Sie sind auf Seite 1von 8

The eect of processing on the mechanical properties and interfacial

strength of aluminium/TiC MMCs

A.R. Kennedy*, S.M. Wyatt
Department of Materials Engineering and Materials Design, University of Nottingham, Nottingham NG7 2RD, UK
Received 28 May 1999; received in revised form 1 August 1999; accepted 9 August 1999
The microstructure and mechanical properties of aluminium/TiC MMCs made by powder processing (PM), by a ux-casting
process, and by melting PM material, have been compared. Particle clustering is more prevalent in cast than in PM composites, but
the grain-rening nature of TiC particles signicantly reduces the degree of clustering commonly observed in cast MMCs. Melting
PM material enables oxide lms to `trawl' the particles into large clusters. The stiness and ductility are similar for cast and the PM
composites but melting the PM material results in signicant reductions in strength and ductility. In all cases, composite ductility is
enhanced by extrusion through the removal of porosity and the break-up of particle clusters. Modulus measurements as a function
of plastic strain indicate that rates of damage accumulation are lowest, and hence interfacial bonding is strongest, in cast compo-
sites as a result of the attainment of intimate contact and strong chemical bonding between the two phases. #2000 Elsevier Science
Ltd. All rights reserved.
Keywords: A. Metal-matrix composites (MMCs); B. Mechanical properties; B. Interfacial strength; B. Microstructure; Processing
1. Introduction
The nature of the technique used to fabricate particle-
reinforced metal-matrix composites (MMCs) has been
shown to have a signicant eect on the product's
mechanical properties and hence the applications for
which it is suitable [14]. Composite products manu-
factured by power metallurgy (PM) are mainly targeted
for use as structural components. Cast composites, on
the other hand, are mainly considered as replacements
for conventional materials in non-critical, wear resistant
applications. The breakthrough of MMCs into large
volume markets, such as the automotive and aerospace
industries, is reliant upon achieving the mechanical
properties of PM materials at cast composite prices.
Composite mechanical property enhancement is not
only a function of the volume fraction, size, shape and
spatial distribution of the reinforcement, but is also
dependent upon how well the externally applied load is
transferred to the reinforcing phase. Stronger adhesion
at the particle/matrix interface improves load transfer,
increasing the yield strength and stiness increases and
delays the onset of particle/matrix de-cohesion [3,57].
The integrity of the reinforcement/matrix interface
can be quantied by measuring the decrease in modulus
with increasing plastic strain. This decrease is then
taken as a direct measure of the extent of progressive
damage to the reinforcement [8,9]. Damage takes the
form of either reinforcement cracking or de-cohesion at
the particle/matrix interface. At low strains, and for
particles less than 20 mm in diameter, particle cracking
seldom occurs [5,6,8,9]. Thus for similar composites
containing ne reinforcements, dierent rates of
damage accumulation can be linked to dierent eases
with which the particle/matrix interface de-coheres.
The mechanical properties of cast composites tend to
be poorer than those produced by PM [7,1015]. The
main problem associated with developing cast MMCs
with mechanical properties similar to their PM equiva-
lents, is not so much an issue of poor interfacial bond-
ing, but the formation of clusters of reinforcement
during solidication processing [1419].
Fewstudies have enabled the eect of primary processing
route on the mechanical properties and microstructure of
0266-3538/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PI I : S0266- 3538( 99) 00125- 6
Composites Science and Technology 60 (2000) 307314
* Corresponding author. Tel.: +1-44-115-951-3744; fax: +1-44-115-
E-mail address: (A.R. Kennedy).
similar composites to be determined. Most studies have
investigated aluminium/SiC systems but the particle
sizes and spatial distributions often dier between com-
posites. Reaction between SiC particles and most mol-
ten aluminium alloys means that the chemical
compositions of PM and cast composite matrices also
A ux-assisted casting method has been developed
[2022] which enables transition metal compounds such
as TiC to be incorporated into molten aluminium. The
role of the ux is to remove oxides from the melt and
reinforcement surfaces thus enabling intimate contact
and encouraging wetting [23]. Particle clustering in the
Al/TiC system is signicantly reduced and is attributed
to good wetting of the particles by the liquid metal and
the nucleation of solid aluminium on the reinforcing
particle surfaces [22,24]. The as-cast Al/TiC composite
particle distributions more closely resemble those
observed in powder-route processed composites,
enabling more meaningful comparisons between their
mechanical and physical properties.
