ARTICLE IN PRESS

Journal of the Mechanics and Physics of Solids

54 (2006) 14011425
www.elsevier.com/locate/jmps

Solutions of inhomogeneity problems with graded

shells and application to coreshell nanoparticles
and composites
H.L. Duan, Y. Jiao, X. Yi, Z.P. Huang, J. Wang
LTCS and Department of Mechanics and Engineering Science, Peking University, Beijing 100871, PR China
Received 6 January 2006; accepted 18 January 2006

Abstract
This paper rst presents the Eshelby tensors and stress concentration tensors for a spherical
inhomogeneity with a graded shell embedded in an alien innite matrix. The solution is then
specialized to inhomogeneous inclusions in nite spherical domains with xed displacement or
traction-free boundary conditions. The Eshelby tensors in the innite and nite domains and the
stress concentration tensors are especially useful for solving many problems in mechanics and
materials science. This is demonstrated on two examples. In the rst example, the strain distributions
in coreshell nanoparticles with eigenstrains induced by lattice mismatches are calculated using the
Eshelby tensors in the nite domains. In the second example, the Eshelby and stress concentration
tensors in the three-phase conguration are used to formulate the generalized self-consistent
prediction of the effective moduli of composites containing spherical particles within the framework
of the equivalent inclusion method. The advantage of this micromechanical scheme is that, whilst its
predictions are almost identical to the classical generalized self-consistent method and the third-order
approximation, the expressions for the effective moduli have simple closed forms.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: Eshelby tensor; Stress concentration tensor; Finite domain; Coreshell nanoparticle; Effective modulus

Corresponding author. Tel.: +86 10 6275 7948.

E-mail address: jxwang@pku.edu.cn (J. Wang).

0022-5096/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jmps.2006.01.005

ARTICLE IN PRESS
1402

H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1. Introduction
The synthesis and characterization of particles with coreshell structures have attracted a lot
of attention in many areas of science and technology. In materials science and engineering,
these particles have been used as reinforcements and tougheners in composites. In solid-state
physics, coreshell nanoparticles are found to exhibit novel physical effects and properties,
such as quantum connement effect, and novel electronic, magnetic and optical properties (e.g.
Zhou et al., 1996; Rockenberger et al., 1998; Brongersma, 2003; Goncharenko, 2004).
Coreshell particles can be used as functional devices on their own, besides being a constituent
part of a composite medium (e.g. Williamson and Zunger, 1999; Lauhon et al., 2002; Abe and
Suwa, 2004; Goncharenko, 2004). Many researchers have studied the strain distributions in
heterogeneous electronic structures of ne scale, e.g. quantum dot structures (Gosling and
Willis, 1995; Freund and Johnson, 2001; He et al., 2004) and shown that the strain affects the
optical properties of these structures by modifying the energies and wave functions of the
conned carriers. For coreshell nanoparticles, as pointed out by Little et al. (2001), and PerezConde and Bhattacharjee (2003), the mist strain, the surface stress and the applied external
pressure all modify the strain elds in them, which in turn affect the electronic structures, and
hence their physical properties.
Coreshell structures also widely exist in conventional particle-reinforced composites
and nanocomposites due to complicated interactions between the particle surface and the
matrix (e.g. Theocaris, 1987; Tzika et al., 2000) and the need for good bond between the
reinforcement and the matrix. The elastic properties of the interphase can be uniform or
variable through its thickness (Ostoja-Starzewski et al., 1996). The inhomogeneity
problems with graded interphases have attracted a lot of attention (e.g. Lutz and
Zimmerman, 1996; Wang and Jasiuk, 1998). However, almost all the existing works on
inclusion/inhomogeneity problems with graded (inhomogeneous) interphases are concerned with the solutions of stress elds under special loading conditions or with the
predictions of effective elastic moduli. It is noted that Ding and Weng (1998), and Weng
(2003) have predicted the effective bulk moduli of composites containing spherical particles
and graded matrices using a three-phase model containing a graded interphase.
The Eshelby formalism (Eshelby, 1957, 1959) for an inclusion/inhomogeneity is one of the
cornerstones in the solutions of many problems in materials science, solid-state physics and
mechanics of composites. The classical Eshelby formalism is for an inclusion/inhomogeneity
without an interphase in an innite matrix. In this paper, we shall give the solution of the
Eshelby formalism for a spherical particle with a graded interphase embedded in an innite
medium. The Eshelby tensors in the whole region when an eigenstrain is prescribed in the
particle and the stress concentration tensors under remote loading will be presented. When
the stiffness of the innite medium is set to be innite or zero, the Eshelby tensors in a nite
domain with a xed displacement or traction-free boundary condition are given. The
application of the Eshelby formalism in the nite and innite domains is demonstrated on
two examples, namely, the calculation of the strains in coreshell nanoparticles and the
prediction of the effective moduli of particle-reinforced composites.
2. Solution of spherical inhomogeneity with graded interphase
Consider a spherical inhomogeneity with a graded interphase embedded in an innite
elastic matrix, as shown in Fig. 1. The radius of the inhomogeneity and the outer radius of

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1403

Fig. 1. A spherical inhomogeneity (1) with an interphase I in an innite matrix (2).

the interphase are denoted by a and b, respectively. For brevity, in the following, all length
scales are regarded as being normalized by the radius of the inhomogeneity. The interface
between the inhomogeneity and interphase is denoted by G1I (r 1, and the interface
between the interphase and matrix is denoted by GI2 (r b=a. The inhomogeneity and
matrix are homogeneous, linearly elastic and isotropic, characterized by the bulk modulus
kk , the shear modulus mk and the Poisson ratio nk . Here, and in the following, the subscript
and superscript k 1; I; 2 denote the inhomogeneity, the interphase and the matrix,
respectively. For expediency, dene modulus ratio gij mi =mj i; j 1; I; 2. In the
spherical coordinate system (r; y; j, with the origin coinciding with the centre of the
spherical inhomogeneity, the bulk and shear moduli of the interphase are assumed to be
power-law functions of r
kI r k0 rQ ;

mI r m0 rQ ;

nI const:;

(1)

where k0 ; m0 and Q are constants.

Assume that the inhomogeneity is subjected to a uniform eigenstrain e
. It is expedient
to split the eigenstrain tensor into its dilatational part e
m I2 and deviatoric part e
e , i.e.
e
e
m I2 e
e ,

(2)

where e
m tr e
=3. Thus, the solution under the uniform eigenstrain e
is obtained by the
superposition of the solutions under e
m I2 and e
e , respectively. In the following, the
solution under e
m I2 will be called a dilatational solution, and that under e
e a deviatoric
solution.
We solve rst the inhomogeneous inclusion problem with the graded interphase when the
only non-zero component of e
is e
zz . Obviously, the displacements in the interphase for the
dilatational part of e
zz have the following form in the spherical coordinate system:
uIr oIr r;

uIy 0;

uIj 0.

