of a
planar interface. They have tabulated the required values of the interface
strength and fracture toughness for delamination in number of ceramic, metal,
intermetallic, and polymer matrix composites. We discuss their results briefly.
Figure 9.8a shows a crack terminating normal to the interface between two
orthotropic materials. Both materials are treated as homogeneous, linear elastic
solids. Fibers are anisotropic while the coatings are isotropic. Gupta et al.
(1993) find that the stress and deformation fields depend on two material
parameters, ').. and p, for the two media and two bimaterial constants, a and B.
For plane stress condition, these parameters depend only on the elastic
compliances, Sij of the two media. These parameters are defined as follows:
(9.2a)
(9.2b)
rss
Y = V~22~11' n= ~
~~ (9.2c)
where the subscripts 1 and 2 outside the parentheses in Eq. 9.2b refer to the two
media, with the subscript 2 designating the cracked medium. Note that the
parameters A. and p become unity for an isotropic medium. We use these
parameters in describing the important results of Gupta et al. (1993).
9.2.3 Stress criterion
According to this criterion, the crack will deflect along the interface if
where o*j is the interface strength, O*f is the fiber strength, Oyy{QO) is the stress
at the interface, and oxx{9QO) is the stress in the fiber. Figure 9.8b depicts this
strength criterion for delamination. Knowing the elastic constants of the fiber
and matrix, we can calculate a. and p. For these values of a. and p, the ratio
o*/o*c can be obtained from Fig. 9.8b. We can obtain the fiber strength, o*c,
from the manufacturer or determine it experimentally. Knowing these three
quantities, we can obtain the maximum permissible interface strength value for
delamination.
9.2.4 Energy release rate criterion
The energy release rate criterion for crack deflection considers two possible
directions of crack propagation at the interface: along the interface or across the
interface. If the ratio of energy release rate for crack growth along the interface,
Gd, to growth perpendicular to the fiber, Gp, is less than the ratio of work of
delamination of the interface, Gci , to the work offracture across the fiber, Gcft,
then the crack will deflect along the interface, i.e., the crack will deflect for
Fiber (2)
-
Matrix
y,2
Reaction -.l;~;;:I@~F==-T--,-
product Gcft
wedge
(a)
o cf» (0',/') I (O"f 'It)

o
C> r:Tyy(OO)= OjO
0.4
bK Interface delominolion
"
;;-
o
0.2
-0.5 0.0 0.5 1.0

a
(b)
Fig. 9.8 (a) A crack tenninating nonnal to the interface between two orthotropic
materials (b) strength criterion for interface delamination. (After Gupta et ai., 1993)
The main crack may be deflected singly or doubly, Fig. 9.9 (a,b). According to
Gupta et aI., for most pairs of materials, the energy release rate is higher for the
doubly deflected crack than that for the singly deflected case. Figure 9.10 plots
G/Gp for a doubly deflected crack as a function of the bimaterial parameter a
for a series of 1-,1 values.
According to these authors, a generalized interface delamination chart based on

the energy criterion cannot be made. However, the authors do provide the ratio
ofGiGp for a few chosen interface systems from metal, ceramic, and polymer
matrix composites. Table 9.2 gives the information for some ceramic systems.
The first column is the reinforcement (medium 1), the second column is either
a coating or a matrix. Columns 3 and 4 give the values of a and ~, respectively,
for the material pairs in columns 1 and 2. Columns 5 and 6 give the calculated
energy release rate ratios for the singly deflected and the doubly deflected
cracks, respectively. Column 7 gives the maximum of the two values and thus
provides a bound on the ratio GjGeft for interface delamination. Column 8
gives the ratio of the mode 2 to mode 1 components of the stress intensity
factors at the deflected crack tip. Note that almost all interfaces are governed
by the doubly deflected crack.
The strength determined in the laser spallation experiment is thought to be

independent of any inelastic processes because the interface separation takes
place at a very high strain rate, about 106 S-I. The tensile strength determined
in this fashion is the related to intrinsic interface toughness through the
following relationship (Rose et aI., 1983; Rose et aI., 1984; Gupta, 1991):
2
°max.
m f m f
( (
\ .
(a) (b)
Fig. 9.9 Crack deflection at the interface (a) singly deflected and (b) doubly deflected.
where h is the unstressed separation distance between the planes meeting at the
interface, Eo is the tensile modulus of the interfacial layer, and other symbols
have the same significance given earlier. This kind of information does provide
some fundamental input into the behavior of a given pair of components of a
composite and the interface thereof.
9.2.5 Matrix Cracking: Aveston, Cooper, and Kelly (ACK) Model
The strain value at which the first matrix crack appears in a unidirectionally
reinforced, fibrous CMC and the corresponding first matrix crack stress value
can be important parameters. Aveston, Cooper, and Kelly (1973) made an
energy balance analysis of a composite consisting of unidirectionally aligned,
continuous fibers, in a matrix whose failure strain is less than that of the fibers.
The ACK model makes the following assumptions:
pt== AF P2= 1.0; p= O.

2
* AI= 0.2
+ AI= 1.0
X AI= 2.0
1.5
o AI= 5.0
Lnrerface
O.S
.\ -0.8 -0.6 -0.4 -0.2 o 0.2 0.4 0.6 0.8

(J.
Fig. 9.10 Plot of G/Gp for a doubly deflected crack as a function of the bimaterial
parameter a for a series of A) values (after Gupta et aI., 1993).
Table 9.2 Energy criterion for interface crack deflection for various metallic, ceramic, intermetallic and polymer matrix composites
(adapted from Gupta et a1., 1993)
Medium] Medium 2 GJG~

ran-1(KrlK;
Fiber Matrix Coating a f3 G.,jGp* Gw/G/ ~ (degrees)
a-Al,03 SnO, -0.2096 -0.0907 0.2369 0.2281 0.2369 -51.79

a-Alp3 C (amorphous) 0.7242 0.2111 0.7276 0.7452 0.7452 -22.33
a-Al,03 LAS glass 0.2658 0.0380 0.3223 0.3357 0.3357 -30.75
AlP3 sapphire Pt 0.4863 0.1098 0.4174 0.4343 0.4343 -16.08
Al,03 sapphire Mo 0.2049 0.0330 0.2907 0.3024 0.3024 -16.08
SiC SnO, 0.1294 0.0115 0.2815 0.2913 0.2913 -34.55
SiC LAS glass 0.6367 0.2375 0.5544 0.5714 0.5714 -23.35
SiC Soda glass 0.7641 0.2610 0.8061 0.8226 0.8226 -21.42
Al,03 sapphire Ni 0.3949 0.1065 0.3614 0.3776 0.3776 -16.D7
AI'03 sapphire Nb 0.5291 0.1409 0.4460 0.4633 0.4633 -16.19
SiC Pt 0.4519 0.1120 0.4104 0.4274 0.4274 -26.85
SiC Nb 0.4966 0.1425 0.4380 0.4555 0.4555 -26.05
SiC Mo 0.1623 0.0378 0.2895 0.3010 0.3010 -33.68
Nicalon SiC (bulk) -0.4192 -0.1666 0.2440 0.2173 0.2440 -62.17
Pitch-55 carbon SiC (bulk) -0.7310 -0.2456 0.3226 0.2649 0.3226 -49.68
SiC Si3N. 0.1838 0.0438 0.2954 0.3076 0.3076 -33.08
SiC Zirconia 0.0263 -0.0152 0.2603 0.2659 0.2659 -37.84
SiC Al203 0.0553 0.0127 0.2656 0.2733 0.2733 -37.00
Al,03 sapphire LAS glass 0.6622 0.2133 0.5833 0.6004 0.6004 -17.77
Nicalon LAS glass 0.2987 0.0870 0.3327 0.3473 0.3473 -30.18
Nicalon Si3N. -0.2529 -0.1199 0.2350 0.2220 0.2350 -53.74
SiC TiAl 0.4436 0.1560 0.3961 0.4l35 0.4l35 -27.04
Nicalon Ti3Al 0.1082 0.0190 0.2764 0.2859 0.2859 -35.26
SiC Ti3Al 0.5028 0.1~~2 0.4374 0.4520 0.455Q -25.91
*Singly deflected; tdoubly deflected
• the fiber and the matrix both have well-defined stress - strain curves and they
are elastic to fracture.
• the fiber/matrix interface is under a constant, frictional shear stress.
• the interface debonding energy is neglected.
• the situation of isostrain prevails.
• the strain to failure of the brittle matrix in the composite is less than that of
the fiber.
In such a composite, with only frictional bonding between the fiber and the
brittle matrix, the ACK analysis shows that the matrix failure strain is enhanced
by the presence of fibers and that the matrix will be traversed by a series of
more or less parallel cracks and perpendicular to the fibers. The crack
separation will be between x and 2x. Consider the loading of such a composite
in tension. The brittle matrix will crack first. The ACK model makes another
assumption at this point, to wit, the matrix crack propagates across the whole
matrix, causing a planar fracture in the matrix and the broken halves of the
matrix are bridged by the fibers. When the first matrix crack appears, the stored
elastic energy in the matrix is released. At the same time, an additional load is
thrown onto the fibers in the vicinity of the crack. An energy balance done by
ACK indicates that the matrix failure strain of such a CMC will be enhanced
when the following inequality holds
(9.3)
where Emu is the failure strain of the unreinforced ceramic; 't is the interfacial
frictional bond strength; Ym is the matrix fracture energy; r is the fiber radius;
Yrand Ymstand for fiber and matrix volume fraction, respectively; Er, Em, and
Ec are the Young's moduli of the fiber, matrix, and composite, respectively.
If the fiber radius, r, is large and we have a small fiber volume fraction, Yr, the
right hand side in Eq.(9.3) will be small and Emu will be greater than the right
hand side in Eq.(9.3). In this case, the matrix strain in the composite Emuc will
be equal to the matrix strain of the unreinforced ceramic, Emu' When we have a
small fiber radius, r, and a large fiber volume fraction, Yr, then Emu will be less
than the value given by the equality in Eq.(9.3) and we will have Emuc> Emu'
We now obtain an expression for the crack spacing, x. Assuming only sliding
friction between the fiber and matrix, i.e., physical or chemical bonding,
consider a single crack in the matrix. Over some characteristic distance x from
the crack, the load on the intact fiber will be greater than the average load on
the fiber. This distance x depends on the shear strength of the interface, t. Let
us say there are N fibers per unit area, then
Vr = Nnr or Nnr = VIr.
The load transferred from a fiber to the matrix over a distance x is 21trxt. This
must equal the tensile load carried by the matrix V mam' i.e.,
21tfNtx = 2(Vf r)tx = V mam (9.4)
where am is the tensile stress in the matrix at a distance x from the crack face.
At some point along the fiber length, the fiber and matrix will experience the
same strain.
2V
_f'CX
I
r
(9.5)
With increasing applied load, the load on the central region of the matrix
increases until it fractures. Of course, the matrix fracture will occur only if
enough load transfer length is available for the matrix fracture strength to be
attained. Let us say that at x = XI' am = a mu' the fracture strength of the matrix.
Then, from Eq.(9.4), we have The ACK model does have some problems.
Consider Eq. (9.3). It indicates that the matrix strain to failure goes to zero at
V f = 0, which clearly not tenable. The model also predicts that Cmuc increases
with interfacial strength. Thus, a high interfacial strength would be desirable.
This ignores the fact that in the case of a strong interface, fiber/matrix
debonding may not even occur.
9.2.6 Budiansky, Hutchinson, and Evans (BHE) Model
Budiansky, Hutchinson, and Evans (1986), hereafter known as BHE, analyzed

the steady state cracking (SSC) in CMCs including frictional and residual
stresses. The SSC condition is when crack grows with no increase in load. The
BHE model considers two types of interfaces:
(i) Unbonded fibers, held in the matrix under the action of thermal or other
mismatches, but subject to frictional sliding.
(ii) A weak fiber/matrix interface which becomes debonded under the action of
stresses at the crack tip in a matrix.
Both ACK and BHE models predict that above a certain fiber volume fraction,
the matrix cracking strain (and stress) will be higher than that of the
unreinforced material. It is worth pointing out that both these models consider
a very special case of composite geometry, loading and failure mode. Details
of crack growth are ignored and an instantaneous planar crack formation in
matrix is assumed. The so-called SSC condition does not always exist in
composites. It is not unusual to observe small but distinct cracks in the matrix
well before extensive, large-scale cracking occurs.
9.2.7 Fiber Pullout
We now consider the phenomenon of fiber pullout from a matrix. Consider a

fiber of radius r embedded in a matrix and being pulled out as shown in Fig.
9.11. The longitudinal tensile force on the fiber will generate shear at the
fiber/matrix interface. Neglecting the shear stress concentration at the fiber end,
two things can happen: either the fiber will break due to the tensile stress or it
will get pulled out of the matrix due to the shear at the fiber/matrix interface
without being broken. A simple force balance, shown in Fig. 9.11, assuming a
constant shear stress and ignoring the shear stress concentration, can tell us
which one of these two will happen. Let 0ru be the fracture stress ofthe fiber in
tension and 'tj be the shear stress at which the interface fails. Then, for the fiber
to break, the tensile force on the fiber at breakage must be less than the shear
force corresponding to the interface failure, i.e.,
For fiber pullout, on the other hand, the above inequality will be reversed.
1 2 crt'u
(9.6)
d 4 't,
The aspect ratio (~/d) corresponding to the equality in Eqs. (9.6) is called the
critical aspect ratio. One can write
(9.7a)
If the diameter is fixed, we can write Eq. 9.4 as
(9.7b)
Once the inequality condition for fiber pullout is satisfied, the fiber will be
pulled. The work done in this pullout process is discussed in the next section.
9.2.8 Work Done in Fiber Pullout
Consider the situation depicted in Fig. 9.12. Let us say that the fiber breaks at
some distance k below the main crack plane, such that 0 < k < ~/2. There is
local debonding of the fiber/matrix interface. When the fiber is pulled out of the
matrix, an interfacial frictional shear stress, 'ti will be generated. Recall that in
our simple analysis here, we have assumed that this shear stress, 'ti, resisting the
fiber sliding is a constant. More complex treatments involving a Coulomb
friction law governing the fiber sliding resistances (Shetty, 1988; Gao, et aI.,
1988) and treatments taking into account residual stresses as well as a Coulomb
friction law are available in the literature (Cox, 1990; Hutchinson and Jensen,
1990; Kerans and Parthasarathy, 1991). We derive a simple expression for the
work done in the process of fiber pullout.
Let the fiber be pulled out a distance x. The interfacial shear force opposing this
motion of fiber is 'rjnd(k-x). If the fiber is pulled out through a small distance
dx, then the work done by the interfacial shear force is 'rjnd(k-x)dx. The total
work done in the fiber-pullout process over the length k is obtained by
integrating.
k k k
f'rind(k-x)dx 'r;Ted fkdx - fXdx
o o 0
(9.8)
The length of the fiber pulled out will vary between 0 and Q/2. The average
work done in the fiber pullout process is
0<
Wfp _
- (-1 ) JT,xd k' dk
Q/2 2
r
c 0
~ (~,:2) [T,;d ~3
2 2
'rind Qc 'rind Qc
--- (9.9)
6 4 24
Matrix
}I'
Pull
Fig. 9.11 A fiber of radius r embedded in a matrix and being pulled out. The
longitudinal tensile force on the fiber generates shear at the fiber/matrix interface.
-T
Principal crack
1
Fig. 9.12 Fiber pullout
This equation assumes that all of the broken fibers are pulled out. Experimental
observations, however, show that only fibers with broken ends within a distance
V2 from the principal fracture plane undergo the pullout process. Thus, we
should expect a fraction (e,/O of fibers to pullout. Therefore, the average work
done per fiber in fiber pullout can be written as
Wjp (9.10)
e 24
Ifthere are N fibers in the area of the crack, A, then we can write the following
for the fiber volume fraction, V f
(9.11 )
where d is the fiber diameter.
Pullout work done per unit area of the crack can then be obtained from Eqs.
(9.8) and (9.10) as
Rearranging,
2
- N 1td 2 T;~c
n
W=---
A 4 6d
2
T·~c
V-'- (9.12)
f 6d
It follows from Eq. (9.12) that in order to maximize the fiber pullout work, we
should have a large fiber volume fraction, high interfacial strength, high ~c, and
small fiber diameter. Here we have tacitly assumed that fiber/matrix debonding
has already occurred and that the fiber pullout work is due to sliding at the
interface. Recalling from Eq. (9.6) that for a fiber being loaded from both ends,
the critical aspect ratio is given by
(9.13)
Using Eq. (9.12), we can eliminate the critical length, and write the pullout
contribution to toughness per unit area of the crack as
d2
-.2. (J/u
2
W = V.
f 6d 4 '[2
(9.14)
9.2.9 Kerans and Parthasarathy Model
Kerans and Parthasarathy (1991) proposed a comprehensive model to explain

the results of fiber pullout and fiber pushout. Their model includes chemical
bonding, Poisson contraction or expansion effect on frictional resistance,
residual axial strain in the fiber, and the effect of interface roughness on fiber
sliding after debonding. Poisson contraction during a fiber pullout test will add
a radial tensile stress component aiding the debonding. Poisson expansion, on
the other hand, will come into play during a fiber pushout test and will provide
a radial compressive stress inhibiting the debonding. Kerans and Parthasarathy
(1991) predict two types of interface failure: a catastrophic failure and a
progressive failure. Catastrophic failure occurs when the interface debond
initiation load is high or when the residual stress is low. In the progressive
debonding case, as expected, the load-displacement curve is nonlinear. Interface
parameters such as the coefficient of friction, radial stress at the interface,
interface fracture toughness, and residual axial strain in the fiber are used to
predict the load-displacement curve from the first nonlinearity to near the peak
load. Their model considers a fiber of radius r embedded in an infinite cylinder
of matrix. A length eo of the fiber protrudes out of the fiber. Their final
complicated expression relating the load, p. to the displacement, 0 at the fiber
end during progressive debonding is as follows:
0= Qo P/v Er+ [(1 - 2uf k)/(2f! k 7tf Ef )][ P d + Pr - p.

+ (p' - Pr) In {po -(Pd + Pr)}/(p' _p.)] (9.15)
The significance of various symbols in this equation is as follows. E f and Em are

the elastic moduli, U f and Urn are the Poisson's ratios of the fiber and matrix,
respectively, k = Em ul {(Er (l + Urn) + Em (1 - ur))], and f! is coefficient of
friction. Pro the residual axial load on the fiber due to thermal mismatch between
the fiber and matrix, and is given by
Pr = -~a ~T 1t rEf {(1+ 2k)/(1-2uf k)}

where ~a equals (a,., - ar) and ~T is the temperature interval. P d is the critical
axial load in the fiber at the debond crack tip necessary to make the debond
crack grow and is related to the critical strain energy release rate of the
interface by the following relationship:
Finally, p. is the critical axial load on the fiber at which the Poisson contraction
balances the residual normal stress and the roughness effects such that no
frictional resistance remains to sliding. p. is related to radial stress, O'r at the
interface
(9.16)
Parthasarathy et ai. (1991) showed how one can extract the key interface
parameters from their model and a single fiber pullout test. It should be
recognized, however, that the large scatter in the data poses problems in
obtaining the interface parameters with high reliability (Jero, et aI., 1992).
Mumm and Faber (1992) used a special single fiber pullout experiment on SiC
monofilament (SC S-6)/soda-borosilicate glass composite system to evaluate this
progressive debonding model. Their single fiber sample was notched to give a
reduced cross section at the midpoint of the sample. They successfully used the
Kerans and Parthasarathy analysis to obtain interfacial parameters from the
progressive debonding portion of load-displacement curves. The calculated
interfacial parameters were consistent with independent measurements
available in literature.
9.2.10 Crack Bridging by Fibers or Whiskers
Crack deflection and crack bridging by whiskers or fibers can be major

toughening mechanisms, especially in whisker reinforced composites such as
the silicon carbide reinforced alumina that are used as cutting tool inserts (see
chapter 11). Whisker/matrix debonding and whisker pullout occur behind the
crack front and exert closure forces on the crack (Becher and Wei, 1984; Wei
and Becher, 1984; Becher, et aI., 1988; Becher, 1990). The increase in
toughness due to whisker reinforcement, according to Becher et al. (1988),
~K.vr is given by
1
!l.K =-
wr 2
where K", is the matrix toughness.
For a very brittle matrix, we can take Km '" 0, and rewrite .1Kwr as
Vwr Ee Gm
(9.17)
6(1 -v 2) Ew G;
where Ee = Young's Modulus of the composite (I;.Vw + EmVm)

Ew = Young's Modulus of the whisker
Gm = initial strain energy release rate of the matrix
Gi = critical strain energy release rate of the interface
r = whisker radius
v = Poisson's ratio
Vw = whisker volume fraction
O'wu = whisker ultimate strength
Crack bridging is an important toughening mechanism in CMCs having a

frictional bonding at the interface. Figure 9.13a shows the idea of crack
bridging schematically, Fig. 9.13b shows an example of crack bridging by
silicon carbide whisker in alumina matrix, while Fig. 9.13 c shows the toughness
enhancement obtained by silicon carbide whisker reinforcement of in different
matrix materials. In Fig. 9 .13c, .1Kfb and .1KPO represent frictional bridging and
pullout contributions, respectively. The room temperature toughness of
monolithic alumina has been nearly doubled by incorporating about 20%, by
volume, of SiCw ' The reader will recall from chapter 7 that fracture strength
also increased as function of SiCw in SiCjAl2 0 3 composites. In a later paper,
Becher et al. (1991) showed that the fracture toughness of whisker reinforced
composites increased with increases in matrix grain size as well as whisker
content, a combined result of grain bridging and whisker bridging. The matrix
microstructure (i.e., grain size) contribution to the toughness should be included
in an overall evaluation of the toughness of a composite. Giannakopoulos and
Breder (1991) analyzed the synergism among the various toughening
mechanisms that are possible in a whisker reinforced composite by using an
energy release rate balance equation. Their basic assumption is that the load
transfer occurs via a Coulombic friction at the whisker/matrix interface. When
their model is applied to SiCjAl20 3 composites, it predicts that fracture

toughness increases with increasing whisker volume fraction, reaches a
maximum after which it starts to drop and goes to zero at 100 % whisker
volume fraction. Using a coefficient of friction, Il = 0.007, which the authors
obtained by a finite element silmulation, their prediction fitted very well the
experimental values obtained by Becher et al. (1988).
9.3 INTERFACE ENGINEERING FOR ENHANCED TOUGHNESS
It should be clear to the reader by now that a proper control of the interface
characteristics is the key to obtain a tough ceramic composite. The components
of the composite must be chemically stable at the processing temperatures as
well as at the service temperatures. Any chemical interaction between the
matrix and the reinforcement during processing will influence the interface
bond strength, and consequently the ability to debond at the fiber/matrix
interface and cause fiber pullout.
Therefore, an optimum interface strength and fracture toughness (or strain

energy release rate) are required. The interface should be strong enough for
axial load transfer and strength in the transverse direction, yet week enough to
make a transverse crack deflect along the interface leading to fiber pullout, and
thus provide toughness. The interface characteristics can be controlled in two
ways:
(i) proper selection of the fiber and matrix such that they form a
thermodynamically stable interface during processing and service.
(ii) incorporation of appropriate interfacial coatings will prevent chemical

reaction between fiber and matrix at the interface. Choice of a coating depends
on the fiber, matrix, processing technique, and service requirements.
Interfacial coating provides an easy and effective way to tailor the interface
region. The coating on fiber could also protect the fiber from mechanical
degradation during processing and handling. Ideally, one would like to have an
interfacial coating that does not interact with either the fiber or the matrix and
is stable at the fabrication temperature and time as well as during the service
conditions of time and temperature. Clearly, coating thickness will be an
important parameter. Coating thicknesses in the range of 0.1 to 1 Ilm are
commonly used.
(a)
SiC
Whisker
(b)
Fig. 9.13 (a) Schematic of crack bridging by whiskers. (b) A silicon carbide whisker
bridging a crack in alumina matrix. (Courtesy of P.F. Becher). (c) Toughness
enhancenment obtained by silicon carbide whisker reinforcement of different matrix
materials (after Becher, 1990).
S
?d ....0 .iKfb+.iKPO
p.,
~
.,<Ii' 00
=
<U
.c:
01)
;::!
0 -.a
1-<
....
<U
....<.>
;::!
..,.
....d
.... .iKfb
.~ N
'"
0
Q.
S
0 0
U
0 10 20 30 40
SiC Whisker Content, Volume %
(c)
Fig 9.13 -contd.
In chapter 8 we discussed the subject of thermal stresses in composites and

pointed out that it is almost impossible to eliminate the thermal expansion (or
contraction) mismatch between the components of a composite material. What
one can do is to exploit this inherent characteristic of a composite material to
obtain a desirable thermal stress distribution in the composite by choosing the
components such that the residual stress pattern existing at the end of the
processing is one that gives the desired final characteristics in the composite.
Specifically, in ceramic matrix composites, one would like to enhance crack
deflection at the fiber/matrix interface by making it easy for the interface to
debond. Such debonding and crack deflection at the interface are prerequisites
for fiber pullout, all of which contribute to enhanced toughness in the
composite. The type of interfacial bonding (chemical or mechanical) as well as
the nature and magnitude of thermal stresses remaining at the end of processing
of a composite will be of great importance in this regard. Since the choice ofthe
matrix and fiber components of a composite is dictated by application, one
frequently incorporates a fiber coating, appropriate for a given composite
system, to optimize the stress distribution and bonding at the fiber/matrix
interface. In ceramic matrix composites, suitable interfacial coatings are chosen
such that the bonding is primarily mechanical in nature. In a mechanically
bonded interface, the degree of interfacial roughness becomes an important

parameter. A general approach of interface engineering in continuous fiber
reinforced ceramic matrix composites is illustrated in Fig. 9.14. Compatibility
(chemical, mechanical, thermal, etc.) between constituents is of obvious
importance. Various attempts have been made involving continuous carbon and
silicon carbide fiber reinforced carbon and silicon carbide matrix composites.
These have resulted in favorable mechanical properties at room temperature or
at high temperatures in the absence of air. However, all these nonxide materials
invariably suffer from lack of oxidation resistance at high temperatures.
Therefore, nonoxide CMCs can on be used for very short durations at high
temperatures and in oxidizing atmospheres. A possible alternative is the use of
oxide fiber reinforced oxide matrix composites. In this regard, a suitable matrix
material can be alumina because of its high temperature stability in air. The
fibers for such composites should have high strength, good creep resistance, and
favorable thermal shock characteristics. Polycrystalline or single crystal fibers
of the alumina-silica chemical system best achieve such characteristics. To
prevent strong interfacial bonds between the fiber and matrix, there is a very
crucial need for interface engineering of these composites via interfacial
coatings. Examples of a carbon coated Nicalon fiber and a tin dioxide coated
alumina fiber (Saphikon) are shown in Fig. 9.15 (a,b). Interphases such as C or
BN have the desirable feature of a layered structure. If coatings of C or BN are
deposited in a manner that the easy cleavage planes lie parallel to the
fiber/matrix interface then they will be very propitious in deflecting a crack
propagating perpendicular to the interface. There is a paucity of oxide
interphase materials. We describe these results in this section.
A number of researchers (Bender et aI., 1984, 1986; Lowden and Stinton, 1988;
Brun and Singh, 1989; Ha and Chawla, 1993; Ha et aI., 1993; Chawla et aI.,
1993, 1997,2000; Lewis, 2001; Lowden, 1991; Morgan and Marshall, 1993;
Naslain et aI., 1991) have pursued this route to obtain tough ceramic
composites. Bender et ai. (1984) observed that BN-coated (Nicalon) SiC fibers
in a zirconia based matrix (Zr0 2 + 50 mol% Si02 and zr02 + 50 mol% Ti0 2)
resulted in substantial improvements in strength and toughness of the
composites. The BN coating prevents the strong bonding between the silica
reaction layer in the matrix and the fiber. Composites made with uncoated
fibers showed a very brittle behavior. Improvement in room temperature
toughness of a Nextel 480 (essentially mullite) in a glass matrix by
incorporating a BN coating at the interface was observed by Vaidya et
aI.(1992). Figure 9.16 (a,b,c) taken from their work shows the difference in the
fracture behavior of the composites containing uncoated Nextel480 (mullite)
(M~~ ~ Chemical Thermal ~

