Composites: Part B 36 (2005) 513523

www.elsevier.com/locate/compositesb

New models for effective Youngs modulus of particulate composites

Rajinder Pal*
Department of Chemical Engineering, University of Waterloo, Waterloo, Ont., Canada N2L 3G1
Received 2 April 2004; accepted 7 February 2005
Available online 8 June 2005

Abstract
Four new models for relative Youngs modulus of concentrated particulate composites are developed using a differential scheme along
with the solution of an infinitely dilute dispersion of particles in a solid matrix. The solid matrix phase is assumed to be incompressible in the
derivation. Relative Youngs modulus of concentrated particulate composites is a function of three variables according to the first two models
developed in the paper; the three variables are: dispersed-phase Poissons ratio, ratio of dispersed phase Youngs modulus to matrix phase
Youngs modulus, and volume fraction of particles. The remaining two models include an additional parameter, that is, the maximum
packing volume fraction of particles. The proposed models are evaluated using seven sets of experimental data on Youngs modulus of
concentrated particulate composites.
q 2005 Published by Elsevier Ltd.
Keywords: (B) Elastic properties; (B) Mechanical properties; Particulate composites

1. Introduction
Interest in particulate composites has expanded in recent
years as these multi-phase mixtures often provide an
advantageous blend of properties of the individual
materials. For example, most polymers in homogenous
form are glassy and brittle; the addition of rubber particles
to a polymer matrix can greatly improve the impact
resistance of the material [1,2]. Likewise, the addition of
rigid fillers (for example, carbon black) to rubberlike
elastomers can greatly improve the stiffness and strength
of the material [3]. While fillers are commonly added to a
polymer matrix to enhance the mechanical properties of the
material, this is not the only reason for filler addition; in
some situations, fillers are added to a polymer to improve
other properties such as electrical conductivity [4].
Electrically conducting polymers find many applications
such as in electromagnetic shielding and in electronic
packaging. Polymer based particulate composites are also
widely used as dental restorative materials [5,6]. Dental
composites consist of blends of fillers and resins, and often

* Corresponding author.
E-mail address: rpal@cape.uwaterloo.ca.

1359-8368/$ - see front matter q 2005 Published by Elsevier Ltd.

doi:10.1016/j.compositesb.2005.02.003

have a coupling agent between the filler and matrix phases.

The filler volume fraction in dental composites is generally
high (O0.50). The solid propellants commonly used in
aerospace propulsion are particulate composites consisting
of particles of solid oxidizer, typically ammonium perchlorate, dispersed in a matrix of plastic fuel such as
polybutadiene [79].
While the most common advanced composites are based
on polymer matrix (and are called polymer matrix
composites), there are other important particulate composites based on non-polymeric matrix such as: metal matrix
composites and ceramic matrix composites [10].
According to Cohen and Ishai [11], two-phase particulate
composites can be divided into three major classes in terms
of their filler-to-matrix modulus ratio: (1) composites with
high filler-to-matrix modulus ratio (Ed/EmO20); these
include most filled or reinforced polymer composites; (2)
composites with low filler-to-matrix modulus ratio (1!Ed/
Em!5); these include most cement concretes and metal
matrix composites. Note that concrete is a composite
material consisting of cement paste matrix and particles of
various sizes ranging from sand grains of diameter 100 mm
to large rock particles of diameter 1020 mm [12]; and (3)
composites with soft particles (0!Ed/Em%1) or with voids
(Ed/EmZ0).
Over the past 50 years or so, a number of experimental
and theoretical studies have been published on

514

R. Pal / Composites: Part B 36 (2005) 513523

the mechanical properties of particulate composites [1,13

45]. However, the understanding of the relationship between
the macroscopic mechanical behavior and the microstructural properties (volume fraction of particles, size distribution of particles, etc.) is far from satisfactory. In the
existing literature, while there are exact micromechanics
models available for the infinitely dilute systems, there are
no exact micromechanics models available for the concentrated systems. In concentrated systems where particle
particle interactions are important, it is difficult to solve the
fundamentals equations of mechanics and hence no exact
analytical solutions exist for the effective properties of
concentrated particulate composites. Only approximate
micromechanics models are available for the concentrated
systems. Based on different modeling approaches, a number
of expressions have been proposed to describe the elastic
properties for concentrated particulate composites. However, no one equation has been found to adequately
represent all the systems.
One approach to predict the effective mechanical
properties of a two-phase composite is to apply the law of
mixtures. According to the law of mixtures, each of the
constituent phases of the composite contributes to the
effective mechanical property to an extent depending on the
volume fraction of the constituent phase. If it is assumed
that each component of the composite undergoes the same
strain (isostrain or action-in-parallel situation), the effective
mechanical property can be calculated from the Voigt law of
mixtures [39]. If it is assumed that the each component of
the composite experiences the same stress (isostress or
action-in-series situation), then the Reuss law of mixtures
[40] can be used to estimate the effective mechanical
property of the composite. The rule of mixtures formulae for
the effective mechanical properties of composite materials
provide only crude estimates. The actual value of the
mechanical property shows large deviation from the rule of
mixtures.
Several researchers have applied the variational theorems of the theory of elasticity to obtain upper and lower
bounds for the effective modulus. A composite material
composed of isotropic components distributed randomly in
the material space will have isotropic effective properties,
which are restricted by definite lower and upper bounds.
Regardless of the geometry of the inclusions (dispersedphase), the effective property of the composite lies somewhere in the interval between the lower and upper bounds.
Such bounds depend only on the volume fractions of the
components and the properties of the components. This
approach was utilized by Hashin and Shtrikman to obtain
the celebrated HashinShtrikman bounds [16]. However,
the HashinShtrikman bounds yield satisfactory estimate
for the effective moduli (in that the upper and lower bounds
are close) only when the particles and the matrix material
have similar moduli. When the moduli of the two phases are
very different, the bounds become too wide to be of any
practical value.