Substituting Eqs. (1) and (3) into the equilibrium equations gives


 I
d2 oIr 2 Q doIr
2nI Q
or


2
0.
r
1  nI
dr2
dr
r2

(3)

(4)

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1404

The solution of Eq. (4) is

oIr r F Izz rh5 G Izz rh6 ,
s#
"
1
2  10nI
h5 ; h6  Q 1  Q 2 9 Q
,
2
1  nI

where F Izz and G Izz are constants to be determined by the continuity and boundary
conditions, and the subscript zz indicates that the solution is for the eigenstrain
component e
zz .
Now we consider the deviatoric solution for e
zz . By analogy with the solutions for
homogeneous media, we assume that the displacements for the graded interphase have the
following form in the spherical coordinate system:
uIr r; y U Ir rP2 cos y;

uIy r; y U Iy r

dP2 cos y
;
dy

uIj r 0,

(6)

where P2 cos y is the Legendre polynomial of order two, and U Ir r and U Iy r are
unknown functions of r. Substituting Eq. (6) along with Eq. (1) into the equilibrium
equations gives
qU Ir
 Q 12  2nI Q 6U Iy
qr
qU Iy
q2 U Iy
1  2nI r2
Q 21  2nI r
0,
qr
qr2
qU Ir
q2 U Ir
2nI Q 4  5U Ir Q 21  nI r
1  nI r2
qr
qr2
I
qU y
0.
33  2nI Q 2U Iy  3r
qr
Q 4  2nI Q 2U Ir r

To solve Eq. (7), we introduce the operator (Wylie and Barrett, 1982)
rm

qm U
hh  1 . . . h  m 1U.
qrm

(8)

Substituting Eq. (8) into Eq. (7), while eliminating U Ir and U Iy , gives a characteristic
equation in h
1  nI h4 2  2nI 2Q  2QnI h3 Q2  Q2 nI Q
QnI 13nI  13h2 3Q2 nI  Q2 17QnI  15Q 14nI  14h
4Q2 nI 4QnI  4Q  24nI 24 0.

Eq. (9) has two sets of roots. The rst set contains real roots, and the second contains
complex ones. The four real roots h1 , h2 , h3 and h4 are
p
p
RT
RT
Q1
Q1
p ; h2 
 p ,
h1 
2
2
2 1  nI
2 1  nI
p
p
Q1
Q1
RT
RT
p ; h4 
 p ,
h3 
10
2
2
2 1  nI
2 1  nI

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1405

where
R 1  nI Q2 4Q 29  4QnI ,
q
T 2 1  nI 25Q2 nI 44Q  100QnI nI  1 Q2 3nI 1.

11

The corresponding displacement elds in the graded interphase are given in Eq. (6) with
U Ir r AIzz rh1 BIzz rh2 C Izz rh3 DIzz rh4 ,
eI rh1 B
e I rh 3 D
eI rh2 C
e I rh 4 ,
U Iy r A
zz
zz
zz
zz

12

eI , B
eI and D
eI , C
eI are constants to be determined from the
where AIzz ; BIzz ; C Izz ; DIzz ; A
zz
zz
zz
zz
continuity and boundary conditions. Substituting Eq. (12) into Eq. (7) gives the
eI , B
eI ; D
eI ; C
eI
corresponding relations between AI ; BI ; C I ; DI and A
zz

ei
Xi X

zz

zz

zz

zz

zz

zz

zz

12  12nI Q1  2nI  1  2nI hi Q 1 h2i 

,
2QnI  Q  hi  4 4nI

(13)

where i 1; 2; 3; 4, X1 ; X2 ; X3 and X4 stand for AIzz ; BIzz ; C Izz and DIzz , respectively, and
eI ; B
eI and D
e1; X
e2; X
e 3 and X
e 4 for A
eI ; C
eI , respectively. The complex roots (two pairs of
X
zz
zz
zz
zz
complex conjugate roots) of Eq. (9) are
h1 m1 id 1 ;

h2 m1  id 1 ;

h3 m2 id 2 ;

h4 m2  id 2 ,

(14)

where
Q1
;
m1 m2 
2

p
RT
d 1 p ;
2 1  nI

p
RT
d 2 p .
2 1  nI

(15)

The displacement elds in the graded interphase corresponding to these complex roots are
again given by Eq. (6) with
U Ir r AIzz cosd 1 ln rrm1 BIzz sind 1 ln rrm1
C Izz cosd 2 ln rrm2 DIzz sind 2 ln rrm2 ,
eI cosd 1 ln rrm1 B
eI sind 1 ln rrm1
U Iy r A
zz
zz
eI cosd 2 ln rrm2 D
eI sind 2 ln rrm2 .
C
zz
zz

16

Substituting Eq. (16) into Eq. (7) gives the corresponding relations between AIzz ,
eI ; B
eI ; D
eI ; C
eI , i.e.
BIzz ; C Izz ; DIzz and A
zz
zz
zz
zz
AIzz

HA
X

(17)

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1406

in which
eI f1  2nI 2 Q2 161  nI 1  2nI Q 481  nI 2
HA A
zz
 m1 Q2 1  2nI 2 2Q1  2nI 2  3nI  81  n2I 
1  2nI 2m21 1  nI Q 2  m21  m31  d 21 2QnI
eI f1  2nI 2 Q2  61  2nI 1  nI Q
 3 4nI m1 g B
zz
 81  nI 2  nI d 1  1  2nI 2d 1 m1 Q  2QnI 4  4nI
d 1 m21 d 31 g,
X d 21 m1 Q 4  2QnI  4nI 2 .

18

eI , B
eI and d 1 , respectively. C I
eI and d 1 in H A with B
eI , A
BIzz is obtained by replacing A
zz
zz
zz
zz
zz
eI , B
eI , D
eI , m1 and d 1 in H A and X with C
eI , m2 and d 2 ,
is obtained by replacing A
zz
zz
zz
zz
eI , B
eI , m2 and
eI , m1 and d 1 with D
eI , C
respectively, and DIzz is obtained by replacing A
zz
zz
zz
zz
d 2 , respectively.
In the inhomogeneity and matrix, the elastic solutions are also given by Eqs. (5), (6) and
(12) but with Q 0, since their elastic moduli are constant. This means that h1 3; h2 1;
h3 2; h4 4; h5 1; h6 2. The solutions in Eqs. (5), (6) and (12) as applied to the
inhomogeneity and matrix are identied with the superscripts 1 and 2, respectively.
In the Cartesian coordinate system, the deviatoric solution in Eq. (6) under e
zz has the
following form:

2

1 I
z
x
xz2
 3U Iy 3 ,
Ur 3 2  1
2
r
r
r

2

1
z
y
yz2
 3U Iy 3 ,
uIy U Ir 3 2  1
2
r
r
r

2




1 I
z
z
z2 z
I
I
3U y 1  2
,
uz U r 3 2  1
2
r
r
r r
uIx

19

where U Ir and U Iy are given in Eq. (12) for the case of real roots of Eq. (9), or Eq. (16) for
the case of complex roots. Thanks to the spherical symmetry of the inhomogeneity
problem under consideration, the deviatoric solutions under e
xx and e
yy can be obtained by
the simultaneous permutation of the subscripts and the coordinates x; y and z from the
displacements in Eq. (19). This procedure can also be found in the paper of Duan et al.
(2005). Thus, it will not be reproduced here. However, as can be seen from Eq. (19), the
displacements under e
xx and e
yy will also contain the functions U Ir and U Iy given in Eq. (12),
but the subscript zz should be replaced by xx and yy, respectively.
Now the solutions under shear eigenstrains e
xy , e
xz and e
yz will be sought. We consider
rst the solution under shear eigenstrain e
xy . Following a procedure similar to those in the
works of Christensen and Lo (1979), and Duan et al. (2005), namely, by assuming that the
displacements contain some unknown functions of r and substituting them into the
equilibrium equations to solve the unknown functions, the displacements for the graded

