Lloyd1994 PDF

Particle reinforced aluminium and magnesium
matrix composites
D. J. Lloyd
application as selective reinforcement in the ring land

Particle reinforced metal matrix composites are area of diesel pistons," and whisker reinforcement is
now being produced commerically, and in this under development for aerospace applications.
paper the current status of these materials is Particle reinforced light metals, with their potential
reviewed. The different types of reinforcement as low cost, high modulus and strength, high wear
being used, together with the alternative resistance, and easily fabricated material, are just
processing methods, are discussed. Depending on reaching the commercial production stage. .Under-
the initial processing method, different factors
standing the factors that influence the physical and
have to be taken into consideration to produce a
, mechanical properties of these materials presents quite
high quality billet. With powder metallurgy
processing, the composition of the matrix and the a challenge, because they are sensitive to the type of
type of reinforcement are independent of one reinforcement, the mode of manufacture, and the
another. However, in molten metal processing they details of any fabrication processing of the composite
are intimately linked in terms of the different after initial manufacture. While there are still many
reactivities which occur between reinforcement areas which are poorly understood, the work of the
and matrix in the molten state. The factors past five years or so has identified many of the key
controlling the distribution of reinforcement are factors which have to be considered to achieve opti-
also dependent on the initial processing method. mum properties.
Secondary fabrication methods, such as extrusion
The objective of the present paper is to give an
and rolling, are essential in processing composites
overview of the current understanding of particle
produced by powder metallurgy, since they are
required to consolidate the composite fully. Other reinforced light metal composites, identify some of
methods, such as spray casting, molten metal the key factors which need to be controlled - which
infiltration, and molten metal mixing give an may be specific to the mode of manufacture or
essentially fully consolidated product directly, but processing - and consider some of the main problem
extrusion, etc., can improve the properties by areas which need to be overcome if these materials
modifying the reinforcement distribution. The are to reach their full commerical potential.
mechanical properties obtained in metal matrix In a field which typically has about 20 publications
composites are dependent on a wide range of a month, it is inevitable that a review such as this
factors, and the present understanding, and areas
has to be highly selective. It will only consider Al-
requiring further study, are discussed. The
and Mg-based matrices, even though Ti-based sys-
successful commercial production of metal matrix
composites will finally depend on their cost tems have considerable potential, particularly for high
effectiveness for different applications. This temperature applications. Most of the published work
requires optimum methods of processing, has considered AI-based composites, with their attrac-
machining, and recycling, and the routes being tions of low density, wide alloy range, heat treatment
developed to achieve this are considered. capability, and processing flexibility. Many of these
IMR/255 features are also exhibited by Mg-based systems, and
with its lower elastic modulus Mg often achieves a
1994 The Institute of Materials and ASM International. larger property improvement with reinforcement than
The author is with Alcan International Ltd, Kingston
Research and Development Centre, Kingston, ON, does Al, Also, many of the composite fabrication
Canada. processes are common to both AI- and Mg-based
systems.
Introduction The reinforcement

The development of metal matrix composites (MMCs) A recent overview of the short fibres and whiskers
has been one of the major innovations in materials available for metal matrix composites has been given
in the past 20 years. Many of the social and techno- by Stacey." Since most ceramics are available as
logical factors influencing this development have been particles, there is a wide range of potential reinforce-
reviewed by Kelly.1 The early work on composites ments for particle reinforced composites. Rohatgi and
considered continuous fibre reinforcement, and while co-workers have used mica," alumina, silicon carbide,
work in this area continues.' it was soon apparent and clay," zircon," and graphite, 10 and other
that the cost of continuous fibres, complex fabrication reinforcements, such as boron carbide!' and titanium
routes, and limited fabricability, would restrict their diboridc.F have also been examined. However, the
use to those applications requiring the ultimate in choice of reinforcement is not as arbitrary as this
performance. This led to the development of discon- list of composites might suggest, but is dictated by
tinuously reinforced composites, particularly short several factors:
staple Al, 03 fibre" and SiC whisker" reinforced com- 1. The application: if the composite is to be used
posites. Discontinuous fibres have found commercial in a structural application, the modulus, strength, and
International Materials Reviews 1994 Vol. 39 No.1

2 Lloyd Particle reinforced AI and Mg matrix composites
density of the composite will be important, which

requires a high modulus, low density reinforcement.
Particle shape may be important, since angular
particles can act as local stress raisers, reducing
ductility. If the composite is to be used in thermal
management applications, the coefficient of thermal
expansion and thermal conductivity are important.
The coefficient of thermal expansion is generally Cold Isostatic Compaction
important because it influences the strength of the
composite, as discussed in the section 'Strength' below.
2. The method of composite manufacture: as
considered in the next section, there are two generic
methods for composite manufacture, powder metal-
lurgy and methods involving molten metal. In the
case of powder metallurgy, the matrix alloy powder
is blended with particles of the reinforcement to
achieve a homogeneous mixture. To achieve this, the
sizes of the metal and ceramic powders need to be ~Degassing
carefully chosen so that agglomerates are not left after
blending, and carryover into the final product. The
appropriate size ratio will depend on the blending
process used, but in one case a SiC/AI particle size
ratio of 07: 1 gave a more uniform reinforcement
distribution than a ratio of O'3 : 1 (Ref. 13). In powder
~~
metallurgy processing, the brittle ceramic particles
are also susceptible to particle fracture, which is Evaluation
dependent on the particle aspect ratio and flaw den-
sity. Typically, the atomised aluminium powder 1 Powder metallurgy manufacturing route
particle size is in the range 20-40 urn, and reinforce-
ment particle sizes are 3-20 urn with aspect ratios
of < 5: 1. With these considerations in mind, the <, two
reinforcements receiving the most attention are SiC
For composites processed in the molten state, there
are different considerations. In some of these pro-
and Al2 3, and some of the properties of these two
reinforcements are given in Table 1.
cesses, the ceramic particles can spend considerable
time in contact with the molten alloy matrix, and this
can result in reaction between the twO.14 For example, Composite manufacture
SiC is thermodynamically unstable in most molten
While the details of the different manufacturing routes
Al alloys, reacting to form aluminium carbide, Al4 C3,
are often proprietary, the important features are quite
whereas it is stable in many molten Mg alloys. On
well established.
the other hand, Al2 03 is stable in most Mg-free Al
alloys, but unstable in Mg ailoys, reacting to form
spinel, Al2 MgO 4' Reaction of the reinforcement can Powder metallurgy
severely degrade the properties of the composite, so Because of the difficulty in wetting ceramic particles'
the reinforcement has to be chosen after considering with molten metal, the powder metallurgy route was
the matrix alloy, and the processing time and temper- the first method developed. The essential features of
ature. The reinforcement particle size is also important the powder route are shown in Fig. 1. The metal
because, while it is generally easier to incorporate powder is usually prealloyed atomised powder, in the
coarser particles into the melt, large particles are 20-40 urn size range, but it can also be a blend of
more susceptible to gravity settling and can result elemental powders, or rapidly solidified chopped
in a heavily segregated casting.P However, finer ribbon or flake, as in the Allied-Signal process.l" As
particles increase the viscosity of the melt, making noted previously, the blending step is important since
processing difficult. Most molten metal processes use this establishes the initial homogeneity of the com-
ceramic particles in the 10-20 urn size range. posite. Because the metal powder particles have a
3. Cost: a major reason for using particles is to hydrated oxide film, it is necessary to remove the
reduce the cost of the composite, so the reinforcement associated water molecules before consolidation if
has to be readily available in the quantities, size, and subsequent gas porosity is to be avoided. The vacuum
shape required at low cost, i.e, about US$5/kg. hot pressing consolidates the composite to over 95%
Table 1 Properties of SiC and AI203 "reinforcement

Elastic Coefficient of Specific Thermal
modulus, Density, thermal expansion, heat, conductivity, Poisson's
Particle GN m-z 9 cm-3 K-1 J kg-1 K-1 Wm-1 K-1 ratio
SiC 420-450 32 4'3x10-6 840 10-40, at 11 OOC 017

Alz03 380-450 396 70 x 10-6 1050 5-10, at 1000C 025

Lloyd Particle reinforced AI and Mg matrix composites 3
dense, and can be carried out below the solidus of

the alloy, or in the liquid-solid region. The kinetics
of densification are higher using liquid phase sintering,
but it has the disadvantage that reaction can more
readily occur between the reinforcement and the
liquid phase to form undesirable intermetallics at the
ceramic particle interface. Liquid phase sintering will
also degrade the microstructure of the rapidly solidi-
fied powder particles, and generate coarse eutectic
intermetallic phases in the melted regions. Completely
solid state consolidation enables advantage to be
taken of the supersaturated metastable alloy compos-
itions that can be obtained by the rapid solidification
atomisation and ribbon casting processes.
The final wrought product is obtained by extrusion,
with an extrusion ratio of about 20: 1 or higher. A
high extrusion ratio is required to disrupt the oxide 2 Microstructure of Duralcan 6061-
film between metal powder particles, allowing metal 20 vol.-%AI203 composite after extrusion
to metal contact and the development of a good bond
between the metal particles. A high extrusion ratio
also improves the distribution of reinforcement, since ation. The wetting angle is usually measured by the
the plastic flow associated with extrusion tends to sessile drop method, and is influenced by several
disperse any clusters of reinforcing particles. However, variables including the heat of formation, stoichi-
the extrusion ratio and temperature must be carefully ometry, valence electron concentration, interfacial
controlled to avoid reinforcement particle cracking, reactions, temperature, and time.i" In general, how-
and degradation of the matrix if rapidly solidified ever, molten aluminium does not wet most ceramic
powders are being used. particles at typical casting temperatures, i.e. < 800C,
The powder metallurgy route has several attract- and the molten metal methods attempt to improve
ive features: this wetting behaviour.
1. It allows essentially any alloy to be used as
the matrix.
Mixing methods
2. It also allows any type of reinforcement to be
used because reaction between the matrix and The early mixing method of Surappa and Rohatgi "
reinforcement can be minimised by using solid state introduced ceramic particles through the sides of a
processing. . vortex created in the melt with a mechanical impeller.
3. Non-equilibrium alloys can be used for the This method is helped by the addition to the melt of
matrix by using rapidly solidified material. This is surface active elements, such as magnesium to alumin-
ium melts,26,27 or metal coated particles." However,
particularly important where the composite is to be
used for high temperature applications, and rapidly the process is limited to coarse ceramic particles,
solidified alloys have much better elevated temper- > 50 um, and low volume fractions, < 10 vol._%.
ature strength than conventional alloys."? The major breakthrough in mixing processes came
4. High volume fractions of reinforcement are poss- with the development of the Duralcan (Duralcan
ible, thus maximising the modulus and minimising owned by Alcan Aluminum Corp.) process by Skibo
the coefficient of thermal expansion of the composite. and Schuster.'? This process is applicable to conven-
The powder route also has some major disadvan- tional aluminium alloys, using uncoated ceramic
tages: it involves handling large quantities of highly particles of about 10 urn and larger, and produces
reactive, potentially explosive powders and the manu- reinforcement levels of up to 25 vol._%. The typical
facturing route is relatively complex and limited in microstructure of a 6061/AI203/20p* Duralcan com-
the initial product forms it can produce. As a result, posite after casting and extrusion is shown in Fig. 2.
the product is expensive in comparison with conven- The Duralcan material is now being commercially
tional wrought aluminium alloys, and currently costs produced in batches of up to 6800 kg, and cast as pig
over US$100/kg, but this should decrease with future or direct chill (DC) ingot.
scale-up. Two other aluminium companies have announced
the development of a molten mixing method. Hydro
Aluminium AS have discussed composites which
Molten metal methods appears to be comparable with Duralcan material. 30
Early attempts to incorporate ceramic particles into Comalco has recently introduced Comral, which
metallic melts had limited success because most metals is 6061 reinforced with a spherical, mixed oxide,
do not wet ceramic particles, and this results in AI203-Si02, reinforcement." The extruded micro-
rejection of the particles from the melt.18,19 The basic structure of this composite is shown in Fig. 3.
thermodynamics associated with incorporating a The molten metal mixing method is attractive
single particle into a melt have been considered in because, in principle, it allows all the conventional
some detail,20-23 and demonstrate that the contact
angle between molten aluminium and the ceramic *The suggested Aluminum Association designation for composites:
particle must be less than 90 for successful incorpor- matrix/reinforcement/volume fraction and reinforcement type
(p refers to particle reinforcement) is used throughout.

