Sie sind auf Seite 1von 31

Composites Science and Technology 42 (1991) 25-55

Interfacial Mechanics in Fibre-Reinforced Metals


T. W. Clyne & M. C. W a t s o n
Department of Materials Science and Metallurgy, University of Cambridge,
Pembroke Street, Cambridge CB2 3QZ, UK
(Received I0 August 1990; revised version received 20 September 1990;
accepted 29 October 1990)

ABSTRACT
It is well known that mechanical character&tics of the fibre-matrix interface
in MMCs have a strong influence on various properties. However, there is
much uncertainty surrounding the question of how best to control the
structure of the interfacial region in order to optimise particular types of
composite performance. In this paper current thinking on how interfacial
characteristics affect composite properties is briefly summarised. The
importance of different types of stress relaxation processes is emphasised.
This is followed by some observations about testing procedures designed to
measure bond strength in MMCs. It is noted that these invariably involve
predominantly shear loading, whereas there is a need to explore the response
of interfaces to tensile and mixed mode conditions. Finally, some observations
are presented on interfacial chemical reactions and the development of fibre
coatings.

1 INTRODUCTION
When compared with material produced in the early 1980s, significant
improvements in the mechanical properties exhibited by metal matrix
composites (MMCs) have been obtained over the last few years. This has for
the most part been achieved by processing route improvements, leading to a
reduction in the level of gross defects such as matrix porosity and
inhomogeneities in the spatial distribution of reinforcement. Further
progress now requires a fundamental understanding of the interplay
25
Composites Science and Technology 0266-3538/9I/$03.50 1991ElsevierSciencePublishers
Ltd, England. Printed in Great Britain

26

T.W. Clyne, M. C. Watson

between microstructural features and mechanical performance in these


materials. Prominent among the features of critical importance are the
factors controlling the mechanical characteristics of the interface, including
the stress state resulting from externally applied loads and temperature
changes. As a simple example, a high bond strength (often favoured by a
localised chemical reaction) will tend to improve the stiffness and workhardening rate but give a low toughness and ductility. In general, it is likely
that a compromise will need to be struck between achieving efficient load
transfer with a strong bond and promoting greater ductility via inelastic
processes occurring at the interface.
In this paper a brief survey is first given of how interfacial characteristics
can affect the performance of metal matrix composites. This is followed by
an outline of certain tests designed to characterise the mechanical response
of the interface. Finally, some examples are presented of means by which a
degree of control can be exercised over interracial structure. Much of the
treatment is oriented towards short-fibre composites. Particulate and longfibre reinforcement may be considered as special limiting cases; in most
instances the effect is a fairly obvious simplification.

2 THE ROLE OF T H E I N T E R F A C E IN M M C P E R F O R M A N C E

2.1 The meaning and significance of bond strength


A c o m m o n problem in characterising interfacial behaviour in MMCs lies in
incomplete identification of the various parameters which may be
significant. Relevant system properties include a critical shear stress for
shear debonding, z., and a coefficient of sliding friction, #. The value of r .
will dictate the onset of frictional sliding, which will then progress at a rate
determined by the shear stress rfr, given by
g f r ~--- - - ~ 0 " r

(1)

where a t is the radial stress (normal to the interface) at the point concerned,
which must be negative (compressive) if rr~ is to have a non-zero value. (This
is usually the case in MMCs as a result of differential thermal contraction.)
Under applied load parallel to the fibre axis (x-direction) differential Poisson
contraction effects often cause the value of a r to vary along the length of a
(discontinuous) fibre:
or(x) = a~aT + O',av(X)
(2)
where cry,,T is the radial stress from differential thermal contraction (see
below). It is c o m m o n to assume that the shear stresses at which both
debonding and frictional sliding take place are independent of other

Interfacial mechanics in fibre-reinforced metals

27

components of the stress state. (This is somewhat analogous to the basis of


the Tresca criterion for plastic flow.)
While the ease of debonding and frictional sliding are important for the
behaviour of the composite, particularly in terms of energy absorption and
fracture mechanics, there is also interest in the conditions under which
decohesion may occur, causing interfacial cavitation and/or the opening of a
crack along the interface. It might be expected that it would be possible to
identify a minimum (tensile) value of the radial stress, at., or, more probably,
the hydrostatic component of the stress state, all,, necessary for decohesion
to take place. Among other interfacial processes essentially governed by the
local stress field are mass transport by diffusion and local dislocation
rearrangements. While these do not involve the breaking of a fibre-matrix
bond, they may be sensitive to the structure of the interface. A further point
to note is that the stress field will be determined not only by the partitioning
of the applied load between the two phases but also by differential thermal
contraction (and possibly by prior plastic deformation of the matrix). In Fig.
1 a schematic illustration is given of the nature of these stresses and of the
interfacial processes triggered by them.
Drawing a distinction between interfacial decohesion and sliding in the
way described above is equivalent in conventional fracture mechanics to
identifying mode I and mode II loadings respectively acting on a crack front.
(The equivalent of mode III loading, which would arise from a twisting of the
fibre about its axis, would be very unusual in a fibre composite.) In practice,
an interface will often be subjected to a stress state tending to cause both
shearing and decohesion at the interface, i.e. a mixed mode loading situation.
The propagation of a planar crack constrained to follow an interface
between two dissimilar materials under mixed mode elastic loading
conditions can be analysed to give an interfacial tYacture toughness. L- 3 This
approach can in principle be extended to treat fibres in a metallic matrix, but
there are various complications, particularly as matrix plastic flow is likely
to occur. Evans 4 has identified the ratio of the critical strain energy release
rate for the interface to that for the matrix, Gic/Gm,as determining the ease
of debonding in ceramic matrix composites, with a small value favouring
propagation of the interfacial crack. There is as yet, however, very little
information available on Gic values in MMC systems and reliable
measurement of this parameter presents certain difficulties with a nonplanar interface.
Few direct correlations have been established between interfacial
characteristics and composite performance indicators. However, poor
bonding has been shown to result in a reduced elastic modulus 5 and workhardening rate. 6 In the following sections the implications of interface
structure for a number of key areas are discussed in general terms.

28

T ~ Clyne, M. C. Watson

~ C t AT stresses]

+ applied

!iiiii /1

load)

,
Interfacial Vacancy

Interracial Debonding

Diffusion ~ ~CIi.~/~x

'~,

Interracial Frictional
Sliding, ~:fr= p. ~r

Interracial
Decohesion
/Cavitation
GH.

Fig. 1. A schematic view of the interracial stresses in a short-fibre MMC subjected to (a) a
temperature decrease and (b) a superimposed axial tensile load. Also shown are schematic
illustrations of some of the processes which might take place at the interface under the
influence of the stress field.