~ ~ Compatibility ~mpatib~
Selection of Composite System
Design of Interface Microstructure
Interphase Incorporation
Characterization of Interlaces
------"---
I'ropeliy I:valuation of the
Lngineered Composite
Fig. 9.14 General approach of interface engineering in continuous fiber reinforced

ceramic matrix composites.
fiber(brittle, planar fracture) and those containing BN-coated fibers (extensive

fiber pullout). The hexagonal-BN coating in this case, made by CVD, had a
thickness of 0.2 Jlm and a nanocrystalline structure as seen in a transmission
electron microscope, Fig. 9.17. In another case, however, the hexagonal BN had
a graphitic layered structure. Figure 9.18 shows a BN coated Nicalon fiber in
a silicon carbide matrix, with the inset, a high resolution TEM micrograph,
showing the layered structure. Ha et al. (1992, 1993) used rather thick BN
coating, about 1 Jlm, as well as duplicate SiC/BN coatings to obtain tough
mullite fiber/mullite matrix composites. The idea of using a thicker or double
coatings is to use a part of the coating as a sacrificial material during the high
temperature processing, generally by hot pressing. In the duplicate coating, SiC
protects BN during processing. Nicalon fiber reinforced alumina composites
showed degradation of fiber strength at elevated temperatures (Celemin et aI,
1997)
Gouadec et al. (2001) used Raman spectroscopy to characterize Hi-Nicalon

fibers, coated with BN and SiC coatings, embedded in celsian matrix, which is
a refractory barium aluminosilicate (melting point -1760 QC). Figure 9.19
summarizes their results. The BN layer provided a weak, easily cleavable
region while SiC provided oxidation resistance for BN. Raman spectra were
recorded with a 514.5 nm excitation wavelength across the cross-section of
fiber, interphases, and matrix. The Raman spectra allowed detection of
SiC COATING
(a)
Fig. 9.15 Examples of coated fibers: (a) carbon coated (Nicalon) fiber (courtesy ofR.
Lowden) (b) tin dioxide (white layer) coated alumina (Saphikon) fiber, slightly oval in
shape.
(b)
Fig. 9.15 -contd
monoclinic and hexagonal polymorphs of the celsian matrix and free silicon in
the SiC region. It was proposed that the following reaction occurred at the
interface:
BN + SiC -» Si + BN-C
BN-C represents some kind of a solution of carbon in BN.
It should be pointed out that what is important for high temperature use is the
stability of the coating at the service temperature. In this regard, carbon
coatings in air and at moderately high temperatures will not be suitable. Boron
nitride coating also tends to oxidize in air at about 600°C or so, i.e., at
temperatures slightly above those for carbon. Silicon carbide coatings tend to
form silica at temperatures above 1200 °C. Lowden (1991) studied the effect of
different thicknesses of graphitic carbon interfacial coating in NicaloniSiC
composites. Carbon coatings were deposited isothermally from an argon and
(a)
(b)
Fig. 9.16 Difference in the fracture behavior of a composite containing uncoated

Nextel480 (essentiallymullite composition) fiber and one containing BN-coated fibers.
(a) brittle planar fracture in the uncoated fiber composite, (b) and (c) show extensive
fiber pullout and multiple matrix cracking in the coated fiber composite.
(c)
Fig. 9.16 -contd
polypropylene mixture at 1100 °C and 3.3 kPa pressure. By varying the

deposition time, coatings of thicknesses in the range of 0.1 to 1.0 f.1m were
obtained. The composite was fabricated by the forced-flow, thermal-gradient
chemical vapor infiltration described in chapter 4. Some of the samples with
0.52 f.1m carbon coating were oxidized to remove the carbon coating. Figure
9.2 Oshows flexure load-displacement curves for these composites with varying
carbon thickness. The uncoated fiber reinforced composite, as expected,
showed poor strength and a brittle fracture devoid of any fiber pullout. The
specimens subjected to oxidation treatment also showed low strength, but they
showed high strain to failure. Incorporation of carbon coating resulted in a
graceful failure and the phenomenon of fiber pullout came into play. An
increased fiber pullout length occurred with increasing carbon coating
thickness. The inverse relationship between interfacial shear stress and the
carbon coating thickness is shown in Fig. 9.21. The uncoated fiber composite
with strong chemical bonding showed the highest interfacial shear strength
Fig. 9.17 CVD Hexagonal-BN coating on a fiber, showing a nanocrystalline structure.

TEM. (courtesy of 3M Co.).
while the composite with an oxidized interlayer showed no interfacial bonding

because the oxidation of carbon produced a gap between the fiber and the
matrix. In the NicalonlSiC composite, the coefficient of thermal expansion of
matrix is greater than that of the fiber, resulting in a radial compressive stress
at the interface, making the sliding of the fiber difficult. Incorporation of a low
modulus interfacial coating, such as carbon, can result in a reduction of thermal
stresses. The carbon coating deforms in response to the thermal stresses
generated on cooling from the processing temperature. Thicker carbon coatings
are able to reduce the thermal stresses more than thinner coatings, hence the
inverse relationship between interfacial shear strength and the coating
thickness.
Now let us look at some examples of interface engineering in oxide/oxide

composites. Among the continuous oxide fibers available for use as
reinforcements for ceramic or glass matrices, alumina type fibers are the most
promising because they possess high stiffness, adequate strength, and excellent
thermal stability in oxidizing atmospheres even at elevated temperatures.
Consider the alumina type fiber/glass (silica-based) matrix system, an
oxide/oxide composite system. In the alumina/silica system, a series of
intermediate compounds can form (Aksayand Pask, 1975), leading to a strong
chemical bond at the fiber/matrix interface (Michalske and Hellmann, 1988;
Fig. 9.18 BN coated Nicalon fiber in a silicon carbide matrix, with the inset, a high resolution TEM micrograph, showing the layered
structure of the BN coating (Lowden and Stinton, 1988)
BN SiC Celsian Matrix

20000
18000
16000 • Free silicon
i'
=
=0
14000
-<>-- SiC TO mode
-::, 12000
,e. 10000 -4- SiC LO mode
';J
= 8000
...i 6000
--<>- HexaCelsian
4000 -0-- MonoCelsian
2000
0
-4 -2 0 2 4 6
Displacement perpendicular to the fibers (/-lm)

Fig. 9.19. Raman spectra across the cross-section of fiber, BN and SiC interphases, and
celsian matrix. Note the presence of monoclinic and hexagonal polymorphs of the
celsian matrix and free silicon in the SiC region (after Gouadec et aI., 2001)
300
250
200
...... -
~
0 150
....._-.\
«
0
-'
OXIDIZED
...
"'\ TO 0.13
100
'.... -...... .-........
'---
50
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.Q7
DISPLACEMENT (em)
Fig. 9.20 Flexure load-displacment curves for Nicalon/SiC composites with varying
carbon thickness. The uncoated fiber composite showed poor strength and a brittle
fracture devoid of any fiber pullout. The oxidation treatment resulted in low strength.
Incorporation of carbon coating resulted in a graceful failure (Lowden, 1991).
.
500
Il. .,,- .,,-

~ 400 ./
en
en
w
a:
I-
en 300
a:
«
w
J:
en
...J
200
~
u
«
LL
a:
w 100
I-
:!!:
0
0 2 4 8 10 12
Fig. 9.21 The inverse relationship between interfacial shear stress and the carbon
coating thickness for the NicalonlSiC composites (Lowden, 1991).
Maheshwari et aI., 1989). Strong interfacial bonding results in a brittle failure

of the composite similar to that of monolithic ceramics. There are, however, a
few oxide/oxide systems of simple eutectic type, such as AI20 3-Sn02 where
there is no solubility between the components (Barczak and Insley, 1962), that
can be exploited to produce tough composites that are stable in air. In the case
of alumina/silica systems, solubility studies indicate that tin dioxide has low
solubility in silica (56) and it has no mutual solubility with alumina up to 1400
°C (Barczak and Insley, 1962). These results were confirmed by quantitative
electron microprobe work (Maheshwari et aI., 1989). This low solubility would
be expected to promote relatively weak bonding between the fiber and coating,
and thus produce conditions suitable for enhanced toughness. Chawla and
coworkers (Chawla et aI., 1993, 2000) applied this interface engineering
approach to this alumina/glass system. They used two types of fibers: an
alumina + 15 wt. % zirconia fiber, PRD-166, and a single crystal alumina fiber,
Saphikon. Both react with the silica-based glass matrix and thus produce a
strong chemical bond. Tin dioxide coating on these fibers acts as a diffusion
barrier and prevents the chemical reaction. The bonding at the
(alumina+zirconia) fiber/Sn0 2 interface is purely mechanical while that
between Sn02 and glass is chemical cum mechanical (Michalske and Hellman,
1998; Maheshwari et aI., 1989). The big difference between these two fibers,
however, is the degree of surface roughness. Figure 9.22 (a and b) shows the
nature of the interface in PRD-166 fiber/tin dioxide /glass matrix and
Saphikonltin dioxide/glass matrix composites, respectively. The degree of
interfacial roughness is quite high at the PRD-166 fiberlSn0 2 interface as

compared to Saphikonltin dioxide interface. The thermomechanical
characteristics of the components of the two systems are given in Table 9.3. A
general expression for the radial stress due to the thermal mismatch between the
components and the compressive stress developed at the fiber/matrix interface
due to fiber surface roughness is as follows (Kerans and Parthasarthy, 1991):
where .1.T is the temperature change during cooling, q is an adjustable

parameter (equal to 1 for an infinite matrix), A is the amplitude of roughness,
and r is the radius of the fiber. Figure 9.23 relates the roughness amplitude, A,
to interfacial sliding. Figure 9.24 shows two different measures of roughness
amplitude: root mean square roughness and peak to valley roughness. The peak
to valley roughness amplitude, A, of PRD-166 and Saphikon fiber was
evaluated from tracings of the SEM micrographs shown in Figs. 9.22
respectively. SEM micrographs and surface roughness profiles of three alumina
fibers, PRD-166, Nextel610, and Saphikon fibers are shown in Fig. 9.25. Here,
we have included the surface trace and an SEM micrograph of a third alumina
fiber, a polycrystalline fiber, Nexte1610. This fiber, although polycrystalline,
is very smooth because of its extremely fine grain size. In general, the surface
roughness of a polycrystalline fiber scales as its grain size. The values of Air
for PRD- 166 fiber and Saphikon filament are given in Table 9.4. The Air value
ofPRD-166 fiber is about nine times greater than that of Saphikon fiber. We
can qualitatively compare the effects of interfacial roughness and thermal
stresses by comparing the magnitudes of the following two terms (from
Eq.(9.17)):
Table 9.3 Thermomechanical characteristics ofPRD-166 fiber, Saphikon fiber, glass

matrix and tin dioxide.
Material Young's Thermal Expansion Poisson's

Modulus, E Coefficient, a Ratio, v
CGPa) 0(}"60C- 1)
Alumina (PRD-166) 380 9.0 0.25
Sn02 233 5.23 0.25
N51A Glass 70 6.0 0.25
Saphikon 380 9.15 (parallel to c-axis) 0.25
7.94 (perpendicular to the c-
axis)
(a)
(b)
Fig. 9.22 (a) A longitudinal section of the composite showing the rough nature of the
PRD-166 fiberlSn0 2 interface; (b) A longitudinal section of the composite showing the
relatively smooth Saphikon fiberlSn0 2 interface. F, C, and M indicate fiber, coating,
and matrix respectively. SEM.
(a) (b) (c)
Fig. 9.23 A schematic of the interface roughness (periodic) with an amplitude A. f and
m denote the fiber and the matrix, respectively.
TA
..... l
Rrms = Root Mean Square Roughness _ j"t.. X2
Rp" = Peak to Valley Roughness _ :z ji." z2

=
where n number of peaks or valleys
x = ordinate of peak or valley
R. = Mean Roughness L
tf /(z) tbt
where
•
L = length of roughness curve
f(x) =roughness curve
Rmax = Maximum Roughness
difference between the highest and lowest point
Rz =Ten-Point Average
Fig. 9.24 Root mean square and peak to valley roughness amplitude, A, of the fiber.
- interfacial roughness strain --;. AIr

- thermal strain --;. (elf - tlm) /j. T
The values obtained for both PRD-166 fiber/tin dioxide/glass matrix and
Saphikonltin dioxide/glass matrix composites are shown in Table 9.5. It is
interesting to note that in PRD-166 fiber/tin dioxide/glass matrix composites the
compressive radial strain induced due to fiber roughness is more than an order
of magnitude greater than the thermal strain. In Saphikon/tin dioxide/glass
matrix composites, the roughness induced strain is only three times greater than
the thermal strain. This is advantageous since some bonding is required to
maintain the integrity of the composite.
Fracture surface features help elucidate the effects of interfacial roughness and
thermal stresses in these composites (Venkatesh et aI., 1992; Chawla et aI.,
1993). The fracture surface of a PRD-166 fiber/glass composite, Fig.9.26,
shows a characteristically flat, brittle fracture. Figure 9.27 shows the fracture
surface ofPRD-166 fiber/tin dioxide/glass matrix composites. Note the partial
debonding of the fibers from the matrix and the characteristic rough fiber
surface that results in strong mechanical bonding, see Fig. 9.27. Figure 9.28
(a,b) shows the fracture surface of Saphikonltin dioxide/glass matrix
composites. Note the fiber/coating debonding and the neat fiber pullout at the
alumina/tin dioxide interface. This is as expected since alumina and tin dioxide
show no solid solubility which precludes the chemical bonding but more
conspicuously the smooth interface results in a weak mechanical bonding.
Major toughening mechanisms in case system include crack deflection,
fiber/matrix debonding, and fiber pullout. Similar effects of interfacial
roughness were verified in Saphikon fiber/ lanthanum phosphate/alumina
matrix composites (Chawla et aI., 2000). We digress here a bit to mention the
usefulness of atomic force microscope in monitoring the fiber surface
morphology and to obtain the roughness profiles (Chawla et aI., 1993; Chawla
et aI., 1995). Very briefly, in an AFM, the probe tip in the form of a
microcantilever is placed a few nm from the sample surface. The cantilever tip
is deflected by the interatomic force between the atoms on the tip and the atoms
on the sample surface. A laser beam measures the deflection of the tip. The
deflection of the cantilever tip monitored dn used in a feedback loop. An
example of the surface profile ofNicalon fiber is shown in Fig. 9.29.
An analysis of the processing induced thermal stresses in uncoated PRD-166

(alumina-zirconia) fiber/tin dioxide/glass matrix composite and Saphikon fiber
coated with tin dioxide/glass matrix composite is shown in Figs. 9.30 and 9.31
respectively. Here we have used the equations derived in chapter 8. The
Saphikon
1 .. m
Fig. 9.25 SEM micrographs and surface roughness profiles of PRD-166, Nextel 610,
and Saphikon fibers.
Fig. 9.26 Fracture surface of uncoated PRD-166 fiber/glass composite showing a flat,
brittle fracture.
Fig. 9.27 Fracture surface of PRD-166/SnO/glass composite showing a non-planar

fracture surface but only a partial debonding at the fiber/matrix interface. Note also the
rough fiber surface.
distribution of thermal stresses in alumina (PRD-166) fiber/glass matrix and

alumina (PRD-166) fiber/SnO/glass matrix composites, for a I1T = -450°C is
shown in Fig. 9.30, (a) radial stresses, (b) tangential stresses, (c) axial stresses,
while that of Saphikon fiber/SnO/glass matrix composites, for the same
temperature interval is shown in Fig. 9.31, (a) radial stresses, (b) tangential
stresses, (c) axial stresses. This analysis shows radial tensile stress components
at the fiber/coating and coating/matrix interfaces, which is desirable from the
toughness point of view. However, this must be taken with the roughness
induced radial stress to determine the net radial stress existing at the interface.
It turns out that the roughness induced radial stress is much greater in PRD-166
fiber/tin dioxide glass matrix composite as compared to Saphikon/tin
dioxide/glass matrix composites. The effect of fiber roughness affects the
fracture behavior of these composites. Crack deflection is the predominant
toughening mechanism in PRD-166/tin dioxide/glass matrix composites while
fiber debonding and subsequent fiber pullout are the major toughening
mechanisms in Saphikon/tin dioxide/glass matrix composites. One can further
conclude that in order to produce tougher CMCs, fiber roughness should be
included as a design parameter along with thermodynamic stability and
mechanical compatibility between fiber and matrix.
Table 9.4 Roughness amplitude (A), fiber radius (r), and the ratio (Air) for PRD-166
and Saphikon fibers.
PRD-166 Sapphire
A, pm 0.185 0.0077
r, /.lm 10 40
Air 0.0185 0.0002
Table 9.5 Comparison of roughness induced strain (Air) and thermal strain (~a ~T)
in PRD-166 fiberlSn0 2 and SaphikonlSn02 interface
PRD-1661Sn02 interface Saphikon/Sn02 interface
~a~T 0.0014 0.0009
Air 0.0185 0.0002

(a)
(b)
Fig. 9.28. Fracture surfaces ofSaphikon/SnO/glass composite. Note the fiber/coating
debonding and extensive fiber pullout at the fiber/coating interface.
200
(nm) 100
o o
2
(jim)
Fig. 9.29 Surface texture of a Nicalon fiber. AFM. (Courtesy ofN. Chawla)
120
100
l PRD-166 FIBER
e
til
80
til
~ 60
1;;
....l
~ 40
Q
;,2 20 COATING
o~ ______+-______ ~ ______ ______

~ ~~~~~
o 5 ~
DISTANCE FROM THE CENTER OF THE FIBER (JIm)
(a)
100
..•••••........••••••....
~ 0~
c..
~
;; -100 ~
til
~
PRD-166 FIBER -- ~
MATRIX
c=::
ti; -200 ~
f-<
Z
~ -300 "-
(j
Z COATING- ~
~ -400 ~
-500
II I I
o 5 10 15 20 25
DISTANCE FROM THE CENTER OF THE FIBER (flm)

(b)
Fig. 9.30 Distribution of thermal stresses in alumina (PRD-166) fiber/glass matrix and
alumina (PRD-166) fiber/SnO/glass matrix composites, for a AT = - 450°C. (a) Radial
stresses (b) Tangential stresses (c) Axial stresses.
400
.... -.-_ .... _--------_ ....

~ 200 f-
6 PRD-166 FIBER
<Zl
<Zl
MATRIX
~ 0 f-
f-;
<Zl
.....l
-- .....••.•••...........•......
<
~
< -200 r- COATING- ~
l.-
I I I I
-400
o 5 10 15 20 25
DISTANCE FROM THE CENTER OF THE FIBER (/lm)

(c)
Fig. 9.30 - contd
60 SAPHIKON FIBER MATRIX

~
~
6
<Zl
............... _-- ...........
<Zl
~
40
~
f-;
<Zl
....l
<t:
......
Q 20
<
~ COATING
OL-____- L________ ~U- ____ ~ ________ ~ __ ~~~
o 20 100

(a)
Fig. 9.31 Distribution of thennal stresses in Saphikon fiberlSnO/glass matrix

composites, for a i1T = - 450°C. (a) Radial stresses (b) Tangential stresses (c) Axial
stresses.
-- ---
---
0-
(;j'
Cl... .
6
CI)
-100 -
CI)
~ SAPHIKON FIBER MATRIX
~
f0-
CI)
....
...... -200 -
-<
fo-
Z
~
"-<
Z
fo-
-300 - COATING---t
-400
o 20 40 60 80 100
DISTANCE FROM THE CENTER OF THE FIBER (Jlm)

(b)
300~~~~~~~~~n---------------------~
200-
(;j' 100 ~
Cl...
6 SAPHIKON FIBER MATRIX
CI)
CI)
Or-
~
~ -1oo~
CI)
....
-< -200r-
~ COATING---t
-3001-
_4ooL-______~I--------~I-=~----~I~-------~I------~
o 20 40 60 80 100
(c)
Fig 9.31 - contd.

