Sie sind auf Seite 1von 28

Solidification processing of metal matrix

composites
A. Mortensen and I. Jin

Fundamentals of materials processes wherein a


reinforcing phase is combined with bulk molten
metal are reviewed. Capillary phenomena which
govern the incorporation of the reinforcement into
the matrix are discussed first, with focus on
thermodynamics, mechanics, and chemistry of
wetting. Engineering strategies that have been
adopted to increase reinforcement wettability are
also addressed. Relevant transport phenomena are
covered next, with focus on the infiltration process
and the rheology of composite slurries. The last
section of this review is concerned with
solidification of the matrix after it has been
combined with the reinforcement, including
nucleation and growth of the matrix, and particle
pushing by the solidifying metal. This review
places emphasis on the physical phenomena that
govern the processing and the microstructural
development of cast metal matrix composites.

IMR/235

1992 The Institute of Materials and ASM International.


Professor Mortensen is in the Department of Materials
Science, Massachusetts Institute of Technology, Cambridge, MA, USA, and Dr Jin is with the Kingston
Research and Development Centre, Alcan International
Ltd, Kingston, Ont., Canada.

Introduction
More than 25 years have elapsed since Kelly and
Davies,! and Cratchley2 summarised their and other
pioneering efforts on metal matrix composite materials in this journal. The concept and advantages of
reinforcing a metal are now solidly established, but
it is only over the past few years that reinforced
metals have truly been considered serious contenders
in the arena of engineering materials. The relatively
late blooming of these materials has several causes.
One is certainly that significant advances toward the
production of affordable reinforcing materials have
been made over the past decade. Of at least equal
importance, however, is the stronger emphasis that
has more recently been placed on the processing of
these materials: it is now clear to the materials
community that if metal matrix composites (MMCs)
are to gain industrial viability, their processing must
be rendered economical and reliable, and must be
tailored to produce microstructures that will optimise
critical properties of the composite. The complexity
of metal matrix composite processing, and its critical
incidence on both cost and properties of the resulting
materials, has thus led to a recent surge of scientific
interest in its underlying principles.
A large and. increasing proportion
of MMCs is
presently produced by solidification processing, in
which the metal matrix is molten before it is combined

with the particles, whiskers, or fibres that are to serve


as its reinforcing phase in the final composite material.
The appeal of this route compared with solid state
processes (such as diffusion bonding or powder metallurgy) results from the ready availability of bulk
molten metal, and the relative simplicity of blending
two phases when one is liquid.
This article is a review of the fundamental principles
of solidification processing of MMCs. It is focused
on those processes where the molten matrix is in bulk
form when it is combined with the reinforcement.
Thus processes in which the metal is injected into a
preform of the reinforcing phase, as practiced for
example in the squeeze casting process for piston
production by Toyota 3 and processes in which the
reinforcement is added to the metal in fine divided
form, i.e. as loose particles, whiskers, or -fibres, of
which the Duralcan process4,5 is an example, are
covered. Rapid solidification and spray casting processes (such as plasma spraying, or the Osprey process6), despite a recent increase in their importance
and potential, are omitted, largely for lack of space.
In situ composites, obtained via directional solidification of eutectic alloys, are also excluded from
consideration: a composite is defined here as a material that results from the artificial combination of two
phases to obtain a material that can not result from
conventional alloying.
For brevity, it is assumed that the reader has some
familiarity with the nature and general characteristics
of metal matrix composite materials and with their
processing. An excellent introductory
reference on
these topics can be found in the recent monograph
by Chawla.7 Technical aspects of solidification processes for metal matrix composite fabrication, which
ha ve been covered in several preceding reviews, e.g.
Refs. 8-21, are not reviewed systematically.
This article is divided into three main sections,
concerned with the initial, the intermediate, and the
final steps of metal matrix composite solidification
processes, respectively. The first section thus addresses
wetting of the reinforcement
by the metal, which
governs that step of the process in which the liquid
matrix and the reinforcement
are first combined.
Fluid flow, heat transfer, and solidification phenomena that take place in the composite material
before it is fully solidified are covered in the second
section. In the third main section, final solidification
of MMCs, the last step in forming the material and
its microstructure, is reviewed. The issue of interface
reactivity is omitted, despite its importance in metal
matrix composite solidification processing. This is
because it is too system specific an issue to be treated
otherwise than with a long list of pairs of materials,
their reaction products, and their kinetics. Such a list
is excluded here for lack of space.

International

Materials -Reviews

1992

Vol. 37

NO.3

101

102

Mortensen and Jin

Solidification

processing of metal matrix composites

taneous and work is required, supplied in infiltration


processes via pressure applied on to the molten metal.
A simple calculation22-25 gives the minimum pressure
required

000000000
000000000
000000000
000000000
000000000
000000000
000000000
000000000
000000000
000000000
000000000

composite

1111
1

Infiltration of fibre preform by molten metal.


For quasistatic infiltration,
applied pressure
must oppose capillary
pressure drop at
infiltration front, and constitutes therefore a
measure of preform wettability

Che,mistry and mechanics of wetting


Thermodynamics

and mechanics of wetting

For the producer of a MMC, the manifestation of


poor wettability of a reinforcement by a metal is a
repulsive force exerted between the liquid metal and
the fibres or particles one seeks to combine. A
sufficient - but not necessary - explanation for such
poor wettability is that an overall increase in surface
energy results from combining the two phases.
Wetting in solidification processing of MM Cs can
be quantified by measurement of the minimum pressure AP y that must be applied on the metal, or,
equivalently, of the force that must be applied on to
the reinforcement, to combine the two phases fully at
vanishingly small velocity. In the infiltration of a
preform, illustrated in Fig. 1, when AP y is negative
the metal is said to 'wet' the reinforcement which it
can then infiltrate or engulf spontaneously, as does
molten copper with a bundle of tungsten fibres. This
is unfortunately seldom the case in metal matrix
composite fabrication: wetting is generally unfavourable. Positive pressure, and hence external work, must
then be provided to create the composite.
If for now infiltration is focused on and it is
assumed that it can take place reversibly (as it would
were the metal flowing through a bundle of parallel
capillary tubes of the reinforcing phase without
reacting chemically with the latter), the work necessary for quasi static fabrication of a volume of composite containing 1 m2 of interfacial area is equal to that
of replacing the solid reinforcement/atmosphere surface, of energy CTSA, by the solid reinforcement/liquid
metal surface, of energy CTsv If CTSA > CTsv the metal
can spontaneously infiltrate and therefore wet the
preform according to basic thermodynamic analysis.
Alternatively, if CTSA < CTSL, the process cannot be sponInternational

Materials Reviews

where Sf is the surface area of interface per unit


volume of metal matrix .
Equation (1) also shows that the critical parameter
governing the wettability of a reinforcement by a
metal is the work of immersion, lti = CTSL - CTSA. lti is
related to the contact angle (J of the metal on the
ceramic (measured through the liquid metal) and to
the liquid metal surface tension CTLA by the YoungDupre equation
CTSA -

pressure

1992

Vol. 37

NO.3

(1)

CTSL = CTLA

cos(J

(2)

To achieve an improvement in wetting of the


reinforcement by a molten metal, one seeks to raise
CTSA, and/or, to lower CTSL- Lowering the metal surface
tension CTLA at constant CTSA-CTSL will reduce the wetting
angle (J in wetting systems, but will never cause a
non-wetting system with ()> 90 to become a wetting
system in which (J < 90 (Refs. 24, 26, 27).
With the proper values for lti, equation (1) gives a
lower bound for the pressure required for infiltration,
because irreversible energy losses exist in the process.
Interfacial chemical reactions may result in irreversible energy expenditure, which complicates the energy
balance involved in deriving equation (1). O"SL is now
a function of time and, hence, of position along the
composite. Heat generated or absorbed by the interfacial reaction may also influence the kinetics of wetting
and infiltration which, in turn, will influence the
apparent O"SL- Calculating AP y is thus complicated,
though equation (1) could still be used in conjunction
with a comprehensive model of the infiltration process
and knowledge of (a) the effective value of CTSL as a
function of time and temperature (keeping in mind
that in the presence of a reaction layer at the interface,
the physical meaning of 0" SL is also changed) and (b)
the influence of free energy released by the reaction.
Under conditions of limited interfacial reaction, it has
been proposed28-30 that CTSA -O"SL is given by
(3)
where (O"SA-CTsL)O
corresponds to wetting with no
reaction, ACTr represents the change in interfacial
energy resulting from the replacement of the unreacted
solid/liquid interface with at least one new interface
after reaction, and AGr is the free energy released at
the liquid/solid/atmosphere triple contact line. The
role and evaluation of this last term is very system
dependent and still a subject of some controversy.30,31
A second and more general limitation in the validity
of equation (1) stems from the fact that the mechanics
of wetting of a porous medium generally result in
irreversible energy losses, a fact that is well documented in hydrogeology or reservoir engineering.32-38 Mechanical irreversibility in wetting is
revealed by hysteresis of drainage-imbibation curves,
wherein the volume fraction liquid is plotted versus

Mortensen

and Jin

pressure applied on a liquid that is infiltrating or


dewetting a porous medium (examples are given in
Refs. 32, 33, 37, 38). Such hysteresis occurs because
energy is mechanically lost in the process of wetting
and dewetting the porous medium, even if there are
no chemical reactions between the porous medium
and the liquid. Irreversible energy losses in infiltration
result, among other reasons, whenever the invading
liquid encounters a constriction in the porous
medium, through which it jumps irreversibly into the
pore on the other side of the constriction (these jumps
bear the name of 'Haines jumps'). Another cause for
hysteresis in drainage-imbibation curves is roughness
or chemical inhomogeneity on the reinforcement surface, which cause other, more microscopic, jumps as
the invading fluid passes over the pore walls, and
lead to a difference between the macroscopically
apparent advancing and receding contact angles.39-41
An important practical consequence of mechanical
irreversibility in infiltration is the fact that a wetting
angle lower than 90 does not imply that metal will
spontaneously infiltrate a packed bed of particles or
a bundle of fibres.33,42
The pressure given by equation (1) is therefore the
lowest pressure that may possibly drive the molten
metal into the reinforcement preform. In reality, no
single wetting pressure exists because of the complex
microscopic mechanical phenomena that take place
during wetting of a porous medium. Wetting is gradual, the volume fraction liquid being a step-wise
continuous function of applied pressure, plotted in
drainage-imbibation curves.43 If parallel fibres touch,
for example, the pressure needed to infiltrate contact
lines between two fibres is calculated to be infinite,24
which agrees with observations of voids at fibre
contact points24,44-48 and implies that infinite pressure is required for perfectly full infiltration. Similarly,
asperities on the reinforcement surface may prevent
achievement of full reinforcement contact with a nonwetting liquid.24,40 'Full infiltration' therefore often
means, from a practical standpoint, infiltration of all
but perhaps a very small portion of open porosity in
the preform. Then, with AP y now defined with reference to practically full infiltration, experimental evidence22,32 and various idealised calculations of
capillary pressure49-51 indicate that equation (1) will
only underestimate that pressure relatively slightly
(by a factor of two or so) as long as the proper
contact angle and specific surface are used, and other
energy loss mechanisms such as viscous drag are
properly modelled.25
Turning now to casting processes in which the
reinforcement is added loosely to liquid metal, analysis is relatively simpler because the geometry of the
problem is fairly well defined if the interaction
between neighbouring particles in the process of being
engulfed is neglected. The problem is then that of one
isolated particle at the free surface between a liquid
and a gas, which is a relatively classical problem in
capillarity that has recently been reviewed by several
authors. 52-55 A conclusion from equilibrium configuration calculations is that for quasistatic
engulfment of a smooth particle, an energy barrier
must be overcome even if the contact angle is less
than 90, in fact as long as it is finite (8) 0). This can

Solidification

processing of metal matrix composites

103

Immersion of cube in liquid. Irrespective of


wetting conditions, as long as wetting angle
of liquid on cube is positive, the transition
from step 1 to 2 is spontaneous, while that
from step 4 to 5 is energetically unfavourable
(equation (2), with 9> 0)

be illustrated with elementary analysis of the idealised


case of engulfment of a cubic particle with one face
horizontal, Fig. 2. For the last face to be wetted, one
interface (that between the particle and the gas) must
be replaced with two interfaces (particle/liquid and
liquid/gas). From equation (2), force must then be
applied on to the reinforcement to-produce the composite if 8> O.The very last portion of the surface of
a smooth particle that is engulfed is always horizontal,
as is the top surface of the cube in Fig. 2, so the same
conclusion obtains. Because of the small dimensions
of the particles used in reinforcing metals, gravity
seldom can supply sufficient force to drag them into
the liquid metal. The usual method for producing
composites by adding discrete reinforcement particles
at the melt surface therefore makes use of motion in
the metal, which applies force on to the particles via
viscous drag along their surface to pull them into the
melt (e.g. Refs. 56, 57).
The process of particle engulfment in stirred liquid
metal was recently modelled by Ilegbusi and
Szekely,53 who added a term for viscous drag to the
general force balance proposed by previous researchers, and thus created a link between wetting and
processing parameters. An empirical drag coefficient
for flow of particles within an electromagnetically
stirred turbulent fluid was used after multiplication
by the fraction of wetted particle surface. Results for
spherical boron carbide particles dragged into electromagnetically stirred magnesium are given in Fig. 3.
These show the significant influence of both particle
size and contact angle on the fluid flow velocity
required for incorporation.
Limitations exist to such modelling work, of course,
if only because of the influence of cuspidal points
generally present on the surface of discrete reinforcement materials such as particles, flakes, or short fibres,
where the wetting angle 8 is not defined. The oxide
layer covering the molten metal is also' expected to
be thicker here than in infiltration processes, and thus
to playa significant role in wetting. This is shown in
recent work by Stefanescu et aI.,58,59 who measured
the centrifugal force required to immerse particles in
aluminium. With increasing particle size, a sudden
decrease in the force required for engulfment was
observed, taking place at the same particle radius for
International

Materials Reviews

1992

Vol. 37

NO.3

104

Mortensen and Jin

Solidification

processing of metal

matrix

180

160

composites

---------- ---

140

\
\

4
"";(/)

~ 120

\
\

(])

\
\

100

,
\

C'I

E
>!:::

..J
W

>

30

60

90

. 120

150

80

180

CONTACT ANGLE, deg

60
900

::::>

..J
I.L.

~
::::>
~
z
~

100
PARTICLE DIAMETER,

a influence

of wetting angle of liquid magnesium


dia.; b influence of particle diameter

(b)

on particle 100 ~m

two oxide particles of different density. This sudden


decrease was interpreted by the authors 58 as resulting
from the particle diameter exceeding the thickness of
the oxide skin covering the metal. Particles also have
been found to agglomerate at the metal surface before
incorporation,6o,61
which adds to the mechanical
complexity of the problem. Despite these limitations,
the results from Ref. 53 show that correlation
of
processing parameters with appropriate wetting data
is possible, and agrees with empirical knowledge that
finer particles are harder to stir into molten metal
(e.g. Refs. 62,.63).
Measurement of wettability
The conventional experimental method used for measuring wettability by a metal is the sessile drop experiment. This consists in measuring the shape and
contact angle 8 of a drop of the liquid metal matrix
at rest on a flat substrate of the ceramic reinforcement
material, or in some instances64,65 on fibres. The
sessile drop experiment enables measurement of both
O'LA and 8, and can therefore
allow calculation of
~p y via equations (1) and (2). For example, the wettability
of carbon,64-75
of SiC,52,65.70,76-82 of
B4C,51.69.70,78.83,84 and of AI20352.71,74,75,84-94 by
International

and its alloys has been measured


Materials Reviews

1992

1200

1300

Schematic
variation
of wetting,
angle (J
measured on -flat substrate with temperature
for pure aluminium on alumina for two partial
pressures of oxygen: .2:- - total pressure
",,-10-3 Pa, oxygen. partial pressure >10-6 Pa;
-total pressure 5 x 10-5 Pa, oxygen partial
pressu,re 10-15Pa, Difference between the two
curves arises from presence of oxide layer
covering metal at higher pressures, which
evaporates around 1150 K. From Refs. 89, 97

200

~m

Minimum
fluid velocity
required to engulf
spherical particle of boron carbide into molten
magnesium at vortex created by electromagnetic stirring,
based on calculations
by
lIegbusi and Szekely53

aluminium

1100
T,K

contact angle 150


contact angle 60

-----a

1000

Vol. 37

and
NO.3

found to be poor below about 1223 K. Reviews of


such wetting angle data have been published by
Delannay et al.,44.83 Russell et al., 52 Nicholas,95 and
Naidich,70 this last review being particularly extensive. A general conclusion from the large number of
wetting angle measurements that have been made to
date is that'the technique is precise and reliable, and
yields data amenable to scientific analysis. Examination of sessile drop data relevant to systems of
interest for metal matrix composites reveals that these
also depend on several complicating factors:
(i) with several metals including aluminium and
tin, the presence of a layer of oxide on the metal
droplet prevents it from contacting the substrate
and therefore influences wetting detrimentally
during the exp~riment. Several observations
lead to this conclusion, including visual observation of the drop and the influence on 8 of
metal cleaning before melting, 84.87,88 the presence of abrupt wetting/non-wetting
transition
temperatures that are relatively independent of
the substrate but depend on the oxygen partial
pressure of the atmosphere, 84,87,92,96Fig. 4, and
the strong influence on wetting angle of alloying
additions,66,73,94 or temperature
changes92,96
which are known to affect the oxide covering
the metal
(ii) wetting angle data are often time dependent.
This phenomenon
is attributed
to chemical
reactions
occurring
at the metal/substrate
interface, as well as to the necessity for the
metal to break thro'ugh its surface oxide layer
before achieving intimate contact with the substrate. Examples of this effect can be found in
Refs. 29, 66, 71, 74, 75, 77, 82, 88, 90-92, 96,
98-100.

Mortensen

and Jin

A further possible complicating effect in assessing


wettability is the presence of a precursor film ahead
of the spreading metal, which is reported for metallic
systems in Ref. 101 and modelled for the case of
organic and other liquids spreading on a solid substrate,102 or of reaction between the solid and the
liquid ahead of the wetting line.29 Added complications of a more mechanical nature exist as well,
such as the well known fact that advancing and
receding contact angles differ as a consequence of
surface roughness or variations in the chemical nature
of the wetted substrate.39,40,92,102
The dependence of sessile drop experiment data on
these epiphenomena often invalidates direct transposition of sessile drop wetting angles to metal matrix
composites solidification processes, because the velocity of the three-phase contact line in these processes
is much higher than in a sessile drop experiment,
because the oxide layer covering molten aluminium
is most likely disrupted by the reinforcement as it
combines with the metal,103,104or because of microscopic roughness and chemical heterogeneity on the
surface of many reinforcements (e.g. commercially
available polycrystalline alumina fibres105).These several limitations of the sessile drop technique have
served as an impetus for the development of alternative measurement techniques for wettability, often less
precise and less amenable to detailed scientific
interpretation, but more directly pertinent to metal
matrix composite solidification processing.
Dipping experiments have been proposed to replicate more accurately the dynamic nature of wetting
in composite fabrication. In hanging plate or rod
experiments, the weight of a straight-edged solid
partly immersed in the liquid metal and the shape of
the meniscus near the solid are recorded 106to measure
both aLA and 8. Compared with the sessile drop
method, this technique is somewhat more complicated
but is equally precise, and has the advantage that
dynamic wetting conditions can be studied by moving
the object into and out of the liquid.
A different type of dipping experiment was used by
Choh and co-workers/07-111 who studied the rate of
spreading of molten aluminium and aluminium alloys
on flat discs of carbon and SiC substrates, which were
suddenly completely immersed in the molten metal.
Curves of fraction of disc surface wetted by the molten
metal as a function of time were found to have a
sigmoidal shape typical of thermally activated
nucleation and growth reactions. Their temperature
dependence was explained in terms of possible critical
stepscin the chemical interaction between metal and
substrate. These experiments present, however, the
disadvantage of using flat substrates instead of actual
particulate or fibrous reinforcements.
Experimental procedures using particles of ceramic
material to measure wettability have recently been
established. In most of these,112-117molten metal is
forced under pressure into a packed bed of the powder
held at the same temperature as the metal. The
distance travelled by the metal during a time interval
lit into the packed bed is measured for various applied
pressures, and from extrapolation of these data to
zero infiltration distance, a minimum ('threshold')
pressure for initiation of the metal movement into

Solidification

processing of metal matrix composites

105

the packed bed of particles is recorded as a function


of alloy composition, atmosphere, temperature, and
lit. This threshold pressure does not generally equal
liP y in equation (1) because (a) it is a measure of
initiation of wetting, which can be influenced by
percolation and other preform entrance effects and
(b) the infiltration distance which is recorded frequently includes regions of partially infiltrated composite. These experiments have been performed for
carbide reinforcements and matrices based on aluminium or zinc.112-118A practically important and scientifically interesting finding of these investigations is
that the infiltration distance and consequently the
threshold pressure depend on the time of pressure
application. In particular, for a given set of experimental conditions, an incubation time exists, before
which no infiltration occurs.115-117 This observation
agrees with results of Choh and co-workersl07-111
on the spreading of metal on dipped flat substrates.
Direct measurement of the capillary pressure drop
liP y at the infiltration front, given by equation (1) for
reversible infiltration, requires that data be collected
while the metal is being combined with a preform of
the reinforcing phase, to replicate the dynamic and
mechanical characteristics of wetting in infiltration
processes. Such experimental data have been generated in the wake of recent studies of infiltration
processing. 104,119,120The measured capillary pressures were shown to be interpretable as closely
approximating liP y of equation (1),25and could therefore be compared with wetting angle data via equation
(2). Such comparison yielded values of the apparent
wetting angle that were much closer to sessile drop
wetting angles in high vacuum than to angles measured in air or in lower vacuum. This, in turn, may
indicate that the oxide layer which usually covers
metals such as aluminium during static sessile drop
experiments and in measurement of threshold pressure is washed away by the reinforcement phase during
pressure infiltration, as was first proposed by Cappleman et a1.103 This interpretation would also agree
with observations of a fine layer of alumina along the
fibre/matrix interface in some cast aluminium matrix
composites.121,122These experiments accounted for
the effect of differing metal and reinforcement initial
temperatures, which causes the wetting process to
take place concomitant with intense heat flow
from one phase to another, and possibly matrix
solidification.
Measurement of liP y from infiltration experiments
is only possible when full wetting in partly infiltrated
composites can be unambiguously identified and measured during infiltration, as was the case in Ref. 104.
This is generally not the case, because the geometrical
complexity of the reinforcement preform causes wetting to take place gradually. To account for this,
wetting in infiltration has been recently characterised
by measuring drainage-imbibation curves which give
the volume fraction metal as a function of applied
pressure, as is general practice in hydrogeology or
reservoir engineering.43,123,124As seen in the section
'Infiltration of preforms' below, provided there is no
influence of infiltration velocity on these curves, these
drainage-imbibation curves are useful in modelling
infiltration.
International