2.2 Stress relaxation processes


The stress-strain curve of a short-fibre or particulate M M C can be divided
into several regimes, as depicted schematically in Fig. 2(a). The elastic
portion is short or non-existent, because local plastic flow takes place very
readily in regions of the matrix (near the interface) where there are already
high residual deviatoric stresses as a result o f differential thermal
contraction. As this plastic flow becomes global throughout the matrix,
linear stress-strain behaviour is expected. This regime should exhibit a high
work-hardening rate (even if the matrix itself were to undergo no strain
hardening) because a uniform plastic deformation o f the matrix wilt raise the

29

Interracial mechanics in fibre-reinforced metals

Applied
Stress (MPa)
Global

_ . , UnrelaxedWo~
~

Harlning

Plastg'tY.,'~"~Failure
r'~"~'/
,,:
,:"

I~

~1
Ioo/

Relaxation

U.Zi ...:.r...TF~._.~.
_ ~
[ M - ' ~ " ~ ~ S i C ~ (s=lO) I

~al
El~iic

d.s~

(a)
300Applied
Stress
(MPa)

11o~
(b)

115~ S/rain

i [ as.received SiC ]

200 -

[ pre-oxidisedSiC ] x

100

[ AI (A356-T6) - 10qo SiC ]

2%

4%

6%

Su'ain

tc)
Fig. 2. The influence of stress relaxation processes on the tensile stress-strain behaviour of
short-fibre or particulate MMCs: (a) a schematic curve showing the processes controlling the
shape; (b) three curves for a SiC whisker-reinforced Mg-Li alloy at differenttemperatures and
strain rates;s and (c) two curves for SiC particulate-reinforced AI alloy, with and without the
prior formation of an oxide layer on the SiC. 9
misfit between the (stress-flee) shapes of the fibre and the corresponding hole
in the matrix; this transfers load strongly to the fibre and hence raises the
load-bearing capacity of the composite.
In practice, various stress relaxation phenomena tend to be stimulated by
the resulting sharp gradients of stress within the matrix, these being
processes which transfer load back from the fibre to the matrix. An example
of how stress relaxation can affect the behaviour can be seen in the stressstrain curves of Fig. 2(b). These are for a matrix in which diffusive processes
are fast at room temperature. 7.s Testing at high imposed strain rates or low
temperature gives little scope for diffusive stress relaxation processes. The

30

77. ~|i Clvne, M. C, Watson

effect of this is to raise the work-hardening rates but to reduce the ductility. If
the high fibre stresses caused by matrix plasticit.v are not relaxed in some way
then cavitation or fibre fracture are likely, leading to t:ailure of the specimen.
A similar type of effect is seen in Fig. 21c). In this case the provision ot'a thick
oxide laver on the interti~ce appears to p r o m o t e stress relaxation, raising the
ductility. This is probably 9 due in large part to decreased resistance to
intertitcial sliding, which also has the effect of reducing the stress in the
reinforcement. (It should, however, be noted that magnesium in the alloy
migrates to the oxidised inter[itce, reducing the age-hardening capacity of
the matrix and probably contributing to the observed effect.}
The simplest picture of stress relaxation mechanisms as a group oF
processes is obtained by considering the reduction the) effect in the misfit
strain between the shapes of fibre and hole. This is illustrated in Fig. 3 t'or
punching of prismatic dislocation loops'~'-" '~ inot to be confused with global

I- .4
!- -i

tat

a,'%

1
tbl

Fig. 3. Schematic depictions of t~o stress relaxation mechanisms. (a) Punching of


secondary prismatic dislocation loops and (bi mass transfer by diffusion (vacancy transport).
Both of these are shown relaxing the misfit strain l~r a short-fibre MMC resulting from a
temperature decrease.

lnterfacial mechanics in fibre-reinforced metals

31

plastic flow) and interfacial diffusion (vacancy transport) in response to the


misfit from differential thermal contraction.
Quantitative studies of sliding phenomena have often been hampered by
uncertainties about reinforcement shape effects. Figure 4 shows photoelastic
fringe patterns, obtained by the 3-D 'frozen stress' technique ~2 around an
ellipsoid and a cylinder of the same aspect ratio, stressed parallel to the axis.
It is immediately apparent that high order fringes, corresponding to high
shear stresses, appear at the 'corners' of the cylinder. Evidently local plastic
flow and/or interracial sliding will tend to occur very readily in these regions.

%
(a)

(b)
Fig. 4. Two photoelastic fringe patterns in the matrix around short ellipsoidal and
cylindricalfibresunder axial loading.Note the high shear stresses(high fringeorders)near the
'corners' of the cylinder.

32

77.~i C/yne, M. C. ~ t s o n

2.3 Cavitation and failure

Under most loading configurations high tensile stresses build up at the fibre
ends and these can cause interracial cavitation. Conditions for cavity
nucleation and growth at a stiffinclusion are uncertain, with some attthors, 1:~
for example, predicting an increase in the critical plastic strain for cavitation
with particle size and others t.* the reverse. Nevertheless, it is again evident
that fibre ends are preferred locations, and FEM computations have been
used to explore this. is
Much of the difficulty in laying down ground rules for tailoring interracial
properties lies in establishing a criterion (related to interracial strength
parameters) for the onset of interracial debonding/cavitation lwhich is
thought to lead rapidly to composite failure by growth and link-up of
cavities). One approach is to consider the hydrostatic component of the
stress state, o H, expecting a critical {tensile) value for cavity formation.
dependent on the bond strength (but with a maximum value corresponding
to nucleation within the matrix). A variation o f o Hwith position predicted by
the Eshelby model 16'~v is shown in Fig. 5ta). This confirms the sharp peak in
the matrix at the fibre end. This may be reduced by the differential thermal
contraction stresses expected at room temperature but, on the other hand, it
will be greater when the fibre end is fiat rather than rounded. A cavity
formed at the end of a fibre lying parallel to the applied stress axis can be
seen in Fig. 5(b). It may be deduced from the above that a useful goal for
composite production would be an interface prone to sliding but resistant to
cavitation.
3 BOND S T R E N G T H M E A S U R E M E N T
Techniques for measurement of interfacial characteristics can be classified
according to the phase angle of interracial loading generated during the
test.* (Interracial critical strain energy release rates will vary with this phase
angle.) Compact tension, flexural (4-point bend) and fibre pult-out push-out
tests respectively have phase angles of about 0, re/4 and rr/2--corresponding
to pure opening (I), mixed (I + II) and pure shearing (II) modes (see Section
2.1). Most measurements on M M C s have focused on shear debonding and
sliding (often using simple variants of the shear lag theory to interpret the
data), with little or no attempt to introduce substantial mode I components
to the loading. It might be argued that an interface exhibiting a high shear
debonding stress would also be expected to resist decohesion strongly. That
this is not necessarily true can be seen from the fact that pronounced
interracial roughness is expected to raise the former (in shearing mode) while
having little effect on the latter (in opening mode), This type of argument

Interfacial mechanics in fibre-reinforced metals

33

GH
""I-. I

t"7/
iii

"",..