9.3.1 Coatings involving a phase transformation with volume change
Coatings with compositions that have multiple polymorphs wherein

transformations can be triggered by shear stress have been explored (Kriven,
1995). Under shear stress from an approaching crack tip, a transformation is
induced in the coating, causing a volume or shape change at the fiber/matrix
interface. This is illustrated in Fig. 9.32. Energy from the crack is then
dissipated in a variety of ways:
• Dissipation of crack energy through nucleation of the phase

transformation.
• Possible autocatalytic nucleation of transformation along the
interface promoting long fiber pullout or interfacial debonding.
• Increased surface energy contribution from interfacial microcracking
or shearing from the transformation.
These coatings also offer advantages difficult to obtain with other interphase
types described above. One unique difference is the ability to use a coating with
strong bonding to the fiber and matrix, and thus retain overall good composite
strength, but possessing the ability to convert to localized weak bonding in front
of a crack tip. An example of such a coating is enstatite (MgSi03) (Zhu and
Kriven, 1996). The polymorphs of enstatite and the conditions for
transformation between them are shown in Fig. 9.33. This material undergoes
a shear-induced proto to clino destructive transformation at 96 MPa
accompanied by a molar unit cell volume shrinkage of 5.5% and
crystallographic unit cell shape change of 18.3 0 • Such composites were
fabricated by a co-extrusion method to produce multi-filament arrangements of
fibrous titania (Ti0 2) surrounded by enstatite and then polycrystalline titania.
Three-point flexure tests showed a work of fracture of 1000 J/m2 for the work
of fracture associated with transformation-weakened interfaces. Fractography
revealed pullout of fibrous regions as well as significant fracture along the
transformation weakened interface regions in the annealed specimens.
9.3.2 Magnetoplumbites and other interphases having a layered structure
Magnetoplumbites are yet another class of materials that may provide some
suitable coatings (Cinibulk, 1994). As mentioned earlier, a layered
microstructure with weak bond between layers allows for easy cleavage.
Common examples are, of course, pyrolytic carbon and hexagonal boron
carbide. Compounds having p-alumina or magnetoplumbite structure have such
an anisotropic layered oxide structure. These layered oxides can be regarded as
MATRIX
IH li'l~
TRANSFORMABLE
OXIDE INTERPHASE
CRACK
Fig. 9.32 Under shear stress from an approaching crack tip, a transformation is induced
in the coating, causing a volume or shape change at the fiber/matrix interface (after
Kriven, 1995)
Protoenltatite PE
(orthorhombic>
~ 865"C
fait quenching
4V .. ·5.591.
104~
slow cooling
.~ clinoenstatite C E
(monoclinic)
shearing under
high pressures and
temperatures {2:SOOoC>
Orthoenstatite 0 E
(orthorhombic)
Fig. 9.33 Polymorphs of enstatite and the conditions for transformation between them
(after Kriven, 1995).
structural analogs of C and BN. Potentially, such layered oxides, capable of

easy cleavage along basal planes can be attractive as interfacial coating
materials.
Cinibulk(1994) used a magnetoplumbite, CaAI 1Z0 19 (hibonite) as a coating

material on alumina fibers, which were then consolidated in an Alz03 or
Y3AI50 12 (YAG) matrix by hot pressing. A strong texturing with basal planes
oriented along the fiber surface was observed in the CaO-doped YAG matrix
composite. Such a texture may deflect cracks along the easy-cleaving basal
planes. Lanthanum hexaluminate (LaAIl10 1s) is another oxide with a
magnetoplumbite structure that has the potential to be used as an interfacial
coating in oxide fiber reinforced mullite matrix composites. Some work with
lanthanum hexaluminates has involved additions of divalent cations (Ca2+,
Mn2+) to stabilize the magnetoplumbite structure in dense and porous coatings
of lanthanum hexaluminate on Nextel 610 and SCS-O fibers. A possible
limitation of a p-alumina compound is its thermal stability at high temperatures
due to any loss of alkali, e.g., loss ofNazO during heat treatment of p-alumina
is well known. Sambasivan et al. (1994) used large alkali atoms, such as
rubidium, to stabilize the p-alumina structure. Rubidium stabilized p-alumina
has been suggested as a potential interface coating for oxide/oxide CMCs.
Sambasivan et al. observed morphologically stable coating in a hot pressed
Saphikon fiber/alumina matrix composite. Indentation tests showed interface
debonding but pushout tests did not show interfacial debonding. Another
possible candidate is KCazNb301O, which shows easy fracture along basal
planes. Indentations in bulk KCazNb 30 lO produced cracks parallel to the basal
planes, see Fig. 9.34. However, on making composites by hot pressing in a
reducing atmosphere, the authors observed a loss of rubidium.
Fiber coatings of phyllosilicates or micas have also been tried. They are,
however, unstable above 700°C because the hydroxyl (OH-) groups in their
crystal lattice decompose. The thermal stability of phyllosilicates can be
improved by eliminating the OH- groups. A modification of phyllosilicates,
called phyllosiloxides, having a mica like layered structure but without the OH-
groups has been suggested. Schematic descriptions of mica and a phyllosiloxide
are shown in Figure 9.35.
Yet another method to create multiple weak layers at the fiber/matrix interface
involves interfacial reactions during processing. Unlike the earlier described
interphase materials, in which reactions between the interphase and fiber or
matrix are not desirable, this method actually utilizes reactions to form
multiple, weak, and stable interface layers between the fiber and matrix. An
Fig. 9.34 Indentations in bulk KC~NbPlo showing cracks parallel to the basal planes.
(Courtesy of S. Sambasivan)
602- 602 -
2K+ 2K+
602 - 602 -
3Si4 + + lA1 3+ 4Si 4+
.........................
402- + 20H- 602 -
------------------ 6Mg 2+
------------------ 4Mg2+ + 2Ae+
40 - + 20H-
2 60 -
2
3Si4 + + lAtl+ 4Si4 +

602- 602 -
2K+ 2K+
602- 602-
KMg3(ShAl)OIO(OH)Z KMgzAlSi40 12
(a) (b)
Fig. 9.35 Structure of (a) micaphlogopite and (b) related phyllosiloxide (after Reig et
aI., 1995).
example of an interphase material that will form multiple interfaces through

reaction with the fiber/matrix is BaZr03 in an alumina/alumina composite.
BaZr03will react with alumina at temperatures around 1475 °C to form a series
of oxide compounds: Zr02 ; barium monoaluminate,BaO .AI20 3,and Ba-~-AI203,
which has the composition ofBaO·7.3 Al 20 3 and has a layered type structure
(Chen et aI., 2002). The effectiveness of these interfacial layers in providing
crack deflection is shown in Fig. 9.36, which shows crack propagation through
a laminated A1 20 3- BaZr03 composite.
9.3.3 Porous coatings and porous matrices
Another possibility is to use porous materials, which is based upon the

assumption that the fracture energy of a porous material is lower than its fully
dense counterpart. Consequently, the lower fracture energy of the porous
materials would allow debonding within the interfacial layer itself. In this case,
there is the again problem of inhibiting any tendency of pores to agglomerate
or be removed through densification at high temperatures. Such a porous
Fig. 9.36 The effectiveness of interfacial layers formed during processing of

BaZrO/alumina laminated composite in providing crack deflection (Gladysz et aI.,
1998).
coating, if weak in the as-coated condition, would be also be quite susceptible

to severe damage during composite processing. An additional problem to
consider with porous coatings would be the nature of the interfacial roughness
from the fracture path through the porous interface. It is quite likely to be
extremely rough, which could hinder fiber pullout.
Porosity has also been used in the matrix material of ceramic matrix composites
to improve damage tolerance (Levi et aI., 1999). The notch insensitivity ofthese
porous composites is claimed to be similar to that of conventional ceramic
matrix composites, but they have a failure strain greater than that of
conventional monolithic ceramics. Processing is easier because a high density
is not required and no interphase is needed between the fiber and matrix. Such
composites with a porous matrix have a failure mode different form
conventional ceramic matrix composites. Matrix cracking occurs throughout the
entire loading range with the tensile stress/strain behavior almost linear to
failure.
We should mention here an interesting development in porous oxide composites

(Kanka and Schneider, 2000) called WHIPO)(lM (Wound Highly £orous Oxide
CMC). One and two-dimensional composites have been developed with highly
porous mullite matrices (porosity> 60% by volume) containing oxide fibers
(Nextel 610, 650, and 720) by a computer controlled filament winding
technique. The fiber bundles were infiltrated with an aqueous slurry of matrix.
The fiber rovings were pre-dried in a continuous microwave furnace, which
allowed a high matrix content between fibers and avoided fiber-to-fiber
contacts. The infiltrated yarns were wound using a commercial winding
software on a plastic mandrel in an enclosure with constant humidity. The fiber
volume fraction of the as wound components varied between 25 and 50%.
Rings and tubes of diameters between 20 and 250 mm and length (about 400
mm) were made by drying the green material on the mandrel, followed by
sintering in air at 1300° C for 60 min at a heating rate of 10° Cimino Figure 9.3 7
shows the microstructure of the composite obtained by this process.
Eutectic Oxide Composites
This is an interesting alternative route to create tough high temperature

ceramics. This method relies on the in situ formation of an oxide/oxide
composite by exploiting unidirectional so lidification of a eutectic. The resulting
composite can, under some circumstances, combine high strength combined
with ductility. Waku et al. (1997) investigated a unidirectionally solidified 77
Fig. 9.37 Microstructure of the composite obtained by WHIPOX process. (Courtesy of

H. Schneider).
atomic AIP3/23 atomic % GdAI0 3eutectic mixture. The microstructure of this

composite consisted of an interpenetrating network of single crystal Al20 3 and
single crystal GdAI03, i.e., there were no grain boundaries present. At 1600 °0
this composite showed plastic deformation due to dislocation motion, with a
flexural yield stress of about 695 MPa as measure in a three-point flexural test,
and thus showed a relatively high fracture energy in comparison to conventional
ceramic composites. Unlike the engineered fiber/matrix interfaces we are
discussing in this section, which are designed to be weak, it is believed that the
two phases in this material are bound together strongly at the interface. The
ductility in this system would appear to have its origin in the ability to deform
plastically at high temperature. Clearly, such dislocation-based plasticity would
not be available at room temperature.
REFERENCES
Aksay, LA. and J.A. Pask (1975), J. Amer. Ceram. Soc., 58, 507.
Aveston, J., G.A. Cooper, and A. Kelly (1973) in The Properties o/Fibre Composites,
Conference Proceedings, National Physical Laboratory, IPC Sci. & Tech. Press,
Guildford, Surrey, p. 15.
Aveston, J. (1973) in The Properties of Fibre Composites, Conference Proceedings,

National Physical Laboratory, IPC Sci. & Tech. Press, Guildford, Surrey, 1973, p. 63.
Barczak, V.J. and R.H. Insley (1962) .I. Amer. Ceram. Soc., 45,144.
Becher, P.F., C.-H. Hsueh, P. Angelini, and T.N. Tiegs (1988) .I. Amer. Ceram. Soc.,
71,1050.
Becher, P.F. (1990) Ann. Rev. Mater. Sci., 20,179.
Becher, P.F., E.R. Fuller, Jr., and P. Angelini, (1991) .I. Amer. Ceram. Soc., 74, 2131.
Becher, P.F. and G.C. Wei, (1984) .I. Amer. Ceram. Soc., 67, 259.
Bender, B., D. Lewis III, W.S. Coblenz, and R. W. Rice (1984) Ceram. Eng. Sci. Proc.,
5,513.
Bender, B., D. Shadwell, C. Bulik, L. Incorvati, and D. Lewis III (1986) Amer. Ceram
Soc. Bull., 65, 363.
Bhargava, P. and B. R. Patterson, (1997) .I. Amer. Ceram. Soc., 80, 1863.
Brun, M.K. and R.N. Singh (1989) .I. Amer. Ceram. Soc., 72, 1891.
Budiansky, B., J.C. Amazigo, and AG. Evans (1988) .I. Mech. Phys. Solids,36, 167.
Budiansky,B.,J.W. Hutchinson, andA.G. Evans (1986).1. Mech. Phys. Solids, 34, 167.
Celemfn, J.A., J.Y. Pator, J. Llorca, and M. Elices (1997) .I. Amer. Ceram. Soc., 80,
2569.
Chawla, K.K., C. Coffm, and Z.R Xu (2000) Inti. Mater. Rev., 45, 165.
Chawla, K.K., M.K Ferber, R. Venkatesh, and Z.R. Xu (1993) Mater. Sci. & Eng.,
AI62,35.
Chawla, K.K., H. Liu, J. Janczak-Rusch, and S. Sambasivan (2000).1. Eur. Ceram. Soc.,
20,551.
Chawla,K.K., Z.R. Xu, J.-S. Ha, M. Schmucker, and H. Schneider (1997) Applied
Composite Materials, 4,263.
Chawla, K.K., Z. R. Xu, A. Hlinak, and Y. W. Chung (1993) in Adv. in Ceramic Matrix
Composites. N.P. Bansal (ed.), Amer. Ceram. Soc., Westerville, OH, p. 725.
Chawla, N., J.W. Holmes, andJ.F. Mansfield (1995) Mater. Characterization, 35, 199.
Chen, Z., K.K. Chawla, M. Koopman (2002) Mater. Characterization, in press.
Cinibulk M.K. (1994) Ceram. Eng. Sci. Proc., 15,721.
Cook, J. and J.E. Gordon (1964) Proc. R. Soc. London, A228, 508.
Cox, B.N. (1990) Acta Met. Mater., 38, 2411.
Das Chowdhury, K., RW. Carpenter, and W. Braue (1992a) in Mater. Res. Soc.
Symposium, vol 238, p. 421
Das Chowdhury, K., RW. Carpenter,and M.J Kim (l992b) in Microscopy: The Key
Research Tool, Electron Microscopy Society ofAmerica, Woods Hole, MA" p. 61.
Donald, I.W. and McMillan, P.W. (1976) .I. Mater. Sci., 11,949.
Evans, AG. (1972) Phil. Mag., 26, 1387.
Evans, AG. and D.B. Marshall (1989) Acta Met. 37,2567.
Faber, K.T. and A.G. Evans (1983) Acta Met., 31, 565.
Gao, Y.-C., Y.-M. Mai, and B. Cottrell (1988) .I. App. Math. Phys., 39, 550.
Gladysz, G.M., M. SchmUcker, K.K. Chawla, H. Schneider, D.L. Joslin, and M.K.
Ferber (1998) Mater. Characterization, 40, 209.
Giannakopoulos, AE. and K. Breder (1991) .I. Amer. Ceram. Soc., 74, 194.
Gouadec, G., P. Colomban, and N.P. Bansal, (2001) J Amer. Ceram. Soc., 84,1121.
Green, D.J., R.H.I. Hannink, and M.V. Swain (1989) Transformation Toughening of
Ceramics, CRC Press, Boca Raton, FL .
Gupta, V. (1991) MRS Bulletin, XVI-4, 39.
Gupta, V., J. Yuan, and D. Martinez (1993) J Amer. Ceram. Soc., 76, 305.
Ha, J-S. and K.K. Chawla (1993) J Mater. Sci. Letters, 12,84.
Ha, J.-S., K.K. Chawla, and R. Engdahl (1992) Mater. Sci. and Eng., A161, 303.
He, M. Y. and J. W. Hutchinson (1989) J App. Mech., 56, 270.
Hutchinson, J.W. and H.M. Jensen (1990) Mech. Mater., 9,139.
Janney, M. (1987) Bull. Amer. Ceram. Soc., 66,322.
Jero, P.D., Parthasarathy, T.A, and Kerans, R.J. (1992) Ceram. Eng. Sci. Proc., 13, 54.
Kerans, R.J. and T.A. Parthasarathy (1991) J Amer. Ceram. Soc., 74,1585.
Kanka, B. And Schneider (2000) J Eur. Ceram. Soc., 20, 619.
Kriven.W.M. (1995) J Physique IV 5, C8-101.
Krstic, V.V., P.S. Nicholson, andR.G. Hoagland (1981)J Amer. Ceram. Soc., 64,499.
Lange, F.F. (1970) Phil. Mag., 22, 983.
Levi, C.G., F.W. Zok, J.-y' Yang. M. Mattoni, and J.P.A. L5fvander (1999) Z.
Metallkunde, 90, 12.
Lewis, M.H. (2001) in High Temperature Ceramic Matrix Composites, Wiley-VCH,
Weinheim, Germany, p. 136
Lowden, R.A. (1991) inAdv. Camp. Mater., vol. 19, Amer. Ceram. Soc., Westerville,
OH, p. 619.
Lowden, R.A. and D.P. Stinton (1988) Ceram. Eng. Sci. Proc., 9, 705.
Maheshwari, A., K.K. Chawla, and Michalske, T.A. (1989) Mater. Sci. & Eng., AI07,
269.
Majumdar, A.J (1970)Proc. Roy. Soc. Land., A319, 69.
Marshall, D.B., B.N. Cox, and A.G. Evans (1985) Acta Met., 33, 2013.
Mendelson, M.1. and M.E. Fine (1974) J A mer. Ceram. Soc., 57, 154.
Michalske, T.A. and J. Hellmann (1988) J Amer. Ceram. Soc., 71, 725.
Morgan, P.E.D. and D.B. Marshall, (1993) Mater. Sci. and Eng., A162, 15.
Mumm, D.R. and K.T. Faber (July-Aug., 1992) Ceram. Eng. & Sci. Proc., 13,218.
Naslain, R., O. Dugne, A. Guette, C.R. Sevely, J. Brosse, J.-P. Rocher, and J. Cotteret
(1991)'/' Amer. Ceram. Soc., 74, 2482.
Parthasarathy, T.A., P.D. Jero, and R.I. Kerans (1991) Scripta Met., 25, 2457.
Ramualdi, J.P. and G.B. Baston (1963) Proc. ASCE, J Eng. Mech. Dn., 89,147.
Reig, P., G. Demazeau, andR. Naslain (1995) Eur. J SolidStateInorg. Chem., 32, 495.
Ruhle, M. and A.G. Evans (1988) Mater. Sci. and Eng., AI07, 187.
Rose, J.H., J.R Smith, and J. Ferrante (1983) Phys. Rev. B, 28, 1835.
Rose, J.H., J.R Smith, F. Guinea, and J. Ferrante (1984) Phys. Rev. B, 29, 2963.
Sambell, R.A.J., et. al (1972) J Mater. Sci., 7, 676.
Schoenlein, L.H., R.H. Jones, C.H Henager, Jr., C.R. Schilling, and F.C. Gac (1988)
Proc.: Materials Res. Soc. Symp., Spring 1988, April 5-9, Reno, NV.
Sambasivan, S., T. A. Parthasarathy, F. J. Scheltens, and R. J. Kerans (1994) Ceram.
Eng. Sci. Proc., 15,972.
Shetty, D.K. (1988) J. Amer. Ceram. Soc., 71, CI07.

Taya, M., S. Hayashi" A.S. Kobayashi, and H.S. Yoon (1990) J. Amer. Ceram. Soc.,
73,1382.
Todd, R.I., A.R. Boccaccini, R. Sinclair, R.B. Yallee, and RJ. Young (1999) Acta Met.,
47,3233.
Vaidya, R.U., J. Fernando, K.K Chawla, andM.K. Ferber (l992) Mater. Sci. and Eng. ,
A151.
Venkatesh, R. and K.K. Chawla (1992) J. Mater. Sci. Lett., 11,650.
Waku, Y., N. Nakagawa, T. Wakamoto, H. Ohtsubo, K. Shimizu, and Y. Kohtuko
(Sept. 4, 1997) Nature, 389, 48.
Wei, G.C. and P.F. Becher (1984) A mer. Ceram. Soc. Bull., 64, 298.
Wu S., B.R. Patterson, M.K. Ferber and E.R. Fuller, 1r. (2001) in Ceramic Engineering
and Science Proceedings, Vo1.22, No.3, 25th Annual Conference on Composites,
Advanced Ceramics, Materials, and Structures: A, Cocoa Beach, American Ceramic
Society, p. 269
Wu, S., B.R. Patterson, and M.K. Ferber (2001) in Advances in Ceramic Matrix
Composites VI, American Ceramic Society, 2001, p. 115.
Zhu D. and W. M. Kriven (1996) Ceram. Eng. Sci. Proc., 17, 383.
SUGGESTED READING
Chawla, K.K., C. Coffin, and Z.R. Xu (2000) "Interface Engineering in Oxide

Fibre/Oxide Matrix Composies," IntI. Mater. Rev., 45, 165.
Faber, K.T.. (1997) Ann. Rev. Mater. Sci., 27, 499.
Kim, 1.K. and Y. W. Mai (1998) Engineered Interfaces in Fiber Reinforced
Composites, Elsevier Science, Ltd. New York.
Cliapter 10
MACROMECHANICS
Laminated fibrous composites are made by bonding together two or more

laminae. The individual unidirectional laminae or plies are oriented in such a
manner that the resulting structural component has the desired mechanical
and/or physical characteristics in different directions. In this way, one can
exploit the inherent anisotropy of fibrous composites to design a composite
material having a desired set of characteristics such as elastic constants, thermal
expansion coefficients, etc. This has been employed quite extensively in
polymer matrix composites to design PMCs having highly tailored elastic,
thermoelastic, and strength characteristics, not so much in metal matrix and
ceramic matrix composites. Techniques such as tape-casting and hot pressing
oflaminae can be used to produce laminated CMCs (Prewo and Brennan, 1982;
Bhatt, 1991; Bhatt and Phillips, 1990; Amateau and Messing, 1990;
Velamakanni and Lange, 1991; Gladysz et aI., 1999). In this chapter, we
provide the reader the very basic mathematical tools to analyze such laminated
composites. For greater details on the mechanics oflaminated composites, the
references listed under Suggested Reading should be consulted.
In Chapter 6 we treated the micromechanics of composites, i.e., how to obtain

the properties of a composite when the properties of the matrix and the
reinforcement as well as their geometric arrangements are known. In particular,
we can use the information obtained from a micromechanical analysis of a thin,
unidirectionally reinforced lamina as input for a macromechanical analysis of
a laminated composite. Should there be a lack of such analytical information,
then the properties of a lamina must be determined experimentally. Once we
have determined, the characteristics of a fibrous lamina, analytically or
otherwise, we shall ignore its detailed microstructural nature and simply treat
it as a homogeneous, orthotropic sheet. A laminated composite is made by
stacking a number of such orthotropic sheets at specific orientations to get the
composite materials having appropriate properties. We then use the theory of
laminated plates to analyze such laminated composites.
356 MACROMECHANICS
10.1 PROPERTIES OF A LAMINA
Since a lamina is the basic building block of a laminated composite, we first

discuss the properties of an individual lamina made of a unidirectionally
aligned fibers. An orthotropic material, i.e., one having three mutually
perpendicular axes of symmetry, has nine independent elastic constants as
discussed in chapter 6. For an orthotropic material with the coordinate axes
parallel to the symmetry axes of the material, this array of elastic constants is
given by
S)) S)2 S)3 0 0 0
S22 S23 0 0 0
S33 0 0 0
[Sii] = (10.1)
S44 0 0
S55 0
S66
A lamina being a thin orthotropic material, we can take the through thickness
components to be zero. Then, we can write the stiffness matrix for an
orthotropic lamina by eliminating the terms involving the z- or 3-axis
(10.2)
Using this [Sjj] we can write Hooke's law for a thin orthotropic lamina, with
natural and geometric axes coinciding, as follows
£) S)) S)2 0 a)
£2 = S)2 S22 0 a2 (10.3)
£6 0 0 S66 a3
357 MACROMECHANICS
Conversely, we can write for stresses in terms of strains as
0'1 QIl QI2 0 EI
0'2 = Q 12 Q 22 0 E2 (l0.4)
0'6 0 0 Q66 E6
Here, we have used the symbol Qij rather than Cij for thin materials because of
convention. Qij are called reduced stiffnesses. Qij and Sij are related as follows
S22
QIl = 2
SII S 22 - S 12
S12
Q12 = 2
S11 S 22 - S 12
(10.5)
SII
Q22 = 2
S11 S22 - S 12
1
Q66 =S
66
(ij = 1,2,6)
Equations (10.3) and (10.4), showing terms with indices 16 and 26 to be zero,
represent a case of special orthotropy when the principal material axes of
symmetry coincide with the geometric axes, i.e., the fiber direction (1) and the
direction transverse to it (2) coincide with the principal loading directions. If
the material symmetry axes and the geometric axes do not coincide, then we
have the case of general orthotropy, a fully populated elastic constant matrix,
i.e., 16 and 26 terms are nonzero. These two cases are illustrated in Fig. 10. 1.
The constitutive equation in terms of general orthotropy can be written as
O'x Q 11 Q 12 Q16 Ex
O'y = Q I2 Q 22 Q26 Ey (10.6)
O's Q16 Q26 Q 66 Es

358 MACROMECHANICS
L x,1
(a)
y
2
(b)
Fig. 10.1 Two cases of orthotropy: (a) special orthotropy when the material symmetry
axes (1-2) and the geometric axes (x-y) do not coincide, (b) general orthotropy when
the principal material axes of symmetry (1-2) coincide with the geometric or principal
loading directions (x-y).
where the Qij matrix is called the transformed reduced stiffness matrix because
it is obtained by transforming Qij (specially orthotropic) to Qij (generally
orthotropic), In order to obtain Qij from Qij' the transformation of axes can be
performed as shown in Fig. 10.2. This shows a unidirectional lamina with axes
in 1-2 system not coinciding in the x-y system, Ifwe know the properties in the
1-2 system, we can determine them in the x-y system or vice versa. This
transformation of axes is carried out easily in the matrix form. A
counterclockwise rotation is taken to be positive. The transformation equation
for stress, in matrix form, is
(10.7)
359 MACROMECHANICS
2
y !
X
Fig. 10.2 Transformation of axes to obtain Qij from Qij (counterclockwise rotation is
positive).
while the equation for strain is
(l0.8)
es
where [TJa and [TJ. are the transformation matrices for stress and strain
transformations, respectively, and are given by
m2 n2 2mn
n2 m2 -2mn (10.9)
-mn mn m 2 -n 2
mn
-mn (10.10)
m = cos e and n = sin e

360 MACROMECHANICS
This method of using different transformation matrices for stress and strain
transformations is convenient because it avoids the need of putting the factor
112 before the engineering shear strains to convert them to tensorial strain
components suitable for transformation.
In order to obtain [ Q] from [Q], we proceed as follows. Multiply both sides of

