Sunteți pe pagina 1din 11

Acta mater.

49 (2001) 32193229
www.elsevier.com/locate/actamat

CORRELATION BETWEEN TENSILE AND INDENTATION


BEHAVIOR OF PARTICLE-REINFORCED METAL MATRIX
COMPOSITES: AN EXPERIMENTAL AND NUMERICAL STUDY
Y. -L. SHEN1, J. J. WILLIAMS2, G. PIOTROWSKI2, N. CHAWLA2 and Y. L. GUO1
1

Department of Mechanical Engineering, The University of New Mexico, Albuquerque, NM 87131, USA
and 2Department of Chemical and Materials Engineering, Arizona State University, Tempe, AZ 852876006, USA
( Received 16 April 2001; received in revised form 30 May 2001; accepted 30 May 2001 )

AbstractThe correlation between tensile and indentation behavior in particle-reinforced metal matrix composites (MMCs) was examined. The model composite system consists of a AlCuMg alloy matrix reinforced
with SiC particles. The effects of particle size, particle volume fraction, and matrix aging characteristics on
the interrelationship between tensile strength and macro-hardness were investigated. Experimental data indicated that, contrary to what has been documented for a variety of monolithic metals and alloys, a simple
relationship between hardness and tensile strength does not exist for MMCs. While processing-induced particle fracture greatly reduces the tensile strength, it does not significantly affect the deformation under indentation loading. Even in composites where processing-induced fracture was nonexistent (due to relatively small
particle size), no unique correspondence between tensile strength and hardness was observed. At very low
matrix strengths, the composites exhibited similar tensile strengths but the hardness increased with increasing
particle concentration. Fractographic analyses showed that particle fracture caused by tensile testing is independent of matrix strength. The lack of unique strengthhardness correlation is not due to the particle fractureinduced weakening during the tensile test. It is proposed that, under indentation loading, enhanced matrix
flow that contributes to a localized increase in particle concentration directly below the indenter results in a
significant overestimation of the overall composite strength by the hardness test. Micromechanical modeling
using the finite element method was used to illustrate the proposed mechanisms under indentation loading
and to justify the experimental findings. 2001 Acta Materialia Inc. Published by Elsevier Science Ltd. All
rights reserved.
Keywords: Composites; Aluminum; Hardness testing; Tensile testing; Computer simulation

1. INTRODUCTION

Metal matrix composites (MMCs) are attractive


replacements for many conventional monolithic
alloys because of their higher strength-to-weight ratio
and fatigue resistance. A lightweight metallic matrix
with particulate reinforcement, in particular, can be
manufactured using slight modifications in conventional metallic processing. Such composites are therefore more affordable and exhibit more isotropic
properties than their continuous fiber reinforced
counterparts. In monolithic metallic materials, the
macroscopic indentation response, typically obtained
through a hardness test, is routinely correlated with
the mechanical strength. Although hardness in itself is
not a well-defined material parameter, the correlation
between various hardness scales and tensile strength

To whom all correspondence should be addressed. Tel.:


+1-505-277-6286; fax: +1-505-277-1571.
E-mail address: shenyl@me.unm.edu (Y. -L. Shen)

has been compiled for a variety of metals and alloys


[1]. With the increasing use of particle-reinforced
MMCs in structural applications, the macroscopic
indentation response and its relation to tensile
strength of MMCs is of great practical interest.
While several studies have used indentation, in the
form of macro- and micro-hardness measurements, to
study the matrix strength characteristics in MMCs [2
8], there seems to be very little understanding of the
correlation between macro-hardness and overall
strength of the composite. A recent study of the aging
response of a 6061/SiC/10p aluminum composite [9]
showed a linear relationship between the Rockwell

We follow the standard notation for metallic composites, designated by the Aluminum Association. The
matrix alloy is followed by the reinforcement composition
and reinforcement volume fraction of the composite. The
latter is denoted as a particulate reinforcement by the
subscript p. Thus, a 2080 alloy matrix reinforced with
20% SiC particles would be denoted as 2080/SiC/20p.

1359-6454/01/$20.00 2001 Acta Materialia Inc. Published by Elsevier Science Ltd. All rights reserved.
PII: S 1 3 5 9 - 6 4 5 4 ( 0 1 ) 0 0 2 2 6 - 9