Materials Reviews

1992

Vol. 37

NO.3

106

Mortensen and Jin

Solidification

processing of metal matrix composites

aluminium,14,15,131,140-153oxide coatings for


Engineering approaches
carbon fibres in magnesium 154,155or silicon
On the basis of wetting angle data, spontaneous
carbide in aluminium 118,129
.
wetting of a reinforcement by molten metal is found
to be promoted by. some degree of reactivity of the
(ii) coatings that are designed to react with the
metal with the substrate, 11-13, 70 or by a reduction in
oxide layer covering molten aluminium. These
include loose coatings of K2ZrF 6,13,156-165
the tenacity of the oxide layer on metals prone to
oxidation such as aluminium.13 Based on these obserwhich is a known fluxing agent for aluminium.
vations, much work has been done to design chemical
It has been argued that most coatings that
means of enhancing the generally poor wetting of
cause reaction induced infiltration by aluminreinforcements by metals. A majority of this work
ium with no applied pressure below 900C do
can be classified into three broad categories, namely
so mainly because they disrupt the oxide layer
(a) reinforcement pretreatment, (b) alloying modifion the metal.13,87Also, since it has been docucations of the matrix, and (c) reinforcement coating.
mented that aluminium oxide is present at the
The fibre surface energy O"SA can be raised by
interface of carbon fibres treated by the TiB
changing the chemical nature of the atmosphere
process in aluminium, it can be argued that this
before infiltration. Results have been published on
coating technique also acts by oxygen gettering
the influence of heat treatment of alumina, silicon
or formation of a spinel with the oxide layer
carbide, and graphite particles on their wettability,
on the metal,87 Fig. 7 of Ref. 166. Similarly,
which improved their ease of incorporation into aluMgO coatings that have been used to increase
minium melts. This was attributed to desorption of
wettability of alumina particles by aluminium
gaseous species from the reinforcement surface during
may in fact be active via the oxide on the
the heat treatment. 12,57,125-129
metal. 167
Alloying additions to the matrix have been shown
The dividing line that separates these two categories
to affect the wetting angle on, and ease of incorporof alloying additions or coatings is thus not clear.
ation of, many reinforcements. Effective additions fall
There is nevertheless an established perception that
in two categories:
there are at least two strategies for promoting wet(i) additions that promote reactions between the
tability of reinforcements by metals, particularly alureinforcement and the matrix: Li in Al for
minium: to obtain wetting, one can modify the fibre
alumina fibres,83,105,130,131Li for wetting of
or the matrix with something that will (a) react with
SiC by AI,83,132,133
or Mg,134Ti in AI-Sn alloys
the metal or (b) disrupt its outer oxide layer.
(but not pure AI) for wetting of SiC;133carbide
Many other methods have been developed to
formers in Al for wetting of carbon fibres;135,136 improve wetting, which do not fall into these categorSi in Al for wetting of carbon particles.126 Their
ies, either by nature or for lack of documentation
efficiency agrees rather well with wetting angle
and/ or of understanding of the actual mechanism that
data, which indicate that, in general, alloying
promotes wetting. These include (a) various proc.esses
additions that promote reactivity between the
involving fibre treatment by molten sodium for infilmetal and the substrate lower the wetting
tration of carbon or alumina fibres by aluminium or
angle68,70,83,95,137
magnesium,152,168-170(b) the TiB process involving
(ii) additions to aluminium that do not promote
CVD deposition of Ti-B mixtures on carbon fibres
reactions with the reinforcement, but modify
before infiltration in oxygen free atmospheres by
aluminium or magnesium,171-175 (c) pretreatment of
the characteristics of the oxide layer on the
SiC by dehydrated sodium tetraborate for infiltration
metal surface: Mg in Al with most reinforcements 13,54,60,62-64,73,94,101,113,118,133,138,139
or
by molten aluminium, 176(d) pretreatment of carbon
Li in' AI13,105,131-133(with alumina or oxide
by tetraisopropyltitanate for infiltration by molten
aluminium or magnesium,177 (e) pretreatment of B4C
covered SiC reinforcements, these two categorby one of various alcohols or other organic solvents
ies obviously overlap). Again, their effect correfor infiltration by molten aluminum at elevated temlates well with wetting angle data, in the sense
peratures,178 (f) dispersion of solid magnesium nitride
that the transition temperature from wetting to
between carbon fibres for infiltration by magnon-wetting, as well as the wetting angle, are
nesium,179(g) the Lanxide process, reportedly using
decreased by these additions, e.g. Li, Mg, or Ca
in aluminium 69,71,89all elements known to
magnesium alloy additives, nitrogen containing,
oxygen free atmospheres and non-disclosed temperaffect the oxide covering aluminium.66
atures for infiltration by aluminium of several
These two practical methods either induce reactions
reinforcements,180-184(h) the use of nitrogen in aiding
at the fibre/matrix interface, or modify the oxide layer
wettability of SiC and Al203 particles by magthat usually coats molten aluminium. Reactions
nesium,185,186and (i) sodium tetraborate additions to
between the reinforcement and the matrix are generaid wetting of alumina particles by AI-7Si-O-1 Mg
ally undesirable because they degrade the reinforcealloys*.63 Last, the elegant experiments of Nogi
ment strength. A different approach based on the
et al.,187 who improved the wettability of zirconia by
same principles is to coat the fibre or particle surface.
liquid copper and iron using a dc voltage applied
Such coatings have been extensively used, and can
across the metal/ceramic interface, and interpreted
also be classified broadly into the two categories used
for alloying additions:
this effect as resulting from the dissolution of oxygen
at the zirconia/metal interface.
(i) coatings that are designed to react with the
matrix. These are numerous, and include metallic coatings for various reinforcements in
* All compositions are in weight per cent unless otherwise stated.
International

Materials Reviews

1992

Vol. 37

NO.3

Mortensen

and Jin

Improvements in the wetting of reinforcements


during infiltration or incorporation of particles into
a melt have also been obtained by mechanical means.
As seen above, with parallel fibres, infinite pressure
is theoretically required to infiltrate the line of fibreto-fibre contact. Calculations of capillary forces
between two parallel fibres188,189show that the pressure required to infiltrate fibre contact lines decreases
significantly with even a small separation between
two parallel fibres, and that forces between fibres
during infiltration are large, causing them to cluster
during infiltration. These problems were solved with
'hybrid' fibre preforms, first proposed by Towata and
co-workers, in which parallel fibres are individually
held apart during infiltration by small particles or
whiskers.46,190-199Other examples of improvement
of wetting by mechanical means include the use of
semisolid metal of increased apparent viscosity in the
Compocasting process,200-210the agitation of parallel
fibre bundles using alternative current in a magnetic
field to ease their permeation in gravity driven infiltration211 and vibration induced wetting improvements of alumina fibre preforms by aluminium.212,213

Solidification

processing of metal matrix composites

107

and the positive ions, and have been quite successful


at predicting from first principles the surface tension
and electronic properties of several metals.214-219On
a more macroscopic level, several correlations have
also been experimentally found and modelled, relating
the surface tension of pure liquid metals with their
heat of vaporisation and other macroscopic properties
of the metal. 83,220-226
Most metal surfaces are usually modified by the
presence of adsorbed atoms from the atmosphere or
from within. The effect of such adsorbed elements on
the surface tension of the metal has been modelled
from thermodynamic considerations, ranging from
Gibbs isotherm treatment to more complex treatment, 83,95,217
,219,222,223,226-228
and it is well known
that even minute quantities of alloying elements or
very small partial pressures in the vapour phase can
modify the surface composition, morphology, and
energetics to a considerable extent. Furthermore, the
level of metal surface contamination may depend on
the process, possibly varying for aluminium between
a relatively clean surface (at elevated temperatures in
vacuum or because of skimming in preform pressure
infiltration, for example), to a surface covered with
an oxide crust many atomic layers thick.

Chemistry of wetting

On a more microscopic scale, wetting can be examined


in terms of atomic bonding by comparing the initial
state (one or two free surfaces) with the final state
(the interface). The results may allow one to design
novel strategies for improved wettability, ultimately
optimising mechanical properties of the final cast
material as well, because an understanding of interfacial strength can be gained from knowledge of
interface chemistry.
In the solidification processing of MMCs, one
wishes to create a bond of controlled strength between
a liquid metal and a generally non-metallic reinforcement. The kinetics of this bond creation process will
depend on the initial structure of each surface present,
on the mutual attraction exerted by the two surfaces
as they approach one another, and on the structure
of the resulting final interface. Most relevant work is
theoretical and, with the advent of advanced characterisation techniques such as high resolution microscopy, microstructural. A large fraction of present
research seeks to improve knowledge of initial and
final surfaces and interfaces, in part because of the
somewhat static nature of sessile drop experiments
(which form the bulk of available wetting data), and
of the considerable experimental difficulty involved
in the experimental study of dynamic wetting.
Surface of metals

The origin of the surface tension of a material lies in


the disruption of the bonding state for the atoms at
or near to the surface or interface. In the case of a
metal in contact with vacuum, this manifests itself as
a local disruption of the electron gas density near
that surface. An electric double layer, formed by a
positively charged background and a negatively
charged electron cloud that extends further into space,
results. Several authors have proposed calculations
of the resulting surface energy using various degrees
of sophistication in their description of the electrons

Reinforcement surface

The surface structure of oxides is of interest for two


reasons: because several reinforcements are oxides, or
are coated with an oxide (such as Si02 for SiC56,79)
and because the metal surface is often covered by its
oxide. From an energetic point of view, treatment of
ionic solid surface energies is generally based on
electrostatics, and the surface energy can be correlated
with the cohesive energy of the solid.217,219,225,229,230
From a structural point of view, it is noted that ionic
surfaces can be electrically charged.229,231A structural
model of oxide surfaces by Weyl, in which the cations
at the surface are screened by the anions because of
the much higher polarisability of the anions, is relatively well accepted.7o,83,232-237According to this
model, the surface of an oxide - and a fortiori that
of many (oxidised) metals in air - consists predominantly of highly polarised oxygen atoms.
The surface of clean graphitic carbon is complex
and depends on the orientation of the surface with
respect to the basal planes. The surface energy of
diamond is roughly one-half of its cohesive energy,225
and differs significantly from that of graphite.238 The
orientation of basal planes at the surface of carbon
fibres is known to vary significantly from fibre to
fibre, which results in variations in fibre chemistry
(Refs. 239-241 contain recent reviews). The surface of
carbon fibres is very active and is thus heavily contaminated with adsorbed species, predominantly
oxygen containing atomic groups. It also is often
modified for improved wetting and bonding with
polymer matrices,242,243because these are used in
most applications of carbon fibres.
The surface of carbides or nitrides varies with the
character of the bond (covalent or part metallic), and
is for most practical purposes usually that of an oxide
of one or several of their constituents (for example,
the surface of SiC reinforcements generally comprises
a layer of Si02).
International

Materials Reviews

1992

Vol. 37

NO.3

108

Mortensen and Jin

Solidification

processing of metal matrix composites

Interface chemistry

Depending on the materials and the degree of bonding


achieved, the interface can be one of two kinds:

be rigorously attained until complete transformation


of at least one of the two initial phases.
Models for such an interface vary in nature. Generally, it is described as sharp, i.e. as consisting of a

Reversible bond only, generally physical There is no


chemical interaction between the two materials on
either side of the interface. There is, nonetheless, an
attractive force binding the two, due to Van der Waals
forces. These are most often treated as dispersion
forces. This 'physical bond' is established in a reversible manner, and is generally the only bonding mechanism operating in wetting by organic solvents,
aqueous solvents, or polymers.83,244,245Theoretical
treatment of the physical bond is relatively well
established, and its energy can be calculated by use
of the London formula or other more complete
treatments.70,83,245-250The resulting bond energy is
relatively low and is temperature independent. 70This
type of bond is reversible, since the physical act of
separating the two surfaces after contact can be made
without significant additional energy losses, bond
strengths being low. Reversibility of bonding at the
interface has indeed been observed in polymer matrix
composites.242 Other types of reversible bonds of
higher bond energy have been proposed: the hydrogen
bond;251 or electrostatic attraction due to image
charges across the interface.236,238,250With such
reversible bonding, the science of wetting and that of
adhesion are closely related, so that additional background can be gained from several reviews pertaining
to adhesion.251-254

phases. Most models describe bonding across the


interface in thermodynamic terms to predict interface
energy.70,97,227,232,237,268-273
Other models are based
on quantum mechanical calculations,274-276 or on an
extension of the 'jellium' mode1.277Yet other models
analyse the bond qualitatively based on the bond
within carbide reinforcements,78,98,257,273or estimate
electrostatic forces resulting from electron exchange
between the two solids174,253,254(this contribution
was estimated by Krupp253 to be negligible; however,
electrostatic attraction pressures have been found
experimentally across the interface in composites of
aluminium reinforced with carbon fibres278).Microscopic investigations of interfaces are now possible
with advanced high resolution transmission electron
microscopy, and recent results are reviewed by Riihle
and Evans.279 These include recent investigations of
metal/oxide interfaces, for which it was shown by
Mader280 that oxides produced by internal oxidation
display a layer of oxygen atoms along their low
energy facets with the metal matrix, in agreement
with previous work reviewed by Mader in that same
reference. More detailed reviews of metal/ceramic
interface structure and chemistry can be found in
Refs. 70, 83,279,281.

Chemical bonds On the basis of sessile drop experiments and experience in metal matrix composites
processing, it has become clear that in interfaces
separating a metal from another material, a stronger
chemical bond can be established, resulting from
chemical interaction across the interface. Such bonds
generate works of adhesion (by definition, the energy
liberated on their formation from two free surfaces)
that are temperature dependent, are typically an order
of magnitude higher than for physical bonds and
therefore govern interfacial strength. In practical
terms, their importance translates into the observation
noted above that one of the two methods of improving
wettability in metal matrix composites solidification
processing is to 'add something to the fibre or the
matrix that will promote reactions between the two'.
These are irreversible bonds in the sense that once
established, they cannot be broken without also damaging adjoining material on either side of the interface.
Evidence of the establishment of a chemical bond
after wetting of ceramics by metals has been found
in several studies of wettability by metals using the
sessile drop technique, 68,78,98,99,137,232,255-261
or in
studies of adhesion of metals to ceramics,262-266and
is reviewed by several authors. 52,70,83,98The bond is
no longer constrained to remain in a surface, but may
be several atomic layers thick, or may even extend
across an interfacial region visible under an optical
microscope. The connection between wetting and
adhesion becomes more tenuous because the interface
cannot in practice be undone once it has formed (e.g.
Ref. 267). A theoretical treatment of the interface is
also more difficult for this reason. For example,
thermodynamic equilibrium will in many cases not

Most research on the structure of surfaces and


interfaces is for relatively pure metals and reinforcements. In practice, the metal is generally impure or
alloyed, and the reinforcement surface may be oxidised or otherwise contaminated. When the metal is
alloyed, both its surface tension and the metalj
reinforcement interface energy are affected, especially
if alloying additions tend to segregate at surfaces.
Naidich 70 proposed that oxygen dissolved in metals
can form ionocovalent metal-oxygen clusters which
segregate at ceramic/metal interfaces by coulombian
attraction. This mechanism explains the strong
reductions in wetting angle that result from oxygen
dissolution
in
molten
metals
contacting
oxides.28,70,237This interpretation has recently been
confirmed by Kritsalis et al.282
The influence of alloying on wetting was recently
modelled by Li et al.28,227 assuming a regular solution alloy and using statistical thermodynamic calculations. Their model allows prediction of the influence
of alloying additions at low dilution, and agrees with
experimental data. 28,227,283
In many practical cases, wetting is accompanied by
chemical changes that extend beyond a single metal/
ceramic interface, such as dissolution of the solid in
the liquid, diffusion of the liquid into the solid, or
formation of novel interfacial phases. These phenomena introd uce a time dependence to the wetting
process and complicate its analysis significantly. Reactive wetting has been addressed by several
authors30,31,70,284from a thermodynamic point of
view. Of particular interest is the unsolved question
of the influence exerted by free energy released in

International

Materials Reviews

1992

Vol. 37

NO.3

monolayer of chemical bonds bridging dissimilar

Influence of alloy additions and reactive wetting

Mortensen and Jin

reactions between the liquid and the solid on wetting.


A recent review of theory and experimental data on
reactive wetting is given by Laurent. 30

Solidification

processing of metal matrix composites

Forscheimer equation297-302

f-

VP

= [,uVm(l- gs)K-1

+ BPmlv1
Transport phenomena
This section is concerned with solidification processing steps in which the metal and the reinforcement
have been combined, and the metal matrix remains
at least partly liquid. These precede the final processing step in which solidification of the metal is
completed. As in wetting, a distinction is made
between cases where the reinforcement in the composite constitutes a mechanically self-sustaining preform,
and cases where it consists of individual particles,
short fibres, or whiskers dispersed within the metal
in the final composite. In the former case, the composite is directly cast by infiltration of the preform,
generally to its final general shape. In the latter case,
the composite is generally formed by stirring the
reinforcement into molten or semisolid metal to yield
a free-flowing composite slurry. This slurry is subsequently cast into shape using processes similar to
those for unreinforced metals. These two classes of
processes have been combined,61,285-288 but they
differ substantially in the transport phenomena by
which they are governed. These are reviewed in turn
in this section, beginning with infiltration processing.

109

vsl(vI - vs)

(4)

where f is the local volumetric value of gravitational,


centrifugal, or electromagnetic body forces in ~ V, P
the liquid average pressure in ~ V, ,u the viscosity of
the liquid metal, V m the volume fraction metal, and
gs the volume fraction of the metal that is solid; Vs
and VI are, respectively, the average solid and liquid
velocities..K, the local symmetric permeability tensor
of the preform and B are functions of Vm and of the
volume fraction and morphology of solidified metal.
When the relevant Reynolds number is below a
critical value
(5)

where d is a characteristic length of the reinforcement


(for example, the fibre diameter in fibre preforms, in
which case Rec ~ 1 (Ref. 119)), the second term between
brackets can be neglected. This is most often the
case, and the Forscheimer equation then reduces to
D'Arcy's law
(6)

Infiltration

of preforms

Mechanics of infiltration

The infiltration process has been used for several


decades to produce metallic composite materials.
Early applications of the process, which were focused
on capillarity driven processing for the production of
cermets and particulate metal-metal composites, are
reviewed by Lenel.289Several analyses of infiltration
were proposed for these, in which the porous medium
was modelled as a bundle of cylindrical tubes being
infiltrated by the metal according to the HagenPoiseuille equation.289-291 This assumption, which
has also been made in recent work by several investigators,116,292-294eases conceptual visualisation of the
infiltration process. However, it has no physical justification, it introduces errors (for example, the effects
of inertial losses which arise because of tortuosity in the
porous medium are unaccounted for, or erroneously
ascribed to, turbulence) and it brings no real simplification in analysis. Limitations of straight tube models
are reviewed for example by Scheidegger295 and
Morel-Seytoux,296 and practical difficulties encountered in their use in materials processing are described
in Ref. 292.
Generally, the reinforcing phases (fibres, particles)
are small enough that the preform can be described
as a porous continuum. Theoretical analysis is then
based on consideration of a small volume element
~ V, which contains several pores and reinforcement
elements such as fibres, particles, etc. Within ~ V,
temperature, matrix composition, and fraction solid
are assumed to be essentially uniform, and velocities
of the fluid and solid phases are averaged. Flow of
the liquid metal in ~ V is generally governed by the

which has been used in most recent analyses of


infiltration processing.50,104,119,120,303-312
There are,
however, infiltration processes in which Rec is greater
than one.313
Assuming for simplicity that the densities of liquid
and solid metal are equal and constant, continuity of
matter dictates

oltf
at
= - V(ltfv

(7)

o(V
gs-at=

(8)

s)

m)

-V(Vmgsvs)

(9)

where ltf, Jt;" Vm are, respectively, the local volume


fraction of reinforcement, pores, and metal, such that
ltf + Jt;, + Vm = 1.
Drag due to flow of gas initially present in the
preform is generally ignored because of the much
lower viscosity of the gas. Provided wetting is not
significantly influenced by the velocity of infiltration,
Vm can be considered a function of the difference
between the pressure P in the metal and that of the
atmosphere P g (if any) in the pores within ~ V
.