Eq. (10.7) by [T],,·I. Note that [T],,[T],,·I = [I], the identity matrix. [T],,·I can be
obtained from [T]" by simply substituting -e for e. Thus,
m2 n2 -2mn
[Ttl = n2 m2 2mn (10.11)
where m = cose and n = sine. Stresses in the x-y system are then
(10.12)
Substituting Eq. (10.4) in Eq. (10.12), we obtain
(10.13)
Now we substitute Eq. (10.8) in Eq. (10.13) and get
(10.14)
where
361 MACROMECHANICS
(10.15)
[Q] is the stiffness matrix for a generally orthotropic lamina whose components
in expanded form are written as follows (m = cose, n = sine):
The reader should note that although Qij is a completely filled matrix, only four
of its components are independent: QI6 and Q26 are linear combinations of the
four, Qij.
Consider a generally orthotropic lamina wherein we have nonzero 16 and 26

terms, see Fig. 10.I(b). Ifwe apply a unidirectional normal stress ax to such a
lamina, we will have both normal as well as shear strains as responses and vice
versa: that is, there is a coupling between the normal and shear effects. In the
case of specially orthotropic lamina where the 16 and 26 terms are zero, normal
stresses produce normal strains and shear stresses produce shear strains and
vice versa. In other words, there is no coupling between the normal and shear
components in a specially orthotropic lamina. We shall present more
information about such coupling effects in laminated composites later.
10.1.1 Variation of Lamina Properties with Orientation
We shall now obtain the conventional engineering elastic constants referred to

the geometric axes, x-y (Ex, Ey; Gs' and vx) in terms of engineering elastic
constants referred to the material symmetry axes 1-2 (El> E2, G6, and VI).
Consider Eqs. (10.7) and (10.9) and let ax be the only nonzero stress
component. Then
362 MACROMECHANICS
a 1 =amx
2
(l0.17a)
a 2 =anx
2 (l0.17b)
a 6 = -axmn (10.17c)
Hooke's law gives us the strains in a lamina as
(10.18a)
(10.18b)
(10.18c)
e
1
= ax (m2
E
- v ~l
1 E (10.19a)
1 1
e=a(~-vm21
2 E 2 E
x
(l0.19b)
2 2
ax mn (10.19c)
e =---
6 G6
where we have used the relationship v/E t = viE2 derived in the next section.
Since we have the strain transformation given by Eq. (10.8), we can write the
inverse ofEq. (10.8) as
363 MACROMECHANICS
where [T]e- I can be obtained by substituting -9 for e in Eq. (10.l0).
In an expanded form, we have
ex = m 2e 1 + n 2a 2 - mna6
(10.20a)
ay = n 2a 1 + m 2a2 + mna6 f
(10.20b)
a,I' = 2(a 1 - a2)mn + e6(m 2 - n 2) (10.20c)
Substituting Eq. (10.19) in Eq. (10.20), we obtain
(10.21a)
(10.21b)
Now, Ex = a/ax by definition. Using this with Eq. (10.21a) we obtain
(10.22)
Ey can be obtained from Ex by substituting 9 + 90° for e in Eq. (10.22). Thus,

we can write
364 MACROMECHANICS
(10.23)
Similarly, from Eqs.(10.21) and (10.22) we obtain
Vx = _~ = _ 8y = 2 _(_1 + 2Vl + _1 __1_] m 2n 2

Ex Ex8x O'x El El El E2 G6
where Vx = -e.j8x when O'x is the applied stress.
Or
(10.24)
Similarly, it can be shown that
v =E IV2 + (_1 + _1 + 2Vl - _1 ] m 2n 2 ]

y y E2 El E2 El G6 (10.25)
Taking 0'. to be the only nonzero stress component, and noting that the in-plane
shear strain 8. = O'iGs' and applying Hooke's law, we obtain the following
relationship for in-plane shear modulus as a function of angle e.
1 + v2 - _I ] (10.26)
E2 G6
We can also define two shear coupling coefficients

88
m =--
y E
y
and
365 MACROMECHANICS
e,I'
mx =--
Going through the expressions given in Eq.(10.21), we can arrive at
and
Figure 10.3 shows, schematically, the variations of Ex, Ey, G., VX (= VXy) and Vy
(= vyx), and mx and my with fiber orientation 9. Various researchers have
examined laminated composites made by hot pressing of plies of Nicalon
fabric/glass matrix (Prewo and Brennan, 1982) or monofilament silicon carbide
fiber (SCS-6 type CVD fiber) in a reaction bonded silicon nitride (RBSN)
matrix (Bhatt, 1991; Bhatt and Phillips, 1990). Bhatt and Phillips (1990), in
particular, used the laminate theory and modified constitutive equations to
account for weak bonding in their SiC/RBSN composites, to predict elastic
properties of unidirectional, cross-ply, and angle-ply composites.
10.2 RELATIONSHIPS BETWEEN ENGINEERING CONSTANTS AND REDUCED

STIFFNESSES AND COMPLIANCES
In engineering practice, we commonly use constants such as Young's modulus,

E, shear modulus, G, bulk modulus, K, and Poisson's ratio, v. It is instructive
to obtain relationships between these and the reduced stiffnesses.
Consider the thin lamina shown in Fig. 10.4 with its natural axes coinciding
with the geometric axes. The conventional engineering constants in this case are
Young's moduli in directions 1, EI and direction 2, E 2 ; the principal shear
modulus, G6 ; and the principal Poisson ratio, VI. In an isotropic material, there
is only one Poisson ratio, because the contractions in the two principal
transverse directions are equal. In the case of an orthotropic lamina, there will
366 MACROMECHANICS
o 45 90
(a)
Ply angle, (10
o 45 90
Ply angle, (r
(b)
Fig. 10.3 Variation (schematic) of various elastic constants with fiber orientation e for
a unidirectional fiber composite. (a) Ex and Ey, (b) G" (c) Vx and vy' and (d) mx and my.
367 MACROMECHANICS
o
Ply angle, 0°
(c)
t
E
o 90
Ply angle, 0°
(d)
Fig. 10.3 -contd.

368 MACROMECHANICS
be two Poisson's ratios, as shown schematically in Fig. 10.5. When the lamina
is strained in direction I, we get equal the Poisson's ratio Vb equal to -Ez!E 1.
When the lamina is strained in direction 2, we get the Poisson's ratio v2, equal
to -E/E2•
We wish to relate these five conventional engineering constants to the four

independent elastic constants, reduced stiffnesses, Qij' Let us consider that 1 °
is the only nonzero component in Eq.(10.14). Then we can write
°1 = Qll El + Q12 E2
Solving for EI and E2, we get
0'"6
Ii;
- f- ;\:
):
l - F--
- r- r--- 0'"2
'"
t--
~ \
- f-
:f, .:1:
..
'--- 2
- i1 ~
,.
r- :~
t-- i---
-, l-
\~
f;
.~-;,
;i - ~
~~:
Fig. 10.4 A thin orthotropic lamina with its natural axes coinciding with the geometric
axes.
369 MACROMECHANICS
By definition, we have the Young's modulus in direction I, Et = (J/f,t. Thus,
2
Ql1 Q22 - QI2
EI (10.27)
Q22
c2 Q I2
VI -- (10.28)
f,1 Q22
Now let us repeat the above procedure with (J2 as the only nonzero strain
component in Eq.(lO.14). Then
2
(J2 Q ll Q22 - QI2
f,2 Qll (10.29)
(10.30)
~
---[d---------
....... .
--- -----------------
.....
'L t.
[]
1
·
:
·:
:.
:
I·· I..
:---------!
Fig. 10.5 The two Poisson's ratios for an orthotropic lamina.

370 MACROMECHANICS
Similarly, if we consider that (j6 is the only nonzero component, we can get
(j6
G6 = - = Q66
E6 (10.31)
VI E2 = v2 E I (10.32)
EI VI
(10.33)
E2 v2
We can solve Eqs. (10.27) through (10.30) for Qij to give

VI E2
QI2 =
1 -VI v 2
and Q66 = G6 is given by Eq.(10.31)

Note that only four of the five constants are independent.
Similarly, it can also be shown that the relationships between compliances and
engineering constants are as follows:
1
S =-
II E
I
10.3 ANALYSIS OF LAMINATED COMPOSITES
In the macroscopic analysis of laminated composites, we ignore the individual

identities of fiber and matrix. Each individual lamina is treated as a
homogeneous, orthotropic sheet and the laminated composite is analyzed using
the classical theory of laminated plates.
It would be in order at this point to describe the way a multidirectional laminate

is defined by using a laminate code to designate the stacking sequence of
laminae. Figure 10.6 shows two stacking sequences, an eight-ply laminate made
of eight unidirectionally aligned (0°) and stacked in parallel mode, denoted by
[0]8 and another eight-ply laminate with plies stacked at 0°,90°, +45°, and -45 0
371 MACROMECHANICS
// [0]8
// [0/90/+ 45/ - 45],
Fig. 10.6 Two stacking sequences, an eight-ply laminate made of eight unidirectionally
aligned (0°) and stacked in parallel mode, denoted by [0]8 and another eight-ply laminate
with plies stacked at 0°, 90°, +45°, and -45°, denoted by [0/90/+45/-45],.
in the top half and the bottom half being the mirror image of the top, denoted
by [0/90/+45/-45]., where the subscript s indicates that the stacking is
symmetrical with respect to the geometrical midplane of the laminate (z = 0).
10.3.1 Basic Assumptions
We assume that the laminate thickness, h, is small compared to its lateral

dimensions, Fig. 10.7. Therefore, stresses acting on the interlaminar planes in
the interior of the laminate, that is, away from the free edges, are negligibly
small (we shall see laterthatthe situation is differentatthe free edges). We also
assume that there exists a perfect bond between any two laminae. That being
so, the laminae are not capable of sliding over each other and the continuous
displacements are across the bond. The last assumption gives us the corollary
that a line originally straight and perpendicular to the laminate midplane
remains so after deformation.
Finally, we use the Kirchhoff assumption which states that in-plane

displacements are linear functions of the thickness, and therefore the
interlaminar shear strains, Exz and Eyz , are negligible. With these assumptions we
can reduce the laminate behavior to a two dimensional analysis of the laminate
midplane.
372 MACROMECHANICS
Ifu, v, and ware the displacements in the x, y, and z directions, respectively,

then we can write the strain-displacement relationships as follows:
OU ou
- +
OV
Ex Exy
Ox oy Ox
Ov Ou ()W
Ey - Exz - + - (10.34)
oy oz ()x
()W ()V OW
Ez Eyz - +-
oz oz ()y
For i = j, Eij are the normal strain components while, for i *' j, Eij represent the
engineering shear strain components equal to twice the tensorial shear
components. According to Kirchhoffs assumption, the in-plane displacements
being linear functions of the thickness coordinate, z, we write
T /y,v
~x,u
+h12
- h/2
f
midplane (z=O)
Fig. 10.7 A laminated composite. The geometric axes, x-y, of the laminate are the in-
plane axes while z is the thickness direction. z = 0 is the geometric midplane of the
laminate and h is the thickness of the laminated composite. Also shown are the
displacements U,V, and w along x,y, and z axes, respectively.
373 MACROMECHANICS
U = Uo(X,y) + z F) (x,y)
(10.35)
where Uo and Vo are displacements of the midplane. It also follows from

Kirchhoffs assumptions that interlaminar shear strains €xz and €yz are zero.
Therefore, from Eqs. (10.34) and (10.35) we obtain
()W
€xz = F) (x,y) +- =0
()X
()W
€yz = F2 (x ,y) +- =0
()y
From these two expressions, we get
()W
and F2 (x,y) =- (10.36)
()y
The strain in the thickness direction, €z' is negligible, thus we can write
w =w(x,y)
That is, the vertical displacement of any point does not change in the thickness
direction.
Substituting Eq. (10.36) into Eq. (10.35), we obtain
() U = () U 0 _ Z ()2 W =€ 0 + zK (10.37a)
()X ()X ()X2 x x
()V
€ =- (10.37b)
y ()y
374 MACROMECHANICS
The subscripts xy indicate the in-plane components. We have used subscript s

to denote the in-plane shear strain and curvature components, therefore, for the
ou +-
ov OV" 02w
- 2z - - = s " + zK
oy ox oy ox oxoy xy xy
sake of consistency, we shall denote Sxy by Ss and K,.y by Ks, we can rewrite the
expression for Sxy above as
(1O.37c)
The components sox, SOy, and so, are the midplane strains while Kx, Ky, and K,
are the plate curvatures. We can represent these quantities in a compact form
as follows
(10.38)
and
o-w
--
ox 2
Kx
02w
Ky -- (10.39)
oy2
1\\. 202W
--
~ ~
Equation (10.37) can be put in the following matrix form

375 MACROMECHANICS
(10040)
10.3.2 Constitutive Relations for Laminated Composites
Consider again the laminated composite shown in Fig. 10.7. The geometric
axes, x-y, of the laminate are the in-plane axes while z is the thickness
direction. z = 0 is the geometric midplane of the laminate and h is the thickness
of the laminated composite. Also shown are the displacements u, v, and w along
x,y, and z axes, respectively. The following assumptions are involved:
(a) Each lamina is a homogeneous and orthotropic material.

(b) The individual ply and the laminate obey Hooke's law, i.e., the
displacements and strains are small.
(c) A plane strain situation prevails.
(d) The bond between plies is perfect, i.e., no slipping between plies is allowed.
Then, we can write, for the kth layer, the following constitutive relation
(10041 )
According to the theory of laminated plates, we have the strain-displacement

relationships ofthe composite given by Eq.(10AO). We can rewrite Eq.(10AO)
in a more compact form as
(10042)
Substituting Eq. (10042) in Eq. (10041), we get the following constitutive

relationship for the kth ply
(10043)
376 MACROMECHANICS
In a composite made by stacking layers having fibers in different orientation,

the stresses will vary from ply to ply. The classical theory of laminated plates
is very useful to analyze such a material. This theory has a great simplifying
advantage, viz., a complex loading system across the laminate cross-section is
replaced with a simple, statically equivalent system of forces and moments
acting on the geometrical midplane of the laminated composite. The loading
system with respect to the midplane is described in terms of stress resultants
(Nx, NY' and N.) and moment resultants (Mx, My, and M.), Fig. 10.8. The stress
resultants (Nx, Ny, and N.) have the dimensions of stress times length or force
per unit length awhile the moment resultants (Mx, My, and MJ have the
dimensions offorce times length per unit length (i.e., force). In the most general
case, we have six stress components, O'x' O'y, 0',., O'xy, O'yz, and O'zx. The laminated
composite being in a state of plane stress, we only have O'x' O'y, and O'xy(= 0'.),
the components in the thickness direction, z, being zero. We define the three
corresponding stress resultants as
(10.44)
As can be seen from the above expressions, these stress resultants have the
dimensions of force per unit length and are positive in the same direction as the
corresponding stress components. These resultants give the total force per unit
377 MACROMECHANICS
length acting at the midplane. The three moment resultants are
(10.45)
h
2
~y - M.. fa .. zdz
h
'2
This system of three stress resultants (Eq. 10.44) and three moment resultants
(Eq. 10.45) is statically equivalent to actual stress distribution through the
thickness of the composite laminate.
-v
x M,
x N,
Fig. 10.8 Stress and moment resultants for a laminate.

378 MACROMECHANICS
From Eq.(1 0.43), we can write for the stress resultants as a summation over the
n plies
Q I2 QI6 Kx
Q22 Q26 Ky z dz
Q26 Q66 Ks
Since [SO] and [K] are not functions of z and in a given ply, and [0] is not a
function of z, these can be taken outside the integral sign to simplify the above
expression to
~ t::
1111 1112 11 16 Kx
= (10.47)
~. ~16
where
n
Aij = L (Qij)k(hk -hk_l )
k=1
(10.48)
and
379 MACROMECHANICS
(10.49)
We can rewrite Eq. (10.47) as
[N] = [A][eO] + [B] [K] (10.50)
Similarly from Eq.(l0.43) we can write for the moment resultants

0
Mx BII BI2 BI6 ex Dll D12 DI6 Kx
My BI2 B22 B 26 ey
0
+ DI2 DI6 D 26 Ky (10.51)
M., BI6 B 26 B66

e"
0
D22 D 26 D66 Ks
or [M] = [B][eO] + [D][K] (10.52)
where
Dij = 3"It (Qij)k(hk-h

- 3 k3-I ) (10.53)
k=1
and Bij is given by Eq. (10.49) above.
The stress and moment resultants with respect to midplane strains and plate
curvatures can be put in a compact form
where
[Aij' Bij' Dij] = f h/2 - 2

-h/2 Q/I,z,z )dz (i,j =1,2,6)
380 MACROMECHANICS
and Qij denote the plane stress reduced stiffnesses for the individual plies and
h is the total thickness of the laminate.
A quick examination of Eqs. (10.47) through (10.51) will show that a tensile
stress resultant is a function of the midplane tensile strains (COx and COy),
midplane shear (eo.), bending curvatures (I<.,. and Ky), and twisting (K.). This is
a much more complex situation than that observed in a homogeneous plate
where tensile loads result in only tensile strains. In a laminated plate we have
coupling between tensile and shear, tensile and bending, and tensile and
twisting effects. Specifically, the terms A 16 and A26 bring in the tension-shear
coupling while the terms B 16 and B26 represent the tension-twisting coupling.
The D 16 and D26 terms in a similar expression for Mx represent flexure-twisting
coupling. Under certain special conditions, the stress and moment resultants
become uncoupled.
Figure 10.9 shows that when the deformation is parallel to the material axes in
orthotropic material, then there is no coupling. This will be the case when the
fibers at 0° or 90° degrees to the stress axis. When the fibers are at an angle to
the stress axis, Fig. 10.9, the coupling terms will come into play and even when
a normal stress is applied, the response field will include normal, shear, and
bending strains. We can also examine the expressions for [A], [B], and [D]
terms and arrive at some conclusions. For example, the Aij terms are the sum of
ply Oij times the ply thickness (Eq. 10.40). Thus, Aij will be zero if the positive
contribution of some laminae is canceled by the negative contribution of others.
Now, the Qij terms of a ply are derived from orthotropic stiffness and, due to the
form of transformation equations (Eqs. 10.25), the components 011' 012' 022'
and 066 are always positive, so the components Aw A 12, A22 , and A66 are always
positive. Also, 016 and 026 are zero for 9 = 0 0 and 90 0 , and are positive or
negative for 0 0 < 9 < 90 0 • The components 016 and 026 are odd functions of9.
This means that
016(+9) = -016(-9)
and
026(+9) + -026(-9)
Thus, iffor each +9 ply, we have another identical ply of the same thickness at
-9, then we shall have what is called a specially orthotropic laminate with
respect to in-plane stresses and strains, i.e., A 16 = A26 = O. Relative position of
such plies in the stacking sequence is immaterial.
The Bij terms are sums of terms involving Oij and differences of the square of
the z terms of top (hk ) and bottom (hk_1) of each ply. Thus, Bij terms are even
381 MACROMECHANICS
functions ofhk> which means that they will be zero if the laminate composite is
symmetrical with respect to thickness. In other words, Bij will be zero if we
have for each ply above the midplane, a ply identical in properties and
orientation and at an equal distance below the midplane. Such a laminate is
called a symmetric laminate and will have Bjj identically equal to zero. The
absence of bending-stretching coupling in symmetric laminates eliminates the
problem of warping encountered in non-symmetric laminates when in-plane
forces are induced by thermal mismatch during processing. Symmetric
laminates will only experience thermal strains at the mid plane but no flexure.
The reader should observe that the origin of [B] matrix lies not in the intrinsic
orthotropy of laminae, but in the heterogeneous (non-symmetric) stacking
sequence of the plies. Thus, if were to make a two-ply composite consisting of
different isotropic materials such as alumina and glass sheets, such laminate
will have a nonzero [B].
---
~
~--
Fig. 10.9 The coupling phenomenon in an off-axis lamina. The coupling is zero in an
on-axis lamina.
382 MACROMECHANICS
The bending matrix Dij terms are defined in terms of Qij and the difference
between hk and hk_,. The geometrical contribution (h3k - h3k-I) will always be
positive. Thus, as explained ab~e for~, values ofD w D 12 , D22 , and D66 will
always be positive. Recall that QI6 and Q26 are odd functions ofe. DI6 and D26
will therefore be zero for all plies oriented at 0° or 90° as these plies have QI6
= Q26 = O. DI6 and D26 can also be made zero if for each ply oriented at +e and
at a given distance above the midplane, we have an identical ply at an equal
distance below the midplane but oriented at -e. This follows from the property
of odd functions of e, i.e., QI6 (+e) = - Q16( - e), Q26(+e) =Q26( - e) and (hk- hk_,)
is the same for both plies. Such a laminate, however, will not have a midplane
of symmetry, i.e., Bij 'f. O. In fact, DI6 and D26 are not zero for any midplane
symmetric laminate except for unidirectional laminates (0° or 90°) and cross-
plied laminates (0°/90°). We can, however, make DI6 and D26 arbitrarily small
by using a large enough number of plies stacked at +e. This is because the
contributions of +e plies to 0 16 and D 26 will be opposite in sign to those of - e
plies, and although their location will be at different distances from the
midplane, they will tend to cancel each other.
Ifwe wish to have an isotropic laminate, we can stack identical plies oriented
in such a way that the angle between any two adjacent layers is 21t1n. Such a
laminate will have [A] independent of orientation in the plane. We call such a
stacking sequence quasi-isotropic, as [B] and [D] are not necessarily isotropic.
Aij goes to zero only if the positive contributions are canceled by the negative
contributions.
10.4 STRESS-STRAIN BEHAVIOR OF LAMINATED COMPOSITES
Figure 10.10 shows the tensile stress-strain curves of [0]8' [90/0ls, and [90]8
SiCIRBSN laminates as well as of the unreinforced RBSN (Bhatt, 1991). The
[0]8 laminate curve shows three regions: a linear region, a nonlinear region
containing serrations, and a linear region again. Other laminates show only
linear region fracture. The elastic analysis described above had an important
assumption, viz., the bonding between the fiber and the matrix is perfect. That
assumption is not valid in the case of weakly bonded SiCIRBSN composites.
Accordingly, Bhatt et al. modified the above relationships to obtain the
383 MACROMECHANICS
600
500 - [O]B
m
0...
{ =, Specimen
~
failure
'"
en
<])
tJ
en
400
~
en
.::
<])
+-' 300
"@ [90;OJ,
.~
~
<])
200
+-'
S
0
0
~
100
o 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Axial strain, %
Fig. 10.10 Tensile stress-strain curves of [0]8, [90/0]8' and [90]8 SiCIRBSN laminates
as well as ofthe unreinforced RBSN (after Bhatt, 1991).
variation of the Young's modulus and room temperature strength of the

SiCIRBSN laminate with the fiber orientation, Figs. 10.11 and 10.12,
respectively. Pryce and Smith (1991) investigated tensile behavior of [0]12,
[0/90" [Ozf90 4 ], and [0/90]38 laminates made of Nicalon fibers and calcium
aluminosilicate glass-ceramic matrix made by hot pressing of prep regs. Figure
10.13 shows the stress-strain curves of these laminates. Note the characteristic
knee in the curves indicating the beginning of matrix cracking. The matrix
damage occurs in the form an array of cracks.
When we stress a laminate composite in a given direction, the ply with the least
strength in that direction will fail first. This is commonly referred to as the first-
ply-failure (FPF). In a [0/90] laminate, a 90 ply cracks first while the 0 ply,
having a higher strength than the 0 ply, will be uncracked. If we unload this
laminate and reload, the stress-strain curve will follow the line "partially
degraded" which shows a reduced stiffness of laminate. At higher strains, the
o ply will also crack. The crack density in 0 and 90 plies increases with
384 MACROMECHANICS
c
Q.
(!)
200f\
175
~
......
~
)(
)( 150
LLI
UI 125
:::J
:::J
't:J
0
100
E
-u
UI
c
LLI
75
50
25
0
0 20 40 60 80 100
e, deg.
Fig. 10.11 Variation of Young's modulus as a function of fiber orientation in
SiC/RBSN laminate (after Bhatt, 1991).
increasing strain until a saturation crack spacing is attained. Figure 10.14 shows
crack density as a function of applied strain for different laminates. As we
pointed out in chapter 7, such cracking results in a loss of stiffness which can
used as a measure of damage in the laminate. Figure 10.15 shows the variation
in normalized stiffness (0.02% secant modulus) as a function of applied strain.
The stiffness reduction results from the combined effect of cracking in 90 and
oplies.
Ply delamination is an important failure mode in laminated composites, Fig.
10.16. Delamination cracking in [0/90] laminated ceramic composites,
consisting ofNicalon fibers/LAS-II glass-ceramic matrix made by hot pressing,
has been analyzed (Sbaizero et a!., 1990). This material has a thin layer of
carbon at the fiber/matrix interface which imparts ease of debonding and sliding
and consequently toughness in the longitudinal directions (Prewo, 1986). It was
found that crack growth initiation resistance was dominated by the critical
strain energy release rate of the matrix. Bridging effects associated with intact
fibers and segments of matrix led to an increased resistance as the crack
extension progressed, i.e., a rising R-curve type behavior was observed.
385 MACROMECHANICS
900
800
o Matrix Crocking
Stress
c 700
Q.
~ • Ultimate Tensile
~
~
600 Strength
><
b><
500
.c
C.
c::
400
Q)
!:
III 300
~
'iii 200
c::
~
100
0
0 20 40 60 80 100
e, deg
Fig. 10.12 Variation of room temperature strength with fiber orientation for an
SiCIRBSN laminate (after Bhatt, 1991).
400.-----------------------------------------~-,
350
.,
a.
300
~ 250
...,
If)
-.-
~
iii
200
___.- (0/90)5
~ 150
Q. ....._..--_... (0/90)35
a.
<{
..................-_......._........... __ ._..._ ..- ..-- (02/904)5
.>-/,.------------
100
50
01+-----r----,----~----_r----~----~----~---4
o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 O.B
Strain (%)
Fig. 10.13 Stress-strain curves of Nica10n fiber/calcium aluminosilicate matrix
laminates (after Pryce and Smith, 1991). Note the characteristic knee in the curves
indicating the beginning of matrix cracking.
386 MACROMECHANICS
8 I I I I I I
• • Q -
7
_.
'-
6
• b- b-
I
E
I--
• • •
E • ¢ ¢
-
-
5 r- b-
>-
C/) 4 I--
• • -
"0
c:
Q)
3 r- ¢
b-
~
• (0)12 -
~
c
u
2 I--
•• • (0/90)S
-
u
~
b-
/:).
(02 /904)S
~
II-- ¢¢.b- <> (0/90)3S -
0 ~ I L I I I
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Strain,%
Fig. 10.14 Crack density as a function of applied strain for Nicalon fiber/calcium
aluminosilicate matrix laminates. (After Pryce and Smith, 1991).
1.0 AJrA
-b-
E»
0.9 rP
0 (0)12
C/) -rP A (02 /90 4 )S
-- •
C/)
.~
Q)
c: 0.8 (0190 )s
•.eb-
eb-
C/)
"0
0.7 .ib- b-
•i
Q)
N b- b-
e
c 0.6 b- b-
E • •
• •
~
0 iii
z
0.5 e
0.4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Strain, %
Fig. 10.15 Variation in normalized stiffness (0.02% secant modulus) as a function of
applied strain. (After Pryce and Smith, 1991).
387 MACROMECHANICS
Central
delamination
Edge Edge
- -
delamination / delamination
~ .-
Fig. 10.16 Two different types of delaminations in a laminate.
10.5 EDGE EFFECTS
The general three-dimensional stress state in a body is shown in Fig. 10.17. In

the lamination theory described above, we assumed that the laminate was thin
and only in-plane stresses (O"X' O"y, and O"s, or 0"1,0"2' and 0"6) were nonzero. Such
an assumption is valid for an infinite laminate composite, i.e., the interlaminar
stresses in the z-direction are zero for such a composite. In a real composite
laminate, however, one has free edges where rather significant levels of out-of-
plane stresses can be generated. It has been shown that in the vicinity of such
free edges, extending about a distance equal to the laminate thickness, there can
be present rather significant levels of interlaminar stresses in the z-direction (O"Z'
O"xz' and O"yz). This has been analyzed and experimentally verified in polymer
matrix composites (Pagano and Pipes, 1971, Pipes and Pagano, 1974; Pipes et
aI., 1973; Oplinger et aI., 1974; Herakovich, 1976; Garg, 1988). Figure 10.18
shows the situation in a schematic manner at a flat free edge. Such a situation
can also exist at free edges around a circular hole in a laminate. Clearly,
stresses acting in the direction of thickness would be particularly insidious if
they were of a tensile nature, because they would could cause easy
delamination. A schematic of such a delamination is shown, in an exaggerated
388 MACROMECHANICS
~
33
032
3 °23
°12
1 °11
Fig. 10.17 The general three-dimensional stress state in a body.
Free
edge
o
Load
14 .1
Laminate
thickness
Fig. 10.18 Stresses in the boundary layer of a laminate. Note the stress in the thickness
direction.
389 MACROMECHANICS
manner, in Fig. 10.l9. Compressive interlaminar stresses would be desirable.