3220

SHEN et al.: PARTICLE-REINFORCED COMPOSITES

superficial-scale hardness and the ultimate tensile


strength of the composite. The study did not consider
the effects of reinforcement volume fraction and particle size which, however, certainly play different
roles during indentation and tensile loading.
The tensile behavior of particle-reinforced MMCs,
unlike indentation response, has been extensively
studied and reviewed [1012]. Since the reinforcing
phase is typically much stiffer than the matrix, a significant fraction of the stress is initially borne by the
reinforcement. The incorporation of particles in the
matrix results in an increase in work hardening of the
composites due to lower relative volume of metal and
geometric constraints imposed by the presence of the
reinforcement. When the matrix is significantly work
hardened, the matrix is placed under great constraint
with an inability for strain relaxation to take place.
This causes the onset of void nucleation and propagation, which take place at a lower far field applied
strain than that observed in the unreinforced material.
With an increase in reinforcement volume fraction,
higher elastic modulus, macroscopic yield and tensile
strengths have typically been observed, coupled with
lower ductility. Cracked particles in the composite,
which may result from processing of composites with
fairly coarse particulate reinforcement, do not contribute to load transfer or strengthening and result in
lower strength. The high stress concentration at the
tips of the cracks would also contribute to a lower
ductility in the composite, compared to the unreinforced alloy. The effect of particle size on tensile
behavior, documented by several investigators [13
17], indicates an increase in work hardening rate, tensile strength and ductility with a decrease in particle
size. In addition, a smaller particle size leads to a
lower propensity of processing-induced particle fracture, since a particle with a smaller volume has a
lower probability of containing strength-limiting flaw.
This also contributes to an increase in composite
strengthening.
In the current study, we seek to explore the indentation behavior in detail and compare it to tensile
behavior. We have carried out experimental and
numerical investigations on a 2080/SiCp composite
system, by examining the effects of reinforcement
volume fraction, particle size, and matrix aging
characteristics. Our study is devoted to macroscopic
indentation where the indented area is significantly
greater than the reinforcement particle size. It will be
shown that the hardnessstrength relationship in particle-reinforced metal matrix composites is not as
straightforward as the behavior exhibited by conventional monolithic materials. We will illustrate that, for
SiC particle-reinforced Al matrix composites, similar
macro-hardness for two different materials may not
correlate to the same overall strength, and vice versa.
It will be shown that reinforcement fraction, particle
size, and matrix strength play very different roles during indentation and tensile loading.

2. MATERIALS AND EXPERIMENTAL PROCEDURE

An AlCuMg alloy (2080 Al, age-hardenable,


with composition 3.6 Cu, 1.9 Mg, 0.25 Zr), unreinforced and reinforced with 10 and 20 vol.% SiC
particles, were used in this study. All materials were
processed by a powder metallurgy route and extruded
(Alcoa Technical Center, Alcoa, PA). The average
reinforcement particle size, after extrusion, was about
6 m (designated as F600). Composites containing
larger particles, with an average size of about 23 m
after extrusion (F280), were also examined. The designations F600 and F280 are henceforth used to specify the reinforcement size in this paper. A detailed
characterization of the reinforcement size, distribution
and morphology can be found elsewhere [13, 18].
Extensive particle cracking in the as-processed condition was observed in the F280 composites, but not
in the F600 composites.
A T8 thermomechanical treatment [19] was applied
to the composites and the unreinforced alloy. The
materials were solution treated at 493C, quenched in
water, cold rolled to 5% reduction in thickness, and
aged at 175C for 24 h (to the peak-age condition).
The rolling step provided a homogeneous distribution
of dislocations that serve as sites for heterogeneous
nucleation of precipitates. This resulted in nearly
identical microstructures in the composite and the
monolithic alloy [20]. Thus, indirect strengthening
of the matrix of the composite, due to thermallyinduced dislocations, was homogenized. Peak-aged
materials were also heat-treated to the over-aged conditions at 200, 225, or 250C for 24 h.
Low-stress grinding was used to machine the cylindrical tensile test specimens. The gage diameter of
the specimens was 5.08 mm and an extensometer with
a gage length of 12.7 mm was used. Tensile tests
were conducted on a servohydraulic load frame at a
strain rate of 103/s. The standard Rockwell B-scale
test, featuring a 1.588 mm diameter spherical steel
indenter, was used to characterize the hardness [21].
The Rockwell B-scale hardness test is commonly
employed for a wide variety of Al alloys, due to its
speed, reproducibility, and relatively small size of the
indentation. A minor and major load of 10 and 100
kg, respectively, were imposed on the LT-plane of
the heat-treated material. L and T designate the longitudinal and transverse directions, respectively. The
hardness value (HRB) was directly determined by the
depth of indentation beyond the minor load. While
the Rockwell hardness scale was used in the measurements, other hardness scales are expected to yield
qualitatively similar results to those presented here.
Optical and scanning electron microscopy were used
to examine the deformation and damage characteristics of the tensile and hardness tested specimens. In
order to quantify the extent of particle fracture and
particle pullout during tensile loading, both mating
fracture surfaces were examined. Thus, a given particle found on both mating fracture surfaces was con-

SHEN et al.: PARTICLE-REINFORCED COMPOSITES

Fig. 1. True stresstrue strain curves of the indicated composites and the monolithic 2080 alloy. All materials were
treated to the T8 peak-age state.