Another well-known method to determine the effective

mechanical properties of concentrated composites is the
self-consistent scheme. This approach involves embedding the particle in a homogeneous infinite medium that has
the effective mechanical properties of the composite. The
bond between the particle and the surrounding infinite
medium is assumed to be perfect so that there occurs
displacement and traction continuity across the interface
between phases. The system is subjected to uniform strain or
stress at infinity and the stress and strain fields in the particle
(known to be uniform) are determined. From the relationship between the far-field stresses and strains and the
stresses and strains in the particle, the elastic properties of
the infinite homogeneous medium (composite) are determined. This self-consistent scheme was first utilized by Hill
[18] and Budiansky [19] to develop the equations for the
elastic moduli of particulate composites. The self-consistent
scheme does not give accurate predictions of the moduli
especially when there is a large mismatch in the properties
of the dispersed phase and matrix. Furthermore, this scheme
gives unrealistic results in limiting cases.
In order to overcome the limitations of the self-consistent
scheme, the generalized self-consistent scheme was
proposed to determine the effective mechanical properties
of concentrated composites [32]. According to the generalized self-consistent scheme, the composite is first treated
as an equivalent effective medium which is homogeneous
and has the same macroscopic mechanical properties as that
of the composite. Then a small portion of the effective
homogeneous medium is replaced by the actual components
of the composite. The mechanical properties of the effective
medium are then determined by insisting that if a small
portion of the effective homogeneous medium is replaced by
the actual components of the dispersion, no difference in
mechanical properties could be detected by macroscopic
observations. This approach was utilized by Kerner [20],
van der Poel [22], Smith [25,26], and others [32,44]. The
equations based on the generalized self-consistent scheme
give reasonable predictions at low to moderate concentrations of the particulate phase. At high concentrations of
the particulate phase, a large deviation between the
experimental and predicted values is generally observed.
Furthermore, the equations predict the modulus of the
particulate composite to be independent of the particle size
distribution.
In this paper, new equations for Youngs modulus of
concentrated particulate composites are derived using a
differential scheme along with the exact solution of an
infinitely dilute system. The key features of our models are:
(a) they are closed form expressions; (b) they are valid over
the full range of the particle concentration; (c) they take into
consideration the effects of the particulate (dispersed) phase
compressibility (Poissons ratio) on the Young modulus of
the composite; (c) they consider the effect of the modulus
ratio (defined as the ratio of the dispersed phase Young
modulus to matrix Young modulus) on the Young modulus

R. Pal / Composites: Part B 36 (2005) 513523

of the composite, and are valid over the full range of the
modulus ratio; and (d) they include the effect of the particle
size distribution on the Young modulus of the composite
through the parameter fm, the maximum packing volume
fraction of the particles. The fm value for any composite
depends on the particle size distribution of the particulate
phase. The equations developed in the paper are evaluated
using seven sets of experimental data on Youngs modulus
of particulate composites.

2. Background
For an infinitely dilute dispersion of spherical solid
particles in a solid matrix, the shear and bulk moduli are
given as [1,23,28,32]:



G
151 K nm Gd K Gm
Z1C
f
(1)
Gm
2Gd 4 K 5nm C Gm 7 K 5nm
K
Z1C
Km



3Km C 4Gm
3Km



3Kd K 3Km
3Kd C 4Gm


f

(2)

where G and K are shear and bulk moduli, respectively, of

the composite, Gm and Km are shear and bulk moduli,
respectively, of the matrix, Gd and Kd are shear and bulk
moduli, respectively, of the dispersed-phase (particles), nm
is Poissons ratio of the matrix, and F is the volume fraction
of the particles.
For isotropic materials, Youngs modulus (E) is related
to shear and bulk moduli as [46]:
EZ

9KG
3K C G

(3)

In terms of Poissons ratio (n), Youngs modulus can be

expressed as [46]:
E Z 2G1 C n

(4)