150~P~
(a.)

(b)
Fig. 5. Cavitation is stimulated by high tensile hydrostatic stresses, particularly when these
occur at the interface.and the bond strength is not high. (a) A map of the hydrostatic stress
around an isolated ellipsoidal SiC particle in an Al matrix subjected to 100 MPa tensile load;
this prediction was obtained using the Eshelby method, t6''~ (b) A TEM micrograph showing
a cavity formed at the end of a SiC whisker in an A1 matrix after tensile straining.

34

77. W Clyne, M. C. Watson

provides an immediate insight into the dependence of Gi~ on the phase angle
of loading referred to above. It may also be noted in this context that there
may be practical benefits in promoting interfacial sliding {to encourage
energy absorption at the interface) while inhibiting decohesion and
cavitation (which tend to promote failure of the component). Under these
circumstances more sophisticated tests are required in order to monitor and
control the interracial mechanical characteristics and this is an area of
ongoing research.

3.1 Single fibre pull-out testing


This test, devised for long fibres, has been extensively applied to polymer
composites. It comprises the extraction or" a single fibre, half embedded
within a matrix, under simple tensile forces. Its main virtue is simplicity, with
the load/displacement data normally interpreted from a minor adaptation
of the shear lag theoryJ 8'19 A schematic illustration is given in Fig. 6 of the
axial distributions of normal stress in the fibre and shear stress at the
interface. These distributions are shown at three stages in the process,
corresponding to elastic distortion up until debonding, propagation of the
debonded portion and subsequent pure frictional sliding. Basic assumptions
of the shear lag model, such as no shear strain in the fibre and no transfer of
normal stress across the fibre end, are retained in most treatments of this
problem.
Analysis is usually divided into two distinct parts; the first corresponds to
the point of debonding and the second to the subsequent frictional sliding
process. It is conventional to assume that the peak in the load/displacement
plot corresponds to the debonding event, occurring at an applied stress %..
(This may not necessarily be the case: depending on the values of T,, l~ and
ara T, the load could in principle rise during the propagation of debonding,
shown as stage 2 in Fig. 6. However, Poisson contraction of the debonded
fibre will reduce a r (see eqn (2)), and hence limit the contribution to the load
from the debonded portion, so that this possibility is often ignored.)
For the relatively high values of the fibre aspect ratio, s (= L/2r), typical of
this type of test the stresses are very low along most of the fibre length. Some
variants of the basic model have been published. For example, Hsueh 2
incorporated the possibility of stress transfer across the fibre end and
ensured that the load carried by the free fibre is balanced by that in the
composite. This model leads to more complex equations, but the general
form of the curves are similar. In particular, the ratio of T. to a o. is usually
very close to that for the basic shear lag treatment. None of these models
takes account of the fact that the shear stress in the matrix should fall to zero
at the free surface from which the fibre emerges.

Interfi~cial mechanics in fibre-reinforced metals

(50*~ ( 5

Applied
Stress
(50

x=O
/

35

x=L
.-----.]i-

jdebonded

ent f

G_(x) = G ._+

~v(X)

121

~t (sr (x)"
L'
Fig. 6. Schematic illustration of stress distributions during the pull-out test.

The frictional sliding behaviour has also been analysed, taking account of
the effect of the Poisson contraction of the fibre in reducing the radial
compressive stress. On the basis of a crude assumption that only the fibre
carries an axial normal stress (with the resulting radial strain at the interface
producing a reduction in radial stress proportional to the matrix stiffness),
relationships are obtained between pull-out stress and length remaining

77. ~~ Civne M. C. Hatson

36

X=(

X=X )

debonded

Applied
Stress
G0

Displacement of
Loading Point

Or(X),= (J

(J0s~

! !ii.!

" <
u%~x, .... ~
Fig. 7.

"

+ G

'xl

i!

(o rx
.',~v0; ' 0

~TJ

S c h e m a t i c illustration of stress distributions d u r i n g t h e p u s h - o u t test.

e m b e d d e d , E, It is therefore possible for ,~ and ara- r to be e v a l u a t e d f r o m a


single pull-out load/displacement curve by finding the combination o f these
two parameters giving the best fit.
Although single fibre pull-out testing has been successfully applied to
M M C s (e.g. see Ref. 21), there are practical difficulties in specimen
preparation and handling when applying the procedure to composites with a
relatively stiffmatrix. In general, it is rather more convenient to use the pushout test described below.

Interfacial mechanics in fibre-reinforced metals

37

3.2 Single fibre push-out testing


This test is now being widely applied to MMCs and CMCs. -'2-26 It is usual
to examine a series of specimens with different fibre aspect ratios, all of these
being much smaller than those typical of pull-out testing. In this case the
Poisson effect raises, rather than lowers, the frictional sliding stress.
Schematic illustrations of the stress distributions and load/displacement
curve are shown in Fig. 7. Recording of the latter requires sensitive, purposebuilt equipment (particularly for fine fibres), but the basic operation can be
carried out with a conventional microhardness indenter. The SEM
micrograph in Fig. 8(a) shows a monofilament that has been pushed down in

(a)

(b)
Fig. 8. S E M micrographs of a Ti-SiC monofilament specimen afterpush-out testingwith a
conventional microhardncss machinc, showing (a) the indented fibreand (b) the pushed-out
fibre end protruding from the rcvcrsc facc.

38

T I,~-2C(vne, M. C. ~btson

Interracial
shear stress
(MPa)
i
0.08 i
!