30-.----------------
25
C\J
I
E
V) 20
z
>-
t::: 15
(j)
o
u
(j) 10
s
5
O-+---.----r-~~~~~~~~E....J
o 10 20 30 40 50 60 70 80 90 100
SHEAR RATE 5-1 I
3 Microstructure of Comalco 6061-20 vol.-%

spherical oxide after extrusion 4 Shear rate dependence of viscosity of
A356-15 vol.-%SiC
metal processing routes to be used, and hence mini-

mises the cost. At the present time, composite material evacuated packed bed in the bottom of a crucible is
is available from Duralcan for about US$6/kg. It covered with a melt, and the stirrer shears the interface
does, however, have some problems in terms of reac- between the particles and the melt, resulting in incor-
tivity between the reinforcement and the melt, and poration.P" Combining these two processes, by replac-
particle segregation effects, both of which are consid- ing the ceramic bed with the pressure infiltrated
ered in later sections. The volume fraction of reinforce- composite hardener, may provide a useful way of
ment is also limited in the mixing method because diluting the hardener."?
the viscosity of the melt increases with particle content The pressure infiltration route has the advantage
and becomes non-Newtonian. The viscosity of unre- that it is a means of producing composites with a
inforced aluminium is about 10-3Nsm-2,'but as high volume fraction of reinforcement, and since the
seen from Fig. 4, the viscosity of molten composites system is evacuated, the master composite has very
is much larger, and shear rate dependent.F The low porosity. With the additional dilution stage, the
rheology of molten composites is poorly understood, same factors come into playas with the direct mixing
but appears to be history dependent." and increases route, i.e. reactivity, viscosity, etc. No costs for this
with increasing volume fraction and decreasing process are presently available.
particle size. As a result, the power requirements
necessary for mixing limits the amount of reinforce- Pressu reless i nfi Itration
ment that can be used. A recent molten metal process is the Lanxide Corp.
Primex pressureless infiltration process.?' ,42 In this
Semisolid casting process a packed bed of ceramic powder is infiltrated
The microstructures developed by stirring a semisolid by an AI- Mg alloy, without any applied pressure, in
melt were investigated in the early 1970s.34 The a nitrogen atmosphere. The resulting composite,
partially solidified, non-dendritic microstructure which has a packed bed density of about 55 vol.- %,
developed has high viscosity which inhibits ceramic can then be diluted in the appropriate matrix alloy,
particle settling and floating, and can be used to if desired. Ceramic particles of SiC and A1203, with
retain particles in the melt. 35,36 This method has been particle sizes as fine as about 1 urn have been infil-
developed for magnesium alloys by the Dow Chemical trated in this way, and at infiltration rates of up to
CO.37 The process is restricted to longer freezing the order of centimetres per minute under some
range alloys and, other than this, has the same processing conditions. Processing details of the
limitations as the fully liquid mixing methods. Primex route are proprietary, but it would appear
to be a very competitive process for higher volume
Pressure infiltration fraction composites.
In this process, a packed bed of ceramic particles is
evacuated and then infiltrated by a pressurised melt Spray deposition
to form a 'master composite hardener' containing The spray deposition process for unreinforced alloys
about 50 vol-" of reinforcement. This hardener can was developed by Singer." and put into commercial
then be diluted by adding it to an unreinforced melt, use by Osprey Metals.44,45 It involves atomising a
and the reinforcement redispersed by submerged melt and collecting the semisolid droplets on a sub-
mixing.Pr" However, this dispersion is quite difficult strate. As a result, the process is a hybrid rapid
because of the' high viscosity of the master hardener, solidification process. The 'Spray Co-deposition Pro-
and it is difficult to obtain dispersion without incor- cess' is a variant of the basic process, where ceramic
poration of gas and oxide into the melt. Recently, a particles are introduced into the spray and code-
bottom mixing process has been suggested, where an posited with the alloy droplets/" The deposition rate
International Materials Reviews 1994 Vol. 39 NO.1

PVC Exterior 65G-r--------------------,

Claddll)Q
64
v
(/)~ 63
Pressure Relief ::>
Venls \ o
,~ ToAlr :5
a 62
:J
61
600-+--.......----,----r---.----r---,---r---,--'
ReInforced
Wall
o 0.5 2 3 4
TIME AT 675C, h
6 Decrease of melting point with time at 675C
for 7075-15 vol.-J'oSiC (after Ref. 57)
5 Spray codeposition system
used as a hardener with a pure aluminium matrix,

is 6-10 kg min - 1, and Alcan is presently developing which is then diluted in a melt of the desired matrix
this route commercially, producing 200 kg ingots."? alloy.
A diagrammatical representation of the A1can system An alternative in situ method for producing TiC
is shown in Fig. 5, where the 'solid deposit' is built particles is to inject a carbon-carrying gas into an
up to a billet for subsequent fabrication. The process AI- Ti melt at a sufficient temperature for the exother-
mic reaction to TiC to occur. 52
has the advantage that the contact time between the
melt and the reinforcing particles is brief, so reaction To date, there is little detailed information in the
between the two is limited and a wider range of open literature on the processing aspects of in situ
reinforcements are possible provided the as sprayed composites, so it is difficult to assess the cost of this
material. It has the attraction of producing particles
billets are not remelted. The initial billets are typically
95-98% dense, and require a secondary fabrication which should be inherently wetted by the matrix, and
step to achieve full density. As long as the alloy has therefore provide high interfacial strength. However,
a sufficient freezing range to achieve atomisation at the fine particle size is expected to produce highly
moderate superheats, any matrix composition can be viscous melts, which may make handling and
used, including the advanced aerospace alloys, such dilution difficult.
as the AI-Li 8090 alloy.r" The cost of composites There is, therefore, a range of composite production
produced by spray codeposition should be inter- routes available in varying degrees of commercial
mediate between powder processed composites, and development, and the appropriate choice will depend
material made by the mixing method. on the application and acceptable cost.
XD process Matrix-reinforcement reactivity

This is a rather different approach to composite A recurring theme in the composite literature is the
manufacture than the previous molten metal methods. reactivity between the matrix and the reinforcement
The XD process is a patented composite manufac- since it can have a significant effect on the interfacial
turing method developed by Martin Marietta Corp., strength,53,54 and hence the deformation and fracture
in which ceramic particles are produced in situ in a of the composite. However, reaction between the
melt.'" The process consists of adding to a solvent reinforcement and the matrix can also result in
metal, such as aluminium, compounds which will changes in the matrix alloy metallurgy which will, in
react exothermally to produce the required reinforcing turn, influence such basic properties as the melting
particles. A wide range of ceramic compounds can be point of the alloy'" and its strength.
formed by this process, 50 but the two which have The thermodynamic stability, in aluminium and
received most attention are TiB2 and TiC, which can magnesium alloys, of most of the reinforcements of
be formed by the following reactions interest has recently been reviewed.!" In silicon free
alloys, SiC is thermodynamically unstable above the
2B + Ti + Al ~ TiB2 + Al . (1)
melting point, reacting to form aluminium carbide,
and Al4 C3 (Ref. 56)
C + Ti + AI~TiC + Al . (2) 4AI + 3SiC~ Al4 C3 + 3Si . (3)

The particles are typically single crystal, and should with an increase in the silicon level of the matrix. As
have clean, unoxidised interfaces because they are the reaction proceeds the silicon level increases, and
formed in situ. By varying the process parameters, the melting point of the composite decreases with
such as reaction temperature, the reinforcement size time, as shown in Fig. 6 for an AI-Zn- Mg matrix
can be varied from ~ 02 to 10 urn, though the composite,7075jSiCj15p.57
material reported in the literature has particles in the The aluminium carbide reaction can be avoided by
025-15)lm range." The initial composite is then using high silicon alloys for the matrix, as shown in
14 2.0 +----+---+---+--t--+----+--+---+---I---t 0.10
o'!-
-t-
13 I
I
0.08 "3
12
~ 1.5 MgO
~ o
M
o C\I
0.06 C\I
-.g 11 <( ~
I ~
CJ)
1.0
eC\I
-ri 10 CJ) 0.04 ~
co 2 CJ)
L
tf) 9 .9 2
CJ)
0.02 .9
2 0.5 CJ)
8
2
L---'--4-----+--+--f----t--+---+--+--+ 0.00
7 920 940 960 980 1000 1020 1040 1060 1080 1100 1120
TEMPERATURE, K
~oo 700 800 8 Thermodynamic stability of AI-Mg oxides in

AI-Mg alloys (after Ref. 59)
TEMPERATURE,oC
7 Silicon level required at different tem-
peratures to prevent aluminium carbide shows the spinel crystals formed on the surface of
reaction (after Ref. 58)
Al2 03 particles after reaction above the melting point
in an AI- Mg alloy.
Obviously, from equation (5), Al203 will be
Fig. 7.58 So for processing routes involving long con-
unstable to some extent in magnesium alloys, but
tact times between the reinforcement and the melt,
other oxides, such as MgO and Y 203 will be stable.
high silicon aluminium alloys are p~efe~red. In p~~der
While these thermodynamic considerations show
processing using solid state consolidation, alum~n.tum
the tendency for reaction to occur, it is the kine~ics
carbide formation is not a factor because silicon
and the extent of reaction which are of practical
carbide is stable below the solidus; however, if liquid
importance. There are three stages of the reaction
phase sintering is involved there is th~ potential for
which may be attributed to: .
reaction. It is the molten metal processmg routes that
1. Nucleation of reaction product at preferred SItes
are particularly prone to reaction, since the p~rticle-
on the reinforcement.
liquid metal contact times can be long, partt.cularly
2. Continued dissolution of reinforcement in direct
when large scale casting processes, or remelting are
contact with liquid aluminium.
involved. The spray codeposition method is the least
3. Dissolution of the reinforcement separated from
susceptible, since in this case the contact time is only
the liquid aluminium by reaction product and, in
in the order of seconds.
some cases, a solute enriched region.
Other carbides, such as boron carbide and titanium
Gabryel and McLeod 59 have recently con~idered
carbide are also thermodynamically unstable in
the reactivity of Al203 in AI- Mg alloys USIng an
molten aluminium, but often react in a more complex
manner. 14 approach which may be applicable to. other reinfo~ce-
ments. Expressing the extent of reaction, a(t),at ume
Magnesium has no stable carbide, so ceramic car-
t as
bides are stable in pure magnesium. However, many
of the magnesium alloys of interest contain alloying Wo - w(t)
elements such as aluminium, which will form car- a(t)=--- . (6)
WO-We
bides, and in these magnesium alloys reaction may
occur if the contact times are sufficiently long. where is the initial Mg concentration at t = 0, w(t)

Aluminium oxide, Al2 3, is stable in pure alumin-
ium, but reacts with magnesium in AI- Mg alloys
Wo
3Mg+AI203~3MgO+2AI . . (4)
and
3Mg+4AI203~3MgAI204+2AI . (5)
The magnesium in equilibrium for equations (4) and
(5) has been calculated, and is shown in Fig. 8.59 At
high magnesium levels, and lower temperatures, MgO
may form, while the spinel will form down to very
low magnesium levels. It is not surprising, therefore,
that Al203 is not thermodynamically stable in most
aluminium alloys. Other oxides, such as MgO, are
expected to be stable. It should also be ,noted t?a~,

unlike SiC which is stable below the sohdus, this IS
not the case for Al2 3, and reaction can continue in
the solid state. So solid state processing may still 9 Spinel crystals on surface of extracted AI203
result in reinforcement reaction in this case. Figure 9 crystals

~ that using a mixed oxide, mullite, in the crystallised,

~ 1.0 948 K
+I 0973 K fully dense condition greatly reduces reactivity;" but
$ 0998 K no detailed kinetic data have been published.
A 1023K . Interface reaction can have several undesirable
Z"0.8 e 1048 K
Q 1073K effects. Al4 C3 dissolves in water, degrading the cor-
t- rosion behaviour of the composite, and its formation
<{ 0.6
0:: involves release of Si, modifying the matrix compos-
t-
ition. Al; Mg04 is not expected to affect the corrosion
~ 0.4
behaviour directly, but will modify matrix compos-
U
Z ition. Matrix composition modifications can in princi-
00.2 pal be allowed for, but the interface reaction products
U
CJ)
may also modify the mechanical properties of the
20.0 interface.l'v'" However, for those composites under-
o 50 100 150 200 250 300 going melt processing either in primary processing as
TIME, min in the melt mixing process, or in secondary processing
10 Experimental results for variation in Mg such as shape casting, it is the effect that reaction has
concentration in solution after different on the viscosity of the melt, and hence on casting
reaction times compared with predictions of fluidity, that is particularly important.
kinetic model (continuous lines) (after Casting fluidity is usually assessed by a test, such
Ref. 59)
as the spiral test, which measures the distance the
melt will flow before solidification occurs. So the test
involves both rheological factors, such as viscosity,
the Mg concentration at time t, and We the equilibrium
and solidification factors, such as latent heat, surface
Mg concentration, and applying Mampel's model
tension, superheat, cooling rate, and the freezing range
for deceleratory nucleation caused by overlapping
of the alloy. For an unreinforced metal, the distance
nucleations.?" they obtain
a melt will flow along a channel before solidification,
In(I - a) = C - 4nkNorZt . . . . . . . (7) Lis given by63
where C is a constant, k a rate constant, No the initial . . . (9)
surface concentration of nucleation sites, and r the
particle radius. For fused Al, 03 they found that the where
weight per cent of magnesium remaining in solid C = heat capacity
solution after reacting for a time t (s), at a temperature d = diameter of channel
T (K), is given by h = heat transfer coefficient
H = latent heat
w = 003 + (wo - 0'03) Ps = density of solid, taken to be equal to density
x exp{50[exp(-I03 OOO/RT)](to - t)} (8) of liquid
Tm = melt temperature
where the equilibrium concentration of magnesium To = mould temperature
Wo, is taken as 003 wt_%. Figure 10 shows that the I1T= melt superheat
model fits the data well within the temperature-time v = velocity of liquid.
regimes where MgO does not form. The results dem-
onstrate the importance of minimising the processing For composites several terms will be modified. The
temperature, and in the case of the spinel reaction, density, Ps, will be replaced by the density of the
keeping the magnesium content of the alloy as low composite, Pc
as possible, commensurate with achieving the desired Pc = Pm(1- J!;,) + Pp J!;, . . . . . . . . (10)
strength.
The kinetics of particle-matrix reaction can be where
influenced by several methods. Undesirable Al4 C3 Pm = density of matrix
can be a voided by appropriate choice of the silicon Pp = density of particle
content of the alloy, which, while restricting the choice J!;, = volume fraction of particle.
of matrix alloy, is the most convenient approach for
those SiC reinforced composites involved in molten The latent heat involved in solidification will be
metal processing. Another approach is to oxidise the reduced because the particles are not involved in the
surface of the SiC, forming an outer layer of SiOz. In solidification process; the effective latent heat, He' is
this case the early stages of the reaction involve given by
reducing the SiOz, rather than dissolving the SiC. He = H(I- Vp). (11)
The reaction product will depend on the matrix alloy,
being Alz 03 for pure aluminium, and MgO and Similarly, the effective specific heat of the composite,
AlzMg04 for Mg-containing alloy.?' This is a not c.. is given by
very satisfactory approach, since there is still an
Ce = Cm(1 - ~) + Cp Wp (12)
interface reaction product, and magnesium is lost
from the matrix, reducing the age hardening where
response." For Al203 reinforcement, the spinel reac- Cm = specific heat of matrix
tion can be minimised by using a low Mg-content Cp = specific heat of particle
matrix alloy. There is also some recent information ~ = weight fraction of particle.
70 ~------------.., 70
A
60-
E sa-
u
o