(10)

where the function F describes the first drainageimbibition curve of the preform, as described in the
previous section. If back pressure of unvented gas
builds up, P g may be significant.188~189,314
International Materials Reviews 1992

Vol. 37

NO.3

110

Mortensen and Jin

Solidification

processing of metal matrix composites

Immiscible flow of several liquids in porous media


has been studied in several branches of engineering
science, including reservoir engineering and hydrogeology.34,38,296,315,316
In the particular case of uni-

are open channels left between the reinforcement and


the solid metal.
Modelling heat transfer and matrix solidification
during infiltration is simplified by the fact that heat

directional infiltration driven by a constant pressure

exchange between the reinforcement and the metal is

difference APT between the preform entrance and the


gas phase, if Re < Rec and the body forces can be
neglected compared to the pressure gradient in equation (6), distance x and time t can be combined in
Darcy's law using the Boltzmann transformation. 34,317,318 The system of partial differential
equations is then transformed to one of ordinary
differential equations, which simplifies considerably
the mathematics describing fluid flow.
A simplified treatment of fluid flow during infiltration can be made by assuming that the infiltration
front is sharp, or, in other words, that F(P-Pg)=
(1- Jtf )H(P - APy), where H(x) is the Heaviside function (e.g. Refs. 104, 119,319). APy is then an apparent
capillary pressure drop, given approximately by equation (1) if the infiltration front is relatively flat. When
there is no preform compression, the pressure distribution in the composite during infiltration obeys the
Laplace equation at all times if the permeability is
constant.
If compression of the preform occurs during infiltration, Jtf, F(P-Pg), and the permeability K in AV
are functions of the previous stress-strain history of
the preform in AV. The rheology of wet porous media
has been a subject of much research in biomechanics
and soil mechanics.320-329 This work has led to the
definition of an effective stress, being the component
of the local stress tensor that is responsible for
deformation of the porous medium in AV. When
inertial effects are negligible, and the solid phase
making the porous preform is very rigid compared
to the preform itself, the effective stress equals the
total stress minus the pressure in the liquid P.
Preform deformation has been observed in infiltration processing of MMCs.17,330-335 Clyne and
Mason have analysed dry fibre preform deformation303 and the infiltration of deformable preforms is
currently being investigated at MIT.336

generally very rapid, so that temperature can be


assumed to be uniform within A V. This is because
the reinforcement and the matrix are intimately
bonded with one another along at least a portion of
their interface, and because the scale of the reinforcement is sufficiently small for the time for equalisation
of temperature within AV to be very short (this is
ascertained by estimating te = d2/rt.r where d is
reinforcement diameter and rt.c is thermal diffusivity
of the reinforcement: with d = 100 Jlm, characteristic
of the largest reinforcements, and rt.c= 10-6 m2 s-1,
characteristic of a poorly conducting ceramic, te is at
most of the order of 1 ms).
Heat transfer within AV is governed by conduction
and convection, as well as by local changes in fraction
solid metal gs

Thermal and solidification

effects

In most cases, infiltration is not isothermal. Dies are


often significantly below the metal liquidus or melting
point, to prevent them from sticking to the metal, to
avoid melt leakage at die parting lines, and to solidify
the metal in a short time. The preform is also frequently at a temperature lower than the metal liquidus
or melting point, either because of cooling in contact
with the die, or by design to minimise metal/reinforcement chemical interaction.305
It is. often possible to produce a composite by
infiltration when the initial preform temperature is
lower than the metal liquidus because, as Nagata and
Matsuda 337,338 and Fukunaga et ai.306,307 have
shown, the matrix solidifies when it contacts the
reinforcement, which releases heat within A V. This
serves to heat the preform at the infiltration front to
a temperature where liquid and solid matrix can
coexist, and allows flow of the metal through the
infiltrated portion of ,the composite provided there
International

Materials Reviews

1992

Vol. 37

NO.3

V(kc VT)pccc

aT + [Pmcm V (1at
m

gs)]Vt- VT

+ (Pccc Jtf + PmCmVmgs)vs - V T - PmAH

a(gs Vm)

at

(11)

where kc is the thermal conductivity tensor of the


composite, AH the latent heat of solidification of the
alloy, P the density, C the heat capacity, and the
subscripts m, f, C refer to metal, reinforcement, and
composite, respectively. The heat capacity of the gas
phase is assumed to be zero.
Mass flow in and out of A V by diffusion is neglected
by comparison to convection. The governing equation
for mass transport is then
(12)

where E is the local average matrix composition, CL


and Cs are, respectively, the composition of the liquid
and the solid phases. CL and Cs are linked by the
phase diagram and are functions of the local temperature T.
Initial conditions for the infiltration process are
imposed by process parameters: the preform, mould,
and metal initial temperatures, the initial volume
fraction of reinforcement, and the applied pressure or
infiltration velocity. Boundary conditions between the
composite system and its surroundings, or between
various regions that may appear within the composite,
are derived from physical considerations including
continuity, heat and mass conservation.119,311
General features of infiltration by a pure metal or
an alloy with concommittant matrix solidification are
illustrated in Figs. 5 and 6, respectively, from work
in Refs. 25,104,119,120,311,339-343.
Matrix solidification can be induced both by the fibres and the
die. Partially solidified metal can be remelted if there
is superheat in the incoming metal. This takes place
at a sharp remelting front, which may be unstable

Mortensen

and Jin

Solidification

processing of metal matrix composites

111

region 2

region 2
region 3

vent

infiltration
front

infiltration
front,

0:::

0:::

:::>
~ To

To

0:::

a.. Tm

2:

I-

region 3: region 1
I

I
I

I region 4

TE

region

:
~

Ii;

3l region 1

.
1

i region 4

A'

region 5

0:::

A'

DISTANCE

Schematic
illustration
of
infiltration
of
preform by pure metal, under conditions such
that solid metal forms at infiltration
front in
contact with fibres and along mould wall
because of external cooling. Fibre preform
then contains four regions, shown in figure:
region 1 is composed of fibres, solid metal,
and flowing liquid metal, region 2 of fully solid
metal, region 3 of fully liquid and flowing
metal, and region 4 the uninfiltrated
portion
of preform. When preform or die are above
melting
point of metal, regions 1 or 2,
respectively, are absent

and form 'fingers' of liquid matrix extending into the


region of semisolid matrix.120,340In the case of unidirectional adiabatic infiltration driven by a constant
applied pressure, it was also shown 119,311that the
Boltzmann transformation is compatible with thermal
and solute transport as well as fluid flow equations.
With an alloy, matrix solidification during infiltration results in macrosegregation, featuring significant solute enrichment at the infiltration
front. 311,339,341-343
This occurs because solute is partitioned unequally between liquid and solid during
solidification. Since the solid phase is trapped within
the preform while liquid metal flows further downstream, different chemical elements concentrate
upstream and downstream. With a hypoeutectic
binary alloy, solute enrichment is found downstream
in the composite, as illustrated in Fig. 6.
Where solid metal has formed during infiltration
by cooling at the fibres (regions 1 and 5 in Figs. 5
and 6), the final solid matrix grain size is relatively
small, of the order of interfibre spaces with fine
alumina fibres and aluminium alloy matrices. This is
because of rapid cooling by the fibres and ensuing
high nucleation rates in the matrix. In the same
composites, large grains are found in the region of
remelted matrix (region 3 in Figs. 5 and 6). Quantitative treatment of these phenomena allows prediction
of infiltration kinetics, matrix microstructure, and
solute distribution. 119,307,311,339,341-343
Process parameters can then be tailored, for example, to maximise
uniformity of microstructure and composition in the
composite.

~ CEr~V~;HHHHHH~gion
~ Co

~ region

,.

31 region 1

l ..~
~ ~region 4

. .

A'

A
DISTANCE

Schematic
illustration
of
infiltration
of
preform by binary hypoeutectic alloy, under
conditions such that solid metal forms at
infiltration
front in contact with fibres and
along mould wall due to external cooling. Fibre
preform then contains same four regions as
with pure metal, with region 1 defined as
region where solid primary metal is present.
In region 5, which is only present when
preform temperature
is significantly
below
that of eutectic, the solid metal is eutectic.
When preform or die are above liquidus of
metal, regions 1 and 5, or 2, respectively, are
absent. The two curves are schematic plots of
temperature and average matrix composition
in the composite
along line A-A'.
To is
temperature of incoming metal, TE is eutectic
temperature, Co is nominal alloy composition,
CE is eutectic composition

Matrix solidification during infiltration also interferes with preform deformation because solid metal
adds strength to the preforms. This may either prevent
preform compression by locking the preform to the
die wall,104,336or prevent relaxation of compressed
regions of the preform.336,339,343
Processing of metal matrix composite
slurries

When the reinforcement consists of isolated elements


of the reinforcing phase dispersed in the matrix, the
composite forms a free flowing slurry if a sufficient
portion of the matrix is liquid. Such composite "Slurries
are generally fabricated by stirring the reinforcement
into liquid or semisolid matrix and lend themselves
to ordinary casting processes once they are formed.
The presence of a significant volume fraction of solid
phase in the flowing composite modifies their behaviour during these casting processes and creates an
International

Materials Reviews

1992

Vol. 37

NO.3

112

Mortensen and Jin

Solidification

processing of metal matrix composites

Rheology of composite slurries

pasites, the fluidity increases with increasing temperature up to 750C, then decreases with increasing
temperature. This is because SiC particles are relatively stable in AI-7Si belaw 750C, while they react

It is well known that molten metals and alloys are

at higher temperatures.

Newtanian flui~s, with a viscasity in the range


10- 3-10 - 2 Pa s that is independent .ofshear rate and
decreases with increasing temperature fallawing the
Arrhenius relation. When salid particles are dispersed
in a liquid metal, twa types .ofinteractians can occur:
an hydradynamic interactian between liquid and the
particle, and a non-hydradynamic interactian between
the particles themselves. Bath interactians praduce
an increase in the apparent viscasity .ofthe slurry, an
effect that has been canfirmed in several experimental
studies.55,156,203,210,344-350
These shaw that the
apparent viscasities .ofvariaus campasite slurries are
significantly higher than the unreinforced matrix allay
viscasities, .often by .orders .ofmagnitude. The apparent viscasities .ofthe campasites furthermore decrease
strongly with increasing shear rate aver a wide range
of shear rates, carrespanding ta nan- Newtanian
pseudaplastic behaviaur. This last abservatian indicates that the suspended particles in metal matrix
compasite slurries interact with each ather. This gives
rise ta particle agglomeratian and these aggregates
are apparently respansible for the .observed shear rate
dependent behaviaur .of the slurries. Hawever, the
nature .of the particle interaction itself is nat clearly
understaod. In the temperature range where the
matrix is semisolid, the effect .of temperature an
campasite slurry viscasity is much the same as in the
matrix allay melt: the viscasity decreases with increasing temperature predaminantly because .of the
decreasing valume fractian salid, as in the unreinforeed partially salidified allay.55,210,305,345,349-351
With campasite slurries, hawever, additianal variations in apparent viscasity may result fram chemical
interactian between the reinforcement and the matrix
which alter the shape and valume fractian .of the
reinfarcement. This effect has been shawn ta increase
the apparent viscasity .of AI-SiC particulate slurries.352,353One tapic which is cantraversial in the
literature is that .ofthixatrapic behaviaur: Mada and
Ajersch349rep art that an AI- 7Si allay reinfarced with
SiC particles does nat exhibit any thixatrapic behaviaur, whereas Maan and ca-warkers344,350 .observed
a time dependent viscasity change in a similar
compasite system.
The casting fluidities (in the present cantext, a
measure .of the mauld filling ability and nat the
inverse .ofviscasity) .ofvariaus campasite slurries have
been measured with the spiral test cammanly used
far canventianal faundry allays.12,352,354,355As
expected from viscasity data, spiral fluidity decreases
with increasing particle valume fractian, and with
decreasing particle size far a given particle valume
fractian.354,355 The effect of temperature change is
more camplex. In conventional faundry allays, the
casting fluidity increases with increasing melt temperature, whereas the fluidity .ofAA 6061 reinfarced with
15 val.-%SiC decreases with increasing temperature.
This is due ta Al4C3 formatian, which increases slurry
viscasity and encaurages reinfarcement particle
agglameratian.352 In SiC reinfarced AI-7Si allay cam-

With the additian of particles ta molten metal, the


viscosity increases and the casting fluidity decreases.
Hawever, if the matrix allay campositian and the
type .ofparticles are tailared ta avoid interface chemical reactians, compasite slurries can still be cast using
most canventianal casting techniques in praduction
faundries.356-358 In same instances, improvements in
casting quality aver unreinfarced alloys may even
result fram the increased viscasity. An impartant
example .of this is the reduced parasity faund in die
cast compasites,356,358an effect similar to that faund
with campacast or thixacast unreinfarced metals.359

added cancern because the reinfarcement distributian


must alsa be controlled.

International

Materials Reviews

1992

Vol. 37

NO.3

Particle migration

When a camposite slurry is at rest, density differences


between matrix and reinfarcement induce settling or
flaating .of the reinfarcement at a rate that depends
an lacal valume fractian and can be raughly predicted
by theary.353 This results in variatians .of reinfarcement valume fractian
within the campasite.12,61,352,356
In .one study, campasite slurries were
held in space abave the matrix liquidus.360 Comparisan .ofthe material processed in space with the same
material processed similarly an the ground shawed
much greater microscapic clustering of the reinfarcement in the graund pracessed material, in agreement
with other data 353which shaw that particle clustering
is present even in well stirred ground pracessed
particle reinforced aluminium. These data indicate
that same local redistributian .of the particles in the
liquid matrix, an a scale much smaller than the
casting, can result from gravity induced particle
matian .or fluid flaw.
Often, there is matrix salidificatian cancamitant
with reinfarcement migration. If the composite slurry
is stirred inta the temperature range where the matrix
itself is partly salid (as in campacasting), little .or na
gravity induced segregatian .of the reinfarcement
occurs even if the slurry is at rest.345,361-363This is
because the salid matrix phase has abaut the same
density as the liquid metal, sa it neither settles nor
floats in the slurry and halds the reinfarcement in
place. Since allayed matrices salidify dendritically in
mast casting pracesses, the reinfarcement particles
will nat migrate aver significant distances .once the
lacal temperature falls belaw the liquidus, regardless
.of whether it is pushed .or engulfed by the moving
liquid/solid interface .ofthe matrix. This .observation
has been used ta madel particle segregatian within
gravity cast364,365 .or centrifugally cast compasite
slurries,366,367ta obtain generally gaad agreement
with experiment. 364-371

Solidification
composites

of cast metal matrix

The influence exerted by the matrix micra structure


an the an mechanical praperties .of MM Cs, even
parallel ta an aligned fibre reinfarcement, has been
repeatedly emphasised with experimental evidence

Mortensen and Jin

(examples can be found in Refs. 372-377). This, in


turn, highlights the practical importance of solidification, which governs to a large extent' the final
composite microstructure. Direct transposition of
rules developed for microstructural control in the
solidification of unreinforced metals is not possible
with MMCs, because the reinforcing phase frequently
modifies solidification of the matrix. It is the interference of the reinforcement with matrix solidification
that is addressed in this section, for a composite that
has already been formed an4 is at rest. Nucleation of
the matrix is dealt with, then growth of the solid
phase, starting with the case where the reinforcement
is fixed in space. Subsequently, effects that arise when
the reinforcement can move in the liquid metal, in
which case redistribution of the reinforcement may
occur, are treated.
It is assumed in what follows that the reinforcement
is chemically inert in the matrix, mainly because very
little fundamental work has been done on the interaction between matrix-reinforcement chemical reactions and matrix solidification. The' discussion is
focused on the case of an alloyed matrix, because a
highly pure metal matrix is not frequently used in
practice.
Nucleation of a reinforced metal

The solid reinforcing phase can reduce the grain size


of the matrix significantly if it catalyses heterogeneous
nucleation of the primary metal phase. This rarely
seems to occur with aluminium, since grain sizes far
in excess of reinforcement diameter have frequently
been observed in aluminium reinforced with alumina 120,303,312,339,341,343,378-380
carbon 381and sili378
con ~arbide fibres
,382,383
or particles.352,384When
the reinforcement does provide a propitious site for
nucleation of the matrix, however, its effect on grain
size of the matrix can be quite strong. Thus, the grain
size of AI-45Cu is reduced by several orders of
magnitude on going from SiC or Al203 fibre
reinforcements to a reaction sintered porous TiC
reinforcement (known to act as a heterogeneous
nucleation catalyst for aluminium)385,386 processed
and solidified identically378 (TiC reinforced aluminium was also investigated by Baturinskaya
et al.,387,388but though it is stated that the 'microstructure is refined' in the conclusion of these articles,
it is unclear whether any decrease in grain size was
observed). A second example of matrix grain refinement due to nucleation catalysis by the reinforcement
is that of hypereutectic aluminium-silicon alloys,
wherein the primary phase, silicon, has been shown
to nucleate preferentially on carbon, SiC, Si02, and
AI20312,50,139,144,380,389-391
As a consequence, the
number of primary Si crystals per unit volume in
these alloys is increased in the composites compared
with the unreinforced alloy.12,380Some grain refinement has also been found in Ti-515AI-I4Mn
reinforced with less than 10 vol.- otic, TiB2 particles,
which was attributed to nucleation of the primary
phase on a small fraction of the reinforcing
particles.392 Braczynski393 proposed that intermetallic
alloy phases coat reinforcements to promote heterogeneous nucleation of the primary phase in Cu-Pb-

Solidification

processing of metal matrix composites

113

Ti alloys with graphite par~icles, and AI-Cu- Ti alloys


with alumina particles. A quantitative interpretation
of differential scanning calorimetry data was presented to show that nucleation takes place faster in
the composite, but no reason for why the intermetallic
would precipitate before tlie primary phase was given.
Grain refinement of the matrix may also result
from exchange of heat between reinforcement and
matrix during infiltration. As seen in the section
'Thermal and solidification effects' above, infiltration
of a preform initially at a temperature below the
matrix liquidus results in rapid solidification of a
portion of the matrix during infiltration. Unless this
solid metal remelts, it will produce a fine equiaxed
grain structure in the matrix.119,120,339,342,343
When the reinforcement does not induce nucleation
of the primary phase of the matrix by catalysis or
heat transfer, the grain size in composite castings is
likely to be somewh~t greater than that of an identical
casting of theunreinforced matrix. This is because
the reinforcement impedes convection of the liquid
metal, whether the reinforcement is stationary or is
discretely distributed in the matrix. Many mechanisms
responsible for formation of fine grained, equiaxed
dendritic .structures in castings depend on fluid
flow.394For this reason, if there is significant convection during solidification of a similar unreinforced
casting and matrix nucleation is sluggish, an increased
propensity for columnar dendritic solidification is
expected in composites. This effect is illustrated in
the experiments. of Cole and Bolling,395 who elucidated the' effect of fluid flow in ingot solidification by
inserting a grid of metal wires into a mould before
casting. The resulting castings were, in fact, low
volume fraction wire reinforced metals, and showed a
greater columnar zone than ones devoid of the wire
mesh.
Growth of soiid metal with stationary
rei nforcement

As seen in the section 'Thermal and solidification


effects' above, the time for equalisation of temperature
between matrix and reinforcement is at most around
1 ms. This is generally much less than the time for
matrix solidification in casting processes used for the
production of metal matrix composite ingots or parts.
For a solidifying alloyed matrix in a cast composite,
therefore, the reinforcing phase is not a heat source
or sink of much significance, but is primarily an
impermeable barrier to mass transfer.
When the reinforcement is stationary and the
matrix isan alloy, it is therefore equivalent to a very
fine and narrow crucible, within which the matrix
must solidify. As in solidification of unreinforced
metals, the simplest configuration for fundamental
study of solidification of composites is that where the
process takes place at steady state, allowing simple
definition and control of growth parameters:. temperature gradient G at growth rate V. Steady state
solidification requires that the reinforcing phase delineate tiny straight-walled crucibles, which is achieved
in practice when continuous parallel fibre reinforced
metals are solidified at steady state with the temperature gradient G parallel to the fibres.
International

Materials Reviews

1992

Vol. 37

NO.3

114

Mortensen and Jin

Steady state solidification


composites

Solidification

processing of metal matrix composites

of metal matrix
liquid metal

The morphological stability of plane front alloy solidification is analysed theoretically by calculating the

rate of growth of infinitesimal .sinusoidal perturbations of the plane front. 396,397In unreinforced
alloys, the front is assumed to be essentially infinite
in extent, so if any perturbation wavelength shows a
positive rate of growth, the front is unstable. In a
narrow interstice delineated by closely spaced fibres
of a composite, the spectrum of possible perturbation
wavelengths has an upper limit equal to the width of
the interstice. Calculated critical perturbation wavelengths Ac below which an infinite plane front is stable,
and above which perturbations grow, were given in
Ref. 398 to show that with both AI-Cu alloys and
succinonitrile-I3 wt-O/oacetone, the wavelength for
marginal stability at the breakdown of a plane front
is of the order of 100 ~m, equal to or larger than the
interstices left between fibres customarily used for
reinforcing metals. One therefore expects, as was first
pointed out by Sekhar and Trivedi399 that fibres tend
to stabilise plane front solidification when their separation falls below Ac While Shangguan and Hunt400
found no effect of spatial constraint on plane-front
breakdown in succinonitrile-acetone, recent experiments on AI-Cu matrix composites at MIT are in
general agreement with this analysis.401
One instance in which a reinforcement may lower
the stability of a plane front was also pointed out by
Trivedi et al.,402 namely that where the interstice is
a narrow channel between two wide flat plates. In
this case, there is no upper limit to the possible
destabilising perturbation wavelengths, and because
of curvature in the solidification front perpendicular
to the reinforcement plane, the stability of a plane
front is reduced by the reinforcement. However, there
are very few instances in practice where this situation
will occur in MMCs.
More detailed analyses of plane-front growth take
into account the finite contact angle 8 between the
solid metal, the liquid metal, and the reinforcing
phase, Fig. 7, at the jllnction line between the fibre
matrix interface and the liquid/solid interface (ignoring roughness effects, 8 is the same angle as that used
to evaluate potency of the fibre for heterogeneous
nucleation of the matrix). Unless 8=90, a strictly
planar front can therefore not be obtained because
the liquid/solid interface bends near the reinforcement.
Ungar and Brown403 have studied this problem theoretically in two dimensions, i.e. for solidification
between two parallel plates, using finite element
methods. These authors considered contact angles (}
greater than 90, and found that at high and small
interstice widths, there is a continuous series of solutions for the solidification front morphology going
from shallow to deep cellular solidification fronts as
the temperature gradient decreases at fixed growth
velocity V past conditions for plane-front breakdown
in the unreinforced metal. At smaller 8, still greater
than 90, a pseudo-plane front is observed at high G,
which breaks down, with a sudden increase in the
height of stable cells, at decreasing G. This pseudoplane front to cellular transition takes place at higher
G than that for plane-front breakdown with (}= 90.