2. Experimental
Commercial-purity (99.7 wt%) aluminium composites
containing 10 vol% of 10 mm diameter TiC particles
were prepared by three dierent manufacturing meth-
ods. Composites were manufactured using a ux-assis-
ted casting method (henceforth referred to as cast
composite), the details of which can be found elsewhere
[2022]. Composites were also manufactured from
powders, by simple mixing of the two components fol-
lowed by hot isostatic pressing (PM) and by melting and
casting the aforementioned powder-route material
(melted PM). It was hoped that by melting and casting
PM material, interfacial bonding might be improved.
Melts were poured in a cast iron mould to produce
cylindrical specimens 36 mm in diameter and 100 mm
long. Composite billets were extruded to 10 mm dia-
meter bar at 400

C, through shear faced dies, at a

reduction ratio of 13:1 and a speed of 1 mm s
. Den-
sities of the base material and the composites were
measured before and after extrusion using Archimedes'
principle [24]. Specimens were polished and viewed in an
optical microscope in order to observe the distribution
of particles in the matrix. Matrix grain structures were
revealed by anodising in Barker's reagent for 1 min at
24 V and viewing under polarised light conditions.
Mechanical testing was performed on both composite
and unreinforced, ingot and powder-derived materials,
in the as-manufactured and as-extruded conditions. The
tensile and extrusion axes were parallel. The modulus
was measured and averaged over three loading cycles using
a twin strain gauge method [24]. The 0.1% proof stress,
which was interpreted to be a consistently measurable yield
point, ultimate tensile stress and ductility were also
measured. The results in each case represent an average
of a minimum of six test samples.
The change in modulus with increasing plastic strain
was measured using a clip gauge arrangement. The
initial modulus was measured over three loading cycles,
and then averaged, before plastically straining the spe-
cimen by a set amount and re-measuring the modulus in
the same way. This testing operation was then repeated
until necking was observed. Results for three similar
tests were averaged.
3. Results
3.1. Microstructure and particle distribution
Fig. 1 compares particle distributions in cast and PM
composites and it can be seen that although there are
particle-free zones in both materials, there are few
highly clustered regions. There is, however, slightly less
clustering in the PM composite and in the cast material
areas of porosity can be observed. It should be noted
that large, blocky Al
Ti particles, indicative of reaction
between the matrix and the reinforcement, were not
Fig. 1. Optical micrographs showing the particle distribution in (a)
liquid and (b) PM processed CP Al10 vol% TiC composites.
308 A.R. Kennedy, S.M. Wyatt / Composites Science and Technology 60 (2000) 307314
observed in any of the microstructures. Fig. 2 shows the
grain structures for the cast and PM composites and
Table 1 presents the grain sizes for as-manufactured
composites and base materials, measured using a line
intercept method. The cast aluminium matrix is exten-
sively grain rened by the TiC particles and a signicant
proportion of the reinforcing particles are situated
within the metal grains. The grain size of the PM com-
posite is smaller still and is similar to the size of the
aluminium powder, between 20 and 50 mm. The rein-
forcing particles delineate the PM composite matrix
grains but the extent of clustering is small.
As a result of melting and casting PM material, very
large clusters and signicant particle-free regions are
formed. The TiC particles still act as nucleation sites for
solid aluminium and a rened microstructure is
obtained. The grain size is, however, over 30% larger
than that in the cast composite, indicating that a lower
percentage of the particles stimulate grain renement.