I + Eu 50-
60-
,
x
+
..
I
t-
t = :r:
,
t-
<9 40-
Z
A <9 40-
Z
W W
--J
--J

30-
--J
--J

30-
+

a: a: i
a::
(/)
20- a:: 20-
(/)
10- 10-
1
0
(a)
I I I I a
(b)
I , I
.
...
50 100 150 200 250 300 50 100 200
150 250 300
HOLDING TIME, min HOLDING TIME, min
a 750C; b 800C
11 Spiral fluidities of A356-15 vol.-%Sie after different holding times at different temperatures
Substituting the appropriate values for the composite 2. The viscosity increases with increasing volume
into equation (9) will result in a decrease in the fraction of particle.
expected fluidity length with increasing particle 3. The viscosity increases with decreasing particle
content. size.
However, of more importance is the melt velocity 4. The viscosity is dependent on the history of the
term, v, under the applied metallostatic head. As seen melt, in terms of temperature, time, and shear rate.
from Fig. 4, the viscosity increases with increasing 5. The viscosity increases with increasing reaction
particle level, and this will reduce the melt velocity. product at the interface.
For non-metallic liquids containing spherical The most extensive rheological study of unreacted
particles, the viscosity is given by64 composites has been carried out by Moon.i" who
found that they are also thixotropic.
n; = nm(1 + 2'5Vp + 10'05V~) . (13)
In terms of equation (13), interface reaction will
where often result in an increase in the volume fraction of
n; = viscosity of composite solid in the melt. For example, from the stoichiometry
nm = viscosity of unreinforced matrix. of the SiC reaction to Al4 C3, and the lower density
of Al4 C3 to SiC (2'36 cf. 32 g ern - 3), 1 g of SiC
The non-Newtonian behaviour of molten composites
reacted to Al4 C3 will result in a particle volume
makes comparison with equation (13) difficult, but it
increase of 60%. In addition, as pointed out by
is expected to be of the right order at very high shear
Surappa and Rohatgi,66 the viscosity may be depen-
rates, when Newtonian behaviour is approached. At
den t on particle surface area, and this will increase
low shear rates, or when reaction occurs between the
with interface reaction because the reaction products
reinforcement and the melt, it greatly underestimates
tend to be in the form of fine crystals.!" The non-
the viscosity. If extensive reaction occurs the viscosity
Newtonian and thixotropic nature of composite melts
can, in effect, increase to infinity, and the melt will
indicate that the composite melts have a structure
not flow into the mould at all.As a result, rheological
associated with them, presumably reflecting particle
factors dominate the casting fluidity under these
clustering in the melt. Particle clusters could occlude
circumstances. This effect is demonstrated in Fig. 11,
liquid within the clusters, effectively raising the solid
where the spiral fluidities after different holding times fraction of the melt. 67
at 750 and 800C are shown. For 750C the spiral
It is apparent that reinforcement reactivity can
fluidity remains about constant with holding time,
influence the behaviour of the composite from a wide
because the extent of aluminium carbide formation is
range of viewpoints, and needs to be considered both
limited at this temperature in a 7 wt-%Si alloy.
in terms of composite processing and composite use.
However at 800C, the amount of aluminium carbide
increases rapidly with time resulting in a marked
decrease in fluidity until, after 250 min, the composite
will not flow into the mould. Microstructures
As noted previously, the rheological behaviour of The most important aspect of the microstructure is
composite melts is poorly understood, but a few the distribution of the reinforcing particles, and this
general statements can be made: depends on the processing and fabrication routes
1. The viscosity is non-Newtonian, decreasing with involved. However, particles can modify other aspects
increasing shear rate. of the matrix microstructure.

Reinforcement distribution
In powder processed material, the reinforcement dis-
tribution will depend on the blending and consoli-
dation procedures, as well as the relative size of the
matrix and reinforcing particles. If the matrix powder
is large relative to the reinforcement, the reinforcing
particles will agglomerate in the intersticies of the
coarse particles, and be very inhomogeneously distrib-
uted in the final product. With the blending and
consolidation route used in one study.t" an AI/SiC
particle size ratio of 07: 1 gave a more uniform
distribution than a 024: 1 ratio. Any secondary pro-
cessing will also tend to homogenise the particle
distribution.?" (It should be noted that quantitative
determination of particle inhomogeneity and clus-
tering, considering the variable shape and size of
reinforcing particles, is not a trivial task. Initial
approaches have used the Dirichlet tesselation
method. 69. 70)
In composites processed by molten metal mixing
methods, the situation is somewhat more complicated
because the reinforcement distribution is influenced
'by several factors:
1. Distribution in the liquid as a result of the
mixing.
2. Distribution in the liquid after mixing but before
solidification.
3. Redistribution as a result of solidification.
The distribution during mixing will obviously
depend on the mixing process used, and it is essential
to produce as uniform a distribution as possible
without any gas entrapment, since any gas bubbles a slow solidification rate investment casting; b high solidification
will be lined with reinforcing particles. After mixing rate pressure die casting
and before solidification, the particles will segregate 12 Influence of solidification conditions on
due to gravity.P With the relatively high volume particle distribution
fraction of particles and a range of particle size, the
settling will be hindered 71
Jt:: = Vo(l- C)P . (14) generally applicable to planar front solidification of
single phase systems, while in reality it is dendritic
where solidification of multiphase systems which is relevant.
C = particle concentration In these commercial systems the experiments indicate
d = particle diameter that, whenever particles are captured by a solid/liquid
D = container diameter interface, the particles act as heterogeneous nuclei for
p = 465 + 195d/D for R, < 02 the solids." Because this does not generally occur,
=(435+ 175d/D)R;0,03 for 02<Re< I the particle distribution is influenced by the solidifi-
R; = Reynolds number cation rate, as shown in Fig. 12. Secondary fabrication
Jt:: = particle velocity processing, such as extrusion or rolling, can homo-
Vo = Stokes velocity. genise the structure to some extent, as seen from the
previous extrusion microstructures, but minimising
The settling rate will be a function of the particle
reinforcement inhomogeneity during initial processing
density and size, and there is also the possibility that
is important for achieving optimum properties.
particle shape will playa role."? Particles of different
Figure 13 shows how the tensile elongation increases
size and shape will settle at different rates producing
in (AI-Si) A356/SiC/15p with increasing degrees of
agglomeration.
extrusion.
The third factor which influences reinforcement
At present, quantification of particle inhomogeneity
distribution is the solidification process itself.
remains a problem, but recently some authors have
Reinforcing particles do not generally nucleate the begun to develop methods for addressing it.13, 76, 77
primary solidifying phase, 56 though solidification
nucleation may occur in some hypereutectic sys-
tems.P If solidification nucleation does not occur the Grain structure
reinforcing particles are rejected at the solid/liquid Wrought alloy composites are solution treated and
interface, and segregate to the interdendritic regions aged after fabrication, and recrystallisation will usu-
which solidify last. Particle entrapment or rejection ally occur during this heat treatment. Since particles
has been extensively studied, and there are several of diameter larger than about 1 urn will develop an
recent reviews. 22. 74, 75 The entrapment models are associated deformation zone sufficient to generate
15 strain rate sensitivity m in the strain rate equation

14 (J = kern, is required for superplasticity, and m-values
13 in the range 0'4-0'6, which are comparable with those
12 of superplastic unreinforced alloys, were obtained. An
~11 interesting feature of superplasticity in MMCs is that
z'10
e the high m-values are often obtained at high strain
0 9
rates. In conventional superplastic alloys the highest
~ 8
m-value, and optimum superplastic behaviour, is usu-
(9 7
Z ally obtained at low strain rate, about I x 10-4 to
0
.--J
6
5 X 10-3 S-1. However, in MMCs the largest m-
5 e T4-5.7 em DIA. BILLET
W
T4-19.5 em DIA. BILLET values, and greatest elongations are obtained at
4 T4-PM CAST TENSILE
b.
greater than 10- 2 S -1 (Ref. 82) and even as high as
3
b. 10 and 100 s - 1 in mechanically alloyed composites."
2
1
In all cases the elongation is limited by cavitation,
0
and particles increase the extent of cavitation by
10 100 1000 limiting grain boundary sliding and providing stress
EXTRUSION RATIO concentrations at the interface. (There is also the
13 Variation in tensile elongation with extrusion possibility of pre-existing cavities at the interface.P")
ratio in A356-15 vol.-%Sie, T4 condition The extent of cavitation increases with increasing
volume fraction of reinforcement, but it can be
reduced by using a superimposed hydrostatic pre-
recrystallised nuclei;" the reinforcing particles should sure."! Using as fine a reinforcement as possible
produce a high density of nuclei. However, if the should also help to decrease cavitation, and this is in
particles are very closely spaced, the subgrain growth agreement with the recent work on 2124/Si3N4/20p
necessary for recrystallisation nucleation is impeded where the particles were less than I urn in size and
and recrystallisation may not occur. Recrystallisation elongations of 850% were obtained at a strain rate
is impeded when t;,/ d > 01 urn, whereas most com- of 4 x 10 - 2 S- 1 and 515C, in as extruded material.
mercial composites have t;,/d less than this value, The yield strength is dependent on grain size
where t;, is the volume fraction of particles and d the through the Hall- Petch relationship, and the finer
particle diameter. grain size present in composites could contribute
The grain size of the recrystallised composite can considerably to the strength. Aluminium alloys have
be estimated by assuming that each reinforcing a low Hall- Petch slope, and grain sizes of 10 urn and
particle of diameter d acts as a nucleus for a spherical less will be required to influence the strength signifi-
grain D, which is given by 79 cantly, but in reinforced wrought Mg alloys grain size
strengthening could be very significant, because the
D = d[(1 - t;,)/t;,J 1/3 (15)
Hall- Petch slope is large.I"
Any subsequent grain growth will be limited by the
Zener pinning of the particles on the grain boundaries,
giving a limiting grain size of 2d/3 t;,. For 20 vol-" Aging effects
of 10 urn particles equation (15) gives D = '" 15 urn, Another aspect of the microstructure which may be
and the limited grain growth gives D = 33 urn, modified in composites is that of precipitation. Many
Recrystallisation studies on Duralcan (AI-Cu) studies of the age hardening response of metal matrix
2014/AI203/20p demonstrated that the Al203 composites have been reported87-93 and, in general,
particles stimulate the nucleation, accelerating recrys- an enhancement of the aging kinetics has been
tallisation, and decreasing the recrystallisation tem- observed. However, some of the studies demonstrate
perature.P'' The resulting equiaxed grain size was that at low temperatures the aging kinetics are slowed
",15 urn, and after holding the composite for 150 h down, or unaffected by the reinforcement,89,91,94 The
at 500C the grains had only grown to 17 urn. These more rapid aging kinetics were initially attributed to
results show that equation (15) is appropriate for easy precipitation nucleation on dislocations punched
predicting both the recrystallised and limiting grain out from the reinforcement particle interface as a
size, and the high volume fraction of reinforcing result of the coefficient of thermal expansion mismatch
particles is very effective in stabilising the grain size. on quenching from the solution treatment temper-
In powder processed composites reasonable agree- ature. Enhanced precipitation in the neighbourhood
ment was obtained with the theory except for very of reinforcement particles undoubtedly occurs for
coarse particles (D = 40 urn), where the grain size was precipitates, such as 0' and S', which are susceptible
much finer than expected because of multiple to dislocation nucleation." but there are many other
nucleation occurring at each particle. 79 factors which may influence aging. In material pro-
Since the reinforcing particles are very effective in cessed by powder metallurgy, there are also fine oxide
nucleating new grains, and stabilising the resulting particles in the matrix which are known from work
fine grain size, MMCs are potentially superplastic on SAP-type alloys to influence the aging kinetics."
materials. Tensile elongations of 300-8300/0 have been The presence of fine oxide particles, together with a
obtained in powder processed PM-64/SiC/20p,81 fine grain size is likely to reduce the vacancy concen-
6061/Si3N4/20p,82 and (AI-Cu) 2124/Si3N4/20p,83 tration by providing a high vacancy sink density, as
spray processed 2014/SiCj15p and 7475/SiC/15p,84 do the matrix/reinforcement interfaces. Vacancies may
and mechanically alloyed IN9021/SiC/15p.85 A high also be swept up by moving dislocations associated
130,....,----------------,
120
(\J
IE
zllO
o
~ 100
-.J
::J s = 1.5h
~
o
o 90 /./
/./:,./
2 /..
/. /
S = 1.0
u
~ 80
V>

Ljj 70
- -- Rule of Mixtures
60 -- Equivalent Inclusion
--- Tsai-Halpin Equation
14 Precipitation at SiC interface in (AI-Zn-Mg)
7091 alloy (courtesy of J. J. Lewandowski) 50
+------,r-----r----.---~---1
o 5 10 15 20 25
VOLUME, 0/0
with the plastic relaxation of misfit stresses generated 15 Variation of elastic modulus with volume
fraction of SiC particles: 5 is particle aspect
on quenc himg. 91 Ch anges In . the vacancy concen-
ratio
tration will particularly affect G P zone formation and
the low temperature aging response, consistent with
many of the differential scanning calorimetry stud-
ies.90,91,93,94 Solute segregation associated with result from the presence of thermal residual stresses
interfaces may also be present .in the composite,"? caused by differences in the coefficient of thermal
and the solute distribution may also be affected by expansion between the matrix and the ceramic
any residual stresses present in the matrix. The higher particles. In the case of SiC and Al203 particle
dislocation density in composites could also increase reinforced AI, the matrix is in tension. This means
the solute diffusivity, and enhanced diffusivity of Mg that when the composite is loaded, plastic flow occurs
atoms has been suggested as the reason for the higher earlier in tension than in compression, and the total
growth rate of {3' precipitates in 6061-A12 03 compos- strain will consist of both elastic and plastic com-
ites.?" At higher aging temperatures, or longer times, ponents. The situation is further complicated by
precipitation may occur at the particle interface in inhomogeneity in the reinforcement distribution
some alloys. An example of this in a 7091 alloy matrix which can also result in local plasticity. Comparison
is shown in Fig. 14. with theoretical expectations is also somewhat diffi-
There is very little information in the literature on cult due to uncertainty in the appropriate value for
the aging kinetics of Mg matrix composites, but work the modulus of the particle reinforcement.
on SiC reinforced Mg-6Zn indicated comparable Figure 15 shows the increase in Young's modulus
aging behaviour to that of unreinforced material. 99 with volume fraction of reinforcement for a variety of
From all these considerations it is clear that the AI-SiC composites (these values were taken from the
aging response will depend on a range of factors commercial literature for wrought 2000, 6000, and
including t~e particular matrix, the processing history: 7000 alloys). The rule of mixtures expression
and the agmg temperature. However, it should also e, = VpEp + VmEm (16)
be appreciated that in spite of any modification of
agi.ng kinetics in the composites, the peak aged prop- ~here s; Em' s, are ~he elastic moduli of the compos-
ertres are usually obtained within the normal commer- ite, matrix, and p~rtIcle, respectively? and Vm and ~
cial aging practice for unreinforced alloys, particularly the ~olume fractions of the matrix and particle,
for melt processed composites. considerably overestimates the elastic modulus.
The rule of mixtures expression is most appropriate
for continuous reinforcement and it has been modified
Mechanical properties for discontinuous reinforcement in the Halpin- Tsai
equation 101
Elastic modulus
The on~ mechanical property which is always signifi- E = Em(J + 2sqJt;,)
cantly Increased by the addition of reinforcement is . . (17)
e l-qJt;,
the elastic modulus. The quantitative value of the
elastic modulus is somewhat dependent on the where
method of measurement, with dynamic measuring (Ep/Em - 1)
methods tending to give larger values than static q = (Ep/Em) + 2s . . . . . (18)
measurements obtained from the elastic portion of
the tensile stress-strain curve. Static values may also and s is the particle aspect ratio. As seen from
depend on whether the measurements are made in the figure, the Halpin- Tsai equation gives a good
tension or compression.l''? Most of these difficulties representation of the results.