International

Materials Reviews

1992

Vol. 37

NO.3

reinforcing

phase

solid metal

Definition of contact angle (J between solid


metal, liquid metal, and reinforcing phase

At larger interfibre spacings, such that Ac is smaller


than the interfibre spacing by at least an order of
magnitude, the plane-front to cellular transition is
expected to take place as in the unreinforced metal.
Near the fibre, the liquid/solid interface should then
behave essentially as it does near a grain boundary
of the unreinforced solidifying metal, Le. become a
region where instability amplitude increases faster
than along the plane front at the onset of instability.
Experimental work on directionally solidified succinonitrile-acetone alloys illustrates the effect of 8 on
morphology of a plane front near the fibres, showing
interface curvature near the channel walls.399,400,402
At values of the ratio G/V slightly below that for
breakdown of plane-front stability, the unreinforced
matrix solidifies with a cellular morphology. Spatially
constrained steady state cellular growth, similar to
growth between parallel fibres along their axis, has
been treated theoretically by McCartney and
Hunt404-406for a deep single cell centred in a cylindrical interstice, as well as Ungar and Brown407 and
Trivedi et al.402 for a two dimensional cell centred
in a planar channel. McCartney and Hunt found that
narrowing the interstice below the experimentally
measured diameter of free growing cells increases the
cell tip undercooling somewhat, although not dramatically, for the AI- Mg-Si alloy under consideration
in their work. Tip undercooling was found to decrease
by a few kelvin (or about 50% of its initial value) as
the spacing was reduced to half the spacing chosen
by the cells in the unreinforced metal, while tip radius
decreased and tip composition increased at small
interstice radii. Overall, the cell tip undercooling was
primarily determined by solute diffusion along the
cell, as first proposed by Bower et al.,408 while the
Gibbs- Thomson effect increased the undercooling at
small interstice radii by an amount that was somewhat
higher than expected from proportionality of tip
curvature with interstice radius. For more drastic
reductions in interstice diameter, to below 10% of
the experimental cell spacings measured in unreinforced metals, extrapolation of the curves points to
rather high undercoolings. This is in agreement with
calculations by Trivedi et al.,402 who found that the
tip undercooling of a two dimensional cell is drastically increased as the width of the channel is decreased
by an order of magnitude.
Experimental investigations by Sekhar and Trivedi399 using succinonitrile indicate that steady state

Mortensen and Jin

cellular growth in comp<?sites is somewhat more


complex than is assumed in calculations, because cells
do not always position themselves in the centre of
interstices. Instead, half-cells were found on occasion,
growing along one side of the interstice with the solid
contacting the fibres along that side. Similar experiments by Shangguan and Hunt400 on growth cells in
small capillary tubes show marked differences
between cells in the capillary and those growing
between glass plates with no constraint along one
direction. In the capillary tubes, the cells were larger,
and had a lower tip undercooling, consistent with
numerical calculations reviewed above.
At still lower G/V values, the unreinforced matrix
solidifies in a dendritic fashion. If the diameter of the
interstice decreases significantly below the primary
dendrite arm spacing A1 in the unreinforced alloy,
hindering of the formation of secondary dendrite arms
and an increased propensity for cellular solidification
are expected. This was shown experimentally by
Sekhar et al.,399,402 who found that channels much
smaller than the primary dendrite arm spacing of
unreinforced succinonitrile-acetone favoured cellular
or half-cell growth morphologies. Conversely, when
the channel width was slightly larger than the cell
spacing for cellular growth near the cell-dendrite
transition, dendrites were favoured over cells, apparently because of the increased amount of liquid available for secondary dendrite arms growth.402
At G/V values low enough to be well within the
dendritic growth regime of the unreinforced alloy,
dendritic structures form in composites solidified
at steady state, even when the interstices are significantly narrower than Ab with succinonitrileacetone399,402,409,410and AI-Cu alloy matrices.383
In both systems, no significant increase in dendrite
tip undercooling resulted from the geometrical constraint imposed by the reinforcement. The dendrites
that form within interstices much narrower than A1
are, however, much less regular in shape than those
found in the unreinforced metal, and feature contorted
secondary dendrite arms. Depending on the relative
orientation and on channel width, dendrites may not
be able to grow at steady state along the channel:
in one extreme case documented by Sekhar and
Trivedi,399 zigzag growth of a primary dendrite stem
was observed with a dendrite featuring two primary
growth directions both inclined 45 from the interstice
axis.
As with dendrites formed in unreinforced alloys,
the final microstructure found in the casting is largely
determined by coarsening of the initial dendritic
structure. Coarsening effects specific to composite
materials arise if cells or dendrites in the matrix reach
a size on a par with that of the interstices. This was
shown in experiments on AI-45Cu dendrites solidified at steady state within closed interstices left
between touching silicon carbide fibres, with one
dendrite per interstice, featuring a main stem from
which contorted secondary dendrite arms emanated.383 Because the dimensional scale of the matrix
microstructure cannot exceed that of the interstices
left between the fibres, ripening of dendrite arms
(wherein smaller arms remelt to deposit on to larger
ones) within a closed interstice cannot take place

Solidification

processing of metal matrix composites

115

c
a

quenched liquid metal, resulting in fine dendrites which grew


unperturbed by fibres; b from sample solidified at steady state with
gradient G=9100 K m-1 and growth rate R=203 ~m s-1, in square
interstice, contorted dendritic primary dendrite arm is shown by
coring patterns; in triangular interstice, the dendritic nature of matrix
is erased, and coring patterns are parallel to fibres; c G = 4500 K
m-1 and R= 54 ~m s-1, longer solidification
time results in nondendritic
structure in both square and triangular
interstices, in
triangular interstice, microsegregation
is reduced, as seen by missing
1
low Cu coring patterns; d G= 3500 K mand R= 25 ~m S-l. At
still longer solidification
times, microstructure
is featureless
in
triangular interstices, with no coring and no second phase.

S AI-45Cu-SiC

fibre composite
solidified
in
Bridgman furnace along fibre axes. Fibre
diameter is 140 J.1m.From work reported in
Ref. 383

beyond the point where the average dendrite arm


spacing is of the order of the interstice radius. Ripening of dendrite arms then ceases, and further coarsening of the dendrite takes place entirely by dendrite
arm coalescence, which gradually erases the dendritic
character of the solid metal. If the average time tc for
coalescence of dendrite arms is of the order of, or
larger than, the total time for solidification tr, then
the microstructure within the interstice remains den- ,
dritic, with coring patterns indicating the presence of
secondary or higher order arms, Fig. 8a and b. If, on
the other hand, tc is significantly smaller than tr,
coalescence joins the dendrite arms together, and does
so early in the solidification process. The resulting
final microstructure is then cellular in appearance,
with concentric coring patterns running parallel to
the fibre surfaces, Fig. 8c and d.
These changes in dendrite coarsening mechanisms
affect micro segregation in the matrix. Because the
microstructure can only coarsen up to the point where
the scale of the matrix microstructure equals that of
International

Materials Reviews

1992

Vol. 37

NO.3

116

Mortensen and Jin

Solidification

processing of metal matrix composites

the interstices, fibres packed to a high volume fraction


place an upper limit on diffusion distances within the
solid during solidification. Therefore, solid state
diffusion can reduce micro segregation to a greater

extent than in conventional unreinforced castings of


the same alloy.383 This effect is illustrated in Fig. 8d
where at the lowest cooling rate, a fully homogenised
AI-45Cu matrix was directly solidified in the
composite.
Unsteady solidification

pared with the unreinforced alloy, and that as the


particle volume fraction increases, the final globule
diameter decreases. These observations result from
the same coarsening phenomena founel in steady state
solidification experiments:383 as the particles are
added, the matrix cannot ripen beyond the size of
interstices between the particles. Therefore, the final
globule size decreases as the particle volume fraction
increases at constant particle size and, because
increasing the. particle volume fraction decreases the
diffusion distances for coalescence, coalescence of
dendrite arms is accelerated, resulting in earlier transformation of dendrites of the primary phase to spherical globules. Obseryations by Bryant et al.392 of a
reduced grain size in solidified TiB2 particle reinforced
titanium aluminide, which were explained by
enhanced nucleation, may alternatively also be
explained by similar coarsening effects.

Even with directional solidification at constant G and


V, steady state will not be achieved if the reinforcemerit is not parallel to -the growth direction. Trivedi
and co-workers have studied changes in directional
solidification morphology of succjnoriitrile based
alloys with various reinforcement geometries, including single fibres, isolated particles, two converging or
diverging fibres, particles, and a constriction followed
by an enlarging channel. 399,410-412These various
Growth of solid metal with mobile
geometries caused a variety of solidification morphoreinforcement: particle pushing
logy changes to occur, with concomitant variations
in local solidification velocity and resulting local
When a moving solid/liquid interface approaches a
composition of the solid. Observations were all
mobile foreign particle suspended in liquid metal, the
eXp'lained~by analysing the influence exerted by the
particle can be either captured or pushed away by
reinforcement on diffusion of solute in the liquid, to
the interface. If foreign particles are captured by the
which it constitutes a barrier. Provided there is no
growing solid metal, little redistribution of the
enhanced nucleation of the matrix on the reinforcereinforcement will occur during solidification, and
ment, the matrix alloy grows from within the interhence the particle distribution in the solidified materstices left between the reinforcement, and avoids the
ial will be as uniform as it was in the liquid state. On
latter as it grows because the reinforcement constithe other hand, if particles are pushed by the solidifitutes a barrier to solute evacuation. For this reason
cation front, they will be redistributed, to be finally
(and not for thermal reasons as frequently proposed),
segregated in the last pools of liquid matrix to solidify.
the reinforcement is quite generally found to be
Figures 9 and 10 show typical examples of solidifisurrounded with the last phase(s) to solidify in the
cation microstructures resulting from particle pushing
matrix alloy.12,46,48,50,367,372-374,378,381,391,411,413-419
and particle capture, respectively, in cast AI-Si alloys
With sucinonitrile containing 0'5% impurities,412 rereinforced with SiC particles.
The interaction of particles with a solidification
peated interaction between the solute diffusive field
front has been extensively studied by various workers.
and the particles caused oscillatory motion of the
Work up to - 1986 has been reviewed by Rohatgi
solidification front, resulting in cellular growth under
et al.,12 and Russell et al.,52 and more recent reviews
conditions where dendrites form if there are no
have been presented by Stefanescu and Dhindaw,422
particles. Depending on particle size and volume
fraction, dendrite tip splitting, particle trapping, and
and Rohatgi et al.139 This review therefore provides
only a brief outline of the earlier work, to analyse
particle rejection between primary dendrite arms were
observed when dendrites encountered particles.
recent theoretical and experimental studies with a
Most of the features found with steady state experifocus on those relevant to metal matrix composite
ments also apply with unsteady solidification and
solidification.
irregular reinforcements. Several studies have been
Experimental studies
conducted on infiltrated and remelted fibre reinforced
The first systematic work on particle pushing is due
binary aluminium alloys, in which samples were soto Uhlmann et al.,423 who mixed a number of differlidified in shallow temperature gradients, to induce
equiaxed dendrite growth in the matrix.413,414,420,421 ent particles into various transparent organic matrix
materials, and carried out horizontal directional
The microstructural evolution of the matrix in these
growth experiments under an optical microscope.
composites was in agreement with observations and
They found that, in some particle-matrix systems,
theory from steady state columnar dendrite solidifithe particles were not pushed at all, whereas in other
cation studies:383at long solidification times, coarsensystems a critical velocity ~ existed below which the
ing leads to a matrix microstructure where the
particles were pushed and above which they were
dendritic features are lost, and micro segregation is
captured by the interface. This critical velocity was
reduced.
dependent on the type of particle for a given matrix
Changes in coarsening behavior during an isomaterial, and on the kind of matrix material for a
thermal hold in the semisolid temperature range of
given particle. The effect of particle size on ~ was
AI-7Si-0'3Mg with addition of SiC particles were
not quite straightforward: when the particle size was
studied by Bayoumi and Suery.363It was found that
large (hundreds of micrometres in diameter), ~
in the composite, rounding off of dendrites to form
decreased with increasing particle size, while for small
'globules' of primary aluminium is accelerated comInternational

Materials Reviews

1992

Vol. 37

NO.3

Mortensen

and Jin

Solidification

processing of metal ma~rix composites

117

"~

\X~~~4

(lFf~/
.

,:,{i,'

" /

'.

".' .~.

,/

,~.,
" ',..

~" :..t'.~~~ f(~ ...~..... '.

....
......'i:'iJ"",,

,];,12;,f.~:flt~. };".";~I.

lI!!'
..y.'

1'lt\

50 IJ.m

Microstructure
of AI-7Si-15 vol.-%SiC composite solidified at cooling rate of 4 K s-'.
Note that
SiC particles
are pushed to
interdendritic regions by AI dendrites

particles (below 15 Jlm), it was virtually independent


of the particle size.
A series of experimental investigations followed this
study. Cisse and Bolling424,425carried out similar
directional solidification experiments with water and
salol containing various insoluble particles, with the
purpose of examining the effect of liquid viscosity and
particle size on ~. It was found that regardless of the
particle type, ~ decreased when the particle size
increased, and when the liquid viscosity increased.
Zubko et al.426 examined experimentally the effect of
the thermal conductivities of the particle and matrix
on particle capture, following theoretical work by
Chernov and Mel'nikova,427 who calculated the temperature field and interface shape around a particle
at the solidification front. The materials used were
anthracite, glass, AI, and Zn particles in naphthalene,
and Fe, Ni, and Cr particles in Zn, Bi, and Sn. It was
found that when the thermal conductivity of the
particle was higher than that of the liquid, the particles
were captured, and that otherwise the particles were
pushed.
Neumann et al.428 and Omenyi and Neumann429
investigated particle pushing from a thermodynamic
standpoint using' horizontal directional solidification
experiments with materials of known interfacial energies. They found that when the net free energy change
which occurred during the particle transfer from
liquid to solid was negative, the particles were captured even at extremely low solidification rates. Conversely, when this free energy change was positive,
the particles were pushed at low velocities and captured at high velocities. The critical velocities measured were dependent on the particle size as well as
on the value of this free energy change.
The particle pushing phenomenon has. also been
observed in certain biological systems. The interaction
of biological cells with growing ice crystals was studied by Bronstein et al.,430 using cell suspensions of
brewer's yeast and human red blood cells in aqueous
salt solutions. They found that the behaviour of the
biological cells at the solidification front was the same
as that of other organic or inorganic particles, i.e. the

~\"M
~.:.f,... /
10

ftI

"'

..

..'~"."'~,

~'J'

\,.

'.:?'~~I
,.,.~ \ . ~

_--------t

Microstructure
of
AI-16Si-15vol.-%SiC
composite (cooling rate was 7 K s-'). Note
that SiC particles are not pushed by primary
Si crystals

cells were pushed below and captured above some


critical velocity ~. In these experiments, ~ was
dependent on the solute concentration of the solution,
decreasing with increasing solute content.
The influence of the temperature gradient G on the
critical velocity was investigated by Korber and
Rau431 using latex sphere suspensions in distilled
water and in an aqueous solution of NaMn04' The
critical velocity increased with increasing G.
The effect of fluid convection in the melt on particle
pushing has been studied in metallic,432 as well as in
non-metallic,433 systems. Schvezov and Weinberg432
carried out zone refining experiments with pure Pb
and Pb-Sb alloys which contained Fe particles. When
the interface was planar (pure Pb matrix) the particles
were uniformly distributed in the solid, indicating
that the particles were not pushed by the planar
interface. When the interface was cellular (Pb-Sb
alloy matrix), the particles were segregated to the cell
boundaries. In the experiment of Delamore et al.,433
the behaviour of methyl methacrylate and carbon
particles in various organic materials was examined
under natural and forced convection. With a planar
interface, the particles were pushed, and with a nonplanar interface. the particles were deflected near the
cell tips and segregated to the intercellular regions.
These studies indicate that, when convection is present
in the bulk melt and if the solid/liquid interface is
non-planar, convective flow can sweep the particles
away from cell or dendrite tips, resulting in intercellular or interdendritic segregation.
Recent work has focused on particle pushing in the
solidification of various MMCs. The behaviour of
SiC particles at the solid/liquid interface in various
aluminium alloys has been extensively studied using
unidirectional solidification, multidirectional solidification, and rapid solidification. 352,380,417
,434-439The
parameters investigated were alloy compositio~,
particle size, particle volume fraction, growth velocity,
temperature gradient, gravity, and cooling rate. The
solid/liquid interface was non-planar in all experiments except in the case of eutectic alloy solidification. In directional solidification experiments with
International

Materials Reviews

1992

Vol. 37

NO.3

118

Mortensen and Jin

Solidification

processing of metal matrix composites

controlled temperature gradients,437particle behaviour


was studied by examining the quenched solidification
front. However in most other cases, the particle
behaviour was inferred from the particle distribution
in the resulting microstructure after solidification, i.e.
by examining whether particles are located in the
interdendritic regions or inside dendrite cells. This
method gives a good correlation when the dendrite
arm spacing is larger than the particle size, however
when the solidification rate is high and the dendrite
arm spacing is smaller than the particle size, the
microstructure does not give a clear indication of
whether the particles are captured or mechanically
interlocked between dendrite arms. From the experimental results, it is found that in SiC reinforced,
Al alloy composites, the type .of solid/liquid interface which encounters the particle during solidification has a profoupd effect on particle pushing:
Al dendrites in hypoeutectic alloys push SiC
particles,352,380,435-439while Al-Si eutectic and Si
crystal growth fronts capture the particles during
solidification.380 AI-AI3Ni eutecti~ growth fronts can
either push or capture SiC particles depending on
solidification conditions. In particular, the particles
are pushed under normal and low gravity, but captured at a high gravity level (18g).434,437
Other particles systematically investigated in various aluminium alloys. include A1203, graphite, glass,
mica, TiB2, B4C, and ZrB2.12~139,417
,418,436,440-442
Among these, graphite particles behaved very simi...
larly to SiC particles, i.e. they were pushed in hypoeutectic aluminium alloys and captured in hypereutectic Al-Si alloys.12,440A1203, glass, mica, TiB2,
and ZrB2 particles were pushed by aluminium
dendrites;436,44o-442however, B4C particles were not
pushed at all by aluminium dendrites in Al- 3Mg0'004Be matrix alloy.436
Other solidification parameters such as growth
velocity, temperature gradient, particle size, and
particle and matrix thermal conductivities had little
influence on the particle pushing in aluminium alloy
composites. However, these parameters affected the
solid/liquid interface morphology, resulting in a
change in the particle entrapment behaviour in the
interdendritic regions.437-439The buildup of particles
in front of the interface resulted in particle clusters,
which often were captured by the growth front.411,412
In Ti-515Al-14Mn
(at.-%) reinforced with
91 wt- % TiB2 particles, particle pushing by the primary alpha phase resulted in segregation of the
reinforcement into interdendritic areas.392,443Microstructural features of the composites were also interpreted to result from pushing of the particles, followed
by engulfment of particle clusters.above a critical size
by the solidification front.
This wide range of experimental data leads to three
conclusions:
.
1. There are particle-matrix combinations in which
the particles are captured by the solid/liquid interface
at all growth conditions. Examples include the
siliconed glass-biphenyl, polystyrene-naphthalene,
Teflon-biphenyl, and Teflon-naphthalene systems. In
these systems,. the net free energy change in transferring the particles from liquid to solid is negative, i.e.
the particle/solid interfacial energy is lower than the
International

Materials Reviews

1992

Vol. 37

NO.3

particle/liquid interfacial energy. In MMCs, graphite


or SiC particles in eutectic and hypereutectic AI-Si
alloys fall into this category.
2. There are particle-matrix combinations in which
the particles are pushed at growth velocities lower
than a critical velocity ~, and captured at higher
velocities. Examples of this type of behaviour include
the Ni-salol, graphite-thymol, Si02-water, latexwater, and nylon-naphthalene systems. ~ is dependent on the interface chemistry, particle size, liquid
viscosity, temperature gradient, thermal conductivities
of particle and matrix, and solute content. When the
solid/liquid interface is planar, the ~ versus particle
diameter relationship follows the equation
~dn=

(13)

where d is the particle diameter, C a constant which


depends on the system, and the exponent n varies in
the range 05-3.
3. There are particle-matrix combinations in which
particles are pushed at any growth conditions so far
tested. Examples are SiC-AI-2Mg, SiC-Al~7Si, SiCAI-6Ni, and A1203, SiC, TiB2, and ZrB2-AI-3Mg.
Most hypo eutectic Al alloy matrix composite systems
tested with a non-planar interface shape fall into this
class.
Theoretical models

Theoretical models of particle pushing can be divided


into two broad classes: the first predicts whether
particles are pushed or captured, while the second
aims at predicting a critical velocity.
Particle pushing criteria Neumann et al.428,429 analysed the net free energy change which occurs during
transfer of the particle from the liquid to the growing
solid phase. Based on their analysis and experimental
observations, they proposed a criterion for particle
pushing or capture: particles are pushed if
(ips>

(iPL

(14)

where (ips is the particle/solid interfacial energy, and


(iPL the particle/liquid interfacial energy. Their predictions compared well with their experimental results.
Uhlmann et al.423 and Potschke and Rogge444
considered the particle/solid, particle/liquid, and
solid/liquid interfacial energies at the minimum separation between the particle and the solid/liquid
interface. They stated that particle pushing would
occur only if
(15)
where aSL is the solid/liquid interf~cial energy.
These simple' thermodynamic models are closely
related to classical heterogeneous nucleation theory,
since they essentially measure the ability of a particle
to act as a heterogeneous nucleant for the growing
solid. As in nucleation theory, these models give an
insight into the fundamental aspects of the problem,
but are of limited value as a predictive tool for
practical alloys because the interfacial energy values
for most materials, particularly those which contain
solute elements, are not well known.