0"061
0.04

"

Pv'r= 5

"3

~_)xperiment
L

Oo= o.29.,,P

0.02 i
[
r

ob

~.,
I

02

0.4

4-

-9,
P

0.6
0.8
1.0
Fractional distance along fibre

Fig. 9. Photoelastic measurements can be used to explore the validity of models for the
elastic behaviour during bond strength testing, From the comparison shown here it can be
seen that the Hsueh model is in fair agreement with a set of such measurements.

this way: for such large diameter fibres a quick examination in the optical
microscope will reveal whether the fibre has been displaced after the
application of a particular load. However, a difficulty then arises rrom the
uncertainty about whether the load found to be necessary for push-out
corresponds to the onset of debonding (ao,) or to frictional sliding along the
complete length (%s). An advantage of the test. however, is that a
comparison can be made with the load needed to push the partially pushedout fibre back into the matrix. The recording of a load similar to that for
initial push-out suggests that a0~ is being measured.
Again the analysis is carried out in two regimes, corresponding to the
elastic case and to pure frictional sliding. Hsueh 27 has presented treatments,
based on the shear lag analysis, for both of these. Figure 9 shows a
comparison between predicted curves and experimental data obtained with
the 3-D frozen stress photoelastic technique. These data were obtained by
using two photoelastic resins (with a stiffness ratio of about 2) for matrix and
fibre. Measurements of the fringe order, and of the principal stress
directions, as a function of position along the fibre were combined in a
simple Mohr's circle calculation to give the variation along the fibre length
of the shear stress parallel to the interface. It can be seen that this is in
reasonable agreement with the Hsueh model for this case.
The Hsueh model for the frictional sliding behaviour predicts, at least for
MMCs, a much more uniform distribution of shear stress along the fibre
length than in the elastic regime. This is a strong function of the fibre Poisson
ratio and the matrix stiffness. Figure 10 shows plots ofrrr for SiC fibres in Ti.
It is evident that for this combination (and most MMCs) the value of ~rr

Interfacial mechanics in fibre-reinforced metals

39

Interracial
shear
stress 30

(MPa)

20

s= 10

(~raT ~

S=2

10

SiC monofilament in Ti |
R/r = i0

1.1.=0.3
(~r&T = " 50 MPa
I

0.2

0:4

0:6

o18

1:o

Distance along fibre (mm)

Fig. 10. The frictional sliding stress in the pull-out test varies along the length of the fibre as
a result of the Poisson expansion of the compressed fibre. The significanceof this effect is
greater for large values of the fibre Poisson ratio and matrix stiffness.Predicted variations are
shown here for SiC fibres of 100/~m diameter, with various aspects ratios, in a Ti matrix.
remains close to/tara r. A consequence o f this is that the frictional push-out
stress, O-os,is predicted to rise in an approximately linear m a n n e r with fibre
aspect ratio. Predictions for ao. and aos as a function of fibre aspect ratio are
compared in Fig. 11 with experimental data for SiC monofilaments in an assprayed titanium matrix. These data were obtained with a simple
microhardness machine, so that there is initial uncertainty as to whether the
peak loads represent the onset or completion of debonding. However, it can
be seen that the experimental data are not consistent with the predicted ao.(S)
curves but conform to those for frictional sliding with a zfr value of about
35--40 MPa. This conclusion is supported by the observation of similar loads
for push-out and push-back operations. However, one effect of the limited
variation in frictional sliding stress along the fibre length is that good
agreement with experiment can be obtained over a wide range of # and Or/,r
combinations (giving the same product zfr)3.3 The full fragmentation test

A method that has been used for M M C s is the so-called 'full fragmentation'
technique. This procedure for deducing a shear strength involves embedding
a single fibre in a matrix, then imposing a large plastic strain in tension and
measuring the m e a n aspect ratio of the resulting fibre segments. Analysis is
based on a constant z, with the Weibull modulus of the fibre taken into
account. 2s'2s One o f the problems with the method is that it is unclear
precisely what is being measured, although it is presumably related to z..

T. 14/..Clyne, M. C. Watson

40

Debonding

stress
(MPa)

Fibre aspect ratio, s


2

1
.

SiC raonofilamentin '


prayedTi (E - 60 GPa)
r = 50 ~tra
R/r = I0

Is

aft, 80o

600

,g,
"U, = 65 MPa

"

0/7"

~. .......

i:.~==~m.=~,".=di.=Pt.=cedJ,

0.1

0.2

0.3

Specimen

0.4

thickness
(mra)

(a)
Pushout

0
.

stress
(MPa)

Fibre aspect ratio, s


2

1
.

( SiC raonofilamentin ")


isprayedTi (E _ 60 GPa)l

% =oo I
|

r=50 m

R/r = 10

~ "
i

,
.

0.1

-75 MPa

./"

400

(~ = - i00 MPa
raT/

E x p e r i m e n t a l"-l ~50~ MPa

,~4

Maximum u n d i s p l a c e d J .

0.2

J.

0.3

0.4

Specimen thickness (ram)

(h)
Fig. l l . When the push-out test is done with a conventional microhardness tester, there is
uncertainty about whether the recorded minimum push-out loads correspond to debonding
or frictional sliding. It can be seen here that, for these data obtained with SiC monofilaments
in sprayed Ti, the form of the debonding curve (a) is not consistent with the observed results,
while reasonably good agreement is observed with the frictional sliding predictions (b).

3.4 Other tests


A procedure can be applied to push the fibres into a bulk matrix, 22 with
accurate sensing of load and displacement: interpretation of data is complex
and application to M M C s has been limited. Estimates based on
thermodynamic arguments and debonding observations 29'3 have sug-

Interfacial mechanics in fibre-reinforced metals

41

gested a high af. value (several GPa) for AI-SiC: it certainly appears that this
bond is normally stronger than that for Ti-SiC, although the details are
probably sensitive to interfacial contamination or precipitation. More
experimental data are needed in order to confirm such effects. Various other
procedures have also been suggested, although all those for fibres tend to
generate predominantly mode II loading conditions. The development of
mixed mode loading conditions at fibre/matrix interfaces has been
discussed 31 for ceramic matrix composites, in which differential thermal
contraction can generate tensile radial stresses when the matrix has the lower
expansivity.
3.5 Some data for the Ti-SiC system
It is not appropriate here to attempt any systematic survey of published
data, but a limited comparison of results for Ti-SiC monofilament
composites is shown in Table 1. The strengths for an as-sprayed matrix
appear low and somewhat variable. These composites have little or no
interfacial reaction and the matrix also has some porosity (---5-10%),
leading to reduced radial compressive stress. The carbon-rich coatings on
the SCS fibres are expected to shear very readily when unreacted. However,
the pull-out values seem a little low and test data often appear less
reproducible with MMCs than those obtained by the push-out procedure.
The other composites were all given a heat treatment during fabrication
sufficient to form reaction layers. This raises T, compared with the assprayed composites, although further heat treatments have little effect. 23
Strength values obtained from the fragmentation test often seem a little
high; this may be a consequence of the simplifications in the analysis or of
the fact that the continued plastic flow of the matrix will raise the radial
compressive stress.
TABLE 1
Interfacial Shear Strength Data for Ti-SiC Monofilament Composites

Matrix

SiC
fibre

Ti
Ti
Ti
Ti-25AI-10Nb-3V
Ti-6AI-4V
Ti-15V-3AI-3Cr-3Sn
Ti-25Al-10Nb--3V
Ti-rAI-4V
Ti-rAI-4V

Sigma a
SCS-6 a
Sigma
SCS-6
SCS-6
SCS-6
SCS-6
SCS-6
Sigma

Fabrication
route
Spray deposit

Spray deposit
Spray deposit
Powder hot press
Diffusion bond
Diffusion bond
Diffusion bond

Hot press
Hot press

"lO0~am diameter, stoichiometric SiC, W core.


b 140/~m diameter, C-rich surface, C core.