The particle distribution and microstructure can be seen
in Fig. 3, with clusters of particles visible both within
and around the metal grains.
Fig. 4 shows that extrusion improves the homogeneity
of the cast material and little dierence between the
spatial distribution of particles in the cast and PM
extruded products can be observed. Extrusion also
breaks down most of the large agglomerates in the mel-
ted PM composite, as illustrated in Fig. 5(a), but there
are still more clusters present than in either of the other
two extruded composites. For all composites, the grain
structure of the extruded product is ne and elongated,
a typical example of which is shown in Fig. 5(b), and
there are fewer signs of porosity when compared with
as-manufactured material.
Fig. 2. Optical micrographs showing the grain structure in (a) liquid
and (b) PM processed CP Al10 vol% TiC composites.
Table 1
Grain sizes (in mm) for materials in the as-manufactured condition
Al 10TiC
Al 10TiC
Al 10TiC
melted PM
36542 354 7710 326 10217
Fig. 3. Optical micrographs showing (a) the particle distribution and
(b) the grain structure in melted PM processed CP Al10 vol% TiC
A.R. Kennedy, S.M. Wyatt / Composites Science and Technology 60 (2000) 307314 309
3.2. Mechanical properties
Mechanical property data are shown in Table 2.
Composite stiness values in the as-manufactured con-
dition are similar, ranging from 86 to 88 GPa. This
corresponds to a 1.8 GPa approx. increase in modulus
per vol% of particles added. There is, however, a clear
dierence in the yield (0.1% proof) stress for composites
produced by dierent routes. The proof stress and ulti-
mate tensile stress values are highest in the powder-
route material. Melting and casting the PM composite
results in a signicant reduction in strength and ducti-
lity. The ductilities for the cast and PM composites are
high and not signicantly below those for their unrein-
forced equivalents.
Table 3 presents grain sizes, measured on sections
perpendicular to the tensile axis, and calculates the cor-
responding increases in composite yield stress expected
as a result of the reduction in grain size. This estimation
is based on the HallPetch equation, with a k value of
0.15 [6] and assumes that the friction stress does not
vary between materials compared. By subtracting the
strengthening contribution associated with grain-size
reduction from the actual increase, a rough indication
of the contribution from other strengthening mechan-
isms can be obtained.
In the unreinforced material, the dierence in yield
strength between the PM and cast aluminium is attrib-
uted to both a grain-size reduction and a strengthening
contribution (8.5 MPa) from ne oxide on the powder
surfaces. Since in the melted PM composite the con-
tribution from ne oxide is expected to be lost, through
its association with particle clusters, it was compared
with the cast composite. Although this analysis has several
simplications, it suggests that the contribution to the
yield stress from the reinforcement in the cast composite,
13.5 MPa, is greater than in the PMcomposite, 10.5 MPa,
which is in turn greater than that in the melted PM com-
posite, 8 MPa.
Extrusion increases the proof stress, ultimate tensile
stress and ductility for all the materials studied and
slightly increases the stiness along the extrusion direc-
tion. After extrusion, the mechanical properties of
extruded cast and PM composites are similar and the
ductility of the melted PM material recovers to a value
approaching that of the original PM material. Extrusion
also produces increases in the product density, as shown
Fig. 4. Optical micrographs showing the particle distribution in (a)
liquid and (b) PM processed CP Al10 vol% TiC composites after
Fig. 5. Optical micrographs showing (a) the particle distribution and
(b) the grain structure in melted PM processed CP Al10 vol% TiC
composites after extrusion.
310 A.R. Kennedy, S.M. Wyatt / Composites Science and Technology 60 (2000) 307314
in Table 2. Similar values for extruded composite den-
sities indicate little variation in particle volume fraction.