Mortensen and Jin

Solidification

processing of metal matrix composites

119

Other models considered the flow of heat around


a particle. Chernov and Mel'nikova427 calculated the
temperature field and the resulting interface shape
change. To test their calculated results, Zubko
et al.426 conducted experiments with materials of
known thermal conductivities, and observed that
particle capture occurred in all but one of the systems
they investigated when

particle and the growth front is negligible. By equating


this effective contact radius and the maximum possible mass flow distance which is limited by diffusion
through a mono atomic liquid layer gap, they obtained
the critical velocity for a smooth spherical particle

(16)

where 1] is' the liquid viscosity, a the solid/liquid


interfacial energy, R the particle radius, and 0= the
ratio of the particle radius to the interface radius.
The model was extended to take into account the
effect of gravity, the number of contact points, and
the particle roughness. This model was extensively
tested against experimental results,424,425,446and
good agreement was observeq. The authors concluded
that lI:: decreases as the liquid viscosity increases and
the particle size increases, regardless of particle type.
Although this model shows that lI:: is ind,ependent of
particle type, there are experimental results423,429
showing that lI:: varies widely depending on the
particle type in a given liquid material, an effect which
cannot be explained by this model.
Recently, more elaborate models have been pre-.
sented in the literature.437,444,447-453Although there
are differences between them, all consider the repulsive
force between the particle and the solid/liquid
interface, and the viscous drag acting on the particle,
and assume that at steady state these counteracting
forces must balance. The general framework of these
models is the following:
1. The interface shape is calculated, with the
interface temperature 'Ii given by

where kp and kL are the thermal conductivities of the


particle and the liquid, respectively. When kp > kv a
plane solidification front forms a depression under
the particl~, and the latter is easily entrapped. Otherwise, a bump is formed under the particle, which is
rejected as it rolls over the bump. To improve the
agreement with the experimental data of Omenyi and
Neumanrt,429 but with no theoretical justification,
Surappa and Rohatgi445 modified this criterion to
propose that particle capture occurs when
(17)

where c and p are specific heat and density, respectively, and subscripts P and L stand for particle and
liquid.
Models predicting
a critical velocity
Uhlmann
et al.423 considered diffusion in the liquid metal form-

ing the gap between the particle and the solid/liquid


interface. By solving the mass conservation equation
and the chemical potential equilibrium condition in
the gap, the authors arrived at an equation for the
critical velocity

V2
c

= _4k_T_a_a_o_0=(_1_-_0=)_3
9n1]2

R3

(19)

1- 30=

(20)
Lao VoD
Vc = 0'5(n + 1) kTR2

(18)

where n is a positive number near 4 or 5, L the latent


heat of fusion per unit volume, ao the atomic diameter
of the liquid, Vo the atomic volume of the liquid, D
the diffusivity of the liquid, R the radius of irregularities on the particle surface, k the Boltzmann constant,
and T temperature. This model was further extended
to include viscous drag of the fluid. The predicted
and experimentally measured critical velocities for
liquid particles in water showed agreement on the
order of magnitude of lI::; however, no systematic test
was made to verify the effect of each parameter in
the equation. This model is based on mass diffusion
in the liquid gap between particle and solid, which
arises from variations in interface curvature; yet equation (18) does not include an interfacial energy term,
implying that the critical velocity is not dependent
on particle type. Also, equation (18) predicts that all
flat particles (R = (0) should be captured, regardless
of the particle type and the liquid properties. This
limitation arises from the spherical particle shape
assumed in the model.
Bolling and Cisse446 considered a model based on
viscous drag on the particle. They first calculated the
viscous drag on the moving particle as a function of

interface shape, and then the effective contact radius


in the gap beyond which the interaction between the

where TM is the melting temperature of the pure


element, ~Ts, ~Tr,and ~TF are the undercoolings due
to the solute effect, the Gibbs-Thompson effect, and
the interface force, respectively. The temperature
gradient G, and the thermal conductivities of the
particle and the liquid, kp and kL, are also taken into
account in the calculation.
2. The repulsive force which pushes the particle to
the liquid is calculated. In the thermodynamic
approach, this force is calculated based on the gap
geometry and the interfacial free energy change at the
minimum separa tion, ~a 0 = a PS - (a PL + a SL)'
3. The drag force, which presses the particle
towards the growing solid is calculated as a function
of the gap geometry, particle size, fluid viscosity, and
the growth velocity.
4. Forces are equilibrated to calculate the critical
velocity. Since the solution is a steady state. solution
of a free boundary problem, the solution contains
one degree of freedom: the velocity is a function of
the gap distance d. A more complete solution to this
type of problem can be made by considering the time
evolution of the forces and the interface shape.
5. The maximum velocity criterion is invoked to
eliminate this degree of freedom, and to calculate
both the critical velocity and the critical gap distance.
Using this method, it is possible to calculate the
critical velocity with no assumption regarding the gap
International

Materials Reviews

1992

Vol. 37

NO.3

120

Mortensen and Jin

Solidification

processing of metal matrix composites

distance

(one exception is the work of Stefanescu


assumed
a planar
solid/liquid
interface and a gap distance of 1 or 2 atomic diameters). The critical velocity equations obtained by

et al.,58,437 who

this method are of many different forms depending


on the factors included in the model, and the approximations
used in the calculation.
Potschke
and
Rogge444 obtain

(21)

where Coo is the bulk solute concentration,


m} the
liquidus slope, ko the partition coefficient, G the
temperature gradient, and D the diffusion coefficient
of the solute in the liquid. This model shows that the
critical velocity increases as the interfacial free energy
change A(J 0 and the imposed temperature gradient G
increase, whereas it decreases as the fluid viscosity,
the particle radius, the conductivity ratio, and the
solute content increase. In this equation, the only
parameter that can determine whether particle pushing or capture occurs is the interfacial free energy
change A(J 0, since this is the only parameter that can
be positive, zero, or negative. When A(J 0 is zero or
negative, ~ becomes zero or negative, which means
particles are always captured. The thermal conductivity ratio kp/kL has a strong effect on the magnitude
of~; however, this parameter alone cannot determine
particle pushing or capture.
Comparison of models with experimental data

Particle pushing criteria

There is general agreement


that equation (14) and afortiori equation (15) constitute valid thermodynamic
criteria for predicting
whether particle capture will occur at all growth
velocities. Validity of this criterion is confirmed for
non-metallic
systems in experiments by Neumann
et al.428,429 Recent experimental
data for metal
m-atrix composite systems380 also confirm these criteria. In this work, AI-16Si and the same alloy
reinforced with SiC particles were solidified identically. The solidification microstructures
are shown
in Figs. 10 and 11, and reveal that in the composite,
the primary Si grains are about an order of magnitude
smaller and captured the SiC particles. The smaller
Si grain size of the composite indicates that the SiC
acted as a nucleation catalyst for solid Si (see the
section 'Nucleation of a reinforced metal' above), so
equation (14) was obeyed. Conversely, hypoeutectic
alloys show no grain refinement and extensive particle
pushing under similar solidification conditions, Fig. 9.
In Ti-515AI-l4Mn
(at.-%) reinforced with TiB2
particles, both particle pushing and matrix refinement
were observed. Contradiction
of these data with the
criteria in equations (14) and (15) was resolved by
assuming that only a few of the TiB2 -particles were
able to nucleate the matrix primary phase. This
interpretation
is supported by the observation that
there are ma~y more TiB2 particles than primary
grains in the matrix. 392
International

Materials Reviews

1992

Vol. 37

NO.3

Critical velocity
Testing the critical velocity models
with respect to particle size and temperature gradient
in non-metallic
samples presents relatively little
difficulty, because the parameters can be independently varied in an experiment without affecting other
parameters. However,' other important variables such
as A(Jo, viscosity, thermal conductivities, or solute
content cannot be varied independently
of one
another. Tests of functional relationships between Vc
and these variables are therefore practically very
difficult. Published work424,426,431.446 shows that the
observed effect of parameters is as predicted, with
the exception of the effect of matrix solute content.
The experimental results of Bronstein et al.430 for
biological cells in various solutions confirmed that Vc
decreases with increasing solute concentration; however, data of Korber et al. for latex beads in pure
distilled water and distilled water containing 056
mol.-o/o NaMn04 showed no effect of solute content.
As stated, this may, for example, be a result of the
known effects that small amounts of solute can exert
on interfacial energies. 227,454
Recently, several investigators
have conducted
directional solidification experiments with a number
of metal matrix composite systems and attempted to
correlate the results with theory.12,434-439 These are

Mortensen and Jin

important studies of the solidification of particle


reinforced metals; however, precise comparison of
data with theory is hindered by several factors:
1. The chemical nature of the interface between
particle and matrix is often ill-characterised because
of chemical reactions at the particle/matrix interface.
For instance, SiC particles react with molten aluminium under certain conditions to form A14C3; Al203
particles react with Mg in Mg-containing Al alloys
to form spinels; and B4C particles form A14C3, aluminium borocarbide, and MgB2. If a chemical reaction
occurs at the interface, the true particle/matrix
interface during solidification will not be the same as
the interface given by the initial particle/matrix alloy
chemistry and may often be difficult to define
precisely.
2. Particle settling or floating may occur because
of density differences.252Furthermore, particles interact with one another in the composite slurries to form
agglomerates, which give rise to particle segregation
before solidification begins.
3. Directional solidification studies are best conducted at steady state. This is, however, difficult to
maintain when particles accumulate ahead of the
solidification front and disturb solute and temperature
fields in a transient manner.
4. When a directional solidification experiment is
carried out with an alloyed matrix, the determination
of a critical velocity is impaired if plane-front breakdown occurs before ~ is attained. If the solid/liquid
interface is non-planar, the true interface growth
velocity (the normal velocity) varies along the
interface, with a maximum at the tip (the imposed
velocity) and a minimum at the cell groove or the
interdendritic region (virtually zero). Under this situation, particles will experience different local front
velocities at various locations along the interface for
a given imposed overall growth velocity. Measurement of ~ is then blurred.
Unfortunately, directional solidification experiments on metal matrix composite systems to date are
subject to these reservations, leaving comparison of
experiment with theory of particle pushing in the
solidification of MMCs somewhat of an open
question.

Conclusions
Current strategies for increasing wettability are generally based on sessile drop contact angle measurements
and the understanding that wettability is enhanced
by the establishment of a chemical bond across the
interface between the reinforcement and the matrix,
which lowers the apparent matrix/reinforcement
interfacial energy. Wetting is often, but not always,
improved by increasing chemical reactivity between
the metal and reinforcement surfaces, or by disruption
of any oxide skin covering the liquid metal matrix,
most noticeably aluminium. Although this approach
has proven fruitful in practice, it contains the unfortunate implication that detrimental reaction products
and/or degradation of reinforcement properties are
frequently traded off for improved reinforcement wettability. Recent improvements in the theory of surface
and interface chemistry and a growing realisation of

Solidification processing of metal matrix composites

121

the importance of the mechanics of wetting hold


much promise for the design of alternative strategies
for improving wetting in metal matrix composite
processing. Prominent examples include the work of
Naidich 70
and
Eustathopoulos
and
coworkers29,97,227on alloying strategies to improve the
wettability of ionocovalent oxides by non-reactive
metals, and of Towata et ale on fibre preform
hybridisation.190-196
Fluid flow, heat and mass transfer, and solidification effects in the processing of MMCs affect not
only the processes themselves, but also dictate to a
large extent the microstructure of the cast composite.
In the infiltration of a preform of the reinforcing
phase, matrix flow, preform deformation, and gradients in porosity, matrix composition, and matrix
microstructure are strongly influenced by solidification of the matrix in contact with initially cold
fibres. Likewise, the complex rheology of free flowing
slurries of reinforced metals, and segregation of the
reinforcement in the slurries, influence not only handling and casting of the composite, but also its microstructure and its soundness. Significant work exists
in this area, but applied ptocess modelling for
reinforced metals is not on a par with what is achieved
in the processing of unreinforced metals. This is in
part because the physics of underlying transport
phenomena still require some clarification.
The solidification of metals is strongly influenced
by the presence of a reinforcing phase. The reinforcement can affect the solidification mode, coarsening,
micro segregation, and grain size of the matrix, all
essential features of its microstructure. Experimental
and theoretical work allows prediction, in many
instances, of essential microstructural features of a
reinforced metal. Several areas are still unclear or
unexplored, but the microstructure of a cast metal
matrix composite is often understandable.
It appears, in conclusion, that the solidification
processing of MMCs now stands at a point where its
most essential physical phenomena are beginning to
be clarified from a scientific standpoint. The practical
importance of these materials is growing, and it is
now clearly realised that their processing must be
well understood if their properties are to be optimised
and consistent. It is therefore likely that further
research will soon complement the work summarised
here, aiding for example in the design of interface
chemistry for improved wetting and lowered reactivity, allowing the modelling of fluid flow and feeding
in composite castings, or providing the tools needed
to control and optimise matrix microstructure.
Beyond the questions left open to answer, and the
significant advances of the past decade, what emerges
from this review is the interplay that exists between
the various phenomena of interest. Capillary effects,
fluid flow, transport of heat, solute migration, and
matrix solidification are all interrelated in infiltration
processing. With free flowing composites of dispersed
reinforcement, the kinetics of wetting are related to
the rheology of the composites, particle distribution
is related to fluid flow, interfacial phenomena, and
matrix solidification, while particle pushing is related
to nucleation of the solid matrix. Scientific knowledge
pertaining to the solidification processing of MMCs
International Materials Reviews 1992

Vol. 37

NO.3

122

Mortensen and Jin

Solidification

processing of metal matrix composites

is still evolving; however, it is now doing so as a


relatively coherent group of closely related topics.

Acknowledgments
A. Mortensen acknowledges support from the Aluminum Corporation of America (Alcoa) in the form
of his chair at MIT, and wishes to express his gratitude
to Professor N. Eustathopoulos and Dr V. J. Michaud
for helpful discussions and comments on the manuscript. I. Jin is grateful to Alcan International Ltd for
supporting this work, and to Dr D. J. Lloyd for
helpful discussions. The authors also wish to express
their gratitude to Dr S. Fishman, Office of Naval
Research, who initiated commission of this article.
Correspondence related to this article should be
directed to the author primarily responsible for the
relevant section of the artlcle: I. Jin for the sections
'Rheology of composite slurries' and 'Growth of solid
metal with mobile reinforcement: particle pushing';
A. Mortensen all other sections.

23.
24.

A. MORTENSEN

25.
26.

A. MORTENSEN: Metall. Trans., 1990, 21A, 2287.


A. W. ADAMSON: 'Physical chemistry of surfaces',

1.
2.
3.

27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.

40.

A. KELLY and G. J. DAVIES: Metall. Rev., 1965,10, (37),


D. CRATCHLEY: Metall. Rev., 1965, 10, (37), 79-144.

1-77.

T. DONOMOTO, N. MIURA, K. FUNATANI, and N. MIYAKE: SAE


Tech. Paper Ser. No. 830252, 1983.
4. D. M. SCHUSTER, M. SKIBO, and F. YEP: J. Met., 1987, 39, (II),
60.
5. 'Duralcan MMC' and 'Casting Duralcan'; 1988, San Diego,
CA, Dural Aluminum Composites Corp.
6. 'Aluminium metal matrix composites - AI-SiC particulate';
1987, Banbury, UK, Alcan International Ltd.
7. K. K. CHAWLA: 'Composite materials - Science and engineering', Chaps. 1,2,3,4,6, and 12; 1987, New York, SpringerVerlag.
8. A. K. GHOSH: in 'Principles of solidification and materials
processing', (ed. R. Trivedi et al.), 585-611; 1989, New Delhi,
India, Oxford and IBH Publishing Co. Pvt. Ltd.
9. P. K. ROHATGI: in 'Metals handbook', 9 edn, Vol. 15, 840-854;
1988, Metals Park, OH, ASM International.
10. ZHANG ZHU: in 'Advances in cast reinforced metal composites',
(ed. S. G. Fishman and A. K. Dringra), 93-99; 1988, Metals
Park, OH, ASM International.
11. F. A. GIROT, J. M. QUENISSET, and R. NASLAIN: Compos. Sci.
Technol., 1987, 30, 155-184.
12. P. K. ROHATGI, R. ASTHANA, and s. DAS: Int. Met. Rev., 1986,
31, (3), II 5-139 .
13. J. P. ROCHER, F. GIROT, J. M. QUENISSET, R. PAILLER, and
R. NASLAIN: Mem. Sci. Rev. Metall., 1986, 83, (2), 69-85.
14. T. W. CHOU, A. KELLY, and A. OKURA: Composites, 1985, 16,
(3), 187.
15. P. BRACKE, H. SCHURMANS, and J. VERHOEST: 'Inorganic fibres
and composite materials - a survey of recent developments';
1984, Oxford, Pergamon International Information Corp.
16. s. CARON and J. MASOUNAVE: in 'Fabrication of particulates
reinforced metal composites', (ed. J. Masounave et al.),79-86;
1990, Materials Park, OH, ASM International.
17. R. B. BHAGAT: in 'Metal matrix composites: Processing and
interfaces',- (ed. R. K. Everett and R. J. Arsenault), 43-82;
1991, Boston, Academic Press.
18. M. U. ISLAM and w. WALLACE: Adv. Mater. Proc., 1988, 3,
1-35.
19. I. A. IBRAHIM, F. A. MOHAMED, and E. J. LAVERNIA: J. Mater.
Sci., 1991, 26, 1137-1156.
20. s. J. HARRIS: Mater. Sci. Technol., 1988, 4, 231-239.
21. A. MORTENSEN: in 'Metal matrix composites - processing,
microstructure and properties', 12th Rise Int. Symp. Metallurgy and Materials Science, Rise National Laboratory, Roskilde, Denmark, 1991, (ed. N. Hansen et al.), 101~12I.
22. P. C. CARMAN: Soil Sci., 1941,52, 1-14.

International

Materials Reviews

1992

Vol. 37

NO.3

and

Sci., 1982, 90, 536-538.


Trans., 1987, 18A,

J. A. CORNIE: Metall.

1160-1163.

39.

References

L. R. WHITE: J. Colloid Interface

41.
42.
43.
44.

45.
46.

47.
48.

49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.

2 edn,
338-341; 1982, New York, Wiley-Interscience.
w. D. KINGERY: Ceram. Bull., 1956,35, (3), I08-112.
N. EUSTATHOPOULOS, D. CHATAIN, and L. COUDURIER: Mater.
Sci. Eng., 1991, A135, 83-88.
V. LAURENT, D. CHATAIN, and N. EUSTATHOPOULOS: Mater. Sci.
Eng., 1991, A135, 89-94.
V. LAURENT: These de Docteur, 1988, Institut National Polytechnique de Grenoble, France.
P. KRITSALIS, L. COUDURIER, and N. EUSTATHOPOULOS: J. Mater.
Sci., 1991, 26, 3400-3408.
N. R. MORROW: Ind. Eng. Chem., 1970,62, (6),32-56.
w. G. ANDERSON: J. Petrol. Technol., Oct. 1987, 39, 12831300.
D. HILLEL: 'Soil water - Physical principles and processes',
Chap. 5, 103-127; 1971, New York, Academic Press.
E. E. MILLER: in Ref. 34, Appendix 2, pp. 255-258.
E. E. MILLER and R. D. MILLER: J. Appl. Phys., 1956, 27, (4),
324-332.
F. A. L. DULLIEN: 'Porous media - Fluid transport and pore
structure', 8-42; 1979, New York, Academic Press.
J. BEAR: 'Dynamics of fluids in porous media', 441-457; 1972,
New York, American Elsevier.
A. W. ADAMSON: 'Physical chemistry of surfaces', 4 edn,
340-348; 1982, New York, Wiley-Interscience.
N. EUSTATHOPOULOS and D. CHATAIN: Entropie Energ. Dyn.
Syst. Complexes, 1990, 153/154,40-46.
F. A. L. DULLIEN: 'Porous media, fluid transport and pore
structure', ll-I3; 1979, New York, Academic Press.
Y. W. YANG, G. ZOGRAFI, and E. E. MILLER: J. Coli. Interface
Sci., 1988, 122, 35-46.
A. MORTENSEN: Mater. Sci. Eng., 1991, A135, I-II.
F. DELANNAY, L. FROYEN, and A. DERUYTTERE: in 'Advances in
cast reinforced composites', (ed. S. G. Fishman and
A. K. Dhingra), 81-84; 1988, Metals Park, OH, ASM
International.
x. DUMANT, E. BEAUGNON, and G. REGAZZONI: J. Met., 1989,
41, (II), 46-51.
E. A. FEEST, R. M. K. YOUNG, S. I. YAMADA, and s. I. TOWATA:
in 'Developments in the science and technology of composite
materials, ECCM 3', (ed. A. R. Bunsell et al.), 165-170; 1989,
London, Elsevier.
J. KIM and S.-K. LEE: in Proc. 8th Int. Conf. on 'Composite
materials, ICCM 8', (ed. S. W. Tsai and G. S. Springer),
17KI-17K9; 1991, Covina, CA, SAMPE.
F. VESCERA, J.-P. KEUSTERMANS, M.-A. DELLIS, B. LIPS, and
F. DELANNAY: in 'Metal matrix composites - processing, microstructure and properties', 12th Rise Int. Symp. on Materials
Science, Rise National Laboratory, Roskilde, Denmark, 1991,
(ed. N. Hansen et al.), 719-724.
w. H. SUTTON: in 'Whisker technology', (ed. A. P. Levitt),
273-342; 1970, New York, Wiley-Interscience.
T. W. CLYNE, M. G. BADER, G. R. CAPPLEMAN, and P. A. HUBERT:
J. Mater. Sci., 1985, 20, 85-96.
S. NOURBAKHSH, F. L. LIANG, and H. MARGOLIN: Metall. Trans.,
1989,20A, 1861-1866.
K. C. RUSSELL, J. A. CORNIE, and s. Y. OH: in 'Interfaces in
metal matrix composites', (ed. A. K. Dhingra et al.), 61-91;
1986, Warrendale, PA, Metallurgical Society of ATME.
o. J. ILEGBUSI and J. SZEKELY: J. Coll. Interface Sci., 1988, 125,
567-574.
P. K. ROHATGI, R. ASTHANA, R. N. YADAV, and s. RAY: Metall.
Trans., 1990, 21A, 2073-2082.
J. A. CORNIE, H. K. MOON, and M. C. FLEMINGS: in 'Fabrication
of particulates reinforced metal composites', (ed. 1. Masounave
et al.), 63-78; 1990, Materials Park, OH, ASM International.
M. D. SKIBO and D. M. SCHUSTER: US Pat. 4865806, 1989.
S. CARON and J. MASOUNAVE: in 'Fabrication of particulates
reinforced metal composites', (ed. J. Masounave et al.),
107-113; 1990, Materials Park, OH, ASM International.
D. M. STEFANESCU, F. RANA, A. MOITRA, and S. KACAR: in 'Metal
matrix composites', (ed. P. K. Rohgati), 77-89; 1990, Warrendale, PA, Metallurgical Society of AIME.
A. S. KACAR, F. RANA, and D. M. STEFANESCU: Mater. Sci. Eng.,
1991, A135, 95-100.