E Z 3K1 K 2n

(5)

From Eqs. (1)(5), it can be shown that Youngs modulus

of an infinitely dilute dispersion of spherical solid particles
in a solid matrix is given by:
E
Z 1 C 10b1 1 C nm C b2 1 K 2nm f
Em
where b1 and b2 are given as:



a1 Ed =Em K a2
b1 Z
2a3 Ed =Em C a4

b2 Z

a2 Ed =Em K a5 
Ed =Em C 2a6

(6)

(7)


(8)

In Eqs. (7)(8), Ed and Em are Youngs moduli of the

dispersed-phase and the matrix-phase, respectively, and as
are functions of nd and nm, Poissons ratios of the dispersed

515

and matrix phases, respectively:

a1 Z

1 K nm
1 C nd

(9)

a2 Z

1 K nm
1 C nm

(10)

a3 Z

4 K 5nm
1 C nd

(11)

a4 Z

7 K 5nm
1 C nm

(12)

a5 Z

1 K 2nd
1 K 2nm

(13)

a6 Z

1 K 2nd
1 C nm

(14)

For particulate composites of incompressible matrix

(nmZ1/2), Eq. (6) reduces to:



E
15Ed =Em K 101 C nd
Er Z
Z1C
Em
6Ed =Em C 61 C nd



(15)
1 K 2nd
K
f
3Ed =Em C 41 K 2nd
where Er is relative Youngs modulus, defined as the ratio of
composite Youngs modulus to matrix-phase Youngs
modulus.
For composites with voids (pore-solid composites),
Eq. (15) gives relative Youngs modulus as:
 
23
Er Z 1 K
f
(16)
12
For incompressible particles/inclusions (n dZ1/2),
Eq. (15) gives the following expression for relative Youngs
modulus:



5 2Ed =Em K 2
Er Z 1 C f
(17)
2 2Ed =Em C 3
When Ed/Em/N, that is, the particles are rigid, Eq. (17)
reduces to the well-known Einstein relation [13]:
5
Er Z 1 C f
2

(18)

Youngs modulus equations, Eq. (6) or Eq. (15), are valid

only for infinitely dilute systems. They cannot be applied at
finite concentrations of dispersed phase as the interaction
between the particles is ignored in their derivation. In the
following section, new equations are developed for Youngs
modulus of concentrated particulate composites using the
differential effective medium approach originally proposed
by Bruggeman [47], who derived an equation for the
dielectric constant of a concentrated suspension.

516

R. Pal / Composites: Part B 36 (2005) 513523

Thus, Eq. (19) becomes:

3. Development of new Youngs modulus equations

for concentrated systems
Let us consider a particulate composite with a volume
fraction of particles f. Into this composite, an infinitely
small number of new particles were added. The increment
increase in Youngs modulus, dE, resulting from the
addition of the new particles can be calculated from the
dilute system result (Eq. (15)) by treating the composite into
which the new particles are added as an equivalent effective
medium of Youngs modulus E. Thus,



15Ed K 10E1 C nd
1 K 2nd E
dE Z E
K
df
6Ed C 6E1 C nd
3Ed C 4E1 K 2nd
(19)
This equation is separable, since we can write it as:
E

dE
15EdK10E1Cnd
6EdC6E1Cnd

d E
K 3Ed1K2n
C4E1K2nd

i Z df

(20)

After considerable algebra and integration with the limit

E/Em at f/0, Eq. (20) gives
 


E
MEm K 3Ed P C Q NC53=46
Em
ME K 3Ed P C Q
(21)



ME K 3Ed P K Q NK53=46
Z exp2:5f
MEm K 3Ed P K Q
where
M Z 461 C nd 1 K 2nd

(22)

P Z 4 K 23nd

(23)

QZ

q
323n2d K 138nd C 82

N Z 833 K 736nd =46Q

(24)
(25)

Eq. (21), referred to as model 1 in the remainder of the

paper, reduces to the following equation for particulate
composite of incompressible spherical particles (ndZ1/2):
 

E
E K Ed K2:5
Z exp2:5f
(26)
Em
Em K Ed
Model 1 (Eq. (21)) is expected to describe Youngs
modulus of particulate composites of incompressible matrix
at low to moderate values of f (volume fraction of
particles). This is because in the derivation of the
differential equation, Eq. (19), leading to model 1
(Eq. (21)), it is assumed that all the volume of the composite
before new particles are added is available as free volume to
the new particles. In reality, the free volume available to
disperse the new particles is significantly less, due to the
volume pre-empted by the particles already present. The
increase in the actual volume fraction of the dispersed phase
when new particles are added to the composite is df/(1Kf).

dE
df
h
iZ
15EdK10E1Cnd
1K2nd E
1
Kf
E 6EdC6E1Cnd K 3EdC4E1K2nd

(27)