The precise nature of these stresses is determined by the stacking sequence of
the plies. Detailed finite element analyses that out-of-plane stresses become
effective within a distance of about one laminate thickness from the free edge.
This distance over which the interlaminar stresses are effective has been called
the boundary layer. In the inner regions beyond one laminate thickness away
from the free edge, the stresses are in accord with the lamination theory
predictions. Interlaminar stresses can also arise due to thermal mismatch. It is
possible that the interlaminar stresses of thermal origin may be more significant
than due to mechanical loading only (Herakovich, 1976).
10.6 NONFIBROUS LAMINATES
Laminated ceramic composites can also be made from sheet materials, i.e.,
without using fibers. Such laminated multilayer ceramic composites are used
extensively in the electronic industry as well as, to a lesser extent, in other
areas. Laminated glass composites, for example, a composite made of a sheet
of polyvinyl butyryl polymer sandwiched in between two transparent glass
sheets provides high optical quality and safety features and are extensively used
in automotive and architectural windows. Virkaretai. (1987, 1988) made three-
layer composites by dry-pressing powders to obtain surface layers of
unstabilized zirconia and magnesia or alumina and a core layer of matrix oxide
(magnesia or alumina) and stabilized zirconia. When cooled from the
processing temperature, the unstabilized zirconia transformed to monoclinic
zirconia, putting the surface layers in compression. This decreased the
sensitivity to large surface flaws as long as the compressive layer thickness was
greater than the flaw size. Similar ideas can be used to produce high
temperature and corrosion resistance surfaces in nonfibrous laminated
composites. Tape casting technique can be used to produce multilayer laminates
(Amateu, 1990; Amateau and Messing, 1990, Gladysz et aI., 1999). Amateau
and coworkers applied the c1assicallaminate plate theory given above to layers
containing different volume fractions of silicon carbide whiskers to tailor the
residual stress in the SiCwi'Al2 0 3 and SiCwI'mullite laminates. They were able to
exploit the thermal expansion and elastic modulus mismatch between layers to
produce tailored laminates having residual compressive surface stresses and a
reduced tensile stress in the core. Charalambides (1991) also used the laminated
plate theory to analyze the steady-state mechanics of delamination cracking in
symmetric laminated notched beams subjected to an applied moment and an
axial force. Clegg and coworkers (Clegg et aI., 1990; Phillips et aI., 1992a;
Clegg and coworkers (Clegg et aI., 1990; Phillips et aI., 1992a; Clegg and
390 MACROMECHANICS
Seddon, 1992; Phillips et aI., 1992b; Phillips et aI., 1992c) used the concept of
toughening brittle materials by interleaving the laminae with weak interfaces.
The basic idea is a very simple one. They produced SiC laminates with weak
graphitic interlayers by viscous processing. The typical laminate thickness was
about 100 - 300 !lm with graphite layer thickness being about 10 !lm. Figure
10.20 shows load vs. apparent displacement in a
three-point bend test for notched monolithic SiC and for the laminated
composite. The most characteristic feature of these laminates was the
phenomenon of crack deflection at the interfaces between laminae. This
resulted in apparent work of fracture of 6 kJ m- 2 vis it vis 28 J m- 2 of the bulk
SiC produced by the same route.
We provide another example of crack deflection in nonfibrous laminated

ceramic composites, this one being an all oxide system (Gladysz et aI., 1999).
In chapter 9 we discussed the role of Sn02 in an alumina fiber/glass matrix
composite. However, previous attempts to fabricate an alumina matrix
composite using Sn02 as an interface material, i.e., AI20/Sn02 laminated
composites, have been unsuccessful (Norkitis, and Hellmann, 1991). The Al 20 3
- Sn02 phase diagram shows a simple eutectic reaction at 1620 °C. The
difficulty is that when processed in inert or reducing environments (i.e.,
vacuum, Ar, N 2), Sn02 tends to decompose to SnO, which has a low melting
point (»1125 0c). To prevent Sn02 from decomposing at the higher processing
temperatures it is necessary to densify a crystalline ceramic matrix such as
alumina in an atmosphere where there is a significant partial pressure of
oxygen. Gladysz et ai. (1999) were able to fabricate A1 20/Sn02 laminated
composites by tape casting aqueous slurries of the respective powders.
Alternating sheets of each material were then stacked and laminated by pressing
at 90°C and 20 MPa. The organic materials were pyrolized at 500°C for 5
hours. The hot pressing was done in air at temperatures ,at 1400 °C and 30 MPa
pressure for 35-45 minutes. Figure 10.21 shows an example of crack deflection
at interfaces. Sn02 lamina is still seen attached to the Al20 3 lamina that was
explained by the strong mechanical interlocking of Al20 3 into Sn02 grains.
Even though there was no primary chemical bonding at the interface, interfacial
roughness induced gripping was quite strong, indicating thereby the importance
of mechanical texture or roughness of the interface. In spite of this strong
mechanical bonding, the Sn02 layer does seem to serve as a weak interphase as
can be observed from the frequent deflection in the Sn02 layer.
391 MACROMECHANICS
Load Edge
delamination
Fig. 10.19 Schematic of delaminations at the free edge.
250
200
--. 150
6 Laminate
p..
~~
100
0
~
50
0 0 0.4 0.8 1.2

Displacement, u (mm)
Fig. 10.20 Load vs. apparent displacement curves in a three-point bend test for notched
monolithic SiC and for the laminated material. Note the characteristic crack deflection
at the interfaces between laminae represented by successive load drops. (After Philips
et aI., 1992).
392 MACROMECHANICS
Fig. 10.21 Crack deflection at interfaces in Alp/Sn0 2 laminated composite. A and S

indicate alumina and tin dioxide layers, respectively. (From Gladysz et aI., 1999)
REFERENCES
Amateau, M.F. (1990) in Thirty Seventh Sagamore Army Materials Research

Conference on Structural Ceramics, D.T. Viechnicki (ed.), Plymouth, MA, p. 127.
Amateau, M.F. and Messing, G.L. (1990) Center for Adv. Mater. Newsletter, 4, 75.
Bhatt, R.T.(l991) in Proc.: IntI. Con! on Composite Mater.l8 (ICCMl8), Hawaii, p.
23-A-1.
Bhatt, R.T. and R.E. Phillips (1990) J. Composites Tech. & Res., 12, 13.
Charalambides, P.G. (1991) J. Amer. Ceram. Soc., 74, 3066.
Clegg, W.J., K. Kendall, N.M. Alford, D. Birchall, and T.W. Button, (1990) Nature,
347,455.
Clegg, W.J. and L.R. Seddon (1992) in 2nd European Conference on Advanced
Materials, Euromat'91, Cambridge, Institute of Materials, p. 226.
Garg, A.C. (1988) Eng. Fract. Mech., 29, 557.
Gladysz, G.M., M. Schmucker, K.K. Chawla, H. Schneider, D.L. Joslin, and
M.K.Ferber, (1999) Journal ofMater. Sci., 34,4351.
Herakovich, C.T. (1976) Int. J. Mech. Sci., 18, 129.
Norkitis, M.E. and J.R. Hellmann (1991) in High Temp. Rev. 1991. 1991: NASA
Publication 10082.
Oplinger, D.W., B.S. Parker, and F.P. Chiang (1974) Expt. Mech., 14,347.
Pagano, N.J. and R.B. Pipes (1971) J. Composite Mater., 5, 50.
393 MACROMECHANICS
Phillips, A.J., Clegg, W.J., and Clyne, T. W. (March 31-April2, 1992a) in Proc. Fatigue
and Fracture 0/Inorganic Composites, Cambridge.
Phillipps, A.J., Clegg, W.J., and Clyne, T.W., submitted to Acta Met. et Mater., 1992b.
Phillipps, A.J., Clegg, W.J., and Clyne, T. W., submitted to Acta Met. et Mater., 1992c.
Pipes, R.B., Kaminski, B.E., and Pagano, N.J. (1973) in Analysis o/the Test Methods
for High Modulus Fibers and Composites, ASTM STP 521, ASTM, Philadelphia, p.
218.
Pipes, R.B. and N.J. Pagano (1974) J. Appl. Mech., 41, 668.
Prewo, K.M. (1986) J. Mater. Sci., 21, 3590.
Prewo, K.M. and U. Brennan (1982)J. Mater. Sci., 17,1202.
Pryce, A.W. and P.A. Smith (1991) in Proc.: Inti. Con! on CompOSite Mater.l8
(JCCM/8), Hawaii, p. 24-A-1.
Sbaizero, 0., P.G. Charalambides, and A.G. Evans (1990) J. Amer. Ceram. Soc., 73,
1936.
Velamakanni, B.V. and F.F. Lange (1991) J. Amer. Ceram. Soc., 74, 166.
Virkar, A., J.L. Huang, and R.A. Cutler (1987) J. Amer. Ceram. Soc., 70, 164.
Virkar, A., J. Jue, J. Hansen, andR.A. Cutler (1988) J. Amer. Ceram. Soc., 71, C148.
SUGGESTED READING
Calcote, L.R. (1969) Analysis 0/ Laminated Composite Structures, Van Nostrand

Reinhold, New York.
Christensen, R.M. (1979) Mechanics o/Composite Materials, John Wiley & Sons,
New York.
Daniel, I.M. and O. Ishai (1994) Engineering Mechanics 0/ Composite Materials,
Oxford Unviersity Press, New York.
Jones, R.M. (1975) Mechanics o/Composite Materials, Scripta Book Co., Washington,
DC.
Tsai, S.W. and H.T. Hahn (1980) Introduction to Composite Materials, Technomic,
Westport, CT.
Cfiapter 11
APPLICATIONS
In this chapter we describe some of the commercial applications of ceramic

matrix composites, emphasizing the salient requirements in each application
and how they are met by specific CMCs. Let us recall that the main attributes
ofCMCs are: high strength and modulus, low density, capability of being used
at high temperatures, and a greater toughness than that of monolithic ceramics.
In order to drive home some of these advantages of CMCs, we present some
examples. Some ofthe novel processing techniques described in Chapter 4 have
been used very profitably to make CMCs in useful forms. Figure 11.la shows
ceramic matrix composite turbine blades made by sol-gel vacuum impregnation
of fibrous preforms while Fig. 11.1 b shows a filament
'"
(a)
Fig. 11.1 (a) Ceramic matrix composite turbine blades made by sol-gel vacuum
impregnation of fibrous preforms, (b) a filament wound tube of CMC, (c) hammering
a nail through a CMC shows the damage tolerance, and (d) an oxyacetylene flame
impinging on the wall of a tube made of CMC shows the ability to withstand high
temperatures (courtesy of A.R. Hyde, 1989).
396 ApPLICATIONS
(b)
(c)
Fig. 11.1 - contd
397 ApPLICATIONS
Cd)
Fig. 11.1 - contd
wound tube made of CMC (Hyde, 1989). The characteristic of toughness or

damage tolerance is depicted by hammering a nail through a CMC (Fig. 11.1 c)
while the ability to withstand high temperatures is shown by an oxyacetylene
flame impinging on the wall of a tube made of CMC (Hyde, 1989).
A plot of strength/density vs. temperature is presented in Fig. 11.2. to compare

different types of composites The parameter strength/density is often referred
to as specific strength. It serves to normalize strength on a mass basis. In this
figure, we also give the Mach number, which is a measure of speed commonly
used in aerospace terminology. A Mach number 1 equals the speed of sound at
a given altitude. We note in Fig. 11.2 that ceramic matrix composites are the
only materials that have any strength left above 900°C. Chief among these are
carbon/carbon composites, silicon carbide whisker/alumina composites, and
continuous carbon fiber or silicon carbide fiber reinforced ceramic matrix (SiC
or Si3N 4 ) composites.
398 ApPLICATIONS
Mach number
2.4 2.8 3.2 3.8 4.5 6.0 10.0
.
c:
3.0
High- modulus carbon fiber-reinforced

"<=" 2.5 polymer composites
E
:e 2.0
"0
~ 1.5
l:'
.;;;
c: High temperature MMCs
GO
1.0
"....
~
c;.
c:
~ 0.5
Vi
0
0 150 300 450 600 750 900 1050
Temperature I ·C
Fig. 11.2 Comparison of different types of composites in terms of strength/density vs.

temperature. Also given is the Mach number.
In what follows, we first describe some general applications ofCMCs and then
examine, in detail, some specific applications, highlighting the salient features
regarding processing, microstructure, and properties of these composites.
11.1 GENERAL SCOPE OF ApPLICATIONS OF CMCs
It is convenient to divide the general applications of CMCs in terms of

aerospace and nonaerospace applications. In the aerospace area, performance
is the foremost consideration while in the non-aerospace fields cost
effectiveness is the prime consideration.
Aerospace applications, in general, demand high thrust-to-weight ratios, faster

cruising speeds, increased altitudes, and improved flight performance. These
goals translate into material requirements involving increased strength-to-
density, stiffness-to-density, and improved damage tolerance -all at
significantly higher temperatures. High-temperature structural composites
represent a key technology for advanced aerospace systems. Continuous fiber
reinforced ceramic composites potentially offer higher specific mechanical
properties which can be utilized in a variety of high-temperature aerospace
applications. Silicon carbide coated carbon/carbon composites are used as a
thermal protection material in the US space shuttle (Korb et aI., 1981). Carbon
fiber/silicon carbide composites are candidate materials for a variety of space
399 ApPLICATIONS
plane programs (Desnoyer, 1991). Besides the space plane, other applications
for ceramic matrix composites include a variety of high speed airplane, various
defense related projects such as Advanced Tactical Fighter (ATF), many
existing fighters, missiles, hypersonic radomes, hard armor and turbine engines.
It is expected that a number of propulsion related components in the high speed
airplane will use a variety of CMCs. Requirements such as high thrust, high
temperatures, low airport noise, reduced NO x emissions, etc. mean that CMCs
will have to be used. As a rule, aerospace applications require high performance
materials in large, complex shapes.
There are four main materials-related drivers for advanced aerospace

applications (Bray, 1988):
• Cost reduction in terms of fabrication and maintenance requirements.

• Weight reduction to increase specific fuel consumption.
• Higher operating temperatures.
• Signature reduction.
While the ultimate temperature requirement may exceed 2500 °C in some rare
applications, one can list the following three temperature categories where
different materials can meet the requirements:
• less than 800°C

.800 -1650 °C
• greater than 1650 °C
For ceramic matrix composites the last two categories are relevant, i.e;, the
service temperature range is 800 -1650 °C. This temperature range which
represents the extremes for a turbine engine exhaust under normal and after-
burner operations and structural elements for some of the conceptual vehicles.
A structural fiber reinforced composite with balanced in-plane tensile and
compression strength in the range of 175-350 MPa and moduli in the range of
100-175 GPa throughout the temperature range would offer a substantial
payoff in turbine engine, space satellite and transport, and hyper velocity
vehicle technology.
Among the nonaerospace applications of CMCs, engine components at high

temperatures and in corrosive environments, cutting tool inserts, wear resistant
parts, nozzles and exhaust ducts, energy related applications such as heat-
exchanger tubes, etc. are the prime areas. For such applications, the components
can range from simple to complex and tend to be smaller in size. Thus, it is not
400 ApPLICATIONS
surprising that for applications related to wear, cutting tool inserts, and heat
engines, there are commercially available dense, wear-resistant, particle- and
whisker-reinforced ceramic matrix composites. Successful inroads into cutting
tool inserts market have been made by CMCs such as TiC particle reinforced
Si3N4 and AIP3' and SiCjAI20 3. The competition in this case is mainly
tungsten carbide/cobalt, which accounts for about 50 % of the u.S. metal-
cutting tool market of $650 million. The main advantages of ceramic cutting
tools include chemical stability, superior hardness, and ability to operate at
higher temperatures which translate into higher cutting speeds.
In the area of wear resistant parts, ceramics offer high hardness, low friction,
superior abrasion resistance, good elevated-temperature capabilities, and good
mechanical performance at high speeds. Applications include vanes, seals,
nozzles, bearings, and wear guides. Among wear resistant parts, toughened
zirconia, and whisker and continuous fiber reinforced composites find
applications.
In the energy related areas involving heat transfer, storage, and recovery, heat
exchangers and recuperators made from monolithic ceramic materials can
operate at higher temperatures and in a more aggressive environments than
metals. However, a major drawback of monolithic ceramics is their poor
thermal shock resistance. Here continuous fiber reinforced ceramic matrix
composites can be used profitably. For example, continuous fiber reinforced
hot-gas filters for particle control purposes can be very efficient in hot, coal-
derived gas streams. Such gas streams can be found in pressurized fluidized-bed
combustors, direct coal-fueled combustors, and combined gasification systems.
The use ofCMC components in a variety of applications makes economic sense
because it will result in higher operating temperature, higher efficiency,
benefits in the energy and environmental (reduced emission) areas, longer
service life, etc. We describe below some of these applications
Porous radiant burners
Radiant burners transfer a substantial fraction of their energy input directly to

the process load as infrared radiation. Increased radiant output and fuel
efficiency are the main benefits of using CMCs in radiant burners.
Hot-gas filters
These are porous, closed-end tubes used to remove fine particles in a variety of
operations, such as fly ash in coal gasification and in pressurized fluidized-bed
401 ApPLICATIONS
combustion. Conventional filters fail easily during cleaning operations, which

involve backpressure pulses. Application of CMCs in such processes can
exploit their higher toughness and thermal shock resistance than monolithic
ceramics.
Candle-type filters consisting ofNextel312 or Nextel 610 ceramic fibers in a

silicon carbide or alumina matrix have been made for removing particulate
matter from high temperature gas streams up to 1000°C. The collected
particulate is removed by reverse pulse jet cleaning. The high temperature
capability of such filters can eliminate the need to cool the gas stream prior to
filtration, which may increase process efficiency and eliminate the cost and
complexity of gas dilution, air scrubbers, or heat exchangers. The ceramic
fibers toughen the composite and result in a filter with excellent resistance to
thermal shock and catastrophic failure. The light weight of such a filter reduces
the strength requirements of the tube sheet, and the excellent thermal shock
resistance provides protection during upset conditions. An example is a ceramic
composite filter designed for advanced coal-fired power generation systems,
such as pressurized fluidized bed combustion (PFBC) and integrated
gasification combined cycle (IGCC). Among its features are:
• High Temperature Capability

• Resistant to Thermal Shock
• Lightweight
• Resistant to Catastrophic Failure
Potential applications of such filters include:
• Pressurized Fluidized Bed Combustion (PFBC)

• Integrated Gasification Combined Cycle (IGCC)
• Incineration
Combustor liners
The use of CMC as combustor liners in gas-turbine power systems will result
in reduced emissions of nitrous oxides, commonly indicated by NOx, and fuel
savings because of operation at higher inlet temperatures.
Automotive area
There is always interest in the use of advanced ceramics in automobile engines

because of the large market. The driving force here is to increase operating
402 ApPLICATIONS
temperature and thereby increase fuel efficiency. The requirements for ceramics
in heat engine applications are good heat insulation, high strength, low abrasion
and corrosion, high toughness, low coefficient of friction, low density, good
thermal shock resistance and low cost.
Table 11.1 summarizes some industrial applications for continuous fiber

reinforced ceramic matrix composites (Sheppard, 1990). Indicated in this table
are a variety of product areas, examples from each product area, and possible
industrial markets. Having given the general picture of the applications of
ceramic matrix composites, let us now look at some specific composites in
some detail.
11.2 CARBON/CARBON COMPOSITES
Carbon/carbon composites have carbon fibers distributed in a carbon or, more

precisely, a carbonaceous matrix. Carbon is an excellent high temperature
material when used in an inert or non-oxidizing atmosphere. Unlike other
nonoxide ceramics, carbon powder cannot be sintered. Thus, the carbon matrix
is generally obtained by CVD or thermal decomposition of a carbon source
such as pitch or phenolic resin. Generally, many pores form during this
conversion of a hydrocarbon to carbon. Thus, a dense and strong pore-free
carbon/carbon composite is not easy to fabricate. A very wide range of
characteristics, however, can be obtained in these composites, the principal
characteristic being the high temperature capability ofthese composites in inert
environments. A comparison of high temperature capability of carbon/carbon
composites with metallic alloys is shown in Fig. 11.3, in terms of strength to
weight ratio as a function of temperature (Strife and Sheehan, 1988).
Carbon/carbon composites show superior specific strength at higher
temperatures in inert atmospheres.
11.2.1 Thermal Conductivity
Thermal conductivity is a very important characteristic for applications at high

temperatures. Carbon fibers can show a range of thermal conductivity values
depending on the precursor used. Low modulus, rayon-based carbon fiber has
a thermal conductivityof38 W/mK while high modulus, rayon based can be as
high as 122 W/mK (McAllister and Lachman, 1983). PAN-based carbon fibers
can show thermal conductivity values of2.1 and 70 W/mK for the high strength
and high modulus fibers, respectively. Pitch-based carbon fiber has a high
thermal conductivity of about 100 W/mK. Thus, it would appear that an
403 ApPLICATIONS
Table 11.1: Potential industrial applications for continuous fiber reinforced ceramic
composites (after Sheppard, 1990).
Product area Examl2.les Likel~ industrial market&l