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1407

interphase in the spherical coordinate system can be obtained

3
uIr U Ir r sin2 y sin 2j,
2
3
I
uy U Iy r sin 2y sin 2j,
2
uIj 3U Iy r sin y cos 2j,

20

where U Ir r and U Iy r are obtained from Eq. (12) or (16) by replacing the subscript zz with
xy. Wang and Jasiuk (1998) gave the elastic solution of a spherical inhomogeneity with an
inhomogeneous interphase (with the power-law variation in elastic moduli) in an innite
matrix under non-vanishing remote shear stress s0xy . Their solution is similar to those in
Eq. (20). Luo and Weng (1987) gave the elastic solution of a spherical inhomogeneity with
a homogeneous interphase in an innite matrix under non-vanishing eigenstrains
e
xx e
yy . Their solution can be obtained directly from Eqs. (20) after a p=4 coordinate
transformation and setting Q 0. Using the same procedure as above, we get the
displacements under shear eigenstrain e
xz
3
uIr U Ir sin 2y cos j; uIy 3U Iy cos 2y cos j;
2
and those under shear eigenstrain e
yz

uIj 3U Iy cos y sin j

(21)

3
uIr U Ir sin 2y sin j; uIy 3U Iy cos 2y sin j; uIj 3U Iy cos y cos j,
(22)
2
where U Ir r and U Iy r in Eqs. (21) and (22) are obtained from Eq. (12) or (16) by replacing
the subscript zz with xz and yz, respectively.
We have obtained above the basic solutions under six different eigenstrains. Using the
relations between the displacements and strains, and the Hookes law, the stress elds can
be obtained. The constants Akpq ; Bkpq ; C kpq ; Dkpq ; F kpp and Gkpp p; q x; y; z under six different
eigenstrains are determined from the continuity and boundary conditions
u1 e
 x uI ;
I

u u ;
2

u 0;

r1  N1 rI  N1

r  N2 r  N2
2

r 0

at G1I ,

at GI2 ,

at r ! 1,

23

where N1 and N2 are the unit normal vectors to G1I and GI2 , respectively. It is found that under
e
xx 1; e
yy 1; e
zz 1; e
xy 1; e
xz 1 and e
yz 1, respectively, the constants Akpq p; q
x; y; z are equal for each of k 1; I; 2, e.g., Akxx Akyy Akzz Akxy Akxz Akyz . The
constants Bkpq , C kpq , Dkpq , F kpq and G kpq also obey their own respective identities. Therefore, for
brevity, we introduce constants Ak , Bk , C k , Dk , F k and Gk for the inhomogeneity and matrix
i.e. k 1; 2 such that
Ak 

Akpq
;
12nk

Bk 

Bkpq
;
2

F k  F kpp ; Gk  G kpp ,

Ck 

C kpq
;
25  4nk

Dk  

Dkpq
,
3
24

where pp xx; yy; zz and pq xx; yy; zz; xy; xz; yz denote the loading cases e
xx a0; e
yy a0;
e
zz a0; e
xy a0; e
xz a0 and e
yz a0, respectively. In the inhomogeneity (k 1), C 1 ; D1 and G 1

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1408

vanish; in the matrix (k 2), A2 ; B2 and F 2 vanish. For the graded interphase we introduce
constants M i i 1; 2; . . . ; 6 and N j j 1; 2; 3; 4 such that
M 1  AIpq ;

M 2  BIpq ;

M 3  C Ipq ;

M 6  GIpp ;

eI ;
N1  A
pq

eI ;
N2  B
pq

M 4  DIpq ;
eI ;
N3  C
pq

M 5  F Ipp ,
eI .
N4  D
pq

25

From Eq. (25), it follows that the relations between M i and N i i 1; 2; 3; 4 obey Eq. (13)
when the roots are real (e.g. homogeneous interphase), and Eqs. (17) and (18) when they are
complex.
Therefore, the nal elastic elds in the inhomogeneity and matrix contain the constants
Ak , Bk , C k , Dk , F k and Gk (k 1; 2), and those in the interphase the constants M i i
1; 2; . . . ; 6 and N j j 1; 2; 3; 4. These constants are easily obtained from the corresponding continuity conditions and remote boundary conditions in Eq. (23), but the expressions
are generally lengthy for the three-phase conguration of Fig. 1. Therefore, they are not
reproduced here. Knowing these constants, the Eshelby tensors for the three phases can be
calculated from the formulas given in the next section, where we shall also discuss their
general properties. Detailed expressions for the constants for spherical inhomogeneous
inclusions in nite domains will however be given in Section 4.
3. Eshelby tensors in three phases
The Eshelby tensors relate the total strains ek in the three phases to the prescribed
uniform eigenstrain in the inhomogeneity, i.e.
ek r Sk r : e
;

k 1; I; 2.

(26)

Because of the geometrical and physical symmetry of the problem under consideration, the
Eshelby tensors in the three phases are all transversely isotropic tensors with any of the
radii being an axis of symmetry. Moreover, these Eshelby tensors here are generally
position-dependent. Using the Walpole notation (Walpole, 1981) for transversely isotropic
tensors, a fourth-order tensor Sk r with the above-mentioned radial symmetry can be
expressed in a concise matrix form
k

e r  E
e ,
Sk r S

(27)

in which
ek r Sk1 r S k2 r S k3 r
S
e E1
E

E2

E3

E4

E5

Sk4 r S k5 r S k6 r ,
E6 ,

(28)

(29)

where r (r rn) is the position vector. n ni ei is the unit vector along the radius passing
the material point at which the Eshelby tensor is calculated. ni is the direction cosine of r
and i 1; 2; 3 denote x-, y- and z-directions, respectively. S kp r p 1; 2; . . . ; 6 are
functions of r, and Ep p 1; 2; . . . ; 6 are the six elementary tensors introduced by
Walpole (1981).

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1409

From the elastic solutions in Section 2, the Eshelby tensors in the inhomogeneity and the
e2 r being
e1 r and S
matrix are given by Eq. (27) with S
3T
1 B1 2F 1 37  8n1 A1 r2
7
6
6 1 2B1 F 1 36n1 A1 r2 7
7
6
6 1 3B1 37  4n1 A1 r2 7
1
7
6
e r 6
S
7 ,
6 1 3B1 37 2n1 A1 r2 7
7
6
7
6
B1 F 1  18n1 A1 r2
5
4
2
B1 F 1  37  8n1 A1 r

(30)

3T
6D2 =r5 2G2  21 n2 C 2 =r3
7
6
6 12D2 =r5  2G2 25  4n2 C 2 =r3 7
7
6
7
6
3D2 =r5 6C 2 1  2n2 =r3
2
7
6
e r 6
S
7 ,
5
3
12D2 =r 6C 2 1 n2 =r
7
6
7
6
6 6D2 =r5  2G 2  5  4n2 C 2 =r3 7
5
4
6D2 =r5 G 2 41 n2 C 2 =r3