Upon integration with the limit E/Em at f/0, Eq. (27)

gives:
 


E
MEm K 3Ed P C Q NC53=46
Em
ME K 3Ed P C Q
(28)



ME K 3Ed P K Q NK53=46
Z 1 K fK2:5
MEm K 3Ed P K Q
Eq. (28), referred to as model 2 in the remainder of the
paper, reduces to the following equation for particulate
composite of incompressible spherical particles (ndZ1/2):
 

E
E K Ed K2:5
Z 1 K fK2:5
(29)
Em
Em K E d
According to models 1 and 2 (Eqs. (21) and (28)), it is
possible that volume fraction of the dispersed phase can
reach unity as more and more particles are added to the
composite. This is physically unrealistic, especially for
composite systems of rigid particles. In reality, there exists
an upper limit for f, referred to as maximum packing
volume fraction of particles (fm). The value of fm varies
with the type of packing arrangement of particles. For
example, it is 0.637 for random close packing of monosized
spherical particles. For hexagonal close packing of uniform
spheres, fm is 0.74.
Mooney [48], in the derivation of his well-known
equation for the viscosity of concentrated suspensions,
contended that the incremental increase in the volume
fraction of the dispersed phase when infinitesimal amount of
new particles are added to an existing suspension of
dispersed phase volume fraction f, is d[f/(1Kf/fm)] rather
than df/(1Kf) because of the packing difficulties of
particles. Thus, Eq. (19) becomes:
"
#
dE
f
h
i Zd
(30)
1 K ffm
E 15EdK10E1Cnd K 1K2nd E
6EdC6E1Cnd

3EdC4E1K2nd

Upon integration, Eq. (30) gives:




E
MEm K 3Ed P C Q NC53=46
Em
ME K 3Ed P C Q

ME K 3Ed P K Q
MEm K 3Ed P K Q

NK53=46

"

2:5f
Z exp
1 K ffm

(31)

Eq. (31), referred to as model 3 in the remainder of the

paper, reduces to the following equation for particulate
composite of incompressible spherical particles (ndZ1/2):
"
#
 

E
E K Ed K2:5
2:5f
Z exp
(32)
Em
Em K E d
1 K ffm

R. Pal / Composites: Part B 36 (2005) 513523

517

Table 1
Summary of the new models developed in the present work
Model no.
1 (Eq. (21))

Model description
h
iNC53=46 h

2 (Eq. (28))

h

3 (Eq. (31))

h

4 (Eq. (34))

h

E
Em
E
Em

E
Em

E
Em

MEmK3Ed PCQ
MEK3Ed PCQ
MEmK3Ed PCQ
MEK3Ed PCQ

MEmK3Ed PCQ
MEK3Ed PCQ

MEmK3Ed PCQ
MEK3Ed PCQ

iNC53=46 h
iNC53=46 h
iNC53=46 h

Comments
MEK3Ed PKQ
MEmK3Ed PKQ
MEK3Ed PKQ
MEmK3Ed PKQ

MEK3Ed PKQ
MEmK3Ed PKQ

MEK3Ed PKQ
MEmK3Ed PKQ

iNK53=46
iNK53=46

Z exp2:5f
Z 1K fK2:5

iNK53=46


2:5f
Z exp 1K
f

iNK53=46

K2:5fm
Z 1K ffm

fm

This model is valid at low to moderate values of f. It does not

take into account the effect of the particle size distribution
This model covers a broader range off as compared with model 1.
However, this model, like model 1, does not take into account
the effect of the particle size distribution
This model covers a broad range of f, that is 0!f!fm. It is
capable of taking into account the effect of the particle size
distribution
Like model 3, this model covers a broad range of f, that is
0!f!fm, and it is capable of taking into account the effect
of the particle size distribution

Note that M, P, Q, and N present in the above model equations are defined in Eqs. (22)(25).

It is interesting to note that Krieger and Dougherty [49]

in the derivation of their well-known equation for the
viscosity of concentrated suspensions, contended that the
incremental increase in the volume fraction of the dispersed
phase when infinitesimal amount of new particles are added
to an existing suspension of concentration f, is df/(1Kf/
fm) rather than d[f/(1Kf/fm)] as thought by Mooney [48].
Hence, Eq. (19) should be:
df
iZ

1K2nd E
dK10E1Cnd
E 15E
1 K ffm
6EdC6E1Cnd K 3EdC4E1K2nd
h

dE

Upon integration, Eq. (33) gives:




E
MEm K 3Ed P C Q NC53=46
Em
ME K 3Ed P C Q





ME K 3Ed P K Q NK53=46
f K2:5fm
Z 1K
MEm K 3Ed P K Q
fm

Youngs modulus (Em), f is volume fraction of particles,

and fm is maximum packing volume fraction of undeformed particles.