400.------------------------,
100
C\I
IE A356 C\I 350
A356/SiC120p
Z 90- IE 300
o z

(/)'
:3 80-

Z.250
(/)
(/)
::::> W200
0
0
0::
l-
(/) 150
2: 70- W
(/) ~ 100 :
~ I-
360-
50
0
>- O-+---+----+--+---l'-+---+----+--+---l'-+---+----+--+--lf---l
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14
50 -t----.------.-----.----.,.----.II'I---j
I I
I I I TRUE STRAIN
o 100 200 300 400 500 600
TEMPERATURE,oC 17 Stress-strain curves for extruded A356 and
A356/SiC/15p with different particle sizes
16 Temperature dependence of elastic modulus and tempers
in A356 and A356/SiC/20p
reinforcement with an aspect ratio below that necess-

The elastic modulus can also be calculated using ary for full reinforcement loading, gives the composite
the Eshelby equivalent inclusion method.l'" and this strength as109
approach is also in good agreement with the data.
The dominant factor in controlling the elastic . (19)
modulus is the volume fraction of reinforcement, and
it is relatively insensitive to the particle distribution, where (J m is the yield stress of the matrix, and s the
while variations in the type and shape of the reinforce- aspect ratio.
ment can be accounted for by the different expressions. For the aspect ratio typically used in particle
The improvement in Young's modulus is retained MMCs, which are in the 1-5: 1 range, equation (19)
at higher temperatures, as shown in Fig. 16, where underestimates the strength, but Nardone and
results for unreinforced matrix and composite are Prewo '?" have suggested that better agreement is
shown up to 500C. The temperature dependence of obtained if the equation is modified to allow for end
the composite elastic modulus reflects that of the loading effects. The difficulty with this continuum
unreinforced matrix. The higher elastic modulus at approach is that it ignores the influence of particles
temperature is important for higher temperature on the micromechanics of deformation, such as the
applications. very high work hardening at low strains, and modifi-
cations in microstructure, such as grain size and
Strength dislocation density.
The first extensive study of the strength of discontin- In the micromechanics approach, the possible
uously reinforced Al alloys was carried out by strengthening mechanisms are:
Mcfranels.l'" who investigated SiC whisker and 1. Orowan strengthening.
particle reinforcement in several different alloy 2. Grain and substructure strengthening.
matrices. He reported up to a 60% increase in yield 3. Quench hardening resulting from the dislo-
and ultimate tensile strengths, depending on the cations generated to accommodate the difference in
volume fraction of reinforcement, the type of alloy, coefficient of thermal expansion between the reinforc-
and the matrix alloy temper. Subsequent work has ing particles and the matrix.
generally confirmed these findings, but the reported 4. Work hardening, due to the strain misfit between
experimental results show an extremely large degree the elastic reinforcing particles and the plastic matrix.
of scatter, presumably reflecting the quality of the The extent to which the different mechanisms oper-
material and differences in processing. When consider- ate will depend on the microstructure and processing
ing the yield stress there is a more fundamental of the particular composite. In powder processed
difficulty, as pointed out by Humphreys.l'" The yield composites, the grain size can be extremely small and
strength is usually quoted as the 0.2% proof stress, significantly contribute to the strength, whereas in
and since composites work harden extremely rapidly melt processed MMCs the grain size more closely
at low strains, this may not be equivalent to a approaches unreinforced alloys. At typical volume
conventional yield stress. However, composites are fractions, Orowan strengthening is not a major factor
now commercially available which will meet a with the 5 urn and larger particles usually used, but
strength specification, and Table 2 lists the properties particles of this size can result in quench hardening
of some of the composites now available from different and enhanced work hardening because of elastic misfit
producers. For comparison some typical properties back stress hardening.111-113 From these considera-
of unreinforced alloys are given in Table 3. tions, the strength of particle reinforced composites
The strengthening mechanisms which may operate is most strongly dependent on the volume fraction of
in particle reinforced MMCs have been considered reinforcement with a somewhat weaker dependence
in several recent publications, and the behaviour on particle size. lOS Normal matrix strengthening due
has also been extensively modelled mathemat- to solution and precipitation hardening will give
ically.77,104-108 The continuum shear lag models for additional strength to the composite.
Table 2 Typical properties of some commercially available metal matrix composites

Elastic
YS,t UTS, Elongation, modulus,
Composite* MN m-2 MN m-2 % GN m-2 Supplier
Wrought
AI-M92Si
6061 /AI203/1 Op (T6) 296 338 75 81 Duralcan, Alcan
6061/AI203/15p (T6) 317 359 54 87 Duralcan, Alcan
6061/AI203/20p (T6) 305 330 34 85 Comral 85, Comalco
6061 /SiC/15p (T6) 342 364 32 91 Cospray, Alcan
6061 /SiC/15p (T4) 405 460 70 98 DWAt
6061/SiC/20p (T4) 420 500 50 105 DWAt
6061/SiC/25p (T4) 430 515 40 115 DWAt
AI-Cu
2014/AI203/15p (T6) 476 503 23 92 Duralcan. Alcan
2014/ AI2 03 /20p (T6) 483 503 10 101 Duralcan, Alcan
2014/SiC/15p (T6) 466 493 20 100 Cospray, Alcan
2124/SiC/178p (T4) 400 610 5-7 100 BP~
2124/SiC/25p (T4) 490 630 2-4 116 BP~
2124/SiC/20p (T4) 405 560 7 105 DWAt
AI-Zn-Mg
7090/SiC/20p (T6) 665 735 2 105 DWAt
AI-Li
8090/SiC/17p (T4) 310 460 4-7 103 BP~
8090/SiC/17p (T6) 450 540 3-4 103 BP~
Cast
AI-Cu
201/TiC/20p (T7) 420 20 105 XD, Martin Marietta
AI-Si
356/SiC/10p (T61) 287 308 06 82 Duralcan, Alcan
356/SiC/1 5p (T61 ) 329 336 03 91 Duralcan, Alcan
356/SiC/20p (T61) 336 357 04 98 Duralcan, Alcan
380/SiC/10p (F) 245 332 10 95 Duralcan, Alcan
380/SiC/20p (F) 308 356 04 114 Duralcan, Alcan
Mg-AI-Zn
AZ91/SiC/94p 191 236 2 475 Dow
AZ91 /SiC/151 p 208 236 1 54 Dow
AZ61/SiC/20p 260 328 25 80 Dow
* Composite designation: matrix/reinforcement/volume fraction of particles.

t 02% offset yield stress.
t Composite Specialities lnc., Chatsworth, CA.
~I British Petroleum.
Figure 17 shows the stress-strain curves for A356 SiC particle size. At a 0'2% offset strain the composite
and A356/SiC/15p, for two SiC particles sizes and work hardening rate in the T4 condition is around
two tempers, all materials being extruded bar. The 50 GN m -2, which is greater than the shear modulus
apparent work hardening rate at low strains is higher of the matrix. Therefore, the higher initial work
in the composites, and also increases with decreasing hardening rate may, to some extent, reflect differences
in the elastic-plastic transition between composites
and unreinforced material. This is not too surprising,
Table 3 Typical unreinforced alloy properties considering the presence of a relatively inhomo-
geneous distribution of elastic particles. After strains
Elastic
YS,* UTS, Elongation, modulus,
of about 3 % the stress-strain curves of the composite
Alloy MN m-2 MN m-2 % GN m-2 and unreinforced alloy are essentially parallel, so all
6061 (T6) 275 310 20 69
the strengthening is associated with the first few
2014 (T6) 476 524 13 73 percent strain. It is also apparent from Fig. 17 that
2124 (T6) 325 470 12 72 the work hardening rate increases with a decrease in
2618 (T6) 370 470 9 74 particle size from 16 urn (400 grit) to 78 urn (800 grit).
7075 (T6) 505 570 10 72
This is consistent with current models for. coefficient
8090 (T6) 415 485 7 80
A356 (T6) 205 280 6 76
of thermal expansion and back stress strengthen-
A380 (F) 160 320 35 72 ing.112,114 Particle shape, in terms of aspect ratio, will
AZ61 157 198 30 38 influence composite strength, but for the typical aspect
AZ91 168 311 21 49 ratio range of up to 2: 1 it is not expected to be a
* 02% offset yield strength. major factor.l!"

601lJ-r---------------------, 20,u,--------------------,
C\I
18 --- A356
IE 50 C\I
z IE 16 - A3561l5SiC
2 40 z 14
(/) L
(/) ,12
(/)
~ 30
~
(/)
~1O
0::
~ 8
o 20 ........- 606l1SiClfT6 (/)
..J
W --+- 6061/SiCfT4 a
..J
6
>=10 ----&- 2124/SiCfT4 W 4
>=
o 10 15 20 25. 30 35
VOLUME FRACTION, 010 o 50 100 150 200 250 300 350
18 Variation in yield stress with volume fraction TEST TEMPERATURE. DC
of reinforcement for powder processed 19 Variation in yield strength with temperature
composites after 100 h at temperature for A356 and
A356/SiC/15p
The predominant strengthening factor is the volume

fraction of reinforcement, and this is shown for
strength is primarily controlled by the higher temper-
powder processed composites in Fig. 18. Powder
ature strength of the matrix. However, a small
processed material tends to give somewhat higher
improvement in strength over the unreinforced alloy
strengths than melt processed composites, probably
is retained at higher temperatures, as shown in Fig. 19.
because of additional strengthening from oxide disper-
This figure compares the strength at temperature of
soids, and the somewhat finer grain size. Note that
unreinforced A356 with A356/SiC/15p after a 100 h
the strength increase appears to be reduced at the
soak at temperature. The composite has about a
very high volume fractions of reinforcement. This
10 MN m -2 higher strength than the matrix at 300C,
probably reflects mixing difficulties in obtaining a
and the decrease in composite strength with temper-
uniform particle distribution at high volume fractions.
ature follows that of the matrix. By using powder
The extent of strengthening is also dependent on
processing of rapidly solidified powders, improved
the matrix alloy microstructure, being lowest in peak
high temperature strengths can be achieved..i i s
aged material. Age hardened alloys tend to be more
Another alternative is to use a very high volume
susceptible to strain localisation, and the work hard-
fraction of reinforcement in a normally solidified
ening rate decreases very rapidly with strain. This
matrix. By going to a high volume fraction of particles
effect will also occur in composites, and it means that
the Orowan strengthening component becomes sig-
the strengthening observed is alloy and heat treatment
nificant, and this is essentially independent of temper-
dependent. While dislocation density and elastic misfit
ature other than the temperature dependence of the
strengthening models are in agreement with the gen- modulus. The Lanxide Corp. Primex pressureless
eral trends of the strengthening results, they are not
metal infiltration technology allows high volume frac-
sufficiently detailed to account for particle size and
tions to be obtained in near net shapes, and Fig. 20
distribution variations, and how the different micro-
shows the temperature dependence of Lanxide NX,
structural factors interact.
which is essentially 3360 alloy (AI-Si) containing
The ability to achieve any strengthening in a com- 60 vol._%SiC.1l6 The high temperature strength
posite .is depend ent on the ability to transfer stress
advantage of the composite is very significant, but
from the matrix to the stronger reinforcing particles.
there is, of course, a sacrifice in ductility.
This, in turn, is dependent on achieving a strong
interfacial bond between the matrix and the reinforce-
ment. If the interfacial bond is weak the interface will Elongation
fail before any effective stress transfer to the particle The major limitation in the mechanical properties of
can occur, and no strengthening is achieved. The composites is the rather limited ductility, as seen from
composite may, in fact, appear to be weaker than the Table 2. The tensile elongation decreases rapidly with
unreinforced matrix because of the reduced effective the addition of reinforcing particles, Fig. 21, and it
area supporting the load. As noted previously in the also decreases with increased aging time in the heat
discussion of elastic modulus, the presence of thermal treatable alloys. Recent work has demonstrated that
residual stresses, and inhomogeneity of reinforcement composite failure is associated with particle cracking
distribution, result in the early onset of plasticity and and void formation in the matrix within clusters of
no strengthening at very low strains, i.e, close to the particles.68,llS-12o Particle fracture is more prevalent
proportional limit of the stress-strain curve. in coarser particles, and this would be expected from
A final aspect of strength, which is of commercial two points of view. The larger the particle the more
importance, is the behaviour at elevated temperatures. it will be loaded by conventional fibre loading and
While the presence of particles improves the modu- end loading mechanisms. In addition, the coarser
lus at higher temperatures (Fig. 16) they do not add ceramic particles will have a higher probability of
significantly to the high temperature strength. The containing fracture initiating defects. Failure associ-
reason for this is that the strengthening mechanisms ated with particle clusters can be attributed to the
operating in composites at lower temperatures are higher stress triaxiality generated in these regions.
relaxed at higher temperatures, and the composite Drucker'?' noted that matrix deformation between
25u,-------------------, 20
o Iti um
-.- 336/60SiC 18 D 7.5Jlrn 0
-.- ALLOY336 16
~ 20
E cf!.
14 []
z
D
215 612
0
I ~1O
B
I-
(!)
~1O z 8 8 0
W 0
...J
D
0::: 6
l- w
(/)5 4
0+----+--+---+--+--t---f--t--+---+--4--1
o 100 200 300 400 500 600
-200 -150 -100 -50 0 50 100 150 200 250 300 350
TEST TEMPERATURE,oC TEST TEMPERATURE, C
20 Temperature dependence of strength of 336 22 Influence of testing temperature on tensile
with and without 60 vol.-%SiC elongation for extruded A356/SiC/15p with
400 and 800 grit sizes
closely spaced elastic particles would be highly con-