Test
procedure

r,
(MPa)

Pull-out
Pull-out
Push-out

50
5
-110
150
120
120
180
345

Push-out
Push-out
Push-out
Push-out
Fragment
Fragment

rfr
Reference
(MPa) number
12
1
35
60
90
80
50
---

32
32
33
23
24
24
24
34
34

T. W. Clyne, M. C. Watson

42

4 INTERFACIAL CHEMICAL ATTACK

4.1 Types of attack


In some cases reaction is limited because one of the reactants becomes
consumed. An example is provided by 'Saffil' alumina fibres, which contain a
few per cent of SiO2, concentrated at the grain boundaries and on the free
surface. This silica is readily attacked by a strong reducing agent, such as
magnesium, present during fabrication. A c o m m o n complication is that
Saffil composites are often made by infiltration of a fibre preform held
together by a silica-based binder, which also tends to react during
processing. 35 The surface analysis data 36 s h o w n in Fig. 12 give
concentrations in the top few nanometres of the fibres after various
treatments. (Stripping with H F removes several nanometres of material.)
These data confirm that during infiltration magnesium from the melt
penetrates the fibres to a depth o f a few nanometres, corresponding to the
silica-enriched layer. This localised attack o f the fibre surface does not
appear to occur with magnesium-free melts, and this has been correlated
with a significantly lower interracial bond strength exhibited by such
composites. 3
There are a number of cases where uninhibited interfacial reaction causes
progressive degradation of properties. This is true of most titanium
composites, in which reactions are difficult to avoid. The reaction between
titanium and SiC to form TiC and TisSi338'39 has a large thermodynamic
driving force. It occurs during fabrication of SiC-reinforced titanium and
under high temperature service. Figure 13(a) and (b) shows reaction layers
formed with Sigma and SCS6 fibres. `* The latter has a carbon-rich surface,

As-Received

-SiO2 binder added -SiO2binder added


-AI-Mg infiltrated -AI-Mginfiltrated
-matrix d sso red -matrixd ssolved
-HI: stripped

AI

25-

-HF stripped
-SiO2 binder

.hi5
10

'0

Fig. 12. Shown here are XPS analyses of AI, Si and Mg levels in the surface layers of'Saffil'
fibres after various treatments. A brief exposure to HF acid is used to remove a thin surface
layer on the fibre, while prolonged immersion of a composite in sodium hydroxide solution
dissolves away all the matrix. These data confirm that the thin silicon-rich surface of the fibre
becomes impregnated with Mg during manufacture of the composite.

blterfacial mechanics in fibre-reinforced metals

43

(a)

(b)
Fig. 13. SEM images of(a) Sigma (stoichiometric) SiC and (b) SCS-6 (C-rich surface) SiC
monofilaments after 50 h at 800C in titanium, showing thick reaction layers and extensive
cracking.

77 14":Clyne, M. C. Watson

44

often thought to be protective; in I:act, it is consumed at a similar rate to


stoichiometric SiC, but it appears to act as a mechanical buffer against the
propagation of cracks into the fibre.
In many cases fibre-matrix reaction is very limited because, although
thermodynamically favoured, its kinetics are such that it progresses very
little during the heat treatment needed for fabrication or in service. Most A1SiC composites fall into this category, although a fabrication route involving
prolonged exposure of SiC to the aluminium melt causes extensive
reactions. "*~ The extent of reaction can be reduced by raising the level of
silicon in the aluminium alloy. Chemical reactivity at reinforcement/matrix
interfaces is reviewed in several papers in a recently published conference
proceedings/2
4.2 Effect on mechanical behaviour

A thick reaction layer normally impairs composite performance. This again


may be illustrated by reference to the Ti-SiC system. The effect is clear in Fig.
14, showing fracture energy as a function of reaction layer thickness for
particulate composites. One interpretation of such data involves a simple
fracture mechanics approach, assuming flaws to be present of a size equal to
the reaction layer thickness. "~3-'~5 In fact, fracture appears to occur rather
differently in some systems when thick layers are present. For example, in
Ti-SiC particulate composites there is a strong tendency (apparent in Fig.
15) for the crack to follow the boundary between the unconsumed SiC and
the reaction layer) 6 It is thought that the deleterious effect of the reaction
layer may in this case be aggravated by the 5% volume contraction
associated with this reaction, which sets up tensile interfacial stresses, These
counteract the effects of differential thermal contraction and even a
relatively thin layer is predicted to put the interface into net tension before
any external stress is applied. '~6
nominal
fracture
energy 300
(10 m2 )
2OO

I00

reacuon layer
thickness (p.m)

Fig. 14. N o m i n a l fracture energy of T i - S i C particulate composites d u r i n g impact loading.


as a function o f the thickness o f the interracial reaction layer p r o d u c e d by heat treatments. ~6

Interfacml mechanics in fibre-reinforced metals

45

(a)

(b)
Fig. 15. Optical micrographs of Ti-SiC particulate composites after impact testing,
sectioned normal to the fracture surface, showing specimens with reaction layer thicknesses
of about (a) 0-2pm and (bt 5pro.

5 FIBRE COATINGS

5.1 Coating techniques


Certain surface modifications, such as the promotion of an oxide layer, can
often be achieved by a simple procedure such as heating. However, carefully
specified structures may require the use of highly specialised techniques.

T. ~~ Clyne. M. C. Watson

46

Coatings have been deposited by techniques such as sputter deposition,


physical vapour deposition ~" (PVD) and (plasma-assisted) chemical vapour
deposition (CVD). The CVD process relies on the thermal decomposition of
a gas on a hot fibre. This is used to manutacture the SiC monofilaments
considered above, so that the prospect of incorporating the coating
operation into the fibre manufacturing process seems commercially
attractive. Problems may arise in finding a suitable carrier gas and
establishing a deposition procedure for selected coating materials. In
addition, control over factors such as deposition temperature, and the
microstructure and stress state of the deposit, will in general be limited.
Sputter deposition is an alternative technique, offering good control over
deposit density, stress state, "ts composition and thickness, but at a penalty in
cost and processing time.