3.3. Modulus change as a function of strain
Figs. 6 and 7 show the modulus ratio, the modulus at
a given strain divided by the modulus at zero strain, as a
function of strain. Scatter values for the modulus ratio
are not shown but are typically 2%. In all cases the
modulus ratio, and hence the modulus, decreases with
increasing plastic strain. Both Figs. 6 and 7 indicate that
cast composites exhibit the least amount of damage at
any given strain.
Above strains of about 1% for as-manufactured and
about 2% for as-extruded composites, the decrease in
the modulus ratio, or increase in damage, is approxi-
mately linear. Table 4 presents values for the gradients
of the linear parts of these plots. It is clear from both
the table and the plots that the rates of decrease in the
modulus ratio are similar for PM and melted PM com-
posites, in both the as-manufactured and as-extruded
conditions, and that the damage accumulation rate is
signicantly lower for composites made by casting.
Extrusion, in all instances, decreases the damage-accu-
mulation rate.
Fig. 8 presents the modulus ratio as a function of true
stress for as-extruded materials. By extrapolating the
linear parts of these plots to zero damage, approximate
damage initiation stresses of 120, 110 and 100 MPa are
obtained for cast, PM and melted PM composites
respectively, re-arming that the cast composite is most
resistant to damage. Whilst the dierences in initiation
stress may be small, dierences in the modulus ratio are
signicant. For example, at 10% plastic strain a mod-
ulus ratio dierence of 0.1 is observed between the PM
and the cast-extruded composites, corresponding to a 9
GPa dierence in stiness.
4. Discussion
4.1. Microstructure and particle distribution
Although not quantied, it is clear that there is less
clustering in PM than cast composites. Since for the PM
composite, the matrix and reinforcement powder parti-
cle sizes are comparable, as long as mixing is thorough,
the generation of severe clusters will be avoided. Simple
mixing and isostatic compaction does not, however, enable
the particles to be enveloped within the matrix grains.
Grain renement of aluminium by TiC is expected
through close face-centred cubic-crystal-structure
matching [25,26]. The grain rening nature of TiC par-
ticles may reduce the degree of clustering commonly
observed in cast MMCs because solid metal nucleating
and growing from the TiC particle surfaces can result in
many of the particles being located within the matrix
Table 3
Comparison of predicted and actual yield stress variations caused by reductions in the grain size
Grain Size (mm) Predicted
increase (MPa) Actual
increase (MPa) Dierence (actualpredicted) (MPa)
Al cast 36542
Al 10TiC cast 7710 9.5 23 13.5
354 17.5 26 8.5
Al 10TiC PM
326 1.2 12 10.8
Al 10TiC melted PM
10217 7.0 15 8.0
Compared to Al cast.
Compared to Al PM.
Table 2
Mechanical properties for materials in the as-manufactured and as-extruded conditions
Material Modulus (GPa) 0.1% PS (MPa) UTS (MPa) Elongation (%) Density (kgm
Al cast 690.5 343 663 332 2.707
Al cast+ext 700.5 412 893 403 2.713
Al 10TiC cast 871.0 573 1094 242 2.915
Al 10TiC cast+ext 900.5 692 1235 302 2.929
Al PM 700.5 603 755 325 2.701
Al PM+ext 710.5 672 924 392 2.710
Al 10TiC PM 880.5 723 1204 253 2.920
Al 10TiC PM+ext 890.5 792 1294 322 2.927
Al 10TiC melted PM 861.5 494 886 83 2.912
Al 10TiC melted PM+ext 881.0 674 1177 223 2.932
A.R. Kennedy, S.M. Wyatt / Composites Science and Technology 60 (2000) 307314 311
Melting PM material has a marked eect on particle
clustering. It is likely that ne oxide lms present on the
aluminium powder particle surfaces are released and
mobile after melting and cause `trawling' of the particles
into clusters. It is expected that TiC particle surfaces are
not as clean as those introduced via the ux-assisted
casting method, and hence not wet so well by molten
aluminium, and hence have an increased tendency to
agglomerate. It should be noted that particle-melt
interactions in the melted PM system are not so dier-
ent that grain renement becomes impossible, although
the increased level of clustering seemingly results in a
larger grain size than that for the cast composite. Reac-
tion between TiC and aluminium has been reported [27]
but it is believed that in this instance the short proces-
sing times minimise particle/matrix reaction.