strained, resulting in local stress levels many times nucleation, (J'e,is given by
the matrix flow stress. This behaviour has been con- . . . . . . . (21)
firmed by recent continuum modelling.106,122Since
ductile failure mechanisms by void nucleation and where a.; is the equivalent stress and (J'rnthe mean
coalescence are very sensitive to the triaxial stress stress ((J'rn= (J'ii/3).Void nucleation will therefore be a
level,123fracture nucleation within particle clusters is function of the matrix flow stress, which will be
expected. influenced by the volume fraction of particles, heat
Many models have been developed for void treatment, etc. In the Argon et al. model the critical
nucleation. Particle cracking by catastrophic propa- stress is independent of particle size, but particle size
gation of an interal defect is given by the Griffith effects would come into the picture if the particle
equation distribution is a function of particle size, and also
through the particle size dependence for particle
(Jf = (~cy)1/2
,'- (20) cracking, equation (20).
In unreinforced material, voids grow after being
where (J'fis the stress on the particle, y the fracture nucleated until they are of a sufficient size to coalesce,
surface energy, E the Young's modulus of the particle, resulting in final fracture. However, with the high
volume fraction of particles in composite, extensive
and C the internal crack length.
For coarser particles there is a higher probability void growth does not occur. The limit load failure
model of Thomason 123 predicts the condition for
of their containing a defect of length C sufficient to
spontaneous ductile fracture at the void nucleation
give particle fracture. It is worth noting that the
particles are being loaded through the particle/ strain as
matrix interface, so a high degree of particle fracture 03[(n/4 Jtf)1/2 - IJ + 06 1 (J'rn
is indicative of a high interfacial strength, particularly 1 - Jtf = 2 + 2k . . . . (22)
when cracking is occurring in finer particles,
<10 urn, where Jtf is the volume fraction of nucleated voids,
A criterion for particle-matrix decohesion appro- (J'rnthe mean stress, and k the matrix shear flow
priate for coarse particles has been developed by stress.
Argon et al. 124
The critical stress for interfacial void If the mean stress, (J'rn= 2k, this expression predicts
that a void volume fraction greater than 0055 would
result in spontaneous fracture without extensive void
growth. In composites with typical particle volume
2
fractions greater than 0'1, the fracture process would
18
-+- 6061/Sicrr6 be nucleation controlled, consistent with the limited
-+- 60611A~03 /rs void growth seen experimentally.
~1 Since particle fracture is readily observed in most
6 12 failed composite, it is often suggested that this is the
~1 dominant failure process. However, particle fracture
~ 8
occurs from very early strains, as demonstrated by a
0
...J
detectable decrease in modulus with increasing
W strain.119,120,125It is also apparent, in melt processed
composites, that very different particle fracture behav-
2
iour is not reflected in differences in the tensile
0 10 15 20 25 30 35 40 elongation. Figure 22 shows the variation in tensile
VOLUME FRACTION, 010 elongation with test temperature for A356/SiC/15p
21 Variation in tensile elongation of 6061 with with two grit sizes. There is very little difference in
volume fraction of reinforcement the elongation between the two grit sizes, but the
~ 0.20,---------------------. 4. Control of particle shape .

..J o
U 0.18 o lti urn 5. A ductile matrix.
F= o 7.5Jlrn Composite fabrication and processing will influence.
~ 0.16
the degree of flexibility available to meet these
n; 0.14
o
o requirements.
W 0.12
o
~
~ 0.10
Fracture toughness
50.08
The fracture toughness of particle reinforced compos-
~ 0.06 o ites has been reviewed in two recent publi-
Z 0.04 DOD cations.126.127 While the toughness mirrors to some
o o
n extent the tensile elongation, decreasing with increas-
~ 0.02
~ 0.00 +-0----G~ ....
_ _+___+___+-l____+__+___l:jl--_+__+__+-+___+__j ing reinforcement, the decrease is most significant
E 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 from zero to 100/0 reinforcement, with only a slight
UTS, MN m-2
decrease for higher reinforcement loadings. The frac-
23 Fraction of cracked particles in A356/SiC/15p ture toughness is also much less sensitive to the degree
composites fractured at different tem- of aging than is the tensile elongation, and there
peratures, and with grit sizes of 400 and 800
are some data suggesting that coarser particles are
beneficial. With typical' toughness values in the
fraction of cracked particles is much larger in the 15-20 MN m -3/2 range, it is apparent that particle
coarser grit size, as shown in Fig. 23. The minimum composites do provide a reasonable degree of fracture
in tensile elongation at 200C is due to precipitation toughness.
hardening occurring during the test. It is interesting In the continuum mechanics approach to fracture,
to note that the minimum in tensile elongation is not the fracture toughness of a' material is normally
reflected in a maximum in the extent of particle assessed in terms of some crack tip parameter at the
cracking, which demonstrates the importance of the initiation of crack growth. Specifically, fracture will
matrix in composite fracture. At temperatures above occur when the crack tip strain, et, is exceeded over
250C general cavitation becomes the predominant some microstructurally significant characteristic dis-
failure mechanism, and very little particle fracture tance, 10, ahead of the crack tip. Under conditions of
occurs. small scale yielding
Obviously coarse particles, whether reinforcing et = co] . . . . (23)
particles or intermetallics, should be avoided to mini-
mise particle fracture. A high interfacial strength is where b is the crack tip opening displacement, x the
needed to minimise cracking along the interface and distance ahead of the crack, and c a constant of the
to load the particles effectively. However, the micro- order of 1.
structure of the matrix is also important, because At fracture initiation, et = ef over a distance x = 10,
local matrix failure appears ultimately to control the when b = be and, for small scale yielding
fracture. In addition, the particle distribution will also be = CK2 / Eay (24)
influence the composite ductility. The fracture models,
discussed previously, assumed a uniform particle dis- where K is the stress intensity factor, E Young's
tribution, and the mean stress in equations (21) and modulus, ay the yield stress, and C a numerical
(22) is a far field stress. With the inhomogeneous constant depending on the work hardening exponent
distribution of particles in real composites, intrinsic n, and the stress state, typically '" 0'5-0,6.
local stress triaxiality is generated in the clustered The problem is then reduced to relating be to some
regions which may dominate any far field stresses, microstructural feature in the composite. One sugges-
and make it difficult to compare the models with tion is that be scales with some average interparticle
experiment. Recent work on notched tensile bars spacing, but for 2000 series alloys reinforced with
shows that the extent of far field triaxiality has Al203 particles the agreement is poor, especially for
only a small influence on the tensile elongation in particle sizes typical of most commercial composites.
composites, compared with the major influence in Hahn and Rosenfield 128assumed that crack exten-
unreinforced alloys."? The importance of particle dis- sion occurs when the extent of the heavily deformed
tribution is also demonstrated by the increase in region ahead of the crack tip is comparable with the
tensile elongation with extrusion ratio, as shown width of the unbroken ligaments separating cracked
in Fig. 13. particles. They related the critical stress intensity
Considering all these different factors it is clear that factor for fracture to the volume fraction of reinforce-
the fracture process in particle composites is quite ment by the expression
complex, and a quantitative understanding is lacking. K1C = [2ayE(n/6)1/3d]1/2j;1/6 . . . (25)
It is also apparent that different composites may be
dominated by different fracture processes, but to where K1C is the critical stress intensity factor (i.e.
maximise ductility for, a particular volume fraction, fracture toughness), a y' the yield stress, E Young's
the composite should have: modulus, d the particle diameter, and fp the volume
1. Uniform particle distribution. fraction of particles.
2. A fine 10 urn), uniform particle size A problem with this model is that it predicts an
distribution. increase in toughness with increasing strength; a
3. A high interfacial strength. prediction which is not generally valid in unreinforced

alloys. In Al203 reinforced 6061 and 2014 alloys Be ~ Bf, the fracture strain, may be a reasonable
Klimowicz and Vecchio V? found that the fracture assumption.
toughness decreased with aging time, and hence Table 4 compares the fracture toughness predic-
increasing strength, and also continued to decrease tions of equations (25) and (27) with the experimental
even in the overaged condition where the strength results of Klimowicz and Vecchio 129 for 6061/
was decreasing. However, Manoharan and A1203/15p and (AI-Cu) 2014/AI203/15p. In com-
Lewandowski r''' obtained reasonable agreement with paring the data the particles were assumed to be
this model for underaged AI-Zn- Mg-Cu matrix com- homogeneously distributed, with a 10 urn particle size,
posites containing 15 and 20 vol.-oiO SiC particles of and the crack radius was taken as 50 urn,
13 urn average diameter. The equation also predicts As expected the Hahn and Rosenfield equation
an increase in fracture toughness with increasing predicts an increase in fracture toughness with
particle size, which is also not in agreement with increased aging which is contrary to experiment. The
some experiments on unreinforced alloys.P! Some limit load approach of Thomason correctly predicts
studies in composites have reported an increase in the general trend with aging time, but tends to
toughness with increasing particle size, 126,132 while underestimate the fracture toughness, particularly for
other studies report no affect of particle size. 11 7 In the 2014 matrix composite. However, the extent of
general, the fracture toughness does not show a strong agreement does suggest that models based on
dependence on reinforcement particle size.P? nucleation controlled fracture may be appropriate for
Kraffr'P" also proposed that fracture was controlled particle composites. In this case control of particle
by failure of ligaments ahead of the crack tip and volume fraction and distribution will be important
suggested that ligament failure occurred at the insta- for optimum fracture toughness.
bility strain observed in a smooth bar tensile test. There is not much information on the influence
This model gives the expression of temperature on fracture toughness, but in
2009/SiC/20p there is a small increase between 25
K1C = En(2nA)1/2 . . . . (26)
and 200C, and a decline above 220C.136 A micro-
where E is Young's modulus, n the work hardening mechanical model based on a critical strain concept
exponent, and A the process zone size. was used to predict the temperature dependence, with
Recently, Topur134 has compared equation (26) the local fracture strain being inferred from tensile
with fracture toughness data from Al203 reinforced data.
2014 (AI-Cu) and obtained a process zone size It is apparent from this discussion that the present
A ~ I urn. She concluded that this was too small for models of fracture toughness need extensive develop-
the 15% of 9 urn diameter particles studied, and ment to explain the toughness of MMCs. It is difficult
suggested a 'brittle' variant of the model by replacing to take account of reinforcement inhomogeneity,
the strain hardening exponent with the fracture strain particle size distribution, and any residual stresses
of the composite. This approach gives A 'in the range present in the composite, all of which are likely to
10-100 urn, which is comparable with the spacing influence the fracture toughness.
between particle clusters in the composite Finally, it should be noted that obtaining valid
in ves tiga ted. plane strain fracture toughness values for composites
The Krafft model is a limit load failure model can be a problem and many of the results in the
which has been further developed by Thomason 123,135
to represent the strain distribution at the crack tip
more accurately. For fracture which is nucleation Table 4 Comparison between theoretical Kth and
controlled the fracture toughness is given by experimental Kexp fracture toughness
values
K1C = 258(pG'yEBe)1/2 . (27)
cry, Elongation, Kexp, ~h'
where p is the crack tip radius, typically '" 50 urn, G'y Composite MN m-2 % MN m-3/2 MN m-3/2
the yield stress, E Young's modulus, and Be the Thomason equation
microvoid nucleation strain, which is dependent on AI-Cu
the mean normal stress, the volume fraction of void 2014/AI203/15p
nucleating particles, and the particle-matrix inter- 3 h, 160C 331 6 242 254
7 h, 160C 441 2 217 120
facial strength. 16 h, 160C 469 1 195 124
Notice that in equation (27) the decrease in ductility' 40 h, 160C 420 1 190 117
with increasing yield stress will counter the predicted 48 h, 160C 372 2 181 156
increase in toughness with increasing stress, unlike AI-M92Si
6061/AI203/15p
the situation in the Hahn- Rosenfield model.
1 h, 17rC 221 7 247 215
However, equation (27) requires an expression for the 3 h, 17rC 317 5 232 218
critical strain for void nucleation in terms of matrix 10 h, 17rC 345 3 227 176
strength, particle size, distribution, and stress state 25 h, 17rC 331 3 215 172
before the fracture toughness can be related to the 100 h, 177C 276 4 212 180
microstructural features of the composite. As noted Hahn and Rosenfield equation

AI-Cu
previously, the fracture process is not sufficiently 2014/AI203/15p
understood for this to be done at present. However, 3 h, 160C 331 6 242 311
recent experiments on notched tensile bars showed AI-Mg2Si
that the fracture strain was relatively insensitive to 6061/AI203/15p
1 h, 177C 221 7 247 248
the degree of far field triaxiality, suggesting that

280-r---=-----------------,
1O-4.-r-----------------~
(\J 260-
IE o 6061~T6
Z 240- D 6061/15A1203,T6 10-5