Mortensen

60.

and Jin

s. HANUMANTH and G. A. IRONS: in 'Fabrication of particulates reinforced metal composites', (ed. J. Masounave et al.),
41-46; 1990, Materials Park, OH, ASM International.
61. E. M. KLIER, A. MORTENSEN, J. A. CORNIE, and M. C. FLEMINGS:
J. Mater. Sci., 1991, 26, 2519-2526.
62. T. SRITHARAN, K. XIA, J. HEATHCOCK, and J. MIHELICH: in 'Metal
and ceramic composites: Processing, modeling and mechanical
behavior', (ed. R. B. Bhagat et al.), 13-22; 1990, Warrendale,
PA, The Minerals, Metals and Materials Society.
63. L. V. RAMANATHAN and P. c. R. NUNES: in 'Metal matrix composites - processing, microstructure and properties', 12th Ris0
Int. Symp. on Materials Science, RiS0 National Laboratory,
Roskilde, Denmark, 1991, (ed. N. Hansen et al.),611-616.
64. Y. KIMURA, Y. MISHIMA, S. UMEKAWA, and T. SUZUKI: J. Mater.
Sci., 1984,19,3107-3114.
65. K. SHINOHARA and s. UMEKAWA: Bull. Res. Lab. Precis. Mach.
Electron., Sept. 1984, 54, 17-21.
66. N. EUSTHATOPOULOS, J. c. JOUD, P. DESRE, and J. M. HICTER:
J. Mater. Sci., 1974, 9, 1233-1242.
67. M. G. NICHOLAS and D. A. MORTIMER: in 'Carbon fibres - their
composites and applications', 129; 1971, London, The Plastics
Institute.
68. Y. V. NAIDICH and G. A. KOLESNICHENKO: Soviet Powder Metall.
Met. Ceram., 1964, 1, (19), 191.
69. c. R. MANNIG and T. B. GURGANUS: J. Am. Ceram. Soc., 1969,
52, (3), 115-118.
70. J. v. NAIDICH: Prog. Surf. Membr. Sci., 1981, 14, 353-484.
71. N. MORI, H. SORANO, A. KITAHARA, K. OGI, and K. MATSUDA:
J. Jpn Inst. Met., 1983,47, 1132-1139.
72. v. I. KOSTIKOV, YU. I. KOSHELEV, E. F. FILIMONOV, E. M.
TATIEVSKAYA, and R. N. PONKRATOVA: Soviet Powder Metall.
Met. Ceram., 1981,20, (10),732-735.
73. C. ZHONGYU, W. JINBO, and H. XIANGUI: in 'Interfaces in metalceramics composites', (ed. R. Y. Lin et al.), 233-240; 1989,
Warrendale, PA, Metallurgical Society of AIME.
74. H. FUJII, H. NAKAE, K. OKADA, and M. OGUNI: in Proc. 8th Int.
Conf. on 'Composite materials, ICCM 8', (ed. S. W. Tsai and
G. S. Springer), 19BI-19B8; 1991, Covina, CA, SAMPE.
75. D. A. WEIRAUCH and w. F. KRAFICK: Metall. Trans., 1990, 21A,
1745-1751.
76. E. IGNATOWITZ: Aluminium, 1974,50, (5),334-338.
77. V. LAURENT, D. CHATAIN, and N. EUSTHATOPOULOS: J. Mater.
Sci., 1987, 22, 244-250.
78. G. V. SAMSONOV, A. D. PANASYUK, and G. K. KOZINA: Soviet
Powder Metall. Met. Ceram., 1968,71,874-878.
79. R. WARREN and c. H. ANDERSSON: Composites, 1984, 15, (1),
101-112.
80. K. NOGI and K. OGINO: Trans. Jpn Inst. Met., 1988,29, 742-747.
81. M. SHIMBO, M. NAKA, and I. OKAMOTO: J. Mater. Sci. Lett.,
1989, 8, 663-666.
82. V. LAURENT, D. CHATAIN, N. EUSTATHOPOULOS, and x. DUMANT:
in 'Advances in cast reinforced composites', (ed. S. G. Fishman
and A. K. Dhingra), 27-31; 1988, Metals Park, OH, ASM
International.
83. F. DELANNAY, L. FROZEN, and A. DERYTTERE: J. Mater. Sci.,
1987, 22, 1-16.
84. J. H. HAUSNER: J. Mater. Sci. Lett., 1986,5, 549-551.
85. W. DAWIHL and H. FEDERMANN: Aluminium, 1974, 50, (9),
574-577.
86. o. B. KOZLOVA and s. A. SUVOROV: Refractories, 1976, 17,
763-767.
87. L. COUDURIER, J. ADORIAN, D. PIQUE, and N. EUSTHATOPOULOS:
Rev. Int. Hautes Temp. Refract., 1984,21, 81-93.
88. J. J. BRENNAN and J. A. PASK: J. Am. Ceram. Soc., 1968,51,
(10), 569-573.
89. s. M. WOLF, A. P~ LEVITT, and J. BROWN: Chem. Eng. Prog.,
1966, 62, (3), 74-78.
90. J. A. CHAMPIO~, B. J. KEENE, and J. M. SILLWOOD: J. Mater.
Sci., 1969, 4, 39-49.
91. R. D. CARNAHAN, T. L. JOHNSTON, and c. H. LI: J. Am. Ceram.
Soc., 1958, 41, (9), 343-347.
92. V.
LAURENT,
D.
CHATAIN,
C.
CHATILLON,
and
N. EUSTATHOPOULOS: Acta Metall., 1988,36, 1797-1803.
93. J. G. LI, D. CHATAIN, L. COUDURIER, and N. EUSTATHOPOULOS:
J. Mater. Sci. Lett., 1988, 7, 961-963.
94. z. LIJUN, W. JINBO, Q. JITING, and N. QIU: in 'Interfaces in
metal-ceramics composites', (ed. R. Y. Lin et al.), 213-226;
1989, Warrendale, PA, The Minerals, Metals and Materials
Society.
G.

Solidification

processing of metal matrix composites

95.
96.

M. G. NICHOLAS: Mater.

97.

D. CHATAIN, L. COUDURIER,

I. RIVOLLET, D. CHATAIN,

123

Sci. Forum, 1988,29, 127-150.


N. EUSTATHOPOULOS: Acta Metall.,

and

1987, 35, 835-844.


and N. EUSTATHOPOULOS: Rev. Phys.
Appl., 1988,23, 1055-1064.
98. L. RAMQVIST: Int. J. Powder Metall., 1965, 1, (4), 2-21.
99. L. RAMQVIST: Int. J. Powder Metall., 1966, 2, (2), 34.
100. H. FUJII and H. NAKAE: Rep. Mater. Sci. Tech., Waseda Univ.,
1990, 41, 17-28.
101.

J. H. AHN, N. TERAO,

and

A. BERGHEZAN: Scr. Metall.,

1988,

22, 793-796.
102.
103.

A. M. CAZABAT: Contemp. Phys., 1987, 28, 347-364.


G. R. CAPPLEMAN, J. F. WATTS, and T. W. CLYNE: J. Mater.

Sci.,

1985, 20, 2159-2168.


104. A. MORTENSEN and T. WONG: Metall. Trans., 1990, 21A,
2257-63.
105. A. R. CHAMPION, W. H. KRUEGER, H. S. HARTMANN, and
A. K. DHINGRA: in Proc. 2nd Conf. on 'Composite materials,
ICCM 2', (ed. B. R. Noton et al.), 883-904; 1978, Warrendale,
PA, Metallurgical Society of AIME.
106. I. RIVOLLET, D. CHATAIN, and N. EUSTATHOPOULOS: J. Mater.
Sci., 1990, 25, 3179-3185.
107. T. OKI, T. CHOH, and A. HIBINO: J. Jpn Inst. Met., 1985, 49,
1131-1137.
108. T. CHOH, R. KAMMEL, and T. OKI: Z. Metallkd., 1987, 77,
286-290.
109. T. OKI, T. CHOH, and A. HIBINO: Keikinzoku (Light Met. Jpn),
1985, 35, 670-677.
110. T. OKI, T. CHOH, and A. HIBINO: Keikinzoku (Light Met. Jpn),
1985, 35, 663-669.
111. T. CHOH and T. OKI: Mater. Sci. Technol., 1987, 3, 378-385.
112. s. Y. OH, J. A. CORNIE, and K. C. RUSSELL: Metall. Trans., 1989,
20A, 527-532.
113. s. Y. OH, J. A. CORNIE, and K. C. RUSSELL: Metall. Trans., 1989,
20A, 533-541.
114. T. R. FLETCHER, J. A. CORNIE, and K. C. RUSSELL: in 'Advances
in cast reinforced composites', (ed. S. G. Fishman and
A. K. Dhingra), 21-25; 1988, Metals Park, OH, ASM
International.
115. G. R. EDWARDS and D. L. OLSON: Report MT-CWR-089-038,
Annual Report to ONR contract NOOOI4-88-K-0500,
Center for Welding Research, Colorado School of Mines,
Golden, CO.
116. P. B. MAXWELL, G. P. MARTINS, D. L. OLSON, and G. R. EDWARDS:
Metall. Trans., 1990, 21B, 475-485.
117. J. D. SEITZ, G. R. EDWARDS, G. P. MARTINS, and P. Q. CAMPBELL:
in 'Interfaces in metal-ceramic composites', (ed. R. Y. Lin
et al.), 197-212; 1989, Warrendale, PA, The Minerals, Metals
and Materials Society.
118. A. BARDAL and R. H0IER: in 'Metal matrix composites processing, microstructure and properties', 12th Ris0 Int.
Symp. on Materials Science, Ris0 National Laboratory, Roskilde, Denmark, 1991, (ed. N. Hansen et al.),205-210.
119. A. MORTENSEN, L. J. MASUR, J. A. CORNIE, and M. C. FLEMINGS:
Metall. Trans., 1989, 20A, 2535-2547.
120. L. J. MASUR, A. MORTENSEN, J. A. CORNIE, and M. C. FLEMINGS:
Metall. Trans., 1989, 20A, 2549-2557.
121. L.-J. FU, M. SCHMERLING, and H. L. MARCUS: in 'Composite
materials: Fatigue and fracture', (ed. H. T. Hahn), STP 907,
51-72; 1986, Philadelphia, PA, American Society for Testing
and Materials.
122. Q. LI: MS thesis, Figs. 51 and 52, MIT, Cambridge, MA, 1988.
123. H. KAUFMANN, T. SCHMIDT, and H. P. DEGISCHER: in 'Metal
matrix composites - processing, microstructure and properties', 12th Ris0 Int. Symp. on Materials Science, Ris0
National Laboratory, Roskilde, Denmark, 1991, (ed.
N. Hansen et al.),441-448.
124. H. KAUFMANN and A. MORTENSEN: Metall. Trans. A, in press.
125. A. BANERJI, P. K. ROHATGI, and w. REIF: Metall, 1984, 38, (7),
656-661.
126. B. P. KRISHNAN, M. K. SURAPPA, and P. K. ROHATGI: J. Mater.
Sci., 1981,16, 1209-1216.
127. R. H. SIFFERLEN: US Pat. 3 858 640, 1975.
128. V. AGARWALA and D. DIXIT: Trans. Jpn Inst. Met., 1981, 22,
521-526.
.
_129. M. D. SKIBO and D. M. SCHUSTER: US Pat. 4865 806, 1989.
130. w. H. HUNT: in 'Interfaces in metal matrix composites', (ed.
A. K. Dhingra et al.), 3-25; 1986, Warrendale, PA, Metallurgical Society of AIME.
International

Materials Reviews

1992

Vol. 37

NO.3

124

Mortensen and Jin

Solidification

processing of metal matrix composites

131.
132.
133.

T. WADA, D. J. ADENIS, and T. B. cox: US Pat. 4662429, 1987.


M. KOBAsm and T. CHOH: Mater. Trans. - JIM,
1990, 31,

134.

T. WADA, G. T. ELDIS,

A. K. DHINGRA: Phi/os.

Trans. R. Soc., 1980, A294, 559-564.

1101-1107.
and

D. L. ALBRIGHT:

US Pat. 4657065,

1987.
135.
136.
137.

and H. ARAKAWA: US Pat. 3 871 834, 1975.


R. V. SARA: US Pat. 4347083, 1982 or 4402 744, 1983.
J. A. PASK and R. M. FULRATH: J. Am. Ceram. Soc., 1962,45,
(12), 592-596.
138. B. C. PAl, S. RAY, K. V. PRABHAKAR, and P. K. ROHATGI: Mater.
Sci. Eng., 1976, 24, 31-44.
139. P. K. ROHATGI, R. ASTHANA, and F. YARANDI: in 'Solidification
of metal matrix composites', (ed. P. K. Rohatgi), 51-75; 1990,
Warrendale, PA, The Minerals, Metals and Materials Society.
140. M. F. AMATEAU: J. Compos. Mater., 1976, 10, 279.
141. A. G. KULKARNI, B. C. PAl, and N. BALUSUBRAMANIAN: J. Mater.
Sci., 1979, 14, 592.
142. E. G. KENDALL: in 'Composite materials, Vol. 4, Metallic matrix
composites', (ed. J. L. Broutman et al.), 319; 1974, New York~
Academic Press.
143. M. SITTIG: in Chern. Technol. Rev., (162), 220-226; 1980, Parks
Ridge, NJ, Noyes Data Corp.
144. B. K. CHEN, T. MA, and Y. S. wu: in Proc. 5th Int. Conf. on
'Composite materials, ICCM 5', (ed. W. C. Harrigan), 709;
1985, Warrendale, PA, Metallurgical Society of AIME.
145. H. MORITA and H. BABA: J. Jpn Inst. Met., 1973,37, (3), 315.
146. A. A. ZABOLOTSKII and s. E. SALIBEKOV: Met. Sci. Heat Treat.,
1978, 20, 841.
147. B. C. PAl and P. K. ROHATGI: Mater. Sci. Eng., 1975, 21, 161.
148. R. W. SEXTON and D. M. GODDARD: US Pat. 4341 823, 1982.
149. G. IDE, W. GRUHL, and H. GEBHARDT: in 'Faserverbundwerkstofi'e', (ed. W. Bunk et al.), 27-74; 1980, Berlin, SpringerVerlag.
150. E. DE LAMOTTE, K. PHILLIPS, A. J. PERRY, and H. R. KILLIAS:
J. Mater. Sci., 1972, 7, 346-349.
151. R. V. SARA: US Pat. 3473900, 1969.
152. H. BEUTLER, E. DE LAMOTTE, and A. J. PERRY, US Pat. 3 720257,
1973.
153. C. VAIDYANATHAN and R. ROHRMAYER: in Proc. 8th Int. Conf.
on 'Composite materials, ICCM 8', (ed. S. W. Tsai and
G. S. Springer), 17Dl-17DI0; 1991, Covina, CA, SAMPE.
154. H. A. KATZMAN: US Pat. 4376803, 1983.
155. H. A. KATZMAN: J. Mater. Sci., 1987,22, 144-148.
156. J. P. ROCHER, F. GIROT, J. M. QUENISSET, and R. NASLAIN: in
Proc. 1st European Conf. on Composite Materials, ECCM-l,
'Developments in the science and technology of composite
materials', Bordeaux, 1985, AEMC, (ed. A. R. Bunsell et al.),
634-639.
157. J. P. ROCHER, J. M. QUENISSET, and R. NASLAIN: J. Mater. Sci.
Lett., 1985,4,1527-1529.
158. J. P. ROCHER, J. M. QUENISSET, R. PAILLER, and R. NASLAIN:
European Pat. 0 105 890, 1983.
159. J. P. ROCHER, J. M. QUENISSET, R. PAILLER, and R. NASLAIN: US
Pat. 4659593, 1987.
160. J. P. ROCHER, J. M. QUENISSET, and R. NASLAIN: J. Mater. Sci.,
1989, 24, 2697-2703.
161. H. LUNDIN: US Pat. 2686354, 1954.
162. P. G. KARANDIKAR and T. W. CHOU: in 'Fundamental relationships between microstructure and mechanical properties in
metal matrix composites', (ed. P. K. Liaw et al.), 59-69; 1990,
Warrendale, PA, The Minerals, Metals and Materials Society.
163. P. V. HEGDE, V. GOPINATHAN, and P. RAMAKRISHNAN: in 'Metal
and ceramic matrix composites: Processing, modeling and
mechanical behavior', (ed. R. B. Bhagat et al.), 117-126; 1990,
Warrendale, PA, The Minerals, Metals and Materials Society.
164. s. SCHAMM, R. FEDOU, J. P. ROCHER, J. M. QUENISSET, and
R. NASLAIN: Metall. Trans., 1991, 22A, 2133-2139.
165. w. D. JOHNSTON and I. G. GREENFIELD: in Proc. 8th Int. Conf.
on 'Composite materials, ICCM 8', (ed. S. W. Tsai and
G. S. Springer), 19EI-19EI2, 1991, Covina, CA, SAMPE.
166. H. L. MARCUS, D. L. HULL, and M. F. AMATEAU: in 'Failure
modes in composites IV', (ed. J. A. Cornie and
F. W. Crossman), 308-318, Fig. 7; 1977, Warrendale, PA,
Metallurgical Society of AIME.
167. N. I. ABDUL-LATTEF, A. R. ISM~IL KHEDAR, and s. K. GOEL:
J. Mater. Sci. Lett., 1985, 4, 385-388.
168. R. T. PEPPER and E. G. KENDALL: US Pat. 3770488, 1973.
169.: D. M. GODDARD: J. Mater. Sci., 1978, 13, 1841-1848.
K. KUNIYA

International

Materials Reviews

1992

Vol. 37

NO.3

170. A. P. LEVITT and H. E. BAND: US Pat. 4 157409, 1979.


171. w. C. HARRIGAN, R. H. FLOWERS, and s. P. HUDSON: US Pat.
4223075, 1980.
172. w. L. LACHMAN, R. A. PENTY, and A. F. JAHN: US Pat. 3 894863,
1975.