(33)

(34)

Eq. (34), referred to as model 4 in the remainder of the

paper, reduces to the following equation for particulate
composite of incompressible spherical particles (ndZ1/2):
 



E
E K Ed K2:5
f K2:5fm
Z 1K
(35)
Em
fm
Em K Ed
Table 1 summarizes the key aspects of the models
developed in this section.

4. Predictions of new Youngs modulus equations

According to new Youngs modulus equations for
particulate composites derived in Section 3, Youngs
modulus of a particulate composite of incompressible
matrix can be expressed as:
Er Z f nd ; l; f; fm

(36)

where Er is relative Youngs modulus (defined as E/Em), nd

is Poissons ratio of the dispersed phase, l is the ratio of
dispersed-phase Youngs modulus (Ed) to matrix-phase

Fig. 1. Relative Youngs modulus (Er) as a function of volume fraction of

particles (f) predicted by the proposed models. The number shown on the
curve refers to the corresponding model number. The letter d on the curve
refers to dilute system model (Eq. (15)).

518

R. Pal / Composites: Part B 36 (2005) 513523

Fig. 2. Effect of dispersed-phase Poissons ratio (nd) on Er versus f behavior, as predicted from model 4 (Eq. (34)) for different values of modulus ratio l.

Fig. 1 shows the comparison between the predictions of

different models for two different values of l; in one case
lZ5, that is, greater than unity, and in the other case,
lZ0.1, that is, less than unity. Poissons ratio of the
dispersed-phase is kept the same in two cases, that is, ndZ
0.4. The maximum packing volume fraction of particles
(fm) is taken to be 0.637 (corresponding to random close

packing of uniform spheres) in the calculations for models 3

and 4. The number shown on the curve refers to the
corresponding model number. The letter d on the curve
refers to dilute system model, that is, Eq. (15).
When lZ5, all models predict an increase in relative
Youngs modulus (Er) with the increase in the volume
fraction of the particles (f). The predictions of Er from

R. Pal / Composites: Part B 36 (2005) 513523

Fig. 3. Effect of modulus ratio l on Er for different values of dispersedphase Poissons ratio (nd), at a fixed f of 0.50.

different models at any given f are in the following order: Er

(model 3)OEr (model 4)OEr (model 2)OEr (model 1)OEr
(dilute system model).
At a low l value of 0.1, all models predict a decrease in
relative Youngs modulus with the increase in the volume
fraction of the particles. The predictions of Er from different
models at any given f are now in the following order: ErO

519

(model 1)OEr (model 2)OEr (model 4)OEr (dilute system

model)OEr (model 3).
Fig. 2 shows the effect of dispersed-phase Poissons ratio
(nd) on Er versus f behaviour at different values of modulus
ratio l. The calculations are based on model 4 (Eq. (34))
with fm value of 0.637. When l is high, there is little effect
of nd on Er versus f behavior of particulate composites.
However, at not too high values of l (l!100), the effect of
nd on Er versus f behavior is significant; with the increase in
(nd) from K1 to C0.5, relative Youngs modulus of a
particulate composite decreases at any given value of f. It is
interesting to note that for negative values of dispersedphase Poissons ratio (nd), relative Youngs modulus (Er)
can be larger than unity even when the modulus ratio l is
less than unity. Thus, the widely held belief that when l!1,
Er is less than unity and that Er decreases with the increase
in f, is valid only for positive values of dispersed-phase
Poissons ratio (ndO0).
The effect of modulus ratio l on relative Youngs modulus
Er for different values of dispersed-phase Poissons ratio nd is
shown in Fig. 3. The calculations are based on model 4
(Eq. (34)) with fm value of 0.637. The volume fraction of the
particles (f) is fixed at 0.50. The following points should be
noted from Fig. 3: (1) at low values of l(l!10K3), nd has no
effect on relative Youngs modulus of particulate composites;
(2) at high values of l(lO1000), nd has no effect on Er; (3)
Poissons ratio of the dispersed phase (nd) has a significant
effect on Er in the intermediate range of l, that is, 10K3!l!
103; (4) at any given value of l in the intermediate range (10K
3
!l!103). Er decreases with the increase in nd; and (5) the
crossover value of the modulus ratio l where relative Youngs
modulus of the particulate composite crosses from the region
ErO1 to the region Er!1, increases with the increase in
Poissons ratio nd. The crossover value of l is unity only when
ndZ0.50.

Table 2
Summary of various particulate composites considered in the present work
Set. No.

Range of f

Modulus ratio (l)

Ref. no.