~ Q)
, 220- u
(j)
(j) ~
~ 200-
~ 10-6
t- EU
(j) 180-
0% A~03 .
~
:J
~
X
160-
140-
-
~ 10-7
ro
"0
(/~~/ 15% AIz0,
10-8
~ 120- I
i
I
I
103 1b4 105 1~6 107 I
1O-9-t-----r---.----,---,-r--r-.--.-----.--~"T""""""T"-_._____._J
CYCLES TO FAILURE 3 5 10 20
24 S-N curve for 6061- T6 and 6061/AI203/15p, STRESS INTENSITY RANGE (~K), MN m-3/2
T6 at R= -1 25 Fatigue crack growth rate for 6061- T6 and
6061/AI203/15p, T6 at R=-1
literature are acknowledged to be invalid according

to standard testing practice, e.g. ASTM E399 and shaped curve of crack growth rate against stress
BS 5447. Residual stresses and inhomogeneity of intensity range exhibited by unreinforced alloys. The
particle distribution often result in excessive curvature curve is bounded at the low crack growth end by the
in the initial fatigue crack which invalidates the test. threshold stress intensity range for initiating growth
Various aspects of fracture toughness testing of (AKth) and at the high growth rate end by the stress
particle metal matrix composites have been recently intensity for unstable crack growth, i.e. Kmax
considered by Roebuck and Lord.137 approaches the fracture toughness K1C' The data
generally shows that the threshold stress intensity
Fatigue range is higher in MMCs than in unreinforced mater-
Many of the potential applications for composites ial, and an example of this is shown in Fig. 25, again
require a resistance to cyclic loading, and several for 6061/AI203/15p, T6. This higher threshold is
attributed to crack closure effects. You and Allison 144
fatigue studies in MMCs have been reported - see
for example, Refs. 138-141. Intuitively, one would found that in 2124/SiC/20p, particle-matrix deco-
expect the low cycle fatigue behaviour of MMCs to hesion was prevalent at near threshold, while there
be somewhat worse than unreinforced alloys because was increasing evidence of particle cracking at higher
of the lower ductility in composites, whereas the high AKs. This study also found that the fatigue crack
cycle performance should be improved because of the growth rate was relatively insensitive to aging.
higher modulus. While there is considerable scatter However, in 7XXX/SiC/20p the situation was found
in the published data, reflecting variations in pro- to be more complex, with the behaviour depending
cessing and material quality, these expectations are on particle size and the stress intensity range.l'"
generally fulfilled. Figure 24 shows the 8- N curve for Since the crack growth rate plot is bounded by
AKth and Kmax = K1C, and since AKth is higher and
extruded unreinforced 6061- T6 and 6061/Al2 03/
15p- T6 at R = -1. However, several studies141,142 K1C lower than for unreinforced alloys, the stress
have shown that the improvement in fatigue life intensity range for fatigue crack growth in composites
evident in stress life data is eliminated when com- will be smaller than for unreinforced alloys. Other
than this, the fatigue crack growth rate curve for
pared on a strain life basis. Under constant strain
amplitude conditions the MMC is inferior in the composites has the same general shape as for unre-
low cycle regime where plastic strains dominate, and inforced alloys, and does exhibit a linear stage II
in the high cycle regime the composite is little different regime obeying the Paris law
to unreinforced material. The improvement observed da/dN = BAKs . . . . . . . . . . (28)
in constant stress amplitude tests reflects the fact that
with the higher Young's modulus of the composite, where da/dN is the crack growth rate, AK the stress
the strains in the composite are lower than those in intensity range, and Band 8 are constants. "-
the unreinforced material at the same stress level. Davidson 139 found that for a wide range of different
Constant stress amplitude tests involve both fatigue aluminium composites, which included mechanically
crack initiation and crack propagation. Crack alloyed, powder metallurgy, and molten metal pro-
initiation has been associated with defects in the cessed composites, the parameters Band 8 were
composite, intermetallic particles, and large reinforce- related by the expression
ment particle or particle clusters. It has been suggested In B = -16'1- 228 . . . . . (29)
that fatigue cracks initiate late in the life of compos-
ites.143 Fatigue crack propagation is generally char- This relationship was developed for R = 0'1, and
acterised in terms of fracture mechanics, and is investi- compares with In B = -0,143 - 03758 obtained for
gated by crack growth experiments. Crack growth unreinforced Al alloys.l''" Since the correlation
experiments on composites show the typical sigmoidal between Band 8 is very sensitive to R ratio in
aluminium alloys, equation (29) is also likely to vary that the creep rate decreased with increasing volume
with R. However, such an approach may prove useful fraction of reinforcement, and the threshold stress
in estimating fatigue performance. also increased with volume fraction, from
The results to date show that the fatigue perform- 145 MN m-2 for 10% (1'7 urn) to 328 MN m-2 for
ance of the composite is different to that of the 300~, tested at 350C. This suggests that the creep
unreinforced matrix, but whether the fatigue behav- rate is not controlled by oxide dispersoids. They
iour is improved or not depends on the mode of analysed their results in terms of the structure
testing. Most of the studies indicate that the fatigue invariant model proposed by Sherby et al.153
performance can be comparable with or better than 3
the unreinforced matrix, except at high stress or strain s
= A'DL
b
A (0" - 0"0)8
E (31)
5
amplitudes where the reduced ductility of the compos-
ite influences the behaviour. There are insufficient where A' is a constant, DL the lattice diffusivity, ).,the
data to assess the influence of particle size, shape, subgrain size, b the Burgers vector, and E the Young's
etc., and fatigue crack growth models are not suffic- modulus.
iently well developed to make any predictions. / "This equation describes the creep of metals with a
constant substructure deforming by lattice diffusion
Creep controlled creep. They suggest that the subgrain size
In previous sections it was pointed out that particle is controlled by the average spacing between reinforc-
reinforcement provides an increase in the elastic ing particles, and also stabilised by the reinforcing
modulus at elevated temperatures, and only a rela- phase. For composites with fine, 17 urn reinforcing
tively small increase in the flow stress. for many high particles, they found good agreement with this equa-
temperature applications the creep properties of the tion, and about an order of magnitude improvement
material are important and .several recent creep stud- in creep rate over unreinforced alloys. However,
ies have been reported, primarily on powder processed for coarse reinforcement, 145 and 459 urn, while
composites. the threshold of 158 MN m - 2 is equal to that for the
The steady state creep rate can be expressed as finer reinforcement, the creep rate in terms of the
effective stress, (0" - 0"0) showed no improvement over
8 = AO"" exp(-Q/RT) . . . . . . . . (30) unreinforced alloys. This behaviour can be explained
if it is assumed that the interparticle spacing in the
where 8 is the creep rate, A a constant, 0" the creep case of the coarser reinforcement is too large to
stress, n the stress exponent, Q the activation energy, influence the substructure. However, there is presently
R the gas constant, and T the absolute temperature. no satisfactory explanation for the value of the thresh-
Applying this equation to the tensile creep of old stress observed.
composites gives high n-values, 9,5-20,5, and From the creep experiments to date it would appear
high activation energies, 390-400 kJ mol-l (Refs. that a high volume fraction of fine particle reinforce-
145-149). These values compare with n = 5 and ment can provide significant improvement in creep
Q = 142 kJ mol-l for the self-diffusion of aluminium, resistance up to about 350C. At higher temperatures
A high stress exponent is reminiscent of oxide dis- the reinforcing particles are unlikely to exercise much
persion strengthened (ODS) alloys. A recent study constraint on the plasticity, and the creep behaviour
of ODS Al and AI-Mg alloys showed values of will approach that of the matrix.
n = 15-25, and apparent activation energies of
500-540 kJ mol-l (Ref. 150). To explain this behav-
iour it is usual to introduce the concept of a threshold Commercial aspects
stress, 0"0' and replace the creep stress, 0", by (0" - 0"0) As noted in the introduction, particle MMCs are now
in equation (30). Park et al.151 used this approach in at the commercial production stage, and a whole
analysing the creep in shear of 6061/SiC/30p. Their range of factors have to be addressed to produce a
measurements extended over seven orders of magni- cost competitive component. Table 5 lists the factors
tude of strain rate and exhibited two regimes of involved in determining the finished component costs.
behaviour. At low stress and strain rates, n is high Considering that MMCs contain hard ceramic
and increases with decreasing stress, indicative of a
threshold stress of 81 MN m - 2. At high stresses and
strain rates n tends towards a constant value of 74. Table 5 Finished component cost factors for an
Substituting the threshold stress of 81 MN m - 2 into . extruded wrought product
the creep equation gives a value for n of 5, consistent
Extrusion billet cost
with unreinforced alloys. Park et ale suggested that Material
the threshold stress was due to dislocation interaction, Sawing
Extrusion cost
not with the SiC reinforcing interaction, but with fine Extrusion speed
oxide particles incorporated into the composite Die cost
during powder processing. Die wear
Recoveries
One problem with these tensile creep studies is that Cutting
the extent of steady state creep is quite limited, raising Machining
questions regarding the validity of some of the data. Bending and forming
Welding
To overcome this difficulty Pandey et al.152 have Surface finishing - anodising or painting
tested AI/SiC/lOp, 20p, and 30p in compression, and Transportation
Recycling
with two particle sizes, 17 and 145 urn. They found
particles, have high wear resistance and limited duc-

25.----------------------,
tility, even the most straightforward operation, such ....J
as sawing, can require modification of the conven- AI/SiC

2
tional equipment. Extensive sawing and machining ffi <D 20
studies have been carried out, and viable commercial ~~

practices established.P" I.L X. 15
OZ
Most applications require some degree of machin- I-Q
Z(j)
ing, such as sawing, milling, drilling, reaming, and ~zlO
tapping. The ceramic particles present in composites ~~
I.LX
I.LW
have hardnesses approaching that of the machine tool W
material, which means that abrasive wear of the o
u
tooling is the main issue. To minimise abrasive wear
01+-r-,__r_r_1""""T""T'""T""T""T'"
.........
"""T""'T""'1"""T"""T"""............-~..,...,_,._r_T"""T__r_r_.....,....,_T'""'T""T'"
.........
""T"""r'T_r_T""'~
the area and contact time between the tool and the o 10 20 30 40 50 60 70 80 90 100
workpiece should be kept as low as possible during VOLUME FRACTION, 0/0
the machining operation. This means that MMCs 26 Variation of coefficient of thermal expansion
should be machined at higher feed rates and depth of with volume fraction of SiC in variety of AI
cut than traditionally used with unreinforced Al alloy matrices
alloys. As an example, face milling of unreinforced Al
is typically carried out at a feed rate of around
0013 ern per tooth. Increasing this by a factor of 4 in the majority of components. In addition, a key
for the composite not only increases the removal rate, cost parameter in all processing routes will be the
but also increases tool life by a factor of 3. Low ability to recycle the scrap and/or the final component
cutting speeds, which minimise the temperature rise when its useful life is over. This is particularly import-
during machining, also increases tool life. ant for aluminium matrix composites, because the
The hardness of the tooling is an important factor whole Al business philosophy is based on an ability
in controlling tool life. High speed steel tools are to recycle.
dulled in seconds, while conventional and coated There are two strategies which can be used to
carbide tooling last only a few minutes. By far, the handle scrap MMCs:
most cost effective tooling is polycrystalline diamond 1. Recycle: for reuse as composite.
(PCD), provided it is of high quality. Unfortunately, 2. Reclamation: reclaim the individual components
PCD tooling is not yet available for taps or very of the composite, i.e. separate the composite into
small diameter drills and reamers, so carbide tools its two constituents parts, the matrix alloy and the
must be used for these operations. ceramic particles.
The composition of the MMC, in terms of type, Both strategies have to be cost effective and
shape, size, and volume fraction of reinforcement will environmentally safe.
affect machine ability. A matrix with a soft reinforce- To date, no information has been reported on
ment, such as graphite, will be easier to machine than recycling powder metallurgy material, but it will

a matrix reinforced with Al2 3, The shape of the
reinforcement may influence the ease with which the
presumably involve melting. Composites initially pro-
duced by the molten metal route lend themselves to
ceramic particles are sheared, but there is no inform- recycling by remelting, since most of the issues
ation available on this. Increasing the reinforcement involved, such as reactivity between the particle and
particle size and volume fraction will increase the the molten metal, have already been considered in
abrasiveness of the composite, but the uniformity of the original processing. However, melt cleanness and
the reinforcement distribution will also be important. degassing are more difficult problems when dealing
Large clusters of ceramic particles will act as large with scrap, and scrap sorting is very important.
inclusions with regard to tool damage. Fluxing and degassing technologies have been devel-
Many factors, therefore, affect the machining of oped by Duralcan, and presumably other composite
MMCs, but guidelines are being developed which will producers will develop similar technologies.
enable MMCs to be commercially machined, see for The reclamation route is an alternative approach
example, Ref. 155. which is also available to composites which can be
Many of the early composite extrusions used lubri- remelted. Using combined argon and salt fluxing, SiC
cated conical dies to facilitate extrusion, but this is particles can be removed from the melt, and 85-90%
not commercially viable except for very simple, sym- of the Al can be recovered.
metrical shapes. Fortunately, the conventional direct
extrusion process with unlubricated shear faced dies
is feasible with appropriate die design and extrusion Some general comments
practice, though speeds are lower and extrusion press- This review has primarily been concerned with the
ures higher.156.157Excessive die wear can be mini- factors influencing the microstructural-mechanical
mised by using high wear resistant inserts. However, properties relationship of composites. It should be
appropriate extrusion die material is still an area appreciated however, that in some applications Jor
requiring development, particularly for complex composite the physical properties are of paramount
shapes. importance. Addition of SiC particles to aluminium
Directly casting to final shape, and other near net alloys can reduce the coefficient of thermal expansion
shape processing reduces some of these production of the alloys, while still maintaining a thermal con-
difficulties, but a degree of final machining is necessary ductivity equivalent to the matrix alloy. Figure 26
shows the influence of SiC on the coefficient of commercial component runs, and meeting specifica-
thermal expansion of a variety of Al alloys, and the tions with' high yields and at acceptable cost, now
reduction is close to rule of mixture behaviour. Low become the key factors in successful exploitation. The
coefficient of thermal expansion and high thermal technologies are now available which should be able
conductivity is an attractive combination for appli- to meet these requirements, and the next 5 years or
cations requiring dimensional stability. so will demonstrate whether particle MMCs will
While wear resistance is a systems property rather become a general use engineering material, or be
than a materials property, there are many situations limited to niche markets, which is the role they have
where the wear resistance is much higher in the played to date.
composite than in the unreinforced alloy. iSS This
makes MMCs attractive for bearings, bushings, cylin-
der liners, and break rotors. In some cases this
Acknowledgments
property is only required at the surface, and generat- The author is. grateful to Alcan International Ltd for
ing a surface composite layer by spray deposition, or permission to publish, and to his colleagues at the
some other route, may be the appropriate means Alcan Research Centres in Kingston, Banbury, and
of use. Arvida, and at Duralcan USA, San Diego and Dubuc
In many applications the corrosion behaviour of plants, for their contributions to the ideas and inform-
the composite is important. Since many ceramic ation in this article.