sidered to be a fractured particle, whereas a particle


found on only one of the mating fracture surfaces was
considered to be pulled out. A statistically significant
number of particles was counted and the fraction of
cracked particles was determined for each aging condition of the 2080/SiC/20p composites.
3. EFFECTS OF REINFORCEMENT PARTICLE SIZE

We first present some observations on the effects


of reinforcement particle size on the mechanical
properties. Figure 1 shows the true stresstrue strain
behavior of 2080/SiC/20p (F280) and 2080/SiC/20p
(F600) composites along with the unreinforced 2080
alloy. All materials discussed in this section were
given a T8 peak-age treatment. A clear increase in
elastic modulus, over the unreinforced alloy, was
observed with the addition of SiC. In the composite
containing large SiC particles (F280), however, the
yield strength and ultimate tensile strength were significantly lower compared to the composite with
small particles (F600). This can be attributed to the
higher propensity of particle fracture of the F280
composite. This phenomenon and its influence in
degrading the composite mechanical properties have
been documented for the composite used in the current study [13] and in other Al/SiCp systems as well
[2225].
A comparison of hardness, 0.2% offset yield
strength and ultimate tensile strength is shown in
Table 1. While 2080/SiC/20p (F280) had much lower
yield strength and tensile strength, its hardness was
comparable to that of the other composite with a
smaller reinforcement size (F600). Compared to the

3221

Fig. 2. Optical micrograph of the 2080/SiC/20p (F600) composite taken near the indentation site.

unreinforced material, the 2080/SiC/20p (F280) composite also exhibits lower yield strength and tensile
strength, but higher hardness. Clearly, while particle
size had a significant effect on tensile behavior, it did
not significantly influence the hardness of the composites.
Figures 2 and 3 show the microstructures of
2080/SiC/20p (F600) and 2080/SiC/20p (F280),
respectively, after indentation. The region directly
below the indentation is shown. It is evident that,
while the F600 composite has very few fractured SiC
particles, abundant fractured particles are seen
throughout the F280 composite. It should be noted
that these fractured particles were not induced by the
indentation, but were existent prior to the hardness

Fig. 3. Optical micrograph of the 2080/SiC/20p (F280) composite taken near the indentation site. Some fractured particles
are highlighted. The extrusion direction is horizontal.

Table 1. Measured yield strength, ultimate tensile strength and hardness of the peak-aged materials

Yield strength (MPa)


Ultimate strength (MPa)
Hardness (HRB)

2080

2080/SiC/20p (F600)

2080/SiC/20p (F280)

503
554
83

521
563
90

461
485
88

3222

SHEN et al.: PARTICLE-REINFORCED COMPOSITES

test (due to the extrusion process). The propensity of


fractured particles is the same in regions directly
underneath the indentation as well as away from the
indentation. This implies that indentation does not
induce particle fracture and that, in the composites
investigated here, the resistance to indentation is not
sensitive to the main damage mechanism during tensile testing, i.e., fracture of the particulate reinforcement.
The discrepancy between hardness and tensile
strength shown in this section can be understood by
the fundamental difference in the two types of loading
[26]. In a hardness test the localized pressure causes
the material directly under the indented surface to
support most of the load. Thus, the stress field directly
below the indenter is predominately compressive with
the matrix under a severe triaxial pressure. This is in
contrast to deformation in tension where the entire
specimen is under nominal tensile loading. When the
composite is subject to tensile loading, the precracked particles cannot carry the tensile stress and
thus will separate, causing weakening of the composite. With increasing tensile loading, the propensity
for particle fracture also increases. In the hardness
test, however, a pre-cracked particle can still support
the compressive traction across the crack surface.
Plastic flow of the matrix between the particles
accommodates the deformation caused by the indentation pressure, so that the particles are pushed into
the material. Thus, the material response under indentation is not significantly affected by the processinginduced fractured particles, so the F280 composite
exhibits similar hardness to F600 composite, but
lower tensile strength. In this manner, the hardness
test can significantly overestimate the overall tensile
properties of the composite, particularly for composites with relatively large particle size.
4. EFFECTS OF PARTICLE VOLUME FRACTION
AND MATRIX STRENGTH

For the purpose of investigating the effects of


reinforcement volume fraction and matrix strength on
the hardnessstrength correlation, we eliminate the
influence of pre-existing particle fracture by focusing
on the 2080/SiCp (F600) composites exposed to various aging conditions. Figure 4 shows the true ultimate
tensile strength as a function of hardness (HRB) for
the composites 2080/SiC/10p and 2080/SiC/20p and
the monolithic 2080 alloy. The points in the upperright end of the three curves correspond to the peakaged materials, showing the increase in tensile
strength and hardness with increasing reinforcement
volume fraction. As expected, the tensile strength and
hardness of the composites decrease significantly with
the severity of over-aging treatment. The most
important finding shown in Fig. 4 is that the data do
not follow a simple relation as with monolithic metals. With decreasing matrix strength (increasing overaging) the difference in tensile strength of the three