(31)

where A1 ; B1 , F 1 , C 2 ; D2 and G 2 are given in Eq. (24). The Eshelby tensor in the graded
eI r being
interphase has a more complicated form with S
2

4
P

6
P

3T

2li ci ei  4li M i
7
6
i5
i1
7
6
7
6 4
6
P
7
6P
6 li ci 2ei f i gi 4ki l i  2li hi M i 7
7
6 i1
i5
7
6
7
6
4
P
7
6
7
6
2l
e
i i
7
6
i1
I
1
7
6
e r  6
S
7 .
4
P
7
26
7
6
2li ei ki
7
6
i1
7
6
7
6
4
6

 P
P
7
6
7
6
li ci gi  2li hi M i
7
6
i5
i1
7
6
7
6
4
6
P
P
5
4
li ci f i  2li M i
i1

(32)

i5

When the roots of Eq. (9) are real, li , ci ; ei ; f i ; gi ; ki and l i i 1; 2; 3; 4 in Eq. (32) are
li rhi 1

i 1; 2; . . . ; 6,

ci M i ;

ei 3N i ;

ki

32 M i

f i 3M i 6N i ;

N i hi  3;

gi M i hi  1,

l i 3M i hi  3 6N i hi  3.

33

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1410

The constants M i i 1; 2; . . . ; 6 and N j j 1; 2; 3; 4 are given in Eq. (25). When the

roots are complex, li in Eq. (32) are
l1 rm1 1 cosd 1 ln r;

l2 rm1 1 sind 1 ln r;

l4 rm2 1 sind 2 ln r;

l5 rh5 1 ;

l3 rm2 1 cosd 2 ln r,

l6 rh6 1 .

34

Then ci ; ei ; f i ; gi ; ki and l i i 1; 2; 3; 4 are given below. For i 1

c1 M 1 ; e1 3N 1 ; f 1 3M 1 6N 1 ,
g1 M 1 m1  1 M 2 d 1 ; k1 32 M 1 N 1 m1  3 N 2 d 1 ,
l 1 3M 1 m1  3 6N 1 m1  3  3M 2 d 1 6N 2 d 1 .

35

For i 2 the constants can be obtained by replacing M 1 , N 1 , M 2 , N 2 , m1 and d 1 in

Eq. (35) with M 2 , N 2 , M 1 , N 1 , m1 and d 1 , respectively, for i 3 by replacing them with
M 3 , N 3 , M 4 , N 4 , m2 and d 2 , respectively, and for i 4 with M 4 , N 4 , M 3 , N 3 , m2 and d 2 ,
respectively.
When the interphase is homogeneous (non-graded, Q 0), h1 3; h2 1; h3 2;
h4 4; h5 1; h6 2. M i and N i i 1; 2; 3; 4 are related to each other through
Mi Ni

12  12nI  1  2nI hi h2i 

;
hi 4  4nI

i 1; 2; 3; 4.

(36)

For consistency with the expressions in Eqs. (30) and (31), for the homogeneous
interphase, we introduce constants AI ; BI ; C I ; DI ; F I and G I . The relations between them
and M 1 , M 2 , M 3 , M 4 , M 5 and M 6 in Eq. (25) are
AI 

M1
;
12nI

BI 

M2
;
2

CI 

M3
;
25  4nI

DI  

M4
;
3

F I  M 5;

GI  M 6 .
(37)

e r is given by
Therefore, for the homogeneous interphase (Q 0), S
3T
2
BI 2F I 37  8nI AI r2 2G I  21 nI C I  r13 6DI r15
7
6
6 2BI F I 36nI AI r2  2GI 25  4nI C I  r13 12DI r15 7
7
6
7
6
3BI 37  4nI AI r2 61  2nI C I r13 3DI r15
7
6
I
e
7 .
6
S r 6
2
1
1
7
3BI 37 2nI AI r 61 nI C I r3  12DI r5
7
6
7
6
6 BI F I  18nI AI r2  2G I  5  4nI C I  r13  6DI r15 7
5
4
BI F I  37  8nI AI r2 G I 41 nI C I  r13  6DI r15

(38)

Under dilatational eigenstrain e
e
m I2 , the total strain in the inhomogeneity is given by
e1 e
m S1 : I2 . It can be veried that S1 : I2 is a constant tensor and thus the stress eld in
the inhomogeneity is uniform even in the three-phase conguration with a graded
interphase. If the inhomogeneity, the interphase and the matrix have the same elastic
moduli, then the Eshelby tensors given in Eqs. (30) and (31) reduce to the classical interior
and exterior Eshelby tensors for a spherical inclusion.
As we shall demonstrate in Section 7, the volume average Eshelby tensors can be used to
predict the effective moduli of composites. Here, we give the volume average of the

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1411

Eshelby tensor for the spherical inhomogeneous inclusion in the studied three-phase model
when the interphase is a (non-graded) homogeneous material. The average tensor is
dened as
Z
1
1
e1 rE
e T dV ,
S
(39)
S
V 1 V1
where V 1 is the volume of the inhomogeneity. Performing the volume integration, it is
1
found that S is an isotropic tensor



63A1
1
S 1 3F 1 J 1
3B1 K,
(40)
5
where
J 13 I2  I2 ;

K 13 I2  I2 I4s ,

(41)

with I2 and I4s being the second- and fourth-order symmetric identity tensors,
respectively. For an inhomogeneous inclusion with a non-graded interphase, the procedure
for the solutions of the constants F 1 , A1 and B1 has been depicted above. Their expressions
will not be given here for they are lengthy. Instead, we give the detailed expressions for
these constants when the spherical inhomogeneity has the same elastic constants as the
interphase, namely, when the inhomogeneous inclusion degenerates into an inclusion in a
spherical region which, in turn, is embedded in an alien innite matrix. In this case,
g1I m1 =mI 1, g2I m2 =mI g21 m2 =m1 and n1 nI , and the innite matrix material
has different elastic moduli. Then the constants F 1 , A1 and B1 are
21  2n1
f1  r3 1  g21 1 n1  4g21 n1 2g21 g,
3H 1
21  r2 r5 1  g21
A1
,
H2
2r3 H 12
7  5n1

B1
,
15H 2 8  7g21 52 g21 n2  45  45n1
F1 

42

with r a=b and

H 1 31  n1 1 2g21 1  4g21 n1 ,
H 2 31  n1 71 4g21 51  8g21 n1 ,
H 12 200n21  300n1  63r2 1757  5n2 g221 3g21 f75n1 7  9n2
 25n21 13  15n2  725 3r2  55 3r2 n2 g
4  5n2 25n21 126r2  175.

43

1
When g21 1, n1 n2 and r 1, S1 in Eq. (27) and S in Eq. (40) degenerate into the
classical Eshelby tensor, denoted by S0 , for a spherical inclusion in an innite homogeneous
medium.