Advanced heat Combustors, liners, High-temperature gas turbines;
engines wear parts, etc. possibly adiabatic diesels, gas turbine
combustor retrofits
Heat recovery Air preheaters, Indirect heating uses; energy intensive

equipment recuperators industrial internal processes (e.g.,
aluminum remelters, steel reheaters,
glass melters)
Burners and Radiant tube Potentially any indirect-fired, high-

combustors burners temperature and/or controlled
atmosphere heating/ melting/heat
treating industrial application
Burners and Combustors Low-NOx clean fuel heating

combustors applications, including gas turbine
combustors, industrial process heat
Burners and Low-temperature Low-NOx clean fuel heating

combustors radiant combustors applications, including small scale
(space heating) and large scale
(industrial process) applications
Process equipment Reformers, reactors, Chemical process industry, petroleum

HIP equipment refining
Waste incineration Handling Conventional and advanced

systems equipment, toxic/hazardous waste facilities, with
internals, or without energy recovery
cleanup
Separation! Filters, substrates, Gas turbine, combined cycle, and

filtration centrifuges other configurations; particulate traps
for diesel exhausts, molten metal s,
sewage treatment
Refractories and Furnace linings, High-temperature industrial

related crucibles, flasks etc. heating/melting!heat treating processes
Structural components Beams, panels,

decking, containers
404 ApPLICATIONS
important factor in regard to thermal conductivity is the type of carbon fiber

used.
It is worth pointing out that the most important parameter in determining the
thermal characteristics of a carbon fiber is the degree of crystalline orientation
along the fiber axis. The higher the degree of crystalline orientation along the
fiber axis, the higher will be the modulus along the fiber, the lower will be the
coefficient of thermal expansion, and the higher will be the thermal
conductivity. Some pitch-based carbon fibers such as the series K fibers and
another form of carbon fiber, called vapor grown carbon fiber (VGCF), have
much more ordered crystalline structure than PAN-based carbon fibers and,
consequently, the former can show extraordinarily high thermal conductivity
values (> 1000 W/mK).
11.2.2 Oxidation Protection of Carbon/Carbon Composites
By far, the greatest drawback of carbon/carbon composites is the formation of

gaseous oxides of carbon upon reaction with oxygen in air at temperatures as
low as 450°C. Potential applications of carbon/carbon composites may call for
use in service attemperatures exceeding 1000 DC and even approaching 2200 DC
for times ranging from 10 hours to a few thousand hours. There are two main
approaches to protect carbon/carbon composites against oxidation (Strife and
Sheehan, 1988; Luthra, 1998; McKee, 1986; McKee, 1987; Buckley, 1988).
One is to use inhibitors to slow down the rate of reaction between carbon and
oxygen. Another way is to use diffusion barriers to prevent oxygen from
reaching carbon and reacting with it. The coating should have the following
characteristics:
• oxidation resistance
• low oxygen permeability and volatility
• good adhesion to carbon
• chemical compatibility with carbon
• thermal expansion match with carbon
Figure 11.5a shows the critical factors that should be considered in an oxidation
protection system for carbon/carbon composites (Strife and Sheehan, 1988),
while Fig. 11.5b shows, schematically, a typical multilayer oxidation protection
system on a C/C substrate. Luthra (1988) has considered various rate limiting
steps such as gas phase diffusion, interface reaction, diffusion through liquid
B2 0 3 , and diffusion through cracks in the barrier layers. He calculated the
405 ApPLICATIONS
25
-- - -- ,
- - -- - - --
2D Carbon/Carbon __ / '
/
20
- - - -- - --
15 ~ Nickel based Superalloy
10 ""Single Crystal Superalloy
5 Niobium based Superalloy
a LL.L..L...L.l-l-L.L..L I II II ! I ! I I I ! I ! ) ! 11~~LL~~~
a 500 1000 1500 2000
Temperature, °C
Fig. 11.3 A comparison of high temperature capabilities of carbon/carbon composites
with metallic alloys, in terms of strength to weight ratio as a function of temperature
(Strife and Sheehan, 1988). Carbon/carbon composites show superior specific strength
at higher temperatures in inert atmospheres.
2.2
2.0
HIPIC
r<>
1.8 Process
I
E
t> 1.6
Ol
~ 1.4
>-
+-
·in 1.2
c
Q)
1.0
0
0.8
0.6
0.4
0 234 5 6 7
Cycles
Fig. 11.4 A comparison of high pressure impregnation of carbon (HIPIC) and

atmospheric processing of carbon/carbon composites. For density values over 1.4 g
cm- 3, the HIPIC process shows better efficiency than carbonization at atmospheric
pressure (after Taverna and McAllister, 1971).
406 ApPLICATIONS
virtual maximum rates for these steps and compared them with the acceptable
oxidation rates for short term (20 hours) and long term (2000 hours)
applications. His main conclusions are that the oxidation of unprotected
carbon/carbon composites below 500°C is controlled by interface reaction, at
higher temperatures the rate limiting step changes to a mixed gas phase
diffusion - interface reaction. Finally, above 700 °C, depending on the gas flow
conditions, the reaction is limited by gas phase diffusion. Inhibitors can be
beneficial at temperatures up to 700°C for long term and 850 °C for short term.
His calculations on oxygen transport rates through 100 11m layer of B20 3
indicate that for systems relying on it exclusively, the temperature limits are
about 600°C for short term and 850 °C for long term.
The two approaches are frequently combined in that one uses diffusion barriers
and internal glass forming inhibitors and sealants. Silicon-based ceramics, such
as SiC and Si3N 4, are commonly used as the primary oxygen barriers. Sealants
have the added advantage that they also seal thermal stress cracks originating
because of the thermal mismatch between C/C composite and silicon carbide
coating. Tetraethylorthosilicate (TEOS) is commonly used as an agent to
impregnate such cracks (McKee, 1986; McKee, 1987). TEOS leaves oxidation
resistant silica in the cracks. Ehrburger et al. (1986) studied the oxidation
inhibition by boria, B2 0 3. Their results are shown as weight loss of untreated
and B2 0 3 treated samples as a function of oxidation time at 980 K. The
presence ofBP3 significantly decreased the oxidation rate. McKee (1986) used
a ammonium borate - boric acid solution to protect the carbon/carbon
composite. Figure 11.6 shows the weight loss curves vs. time for the untreated,
2% B20 3 and 7% BP3 - treated carbon/carbon in flowing air at 1000 °C
(McKee, 1987). The borate-coated material showed a very small weight loss.
Extended oxidation protection of carbon/carbon composites can be obtained by
using boria containing dense Si3N4 or SiC coatings and boron-rich inner
coating. Borate glasses act as sealants and fill up the cracks that form in the
outer coatings because ofthermal expansion mismatch. An oxidation protection
system used for extended protection under thermal cycling conditions, with
peak temperatures reaching 1400 °C, is shown in Fig. 11.7 (Strife and Sheehan,
1988). Moisture sensitivity of borate glasses, corrosion of outer coatings by
glasses, glass volatilization, and an unacceptably high oxygen permeability are
some of the negative factors. Alvey and George (1991) used multilayer Zr/Pt
coating that formed ZrPt3 on annealing on carbon/carbon composites. Their
results showed that this coating protected carbon and a carbon fiber/phenolic
resin composite from a 2800 °C flame. They attributed this performance to the
oxidation resistance of ZrPt3 and to the reflectivity of the coating, which
reduced the thermal flux on the substrate to a short time.
407 ApPLICATIONS
Oxygen In Volatility
Adherence:!:;:
t;~~~~~~~~~~~~~~~~r-Multilayer
Coating
CtC
Substrate
(a)
Borosilicate Overglaze
(optional)
CVD SiC
Oxidation Barrier
Boride Sealant
. CjC Substrate
(b)
Fig. 11.5 (a) Critical factors that should be considered in an oxidation protection system
for carbon/carbon composites (after McKee, 1987), (b) schematic of a typical multilayer
oxidation protection system on a C/C substrate.
408 ApPLICATIONS
40 Untreated
I
I
I
I
on
on 2%B20 3 I
0
....l 30 I
I
.....
..c I
.....
bI) I 7%B20 3
~.....
I
/
1=1 20 I
<l) I
~
<l)
I
I
0..
I
10 /
/
,,
,
/ ,
/ "
/ "
//
-::-- -
-','
00 5 10 15 20
Time, h
Fig. 11.6 Weight loss curves vs. time for the untreated, 2% BP3 and 7% BPl - treated
carbon/carbon in flowing air at 1000 °C. The borate-coated material shows a very small
weight loss (after McKee, 1986).
Inner Coating of
Glass Formers Cracked Outer
Coating
/
7
o.. . o
..
·0 •
·0 •
..
.. . o
•• 0
Continuous
Carbon Fibers
Glass Formers in
Carbon Mattix
Fig. 11.7 A schematic of the protection system for extended protection ofC/C under
thermal cycling conditions, with peak temperatures reaching 1400 °C. Borate glasses act
as sealants and fill up the cracks that form in the outer coatings because of thermal
expansion mismatch (after McKee, 1986).
409 ApPLICATIONS
11.2.3 Carbon/Carbon Composite Brakes
Brakes are one of the major applications of carbon/carbon composites. An

example of a carbon/carbon brake assembly used on a Boeing 767 plane is
shown in Fig. 11.8. Brakes, aircraft or automobile, have the following general
requisites (Awasthi and Wood, 1988):
(a) Oxidation resistance

(b) High thermal capacity
(c) Good strength, impact resistance, strain to failure
(d) Adequate and consistent friction characteristics
(e) High thermal conductivity
The brake design must consider the following items (Awasthi and Wood,
1988):
• Friction members must generate stopping torque (over a range of

environmental conditions: rain, snow, etc.)
• Heat sinks must absorb the kinetic energy of the aircraft
• Structural elements should be able to transfer torque to the tires.
When the brakes are applied, multiple-disk brakes with alternating rotors and
stators are forced against adjacent members by hydraulic pressure. Friction
between rotating and stationary. disks causes them to heat up to 1500 °C
(surface temperature can be as high as 3000 0c), so good thermal shock
resistance is required. In view of the requisites listed above, any braking
material must be a good structural material, an efficient heat sink, and must
have excellent abrasion resistance. Let us consider an example from the aircraft
field which will help illustrate the formidable demands made on a braking
material. The Concorde supersonic plane and the Boeing 767 are among the
civilian aircraft that use carbon/carbon composite brakes. Let us take Boeing
767 as an example to help illustrate the formidable demands made on a braking
material. It has a mass of about 170,000 kg. Let us say that it has a take-off
velocity of about 320 kmIh. This will give us a kinetic energy at take-off of 670
MJ. In the event of an aborted take-off, this energy must be dissipated in about
30 seconds by the eight brakes on the aircraft. A rejected take-off (RTO) is,
indeed, the worst case scenario but then the braking material must be able to
meet such requirements. It is estimated that about 3000 RTOs occur every year
in the US.
410 ApPLICATIONS
Fig. ll.S An example of a carbon/carbon brake assembly used on a Boeing 767 plane
(Courtesy of AlliedSignal Corp.).
An example of the weight savings that result from replacement of conventional

brakes by carbon/carbon brakes easily explains why the aircraft industry is
interested in C/C brakes (Awasthi and Wood, 1988). In a large aircraft, multiple
stator and rotor arrangement (a sintered high friction material sliding against a
high temperature steel) weighs about 1100 kg. Carbon/carbon brakes (both the
stator and the rotor being made of carbon/carbon composite) weigh about 700
kg, resulting in a weight savings of 400 kg. Also, as a braking material,
carbon/carbon composites are superior to high strength bulk graphite.
Carbon/carbon composites have much higher strength at the same density.
411 ApPLICATIONS
11.2.4 Other Applications of Carbon/Carbon Composites
Among other aerospace applications of carbon/carbon composites, one may cite

their use in rocket propulsion components and reentry thermal protection of a
spacecraft. Figure 11.9 shows the frustum of a cone of a rocket made of a three-
dimensional carbon/carbon composite. The concept of thermal protection is
illustrated in Fig. 11.10 (Meetham, 1989). The top layer in contact with the high
temperature environment gets eventually sacrificed. Among non-aerospace
applications, there are examples of use ofcarbon/carbon composites as implants
as well as internal fixation of bone fractures because of their excellent
biocompatibility. They are also used for making molds for hot pressing.
Carbon/carbon molds can withstand higher pressures and offer a longer life
than polycrystalline graphite. Figure 11.11 shows an example of a mold made
ofcarbon/carbon composite that replaced a conventional heat resistanttool steel
material. In general, the high cost of carbon/carbon composites limits
applications to aerospace and other specialty applications. The limitation oflow
oxidation resistance is a serious one, and most applications do need some sort
of protection against oxidation.
11.3 SILICON CARBIDE WHISKER!ALUMINA COMPOSITES
Conventional cutting tools for metal machining are made of tungsten

carbide/cobalt. Inserts made of these materials are commonly used for
performing turning and milling operations on metals. Figure 11.12a shows
ceramic tools made by hot pressing of SiC whiskers in an Al20 3 matrix, while
Fig. 11.12b shows the microstructure of such a material. The following
characteristics are desired in a tool insert
• Abrasion resistance
• Thermal shock resistance
• Strength
• Fracture toughness
• Thermal conductivity
Ceramic tool bits retain high strength and hardness to elevated temperatures.
This allows the use of ceramic tool bits at much higher machining speeds than
is possible with tungsten carbide based tool bits. However, ceramic tool bits
have a toughness lower than that of tungsten carbide tool bit materials. This
restricts the use of ceramic tool bits to turning and milling operations. Table
11.2 gives a summary of room temperature properties of different cutting tool
412 ApPLICATIONS
Fig. 11.9 Frustum of a cone of a rocket made of a three-dimensional carbon/carbon

composite (courtesy of FMI).
413 ApPLICATIONS
High Thmperature Environment
Fig. 11.10 Concept of thermal protection (after Strife and Sheehan, 1988).
Fig. 11.11 A mold made of carbon/carbon composite (courtesy of SIGRI).

414 ApPLICATIONS
Table 11.2 Room Temperature Properties of Cutting Tool Materials (after Burden et
aI., 1988)
Tool Material Flexural Strength Hardness Fracture

(MPa) RA ness K/c
ToutlPam'/,
AIP3 500-700 93-94 3.5-4.5

AI 20 3-Zr02 700-900 93-94 5.0-8.0
AI20 3-TiC 600-850 94-95 3.5-4.5
AIP3- SiCw 550-750 94-95 4.5-8.0
Si3N 4 70-900 92-94 6.0-8.5
Sialon 700-00 93-95 4.5-6.0
WC-Co alloys 1250-2100 91-93 10.0-13.5
materials (Burden et aI., 1988).
Ceramic oxide matrix composites are used as cutting tool inserts, wear parts
such as wire drawing or extrusion dies, valve seats, high precision balls,
bearings for corrosive environments, and plungers for chemical pumps.
Partially stabilized zirconia is by far the largest of the ceramic material market
in wear and aerospace applications. Silicon carbide whisker or fiber reinforced
alumina, titanium carbide particle reinforced alumina, partially stabilized
zirconia are examples of ceramic composites used for machining. Typically,
SiC)Al20 3 cutting tools contain 30-45 % SiC whiskers. Cutting ofInconel718
with SiCjAI 20 3 composite tools shows a performance 3 times better than
conventional ceramic tools and 8 times better than cemented carbides. Typical
physical properties and performance parameters of AI 2 0 3 and SiC)AIP3
composites used for cutting tool inserts are given in Table 11.3 (Billman et aI.,
1988). During metal removal, a large amount of shear stress is required to
plastically deform and shear the metal, which results in high temperatures. As
much as 80% of the heat generated in cutting is due to this action (Whitney and
Vaidyanathan, 1988; Smith, 1986). This large amount of heat can cause failure
of the cutting tool and thus necessitates a reduction in the cutting speed of the
tool made of conventional materials. SiC whisker reinforced Al 20 3 composites
can withstand high temperatures while maintaining reasonable strength and
toughness. With these tool materials, the heat generated in the shear zone ahead
of the tool can be exploited to obtain optimum speeds at temperatures greater
than 1400 °C. In light finishing of nickel based alloys, it is claimed that a chip
415 ApPLICATIONS
is not produced, the metal being removed is so hot that it oxidizes to a powder
(Whitney and Vaidyanathan, 1988; Smith, 1986).
We discussed the fabrication of whisker reinforced composites in chapter 4 and

the toughening mechanisms for whisker reinforced ceramic matrix composites
in chapter 9. To recapitulate, the toughening mechanisms in these composites
involve crack deflection at the whisker/matrix interface, whisker bridging of the
matrix crack, and whisker pullout (Becher, et aI., 1988; Faber and Evans, 1983;
Giannakopoulos and Breder, 1991).
(a)
Fig. 11.12 (a) Ceramic tools made by hot pressing of SiC whiskers in an AIP3 matrix
(courtesy of Greenleaf Corp.); (b) microstructure of this material.
416 ApPLICATIONS
(b)
Fig 11.12 -contd
11.4 SILICON CARBIDE MATRIX COMPOSITES
Chemical vapor impregnation (CVI) techniques have been used to make

ceramic matrix composites such as SiCISiC and CISiC composites for
commercial applications (Lamicq et aI., 1991; Naslain and Langlais, 1986;
Fitzer et aI., 1980; Cavalier et aI., 1989; Lacombe and Rouges, 1990; Lacombe
and Bonnet, 1990). We described a variety ofCVI techniques in Chapter 4. The
salient feature of the CVI techniques is that most any complex shape can be
obtained starting with a fiber preform made up of stacked two-dimensional
fabric or three-dimensional woven preforms. Weaving in four or five directions
can also be used. The matrix is deposited onto and in between the fibers from
decomposition of gaseous reactants. Two main processes control the growth of
matrix in this process:
• mass transfer of reactants and products through the pores by isothermal

diffusion.
• kinetics of the chemical reaction.
Under low pressure and temperature conditions (Naslain and Langlais, 1986;
Fitzer et aI., 1980), chemical reaction kinetics control the matrix growth and
good matrix can be obtained. CVI techniques have been successfully used to
417 ApPLICATIONS
deposit C, SiC, Si3N 4 , TiC, B 4C, BN, A120 3, Zr0 2 as matrix materials (see
Chapter 4). Also, invariably, multiple impregnation cycles combined with
intermediate surface removal are required to obtain an adequate density. The
process, however, is very slow. The French aerospace company, SNECMA, has
specialized in using flexible, small diameter fibers, such as carbon or Nicalon,
in a fabric form to make any complex shaped preform. The carbon fiber/silicon
carbide matrix (trade name SEPCARBINOX), silicon carbide fiber/silicon
carbide matrix (trade name CERASEP), and carbon fiber reinforced carbon
(trade name SEPCARB) are some of the commercial varieties made by
SNECMA. A two-dimensional preform is made of balanced (0°/90°) woven
fabric of carbon (T300 high strength fiber) or silicon carbide (Nicalon) fiber.
Three dimensional preforms, fibers woven manually or in a machine in three
orthogonal directions have also been used. Table 11.4 gives the physical
properties of various C/SiC and SiC/SiC composites made by SNECMA. Table
11.5 gives mechanical and thermophysical properties of two-dimensional C/SiC
and SiC/SiC and three-dimensional C/SiC composites at room temperature,
1000 °C, and 1400 °C (Lacombe and Rouges, 1990). Exceptionally high fracture
toughness have been obtained in these composites. Tensile strength of C/SiC
and SiC/SiC as a function of temperature is shown in Fig. 11.13, while thermal
Table 11.3 Properties and Perfonnance of A1 20 3-based Cutting Tools (after Billman et
aI., 1988)
Property A120 3 SiCjA120 3

Vickers microhardness at 18.5-kg load (GPa) 17.2 19.7
Fracture toughness (MPa m ll.) 4.5 6.0
Thennal-shock resistance ('Yo) 54 100
Workpiece diameter (mm) 152.4 152.4
Depth of cut by length of cut (mm) 2.5 by 135.9 2.5 by 135.9
Feed rate (mm per revolution) 0.18 0.13
Cutting speed (m S·I) 4.6 4.8
Total cost per piece (insert + machine) ($) 4.99 1.34
Estimated savings ('Yo) 73
418 ApPLICATIONS
Table 11.4 Physical properties of two dimensional (2D) C/SiC and SiC/SiC composites
and three dimensional (3D) C/SiC composites (after Lacombe and Rouges, 1990).
Composite Fiber Vol. fraction Density Residual Porosity

(%) (gem- 3) (Vol%)
2D C/SiC 45 2.1 10
2D SiC/SiC 40 2.5 10
3D C/SiC 24 2.3 12
conductivity and thermal diffusivity of these composites in the plane and in the
thickness direction are shown in Figs. 11.14 and 11.15, respectively. Note the
high degree of anisotropy in two-dimensionally reinforced composites.
These CMCs are mainly meant for use in rocket or jet engines:
• Liquid propellant rocket motors
The composites can withstand, uncooled, the temperatures of liquid

oxygen/hydrogen combustion. Equal or better lifetimes than refractory alloys
and at higher temperatures have been claimed (Lamicq et aI., 1991).
• Turbojet parts
These are meant mainly for the afterburner area and have the benefits of lower
weight, high thermal shock resistance and less maintenance compared to the
conventional materials. Actual flights have been performed on the Mirage 2000
using components made of these composites (Lamicq et aI., 1991).
• Thermal structures for space plane and other space probes
Panels having aerodynamic shapes and dimensions as large as 1.8 x 0.8 x 0.3
m, have been formed with these CMCs. The main limitation on the dimensions
comes from the size of the reactors available. One major application of
SEPCARBINOX (C/SiC) will be in making of thermal structures in the
European space plane, Hermes, which will experience high mechanical loading
and surface temperatures as high as 1300 °C. These panels will be used in the
Hermes space plane on the external surfaces, where the thermal fluxes are very
419 ApPLICATIONS
450
400
2D C/SiC /
--
~
350 -
tr:: 300
1l
IZl 250
~r:: 2D SiC/SiC
200 - - - - - - -
~
150
3D Novollex
C/SiC Composile
100
500 250 500
Temperature, °C
Fig. 11.13 Tensile strength ofC/SiC and SiC/SiC as a function of temperature (after
Lacombe and Rouges, 1990).
25
In Plane
20
1
-5
~
15
~
o~
0p
U
.g
I':
0 10
U
] 5
Through Thickness
Temperature, DC
Fig. 11.14 Thermal conductivity, in-plane and in thickness direction, of C/SiC and
SiC/SiC composites (after Lacombe and Rouges, 1990)0
420 ApPLICATIONS
high. It is claimed that the overall weight of this is lower that the system used
in the US space shuttle where light weight but thick tiles are used (Lamicq, et
aI., 1991). Carbon/carbon thermal structures and protection are planned to be
used in the Huygens probe that will explore the moon Titan. The Titan
atmosphere is nonoxidzing, therefore, carbon/carbon are appropriate for this
mISSIon.
11.5 CERAMIC MATRIX COMPOSITES VIA POLYMERIC PRECURSORS
Prototype CMC components wherein the ceramic matrix is obtained via

polymeric precursor have been made. Just to give an example, Blackglas™ is
a trade name of AlliedSignal for a ceramic that has rather high levels of carbon
distributed in it and has been used to make composites. This material is
amorphous and has a chemical composition of silicon carboxide (SiCxO y) with
carbon (about 15 - 3 0%) being present not as graphitic particles but as bonded
to silicon atoms. This material is formed by a proprietary process using
pyrolysis of thermosetting polymeric precursors. Blackglas™ composites can
then be made via standard polymer composite processing techniques. The
precursor polymer matrix composite can be conveniently formed by a standard
technique for making polymer matrix composites using fiber tows or woven
fabrics. This PMC is then converted into a CMC by pyrolyzing the polymer
matrix into a ceramic matrix. Pyrolysis is done between 800-1000 DC. Multiple
impregnation and pyrolysis cycles can be used to obtain a greater than 90%
theoretical density. In order to obtain a tough fiber reinforced compsoites with
fibers such a Nicalon in this silicon carboxide matrix, a carbon coating on the
fibers seems to be beneficial because it provides a weak interface and leads a
damage tolerant composite. Such a carbon coating, however, will not be very
desirable for high temperature use in air. Prototypes of various structures, such
as gas deflection tube, fairing channel, radome shape, duct section, vane, and
various engine parts have been made by this material. According to
AlliedSignal, even after 51 hours at 1350 °C in stagnant air, the material
showed only minor amounts of crystalline SiC and Si02 •
11.6 PROTECTION OF CERAMICMATRIX COMPOSITES BY ENVIRONMENTAL