Fig. 4. Relationship between true ultimate tensile strength and


Rockwell B-scale hardness of the composites and monolithic
alloy. Data from the peak-aged and over-aged states are indicated.

materials is reduced, but the disparity in hardness


increases significantly. In the over-aged condition, the
composite with a higher reinforcement fraction shows
a much greater hardness value, although its tensile
strength is comparable to those with lower particle
fractions. In other words, when the strength of the Al
matrix is relatively low, a wide range of hardness
exists even when all materials exhibited similar tensile strength. The similar tensile strengths, despite
varying volume fractions of SiC, indicate that in the
over-aged condition, the composite tensile strength is
matrix dominated. On the other hand, the hardness
of the composite in the over-aged conditions is more
significantly influenced by the volume fraction of
reinforcement.
Figure 5 shows the true yield strength (0.2% offset)
as a function of hardness for all the materials and
aging treatments included in Fig. 4. The trends for
yield strength and tensile strength are similar. Once
again, the data do not follow a simple relation.
It is noted that, aside from the effects of plastic

Fig. 5. Relationship between true 0.2% offset yield strength


and Rockwell B-scale hardness of the composites and monolithic alloy. Data from the peak-aged and over-aged states are
indicated.

SHEN et al.: PARTICLE-REINFORCED COMPOSITES

flow strength, the hardness disparity caused by overaging observed in Figs 4 and 5 is not due to the higher
elastic modulus of the composite with increasing
reinforcement concentration, because aging treatment
does not alter the elastic properties of Al matrix in
the respective composites. The possible reasons for
the different mechanisms under tensile and indentation loading are addressed in detail in the following
two sections. It is quite clear, however, that a unique
relationship between hardness and overall composite
strength does not exist in particle-reinforced aluminum matrix composites, especially when the matrix
strength is relatively low.
5. THE ROLE OF PARTICLE FRACTURE DURING
TENSILE LOADING

In this section we concentrate on the role of particle


fracture during tensile damage in 2080/SiC/20p composites with F600 and F280 particle sizes, as a function of matrix strength. Implications to the strength
hardness relationship are also discussed.
Figures 6(a) and (b) show the scanning electron
micrographs of the tensile fracture surfaces of

3223

2080/SiC/20p (F600) in the peak-aged and over-aged


(24 h at 225C beyond peak-age) conditions, respectively. In each case the two matching halves are shown
to adequately determine the extent of particle fracture.
Note that the SiC particles exhibit characteristic
brittle fracture, surrounded by the ductile dimple fracture in the surrounding matrix. While most of the particles appear to be fractured, a small fraction of particles seem to have been pulled-out from the matrix.
Figures 7(a) and (b) show the tensile fracture surfaces
of the composites 2080/SiC/20p (F280) of, respectively, the peak-aged condition and the over-aged (24
h at 250C beyond peak-age) condition. Compared to
Fig. 6, the large particle size for the F280 composites
is evident. Although some fractured particles already
existed in the F280 composites before the tensile test,
the basic features revealed in Fig. 7 are similar to
those in Fig. 6, except that the matrix fracture shows
a somewhat less ductile characteristic in the F280
composites.
The critical criterion that determines tensile loading-induced particle fracture or particle pull-out is the
relationship between particle strength and
particle/matrix interface strength. If the particle

Fig. 6. Scanning electron micrographs of the fracture surface after tensile tests for the composites 2080/SiC/20p
(F600) under (a) the T8 peak-aged condition and (b) the over-aged (24 h at 225C beyond peak-age) condition.
The matching two halves of the fracture surface are shown for each case.

3224

SHEN et al.: PARTICLE-REINFORCED COMPOSITES

Fig. 7. Scanning electron micrographs of the fracture surface after tensile tests for the composites 2080/SiC/20p
(F280) under (a) the T8 peak-aged condition and (b) the over-aged (24 h at 250C beyond peak-age) condition.
The matching two halves of the fracture surface are shown for each case.

strength is higher, then particle pull-out will be the


predominant mode of fracture. On the other hand, if
the interface strength is higher, then particle fracture
will take place prior to interface failure. Table 2
shows the fraction of particle pull-out on the tensile
fracture surface for peak-aged and over-aged composites for both particle sizes. (Particles that were not
pulled out were fractured.) Compared to F600, the
F280 composites show a slightly lower percentage of
pull-out, or higher percentages of particle fracture.
The fraction of pulled out particles in F280 composites remains essentially independent of the aging
condition. For the 2080/SiC/20p (F600) composites,
the fraction of pulled out particles is slightly higher
than that of the F280 composites, and shows a very
modest increase with decreasing matrix strength.