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1412

4. Eshelby tensors in nite domains

4.1. Fixed displacement boundary condition
Letting the moduli of the innite matrix in the three-phase problem in Section 3 tend to
innity, we obtain the solution for a spherical inhomogeneous inclusion in a nite domain
whose outer boundary is xed. In this case, when the shell is homogeneous (Q 0), the
Eshelby tensors in the nite domain are given through Eqs. (27), (30) and (38), and the
constants A1 ; B1 ; F 1 , AI ; BI ; C I ; DI , F I and G I are
1  2n1 2 r3  4  r3 nI 
175g1I r5 1  r2 1  nI
; A1
,
W1
W2
W3
r3 g1I 1 n1 1  2nI
FI
B1
; FI 
; GI  3 ,
W1
3W 2
r
5g1I r5 1  r2
r3 g1I W 4
AI
28  40n1 g1I 7 5n1 ; BI
,
W2
3W 2
5g
2g
C I 1I W 5 r7  W 6 7  10nI ; DI 1I W 5 r5  W 6 7  10nI ,
3W 2
W2

F1 

44

where g1I m1 =mI and W i i 1; 2; . . . ; 6 are

W 1 3g1I 1  r3 1 n1 1  2nI 31  2n1 2 r3  4  r3 nI ,
W 2 41  g1I 4  5nI W 5 r10 25W 7 r7 1  g1I W 6 126r5
 257  12nI 8n2I r3  2W 6 7  10nI 7 5nI  2g1I 4  5nI ,
W 3  44  5nI W 5 r10 25W 8 r7 63W 9 r5 25W 6 7  12nI 8n2I r3
2W 6 49  105nI 50n2I ,
W 4  44  5nI W 5 r7  W 6 175 63r2 300nI  200n2I ,
W 5 7  10n1 7 5nI  g1I 7 5n1 7  10nI ,
W 6  28 40n1  g1I 7 5n1 ,

45

with
W 7 27  10n1 7  n2I  g21I 7 5n1 7  12nI 8n2I
g1I 77  6nI 8n2I n1 49 66nI 20n2I ,
W 8  27  10n1 7  n2I g1I 714  21nI 4n2I n1 7  42nI 20n2I ,
W 9  2W 6  35g1I 1  nI .

46

When the inhomogeneity and the homogeneous shell have the same elastic moduli
(g1I 1; n1 nI ), the solution reduces to that of a spherical inclusion in a nite domain.
4.2. Traction-free boundary condition
Letting the moduli of the innite matrix in the three-phase conguration in Fig. 1
vanish, we obtain the solution for a spherical inhomogeneity in a nite domain with a
traction-free boundary condition. In this case, when the shell is homogeneous (Q 0), the
Eshelby tensors in the nite domain are given through Eqs. (27), (30) and (38), and the

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1413

constants A1 ; B1 ; F 1 , AI ; BI ; C I ; DI ; F I and GI are

21  r3 1  2n1 1 nI
350r5 g1I 1  r2 1  nI
; A1 
,
W 10
W 11
W 12
2r3 g1I 1  2nI 1 n1
F I 1 nI
,
B1
; FI 
; GI 3
W 10
2r 1  2nI
3W 11
10r5 g1I 1  r2 74 g1I  58  g1I n1 
r3 g1I W 13
AI 
; BI
,
W 11
3W 11
5g
g
C I  1I 4W 5 r7 W 6 7 5nI ; DI  1I 4W 5 r5 W 6 7 5nI ,
6W 11
W 11
F1

47

where W 5 and W 6 are given in Eq. (45), and W n n 10; 11; 12; 13 are
W 10  61  r3 1  2n1 1 nI  3g1I 1 n1 1 2r3 1  4r3 nI ,
W 11 41  g1I 7  5nI W 5 r10 50W 15 r7  2521  g1I W 6 r5
251  g1I 7  n2I W 6 r3 W 6 7 5nI 7  8g1I 51 2g1I nI ,
W 12  47  5nI W 5 r10  50W 14 r7  257  n2I W 6 r3
49  25n2I W 6  126108  g1I n1 78  3g1I 5g1I nI r5 ,
W 13  47  5nI W 5 r7 126W 6 r2  25W 6 7  n2I ,

48

with
W 14 n1 720 g1I  42g1I nI  201  g1I n2I 
 7141  g1I 21g1I nI  21 2g1I n2I ,
W 15 772 g1I g21I  6g1I 1 2g1I nI 21 2g1I 2 n2I 
n1 720  7g1I 5g21I  6g1I 11 10g1I nI  201 g1I  2g21I n2I .

49

When the inhomogeneity and the homogeneous shell have the same elastic moduli,
i.e. g1I 1; n1 nI , the solution reduces to that of a spherical inclusion in a free nite
domain.
It is noted that Li et al. (2005), and Wang et al. (2005) recently solved the circular and
spherical inclusion problems in nite circular and spherical regions with xed displacement
or traction-free boundary conditions, and gave the corresponding Eshelby tensors. The
present solutions are different from those of Li et al. (2005), and Wang et al. (2005) in that
the former are for inhomogeneous inclusions.
5. Stress concentration tensors in three phases
In this section, we solve the elastic eld in the three-phase region shown in Fig. 1 when a
uniform stress eld r0 is prescribed at innity. It is noted that previously, Herve and Zaoui
(1993) have solved the elastic eld in an innite medium containing a spherical
inhomogeneity with multiple homogeneous shells. Here, like the case for a general

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1414

eigenstrain, the solution for a general remote stress r0 is also obtained by superposition.
Under the respective remote loadings s0xx a0; s0yy a0; s0zz a0; s0xy a0; s0xz a0 and s0yz a0, the
elastic solutions have the same forms as those under the corresponding eigenstrains.
However, the usual stress and displacement continuity conditions at G1I and GI2 and
the remote boundary conditions should be satised. It is found that under s0xx 1; s0yy 1;
s0zz 1; s0xy 1; s0xz 1 and s0yz 1, respectively, the constants Akpq ; Bkpq ; C kpq ; Dkpq ; F kpp and
Gkpp p; q x; y; z in Eqs. (5), (6), (12) and (16) for the inhomogeneity (superscript 1), the
graded interphase (superscript I) and the matrix (superscript 2) still obey the identities like
Akxx Akyy Akzz Akxy Akxz Akyz . Thus, we introduce constants Ak , Bk , C k , Dk , F k and
Gk k 1; 2 for the inhomogeneity and matrix such that
12nk Ak s0pq  mk Akpq ;

2Bk s0pq  mk Bkpq ;

 3Dk s0pq  mk Dkpq ;

F k s0pp 

25  4nk C k s0pq  mk C kpq ,

21 nk
m Fk ;
1  2nk k pp

Gk s0pp  mk G kpp ,

50

where p; q x; y; z denote different loading cases. In the inhomogeneity (k 1), C 1 ; D1

and G 1 vanish; in the matrix (k 2), A2 and F 2 vanish. For the graded interphase we
introduce constants M i i 1; 2; . . . ; 6 and N j j 1; 2; 3; 4 such that
M 1 s0pq  mI AIpq ;

M 2 s0pq  mI BIpq ;

eI ;
N 1 s0pq  mI A
pq

eI ;
N 2 s0pq  mI B
pq

M 5 s0pp  mI F Ipp ;

M 6 s0pp  mI GIpp .