Extrusion did not produce severe banding of the
reinforcement. Instead, shear processes during extrusion
reduce porosity, facilitate the break-up of both small
and very large particle clusters and reduce the size of
particle-free regions. These observations are in keeping
with those of other researchers [5,10,14,28].
4.2. Mechanical properties
In this study, the matrix composition, reinforcement
volume fraction, shape and size, and in the case of
extruded composites, distribution, are near identical.
Thus, dierences in work hardening rate and damage
accumulation rate are more easily attributed to varying
strengths of bonding at the particle/matrix interface.
Dierences in values for the initial modulus were small
suggesting that at low strains, corresponding to stresses
below the yield point, the load transfer behaviour is
similar. Above stresses greater than about 100 MPa, the
interfacial bonding behaviour diers with manufactur-
ing method. Signicantly lower rates of damage accu-
mulation for composites made by ux-casting indicate
that they exhibit the strongest interfacial bonding.
Reinforcements in the cast composite produce greater
strengthening than those in PM and melted PM com-
posites indicating reduced load transfer and interfacial
bond strength in the same sequence. This conclusion is
consistent with that for damage accumulation rate
Whilst it is clear that hot isostatic pressing produces a
bond suciently strong for load transfer at low applied
stresses, it is apparent that interfacial bonding is stron-
ger in cast composites. Improved interfacial bonding is
attributed to the production of clean interfaces, through
the use of a ux, which enables good wetting by molten
aluminium. As a result of the metallic nature of the
bonding in transition metal carbides such as TiC [29], it
is thought that simply bringing clean TiC and alumi-
nium surfaces together is sucient to enable good
bonding. Clean, coherent interfaces with strong chemi-
cal bonding have been observed in in-situ processed Al/
Fig. 6. Decrease in modulus ratio as a function of plastic strain, for
composites in the as-manufactured condition.
Fig. 7. Decrease in modulus ratio as a function of plastic strain, for
composites in the as-extruded condition.
Table 4
Damage accumulation rates as a function of strain for composites
produced by dierent methods
As-manufactured As-extruded
Al 10TiC cast 1.6 0.7
Al 10TiC PM 2.5 1.7
Al 10TiC melted PM 2.4 1.8
Fig. 8. Decrease in modulus ratio as a function of true stress, for
composites in the as-extruded condition.
312 A.R. Kennedy, S.M. Wyatt / Composites Science and Technology 60 (2000) 307314
TiC composites [26] and it is expected that similar
interfaces to these are produced as a result of ux-casting.
The TiC particles in the PM composite are likely to be
in contact with an alumina layer on the aluminium
powder surfaces, which cannot be removed or by-passed
during hot isostatic pressing. It is this layer which pre-
vents intimate Al/TiC contact and inhibits strong
bonding. Simply melting PM composites is not sucient
to remove the oxide lm, achieve intimate contact and
produce a strong bond. The presence of oxide lms is
also likely to result in microporosity at the non-wetted
particle/matrix interfaces.
Small increases in the modulus after extrusion are
expected to be a result of alignment of reinforcement
with their major axes in the direction of extrusion,
causing an eective increase in the aspect ratio of the
reinforcing phase [5,28]. Reduction in porosity, as evi-
denced by density measurements, is also likely to
increase the modulus. Since the clustering, and the
number of sites for premature void formation, decrease
after extrusion [5,8,14], we expect damage initiation to
be deferred to higher macroscopic strains. The observed
shift in the onset of damage initiation was, however,
only small.