With a decrease in particle size, the fraction of particles with a strength higher than that of the interface
increases, so a higher fraction of particles are pulled
out. This result seems to indicate that the mechanisms
in the matrix, and not particularly at the interface, are
controlling the tensile strength of the composite, in
agreement with the tensile strength data described in
the previous section. Furthermore, a decrease in
matrix strength due to over-aging does not seem to
affect the interfacial strength between particle and
matrix. Thus, the ease of void nucleation in the matrix
but not at the interface of the composite, increases
with a decrease in matrix strength due to the high
triaxial tensile stress imposed by the surrounding particles. Void nucleation and coalescence have been
proposed to take place at lower far field applied

Table 2. Statistics of particle pull-out observed on the tensile fracture surface of 2080/SiCp composites
Material
2080/SiC/20p
2080/SiC/20p
2080/SiC/20p
2080/SiC/20p
2080/SiC/20p
2080/SiC/20p
2080/SiC/20p

Treatment
(F600)
(F600)
(F600)
(F280)
(F280)
(F280)
(F280)

T8-peak-age
Peak+200C 24
Peak+225C 24
T8-peak-age
Peak+200C 24
Peak+225C 24
Peak+250C 24

h
h
h
h
h

Particle pull-out (%)

Number of particles counted

7.80
8.48
9.53
5.70
5.10
5.63
5.50

321
224
294
140
198
231
217

SHEN et al.: PARTICLE-REINFORCED COMPOSITES

strains than in the unreinforced material [15, 27].


With increased over-aging, the tendency for particles
to fracture decreases slightly in the F600 composites.
It should be noted that the fraction of fractured particles decreases significantly with an increase in distance from the fracture surface [28]. This can be attributed to the large local plastic strain on the fracture
plane [29]. A slight overestimation of the extent of
particle fracture may take place since the crack takes
a path of least resistance resulting in a non-planar
fracture surface. Recent results from polished crosssections taken from the center of the gage section of
the tensile specimens, however, indicate that while
the extent of particle fracture is significantly reduced
immediately below and with increasing distance from
the fracture plane, the trends are consistent with those
shown in Table 2 [28].
Since the variation of tensile strength with aging
is controlled by matrix strength (especially when the
matrix strength is low), the disparity of hardness
observed in Fig. 4 for the over-aged composites
should therefore be attributed to the reinforcement
itself. This is treated in the following section, where
the particle distribution under indentation is studied.

3225

can contribute to volumetric contraction of the metal


matrix. As the indenter moves downward during the
test, the pressure is accommodated by the non-uniform matrix flow (see below) along with the localized
increase of particle concentration, which tends to
increase the resistance to deformation. Consequently,
the hardness value increases due to the local increase
in particle concentration associated with indentation.
If there is an intrinsic correlation between hardness
and strength for the metal matrix, then, for the composite, the hardness value will tend to overestimate
the measured tensile strength. In particular, with
decreasing matrix strength the effect of reinforcement
becomes more prominent, as revealed in Figs 4 and 5.
Numerical modeling presented here supports the
mechanisms proposed above. A schematic depicting
our computational approach is shown in Fig. 9. A
square computational domain containing squareshaped particles was used as a two-dimensional
model system. The particles were assumed to be linearly elastic, and the matrix elasticplastic. The
modeling consisted of two steps. A uniform stretching
was first simulated to obtain the overall stressstrain

6. REINFORCEMENT CONFIGURATION UNDER


INDENTATION: NUMERICAL MODELING

The lack of unique correspondence between hardness and tensile strength is due to the fundamental
difference in deformation caused by the two modes
of loading. During a tensile (or compressive) test, the
material within the gage section undergoes nominally
uniform deformation. In a hardness test, however,
severe plastic flow is concentrated in the localized
region directly below the indentation, outside of
which the material still behaves elastically [30].
Directly below the indentation the density of particles
is increased locally, compared to regions away from
the depression. This is schematically shown in Fig. 8
[31]. Note that although plastic deformation itself is
not responsible for volume change, the existence of
very large hydrostatic pressure under the indentation

Fig. 8. Schematic illustrating the local increase in particle concentration due to indentation.

Fig. 9. Schematic showing the general approach of the modeling. Indentation on the two-phase material and the homogenized material was simulated. The homogenized material was
taken to have the same overall stressstrain characteristics as
the two-phase composite.