M 3 s0pq  mI C Ipq ;
eI ;
N 3 s0pq  mI C
pq

M 4 s0pq  mI DIpq ,
eI ,
N 4 s0pq  mI D
pq
51

From Eq. (51), it follows that the constants M i and N i i 1; 2; 3; 4 obey Eq. (13) when
the roots of Eq. (9) are real (e.g. homogeneous interphase), and Eqs. (17) and (18)
when they are complex. Therefore, the nal elastic elds in the inhomogeneity and
matrix contain the constants Ak , Bk , C k , Dk , F k and Gk (k 1; 2), and those in the
interphase the constants M i i 1; 2; . . . ; 6 and N j j 1; 2; 3; 4. These constants are
easily obtained from the corresponding continuity conditions and remote boundary
conditions.
The stress concentration tensors Tk x k 1; I; 2 relate the total stresses rk x in the
three phases to the prescribed uniform remote stress r0 ,
rk Tk r : r0 ;

k 1; I; 2.

(52)

The stress concentration tensors have the same properties as those of the Eshelby tensors;
therefore, Tk r can be expressed as follows:
e k r  E
eT,
Tk r T

(53)

where
e k r T k1 r T k2 r T k3 r
T

T k4 r

T k5 r T k6 r .

(54)

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1415

e 1 r is
In the inhomogeneity, T
2
6
6
6
6
1
6
e
T r 6
6
6
6
4

3T

2B1 2F 1 67 6n1 A1 r2
4B1 F 1  12n1 A1 r2
6B1 67  4n1 A1 r2
6B1 67 2n1 A1 r2
2B1 F 1 6n1 A1 r2
2B1 F 1  67 6n1 A1 r2

7
7
7
7
7
7 .
7
7
7
5

(55)

e 2 r is
In the matrix, T
2

1 12D2 =r5 4G 2  21  2n2 C 2 =r3

6
6 1 24D2 =r5  4G 2 25  n2 C 2 =r3
6
6
1 6D2 =r5 12C 2 1  2n2 =r3
2
6
e
T r 6
1  24D2 =r5 12C 2 1 n2 =r3
6
6
6 12D2 =r5  4G 2  5  n2 C 2 =r3
4
12D2 =r5 2G 2 41  2n2 C 2 =r3

3T
7
7
7
7
7
7 .
7
7
7
5

(56)

e I r is
In the inhomogeneous interphase, T
2

4
P

2li ci ei

6
P

i n2
4li 1h
12n2 M i

3T

7
6
i5
i1
7
6
7
6 4
6
P
7
6P
i n2 
6 li ci 2ei f i gi 4ki l i 2li hi 2h
M
i7
12n2
7
6 i1
i5
7
6
7
6
4
P
7
6
7
6
2li ei
7
6
i1
I
7
6
e r 6
T
7 .
4
P
7
6
7
6
2li ei ki
7
6
i1
7
6
7
6
4
6
P
P
7
6
hi 2hi n2 
7
6
li ci gi 2li 12n2 M i
7
6
i5
i1
7
6
7
6
4
6
P
P
5
4
1hi n2
li ci f i 2li 12n2 M i
i1

(57)

i5

When the roots of Eq. (9) are real, li is given by Eq. (33), and the other constants are
1
1 hi nI M i  6nI N i ; ei 3N i ,
1 2nI
1
31 hi nI M i 61 nI N i ,
fi
1 2nI
gi hi  1M i ; ki 32 M i hi  3N i ,

ci

l i 9  3hi M i  6hi  3N i .

58

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1416

When the roots are complex (two pairs of complex conjugate roots), li i 1; 2; . . . ; 6 are
given by Eq. (34). For i 1,
1
1 m1 nI M 1  6nI N 1 nI M 2 d 1 ; e1 3N 1 ,
1 2nI
1
31 m1 nI M 1 61 nI N 1  3nI M 2 d 1 ,
f1
1 2nI
g1 m1  1M 1  M 2 d 1 ; k1 32 M 1 m1  3N 1 N 2 d 1 ,
c1

l 1 9  3m1 M 1  6m1  3N 1 3M 2 d 1  6N 2 d 1 .

59

For i 2, the constants can be obtained by replacing M 1 , N 1 , M 2 , N 2 , m1 and d 1 in

Eq. (59) with M 2 , N 2 , M 1 , N 1 , m1 and d 1 , respectively, for i 3 by replacing them with
M 3 , N 3 , M 4 , N 4 , m2 and d 2 , respectively, and for i 4 with M 4 , N 4 , M 3 , N 3 , m2 and d 2 ,
respectively.
When the interphase is homogeneous (Q 0), h1 3; h2 1; h3 2; h4 4; h5 1;
h6 2. For consistency with the expressions in Eqs. (55) and (56), we introduce
constants AI ; BI ; C I ; DI ; F I and G I such that
12nI AI  M 1 ;

2BI  M 2 ; 25  4nk C I  M 3 ,
21 nI
 3DI  M 4 ; F I 
M 5 ; GI  M 6 .
1  2nI
e I r is given by
Therefore, for the homogeneous interphase (Q 0), T
3T
2
2BI 2F I 67 6nI AI r2 12DI r15 4GI  21  2nI C I  r13
7
6
7
6
4BI F I  12nI AI r2 24DI r15  4G I 25  nI C I  r13
7
6
7
6
1
1
2
6BI 67  4nI AI r 6DI r5 121  2nI C I r3
7
6
I
e
7 .
6
T r 6
2
1
1
7
6B

67

2n
A
r

24D

121

n
C
I
I
I
I r5
I
I r3
7
6
7
6
2
1
1
7
6
2B

6n
A
r

12D

4G

5

n
C

I
I
I I
I r5
I
I
I r3
5
4
1
1
2
2BI F I  67 6nI AI r  12DI r5 2G I 41  2nI C I  r3

60

(61)

The volume average of the stress concentration tensor T for the inhomogeneity, which is
useful in micromechanical approaches, is given by
Z
1
1
e T dV 3F 1 J 6 21A1 5B1 K.
e 1 rE
(62)
T
T
V 1 V1
5
1
When g1I 1 and nI n1 , T1 in Eq. (53) and T in Eq. (62) degenerate into the classical
0
stress concentration tensor T for a spherical inhomogeneity embedded in an alien innite
matrix.

6. Strain distributions in coreshell nanoparticles

A coreshell nanoparticle is usually composed of two different materials, e.g. CdSe
coated CdS, ZnS coated CdSe, ZnSe coated CdSe, and CdS coated CdSe, etc. Many
coreshell nanoparticles are nearly spherical and have a diameter of a few (35) (Danek
et al., 1996; Peng et al., 1997; Dabbousi et al., 1997) to 11 nanometres (Kim et al., 2005).
Because of the mismatch of the lattices of the materials, signicant mismatch strains may

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

Pex

Pex
CdS

CdTe
Pex

Pex

(a)

1417

Pex

Pex

(b)

CdS
ZnS

Pex

Pex

Fig. 2. (a) CdTe/CdS coreshell particle; (b) ZnS/CdS coreshell particle.