Comments

00.524

0.254

[2]

00.706

19.3

[5]

3
4

00.552
00.68

32
Rigid filler-high l

[6]
[9]

00.625

Rigid filler-high l

[34]

00.235

Rigid filler-high l

[13]

00.206

Rigid filler-high l

[13]

Rubber-toughened poly(methyl methacrylate) (PMMA)

composites; ndZ0.40
Dental restorative composites, a total of 55 composites of
varying filler volume fraction were investigated; ndZ0.20
Dental composites of varying filler volume fraction; ndZ0.20
The composite systems investigated were: (a) polyisobutylene filled with 8.7, 20.3 and 36.7% by volume of glass beads
about 40 mm in diameter; (b) two castable polyurethane
rubbers containing 54% by volume of similar glass beads;
(c) propellant A containing 55, 61 and 68% solids by volume;
(d) propellant B containing 38, 43, 48, 53 and 59% solids by
volume; ndZ0.50
Crosslinked polyurethane rubber filled with glass beads
ranging from 60 to 90 mm in diameter; ndZ0.50
Reinforced-rubber composites; the fillers investigated were:
(a) carbon black (b) Kadox; (c) XX Zinc oxide; (e) Catalpo
clay; ndZ0.50
Reinforced-rubber composites; the fillers investigated were:
(a) Thermax; (b) Gilders whiting; (c) PK33; ndZ0.50

520

R. Pal / Composites: Part B 36 (2005) 513523

Fig. 4. Comparison between experimental data of set 1 (Table 2) particulate

composites and predictions of various models. The number shown on the
curve refers to the corresponding model number. The letter d on the curve
refers to dilute system model (Eq. (15)). For models 3 and 4, fm is taken to
be 0.637.

5. Comparison of new Youngs modulus equations

with experimental data

Fig. 6. Comparison between experimental data of set 3 (Table 2) particulate

composites and predictions of various models. The number shown on the
curve refers to the corresponding model number. The letter d on the curve
refers to dilute system model (Eq. (15)). For models 3 and 4, fm is taken to
be 0.637.

Seven sets of experimental data on Youngs modulus of

particulate composites covering a wide range of particle
volume fraction f and a wide range of modulus ratio l are
considered to evaluate the models. Table 2 gives a summary
of the various particulate composite systems considered in
the present work.

Figs. 410 show comparisons between experimental data

and predictions of various models. Except for set 1 data, all
other sets of data have ErO1. Note that set 1 data has a
modulus ratio l of less than one. The experimental data of
set 1 (Fig. 4) on rubber-toughened PMMA composites can
be described adequately by model 2 (Eq. (28)). Model 1 (Eq.
(21)) overpredicts Er whereas other models (models 3, 4 and
dilute system) underpredict Er.

Fig. 5. Comparison between experimental data of set 2 (Table 2) particulate

composites and predictions of various models. The number shown on the
curve refers to the corresponding model number. The letter d on the curve
refers to dilute system model (Eq. (15)). For models 3 and 4, fm is taken to
be 0.637.

Fig. 7. Comparison between experimental data of set 4 (Table 2) particulate

composites and predictions of various models. The number shown on the
curve refers to the corresponding model number. The letter d on the curve
refers to dilute system model (Eq. (15)). For models 3 and 4, fm is taken to
be 0.85.

R. Pal / Composites: Part B 36 (2005) 513523

Fig. 8. Comparison between experimental data of set 5 (Table 2) particulate

composites and predictions of various models. The number shown on the
curve refers to the corresponding model number. The letter d on the curve
refers to dilute system model (Eq. (15)). For models 3 and 4, fm is taken to
be 0.637.

The experimental data of set 2 (Fig. 5) covers 55

commercial and experimental dental composites. While the
data exhibits significant scatter, it can be described
reasonably well by model 2 (Eq. (28)). Set 3 data on dental
composites (Fig. 6) can also be described adequately by
model 2 (Eq. (28)). The experimental data of set 4 (Fig. 7)
on Youngs modulus of composite propellants and filled

Fig. 9. Comparison between experimental data of set 6 (Table 2) particulate

composites and predictions of various models. The number shown on the
curve refers to the corresponding model number. The letter d on the curve
refers to dilute system model (Eq. (15)). For models 3 and 4, fm is taken to
be 0.25.

521

Fig. 10. Comparison between experimental data of set 7 (Table 2)

particulate composites and predictions of various models. The number
shown on the curve refers to the corresponding model number. The letter
d on the curve refers to dilute system model (Eq. (15)). For models 3 and
4, fm is taken to be 0.637.