5.2 Structure and thermodynamic stability


From a thermodynamic point of view, highly stable oxides emerge as strong
contenders for barrier coatings. As an example, consider the Gibbs free
energy of formation of several candidate oxides (Y203, HfO2 and ZrO2 in
Table 2) for use in titanium matrix composites (the free energy of formation
of TiO2 is also shown). It can be seen that Y 2 0 3 is the most stable of the
candidate oxides. Further data concerning entropy changes during oxygen
dissolution in titanium are needed for detailed predictions, but it may in any
event be noted that there is some evidence "~9 that both ZrO 2 and HfO2 can
undergo appreciable reaction with solid titanium, while Y203 is essentially
stable. These thermodynamic data give no information on the possible
formation of other compounds such as mixed oxides or the possibilit5 of
TABLE 2
S u m m a r y of T h e r m o d y n a m i c and Diffusion D a t a R e l e v a n t
for C h o i c e o f B a r r i e r L a y e r M a t e r i a l s'~-5"

Element
X

For o_ride o f X
AGloooK

D x at tO00 K

Do at IO00 K

( k J m o l e - l)

(m 2 s - ~i

(m" s - l)

Ti

-710

--

--

A1

-893

---

--

Mg
Y

- 996
-1080

--1-3 x 10 -21

-l ' 0 x 10 -Z6

Zr

-840

1 8 x 10 -26

4-t x 10 -1'~

Hf

- 892

NA

NA

Interfacial mechanics in fibre-reinforced metals

47

reaction with the fibre (e.g. carbide formation). Information on possible


fibre/coating reactions are incomplete (especially for yttrium compounds),
although in most cases the high stability of the oxide will prohibit any such
reaction. It may be noted that other types of coating are also being explored
for titanium composites, notably TiB2.

5.3 Stresses and mechanical stability


In practice, the avoidance of mechanical damage to the coating is often the
most important objective. The need for a large grain size (and low porosity)
points to a relatively high coating formation temperature. Sputter
deposition has been used, 5 with a substrate temperature of around 500C.
This leads to the danger of cracking as a result of differential thermal
contraction stresses. The stress field calculations 5t shown in Fig. 16(a)
indicate that large tensile hoop and axial stresses will tend to arise in a yttria
coating on a SiC filament during cooling. Cracking caused by such stresses is
apparent in Fig. 16(b). Matrix plastic deformation close to the interface
during thermal cycling may also cause damage. Mehan et al. 52 demonstrated
the value of a YEO3 layer in preventing attack of SiC in the presence of a
Ni-Cr alloy, but also showed that (on a planar substrate) this layer
underwent spallation after only one thermal cycle.
Fortunately, there is scope during the sputtering process for these stresses
to be offset by the so-called 'atomic peening' effect '*a in which, depending on
the sputtering gas pressure, energetic back-scattered neutral atoms can
cause transient formation of interstitials and hence generate a large
compressive stress. Schematic illustrations of this mechanism and of the
contributions to the final stress state, as a function ofsubstrate temperature,
are shown in Fig. 17. It has been confirmed 53 that a net compressive stress
can be induced, giving good mechanical stability.

5.4 Duplex coatings


Mechanical considerations have led to suggestions 5 for a duplex barrier
consisting of a metallic layer adjacent to the fibre with an overlayer of the
metal oxide. This ductile metallic underlayer offers the possibility of
reducing the deleterious effect of brittle interracial layers, by acting to
prevent crack propagation into the fibre. In the case of titanium, since a
residual oxygen content is always present in the matrix, a suitable choice of
metal/oxide system also offers potential for oxide self-repair by means of
oxygen gettering from the matrix (see below).
In practice, it is preferable to deposit a yttrium coating which can be
partially oxidised to yttria, for example by heating in air. The resultant

77. ~t,: C(vne, M. (.7. Watson

48

Stress
(MPa)
40O

tfoop

30O

200

, /

100

Radial

.~

0
Radial
distance

-I00
0

10

2o

30

4o

50

(~.m)

tai

Fig. 16. Predicted stress distributions ta) are shown here for W-cored SiC monofilament
with a 2-!tin yttria coating, after cooling through 500 K. That these stresses can be sufficient to
cause serious damage is evident from the SEM micrograph (b), which is of a SiC
monofilament with a yttria layer about 1-2 ~m in thickness, after cooling down from 900:C.
Both hoop and axial cracking has taken place.

Interfacial mechanics injbre-reinforced

metals

(TE)

Iorkwtion

Growing
deposit

Sputteting target

In-plane Stress in
Coating at Room
Temperature

Quenched-in
Vacanciesetc

-0.3 T,

t/

I-

I
-- .-

. .*

I. *-

Differential
thermal
contraction

Substrate
Temperature
ROOlll

CompressionTcmpcratun
__._.-._._._._._.-.-.-.-.-

Atomic
Peening

Fig. 17. Atomic peening during sputter deposition can induce large compressive stresses in
a coating. Shown here are (a) a depiction of the process and (b) a schematic illustration of the
effect of formation temperature on the final (in-plane) stress in a sputtered layer. for a case
where the layer has a higher expansivity than the substrate.

is shown in Fig. 18. A better method of actually forming the outer


yttria layer is to allow the yttrium to getter oxygen from solution in the
titanium, so that the protective layer is formed in situ. The X-ray maps
shown in Fig. 19 demonstrate that this gettering operation has taken place.
Finally, Fig. 20 illustrates the protective action of the coating, with the effect
of mechanical damage apparent over a portion of the interface.

structure

Fig. 18.

An SEM micrograph of a duplex Y 'Y203 coating on a SiC monofilament,


produced by heating a Y-coated fibre in a limited oxygen supply

i
I

Fig. 19.

50 t.tm

'1

X - r a y m a p s o f a Y - c o a t e d S i C f i b r e i n T i a f t e r 2 h a t 9 5 0 C , showing the distribution


of yttrium and oxygen.

Interfacial mechanics in fibre-reinJbrced metals

51

Fig. 20. SEM micrograph showing an interface in a composite exposed to 950:C for 2 h.
The coating is intact over most of the circumference; note, however, that there is a region near
the bottom left where it has become mechanically damaged and some attack has taken place.

6 SUMMARY
The nature of the fibre-matrix interface influences the performance of
MMCs in a number of ways. The following points have been identified.
(a)

A strong bond promotes load transfer to the fibres, raising the


stiffness and work-hardening rate for short-fibre and particulate
composites.
(b) The onset of failure in MMCs is frequently provoked by fibre
fracture or cavitation in the matrix at the fibre ends. These arise from
high fibre stresses, which are rapidly stimulated by unrelaxed plastic
flow throughout the matrix. The nature of the interface has a strong
influence on stress relaxation processes, which include interfacial
sliding, interfacial diffusion and generation of secondary dislocations. The promotion of such relaxation mechanisms can
improve the ductility of MMCs by postponing the onset of
catastrophic processes.
(c) It follows that, in measuring the strength of an interface, a distinction
should be drawn between the resistance to shear debonding/
frictional sliding and the resistance to normal debonding/cavitation.
It may be advantageous to reduce the former but raise the latter.
Ideally the interfacial fracture toughness should be measured for
mode I (opening), mode II (shearing) and for mixed mode loading

52

T. ~ Clyne, M. C. Watson

conditions. In practice, although this can be done for planar bimaterial interfaces, virtually all tests developed so far for fibre
composites involve pure shear (mode II) loading.
(d) Fibre-matrix chemical attack can have a strong effect on interfacial
mechanics. A limited degree of reaction usually seems to raise the
bond strength, but substantial attack makes the interface prone to
cracking and can render the composite very brittle. This may be
partly due in some cases to a change in the stress state caused bv the
volume change on reaction.
(e) The design of fibre coatings is a complex area currently under
intensive study. Some information has been given about the factors
to be considered in optimising the structure of barrier coatings
produced by sputter deposition for use in titanium composites.
(f) It has been emphasised throughout that an appreciation of the
interfacial stress state during fabrication and service is of
considerable importance.