particles, such as Al2 3, are insulators, they would
not be expected to affect corrosion behaviour directly. References
SiC, however, is a conductor and changes in corrosion
1. A. KELLY: Compos. Sci. Technol., 1985,23,171-199.
behaviour need to be assessed. Even in the case of 2. T. W. CHOU, A. KELLY, and A. OKURA: Composites, 1985,
insulating reinforcing particles, the composite micro- 16, 187-206.
structure is different to that of the matrix as a result 3. J. DINWOODIE, E. MOORE, C. A. J. LANGMAN, and w. R. SYMES:
of the processing, e.g. finer grain size, coarse inter- in Proc. 5th Int. Conf. on 'Composite materials - ICCM-V',
metallics, reaction products, and this can modify the (ed. W. C. Harrigan, Jr et al.), 671-685; 1985, Warrendale,
PA, The Metallurgical Society of AIME.
corrosion response. There is very little published work 4. s. V. NAIR, J. K. TIEN, and R. C. BATES: Int. Met. Rev., 1985,
on the corrosion behaviour of composite, and no 30, 275-290.
doubt it will depend critically on the matrix-reinforce- 5. T. DONOMOTO, K. FUNATANI, N. MIURA, and N. MIYAKE: SAE
ment combination and the particular environment. Technical Paper 830252, Detroit, MI, 1983.
6. M. H. STACEY: Mater. Sci. Techno!., 1988, 4, 227-230.
With the tendency for grain refinement in MMCs, no 7. DEONATH and P. K. ROHATGI: J. Mater. Sci., 1980, 15,
increase in susceptibility to stress corrosion is 2777-2784.
expected in recrystallised microstructures. However, 8. M. K. SURAPPA and P. K. ROHATGI: J. Mater. Sci., 1981,
it may be more difficult to prevent recrystallisation 16, 983-993.
in those alloy matrices where an unrecrystallised 9. A. BANERJI, M. K. SURAPPA, and P. K. ROHATGI: Metall. Trans.,
1983, 14B, 273-283.
microstructure is required for maximum stress cor- 10. B. P. KRISHNAN and P. K. ROHATGI: Met. Technol., 1984,
rosion resistance. For those alloy-environment com- 11, 41-44.
binations where pitting and intergranular corrosion 11. W. KAI, J. M. YANG, and w. C. HARRIGAN, Jr: Scr. Metall., 1989,
are a consideration, a somewhat inferior performance 23, 1277-1281.
12. B. ROEBUCK and A. E. J. FORNO: in 'Modern developments in
in the composite is expected. In the studies conducted powder metallurgy', Vol. 20, 451-465; 1988, Princeton, NJ,
to date, corrosion does not appear to be a major Metal Powder Industries Federation, American Powder
concern provided the appropriate matrix is chosen, Metallurgy Institute.
and a suitable microstructure developed. 13. J. J. LEWANDOWSKI, C. LIU, and w. H. HUNT, Jr: in 'Processing
In summary, a wide range of particle MMCs are and properties for powder metallurgy composites', (ed.
P. Kumar et a!.), 117-137; 1987, Warrendale, PA, The Metal-
now available in semicommercial and full commercial lurgical Society of AIME.
quantities. Depending on the initial fabrication route, 14. D. J. LLOYD, H. P. LAGACE, and A. D. McLEOD: in 'Controlled
there is almost an unlimited range of matrices, interphases in composite materials - ICCM-III', (ed.
particles, and loadings available, which can provide R. Ishida), 359-376; 1990, London, Elsevier Applied Science.
a broad range of mechanical and physical properties. 15. D. J. LLOYD and B. CHAMBERLAIN: in 'Cast reinforced metal
composites', (ed. S. G. Fishman and A. K. Dhingra), 263-269;
However, the properties will inevitably be inferior to 1988, Metals Park, OR, ASM International.
those obtainable with continuous reinforcement, so 16. c. M. ADAMS and R. E. LEWIS: in 'Rapidly solidified crystalline
they provide a compromise combination of properties alloys', (ed. S. K. Das et al.), 157-183; 1985, Warrendale, PA,
between unreinforced and continuous fibre reinforced The Metallurgical Society of AIME.
17. M. S. ZEDALIS, J. M. PELTIER, and P. S. GILMAN: in 'Light-weight
composite. With this compromise comes the ability alloys for aerospace applications', (ed. E. W. Lee et a!.),
to use conventional fabrication methods and lower 323-334; 1989, Warrendale, PA, The Metallurgical Society
cost. For commercial success large tonnage appli- of AIME.
cations need to be developed, and these are most 18. F. A. BADIA and P. K. ROHATGI: Trans. AFS, 1969, 76, 402-406.
likely to be in the transportation industry. Some 19. A. M. PATTON: J. Inst. Met., 1972, 100, 197-201.
20. A. W. NEUMANN, C. J. van OSS, and J. SZEKELY: Kolloid-Z.u.Z.
commercial application has occurred in the sporting Polym., 1973, 251, 415-423.
and bicycle industry, and small quantities of MMCs 21. P. K. ROHATGI: in 'Interfaces in metal-matrix composites', (ed.
have been used in space and aircraft. However, major A. K. Dhingra and S. G. Fishman), 185-202; 1986, Warren-
applications in the automobile industry, such as brake dale, PA, The Metallurgical Society of AIME.
22. K. C. RUSSELL, J. A. CORNIE, and S.-Y. OH: in 'Interfaces in
rotors and drive shafts, are now in the late stages of metal-matrix composites', (ed. A. K. Dhingra and S. G.
development, and other engineering applications are Fishman), 61-91; 1986, Warrendale, PA, The Metallurgical
also well advanced. Reproducibility of properties in Society of AIME.

23. P. K. ROHATGI and R. ASTHANA: in 'Cast reinforced metal 61. J. G. LEGOUX, G. L'ESPERANCE, L. SALVO, and M. SuERY: in
composites', (ed. S. G. Fishman and A. K. Dhingra), 61-66; 'Fabrication of particulates reinforced metal composites', (ed.
1988, Metals Park, OH, ASM International. J. Masounave and F. G. Hamel), 31-39; 1990, Materials Park,
24. F. DELANNAY, L. FROYEN, and A. DERUYTERRE: J. Mater. Sci; OH, ASM International.
1987, 22, 1-16. 62. H. RIBES and M. SuERY: Scr. Metall., 1989, 23, 705-709.
25. M. K. SURAPPA and P. K. ROHATGI: Met. Technol., 1978, 63. M. C. FLEMINGS: 'Solidification processing'; 1974, New York,
5, 358-361. McGraw-Hill.
26. v. G. GORBUNOV, u. D. PARSHIN, and v. v. PANIN: Russ. Cast. 64. D. G. THOMAS: J. Colloid Sci., 1965, 20, 267-277.
Prod., Aug. 1974, 348. 65. H.-K. MOON: PhD' thesis, Massachusetts Institute of
27. J. TAFTO, K. KRISTIANSEN, H. WESTENGEN, A. NYGARD, J. B. Technology, 1990.
BORRADAILE, and D. O. KARLSEN: in 'Cast reinforced metal 66. M. K. SURAPPA and P. K. ROHATGI: Metall. Trans., 1981,
composites', (ed. S. G. Fishman and A. K. Dhingra), 71-75; 12B, 327-332.
1988, Metals Park, OH, ASM International. 67. J. A. CORNIE: Personal communication, Boston, MA, 1991.
28. F. A. BADIA: Trans AFS, 1971, 79, 347-350. 68. w. H. HUNT, Jr, D. RICHMOND, and R. D. YOUNG: in Proc. 6th
29. Alcan Aluminum Corp.: US Pat. 4786467, 1988. Int. Conf. on 'Composite materials - ICCM VI', (ed.
30. J. B. BORRADAILE, S. SKJERVOLD, and w. RUCH: in Extended F. L. Matthews et al.), 2.209-2.203; 1987, London, Elsevier
Abstracts of Conf. 'Metal matrix composites: property Applied Science.
optimisation and applications', London, 1989, The Institute 69. T. M. OSMAN, J. J. LEWANDOWSKI, and w. H. HUNT, rr: in
of Metals, Paper 10.1. 'Fabrication of particulates reinforced metal composites', (ed.
31. T. SRITHARAN, K. XIA, H. HEATHCOCK, and J. MIHELICH: in J. Masounave and F. G. Hamel), 209-216; 1990, Materials
'Metal and ceramic matrix composites: Processing, modeling Park, OH, ASM International. ,
and mechanical behavior', (ed. R. B. Bhagat et al.), 13-22; 70. c. A. ROGERS: 'Packing and covering', Cambridge
.>
1990, Warrendale, PA, The Metallurgical Society of AIME. Mathematical Tracts, Vol. 54; 1964, Cambridge University
/
32. D. J. LLOYD: in 'High performance composites for the 1990s', Press.
(ed. S. K. Das), 33-45; 1990, Warrendale, PA, The 71. J. F. RICHARDSON and w. N. ZAKI: Chern. Eng. Sci., 1954, 3, 65.
Metallurgical Society of AIME. 72. G. H. GEIGER and D. R. POIRIER: 'Transport phenomena in
33. J. A. CORNIE, H.-K. MOON, and M. C. FLEMINGS: in 'Fabrication metallurgy'; 1973, New York, Addison-Wesley.
of particulates reinforced metal composites', (ed. J. Masounave 73. I. JIN and D. J. LLOYD: in 'Fabrication of particulates reinforced
and F. G. Hamel), 63-78; 1990, Materials Park, OH, ASM metal composites', (ed. J. Masounave and F. G. Hamel),
International. 47-52; 1990, Materials Park, OH, ASM International.
34. R. MEHRABIAN, R. G. RIEK, and M. C. FLEMINGS: Metall. Trans., 74. P. K. ROHATGI, R. ASTHANA, and s. DAS: Int. Met. Rev., 1986,
1974, 5, 1899-1905. 31, 115-139.
35. R. MEHRABIAN, A. SATO, and M. C. FLEMINGS: Light Met., 1975, 75. D. M. STEFANESCU and B. K. DHINDAW: in 'Metals handbook',
38, 177-193. 9 edn, Vol. 15, 142-147; 1988, Metals Park, OH, ASM
36. P. R. GIBSON, A. J. CLEGG, and A. A. DAS: Foundry Trade J., International.
Feb. 1982, 253-263. 76. M. TAYA, K. MURAMATSU, D. J. LLOYD, and R. WATANABE:
37. Dow Chemical Co.: US Pat. 4432936. JSME lnt. J., Ser. 1, 1991, 34, 198-206.
38. E. M. KLIER: SM thesis, Massachusetts Institute of 77. D. J. LLOYD: in 'Metal matrix composites - processing, micro-
Technology, 1988. structure and properties', 12th Rise Int. Symp. on Metallurgy
39. S. CARON and J. MASOUNAVE: in 'Fabrication of particulates and Materials Science, Rise National Laboratory, Roskilde,
reinforced metal composites', (ed. J. Masounave and Denmark, 1991, (N. Hansen et al.), 81-99.
F. G. Hamel), 107-113; 1990, Materials Park, OH, ASM 78. F. J. HUMPHREYS: Acta Metall., 1977,25, 1323-1344.
International. 79. w. S. MILLER and F. J. HUMPHREYS: in 'Fundamental relation-
40. 1. A. CORNIE: Personal communication, Kingston, ON, 1990. ships between microstructures and mechanical properties of
41. A. W. URQUHART: Mater. Sci. Eng., 1991, AI44, 75. metal matrix composites', (ed. M. N. Gungor and P. K. Liaw),
42. M. K. AGHAJANIAN, M. A. ROCAZELLA, J. T. BURKE, and s. D. 517-541; 1989, Warrendale, PA, The Metallurgical Society
KECK: J. Mater. Sci., 1991, 26, 447. of AIME.
43. A. R. E. SINGER: US Pat. 4515864, 1985. 80. M. FERRY, P. MUNROE, A. CROSKY, and T. CHANDRA: in 'Metal
44. Osprey Metals: UK Pat. 1 379261, 1975. matrix composites - processing, microstructure and prop-.
45. Osprey Metals: UK Pat. 1 472939, 1977. erties', 12th Rise Int. Symp. on Metallurgy and Materials
46. United Kingdom Atomic Energy Authority: US Pat. Science, Rise National Laboratory, Roskilde, Denmark, 1991,
4928 745, 1990. (N. Hansen et al.), 337-342.
47. T. C. WILLIS: Met. Mater., 1988, 4, (8), 485-488. 81. M. W. MAHONEY and A. K. GHOSH: Metall. Trans., 1987,
48. J. WHITE, I. R. HUGHES, T. C. WILLIS, and R. M. JORDAN: in 18A, 653-661.
'Proc. 4th Int. Conf. on aluminium-lithium alloys', J. Phys., 82. M. MABUCHI, K. HIGASHI, Y. OKADA, S. TANIMURA, T. IMAf, and
1987, 48, (9), C3, 347-353. K. KUBO: Scr. Metall. Mater., 1991,25, 2003-2006.
49. Martin Marietta Corp.: US Pat. 4 915908, 1990. 83. M. MABUCHI, K. HIGASHI, S. WADA, and s. TANIMURA: Scr.
50. Martin Marietta Corp.: US Pat. 4 710348, 1987. Metall. Mater., 1992,26, 1269-1274.
51. R. M. AIKIN, Jr: 'The mechanism of dispersion strengthening 84. J. PILLING: Scr. Metall., 1989,23, 1375-1380.
and fracture in Al-based XD alloys', Contract Report 4276, 85. K. HIGASHI, T. OKADA, T. MUKAI, S. TANIMURA, T. G. NIEH, and
National Aeronautics and Space Administration, Feb. J. WADSWORTH: se-. Metall. Mater., 1992,26, 185-190.
1990. 86. G. NUSSBAUM, P. SAINFORT, G. REGAZZONI, and H. GJESTLAND:
52. US Pat. 4 808 372. Scr. Metall., 1989, 23, 1079-1084.
53. A. G. METCALFE: 'Composite materials, Vol. 1 - Interfaces in 87. T. G. NIEH and R. F. KARLEK: Scr. Metall., 1984, 18,25:....28.
metal matrix composites'; 1974, New York, Academic Press. 88. s. NUTT and R. W. CARPENTER: Mater. Sci. Eng., 1985, 75,
54. A. MORTENSEN: in 'Mechanical and physical behaviour of 169-177.
metallic and ceramic composites', 9th Rise Int. Symp. on 89. H. J. RACK: in Proc. 6th Int. Conf. on 'Composite materials -
Metallurgy and Materials Science, Rise National Laboratory, ICCM-VI', (ed. F. L. Matthews et al.), 2.382-2.389; 1987,
Roskilde, Denmark, (ed. S. I. Anderson et al.), 141-155. London, Elsevier Applied Science.
55. D. J. LLOYD and I. JIN: Meta 11. Trans., 1988, 19A, 3107-3109. 90. J. M. PAPAZIAN: Metall. Trans., 1988, 19A, 2945-2953.
56. R. WARREN and c. H. ANDERSSON: Composites, 1984, 15, 91. P. B. PRAGNELL and w. M. STOBBS: in Proc. 7th Int. Symp. on
101-111. 'Composite materials - ICCM-VII', (ed. Yunshu et al.),
57. D. J. LLOYD and E. DEWING: in Proc. Int. Symp. on 'Advanced 573-578; 1989, Oxford, Pergamon.
structural materials', (ed. D. S. Wilkinson), 71-77; 1988, 92. s. SURESH, T. CHRISTMAN, and Y. SUGIMURA: Scr. Metall., 1989,
Montreal, PQ, Pergamon Press. 23, 1599-1602.
58. D. J. LLOYD: Compos. Sci. Technol., 1989,35, 159-179. 93. K. K. CHAWLA, A. H. ESMAELI, A. K. DATYE, and A. K. VASUDEVAN:
59. c. M. GABRYEL and A. D. McLEOD: Metall. Trans., 1992, 23A, Scr. Metall. Mater., 1991,25, 1315-1319.
1279-1283. 94. E. HUNT, P. D. PITCHER, and P. J. GREGSON: Scr. Metall., 1990,
60. K. L. MAMPEL: Z. Phys. Chern., 1940, A187, 43-57; 235-249. 24, 937-941.