Shearing at the particle/matrix interface during
extrusion could be responsible for decreasing the
damage rate by generating clean aluminium and TiC
surfaces, much like those produced during friction
welding, and thus increasing the likelihood of forming
strong chemical bonds. Microporosity at the particle/
matrix interface may also be closed. The dierence
between the damage rates in cast and PM composites
was not signicantly reduced after extrusion. It is prob-
able that the extrusion ratio and speed were insucient
to produce sucient shearing to improve greatly the
interfacial bond strength.
5. Conclusions
Fewer particle clusters and particle-free regions are
present in PM Al/TiC composites than in cast ones.
Melting PM material enables oxide lms to become
mobile and seemingly results in `trawling' of the particles
into large clusters.
Mechanical properties are similar for cast and PM
composites, but melting PM material results in severe
particle clustering and signicant losses in both strength
and ductility. In all cases, composite ductility is
enhanced by extrusion through the removal of porosity
and the break-up of particle clusters.
Flux-cast Al/TiC composites exhibit the lowest rates of
interfacial damage with progressive plastic straining and
require higher stresses to initiate damage. This is inter-
preted as being caused by bonding between the reinforce-
ment and the matrix being strongest in these composites.
Enhanced interfacial bonding in cast Al/TiC compo-
sites is in turn attributed to the use of a ux, which
cleans the particle and matrix surfaces, enabling intimate
contact and the formation of strong chemical bonds.
The authors would like to thank Stanley Matthews
for his help with this work.
[1] Mortensen A. Interfacial phenomena in the solidication proces-
sing of metal matrix composites. Mater Sci Eng 1991;A135:111.
[2] Lloyd DJ, Lagace H, McLeod A, Morris PL. Microstructural
aspects of aluminium-silicon carbide particulate composites pro-
duced by a casting method. Mater Sci Eng 1989;A107:7380.
[3] Feest EA. Interfacial phenomena in metal-matrix composites.
Composites 1994;25(2):7586.
[4] Lloyd DJ, Morris PL, Nehme E. Some factors inuencing the
ductility and properties of particulate reinforced MMCs. In:
Masounave J, Dhingra A, editors. Fabrication of particulates
reinforced metal matrix composites. OH: ASM, Metals Park,
1990. p. 23244.
[5] Clyne TW. Metallic composite materials. In: Cahn RW, Haasen
YP, editors. Physical metallurgy, vol. 30. 4th ed. Elsevier, 1996. p.
[6] Humphreys FJ, Basu A, Djazeb MR. The microstructure and
strength of particulate metal-matrix composites. In: Hansen N, et
al., editors. Metal matrix composites processing, micro-
structure and properties, Proc. 12th Riso international sympo-
sium on materials science, Roskilde, Denmark, 1991. p. 5166.
[7] McKimpson MG, Scott TE. Processing and properties of metal
matrix composites containing discontinuous reinforcement.
Mater Sci Eng 1989;A107:93106.
[8] Lloyd DJ. Aspects of fracture in particulate reinforced metal
matrix composites. Acta Metall Mater 1991;39(1):5971.
[9] Lloyd DJ. Factors inuencing the tensile ductility of melt pro-
cessed particle reinforced aluminium alloys. In: Lewandowski J,
Hunt Jr W, editors. Intrinsic and extrinsic fracture mechanisms in
inorganic composite systems. Warrendale: TMS, 1995. p. 3947.
[10] Mortensen A. A review of the fracture toughness of particle
reinforced aluminum alloys. In: Masounave J, Dhingra A, edi-
tors. Fabrication of particulates reinforced metal matrix compo-
sites. OH: ASM, Metals Park, 1990. p. 21733.