elastomers fits well with model 4 (Eq. (34)) using a fm value

of 0.85. Set 5 data on glass beads-reinforced-rubber
composites (Fig. 8) can be described very well by model
4 (Eq. (34)) using a fm value of 0.637. The experimental
data of set 6 (Fig. 9) on Youngs modulus of various fillerreinforced rubber composites can also be described
reasonably well by model 4 (Eq. (34)); however, the fm
value used in model 4 is now much lower, that is, fmZ0.25.
A low value of fm is indicative of flocculation of particulate
filler present in the composite. Indeed, fillers like carbon
black are known to exhibit flocculation in rubber [13]. Set 7
data on reinforced-rubber composites, shown in Fig. 10, can
be described well by model 2 (Eq. (28)).
In summary, the experimental data (seven sets) on
particulate composites considered here can be described
adequately by either model 2 (Eq. (28)) or model 4 (Eq.
(34)). Model 2 is a special case of model 4 with fmZ1.0.
Thus, model 4 appears to be superior to other models
developed in the paper. It should also be noted that model 4
is quite capable of predicting the effect of particle size
distribution on Youngs modulus of particulate composites.
The maximum packing volume fraction of particles (fm),
present in the model, is known to be sensitive to particle size
distribution; fm for particulate composites of uniform
particles is expected to be significantly smaller as compared
with the fm value for polydisperse composites. Model 1
(Eq. (21)) is satisfactory only at low values of f. This is not
unexpected, as this model does not properly account for the
free volume available to the particles. Model 3 (Eq. (31))
generally overpredicts Er when lO1 and underpredicts
when l!1. The exact reason for this behavior is not clear at
the present time.

522

R. Pal / Composites: Part B 36 (2005) 513523

6. Concluding remarks
Starting from Youngs modulus equation for an
infinitely dilute dispersion of spherical solid particles in
an incompressible solid matrix, four new equations have
been developed for Youngs modulus of concentrated
particulate composites using the differential effective
medium approach. Out of the four models developed in
the paper, two models predict relative Youngs modulus
(Er) to be a function of three variables: modulus ratio l
(ratio of dispersed-phase modulus to the matrix phase
modulus), dispersed-phase Poissons ratio nd, and volume
fraction of particles f. The remaining two models include
an additional parameter, i.e. the maximum packing
volume fraction of particles fm. For positive values of
nd (0!nd!0.5), all models predict an increase in Er with
the increase in f provided that the modulus ratio l is
greater than unity (lO1); when the modulus ratio l is less
than unity, all models predict a decrease in Er with the
increase in f. At low values of l (l!10K3), dispersedphase Poissons ratio nd has negligible effect on Er. Also,
at high values of l (lO103), nd has negligible effect on
Er. Only in the intermediate range of l (10K3!l!103),
relative Youngs modulus decreases with the increase in
nd from K1 to 0.5.
The proposed models are evaluated using seven sets of
experimental data on Youngs modulus of concentrated
particulate composites. Model 4 (Eq. (34)) appears to be
somewhat superior to other models (models 13) developed
in the paper when comparisons are made with the
experimental data. Also, model 2 (Eq. (28)) is a special
case of model 4 with fmZ1.0. Furthermore, model 4 is
quite capable of predicting the effect of particle size
distribution on Youngs modulus of concentrated particulate
composites; the maximum packing volume fraction of
particles fm, present in the model, is known to be sensitive
to the particle size distribution.

Acknowledgements
Financial support from the Natural Sciences and
Engineering Research Council of Canada is greatly
appreciated.

References
[1] Christensen RM. Mechanics of composite materials. New York:
Wiley; 1979.
[2] Biwa S, Ito N, Ohno N. Elastic properties of rubber particles in
toughened PMMA: ultrasonic and micromechanical evaluation. Mech
Mater 2001;33:71728.
[3] Sohn MS, Kim KS, Hong SH, Kim JK. Dynamic mechanical
properties of particle-reinforced EPDM composites. J Appl Polym
Sci 2003;87:1595601.