95. P. B. PRANGNELL and w. M. STOBBS: in 'Metal matrix compos- 127.D. L. DAVIDSON: in 'Metal matrix composites: Mechanisms and
ites - processing; microstructure and properties', 12th Rise properties', (ed. R. K. Everett and R. J. Arsenault), 217-234;
Int. Symp. on Metallurgy and Materials Science, Rise 1991, Boston, MA, Academic Press.
National Laboratory, Roskilde, Denmark, 1991, (N. Hansen 128. G. T. HAHN and A. R. ROSENFIELD: Metall. Trans., 1975,
et al.), 603-610. 6A, 653-668.
96. S. CERESARA and P. FIORINI: Powder Metall., 1979,22, (1), 1-4. 129. T. F. KLIMOWICZ and K. S. VECCHIO: in 'Fundamental relation-
97. M. STRANGWOOD, C. A. HIPPSLEY, and J. J. LEWANDOWSKI: Scr. ships between microstructures and mechanical properties',
Metall., 1990, 24, 1483-1487. (ed. M. N. Gungor and P. K. Liaw), 255-267; 1989,
98. and J. L. HAFLEY: Metall. Trans., 1991,
I. DULTA, S. M. ALLEN, Warrendale, PA, The Metallurgical Society of AIME.
22A, 2553-2563. 130. M. MANOHARAN and J. J. LEWANDOWSKI: Acta Metall. Mater.,
99. H. J. RACK and P. K. CHAUDHURY: in 'Light-weight alloys for 1990, 38, 489-496.
aerospace applications', (ed. E. W. Lee et al.), 345-357; 1989, . 131. T. B. cox and J. R. LOW: NASA Tech. Rep. No.3, NGR
Warrendale, PA, The Minerals, Metals and Materials Society. 38-087-003, 1972.
100. R. J. ARSENAULT and s. B. wu: Mater. Sci. Eng., 1987,96, 77-88. 132. J. J. STEPHENS, J. P. LUCAS, and F. M. HOSKINS: Scr. Metall.,
101. J. C. HALPIN: 'Primer on composite materials: Analysis', Rev. 1988,22, 1307-1312.
edn, 130; 1984, Lancaster, PA, Technomic Publ. 133. G. T. KRAFFT: Appl. Mater. Res., 1964,1,88-101.
102. M. TAYA and R. J. ARSENAULT: Scr. Metall., 1987, 21, 349-354. 134. E. E. TOPUR: MSc thesis, Lehigh University, Bethlehem, PA,
103. D. L. McDANELS: Metall. Trans., 1985, 16A, 1105-1115. 1990.
104. F. J. HUMPHREYS: in 'Mechanical and physical behaviour of 135. P. F. THOMASON: Acta Metall. Mater., 1992,40, 241-249.
metallic and ceramic composites', 9th Rise Int. Symp. on 136. B. P. SOMERDAY, Y. LENG, F. E. WAWNER, and R. P. GANGLOFF:
Metallurgy and Materials Science, Rise National Laboratory, in 'Advanced metal matrix composites for elevated temper-
Roskilde, Denmark, (ed. S. I. Anderson et al.), 51-74. ature', (ed. M. N. Gungor), 167-182; 1992, Materials Park,
105. F. J. HUMPHREYS, H. BASU, and M. R. DJAZEB: in 'Metal matrix OH, ASM International.
composites - processing, microstructure and properties', 12th 137. B. ROEBUCK and J. D. LORD: Mater. Sci. Technol., 1990, 6,
Rise Int. Symp. on Metallurgy and Materials Science, Rise 1199-1209.
National Laboratory, Roskilde, Denmark, 1991, (N. Hansen 138. w. A. LOGSDON and P. K. LIAW: Eng. Fract. Mech., 1986,
et al.), 51-66. 24, 737-751.
106. T. 'CHRISTMAN, A. NEEDLEMAN, and s. SURESH: Acta Metall., 139. D. L. DAVIDSON: Eng. Fract. Mech., 1989, 33, 965-977.
198~3~ 3029-305Q 140. D. L. DAVIDSON: Metall. Trans., 1991, 22A, 97-112.
107. v. TVERGAARD: in 'Metal matrix composites - processing, 141. J. K. SHANG and R. O. RITCHIE: in 'Metal matrix composites:
microstructure and properties'. 12th Rise Int. Symp. on Mechanisms and properties', (ed. R. K. Everett and
Metallurgy and Materials Science, Rise National Laboratory, R. J. Arsenault), 255-285; 1991, Boston, MA, Academic Press.
Roskilde, Denmark, 1991, (N. Hansen et al.), 173-188. 142. J. J. BONNEN, J. E. ALLISON, and J. E. JONES: Metall. Trans.,
108. A. LEVY and J. M. PAPAZIAN: in 'Metal matrix composites - 1991, 22A, 1007-1019.
processing, microstructure and properties', 12th Rise Int. 143. D. R. WILLIAMS and M. E. FINE: in Proc. 5th Int. Conf. on
Symp. on Metallurgy and Materials Science, Rise National 'Composite materials - ICCM-V', (ed. W. C. Harrigan, Jr
Laboratory, Roskilde, Denmark, 1991, (N. Hansen et al.), et al.), 639-670; 1985, Warrendale, PA, The Metallurgical
475-482. Society of AIME.
109. M. R. PIGGOTT: 'Load bearing fibre composites'; 1980, Oxford, 144. c. P. YOU and J. E. ALLISON: in Proc. 7th Int. Conf. on
Pergamon Press. 'Fracture', 3005-3012; 1989, Pergamon Press.
110. v. C. NARDONE and K. M. PREWO: Scr. Metall., 1986,20,43-48. 145. J. K. SHANG, W. YU, and R. O. RITCHIE: Mater. Sci Eng. A,
111. R. J. ARSENAULT and N. SHI: Mater. Sci. Eng., 1986,81, 175-187. 1988, 102, 181-192.
112. M. TAYA, K. E. LULAY, and D. J. LLOYD: Acta Metall. Mater., 146. J. P. BAILON, J. MASOUNAVE, and c. BATHIAS: Scr. Metall., 1977,
1991, 39, 73-87. 11, 1101-1106.
113. P. J. WITHERS, W. M. STOBBS, and o. B. PEDERSEN: Acta Metall., 147. T. G. NIEH, K. XIA, and T. G. LANGDON: J. Eng. Mater. Technol.
1989,37,3061-3084. (Trans. ASME), 1988, 110, 77-82.
114. M. TAYA, K. E. LULAY, and D. J. LLOYD: in 'Morris E. Fine 148. K. XIA, T. G. NIEH, J. WADSWORTH, and T. G. LANGDON: in
Symp.', (ed. P. K. Liaw et al.), 153-164; 1990, Warrendale, 'Fundamental relationship between microstructure and
PA, The Metallurgical Society of AIME. mechanical properties of metal matrix composites', (ed.
115. M. S. ZEDALIS, P. S. GILMAN, and s. K. DAS: in 'High performance M. N. Gungor and P. K. Liaw), 543-556; 1989, Warrendale,
composites for the 1990s', (ed. S. K. Das et al.), 61-81; 1990, PA, The Metallurgical Society of AIME.
Warrendale, PA, The Metallurgical Society of AIME. 149. S. PICKARD and B. DERBY: in 'Developments in the science and
116. R. DWlVEDI, G. ALTLAND, P. BARRON-ANATOLIN, J. LEIGHTON, technology of composite materials (ECCM3)', (ed.
and F. G. HAMEL: in 'Int. off highway and powerplant cong.', A. A. Bunsell et al.), 199-204; 1989, Amsterdam, The
Milwaukee, WI, 1991, SAE Technical Paper 911770. Netherlands, Elsevier.
117. Y. FLOM: PhD thesis, University of Maryland, 1987. 150. w. C. OLIVER and w. D. NIX: Acta Metall., 1982,30, 1335-1347.
118. Y. FLOM and R. J. ARSENAULT: in Proc. 6th Int. Conf. on 151. K.-T. PARK, E. J. LAVERNIA, and F. A. MOHAMED: Acta Metall.
'Composite materials - ICCM-VI', (ed. F. L. Matthews et al.), Mater., 1990, 38, 2149-2159.
2.189-2.208; 1987, London, Elsevier Applied Science. 152. A. B. PANDEY, R. S. MISRA, and Y. R. MAHAJAN: Acta Metall.
119. D. J. LLOYD: Acta Metall. Mater., 1991,39,59-71. Mater., 1992,40, 2045-2052.
120. D. J. LLOYD, P. L. MORRIS, and E. NEHME: in 'Fabrication of 153. o. D. SHERBY, R. H. KLUNDT, and A. K. MILLER: Metall. Trans.,
particulates reinforced metal composites', (ed. J. Masounave 1977, 8A, 843-850.
and F. G. Hamel), 235-243; 1990, Materials Park, OH, ASM 154. c. T. LANE: in 'Fabrication of particulates reinforced metal
International. composites', (ed. J. Masounave and F. G. Hamel), 195-201;
121. D. C. DRUCKER: in 'High strength materials', (ed. V. Zackay), 1990, Materials Park, OH, ASM International.
795-830; 1965, New York, Wiley. 155. 'Machining guidelines', 1992, Duralcan Composites, San
122. G. BAO, J. W. HUTCHINSON, and R. M. McMEEKING: Acta Metall. Diego, CA.
Mater., 1991,39, 1871-1882. 156. S. BRUSETHANG, O. REISO, and w. RUCH: in 'Fabrication of
123. P. F. THOMASON: 'Ductile fracture of metals'; 1990, Oxford, particulates reinforced metal composites', (ed. J. Masounave
Pergamon Press. and F. G. Hamel), 173-179; 1990, Materials Park, OH, ASM
124. A. S. ARGON, J. 1M, and R. SAFOGLU: Metall. Trans., 1975, International.
6A, 825-837. 157. P. W. JEFFREY and s. HOLCOMB: in 'Fabrication of particulates
125. T. MOCHIDA, M. TAYA, and D. J. LLOYD: J. Jpn Inst. Met., 1991, reinforced metal composites'. (ed. J. Masounave and
32, (10) 931. F. G. Hamel), 181-186; 1990, Materials Park, OH, ASM
126. A. MORTENSEN: in 'Fabrication of particulates reinforced metal International.
composites', (ed. J. Masounave and F. G. Hamel), 235-243; 158. s. V. PRASAD and P. K. ROHATGI: J. Met., 1987, 39, (1), 22-26.
1990, Materials Park, OH, ASM International.

Lloyd1994 PDF

Uploaded by

Copyright:

Available Formats

Lloyd1994 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lloyd1994 PDF

Uploaded by

Copyright:

Available Formats

Particle reinforced aluminium and magnesium

application as selective reinforcement in the ring land

Introduction The reinforcement

International Materials Reviews 1994 Vol. 39 No.1

density of the composite will be important, which

Table 1 Properties of SiC and AI203 "reinforcement

SiC 420-450 32 4'3x10-6 840 10-40, at 11 OOC 017

International Materials Reviews 1994 Vol. 39 No.1

dense, and can be carried out below the solidus of

International Materials Reviews 1994 Vol. 39 No.1

3 Microstructure of Comalco 6061-20 vol.-%

metal processing routes to be used, and hence mini-

International Materials Reviews 1994 Vol. 39 NO.1

PVC Exterior 65G-r--------------------,

used as a hardener with a pure aluminium matrix,

XD process Matrix-reinforcement reactivity

C + Ti + AI~TiC + Al . (2) 4AI + 3SiC~ Al4 C3 + 3Si . (3)

14 2.0 +----+---+---+--t--+----+--+---+---I---t 0.10

~oo 700 800 8 Thermodynamic stability of AI-Mg oxides in

International Materials Reviews 1994 Vol. 39 No.1

~ that using a mixed oxide, mullite, in the crystallised,

International Materials Reviews 1994 Vol. 39 No.1

15 strain rate sensitivity m in the strain rate equation

International Materials Reviews 1994 Vol. 39 No.1

reinforcement with an aspect ratio below that necess-

Table 2 Typical properties of some commercially available metal matrix composites

* Composite designation: matrix/reinforcement/volume fraction of particles.

International Materials Reviews 1994 Vol. 39 No.1

The predominant strengthening factor is the volume

closely spaced elastic particles would be highly con-

~ 0.20,---------------------. 4. Control of particle shape .

International Materials Reviews 1994 Vol. 39 No.1

microstructural features of the composite. As noted Hahn and Rosenfield equation

International Materials Reviews 1994 Vol. 39 No.1

Z 240- D 6061/15A1203,T6 10-5

literature are acknowledged to be invalid according

particles, have high wear resistance and limited duc-

as sawing, can require modification of the conven- AI/SiC

studies have been carried out, and viable commercial ~~

International Materials Reviews 1994 Vol. 39 No.1

International Materials Reviews 1994 Vol. 39 No.1

International Materials Reviews 1994 Vol. 39 No.1

You might also like