3226

SHEN et al.: PARTICLE-REINFORCED COMPOSITES

response of the composite. This stressstrain relation


was then used as the input properties for a homogeneous continuum, termed the homogenized
material in Fig. 9, which was then subject to indentation. Indentation was simulated by impinging a rigid
circular indenter into the top surface. Indentation was
also simulated directly on the composite with discrete
particles, henceforth termed the two-phase material.
The essence of this approach is that the two systems
under indentation (i.e., homogenized and two-phase)
have exactly the same overall uniaxial tensile stress
strain response, but the composite consists of discrete
reinforcement particles. The motivation for this
model stemmed from the experimental results in Figs
4 and 5, as will be discussed below.
The commercial finite element code ABAQUS [32]
was employed in our modeling. Because of symmetry
only a half of the structure was needed in the calculation. A total of 144 particles were included in the
two-phase model. Our preliminary calculations [33]
showed that, with a constant area fraction of
reinforcement, convergent results on the indentation
loaddisplacement response can be obtained if the
total number of particles included in the calculation
is greater than about 100. Although the modeling was
based on a two-dimensional plane stress formulation,
realistic qualitative trends can still be obtained. In the
simulation of tensile loading the boundary value problem pertains to the commonly adopted unit-cell
approach [34, 35]. In the indentation simulation the
bottom boundary was allowed to slip tangentially.
The top boundary was not constrained; when contact
with the rigid indenter took place, however, a coefficient of friction of 1.0 was applied. The left boundary was the symmetry axis (see Fig. 10), and the right
boundary was free to move but was constrained to
remain vertical. The ratio of indenter radius and initial
side length of the square domain was taken to be 0.79.
The SiC particles were taken to be linearly elastic
with Youngs modulus of 450 GPa and Poissons
ratio of 0.17. The Al matrix phase was taken to be
elasticplastic with Youngs modulus of 70 GPa and
Poissons ratio of 0.33. The plastic portion follows
von Mises plasticity with a piecewise linear hardening response. Two cases of matrix strength in the
plastic regime were considered. In the strong matrix
case, the plastic response was taken to have stresses
of 360, 450, 525 and 750 MPa at the plastic strains
of 0, 0.02, 0.05 and 0.50, respectively. In the weak
matrix case, the stress levels were taken to be twothirds of those in the strong case at the same plastic
strains. As a consequence, two sets of input parameters, resulting from the strong and weak cases,
were used in the homogenized material model. In this
manner the strong and weak matrices qualitatively
simulated the peak-aged and severely over-aged matrices, respectively.
Figures 10(a) and (b) show the contours of constant
effective plastic strain in the homogenized and twophase materials, respectively, at an indentation dis-

Fig. 10. Contours of constant effective plastic strain in (a) the


homogenized and (b) the two-phase materials under indentation. The normalized indentation displacement is at 0.10.

placement of 0.10. The indentation displacement is


defined as the ratio of the penetration depth to initial
height of the model specimen. These plots correspond
to the strong matrix material; the weak matrix
case results in qualitatively similar features. In Fig.
10(b) the particles are well discerned because no plasticity exists in the purely elastic particles. It can be
seen that a large degree of plasticity is localized and
decays very quickly with increasing distance from the
indentation site. The deformation patterns, however,
are very different in Figs 10(a) and (b). The presence
of discrete particles dramatically fragments the plastic
field and forces the deformation into a banded structure. As a consequence, the indentation response
resulting from the two models is quite different. Figure 10(b) also reveals that the particles directly underneath the depression were displaced downward with
the surrounding matrix. In fact, the local density of

SHEN et al.: PARTICLE-REINFORCED COMPOSITES

particles is higher compared to regions away from the


depression. This can be appreciated by noting that the
inter-particle spacing, especially along the vertical
direction, is reduced in the region directly below the
indentation. (We have also used the plane strain
model, rather than plane stress, to simulate indentation loading of the material containing discrete particles. The same qualitative features as in Fig. 10(b)
were obtained. Therefore, the numerical findings of
interest here are independent of the two-dimensional
models chosen for analysis.) The same phenomenon
is depicted schematically in Fig. 8. A greater load is
thus needed to achieve a given displacement (see
below). In the case of homogenized material, this
hardening effect due to particle redistribution is nonexistent, even though the overall stressstrain
behavior of the two materials is the same.
Figure 11 shows the modeled response of load vs.
displacement during indentation, for both the cases of
strong and weak matrices. It is clear that a
stronger matrix requires a greater load to achieve the
same indentation depth. Furthermore, after an initial
displacement of about 0.03, the two-phase model
clearly displays a harder response than the homogenized model. This means that the indentation response
does not uniquely reflect the overall tensile behavior
of the material. When the particles are explicitly
included, a significantly harder response results
irrespective of the fact that the two-phase and homogenized composites have the same overall stress
strain response. This numerical finding is in accord
with the experimental results in Figs 4 and 5, where
the composite materials showed a harder response
than the monolithic materials with the same tensile
strength and yield strength.
With the modeling results available, a numerical
version of the strengthhardness plot can be obtained.
The 0.2% offset yield stresses directly obtained from
the modeling of tensile loading for the cases of
strong and weak matrices were chosen for such

Fig. 11. Calculated loaddisplacement response during indentation. The displacement is normalized by the initial specimen
height.