173. E. G. KENDALL and R. T. PEPPER: US Pat. 4082864, 1978.


174. w. C. HARRIGAN and R. H. FLOWERS: in 'Failure modes in
composites IV', (ed. J. A. Cornie and F. W. Crossman),
319-335; 1977, Warrendale, PA, Metallurgical Society of
AIME.
175. W. MEYERER, D. KIZER, and s. PARPOCKI: in 'Failure modes in
composites IV', (ed. J. A. Cornie and F. W. Crossman),
297-307; 1977, Warrendale, PA, Metallurgical Society of
AIME.
176. T. B. CAMERON, W. W. SWANSON, J. M. TARTAGLIA, and T. B. cox:
US Pat. 4713 111, 1987.
177. T. DONOMOTO, A. TANAKA, M. OKADA, and T. KYONO: US Pat.
4419 389, 1983.
178. D. C. HALVERSON and R. L. LANDINGHAM: US Pat. 4718941,
1988.
179. I. L. KALNIN: US Pat. 4056874, 1977.
180. M. K. AGHAJANIAN, J. T. BURKE, D. R. WHITE, and
A. s. NAGEL BERG: in Proc. Conf. 'Tomorrow's materials today',
Int. Symp. and Exhib., Reno, NA, 1989,34,817-823; Corvina,
CA, SAMPE.
181. M. K. AGHAJANIAN, M. A. ROCAZELLA, J. T. BURKE, and
s. D. KECK: J. Mater. Sci., 1991, 26, 447-454.
182. J. T. BURKE, M. K. AGHAJANIAN, and M. A. ROCAZELLA: in Proc.
Conf. 'Tomorrow's materials today', Int. Symp. and Exhib.,
Reno, NA, 1989,34,2440-2454; Corvina, CA, SAMPE.
183. D. R. WHITE, A. W. URQUHART, M. K. AGHAJANIAN, and
D. K. CREBER: US Pat. 4828 008, 1989.
184. A. W. URQUHART: Adv. Mater. Proc., 1991, 140, 25-29.
185. N. L. HANSEN, T. A. ENGH, and o. LOHNE: in 'Interfaces in
metal-ceramics composites', (ed. R. Y. Lin et al.), 241-257;
1989, Warrendale, PA, The Minerals, Metals and Materials
Society.
186. D. J. LLOYD, A. D. McLEOD, P. L. MORRIS, and I. JIN: US Pat.
5 028 392, 1991.
187. K. NOGI, H. TAKEDA, and K. OGINO: J. Jpn Inst. Met., 1990,
31, 83-90.
188. R. M. K. YOUNG: Mater. Sci. Eng., 1991, A135, 19-22.
189. R. M. K. YOUNG: Mater. Sci. Technol., 1990, 6, 548-553.
190. S.-I. TOWATA, S.-I. YAMADA, and T. OHWAKI: Trans. Jpn Inst.
Met., 1985, 26, 563-570.
191. S.-I. TOWATA, and S.-I. YAMADA: in Proc. Conf. 'Composites
'86: Recent advances in Japan and the United States, CCMIII', Tokyo, 1986, (ed. K. Kawata et al.), The Japan Society
for Composite Materials, Tokyo, 497-503.
192. S.-I. TOWATA, H. IKUNO, and S.-I. YAMADA: J. Jpn Inst. Met.,
1987, 51, 248-255.
193. S.-I. TOWATA, H. IKUNO, and S.-I. YAMADA: in Proc. 6th Int.
Conf. on 'Composite materials, ICCM 6', (ed. F. L. Matthews
et al.), 2.412-2.421;
1987, London, Elsevier Applied
Science.
194. S.-I. TOWATA, H. IKUNO, and S.-I. YAMADA: Trans. Jpn Inst.
Met., 1988, 29, 314-321.
195. S.-I. YAMADA, S.-I. TOWATA, and H. IKUNO: in 'Advances in cast
reinforced metal composites', (ed. S. G. Fishman and
A. K. Dhingra), 109-114; 1988, Metals Park, OH, ASM
International.
196. H. IKUNO, S.-I. TOWATA, and S.-I. YAMADA: J. Jpn Inst. Met.,
1989, 53, 327-332.
197. H. M. CHENG, A. KITAHARA, K. KOBAYASHI, and B. L. ZHOU:
J. Mater. Sci. Lett., 1991, 10, 795-797.
198. S. DE BONDT, L. FROYEN, and A. DERUYTTERE: Mater. Sci. Eng.,
1991, A135, 29-32.
199. 'Tyranno fiber reinforced metal composites (FRM)'; 1987,
Ube City, Japan, Ube Industries.
200. R. MEHRABIAN, R. G. RIEK, and M. C. FLEMINGS: Metall. Trans.,
1974, 5, 1899-1905.
201. c. G. LEVI, G. J. ABBASHIAN, and R. MEHRABIAN: in 'Failure
modes in composites IV', (ed. J. A. Cornie and F. W.
Crossman), 370-391; 1977, Warrendale, PA, Metallurgical
Society of AIME.
202. F. M. HOSKING, F. FOLGAR-PORTILLO, R. WUNDERLIN, and
R. MEHRABIAN: J. Mater. Sci., 1982, 17, 477-498.
203. F. A. GIROT, L. ALBINGRE, J. M. QUENISSET, and R. NASLAIN:
J. Met., 1987, 39, (11), 18-21.

Mortensen and Jin


204.
205.

Solidification

P. K. BALASUBRAMANIAN, P. S. RAO, B. C. PAl, K. G. SATYANARAYANA, and P. K. ROHATGI: Compos. Sci Technol., 1990, 39,

239.

245-259.

240.
241.

and M. SuERY: in Proc. Conf. 'Advanced materials


research and developments for transport - composites', Strasbourg, 1985, (ed. P. Lamicq et al.), MRS-Europe, 241-248.
206. M. A. BAYOUMI and M. SuERY: in Proc. 6th Int. Conf. on
'Composite materials, ICCM 6', (ed. F. L. Matthews et al.),
2.481-2.490; 1987, London, Elsevier Applied Science.
207. M. A. BAYOUMI and M. SuERY: in 'Advances in cast reinforced
metal composites', (ed. S. G. Fishman and A. K. Dhingra),
167-172; 1988, Metals Park, OH, ASM International.
208. P. G. KARANDIKAR and T.-W. CHOU: J. Mater. Sci., 1991, 26,
2573-2578.
209. c. C. JONES and F. E. WAWNER: in 'Fundamental relationships
between microstructure and mechanical properties of metalmatrix composites', (ed. P. K. Liaw and M. N. Gungor),
47-58; 1990, Warrendale, PA, The Minerals, Metals and
Materials Society.
210. M. C. FLEMINGS: Metall. Trans., 1991, 22A, 957-981.
211. M. KATSURA: Japan Pat. 57-25275, 1982.
212. L. O. PENNANDER, J. E. STAHL, and c. H. ANDERSSON: in Proc.
8th Int. Conf. on 'Composite materials, ICCM 8', (ed.
S. W. Tsai and G. S. Springer), 17BI-17BIO; 1991, Covina,
CA, SAMPE.
213. L. PENNANDER and C.-H. ANDERSON: in 'Metal matrix composites - processing, microstructure and properties', 12th Ris0
Int. Symp. on Materials Science, Ris0 National Laboratory,
Roskilde, Denmark, 1991, (ed. N. Hansen et al.),575-580.
214. N. D. LANG and w. KOHN: Phys. Rev. B, 1970, 1, (12), 45554568.
215. J. R. SMITH: in 'Topics in applied physics 4: Interactions in
metal surfaces', (ed. R. Gomer), 1-39; 1975, Springer-Verlag.
216. E. CHACON, F. FLORES, and G. NAVASCuES: J. Phys. F. Met.
Phys., 1984, 14, 1587-1601.
217. J. M. BLAKELY: 'Introduction to the properties of crystal
surfaces', Chap. 5; 1973, Oxford, Pergamon Press.
218. J. HARRIS and R. O. JONES: Phys. Lett., 1974, 46A, 407-408.
219. A. ZANGWILL: 'Physics at surfaces'; 1989, Cambridge University Press.
220. A. S. SKAPSKI: Acta Metall., 1956,4, 576-582.
221. A. s. SKAPSKI: J. Chern. Phys., 1948, 16, (4), 386-389; 389-393.
222. s. H. OVERBURY, P. A. BERTRAND, and G. A. SOMORJAI: Chern.
Rev., 1975, 75, (5), 547-560.
223. N. EUSTATHOPOULOS and J. C. JOUD: in 'Current topics in
materials science', Vol. 4, (ed. E. Kaldis), 281-360; 1980,
Amsterdam, North-Holland.
224. L. E. MURR: 'Interfacial phenomena in metals and alloys'; 1975,
London, Addison-Wesley.
225. A. W. ADAMSON: 'Physical chemistry of surfaces', 4 edn,
260-293; 1982, New York, Wiley-Interscience.
226. J. E. SHOUTENS: J. Mater. Sci., 1989, 24, 2681-2686.
227. J. G. LI, L. COUDURIER, and N. EUSTATHOPOULOS: J. Mater. Sci.,
1989, 24, 1109-1116.
228. N. EUSTATHOPOULOS: Int. Met. Rev., 1983, 28, 189-210.
229. A. M. STONEHAM: J. Am. Cerarn. Soc., 1981, 64, (1), 54-60.
230. G. C. BRENNAN and K. s. YUN: in 'The solid-gas interface',
Vol. 1, (ed. E. A. Alison Flood), 203-269; 1967, Marcel
Dekker.
231. z. A. MUNIR and J. P. HIRTH: in 'Materials science research,
Vol. 14: Surfaces and interfaces in ceramic and ceramic-metal
systems', (ed. J. Pask and A. Evans), 23-34; 1981, New York,
Plenum Press.
232. J. E. McDONALD and J. G. EBERHART: Trans. AIME, 1965,233,
512-517.
233. L. H. VAN VLACK: Met. Eng. Q., Nov. 1965, 7-12.
234. w. D. KINGERY, H. K. BOWEN, and D. R. UHLMANN: 'Introduction
to ceramics', 2 edn, Chap. 5; 1976, New York, WileyIn terscience.
235. w. A. WEYL: in 'Structure and properties of solid surfaces',
(ed. R. Gomer and C. S. Smith), 147-184; 1952,The University
of Chicago Press.
236. A. M. STONEHAM and P. W. TASKER: J. Phys. C, Solid State
Phys., 1985, 18, L543-L548.
237. K. NOGI, K. OISHI, and K. NOGINO: Mater. Trans. - JIM, 1989,
30, 137-145.
238. A. M. STONEHAM and P. W. TASKER: in 'Ceramic microstructures
'S6 - role of interfaces', (ed. 1. A. Pask and A. G. Evans),
155-165; 1986, New York, Plenum Press.
C. MILLIERE

JARLEN

processing of metal matrix composites

DON

and c.

P. JU:

125

Mater. Sci. Eng., 1990, A124,

259-266.
M. ENDO: J. Mater.
A. MORTENSEN: in

Sci., 1988, 23, 598-605.


'Mechanical and physical behaviour of
metallic and ceramic composites', 9th Ris0 Int. Symp. on
Metallurgy and Materials Science, Ris0 National Laboratory,
Roskilde, Denmark, 1988, (ed. S. I. Andersen et al.),141-155.
242. D. HULL: 'An introduction to composite materials', Chap. 1,
3; 1981, Cambridge University Press.
243. D. M. RIGGS, R. J. SHUFORD, and R. W. LEWIS: in 'Handbook
of composites', (ed. G. Lubin), 196-271; 1982, New York, Van
Nostrand Rheinhold.
244. F. M. FOWKES: Ind. Eng. Chern., 1964,56,40-52.
245. A. W. ADAMSON: 'Physical chemistry of surfaces', 4 edn,
232-259; 1982, New York, Wiley-Interscience.
246. E. M. LIFSHITZ: Sov. Phys., 1956, 2, (1), 73-83.
247. R. G. BARRERA, and c. B. DUKE: Phys. Rev. B, 1976, 13, (10),
4477-4489.
248. D. B. HOUGH and L. R. WHITE: Adv. Colloid Interf Sci., 1980,
14, 3-41.
249. D. E. SULLIVAN: J. Chern. Phys., 1981, 74, (4), 2604-2615.
250. A. M. STONEHAM: Appl. Surf Sci., 1982-1983, 14, 249-259.
251. w. H. PRITCHARD: in 'Aspects of adhesion 6', (ed. D. J. Aldner),
11-23; 1969, University of London Press.
252. A. J. KINLOCH: J. Mater. Sci., 1980, 15, 2141-2166.
253. H. KRUPP: Adv. Colloid Interf Sci., 1967,1, 79-110.
254. B. V. DERYAGIN, N. A. KROTOVA, and v. P. SMILGA: 'Adhesion
in solids', 1978, New York, Consultants Bureau.
255. J. A. PASK and A. P. TOMSIA: in 'Surfaces and interfaces in
ceramic and ceramic-metal systems', (ed. J. Pask and
A. Evans), 411-419; 1981, New York, Plenum Press.
256. K. ARATANI and Y. TAMAI: in 'Surfaces and interfaces in ceramic
and ceramic-metal systems', (ed. J. Pask and A. Evans),
433-443; 1981, New York, Plenum Press.
257. G. s. UPADHYAYA: in 'Sintered metal-ceramic composites',
(ed. G. S. Upadhyaya), 41-50; 1984, Amsterdam, Elsevier
Science.
258. L. RAMQVIST: Jernkontorets Ann., 1969, 153, 159-179.
259. R. WARREN: J. Nucl. Mater., 1968, 25, 117-119.
260. s. K. RHEE: J. Am. Ceram. Soc., 1970, 53, (7), 386-389.
261. H. GORETZKI, H. E. EXNER, and w. SCHEUERMANN: in 'Modern
developments in powder metallurgy - Vol. 4: Processes', (ed.
H. H. Hausner), 327-336; 1971, New York, Plenum Press.
262. P. BENJAMIN and C. WEAVER: Proc. R. Soc., 1961, A261,
516-531.
263. P. BENJAMIN and c. WEAVER: Proc. R. Soc., 1959, A252,
418-430.
264. P. BENJAMIN and C. WEAVER: Proc. R. Soc., 1963, A274,
267-273.
265. s. V. PEPPER: J. Appl. Phys., 1976, 47, (3), 801-808.
266. G. PETZOW, T. sU9A, G. ELSSNER, and M. TURWITT: in 'Sintered
metal-ceramic composites', (ed. G. S. Upadhyaya), 3-18; 1984,
Amsterdam, Elsevier Science.
267. G. I. ZHURAVLEV and v. P. ALEKSEEV: in 'Inorganic and
organosilicate coatings', (ed. M. M. Schul'ts), Translation of
Proc. of 6th All-Union Conf. on Heat Resistant Coatings,
Leningrad, 1973,(Russian publisher Nauka Publishers, Leningrad Division, Leningrad, 1975), 13- 20; Published for
National Bureau of Standards, US Department of Commerce
and National Science Foundation, Washington, DC, by Amerind Publishing Co. Pvt. Ltd, New Delhi, 1985.
268. R. WARREN: J. Mater. Sci., 1980, 15, 2489-2496.
269. D. CHATAIN, I. RIVOLLET, and N. EUSTATHOPOULOS: J. Chim.
Phys., 1986,83, 561-567.
270. D. CHATAIN, I. RIVOLLET, and N. EUSTATHOPOULOS: J. Chim.
Phys., 1987,84, 201-203.
271. A. R. MIEDEMA and F. J. A. DEN BROEDER: Z. Metallkd., 1979,
70, 14-20.
272. J. T. KLOMP: Proc. Br. Ceram. Soc., 1984,34, 249-259.
273. A. A. APPEN: in 'Inorganic and organosilicate coatings', (ed.
M. M. Schul'ts), Translation of Proc. of 6th All-Union Conf.
on Heat Resistant Coatings, Leningrad, 1973, (Russian publisher Nauka Publishers, Leningrad Division, Leningrad,
1975), 3-12; Published for National Bureau of Standards, US
Department of Commerce and National Science Foundation,
Washington, DC, by Amerind Publishing Co. Pvt. Ltd, New
Delhi, 1985.
274. P. HICTER, D. CHATAIN, A. PASTUREL, and N. EUSTATHOPOULOS:
J. Chim. Phys., 1988, 85, 941-945.
International

Materials Reviews

1992

Vol. 37

NO.3

126
275.

Mortensen and Jin

K. H. JOHNSON

and s.

Solidification

V. PEPPER: J. Appl.

processing of metal matrix composites


Phys.,

1982, 53,

663-667.
276.

P. JENA, D. SHILLADY,

and R. J. ARSENAULT: in 'Interfaces in


metal-ceramics composites', (ed. R. Y. Lin et al.), 333-345;
1989, Warrendale, PA, The Minerals, Metals and Materials

277.

A. I. GUBANOV

278.

H. MENDEZ, D. FINELLO, R. WALSER,

Society.
and s. M.
1977, 19, (5), 795-797.

DUNAEVSKIi: Sov. Phys.

Solid State,

and H. L. MARCUS: Scr.


1982, 16, 855-858.
279. M. RUHLE and A. G. EVANS: Mater. Sci. Eng., 1989, AI07,
187-197.
280. W. MADER: Z. Metallkd., 1989, 80, 139-151.
281. K. C. RUSSELL, S.-Y. OH, and A. FIGUEREDO: MRS Bull., 1991,
16,46-52.
282. P. KRITSALIS, J. G. LI, L. COUDURIER, and N. EUSTATHOPOULOS:
J. Mater. Sci. Lett., 1990, 9, 1332-1335.
283. J. G. LI, D. CHATAIN, L. COUDURIER, and N. EUSTATHOPOULOS:
J. Mater. Sci. Lett., 1988, 7, 961-963.
284. I. A. AKSAY, C. E. HODGE, and J. A. PASK: J. Phys. Chem., 1974,
78, 1178-1183.
285. N. SUZUKI, K. TANAKA, and M. YAMANASHI: UK Pat. Appl.
GB 2 162 104 A, 1986.
286. E. TANK: US Pat. 4 943 413, 1990.
287. E. KLIER, A. MORTENSEN, J. A. CORNIE, and M. C. FLEMINGS: US
Pat. 4961 461, 1990.
288. J. T. BURKE: European Pat. Appl. 0368788, 1990.
289. F. v. LENEL: 'Powder metallurgy', 313-319; 1980, Princeton,
NJ, Metal Powder Industries Federation.
290. A. J. SHALER: Int. J. Powder Metall., 1965, 1, 3-14.
291. K. A. SEMLAK and F. N. RHINES: Trans. AIME, 1958, 212,
325-331.
292. J. LEZANSKI: Arch. Hutn., 1989,34, 93-108.
293. G. IBE, W. GRUHL, and H. GEBHARDT: in 'Faserverbundwerkstofi'e', (ed. W. Bunk and J. Hansen), 27-74; 1979, K61n,
Springer- Verlag.
294. G. P. MARTINS, D. L. OLSON, and G. R. EDWARDS: Metall. Trans.,
1988, 19B, 95-101.
295. A. E. SCHEIDEGGER: 'The physics of flow through porous media',
3 edn, 124-151; 1974, University of Toronto Press.
296. H. J. MOREL-SEYTOUX: in 'Flow through porous media', (ed.
R. 1. M. De Wiest), 455-516; 1969, New York, Academic
Press.
297. s. ERGUN: Chem. Eng. Prog., 1952,48, (2) 89-94.
298. v. STANEK and J. SZEKELY: AIChE J., 1974,20, (5), 974-980.
299. A. E. SCHEIDEGGER: 'The physics of flow through porous media',
3 edn, 152-167; 1974, University of Toronto Press.
300. R. B. BIRD, W. E. STEWART, and E. N. LIGHTFOOT: 'Transport
phenomena', 196-200; 1960, New York, Wiley.
301. I. F. MacDONALD, M. S. EL-SAYED, K. MOW, and F. A. L. DULLIEN:
Ind. Eng. Chem. Fund., 1979, 18, 199-208.
302. J. SZEKELY: 'Fluid flow phenomena in metals processing',
268-269; 1979, New York, Academic Press.
303. T. W. CLYNE and J. F. MASON: Metall. Trans., 1987, 18A, 15191530.
304. T. W. CLYNE and M. G. BADER: in Proc. 5th Int. Conf. on
'Composite materials ICCM 5', (ed. W. C. Harrigan et al.),
755-771; 1985, Warrendale, PA, Metallurgical Society of
AIME.
305. H. FUKUNAGA and K. GODA: J. Jpn Inst. Met., 1985,49, 78-83.
306. H. FUKUNAGA: in 'Advances in cast reinforced metal composites', (ed. S. G. Fishman and A. K. Dhingra), 101-107; 1988,
Metals Park, OH, ASM International.
307. H. FUKUNAGA and K. GODA: Bull. JSME, 1984,27, 1245-1250.
308. J. M. QUENISSET, R. FEDOU, F. GIROT, and Y. L. PETITCORPS: in
'Advances in cast reinforced metal composites', (ed.
S. G. Fishman and A. K. Dhingra), 133-138; 1988, Metals
Park, OH, ASM International.
309. F. A. GIROT, R. FEDOU, J. M. QUENISSET, and R. NASLAIN: J. Rein.
Plast. Compos., 1990, 9, 456-469.
310. E. LACOSTE, M. DANIS, F. GIROT, and J. M. QUENISSET: Mater.
Sci. Eng., 1991, A135, 45-49.
311. A. MORTENSEN andv. MICHAUD: Metall: Trans., 1990, 21A,
2059-2072.
312. L. J. MASUR, A. MORTENSEN, J. A. CORNIE, and M. C. FLEMINGS:
in Proc. 6th Int. Conf. on 'Composite materials ICCM 6', (ed.
F. L. Matthews et al.),2.320-2.329;
1987, London, Elsevier.
313. R. M. ANDREWS and A. MORTENSEN: Metall. Trans., 1991, 22A,
2903-2915.
Metall.,

International

Materials Reviews

1992

Vol. 37

NO.3

314.
315.

G. A. CHADWICK: Mater. Sci. Eng., 1991, A135, 23-28.


A. E. SCHEIDEGGER: 'The physics of flow through porous

media',
3 edn, 247-290; 1974, University of Toronto Press.
316. F. A. L. DULLIEN: 'Porous media - fluid transport and pore
structure', 300-306; 1979, New York, Academic Press.

317.

J. R. PHILIP:

318.

J. BEAR:

Soil Sci., 1956,83,345-357.