develop in coreshell nanoparticles (e.g. Rockenberger et al., 1998; Little et al., 2001). The
lattice mismatch strain and the thermal strain can be treated as eigenstrains. In addition to
the lattice mismatch, the large surface stress and applied external pressure can also produce
a substantial lattice contraction (Little et al., 2001; Itskevich et al., 1998). In this section,
we will use the Eshelby tensors in the nite domain with the traction-free boundary
condition to analyze the elastic strain eld in a coreshell particle under an eigenstrain e
, a
uniform and isotropic surface stress t and external hydrostatic loading Pex , as shown in
Figs. 2(a) and (b). The strain eld in the core and shell can be obtained by using the
Eshelby tensors in Eqs. (27), (30) and (38) for an arbitrary uniform eigenstrain prescribed
in the core, with the constants A1 ; B1 ; F 1 , AI ; BI ; C I ; DI ; F I and G I given in Eqs. (47) and
(48). The elastic strain tensors e1el and eIel in the core and shell induced by e
, t and Pex are
e1el S1 : e
 e
e1 t; Pex ;

eIel SI : e
eI t; Pex ,

(63)

where S1 and SI are the Eshelby tensors in the core and shell, and they are given in
Eqs. (27), (30) and (38). e1 t; Pex and eI t; Pex in Eq. (63) are the elastic strain tensors in
the core and shell due to t and Pex . The components of e1 t; Pex and eI t; Pex in the
spherical coordinate system are



mI 3kI 4mI
Pex
1
1
1


2t
err t; Pex eyy t; Pex ejj t; Pex 
,
3w
mI
2L2
L2
64
eIrr t; Pex L1  3 ; eIyy t; Pex eIjj t; Pex L1 3 ,
r
r
where t
t=bmI , and
w 41  r3 kI mI k1 3kI 4r3 mI ,






mI 3k1 4mI
Pex
mI k1  kI
Pex




2t
2t
; L2
.
L1 
3w
w
mI
mI

65

Rockenberger et al. (1998) calculated the elastic strain distribution in the CdS coated CdTe
coreshell system (CdTe/CdS, Fig. 2(a)) due to a dilatational eigenstrain e
e
m I2 by
assuming that the elastic constants of the CdTe core and CdS shell are the same within the
framework of continuum mechanics. In order to compare our results with those of
Rockenberger et al. (1998), we consider the strain eld induced by a dilatational
eigenstrain e
e
m I2 . For the coreshell structure, the mist strain induced by the
different lattice constants of the core and shell is e
m aco  ash =ash , where aco and ash are

ARTICLE IN PRESS
1418

H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

Fig. 3. Distributions of normalized elastic strains ezz =e
m and exx =e
m in a CdTe/CdS coreshell nanoparticle
subjected to e
m and t. It is noted that the core (0pzp1:0) is under hydrostatic compression. Thus, the curves of
exx and ezz merge into one in the core for each of the three cases.

the lattice constants of the core and shell, respectively. For example, the mist strains are
11.6% for the CdTe/CdS coreshell structure and 7:0% for ZnS/CdS (Fig. 2(b)). Here,
we give the numerical results for the CdTe/CdS nanoparticle for three cases: (I) The core
and the shell have the same elastic constants with an eigenstrain e
m I2 in the core; (II) The
core and the shell have different elastic constants with an eigenstrain strain e
m I2 in the
core; (III) A surface stress t on the outer surface of the particle is superimposed on the
strain eld of case (I). For the CdTe/CdS coreshell system, the outer radius rsh of the
nanoparticle is 0.9 nm with r 0:84 (Rockenberger et al., 1998). The elastic strains
exx exx eyy and ezz along the z-axis (a radius) are plotted in Fig. 3. It is found that for
this material combination the assumption that the core and the shell have the same elastic
constants generates a relative difference eII  eI=eI 56% of the strain in the core,
because the elastic constants of CdTe (bulk modulus 41.9 GPa, Poisson ratio 0.41) and
CdS (bulk modulus 62.3 GPa, Poisson ratio 0.4) are different. Therefore, the difference in
the elastic constants results in a signicant difference in the strain eld, which may have
great effect on the physical properties. If we assume that there is a surface stress
t 1 N=m, then the effect of the surface stress on the magnitude of the strains is
remarkable, and it may cause a substantial lattice contraction. The CdTe core is under
hydrostatic compression, and the CdS shell is under biaxial tension in the tangential
directions and compression in the radial direction.
7. Micromechanical scheme based on Eshelby and stress concentration tensors of three-phase
conguration
Many micromechanical schemes have been developed to predict the effective elastic
constants of linear composites (e.g. Aboudi, 1991; Bornert et al., 1996; Nemat-Nasser and
Hori, 1999; Torquato, 2002; Milton, 2002). The framework for predicting the effective
properties of nonlinear composites has also been well developed (Talbot and Willis, 1985;
Pontecastaeda and Suquet, 1998; Willis, 2000). The work of Segurado and LLorca (2002)
for linear composites containing non-overlapping identical spheres showed that for rigid
spheres, Torquatos third-order approximation (TOA, Torquato, 1998) gives the effective

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1419

moduli closest to the numerical computation; for a glass-sphere/epoxy composite, the

predictions of the generalized self-consistent method (GSCM, Christensen and Lo, 1979)
and the TOA are practically identical, and they are very close to the numerical
computation; and for spherical voids, the GSCM, the TOA and the numerical
computation give practically identical results. In this case, the prediction of the
MoriTanaka method (MTM, Mori and Tanaka, 1973; Benveniste, 1987) is also very
close to the other two theoretical schemes and the numerical computation. However, it is
noted that the MTM may severely underestimate the effective shear moduli of composites
containing hard inhomogeneities, especially at large volume fractions, as is shown in the
papers of Segurado and LLorca (2002), and Ma et al. (2004). It would therefore appear
that from an overall point of view, the GSCM and the TOA tend to give the best results. It
is noted that the three-phase model (GSCM) takes into account the matrix atmosphere,
which is important in the prediction of the effective properties (Zheng and Du, 2001).
Therefore, in this section, using the three-phase conguration in Fig. 1, we propose a
micromechanical scheme to predict the effective moduli of composites containing spherical
particles based upon the solutions of the inhomogeneity problems obtained in the previous
sections.
We consider a two-phase composite composed of a continuous matrix and randomly
distributed spherical particles. The effective stiffness tensor C
and compliance tensor D

of the composite can be calculated from the following expressions (Hill, 1963):
1,
C
C2 f C1  C2 : E

(66)

1,
D
D2 f D1  D2 : T

(67)

where C1 and D1 denote the stiffness and compliance tensors of the particles, respectively,
1 in Eq. (66) is
and C2 and D2 those of the matrix. f is the volume fraction of the particles. E
1

the strain concentration tensor in a particle, and T in Eq. (67) is the stress concentration
tensor. They are dened as
1
e 1 E : e0 ;

1
r 1 T : r0 ,

(68)

where e0 and r0 are the uniform strain and stress tensors when the material is
homogeneous, and e 1 and r 1 are the volume average strain and stress tensors in a
1
particle, respectively. Among various micromechanical schemes, the GSCM evaluates E
1
or T based on the three-phase model where a particle is embedded in a nite matrix shell
that, in turn, is embedded in an innite equivalent medium with the yet-unknown effective
properties of the composite, as shown in Fig. 4(a). It is noted that the matrix shell (region
2) in Fig. 4(a) corresponds to the interphase I in Fig. 1, and the equivalent homogeneous
medium (region c) corresponds to the innite matrix in Fig. 1. Because we have given the
stress concentration tensor (Eq. (62)) for the inhomogeneity in the three-phase
conguration, we can directly substitute it into Eq. (67) to calculate the effective
compliance tensor of the composite. By doing so, it is found that the obtained effective
bulk and shear moduli of the composite are identical to those given by Christensen and Lo
(1979), and Huang et al. (1994). The effective shear modulus needs to be solved from a
1 and
quadratic equation. Likewise, we can calculate the strain concentration tensor E
substitute it into Eq. (66) to calculate the effective stiffness tensor, which is found to be