3227

a plot. For hardness values, we adopt the formula for


the Rockwell B-scale hardness [36] in our calculation
of modeled hardness number: H = 130500d, where
H is the presently defined hardness number and d
is the penetration depth of the indenter. The hardness
number is calculated with an arbitrarily chosen constant load of 240 N. Note that our two-dimensional
indentation modeling apparently does not conform to
the experimental requirement for the Rockwell hardness test (or any standard hardness test). Nevertheless, the objective here is simply to define a reasonable numerical hardness measure to facilitate a direct
but qualitative comparison with the experimental
results. Figure 12 shows the numerical results of
strength vs. hardness for both the two-phase and
homogenized materials. Comparing Fig. 12 with Figs
4 and 5, a similar trend of strengthhardness relationship is evident. When the particles are explicitly
included in the material (two-phase material in modeling and composite in experiments), greater hardness
values are observed. This is particularly true for the
weaker matrix material. As the matrix strength
increases, the effect of reinforcement particles on
hardness is reduced.
Some general observations regarding the modeling
results can be made. Although the two-phase material
generally exhibits a stronger indentation response, at
the very early stage of the indentation (displacement
smaller than about 0.03), the homogenized material
appears to be slightly harder (Fig. 11). This is
because, at small indentation displacements in the
two-phase model, the resistance to indentation arises
mainly from the top-layer matrix. In the homogenized
model, however, the indenter immediately encounters
a homogeneous material stronger than the pure matrix
(i.e., with the strengthening effects already accounted
for). Consequently, a stronger indentation response is
seen for the homogenized model at the very early
stages of indentation. Once the indenter goes deeper

Fig. 12. Relationship between 0.2% offset yield stress and


hardness number obtained directly from numerical modeling.
Data from the cases of strong matrix and weak matrix are
indicated. The hardness number, defined in the text, is purely
for facilitating a comparison between different material models.

3228

SHEN et al.: PARTICLE-REINFORCED COMPOSITES

into the material, the particle effects become dominant in the two-phase model and the trend is reversed.
The same features were observed in our calculations
incorporating different numbers of particles [33]. The
discrepancy resulting from the two models (twophase vs. homogenized) persists even when the number of particles is increased in the calculation, because
the local increase in discrete particle concentration
cannot be accounted for in the homogenized model.
This local increase of particle concentration is the
main underlying mechanism in affecting the tensile
strengthhardness correspondence of MMCs.
It should be noted that, while the local increase in
particle concentration is believed to occur in actual
composites under indentation, it is somewhat difficult
to directly observe this phenomenon microscopically,
due to the relatively high volume fraction of particles
and their irregular shape and spatial distribution in
the composites we studied. A more quantitative metallography analysis is needed to examine these effects
in more detail.
7. CONCLUSIONS

In an attempt to explore the correlation between


macro-hardness and tensile properties of particlereinforced metal matrix composites, we have conducted experiments on the 2080/SiC/10p and
2080/SiC/20p composites with varying SiC particle
sizes, as well as the monolithic 2080 alloy. Numerical
modeling was also employed to rationalize the experimental results. The major conclusions of this study
are as follows:
1. Unlike most monolithic metals and alloys, hardness does not necessarily scale with the overall
strength of particle-reinforced MMCs. Care must
be taken in applying hardness testing to assessing
the tensile behavior of the composite.
2. The hardness test can significantly overestimate
the overall tensile and yield strengths of composites containing large reinforcement particles,
which are prone to fracture during deformation
processing and/or tensile loading. The predominant local compressive stress state in a hardness
test prevents the pre-existing fractured particles
from weakening the material during indentation.
3. For composites having relatively small reinforcement particles, a unique relationship between
hardness and tensile/yield strength still does not
exist, even when the material is essentially free of
pre-existing fractured particles. This is especially
true in cases where the strength of the Al matrix is
relatively low. For approximately the same overall
ultimate strength and yield strength, the severely
over-aged 2080/SiCp composites showed significantly greater hardness than the monolithic 2080
alloy.
4. Based on fractographic analyses particle fracture
caused by tensile testing is insensitive to matrix

strength. Rather, the decrease in tensile strength


with matrix strength can be attributed to easier
void nucleation and coalescence in the matrix.
Therefore, the lack of unique strengthhardness
correlation cannot be attributed to the particle fracture-induced weakening during the tensile test.
5. The tendency of higher hardness for particlereinforced composites can be attributed to the
localized increase in particle concentration directly
underneath the indenter during hardness testing.
This was illustrated by micromechanical modeling
using the finite element method. Under indentation, the material system with discrete particles
exhibits a higher resistance to deformation than
the homogenized system having exactly the same
overall stressstrain behavior. These findings are
true even when the particle size is much smaller
than the indenter size. The numerical results of
overall yield stress vs. hardness showed the same
trend as that observed in experiments.
AcknowledgementsJ.J.W., G.P., and N.C. gratefully
acknowledge the support of the United States Automotive
Materials Partnership (USAMP) and the Department of Energy.
Y.L.S. and Y.L.G. acknowledge the partial support of National
Science Foundation. The authors are also grateful to W.H.
Hunt, Jr, of Aluminum Consultants Inc., for supplying the
materials used in this study.