'Dynamics of fluids in porous media', 221-222; 1972,


New York, American Elsevier.
319. J. BEAR: 'Dynamics of fluids in porous media', 303 and 519;
1972, New York, American Elsevier.
320. M. A. BlOT: J. Appl. Phys., 1955, 26, 182-185.
321. M. A. BlOT: J. Appl. Mech., 1957, 24, 594-601.
322. X. LI, O. C. ZIENKIEWICZ, and Y. M. XIE: Int. J. Numer. Meth.
Eng., 1990, 30, 1195-1212.
323. G. S. BEAVERS, A. HAJII, and E. M. SPARROW: J. Fluids Eng.,
1981, 103, 432-439.
324. D. E. KENYON: Arch. Ration. Mech. Anal., 1976,62, 131-147.
325. D. E. KENYON: Arch. Ration. Mech. Anal., 1976,62, 117-130.
326. J. R. RICE and M. P. CLEARY: Rev. Geophys. Space Phys., 1976,
14, 227-241.
327. M. K. HUBBERT and w. W. RUBEY: J. Geol. Soc. Am., 1959, 70,
115-166.
328. s. K. GARG and A. NUR: J. Geophys. Res., 1973, 78, 59115921.
329. A. NUR and J. D. BYERLEE: J. Geophys. Res., 1971, 76, 64146419.
330. T. IMAI, Y. NISHIDA, M. YAMADA, H. MATSUBARA, and
I. SHIRAYANAGI: J. Mater. Sci. Lett., 1987, 6, 343-345.
331. Y. NISHIDA, H. MATSUBARA, M. YAMADA, and T. IMAI: in Proc.
4th Japan-U.S. Conf. on 'Composite materials', Washington,
DC, 1988,429-438; Lancaster, PA, Technomic Publishing.
332. N. W. RASMUSSEN, P. N. HANSEN, and s. F. HANSEN: Mater. Sci.
Eng., 1991, A135, 41-43.
333. N. W. RASMUSSEN, P. N. HANSEN, and s. F. HANSEN: in 'Metal
matrix composites - processing, microstructure and properties', 12th Ris0 Int. Symp. on Materials Science, Ris0
National Laboratory,
Roskilde, Denmark, 1991, (ed.
N. Hansen et al.), 617-620.
334. H. FUKUNAGA: in 'Advances in cast reinforced metal composites', (ed. S. G. Fishman and A. K. Dhingra), 101-107, 1988,
Metals Park, OH, ASM International.
335. K. BAN, T. ARAI, T. SAKAKIBARA, and N. MIYAKE: US Pat.
4506721, 1985.
336. J. L. SOMMER and A. MORTENSEN: unpublished work, MIT,
1991.
337. S. NAGATA and K. MATSUDA: Imono (J. Jpn Foundrymen's Soc.),
1981, 53, 300-304.
338. S. NAGATA and K. MATSUDA: Trans. Jpn Foundrymen's Soc.,
1983, 2, 616-620.
339. Ph. JARRY, A. DUBUS, G. REGAZZONI, V. MICHAUD, and
A. MORTENSEN: in Proc. of F- Weinberg Int. Symp. on 'Solidification processing', (ed. 1. E. Lait and I. V. Samarasekera),
195-204; 1990, New York, Pergamon Press.
340. R. B. CALHOUN and A. MORTENSEN: Metall. Trans. A, in
press.
341. v. J. MICHAUD: PhD, thesis, MIT, 1991.
342. v. J. MICHAUD and A. MORTENSEN: Metall. Trans., in press.
343. P. JARRY, V. J. MICHAUD, A. MORTENSEN, A. DUBUS, and
R. TIRARD-COLLET: Metall. Trans. A, in press.
344. H. K. MOON: PhD thesis, MIT, 1990.
345. R. MEHRABIAN, R. G. RIEK, and M. C. FLEMINGS: Metall. Trans.,
1974,5, 1899-1905.
346. F. GIROT: These de Docteur, Universite de Bordeaux I,
1987.
347. T. Z. KATTAMIS and J. A. CORNIE: in 'Advances in cast reinforced
metal composites', (ed. S. G. Fishman and A. K. Dhingra),
47-51; 1988, Metals Park, OH, ASM International.
348. w. R. LOuE and w. H. KooL: in Extended Abstracts of Conf.
'Metal matrix composites: property optimization and applications', London, 1989, The Institute of Metals.
349. M. MADA and F. AJERSCH: in 'Metal and ceramic composites:
Processing, modeling and mechanical behavior', (ed.
R. B. Bhagat et al.), 337-350; 1990, Warrendale, PA, The
Minerals, Metals and Materials Society.
350. H. K. MOON, J. A. CORNIE, and M. C. FLEMINGS: Mater. Sci.
Eng., 1991, AI44, 253-265.
351. L. S. TURNG and K. K. WANG: J. Mater. Sci., 1991, 26, 21732183.
352. D. J. LLOYD: Compos. Sci. Technol., 1989,35,159-179.

Mortensen

and Jin

353. D. LLOYD:in 'Metal matrix composites - processing, microstructure and properties', 12th Rise Int. Symp. on Materials
Science, Rise National Laboratory, Roskilde, Denmark, 1991,
(ed. N. Hansen et al.), 81-99.
354. M. K. SURAPPAand P. K. ROHATGI:Metall. Trans., 1981, 12B,
327.
355. DEONATHand P. K. ROHATGI:J. Mater. Sci., 1980, 15, 27772784.
356. w. R. HOOVER:in 'Fabrication
of particulates reinforced
metal composites', (ed. J. Masounave et al.), 115-123; 1990,
Materials Park, OH, ASM International.
357. s. KENNERKNECHT:
in 'Fabrication of particulates reinforced
metal composites', (ed. J. Masounave et al.), 87-100; 1990,
Materials Park, OH, ASM International.
358. w. R. HOOVER:in 'Metal matrix composites - processing,
microstructure
and properties', 12th Rise Int. Symp. on
Materials Science, Rise National Laboratory,
Roskilde,
Denmark, 1991, (ed. N. Hansen et al.), 387-392.
359. M. C. FLEMINGS,R. G. RIEK,and K. P. YOUNG:Int. Cast Met. J.,
1976, 1, (3), 11-22.
360. S. TAKAHASHI:
in Proc. 5th Int. Conf. on 'Composite materials,
ICCM 5', (ed. W. C. Harrigan et al.), 747-754; 1985, Warrendale, PA, Metallurgical Society of AIME.
361. C. MILLIEREand M. SuERY:in 'Advanced materials: Research
and developments for transport', (ed. P. Lamicq), 241-248;
1985, Paris, Les Editions de Physique.
362. R. M. PILLAI,V. S. KELUKUTTY,R. UPADAHYAYA,
B. C. PAl, and
K. G. SATYANARAYANA:
in 'Principles of solidification and
materials processing', (ed. R. Trivedi et al.), 868-875; 1989,
New Delhi, India, Oxford and IBH Publishing Co. Pvt. Ltd.
363. M. A. BAYOUMIand M. SuERY:in 'Advances in cast reinforced
metal composites', (ed. S. G. Fishman and A. K. Dhingra),
167-72; 1988, Metals Park, OH, ASM International.
364. R. SASIKUMAR
and B. C. PAl:in 'Solidification processing 1987',
451-453; 1988, London, The Institute of Metals.
365. R. SASIKUMAR
and B. C. PAl:in 'Principles of solidification and
materials processing', (ed. R. Trivedi et al.), 851-857; 1989,
New Delhi, India, Oxford and IBH Publishing Co., Pvt. Ltd.
366. L. LAJOYEand M. SuERYIN:in 'Advances in cast reinforced
metal composites', (ed. S. G. Fishman and A. K. Dhingra),
15-20; 1988, Metals Park, OH, ASM International.
367. M. SuERY and L. LAJOYE:in 'Solidification of metal matrix
composites', (ed. P. K. Rohatgi), 171-179; 1990, Warrendale,
PA, The Minerals, Metals an~ Materials Society.
368. M. S. MISRAand s. G. FISHMAN:in Proc. 8th Int. Conf. on
'Composite materials, ICCM 8', (ed. S. W. Tsai and
G. S. Springer), 18AI-18AI2; 1991, Covina, CA, SAMPE.
369. C. VAIDYANATHAN,
R. ROHRMAYER,
and c. R. LOPER:in Proc.
8th Int. Conf. on 'Composite materials, ICCM 8', (ed.
S. W. Tsai and G. S. Springer), 1701-17010;
1991, Covina,
CA, SAMPE.
370. M. N. GUNGOR, A. H. NAKAGAWA,S. D. KARMARKAR,and
A. P. DIVECHA:in Proc. 8th Int. Conf. on 'Composite materials,
ICCM 8', (ed. S. W. Tsai and G. S. Springer), 18GI-18GI5,
Covina, CA, SAMPE.
371. B. N. KESHAVARAM,P. K. ROHATGI, R. ASTHANA, and
K. G. SATYANARAYANA:
in 'Solidification of metal matrix composites', (ed. P. K. Rohatgi), 151-169; 1990, Warrendale, PA,
The Minerals, Metals and Materials Society.
372. H. FUKUNAGAand K. GODA:Bull. Jpn Soc. Mech. Eng., Jan.
1985, 28, 1.
373. Y. ABE,M. NAKATANI,K. YAMATSUTA,
and s. HORIKIRI:in Proc.
1st European Conf. on Composite Materials, ECCM-l,
'Developments in the science and technology of composite
materials', Bordeaux, 1985, (ed. A. R. Bunsell et al.), 604.
374. Y. ABE,S. ;H0RIKIRI,K. FUJIMURA,and E. ICHIKI:in 'Progress
in science and engineering of composites, ICCM 4', (ed.
T. Hayashi), Tokyo, 1982, The Japan Society for Composite
Materials, 1427.
375. T. DONOMOTO,
A. TANAKA,Y. TATEMATSU,
and T. AKAI:US Pat.
4450207, 1984.
376. T. DONOMOTO
and A. TANAKA:Pat. EPO 164536 (EUR), 1985,
cited in Met. Abst., July 1986, abstract no. 62-0217.128.
377. Y. KAGAWAand B. H. CHOI:in Proc. Conf. 'Composites '86:
Recent advances in Japan and the United States, CCM-III',
(ed. K. Kawata et al.), The Japan Society for Composite
Materials, Tokyo, 1986, 537.
378. A. MORTENSEN,
M. N. GUNGOR,J. A. CORNIE,and M. C. FLEMINGS:
J.Met.,
1986, 38, (3), 30.

Solidification

processing of metal matrix composites

127

379. B. R. HENRIKSEN,J. GJ0NNES,and H. MATHIESEN:


in 'Mechanical
and physical behaviour of metallic and ceramic composites',
9th Int. Symp. on Metallurgy and Materials Science, Rise
National Laboratory, Roskilde, Denmark, 1988, (ed. S. I.
Andersen et al.), 397-402.
380. I. lIN and D. J. LLOYD:in 'Fabrication of particulates reinforced
metal composites', (ed. J. Masounave et al.), 47-52; 1990,
Materials Park, OH, ASM International.
381. R. G. DIXON:MSc thesis, MIT, 1985.
382. R. T. PEPPERand R. A. PENTY:J. of Compos. Mater., 1974,8,
29-37.
383. A. MORTENSEN,
J. A. CORNIE,and M; C. FLEMINGS:Metall. Trans.,
1988, 19A, 709-721.
'
384. D. J. LLOYD,H. LAGACE,A. McLEOD,and P. L. MORRIS:Mater.
Sci. Eng., 1989, A107, 73-80.
385. G. J. DAVIES:'Solidification and casting'; 1973, New York,
Halsted Press, Wiley.
386. E. L. GLASSONand E. F. EMLEY:in 'The solidification of metals',
1-9; 1968, London, The Iron and Steel Institute.
387. N. L. BATURINSKAYA,
N. A. KAL'CHUK,M. G. SERVETSKAYA,
and
v. G. CHERNYI:Russ. Metall., 1983,3, 150-154.
388. N. L. BATURINSKAYA,
N. A. KAL'CHUK,Yu. D. SEZONENKO,and
v. G. CHERNYI:Russ. Metall., 1986,3, 145-147.
389. T. A. CHERNYSHOVA,
L. I. KOBELEVA,M. I. TYLKINA,and A. v.
REBROV:Met. Sci. Heat Treat., 1984,26, 592.
390. Y. SAWADAand M. BADER:in Proc. 1st European Conf. on
Composite Materials, ECCM-l, 'Developments in the science
and technology of composite materials', Bordeaux, 1985,
AEMC, (ed. A. R. Bunsell et al.),577.
391. M. YANGand v. D. SCOTT:J. Mater. Sci., 1991,26, 2245-2254.
392. J. D. BRYANT,L. CHRISTODOULOU,and J. R. MAISANO:Scr.
Metall., 1990, 24, 33-38.
393. J. BRACZYNSKI:Mater. Sci. Eng., 1991, A135, 105-109.
394. G. J. DAVIES:'Solidification and casting', 95-116; 1973, New
York, Halsted Press, Wiley.
395. G. S. COLEand G. F. BOLLING:Trans. AIME, 1965,233, 15681572.
396. M. C. FLEMINGS:'Solidification processing', 64-66; 1974, New
York, McGraw-Hill.
397. R. T. DELVES:in 'Crystal growth', (ed. B. R. Pamplin), 40-103;
1974, Oxford, Pergamon Press.
398. A. MORTENSEN:
in 'Solidification of metal matrix composites',
(ed. P. K. Rohatgi), 1-21; 1990, Warrendale, PA, The Minerals,
Metals and Materials Society.
399. J. A. SEKHARand R. TRIVEDI:Mater. Sci. Eng., 1989, Al14,
133-146.
400. D. SHANGGUANand J. D. HUNT: Metaii. Trans., 1991, 22A,
1683-1687.
401. N. F. DEAN,A. MORTENSEN,
and M. C. FLEMINGS:unpublished
work, MIT, 1991.
402. R. TRIVEDI,S. H. HAN, and J. A. SEKHAR:in 'Solidification of
metal matrix composites', (ed. P. K. Rohatgi), 23-37;
1990, Warrendale, PA, The Minerals, Metals and Materials
Society.
403. L. H. UNGARand R. A. BROWN:Phys. Rev. B, 1984, 30, (7),
3993-3999.
404. D. G. McCARTNEYand J. D. HUNT: Metall. Trans., 1984, 15A,
983-994.
405. D. G. McCARTNEY
and J. D. HUNT:Acta Metall., 1987,35,89-99.
405a.D. G. McCARTNEYand J. D. HUNT: Appl. Sci. Res., 1987, 44,
9-26.
406. J. D. HUNT:in 'Solidification processing 1987', 191-197; 1988,
London, The Institute of Metals.
407. L. H. UNGARand R. A. BROWN: Phys. Rev. B, 1985, 31, (9),
5931-5940.
408. T. F. BOWER,H. D. BRODY,and M. C. FLEMINGS:Trans. AIME,
1966, 236, 624-634.
409. R. TRIVEDI:in 'Solidification processing of advanced materials',
Proc. Japan-US Cooperative Science Program Seminar, 1989,
Oiso, Kanagawa, Japan, sponsored by Japan Society for
Promotion of Science and National Science Foundation,
29-38.
410. L. M. FABIETTI,V. SEETHARAMAN,
and R. TRIVEDI:Metall. Trans.,
1990, 21A, 1299-1310.
411. J. A. SEKHARand R. TRIVEDI:in 'Solidification of metal matrix
composites', (ed. P. K. Rohatgi), 39-50; 1990, Warrendale,
PA, The Minerals, Metals and Materials Society.
412. J. A. SEKHARand R. TRIVEDI:Mater. Sci. Eng., 1991, A147,
9-21.
International

Materials Reviews

1992

Vol. 37

NO.3

128
413.

Mortensen and Jin

Solidification

processing of metal matrix composites

and M. C. FLEMINGS: in 'Advances


in cast reinforced metal composites', (ed. S. G. Fishman and
A. K. Dhingra), 39-45; 1988, Metals Park, OR, ASM
International.
414. M. N. GUNGOR, J. A. CORNIE, and M. C. FLEMINGS: in 'Interfaces
M. N. GUNGOR, J. A. CORNIE,

434.

Metall.

International

Materials Reviews

1992

Vol. 37

NO.3

and

P. CURRERI:

Trans., 1988, 19A, 1899-1904.

435.

P. K. ROHATGI, F. M. YARANDI,

436.

J. W. McCOY

in metal matrix composites', (ed. A. K. Dhingra et al.),


121-135; 1986, Warrendale, PA, Metallurgical Society of
AIME.
415. S. NOURBAKHSH, H. MARGOLIN, and F. L. LIANG: in 'Solidification
of metal matrix composites', (ed. P. K. Rohatgi), 103-114;
1990, Warrendale, PA, The Minerals, Metals and Materials
Society.
416. P. K. GHOSH and S. RAY: in 'Solidification of metal matrix
composites', (ed. P. K. Rohatgi), 205-212; 1990, Warrendale,
PA, The Minerals, Metals and Materials Society.
417. B. C. PAl, S. G. K. PILLAI, R. M. PILLAI, and K. G. SATYANRAYANA:
in 'Solidification of metal matrix composites', (ed.
P. K. Rohatgi), 191-203; 1990, Warrendale, PA, The Minerals,
Metals and Materials Society.
418. R. TRUMPER and v. SCOTT: in Proc. 1st European Conf. on
Composite Materials, ECCM-1, 'Developments in the science
and technology of composite materials', Bordeaux, 1985,
AEMC, (ed. A. R. Bunsell et al.), 139-144.
419. A. R. CHAPMAN, V. D. SCOTT, and M. YANG: in Proc. 8th Int.
Conf. on 'Composite materials, ICCM 8', (ed. S. W. Tsai and
G; S. Springer), 19GI-19GI0; 1991, Covina, CA, SAMPE.
420. D.U. KIM, I.M. PARK, and J. KIM: in Proc. 8th Int. Conf. on
'Composite materials, ICCM 8', (ed. S. W. Tsai and
G. S. Springer), 17Ml-17M9; Covina, CA, SAMPE.
421. Q. F. LI, D. G. McCARTNEY, and A. M. WALKER: J. Mater. Sci.,
1991, 26, 3565-3574.
422. D. M. STEFANESCU and B. K. DHINDAW: in 'Metals handbook',
9 edn, Vol. 15, 142-147; 1988, Metals Park, OR, ASM
International.
423. D. R. UHLMANN, B. CHALMERS, and K. A. JACKSON: J. Appl.
Phys., 1964, 35, (10), 2986-2993.
424. J. CISSE and G. F. BOLLING: J. Cryst. Growth, 1971, 10, 67-76.
425. J. CISSE and G. F. BOLLING: J. Cryst. Growth, 1971, 11, 25-28.
426. A. M. ZUBKO, V. G. LOBANOV, and v. v. NIKONOVA: Sov. Phys.
Crystallogr., 1973, 18, 239-241.
427. A. A. CHERNOV and A. M. MEL'NIKOVA: Sov. Phys. Crystallogr.,
1966, 10, 672-675.
428. A. W. NEUMANN, J. SZEKELY, and E. J. RABENDA, Jr: J. Colloid
Interf Sci., 1973, 43, (3), 727-732.
429. s. N. OMENYI and A. W. NEUMANN: J. Appl. Phys., 1976, 47,
(9), 3956-3962.
430. v. L. BRONSTEIN, Y. A. ITKIN, and G. s. ISHKOV: J. Cryst.
Growth, 1981, 52, 345-349.
431-. CR. KORBER and G. RAU: J. Cryst. Growth, 1985, 72,649-662.
432. c. E. SCHVEZOV and F. WEINBERG: Metall. Trans., 1985, 16B,
367-375.
433. G. W. DELAMORE, R. W. SMITH, and w. B. F. MacKAY: AFS
Trans., 1971, 71, 560-564.

B. K. DHINDAW, A. MOITRA, D. M. STEFANESCU,

and Y. LlU: in 'Advances in cast


reinforced metal composites', (ed. S. G. Fishman and
A. K. Dhingra), 249-255; 1988, Metals Park, OR, ASM

International.

437.
438.

439.

440.
441.
442.

443.
444.
445.
446.
447.
448.
449.
450.
451.
452.

453.
454.

and F. E. WAWNER: in 'Advances in cast reinforced


metal composites', (ed. S. G. Fishman and A. K. Dhingra),
237-242; 1988, Metals Park, OR, ASM International.
D. M. STEFANESCU, B. K. DHINDAW, S. A. KACAR, and A. MOITRA:
Metall. Trans., 1988, 19A, 2847-2855.
B. K. DHINDAW, A. KACAR, and D. M. STEFANESCU: in Proc.
ASM Europe Conf. 'Advanced materials and processing tech
niques for structural applications', Paris, France, 1987,
491-500.
A. MOITRA, B. K. DHINDAW, and D. M. STEFANESCU: in 'Solidifi
cation of metal matrix composites', (ed. P. K. Rohatgi),
91-101; 1990, Warrendale, PA, The Minerals, Metals and
Materials Society.
P. K. ROHATGI: in 'Interfaces in metal matrix composites',
(ed. A. K. Dhingra et al.), 185-202; 1986, Warrendale, PA,
Metallurgical Society of AIME.
N. I. ABDUL-LATTEF, A. RAZZAQ, I. .KHEDAR, and s. K. GOEL:
J. Mater. Sci., 1987, 22, 466-472.
B. N. KESHAVARAM, P. K. ROHATGI,
R. ASTHANA, and
K. G. SATHYANARAYANA: in 'Solidification of metal matrix
composites', (ed. P. K. Rohatgi), 151-167; 1990, Warrendale,
PA, The Minerals, Metals and Materials Society.
J. D. BRYANT, J. R. MAISANO, D. T. WINTER, and A. R. H. BARRETT:
Scr. Metall., 1990, 24, 2209-2214.
J. POTSCHKE and v. ROGGE: J. Cryst. Growth, 1989,94, 726-738.
M. K. SURAPPA and P. K. ROHATGI: J. Mater. Sci. Lett., 1981,
16, (2), 562-564.
G. F. BOLLING and J. CISSE: J. Cryst. Growth, 1971, 10, 56-66.
A. A. CHERNOV, D. E. TEMKIN, and A. M. MEL'NIKOVA: Sov. Phys.
Crystallogr., 1976, 21, (4), 369-374.
D. E. TEMKIN, A. A. CHERNOV, and A. M. MEL'NIKOVA: Sov. Phys.
Crystallogr., 1977, 22, (1), 13-17.
A. A. CHERNOV, D. E. TEMKIN, and A. M. MEL'NIKOVA: Sov. Phys.
Crystallogr., 1977, 22, (6), 656-658.
R. SASIKUMAR, T. R. RAMAMOHAN, and B. C. PAl: Acta Metall.,
1989, 37, (7), 2085-2091.
P. F. AUBOURG: PhD thesis, MIT, 1978.
R. SASIKUMAR, T. R. RAMAMOHAN, and B. C. PAl: in 'Principles
of solidification and materials processing', (ed. R. Trivedi
et al.), 843-850; 1989, New Delhi, India, Oxford and IBH
Publishing Co., Pvt. Ltd.
R. SASIKUMAR and T. R. RAMAMOHAN: Acta Metall. Mater.,
1991, 39, 517-522.
L. GOUMIRI and J. c. JOUD: Acta Metall., 1982,30, 1397-1405.

Das könnte Ihnen auch gefallen