ACKNOWLEDGEMENTS
The authors are grateful to a number of their colleagues, notably Dr P. J.
Withers, Dr R. R. Kieschke and Mr A. J. Reeves, for contributions to the
work described and for useful discussions. Financial support for one author
(M.C.W.) is being provided by the Interdisciplinary Research Centre for
High Performance Materials at Birmingham, and the collaboration of Prof.
M. H. Loretto, the Director of the Centre, is gratefully acknowledged.

REFERENCES
1. Hutchinson, J. W., Mear, M. E. & Rice, J. R., Crack paralleling an interface
between dissimilar materials. J. App. 3/Iech. (Trans. ASME), 54 (1987) 828-32.
2. Rice, J. R., Elastic fracture mechanics concepts for interracial cracks. J. App.
Mech. (Trans. ASME), 55 (1988) 98-I03.
3. Charalambides, R G., Cao, H. C., Lund, J. & Evans, A. G., Development era test
for measuring the mixed mode fracture resistance of bimaterial interfaces.
Mech. Mater., 8 (1990) 269-83.
4. Evans, A. G., The mechanical performance of fiber reinforced ceramic matrix
composites. In 9th Rise Syrup., ed. S. I. Anderson, H. Lilholt & O. B. Pedersen.
Rise National Laboratory, Rise, 1988, pp. 497-502.
5. Takehashi, H. & Chou, T. W., Transverse elastic moduli of unidirectional fibre
composites with interfacial debonding. Metall. Trans., 19A (1988) 129-35.
6. Aboudi, J., Constitutive equations for elastoplastic composites with imperfect
bonding. Int. J. Plasticity, 4 (1988) 103-25.

Interfacial mechanics in fibre-reinforced metals

53

7. Mason, J. E, Warwick, C. M., Charles, J. A. & Clyne, T. W., Magnesium-lithium


alloys in metal matrix composites--a preliminary report. J. Mat. Sci., 24 (1989)
3934- 46.
8. Mason, J. F. & Clyne, T. W., Creep behaviour of whisker-reinforced Mg-Li
alloys. Acta Met. (submitted).
9. DaSilva, R., Caldemaison, D. & Bretheau, T., Interface strength influence on the
mechanical behaviour of A1/SiC particle metal matrix composites. In Interfacial
Phenomena in Composite Materials 1989, ed. F.R. Jones. Butterworths,
London, pp. 235-41.
10. Ashby, M. F. & Johnson, L., On the generation of dislocations at misfitting
particles in a ductile matrix. Phil Mag., 20 (1969) 1009-22.
11. Taya, M. & Mori, T., Dislocations punched out around a short fibre in a short
fibre metal matrix composite subjected to uniform temperature change. Acta
Met., 35 (1987) 155-62.
12. Withers, P. J., Smith, A. N., Clyne, T. W. & Stobbs, W. M., A photoelastic
examination of the validity of the Eshelby approach to the modelling of MMCs.
In Fundamental Relationships between Microstructure and Mechanical
Properties of MMCs, ed. P.K. Liaw & M.N. Gungor. TMS-AIME,
Warrendale, PA, 1990, pp. 225-40.
13. Goods, S. H. & Brown, L. M., The nucleation of cavities by plastic deformation.
Acta Metall., 27 (1979) 1-15.
14. Tanaka, K., Mori, T. & Nakamura, T., Cavity formation at the interface of a
spherical inclusion in a plastically deformed matrix. Phil. Mag., 21 (1970)
267-79.
15. Nutt, S. R. & Needleman, A., Void nucleation at fibre ends in A1-SiC
composites. Scripta Met., 21 (1987) 705-10.
16. Eshelby, J. D., The determination of the elastic field of an ellipsoidal inclusion
and related problems. Proceedings of the Royal Society, A241 (1957) 376-96.
17. Withers, P. J., Stobbs, W. M. & Pederson, O. B., The application of the Eshelby
method ofinternal stress determination to short fibre metal matrix composites.
Acta Met., 37 (1989) 3061-84.
18. Lawrence, P., Some theoretical considerations of fibre pullout from an elastic
matrix. J. Mat. Sci., 7 (1972) 1-6.
19. Chua, P. S. & Piggott, M. R., The glass fibre-polymer interface. I: Theoretical
considerations for single fibre pullout tests. Comp. Sci. Tech., 22 (1985) 33-42.
20. Hsueh, C. H., Elastic load transfer from partially embedded axially loaded fibre
to matrix. J. Mat. Sci. Letts, 7 (1988) 497-500.
21. Kieschke, R. R. & Clyne, T. W., Pull-out testing of SiC monofilaments in a
spray-deposited Ti matrix. In lnterfacial Phenomena in Composite Materials
1989, ed. E R. Jones. Butterworths, London, 1989, pp. 282-93.
22. Marshall, D. B. & Oliver, W. C., Measurement of interfacial mechanical
properties in fiber-reinforced ceramic composites. J. Amer. Ceram. Soc., 70
(1987) 542-8.
23. Eldridge, J. I. & Brindley, P. K., Investigation of interfacial shear strength in a
SiC fibre/Ti-24A1-11Nb composite by a fibre push-out technique. J. Mat. Sci., $
(1989) 1451-4.
24. Yang, C. J., Jeng, S. M. & Yang, J. M., Interfacial properties measurement for
SiC fiber-reinforced titanium alloy composites. Scripta Met. et Mat., 24 (1990)
469-74.