BARRIER COATINGS
Materials for turbines, land-based stationary gas turbine (CSGT) for power-and-
steam cogeneration operation or for service in aerospace must withstand long
421 ApPLICATIONS
service times under very aggressive environments. Just as we described above

the need for protection against service environment for carbon/carbon
composites, coatings called environmental barrier coatings (EBC) are needed
for silicon carbide fiber reinforced silicon carbide matrix composites. We
mention here one such example. Under the auspices of the U.S. Department of
Energy (DOE) field tests have been conducted on silicon carbide/silicon carbide
(SiC/SiC) combustor liners in a Solar Turbines Centaur 50S gas turbine, which
accumulated approximately 40000 hours by the end of year. The three-layer
EBC that was applied to both liners consisted of a BSAS top layer, a mullite
(inner liner) or mullite+BSAS (outer liner) intermediate layer, and a Si bond
coat. A layer of silica formed during the engine test due to oxidation of the Si
layer. The three-layer EBC system used on the outer liner performed better than
that on the inner liner. The addition of BSAS to mullite in the intermediate
layer minimized/reduced cracks and porosity in that layer, resulting in reduced
oxidation of the Si layer and, thus, better protection of the liner. In addition,
separation at the interface of the mullite intermediate layer and the silica layer
was greatly reduced with the addition of BSAS. Velocity plays an important
role in volatilization of silica during oxidation of Si and Si-based material.
Mullite was used in the intermediate layer because its thermal expansion
coefficient is close to that of SiC, and its higher temperature capability
compared with BSAS. The stability of mullite in the combustor environment
appeared to be a significant issue. The mullite phase from both liners separated
into silica and alumina phases. When the BSAS top layer fully recessed, the
silica in the intermediate layer was preferentially lost, leaving behind porosity
and alumina. Pinholes had formed at many locations where surface asperities
occurred. The surface asperities in the SiC seal coat caused vertical cracks in
the EBC, which were thought to form either during the coating process or after
thermal cycles from engine start/stops. Once the crack was formed, accelerated
oxidation of the layer occurred, causing the coating to buckle and eventually
spall.Protection of SiC/SiC liners with an EBC led to an extension of life in
industrial gas turbines. The stability of various components such as BSAS and
mullite in the combustion environment must be taken into account.
422 ApPLICATIONS
Table 11.5 Mechanical and thermophysical properties of two-dimensional (2D) C/SiC

and SiC/SiC and three-dimensional C/SiC composites at room temperature, 1000 °C,
and 1400 °C (after Lacombe and Rouges, 1990; Lacombe and Bonnet, 1990).
2D C/SiC
Property Temperature
23 °C 1000 °C 1400°C
Young's modulus (GPa) 90 100 100
Elongation (%) 0.9 0.9
Flexural strength (MPa) (4-point bend) 500 700 700
Compressive strength (MPa)
In plane 580 600 700
Through thickness 420 450 500
Shear strength (MPa) (Interlaminar) 35 35 35
Fracture toughness (MPa mil') 32 32 32
Coefficient ofthermal expansion (10-6 K)
In plane 3 3
Through thickness 5 5
Specific heat (J kg K) 620 1400
Total emissivity 0.8 0.8
2D SiC/SiC
23°C 1000 °C 1400°C

Elongation (%) 0.3 0.4 0.5
Flexural strength (MPa) (4-point bend) 300 400 280
In plane 580 480 300
Shear strength (MPa) (Interiaminar) 40 35 25
Fracture toughness (MPa m'l') 30 30 30
Coefficient of thermal expansion (10-6 K-l)
In plane 3 3
Through thickness 2.5 2.5
Thermal shock resistance (MPa) 315 280
Specific heat (J kg-' K 620 1200
Total emissivity 0.8 0.8
3D (Novoltex) C/SiC
23°C 1000 °C 1400°C

Elongation (%) 0.5 0.6 0.9
Flexural strength (MPa) 300
In plane 650 700 800
Shear strength (MPa) (Interlaminar) 100 40 40
Coefficient ofthermal expansion (10-6 K- 1)
In-plane 2 2
Through thickness 2.5 2.5
423 ApPLICATIONS
1. 20 C/SiC In Plane
2. 20 C!SiC Through Thickness
12 3. 3D C!SiC In Plane
4. 30 C{SiC Through Thickness
5. 20 SIC/SiC In Plane
6. 20 SiC/SiC Through Thickness
10
\ \
\ '- \
\
8 \\ \
\\ \
\
\', , \.,
6 \. \', " "-
" '-
\., ""
, ", '- '-
4 "
--
~, - -- ·3
--------. 5
-4
2 6 -------------
Temperature, °C
Fig. 11.15 Thennal diffusivity, in- plane and in thickness directions, of CISiC and
SiC/SiC composites (after Lacombe and Rouges, 1990).
REFERENCES
Alvey, M.D. and George, P.M.(1991) Carbon, 29, 523.

Awasthi, S. and Wood, J.L. (1988)Adv. Ceram. Mater., 3, 449.
Becher, P.F., Hsueh, C.H., Angelini, P., and Tiegs, T.N. (1988)J. A mer. Ceram. Soc.,
71,1050.
Billman, E., Mehrotra, P., Shuster, A., and Beeghly, C. (1988) Amer. Ceram. Soc. Bull.,
67, 1016.
Bray, DJ. (1988) in Fabricating Composites, SAMPE, paper # EM88-535.
Buckley, J.D. (1988) Amer. Ceram. Soc. Bull., 67, 364.
Burden, S., Rue, J., and Stromsborg, C. (1988) A mer. Ceram. Soc. Bull., 67,1003.
Cavalier, J.e., Lacombe, A., and Rouges, J.M. (1989) in Dev. Sci. Tech. Composite
424 ApPLICATIONS
Mater., 3rd Euro. Conf. on Composite Mater (ECCMl3)., Elsevier Sci. Pub., London"
p.99.
Desnoyer, D. (1991) Mater. Tech. (paris) 79, 57.
Ehrburger, P., Baranne, P., and Lahaye, J. (1986) Carbon, 24, 495.
Faber, K.T. and Evans, A.G. (1983) Acta Metall., 31, 565.
Fitzer, E., Hegen, D., and Strohmeier, H. (1980) Rev. Int. Hautes Temp. Refrac., 17,23.
Giannakopoulos, A.E. and Breder, K. (1991) J. Amer. Ceram. Soc., 74, 194.
Hyde, A.R.(1989) GEC J. Res. 7, 65.
Korb, L.J., Morant, C.A., Calland, R.M., and Thatcher, C.S. (I981)Amer. Ceram. Soc.
Bull, 60, 1188.
Lacombe, A. and Bonnet, C. (October 1990) in AIAA '90, Second Inti Aerospace Planes
Conference '90, Orlando, FL" Amer. Inst of Aero. and Astro., Washington, DC.
Lacombe, A. and Rouges, J.M. (September 1990) in AIAA'90, Space Program and
Technologies Conference '90, Huntsville, AL, Amer. Inst. of Aero. and Astro.,
Washington, DC, AIAA-90-3837.
Lamicq, P., Bonnet, C., and Chateigner, S. (1991) in Designing with Structural
Ceramics, Elsevier App. Sci., London, p.251.
Luthra, K.L. (1988) Carbon, 26, 217.
McAllister, L.E. and Lachman, W.L. (1983) in Fabrication of Composites, North-
Holland, Amsterdam 143.
McKee, D.W. (1986) Carbon, 24, 737.
McKee, D.W. (1987) Carbon, 25,551.
Meetham, G.W. (1989) Mater. and Design, 10, 197.
Miriyala N., J. Kimmel, J. Price, K. More, P. Tortorelli, H. Eaton, G. Linsey, and E.
Sun (2002) in Proc.: 47th ASME IGTI ASME Turbo Expo 2002: Land, Sea, and Air,
GT-2002-30585, Amsterdam, The Netherlands.
Naslain, R. and Langlais, F. (1986) Mater. Sci. Res., 20,145.
Sheppard, L.M. (1990) Amer. Ceram. Soc. Bull., 69, 667.
Smith, K.H. (1986) Carbide and Tool Journal, 8:5, 8.
Strife, J.R. and Sheehan, J.E. (1988) Amer. Ceram. Soc. Bull., 67, 369.
Taverna, AR.and L.E. McAllister (1971) inAdv. Materials: Composites and Carbon,
Amer. Ceram. Soc., Columbus, OH, p.199.
Whitney, E.D. and Vaidyanathan, P.N. (1988) Amer. Ceram. Soc. Bull., 67,1010.
APPENDIX A: WEIBULL STATISTICS
A.l Weibull Statistics
Let P(Vo) be the survival probability, i.e., P(Vo) is the fraction of identical
samples, each of volume Yo, that survives when loaded a stress o. According
to a distribution named after a Swedish engineer. Weibull, this survival
probability is given by
(A.I)
where 0 0 and ~ are constants. Figure A.1 shows a plot ofP(Vo) vIS. o. At 0 = 0,
i.e., no load on the sample, all samples survive and P(V0) ;: 1. As 0 increases,
more samples fail and P(Vo) decreases. Eventually, P(Vo) ~ 0, as 0 ~ 00, Le.,
all samples fail at very high loads. It is easy to see from Eq. (A. I ) that for 0 =
1
P (V) = -; = 0.37 (A.2)
i.e., 0 0 is the stress corresponding to a survival probability of 37%. Pis the so·
called Weibull modulus. The smaller the value of p, the greater the variability
in strength. In other words, the value of ~ tells us how rapidly the strength falls
as we approach 0 0 , Figure A.2 shows schematically the strength distributions
for a series of ~ values.
------1---------
t
1
1
1
.5 ~-~-~-~-~-~
--
.-
~
~ UL-______~____~__
a ---
Fig A.I. Survival probability, P(Vo) as function of stress, 0.
426 ApPENDIX A
t
~=5
OL-__________~~~~__
a --..
Fig. A.2. Strength distributions for a series of Weibull modulus (~ )values.
From Eq. (AI) we can write
CA3)
Or, taking logarithm of both sides once again, we get
(A.4)
Equation (A4) represents a straight line on a double log plot with the slope, ~
and intercept on the y-axis of VN o' Table Al gives some typical ~ values.
Figure A3 shows how the Weibull plot, for a given material, changes with
improvement in, say, processing such that there are smaller defects, which
would result in a higher mean strength but that the distribution of flaws remains
the same, i.e., we have the same value of ~. On the other hand, if the
distribution of flaws changes but the size does not, we shall have a higher value
of ~ but the same value of mean strength. In chapter 3, we have used a
expression for the Weibull distribution that includes the dependence of strength
on fiber length. We can include the volume dependence of material in the
Weibull distribution in the following manner:
427 ApPENDIX A
v (-)P]
P(V)=exp[-- 0 ° (A.S)
o Vo 00
In P (V) = - -v °
(-)P0
Vo 00
In In [_1_] = In ~ + ~ In ( ~) (A.6)
P(V) Vo 00
Equation (A.6) again is of the form y = mx + c, i.e., a straight line on a double

log plot with the slope, ~, the Weibull modulus.
Original
i
....
Q)
i
....
Q)
distribution
I
I
I
I
.= ::l I
--
0 /
~ 00-
I
.
/
00-
-
0 0
>. >. I
I
..0 :0 h /
/
0 0
! Higher
/t1
..0 ..0
....
0 0
.... I
• 13 ,
I
0. 0.
I
same
·
0>
0
0>
0
I
• 0:- ISame {1 ,
...J ...J I
• " higher (f
Log strength- Log strength -
Fig A.3. Effect of improvement in processing results in a higher mean strength but that
the distribution of flaws remains the same, i.e., the same value of ~. If the distribution
of flaws changes but the size does not, we have a higher value of ~ but the same value
of mean strength.
428 ApPENDIX A
Table A.I: Typical values of the Weibull modulus for some materials".
Material (3
Chalk, brick, pottery, cement <3
SiC, AIP3' Si3N4 2 - 10
Steel 100
• indicative values only.
APPENDIX B: SOME IMPORTANT UNITS AND
CONVERSION FACTORS
B.1 STRESS (OR PRESSURE)
1 dyn = 105 Newton (N)

1 N m- 2 = 10 dyn em- 2 = 1 Pascal (Pa)
1 bar = 105 Nm- 2 = 105 Pa
1 heetobar = 100 bars = 108 em- 2
1 kilobar = 108 N m- 2 = 109 dyn em- 2
1 mm Hg = 1 torr == 133.322 Pa = 133.32 N m- 2
1 kgf mm-2 = 9806.65 kN m- 2 = 9806.65 kPa = 100 atmospheres
1 kgf em-2 = 98.0665 kPa == 1 atmosphere
1 lb in-2 == 6.89476 kN m- 2 == 6.89476 kPa
1 kgf em-2 = 14.2233 lb in-2
106 psi == 106 lb in-2 == 6.89476 GN m-2 = 6.89476 GPa
1 GPa == 145,000 psi
B.2 DENSITY
1 g em- 3 == 62.4280 lb ft- 3 == 0.03611b in- 3

lIb in-3 == 27.68 g em- 3
I g em-3 == 103 kg m- 3
B.3 VISCOSITY
1 poise == 0.1 Pa s = 0.1 Nm- 2 s

1 GN m- 2 s = 10 10 poise
B.4 ENERGY PER UNIT AREA
1 erg em- 2 = 1 mJ m- 2 = 10- 3 J m- 2 = 1 dyn em-! = 10- 3 N m-!

108 erg em- 2 == 47.68 ft 1b in- 2 == 572.16 psi in
430 ApPENDIXB
B.5 FRACTURE TOUGHNESS
1 psi in l/2 = 11bfin- 3/2 = 1.11 kN m- 312 = 1.11 kPa m- 1I2

1 ksi in ll2 = 1.11 MPa m- 1I2
1 MPa m ll2 = 0.90 ksi in ll2
1 kgfmm- 3/2 = 3.16 x 104 N m- 3/2
AUTHOR INDEX
Abuhasan, A. 281, 282, 290 Bickerdike, RL. 119, 136

Aksay, l.A. 329, 351 Billman, E. 410, 413, 419
Alvey, M.D. 402, 419 Birchall, lD. 72, 103,389,392,393
Amateau, M.F. 355, 389, 392 Birnie, D.P. 46
Amazigo, lC. 296, 353 Blaszkiewicz, M. 202, 203
Anderson, W.P. 234, 259 Blissett, M.l 286, 289
Andrews, K.W. 266, 289 Boccaccini, A.R. 134, 136, 137, 286,289,
Angelini, P. 151, 167,229,230,258,269, 296,354
271,289,411,419 Bonnet, C. 412, 414, 416, 418, 420
Anselkmi-Tamburini, U. 28, 45 Bordia, R.K. 108, 136, 137
Argon, A.S. 165-167 Bouchetou, M.L. 286, 289
Arridge, A.G.C. 159, 167 Bowen, D.H. 108, 138,225, 226, 227, 258,
Arsenault, RJ. 142, 167,278,289 260
Aveston, J. 164, 165, 167,225,230,258, Bowen, H.K. 10,46, 111, 136
301,308,351,352 Bowman, K.l 286, 290
Avva, V. S. 215,260 Bradbury, l.A.A. 72,103
Awasthi, S. 405, 406, 409 Brandon, D.G. 163, 167
Brandt,]. ll5, 137
Balooch, M. 76,104 Braue, W.293,353
Bansal, S.P. 325, 331, 354 Bray, D. 395,419
Barber, l.R. 130, 132,136 Breder, K. 319, 354, 41 I, 420
Baranne, P. 402, 420 Brennan, lJ. 108, 136, 137,227,228,235,
Barrera, E.V. 281, 289 246,258,355,365,393
Barrett, C.S. 281,282,290 Bright, lD. 159, 167
Barclay, SJ. Ill, 136 Briggs, A. 215,222,229,230,258
Barczk, VJ. 332, 353 Brinker, CJ 46
Barsoum, M. 46 Brinkman, C.R. 215,222, 259
Baston, G.B. 301, 353 Brook, RJ. 129, 137
Beaumont, P.W.R 242, 259 Brooksbank, D. 266, 289
Becher, P. 151,167,229,230,243-246, Brosse,l.323,353
258-260,269,271,286,289,290,318, Broutman, LJ. 151, 167
354,411,419 Brun, MJ. 280, 289,323,353
Beeghly, C. 410, 413, 419 Brown, A.R.G. 119, 136
Behrens, E. 196, 202 Brown, D.R. 134, 136
Beier, W. 286, 289 Brown, W.S. 104
Bender,B.A.246,258,323,353 Budiansky, B. 296, 311,353
Bulik, C. 323, 353
Bennet, S.C. 62, 103 Bullock, E. 252, 258
Bernhart, G.A. 286, 287, 290 Bunsell, A.R. 82, 105
Bertinetti, M.A. 281, 282, 290 Burden,S.410,419
Besmann, T.M. 122, 136, 137 Burger, E.E. 269, 271, 289
Bhargava, P. 298, 352 Burkland, C.V. 119,136
Bhatt, H. 198, 202 Burns, GJ. 128, 136
Bhat R.T 112, 113, 136, 198,202,286,289 Bustamante, W.E. 119, 136
432 Author Index
Butler, E.G. 133, 134, 136 Cotteret, J. 323, 353

Button, T.W. 389, 392, 393 Cox,B.N. 220, 258,259, 299, 300,313,
353,354
Cai, P.Z. 280, 289 Cox, H.L. 205, 258
Calcote, L.R. 393 Cranmer, D.C. 149,150, 167
Calland, R.M. 400,420 Crivelli-Visconti, I. 226, 258
Calvert, L.D. 38, 45 Croft, W.J. 31, 34,45
Caputo, AJ. 122, 136, 137 Cutler, I.E. 104
Carlsson, J.O. 122, 136 Cutler, RA. 389, 393
Carlsson, R. 128, 137 Danchavijit, S. 159, 167
Carpenter, RW. 293, 353 Daniel, I.M. 203, 205, 215, 229, 259, 393
Carroll, D.F. 215, 220, 258, 260 Das Chowdhury, K. 293, 294, 353
Case, E.D. 266, 286, 289, 290 Dauchier, M.M. 286, 287, 290
Cavalier, J.e. 412, 420 Dauger, A. 128, 137
Celemin, J.A. 324, 353 Davidge, RW. 215, 222, 229, 230, 258
Chamis, C. 151,157,167,185,186,195,202 Davis, R.F. 36,45,249,259
Chandra, G. 128, 136 Dawson, D.M. 227, 228,258
Chateigner, S. 412, 414, 416, 420 Day, RJ. 104
Chawla,K.K 10,48,70, 103-105, III, 112, de Arellano-Lopez, A.R. 242,243,260
134,136,137,159, 161, 162, 167,195, DeBolt, H.E. 75, 104
203,260,263,269,271,278,280,286, De Jonghe, L.e. 108, 136, 137
287,289,301,323,332,336,352-355, Desnoyer, D. 395, 420
389,390,392 Demazeau, G. 347, 348, 353
Chawla, N. 104,159, 161, 162, 167,286, Deurbergue, A. 62, 104
287,289,337,342,353-355 Dhingra, A.K. 66, 104
Chen, S.T. 246, 248, 260 DiCarlo, J.A. 104
Chen, Z. 350, 353 DiGregrio, J.F. 281, 282, 289
Chevacharoenkul, S. 249, 259 Dickerson, P.O. 74, 104
Cilia, K.- Y. 215, 260 Diefendorf, R.J. 53, 58, 104
Chiang, F.P. 387, 389 Dinwoodie, J. 72,104
Chiang, Y.-M. 10,71,35,46 Disson, J.P. 126, 128, 136
Cho, B. 74,105 Doerner, M.F. 159, 167
Cho, C. 239,258 Dominguez-Rodriguez, A. 242, 243, 260
Chou, T.-W. 242, 259 Donald,I.W. 301, 353
Chokshi, A. 229, 249, 254, 258, 260 Donaldson, K.Y. 198,202
Christman, T. 278, 289 Dresher, W.H. 50, 104
Chuang, T.-J. 215, 220, 260 Dugne, O. 323, 353
Cinibulk, M.K. 346,348,353 Duncan,S. 350,353
Claussen, N. III, 115, 136,137 Duran, P. 137
Coblenz, W.S. 22,46 Dutta, I. 281, 289
Cockayne,B. 72, 74,104 Dyson, B.F. 215, 222, 258, 259
Coffin, C. Ill, 112, 136, 159, 161,286,
287,289,354 Eaton, H. 417, 420
Cohen, J.B. 290 Eckel, AJ. 286, 289
Collins, J.M. 108, 136 Eldridge,J.l.ll1, 112,136,159,161,162,
Colomban, P. 83-86, 104, 105,325,331,354 167,286,287,289
Conroy, P.J. 126-128, 13 I Ehrburger, P. 411, 420
Cook, J. 149, ISO, 167,301,353 Elices, M. 105,324,353
Cooper, G.A. 164, 167,225,226,230,258, Engdahl, R. 323, 353
308,351 Erdogan,F. 234, 259
Corman, G.S. 252, 254, 258 Erny, T. 128, 136
Cornie, 1. 165·-167 Evans, C.C. 225, 258
Cotterell, B. 159, 167,313,353
Author Index 433
Evans,A.G.220,259,281,289,289,296, Gordon, R.S. 246, 252, 259

300,303,304,311,318,352,353-355, Goretta, K.C. 104, 243, 242, 246, 260
411,420 Gouadec, G. 83-86, 104, 105,325, 331,354
Evans, P.E. 252, 258 Goursat, A. 128, 137
Ezekiel, H.N. 53, 104 Grant, NJ. 252, 259
Ezis, A. 229, 260 Green, DJ. 10,46,280,289,298, 353
Greil, P. 126-130, 136
Faber, K. 296, 318, 353, 354, 355, 411,420 Gresczuk, L.B. 205, 258
Fantozzi, G. 126, 128, 136 Grether, M.F. 252, 259
Farmer, S.C. 74, 104 Groves, G.W. 227, 260
Feldman, L.A. 280, 289 Guekenyesi, J .Z. 286, 289
Felten, JJ. 74, 104 Guette, A. 323, 353
Fett,T.215,258 Guinea, J. 307, 353
Ferasaki, M. 163, 167 Gupta, V. 165-167,303,304-309,353
Ferber,M.K.11l, 136, 159, 167,215,229, Gurley, G.F. III, 137
259,298,352,301,323,354,355,389, Gyekenyesi, J.Z. 126-128, 131
390,392
Fernandez, J.F. 137 Ha, J.-S. 323, 353
Fernando,J.301,323,354 Haggerty, 1.S. 74,104
Ferrante, J. 307, 353 Hahn,H.T.242,258,393
Fine, M.E. 227, 260, 296, 353 Haley, PJ. 232, 258
Firestone, R.F. 252, 258 Han,L.237,258,260
Fisher, R.M. 278, 289 Hannink, R.HJ. 298, 353
Fitzer, E. 119, 129, 136,280,290,412,420 Harade, Y. 255, 256, 258
Flank, A.M. 82, 104 Hashin, Z. 178, 180, 181,183,184,195,
Forrest, C.W. 124, 136 202,203,236,260
Fourdeux,A.62,65,104 Hasselman, D.P.H. 197, 198,202,252,259
Fox,1.R. III, 136 Hayashi, J. 78,105
Frechete, V.D. 290 Hayashi, S. 296, 354
French, J.E. 126-128, 130, 136 He, M.Y. 149, 167,303,353
Frevel, L.K. 104, 127, 137 Hecht, N.L. 104,246,259
Funayama, D. 126, 128, 137 Hegen, D. 119, 136,412,420
Furtak, J.E. 281, 282, 289 Hellmann, J. 329, 332, 353
Helms, H.E. 232, 258
Gac, F.D. 104, Ill, 137, 296, 353 Henager, Jr., C.H.
Gadow, R. 119, 129, 136 Hench, L. 24, 45
Gao, Y.C. 157, 167 Herakovich, C.T. 387,389,392
Galiotis, C. 104 Herbell, T.p. 286, 289
Garboriaud, R.J. 252, 258 Herrmann, K.P. 269, 289
Garg, A.C. 387,392 Hermes, E.E. 246, 258
Gasson, D.G 72, 74, 104 Heuer, A.H. 252, 258
Gao, Y.-C. 313, 353 Hiasyuki, T. 104
Gault, C. 286, 289 Highsmith, A.L. 242, 258
Generazio, E.R. 286, 289 Hill, R. 182-185, 202
George, P.M. 411, 419 Hillig, W.B. 5, 10
Giannakopoulos, A.E. 319, 354,411,420 Hirano, K. 255, 256
Gladysz,G.M. 280,289,350, 353 Hoagland, R.G. 296, 353
Gokoglu, S.A. 122,136 I-lockey, B.I. 215, 220, 244, 260
Gomez, M.P. 234, 259 Hodder, R.S., Jr. 104
Gonon, M.F. 126,128,136 Hoenigman, J.R. 104, 246, 259
Goodier, J.N. 266, 290 Hodder, Jr., R.S. 246, 260
Gordon, J.E. 149, 150,167,225,258,301, Hoffmann, MJ. III, 137
353
434 Author Jndex
Holmes, J.W. 130, 132, 136,239,241,246, Katz, R.N. 9,10,31,32,45