[11] McDanels DL. Analysis of stressstrain, fracture and ductility
behavior of aluminum matrix composites containing dis-
continuous silicon carbide reinforcement. Metall Trans
[12] Whitehouse AF, Clyne TW. Critical stress criteria for interfacial
cavitation in MMCs. Acta Metall Mater 1995;43(5):210714.
[13] Murphy AM, Howard SJ, Clyne TW. The eect of particle clus-
tering on the deformation and failure of AlSi reinforced with
SiC particles: a quantitative study. Key Eng Mater 1997;127
[14] Murphy AM, Clyne TW. The eect of initial porosity and parti-
cle clustering on the tensile failure of cast particulate MMCs. In:
Street K, Poursartip A, editors. Proc. ICCM-10, vol. 2, 1995. p.
[15] Lloyd DJ. Factors inuencing the properties of particulate rein-
forced composites produced by molten metal mixing. In: Hansen
N, et al., editors. Metal matrix composites processing, micro-
A.R. Kennedy, S.M. Wyatt / Composites Science and Technology 60 (2000) 307314 313
structure and properties. Proc. 12th Riso international sympo-
sium on materials science, Roskilde, Denmark, 1991. p. 8199.
[16] Stefanescu DM, Dhindaw BK. Behaviour of insoluble particles at
the solid/liquid interface. In: Metals handbook, vol. 15. Metals
Park (OH): ASM, 1988. p. 14247.
[17] Rohatgi PK. Interfacial phenomena in metal matrix composites.
In: Dhingra AK, Fishman S, editors. Interfaces in MMCs. New
Orleans: AIME, 1986. p. 185202.
[18] Lloyd DJ. The solidication microstructure of particulate rein-
forced aluminium/SiC composites. Comp Sci Tech 1989;35:159
[19] Lagace H, Lloyd DJ. Microstructural analysis of AlSiC com-
posites. Canadian Metall Q 1989;28(2):14552.
[20] Kennedy AR, Karantzalis AE. The incorporation of ceramic
particles in molten Al and the relationship to contact angle data.
Mater Sci Eng 1999;A264:1229.
[21] Kennedy AR, McCartney DG, Wood JV. Homogeneous metal
matrix composites produced by a modied stir-casting technique.
In: Froes FH, Suryanarayana C, Ward-Close CM, editors.
Synthesis/processing of lightweight metallic materials. Warren-
dale: TMS, 1995. p. 26174.
[22] Kennedy AR, Karantzalis AE, Wyatt S. The microstructure and
mechanical properties of TiC and TiB
-reinforced cast metal
matrix composites. J Mater Sci 1999;34:93340.
[23] Karantzalis AE, Wyatt S, Kennedy AR. The mechanical proper-
ties of AlTiC metal matrix composites fabricated by a ux cast-
ing technique. Mater Sci Eng 1997;A237:2006.
[24] Clyne TW, Withers PJ. An introduction to metal matrix compo-
sites. Cambridge University Press, 1993. p. 399443.
[25] Cisse J, Bolling GF. Crystallographic orientations between alu-
minium grown from the melt and various compounds. J Cryst
Growth 1972;13:77781.
[26] Mitra R, Chiou WA, Fine ME, Weertman JR. Interfaces in as-
extruded XD AlTiC and Al/TiB
metal matrix composites. J
Mater Res 1993;8:238092.
[27] Mitra R, Weertman JR, Fine ME. Chemical reaction strength-
ening of Al/TiC metal matrix composites by isothermal heat
treatment at 913 K. J Mater Res 1993;8:23709.
[28] Dixon W, Lloyd DJ. Wrought Duralcan
particle reinforced
metal matrix composites. In: Rohatgi PK, editor. Processing,
properties and applications of cast metal matrix composites.
Warrendale (PA): TMS, 1996. p. 25970.
[29] Naidich JV. Wettability of solids by liquid metals. Prog Surface
Membrane Sci 1981;14:353484.
314 A.R. Kennedy, S.M. Wyatt / Composites Science and Technology 60 (2000) 307314