54

T. W. Clyne, M. C. Watson

25. Verpoest, I., Desaeger, M. & Keunings, R., Critical review of direct
micromechanical test methods for interracial strength measurements in
composites. In Controlled Interphases in Composite Materials, ed. H. Ishida.
Elsevier, New York, 1990, pp. 653-66.
26. Sachse, W., Measurement of the interfacial strength of fibres and thin films. Mat.
Sci. & Eng., A126 (1990) 133-9.
27. Hsueh, C. H., Evaluation of interfacial shear strength, residual clamping stress
and coefficient of friction for fibre-reinforced ceramic composites. Acta Metall.
Mater., 38 (1990) 403-9.
28. Le Peticorps, Y., Pailler, R., Lahaye, M. & Naslain, R., Modern Boron and SiC
CVD filaments: A comparative study. Comp. Sci. & Tech., 32 (1988) 31-55.
29. Flom, Y. & Arsenault, R. J., Interfacial bond strength in an AI alloy 606I-SiC
composite. Mat. Sci. & Eng., 77 (1986) 191-7.
30. Li, S., Arsenault, R. J. & Jena, P., A quantum chemical study of adhesion in AISiC. J. Appl. Phys., 64 (1988) 6246-53.
31. Charalambides, P. G. & Evans, A. G., Debonding properties of residually
stressed brittle matrix composites. J. Amer. Scram. Soc., 72 (1989) 746-53.
32. Kieschke, R. R. & Clyne, T. W., Control over interracial bond strength in Ti/SiC
fibrous composites. In Fundamental Relationships between Microstructure and
Mechanical Properties ofMMCs, ed. P. K. Liaw & M. N. Gungor. TMS-AIME,
Warrendale, PA, 1990, pp. 325-40.
33. Watson, M. C. & Clyne, T. W., The use of single fibre pushout testing to explore
interfacial mechanics in SiC monofilament-reinforced Ti. Acta ~,Iet.
(submitted).
34. Le Petitcorps, Y., Pailler, R. & Naslain, R., The fibre/matrix interfacial shear
strength in titanium alloy matrix composites reinforced by SiC or B CVD
filaments. Comp. Sci. & Tech., 35 (1989) 207-14.
35. Li, C. H., Nyborg, L., Bengtsson, S., Warren, R. & Olefjord, I., Reactions
between SiO2 binder and matrix in 6-AI203/A1-Mg composites. In Interfacia/
Phenomena in Composite Materials 1989, ed. F. R. Jones. Butterworths,
London, 1989, pp. 253-7.
36. Cappleman, G. R., Watts, J. F. & Clyne, T. W., The interfacial region in squeeze
infiltrated composites containing 6-alumina fibre in an aluminium matrix. J'.
Mat. Sci., 20 (1985) 2159-68.
37. Clegg, W. J., Horsefall, I., Mason, J. F. & Edwards, L. F., The tensile deformation
and fracture of A1-Saffil metal matrix composites. A cta Met., 36 (1988) 2151-9.
38. Rhodes, C. G. & Spurling, R. A., Fibre-matrix reaction zone growth kinetics in
SiC-reinforced Ti-6AI-4V as studied by transmission electron microscopy. In
Recent Advances in Composites in the United States and Japan. ASTM STP 684,
ed. J.R. Vison & M. Taya, Am. Soc. Testing & Mats, Philadelphia, 1985,
pp. 585-99.
39. Choi, S. K., Chandrasekaran, M. & Brabers, M. J., Interaction between titanium
and SiC. J. Mat. Sci., 25 (1990) 1957-64.
40. Kieschke, R. R., Titanium reinforced with SiC monofilaments. PhD thesis,
University of Cambridge, 1990.
4t. Lloyd, D. J., The solidification microstructure of particulate reinforced
aluminium-SiC composites. Comp. Sci. & Tech., 35 (1989) 159-80.
42. Ishida, H., Proceedings of Third International Conference on Interfaces in
Composite Materials, Proc. ICCI-III. Elsevier, New York, 1990.

Interfacial mechanics in fibre-reinforced metals

55

43. Metcalfe, A. G. & Klein, M. G., Effect of the interface on longitudinal tensile
properties. In Interfaces in Metallic Matrix Composites: Composite Materials,
Vol. 1, ed. K. G. Krieder. Academic Press, New York, 1974, pp. 310-21.
44. Ochiai, S. & Marakami, Y., Tensile strength of composites with brittle reaction
zones at interfaces. J. Mat. Sci., 19 (1979) 831-40.
45. Ochiai, S., Urakawa, S., Ameyama, K. & Marakami, Y., Experiments on
fracture behaviour of single fibre-brittle zone model composites. Met. Trans.,
IIA (1980) 525-30.
46. Reeves, A. J., Dunlop, H. & Clyne, T. W., The effect of interfacial reaction layer
thickness on fracture of Ti-SiC particulate composites. Metall. Trans.
(submitted).
47. Nathan, M. &Ahearn, J. S., Interfacial reactions in Ti/SiC layered films with
and without thin diffusion barriers. Mat. Sci. & Eng., A126 (1990) 225-30.
48. Hoffmann, D. W., Film stress diagnostic in the sputter deposition of metals.
Proc. 7th ICVM, Iron & Steel Inst. Japan, Tokyo, Japan, 1982, pp. 145-57.
49. Tressler, R. E., Interfaces in oxide reinforced metals. In Interfaces in Metallic
Matrix Composites: Composite Materials, Vol. 1, ed. K. G. Krieder. Academic
Press, New York, 1974, pp. 286-301.
50. Kieschke, R. R., Somekh, R. E. & Clyne, T. W., Sputtered barrier coatings on
SiC fibres for use in reactive metallic matrices. Part l--Optimisation of barrier
structures. Acta Met., 39 (1991) 427-36.
51. Warwick, C. M. & Clyne, T. W., Development of a composite coaxial cylinder
stress analysis model and its application to SiC monofilament systems. J. Mat.
Sci. (in press).
52. Mehan, R. L., Jackson, M. R. & McConnell, M. D., The use ofYzO 3 coatings in
preventing solid-state Si-based ceramic/metal reaction. J. Mat. Sci., 18 (1983)
3195-205.
53. Warwick, C. M., Kieschke, R. R. & Clyne, T. W., Measurement and control of
the stress state in sputtered diffusion barrier coatings on monofilaments. In
Interfacial Phenomena in Composite Materials 1989, ed. F.R. Jones.
Butterworths, London, 1989, pp. 267-75.
54. Janaf Thermochemical Tables, US Department of Commerce, NSRDS-NBS,
Publication 37, 1971.
55. Berard, M. E & Wilder, D. R, Self-diffusion in polycrystalline yttrium oxide. J.
Appl. Phys., 34 (1963) 2318-27.
56. Berard, M. F. & Wilder, D. R., Cation self-diffusion in polycrystalline Y~O3 and
Er20 3. J. Amer. Ceram. Soc., 52 (1969) 85-91.
57. Kofstadt, P., Nonstoichiometry, Diffusion and Electrical Conductivity in Binary
Metal Oxides. Wiley-Interscience, New York, 1972, pp. 268-74.

Das könnte Ihnen auch gefallen