247,256,257,259,260,337,353 Kaya, C. 134,136,137
Holtz, A.R. 252, 259 Keller, K. 215, 252, 258, 259
Homeny, J. Ill, 136,229,259 Keller, L. 104
Honda, K. 163,165 Kelly, A. 164,165, 167,225,230,258,308,
Hsueh,C.H. 108, 136,229,230,258,269, 351
271,289,411,419 Kelly, B.T. 53, 104
Huger, M. 286, 289 Kennedy,P. 124, 136
Hughes, G. 119, 136 Kerans, R.J. 230, 246, 258, 259, 280, 290,
Hull, A.W. 269, 271, 289 313,317,318,333,347,353,354
Hunold, K. 36, 45 Kerner, E.H. 190,191,202
Hurley, G.F. 72,104 Kerr, M.I04
Hurwitz, F. 126-128,131 Khan, A. 162, 167
Hutchinson, J.W. 149, 167,281,290,303, Kim, H.M. 104,246,259
311,313,353 Kim,J.K.354
Hwang, H.J. 72, 104 Kim, R.Y. 242, 258
Kim, Y. 266, 286, 289, 290
Ilston, T.J. 133, 134, 136 Kimmel,J. 417, 420
Inal, O.T. 62, 104 Kingery, W.D. 10,46
Incorvati, L. 323, 353 Kinney, J.H. 76, 104
Insley, R.H. 332,353 Kishi, T. 286, 287, 290
Iosipescu, N. 154,155,167 Klacka, R. 119, 136
Ishai, O. 203, 393 Kobayashi, A.S. 296, 354
Ishirkawa, M. 104 Kohtuko, Y. 350, 354
Isoda, T. 126, 128, 137 Kokini, K. 286, 290
Komarneni, S. 69,105
Jack, K.H. 37, 45 Koopman, M. 100, 112, 136, 159, 161,286,
Jakus, K. 252, 259, 261 287,289,350,353
Janczak, J. 286, 289 Korb, L.J. 400, 420
Janowski, G.M. 350, 353 Krenkel, W. 258, 261
Janney,M. 296, 353 Kristofferson, A. 115, 137
Jarchow, o. 128, 136 Krstic, V.V.
Jayaraman, N. 240,243,260 ,296,353
Jenkins, M.G. 215, 259, 260 Kriven, W.M. 298, 345, 346, 353, 354
Jensen, H.M. 281,290,313,353 Krukonis, VJ. 75,103
Jero, P. 318, 353 Kumagai, M. 66, 103
Johnson, D.D. 252, 259 Kupperman, D. 281, 282, 290
Johnson, DJ. 53, 62, 104 Kurosawa, N. 286, 287, 290
Johnson, L.F. 197, 202
Johnson, W.53,62, 103, 104 LaBelle, H.E. 72, 103
Johnson, W.S. 242, 259 Lachman. W.L. 398, 401, 420
Jones, R. H. 296, 353 Lacombe, A. 411-415,418-420
Jones, R.M. 393 Laffon, C. 82, 104
Jones, R.W. 46 Lagarde, P. 82, 104
Joslin, D.S. 355, 389, 390, 392 Lahaye,J.402,420
Jurewicz, AJ.G. 280, 290 Laird, C. 236, 260
Lamicq, P.J. 286, 287, 290, 412, 414, 416,
Kagawa, Y. 163, 167,286,287,290 420
Kandil, F.A. 215, 258, 259 Lange,F.F. 229,260, 296, 353, 355,393
Kanka, B. 350, 353 Langdon, T.G.42,45
Karandikar, P.G. 242, 259 Langlais
Kardos, J.L. 185, 187, 198,202 ,410,412
Katz, A.P. 104,246,259 Lara-Curzio, E. 77, 103
Author Index 435
Lawrence, P. 205, 259 Martin, R.L. 215, 259

Layden,G.K.227,228,246,260 Mattoni, M. 350, 353
Leatherman, G.L. 31, 32, 45 McAllister, L.E. 398, 401, 420
Leca, N. 62, 103 McCartney, L.N. 205, 259
Lecomte, A. 128, 137 McCullum, J.R. 103,246,259
Lee, H.W. 108, 137 McKee,D.W. 400, 402-404,420
Lee, J.-G. 103 McLachlan, D.S. 202, 203
Lee, J.-W. 205,215,229,259 McLean,M.252,258
Lee, S.M. 157,167 McMillan, P.W. 301, 353
Lee, W.J.266,286,289,290 Meetham, G.W. 407, 420
Lessing, P.A. 246, 252, 259 Mehrotra, P. 410, 413, 419
Levi, C.G. Mendelson, M.I. 296, 353
,350,353 Meschter, P.l 39,45
Lewis, D. 323, 353 Messier, D.R. 31, 34, 45
Lewis, M.H. 323, 353 Messing, G.L. 66, 103,280,289,355,392
Liang, F.L. 66, 103 Metzger, M. 269, 271, 289
Lindemanis, A. 103 Meyers, M.A. 10
Lin, H.-T. 243 Michalske, T.A. 329, 332, 353
Lin, W.246,259 Milewski, J.Y. 103
Linsey, G. 417, 420 Miles, D.E. 252, 258
Lipowitz,1. 103, 127, 137 Miller, R.L. 127, 137
Lipp, A. 36, 45 Miriyala, N. 417, 420
Lipsitt, H.A. 103, 246, 259 Mlavskey, A.I. 72, 103
Liu, D.S. 255, 260 Monkman, F.C. 252, 259
Liu, H.Y. Iii, 137 More, K 417,420
Liu, KC. 215,222,259 Morgan, P.E.D. 323, 353
Liu, W.215,220,260 Morant, C.A. 400, 420
Lohr, R.D. 222, 258, 259 Morrell, R. 222, 258, 259, 261
LLorca, J. 105,353,324 Mortensen, A. 108,136
L6fvander, J.P.A. 350, 353 Moruzumi, H. 126, 128, 137
Lowden, R.A. 119, 122, 123, 136, 137, 159, Moulson, A.J. 22, 45
167,286,290,323,325,326,330-332, Mrosowski, S. 136
353 Mumrn, D.R. 318, 353
Lundberg, R. 115, 128, 137 Munir, Z.A. 28,45
Luthra, KL. 400, 420 Munz, D. 215, 258
Murat, M. 126, 128, 136
Macem, J.G. 286, 287, 290
MacEwen, S.R. 281, 282, 290 Nair, S.Y. 259,261,350,354
Mader, W. 151, 167 Nakayama, N.
Mah, T. 103,246,252,259 Naslain, R. 119, 137,258,261,323,347,
Maheshwari, A. 332, 353 348,353,410,412
Mai, Y.W. 159, 167,313,353,354 Newaz, G.M. 220, 259
Majumdar, A. 301, 353 Newnham, R.E. 202, 203
Majumdar, B.S. 220, 259 Nicholson, P.S.
Majumdar, S. 281, 282, 290 ,296,353
Mann, A.B. 76, 103 Nielsen, L.E. 187, 203
Mansfield, J.F. 337, 353 Niihara, K 72, 104
Margolin, H. 66,103 Nix, W.D. 159, 167
Marom, G. 195, 202 Nixon, R.D. 249,259
Marquis, P.M. 133, 134, 136 Nordine, P.C. 74, 104
Marsh, D.M. 225, 258 Norton, F.H. 252, 260
Marshall, D.B. 321, 225, 281, 289, 299, 300, Nourbakhsh,S.66,103
323,352,353 Noyan, J.C. 290
436 Author Index
Nyc, J.F. 203 Quinn, G.D. 261
Oberlin, A. 62, 103 Rabe,1.A. 103, 127, 137

Ogin, S.L. 242, 259 Rahaman, M.N. 108, 136, 137
Ohji, T. 215, 222, 259 Raj, R. 108, 136, 137
Okamura, K 78,104,105,246,259 Ramakrishnan, V. 242, 243, 260
Omori, M. 78, 105 Ramualdi, J.P. 301, 353
Oplinger, D.W. 387, 389 Ranson, H. 119, 136
Reed, J.S. 46
Pagano, N.J. 163, 167,387,393 Reifsnider, K.L. 242, 258
PaLczer, A.R. 286, 289 Reig, P. 347, 348, 353
Parker, B.S. 387, 389 Riester, L. 159, 167
Parks,D.M. 165-167 Reynolds, W.N. 62,103
Paris, P.C. 234, 259 Riedel, R. 129, 137
Parrot, N.J. 225, 258 Riggs, J.P. 60, 62, 103
Parthasarathy, T.A. 157, 163, 164, 167,230, Roberts,D.E.215,258,260
252,259,313,317,318,333,347,353, Roach, D. 239, 260
354 Rocher, J.-P. 323, 353
Pask, J.A. 163, 164, 167,329,351 Rohr, D.L. 103
Passing, G. 129, 137 Rojas,O.E. 108, 137
Patterson, B.P. 298,353, 355 Romine, J.C. 66,104
Patel, B. 111, 112, 136, 159, 161,286,287, Rose, J.H. 307, 353
289 Rosen, B.W. 104,178,180,181,183,184,
Pator, J.Y. 324, 353 195,202,203,205,236,260
Pearce, D.H. 286, 289 Rosenfield, A.R. 220, 259
Peebles, L.B. 104 Routbort, J.C. 242, 243, 260
Penn, L.S. 157, 167 Rouges, J.M. 412-415, 418, 419, 420
Penty, R.A. 252, 259 Roy, R. 69, 104
Perret, R. 62, 65, 103 Rue, J. 410, 419
Petrovic, J.J. 103, Ill, 137,237,260,281, Ruland, W.62,65, 103
282,289 Ruhle, M. 303, 353
Petzow, G. 111, 137,290
Pfeiffer, K. 286, 289 Sacks, M.D. 108,137
Phalippou, J. 26, 42 Sakka, S. 26,45
Phillips, D.C. 108, 138,226,227,259,260 Salt, F.W. 134, 136
Phillips, RE. 286, 289 Sambasivan, S. 347, 348, 352, 354
Pipes, R.B. 387, 393 Sambell, R.A.J. 108, 138,226,227,260
Pollack, J.T.A. 72,103 Sando,M. 72,104
Pompe, P. 128, 13 7 Sands, R 162, 167
Ponton, C.B. 133, 134, 136,286,289 Sandstrom, J.L. 103
Poritsky, H. 269, 271, 290 Sato, K. 126, 128, 137
Porter, J.R 229, 249, 254, 258, 260 Sayir, A. 74, 104
Prassas, M. 26, 42 Sbaizero, O. 384, 393
Predecki, P. 281, 282, 290 Schadler, L. 104
Preston, R.F. 227, 228, 258 Schapery,R.A. 190, 192, 195,203
Prewo, K.M. 108, 136, 137,227,228,246, Scheltens, F.J. 347, 354
258,260,355,365,384,393 Scherer, G.W. 46, 269, 290
Prochaska,S. 30, 45 Schilling, C.H. 295, 353
Price, J. 417, 420 Schlichting, 1. 119, 136
Pryce, A.W. 383, 385, 393 Schmlicker, M. 350, 353,355, 389,390, 392
Purser, A. 227, 228, 258 Schneider, G.A. 290
Pysher, D.J. 103,246,252,260 Schneider, H. 258, 261, 350, 353, 355, 389,
390,392
Author Index 437
Scholze, H. 10,46 Suresh, S. 237, 238, 240, 258, 260, 261,

Schonfelder, H. 129, 137 278,289
Schrodt, D.S. 246,258 Suwa, Y. 66,104
Schoenlein, L.H. 295, 353 Suzuki, T. 255, 256, 258
Schubert,P.280,290 Swain, M.V. 298, 353
Schwartz, D.S. 39,45 Szweda, A. 130, 132, 136
Schwetz, KA. 36, 45
Seddon,L.R. 390, 392 Takaku, A. 159, 167
Seguchi, T. 103 Talty, P.K 252, 260
Seibold, M. 128, 136 Talreja, R. 242, 260
Seltzer, M.S. 252, 260 Tannery, V.l 258
Seshadri, S.G. 215, 260 Tappin, G. 225, 226, 260
Sevely, C.R. 323, 353 Tattersall, H.G.T. 225, 226, 260
Shadwell, D. 323, 353 Taverna, A.R. 401, 420
Shafry, N. 163, 167 Taya, M. 296, 354
Shalek, P.O. III, 137 Tezuka,A. 126,128,137
Sharp, J.V. 62, 103 Thatcher, C.S. 400, 420
Sheehan,J.E.398,400-402,409,410 Thevenot,F.37,39,45
Sheppard, L.M. 398, 399,420 Tiegs, T.N. 229, 230, 258, 286, 290, 411,
Shennan, J.V. 124, 136 419
Shetty,D.K. 159, 167,313,354 Timoshenko, S. 266, 290
Shibuya, M. 104 Tokarsky, E. 53, 58, 103
Shih, C.J. 229, 260 Tortorelli, P. 417,420
Shimizu, K 350, 354 Towata, A. 72, 104
Shtrikman, SA 181,202 Trojnacki, A. 220, 260
Shuler, S.F. 239,260 Tsai, S.W. 185,187, 198,202,393
Shuster, A. 410, 413, 419 Tressler, R.H. 103,215,246,252,258,260
Simon, G. 82,105 Tulluri, V. 252, 259
Sinclair, R. 296, 354 Tur, Y.K. 130, 132, 136
Singer, L. 53, 58,61,62,104 Turner, P.S. 190, 191,203
Singh,1. 281, 282,290 Tyson, C.N. 53, 103
Singh, R.N. 280, 286, 288-290, 323, 353
Sinha, A. 286, 290 Uhlmann,D.R. 10,46,108,136
Sirieix, F. I, 137 Ullrich, D.R. 24,45
Skaggs, S.R. 103 Urquhart, A.W. 117, 118, 137
Smith, J.R. 307,353
Smith, K.H. 410, 411, 420 Vaidya, R.U. 195,203,301,323,354
Smith, P.A. 242, 259, 286, 289 Vaidyanathan, R. 215, 260
Somiya, S. 36, 45 Vaidyanathan, P.N. 410, 411,420
S0rensen, B. 241,259,260 Varner, J.R. 290
Sowman, 11.G. 71, 104 Vasudevan, A.K 162, 167
Spain, R.G. 53,103 Vaughn, W.L. 111, 136,229,259
Steif, P. 220,260 Velamakanni, B.V. 355, 393
Stemstein, S. 77, 103 Venkatesh, R. 352, 354
Stevens, R. III, 137 Villers, P. 38, 45
Stinton, D.P. 119, 122, 136, 137,323,353 Virkar, A. 384, 393
Strife, 1.R. 398, 400-402, 409, 410 Visser, L.R. 104
Strnad, Z. 10, 46
Strohmeier, H. 412, 420 Wachtman, J.B. 10,46
Stromsborg, C. 410, 419 Wagner, H.D. 104
Studebaker, M.C. 136 Wakamoto, T. 350, 354
Sun, E. 417,420 Waku, Y. 350, 354
Walker, P.L. 136
438 Author Index
Wallace, 1.S. 246, 258 Wright, 1. 280, 290

Wang, I-LV. 286, 288, 290 Wu, S. 298, 355
Wang, L. 142, 167 Wu, X. 115, 137,239,257,259,260
Wang, S.-W. 162,167
Wang, Y.Q. 269, 289 Xu, Z.R. 354
Wang, Y.R. 255, 260
Wang, Z. 236, 260 Yajima, S. 78, 79, 105
Warshaw, S.L 252, 260 Vallee, R.B. 296, 354
Washburn, M.E. 22, 46 Yamamura, T. 104
Watt, W.53,60,61, 104 Yang, J.-M. 119, 136,229,246,248,259,
Wawner, F.E. 75, 103 260,350,353
Wax, S.G. 79, 105 Yang, M. III, 137
Weber, J.K.R. 74, 105 Yasuoka, M. 72, J04
Wei, G.C. 229, 230, 260, 318, 354 Yen, C.F. 236, 260
Weihs, T.P. 76, 77, 103 Yeomans, J.A. 286, 289
Weinberg, A. 195,202 Y oon, H.S. 296, 354
Wereszczsak, A.A. 159,167 Young, RJ. 104,296,354
Westbrook, 1.H. 38,46 Yun, H.M. 74,104
Wetherhold, R.C. 286, 290
Whitney, E.D Zarzycki, J. 26, 46
,410,411,420 Zhu, D. 345, 354
Wiederhorn, S. 215, 220, 244. 258, 260 Zawada, L.P. 286, 290
Williams, J,J. 103 Zok, F.W. 350, 353
Wilson, D.M. 69, J 04 Zwissler, 1.G. 227, 260
Wood. J.L. 405, 406, 419
SUBJECT INDEX
Alumina 3,13, 19,34--36,252-255 see a/so properties 3

under Fiber( s) reaction forming 22
Applications 391-419 self-propagating high temperature
Automotive applications 397 synthesis (SHS) 28,29,131-\33
sintering 20
Biot's modulus 282-286 sol-gel 22-27
Boron carbide 37 via chemical routes 22
Boron nitride 36, 37 composites
types of I
Carbon 38 cordierite 3
fiber(s) creep 41-42
Fabrication 53----62 fracture 42-45
PAN based 54--57 glass 14, 17, 19
pitch based 58-60 borosilicate 13, 17
properties 58-61 quartz 17
structure 62 sodalime 3,13,15,17
fiber/glass transition temperature 14, 15
fiber/SiC glass-ceramics 16,17, 18,19
Carbon/carbon composites 398-409 intermetallics 38-41
brakes 405-406 magnesium oxide 13
thermal conductivity 398-400 mullite 3, 13, 36
oxidation protection 400 reinforcements 47
Ceramic matrix composites ,
applications 9,10,391-419 service temperature 2, 4, 5
via polymeric precursors 416 sialons 37, 38
environmental barriers coatings 416--417 silica 15, 16,17
Ceramic(s) silicon carbide 3, 13,30,31
advanced 2
bonding II silicon nitride 13, 31--34
chemical vapor deposition 22 sol-gel 22-27
coefficient of thermal expansion 4, 7
conventional 2 Chemical vapor deposition 119-124
creep 41, 42 Chemical vapor infiltration 119-124,
crystal structure 11-14 412-416
density 4,6 Coefficient of thermal expansion 188-187
fracture 42-45 Cold pressing 108
fracture toughness 4, 5, 8 Combustor liners 397
hot isostatic pressing (HIP) 21 Coupling phenomena 380-382
injection molding 20 Crack bridging 318-320
materials 12 Crack deflection 147--149,301-308
matrix II
matrix materials 29-41 Damage 215
microwave sintering 27 accumulation 234--237
noncrystalline 13, 14 modes 234
440 Subject Index
zone 234 Eutectic composites 255, 256

Delamination 384, 387, 389, 391
Density 49, 198-200 High temperature capability 249-257
Dielectric constant 201, 202 Hot gas filter 396
Directed oxidation 117, 118 Hot pressing 108, 124, 131, 135
Edge effects 387-389 Interlaminar shear strength, ILSS, see under

Elastic constants 172-187 interface
energy methods 180, 192 Infiltration 115-134, 291-295
exact methods 180 Interface 139-166
micromechanical approach 179, 185 area 139-142
self-consistent field methods 179 bonding 145-147
strength of materials approach 172-176 crystallography 142
Electrical properties 200-202 debonding 301-313
Electrophoretic deposition 134, 135 crack deflection 301-308
Eutectic oxide composites 350, 351 indentation tests 159--164
interlaminar shear strength 155-157
Fatigue 230-242 laser spallation 165, 166
conventional approaches 232-239 roughness 332--344
novel approaches 239-239 strength 139, 142, 145, 148150, 157166
measurement 151--164
Fiber(s)
alumina 62-74 Lamina 356-361
carbon 52--62 effect of orientation 361-365
ceramic 47--103 Laminated composites 280, 370-387
comparison 93-94 Nonfibrous 389-392
definitions 48-50 LANXIDE 117,118
effect of temperature 89-94 Load transfer 205-215
flexibility 50, 51
nonoxide 74-86 Macromechanics 355-389
oxide 62-74 Magnetoplumbite 345
NexteI66-71,101-103 Mechanical testing 215-225
Nicalon 80-9, 108 Modulus 225-230
pullout 156-159,312-318
pushout 159-163 Orthotropy 357-361
Saffil 66, 71
Saphikon 66, 71, 73, 91 Polymer infiltration processing (PIP)
SCS 76-78, 113 126-131
Silicon carbide 75-86, 108 Porous matrix 349, 350
Sigma 75-77 Porous radiant burners 396
strength Processing of CMCs 107--134
Weibull analysis 94--102
Sylramic 82 Raman spectroscopy 82, 85, 90, 325, 326
Tyranno 82, 85, 90 Reaction bonding III-lIS, 118, 124
Yajima process 78-81 Residual stresses 280-282
Filter(s) 396
Shear lag model 205--215
Glass see under ceramic(s) Silicon carbide/silicon nitride composite 113
Graphite Silicon carbide whisker/alumina cutting
compliance 52, 53 tools 407-412
elastic modulus 52, 53 Silicon carbide matrix composites 412-419
stiffuess 52, 53 Sintering 112, 115, 124, 127, 135
structure 52 Slurry 108--111,115,125-128,135
Subject Index 441
Sol-gel 22-27, 125, 126, 135 Eyring equation

Steel 19 Newtonian
Strength 225--230 Voids 198-200,244
Strength vs. density 393, 394 Weibull analysis 94-103
Wettability 142--145
Thennal conductivity 196-198 WHIPOX 350, 351
Thennal fatigue 282-289 Whiskers 82, 86-89
Thennal shock 282-289 rice-hull 87, 88
Thennal stresses 150, 151,263-289 vapor-liquid-solid (VLS) 87, 89
particulate composites 266-269 Work-of-fracture see Toughness
fiber reinforced composites 269-280 Y AG see Yttrium aluminum garnet
Toughening mechanisms 295-352 Yttrium aluminum garnet 166,252-255
Interface engineering 320-349 Young's modulus 171-175,183 see also
Transfonnation 297, 298, 392, 346 modulus
Toughness 225-230
Tyranno see under Fiber(s) Zirconia toughened alumina 297,298
Viscosity

Chawla

Caricato da

Informazioni sul documento

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Chawla

Caricato da

Copyright:

Formati disponibili

CERAMIC MATRIX COMPOSITES

SPRINGER SCIENCE+BUSINESS MEDIA, LLC

Chawla, Krishan Kumar, 1942-

TA418.9.C6 C42 2002

Copyright© 2003 by Springer Science+Business Media New York

Pennission for books published in Europe: oennissions@wkap.nl

Printed on acid-free paper.

Kanika, Anita, Nikhilesh, and Nivedita

Preface to the Second Edition

Materials science and engineering (MS&F,) is by its very nature an

It is a truism that technological development depends on advances in the field

It is pertinent to quote from some other sources about a fundamental change

Designing materials for specific applications is, indeed, the underlying

The plan of the book is as follows. After an introductory chapter, we first

There remains the pleasant task of acknowledging a number of people without

Chawla, K.K. (1987) Composite Materials, Springer-Verlag, New York, p. 1.

Professor Chawla obtained his B.S. from Banaras

He has published extensively in the areas of processing, microstructure, and

He has organized international symposia and edited books in the area do

A composite material is a material that has a chemically and/or physically

Fig. 1.1 Types of composites based on the form of reinforcement.

can be polymeric, metallic, or ceramic. Polymeric matrix composites containing

The choice of individual components, in regard to their synthesis and chemical

(i) Traditional or conventional ceramics: These are generally in a monolithic

Fig. 1.2 Performance of a composite is linked to some important factors : composition

Table 1.1 Properties of some important ceramic materials

Material Young's Poisson's Coefficient of

SiC 420 0.22 4

Plastic deformation being an energy absorbing process, the fracture process in

1.1 CERAMICS VIS-A- VIS OTHER MATERIALS

It is instructive to compare the use or service temperature of different types of

Polymers Metals Ceramics

Improving the toughness of ceramic materials is thus a major objective. One of

6 -- _____________ FeAI, NiAl

Fig. 1.4 Density of some materials.

There have been also major developments in the area of reinforcement

:::::::::::: ~~::V::::::: _~1!!~!~ ___________ SiC

Incorporation of fibers, whiskers or particles in a ceramic matrix can result in

debonding at the fiber/matrix interface, crack deflection, fiber bridging, and

Aluminum alloys 20-50

1.2 POTENTIALITIES OF CERAMIC MATRIX COMPOSITES

exciting potential applications. At the risk of repeating myself, I would like to

Potentially, ceramic components can be introduced into automotive engines,

Other structural part applications, besides the heat exchangers mentioned

Green, D.J. (1998) An Introduction to the Mechanical Properties 0/ Ceramics,

CERAMIC MATRIX MATERIALS

In this chapter we briefly review the basic characteristics of ceramics

2. 1 BONDING AND CRYSTALLINE STRUCTURE

Ceramics exist as both crystalline and noncrystalline compounds. Glasses, a

Table 2.1 Some important ceramic materials.

Single Oxides Alumina (AI 20 3)

2.2 COMMON CRYSTAL STRUCTURES IN CERAMICS

Common crystal structures in ceramics are:

Table 2.2 Physical and mechanical characteristics of some ceramic materials.

Al20 3 3.9 2050 380 7-8 2-4

Figure 2.1 shows some of these common crystal structures in ceramics.

2.3 NONCRYSTALLINE CERAMICS

As we mentioned above, glasses are, microstructurally, not solids, but very

When a liquid is cooled, molecular or atomic rearrangement occurs such that

Silica-based glass is a generic term representing an interesting and versatile

Glass-ceramics represent a special class of ceramics. They are polycrystalline

Table 2.3 Some important types of glasses.

Glass Softening point Density, p, Toughness, KJv

Fused Quartz 1580 2.6