REFERENCES
1. ASM Handbook, Vols 1 (1990), 2 (1991) and 8 (2000),
10th edn. ASM International, Materials Park, OH.
2. Nieh, T. G. and Karlak, R., Scripta metall., 1984, 18, 25.
3. Christman, T. and Suresh, S., Acta metall., 1988, 36, 1691.
4. Suresh, S., Christman, T. and Sugimura, Y., Scripta metall., 1989, 23, 1599.
5. Appendino, P., Badini, C., Marino, F. and Tomasi, A.,
Mater. Sci. Engng A, 1991, 135, 275.
6. Chawla, K. K., Esmaeili, A. H., Dayte, A. K. and Vasudevan, A. K., Scripta metall. mater., 1991, 25, 1315.
7. Dutta, I., Allen, S. M. and Hafley, J. L., Metall. Trans. A,
1991, 22A, 2553.
8. Das, T., Bandyopadhyay, S. and Blairs, S., J. Mater. Engng
Performance, 1992, 1, 839.
9. Gupta, M., High Temp. Mater. Processes, 1998, 17, 237.
10. Suresh, S., Mortensen, A. and Needleman, A., Fundamentals of Metal Matrix Composites. Butterworth-Heinemann,
Stoneham, MA, 1993.
11. Taya, M. and Arsenault, R. J., Metal Matrix Composites
Thermomechanical Behavior. Pergamon Press, Oxford,
1989.
12. Chawla, N. and Shen, Y. -L., Adv. Engng Mater., 2001,
3, 357.
13. Chawla, N., Andres, C., Jones, J. W. and Allison, J. E.,
Metall. Mater. Trans. A, 1998, 29A, 2843.
14. Mummery, P. M., Derby, B., Buttle, D. J. and Scruby, C.
B., in Proc. of Euromat 91, Vol. 2, ed. T. W. Clyne and
P. J. Withers. Cambridge, UK, 1991, pp. 441447.
15. Manoharan, M. and Lewandowski, J. J., Mater. Sci. Engng
A, 1992, 150, 179186.
16. Lewandowski, J. J., Liu, D. S. and Liu, C., Scripta metall.,
1991, 25, 21.
17. Kamat, S., Hirth, J. P. and Mehrabian, R., Acta metall.,
1989, 37, 2395.
18. Chawla, N. and Allison, J. E., in Encyclopedia of
Materials: Science and Technology. Elsevier, in press.

SHEN et al.: PARTICLE-REINFORCED COMPOSITES


19. Krajewski, P. E., Allison, J. E. and Jones, J. W., Metall.
Mater. Trans. A, 1993, 24A, 2731.
20. Chawla, N., Habel, U., Shen, Y. -L., Andres, C., Jones,
J. W. and Allison, J. E., Metall. Mater. Trans A, 2000,
31A, 531.
21. ASTM Standard E 18-84, 1984.
22. Brechet, Y., Embury, J. D., Tao, S. and Luo, L., Acta metall. mater., 1991, 39, 1781.
23. Finot, M., Shen, Y. -L., Needleman, A. and Suresh, S.,
Metall. Mater. Trans. A, 1994, 25A, 2403.
24. Singh, P. M. and Lewandowski, J. J., Metall. Trans. A,
1993, 24A, 2531.
25. Llorca, J., Martin, A., Ruiz, J. and Elices, M., Metall.
Trans. A, 1993, 24A, 1575.
26. Shen, Y. -L., Fishencord, E. and Chawla, N., Scripta
mater., 2000, 42, 427.
27. Lewandowski, J. J., in Comprehensive Composite
Materials, Vol. 3, ed. A. Kelly and C. Zweben. Elsevier,
2000, pp. 151187.
28. Williams, J. J., Piotrowski, G., Saha, R. and Chawla, N.,

3229

Unpublished work, 2001.


29. Singh, P. M. and Lewandowski, J. J., in Intrinsic and
Extrinsic Fracture Mechanisms in Organic Composite Systems, ed. J. J. Lewandowski and W. H. Hunt, Jr. The Minerals, Metals and Materials Society, Warrendale, PA, 1995,
pp. 5768.
30. Shaw, M. C. and DeSalvo, G. J., Trans. ASME J. Engng
Industry, 1970, 92, 480.
31. Shen, Y. -L. and Chawla, N., Mater. Sci. Engng A, 2001,
297, 44.
32. ABAQUS, Version 5.8. Hibbit, Karlson and Sorensen,
Inc., Pawtucket, RI.
33. Shen, Y. -L. and Guo, Y. L., Modelling Simul. Mater. Sci.
Engng, in press.
34. Shen, Y. -L., Finot, M., Needleman, A. and Suresh, S.,
Acta metall. mater., 1994, 42, 77.
35. Shen, Y. -L., Finot, M., Needleman, A. and Suresh, S.,
Acta metall. mater., 1995, 43, 1701.
36. ASM Handbook. ASM International, Materials Park, OH,
1985, Vol. 8, p. 75.

S-ar putea să vă placă și