[4] Berlin Genetti W, Grady B. Crystallization studies of nickel-filled low

density polyethylene composites. Expt Prog Rep 1998;7:52830.
[5] Braem M, Van Doren VE, Lambrechts P, Vanherle G. Determination
of Youngs modulus of dental composites: a phenomenological
model. J Mater Sci 1987;22:203742.
[6] Braem M, Finger W, Van Doren VE, Lambrechts P, Vanherle G.
Mechanical properties and filler fraction of dental composites. Dent
Mater 1989;5:3469.
[7] Smith TL. Elastomeric-binder and mechanical property requirements
for solid propellants. Ind Eng Chem 1960;52:77680.
[8] Blatz PJ. Rheology of composite solid propellants. Ind Eng Chem
1956;48:7279.
[9] Landel RF, Smith TL. Viscoelastic properties of rubber-like
composite propellants and filled elastomers. ARS J 1961;31:599608.
[10] Kaw AK. Mechanics of composite materials. Boca Raton: CRC; 1997.
[11] Cohen LJ, Ishai O. The elastic properties of three-phase composites.
J Compos Mater 1967;1:390403.
[12] Garboczi EJ, Berryman JG. Elastic moduli of a material containing
composite inclusions: effective medium theory and finite element
computations. Mech Mater 2001;33:45570.
[13] Smallwood HM. Limiting law of the reinforcement of rubber. J Appl
Phys 1944;15:75866.
[14] Eshelby JD. The determination of the field of an ellipsoidal inclusion
and related problems. Proc R Soc London A 1957;241:37696.
[15] Hashin Z. The elastic moduli of heterogeneous materials. J Appl Mech
1962;29:14350.
[16] Hashin Z, Shtrikman S. A variational approach to the theory of the
elastic behavior of multiphase materials. J Mech Phys Solids 1963;11:
12740.
[17] Hashin Z. Theory of mechanical behavior of heterogeneous media.
Appl Mech Rev 1964;17:19.
[18] Hill R. A self consistent mechanics of composite materials. J Mech
Phys Solids 1965;13:21322.
[19] Budiansky B. On the elastic moduli of some heterogeneous materials.
J Mech Phys Solids 1965;13:2237.
[20] Kerner EH. The elastic and thermoelastic properties of composite
media. Proc Phys Soc London B 1956;69:808.
[21] Hill R. Elastic properties of reinforced solids: some theoretical
principles. J Mech Phys Solids 1963;11:35772.
[22] Van der Poel C. On the rheology of concentrated dispersions. Rheol
Acta 1958;1:198205.
[23] Walpole LJ. The elastic behavior of a suspension of spherical
particles. Quart J Mech Appl Math 1972;25:15360.
[24] Roscoe R. Isotropic composites with elastic or viscoelastic phases:
general bounds for the moduli and solutions for special geometrics.
Rheol Acta 1973;12:40411.
[25] Smith JC. Correction and extension of van der Poels method for
calculating the shear modulus of a particulate composite. J Res Nat
Burr Stand (USA) 1974;78A:35561.
[26] Smith JC. Simplification of van der Poels formula for the shear
modulus of a particulate composite. J Res Nat Burr Stand (USA)
1975;79A:41923.
[27] McLaughlin R. A study of the differential scheme for composite
materials. Int J Eng Sci 1977;15:23744.
[28] Chen HS, Acrivos A. The effective elastic moduli of composite
materials containing spherical inclusions at nondilute concentrations.
Int J Solids Struct 1978;14:34964.
[29] Christensen RM. Mechanical properties of composite materials. In:
Hashin Z, Herakovich CT, editors. Mechanics of composite materialsrecent advances. New York: Pergamon Press; 1982. p. 116.
[30] Hashin Z. Analysis of composite materials. A survey. J Appl Mech
1962;50:481504.
[31] Norris AN. A differential scheme for the effective moduli of
composites. Mech Mater 1985;4:116.
[32] Christensen RM. A critical evaluation for a class of micromechanics
models. J Mech Phys Solids 1990;38:379404.

R. Pal / Composites: Part B 36 (2005) 513523

[33] Phan-Thien N, Pham DC. Differential multiphase models for
polydispersed suspensions and particulate solids. J Non-Newtonian
Fluid Mech 1997;72:30518.
[34] Bills KW, Sweeney KH, Salcedo FS. The tensile properties of highly
filled polymers-effect of filler concentrations. J Appl Polym Sci 1960;
4:25968.
[35] Munstedt H. Rheology of rubber-modified polymer melts. Polym Eng
Sci 1981;21:25969.
[36] Chantler PM, Hu X, Boyd NM. An extension of a phenomenological
model for dental composites. Dent Mater 1999;15:1449.
[37] Luo JJ, Daniel IM. Characterization and modeling of mechanical
behavior of polymer/clay nano composites. Compos Sci Tech 2003;
63:160716.
[38] Ji S. Generalized means of an approach for predicting Youngs moduli
of multiphase materials. Mater Sci Eng 2004;A366:195201.
[39] Voigt W. Lehrbuch der Kristallphysik. Leipzig: Teubner-Verlag;
1928.
[40] Reuss A. Berechnung der Fliessgrenze von Mischkristallen auf Grund
der Plastizitatsbedingung fur Einkristalle. Z Angew Math Mech 1929;
9:4958.

523

[41] Agarwal BD, Broutman LJ. Analysis and performance of fiber

composites. New York: Wiley; 1990.
[42] Chawla KK. Composite materials. New York: Springer; 1998.
[43] Herakovich CT. Mechanics of fibrous composites. New York: Wiley;
1998.
[44] Halpin JC, Kardos JL. The HalpinTsai equations: a review. Polym
Eng Sci 1976;16:34452.
[45] Christensen RM, Lo KH. Solutions for the effective shear properties in
three phase sphere and cylinder models. J Mech Phys Solids 1979;27:
31531.
[46] Sherman P. Industrial rheology. London: Academic Press; 1970.
[47] Bruggeman DAG. Berechung verschieder Physikalischer Konstanten
von heterogenen substranzen I. Dielekrizitatskonstanen und leifahigkeiten der Mischkorper aus isotropeen substanzen. Ann Phys 1935;24:
63679.
[48] Mooney M. The viscosity of a concentrated suspension of spherical
particles. J Colloid Sci 1951;6:16270.
[49] Krieger IM, Dougherty TJ. A mechanism for non-Newtonian flow in
suspensions of rigid spheres. Trans Soc Rheol 1959;3:13752.