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1420

0 (0 )

0 (0 )

m, m

1, 1
=

2
2

c
(a)

(b)

(c)

Fig. 4. The GSCM (three-phase) conguration for a composite containing spherical particles (a). The Eshelby
equivalent inclusion method in a volume average sense for the three-phase conguration ((b) and (c)). Region 1
denotes a particle in the two-phase composite; region 2 represents the matrix; and region c denotes the equivalent
homogeneous medium (composite).

identical to that obtained from Eq. (67). Thus, the results of the GSCM are conrmed in a
unied tensorial approach.
In the following, instead of using the exact average stress concentration tensor in
Eq. (62), we shall introduce the approximate volume average stress and strain
concentration tensors by applying the Eshelby equivalent inclusion method to the threephase conguration. To this end, we calculate the average stress concentration tensor by
applying the equivalent inclusion method to the region 1, i.e. the spherical inhomogeneity,
in Fig. 4(a). Assume that the volume average stress in this inhomogeneity in the threephase conguration in Fig. 4(a) is r 1 while the remote stress is r0 . For the same remote
stress, when we replace the spherical inhomogeneity with the stiffness tensor C1 by the
matrix material with the stiffness tensor C2 , the volume average stress in the same region is
denoted by r m , as shown in Fig. 4(b). This volume average stress r m can be related to the
remote stress by the relation
r m B : r0 ,

(69)

where the fourth-order tensor B is equal to the classical stress concentration tensor T0 .
Generally, r m is different from r 1 . As in the classical Eshelby equivalent inclusion method,
the spherical matrix region 1 is further given a uniform eigenstrain e
(Fig. 4(c)) such that
the following equivalency condition is satised:
C1 : em e 0 C2 : em e 0  e
,

(70)

where e m D2 : r m D2 : B : r0 . As in the work of Luo and Weng (1987), the disturbed

1
strain e0 is related to the interior Eshelby tensor S through
1
e0 S : e
,
(71)
1
where S is given in Eq. (40). From the above relations, the volume average stress in the
spherical particle can be obtained

: r0 ,
r 1 C1 : e1 C1 : em e0  T

(72)

where
1


T I4s  C1 : S : D1  D2 1 : C1 : D2 : B  a
J b
K.

(73)

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1421

, which relates the volume average strain e 1 to

Likewise, the strain concentration tensor E
0
1

: e0 , is given as
the remote strain e in the relation e E

I4s  S 1 : D2 : C2  C1 1 : A,
E

(74)

where the fourth-order tensor A relates the volume average strain em to the remote strain e0
through the relation em A : e0 (Fig. 4(b)), and A D1 : B : C2 .
It is found that the dilatational component a
of the approximate average stress

in Eq. (73) is identical to that of the exact average stress
concentration tensor T
1 . The expression of the deviatoric component b
is different from
concentration tensor T
that of the exact one. Fig. 5 shows the variations of the ratios b1 =b0 and b
=b0 with the
volume fraction of the inhomogeneities for two composites with different stiffness
contrasts. b0 is the deviatoric component of the classical stress concentration tensor for an
inhomogeneity in an innite medium. In calculating b1 and b
shown in Fig. 5, we need the
effective elastic constants mc and nc of the equivalent homogeneous medium (region c) in
Fig. 4. These constants are calculated using the GSCM. It is seen that the values of b
and
b1 are practically identical for these composites. The numerical results for other
combinations of the stiffnesses of the three phases also exhibit the similar feature.


Therefore, we can conclude that the approximate stress concentration tensor T is very
1

accurate. Thus, we shall replace T in the general expression (67) with T to predict the
effective compliance tensor of composites containing randomly distributed spherical
particles.


When substituting T into Eq. (67), it is found that the obtained effective bulk modulus
1 . Thus, it will not be discussed further. The
is identical to that given by the GSCM using T
effective shear modulus, denoted by mc , needs to be solved from the following quadratic
equation in m
m
mc =m2 :
Am2
Bm
C 0,

(75)

6
5

1/0(*/0 )

1/0 (1/2 = 1/10)

4
3

*/0 (1/2 = 1/10)

1/0 (1/2 = 10)
*/0 (1/2 = 10)

1
0.0

0.2

0.4

0.6

0.8

1.0

f
Fig. 5. Comparison of the deviatoric components (b1 and b
) of the exact volume average stress concentration
1

obtained from the Eshelby equivalent inclusion method in the volume average
tensor T and the approximate T
sense (n1 n2 0:3).

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1422

where
A 126f 7=3  252f 5=3 507  12n2 8n22 f 1  g12 47  10n2 N,
B 252f 7=3  504f 5=3 1503  n2 n2 f 1  g12  37  15n2 N,
C 126f 7=3  252f 5=3 257  n22 f 1  g12  7 5n2 N,

76

with N 7 5n2  2g12 4  5n2 and g12 m1 =m2 . It is seen that the coefcients A, B
and C are much simpler than their counterparts in the classical GSCM. In the following,
we shall compare the effective shear moduli given by the present model with those given by
other methods for various composites to conrm its accuracy.
We rst compare the effective shear moduli predicted by the present model with the
classical HashinShtrikman bounds (Hashin and Shtrikman, 1963), the GSCM
(Christensen and Lo, 1979), the TOA (Torquato, 1998) and the numerical computations
of Segurado and LLorca (2002) for three composites which contain stiff spheres with
m1 =m2 10; n1 n2 0:25; E 1 70 GPa, n1 0:2, E 2 3 GPa, n2 0:38; and rigid
spheres with n2 0:25, respectively. The comparisons are shown in Figs. 6(a)(c). In all
the gures, mc denotes the effective shear modulus of the composite. It is seen that the
effective shear modulus predicted by the present model is practically identical to those
10

4
HS upper bound
HS lower bound
GSCM
TOA
Present model

3
c (GPa)

c / 2

HS lower bound
GSCM
TOA
Present model
Numerical result

2
0.0

(a)

0.2

0.4

0.6

0.8

1
0.0

1.0

0.1

(b)

0.2

0.3

0.4

0.5

4
HS lower bound
GSCM
TOA
Present model
Numerical result

c / 2

1
0.0

(c)

0.1

0.2

0.3

0.4

0.5

Fig. 6. Comparison of the effective shear moduli obtained from the present model with other methods for
composites containing stiff spheres with m1 =m2 10; n1 n2 0:25 (a), glass beads E 1 70 GPa, n1 0:2,
E 2 3 GPa, n2 0:38 (b), and rigid spheres with n2 0:25 (c).

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425
1.0

1.0
HS upper bound
HS lower bound
GSCM
TOA
Present model

0.6

HS upper bound
GSCM
TOA
Present model
Numerical result

0.8
c / 2

0.8

c / 2

1423

0.6

0.4
0.4
0.2

(a)

0.0

0.2

0.4

0.6
f

0.8

1.0

(b)

0.2
0.0

0.1

0.2

0.3

0.4

0.5

Fig. 7. Comparison of the effective shear moduli predicted by the present model with other methods for
composites containing soft spheres with m1 =m2 1=10; n1 n2 0:25 (a), and voids with n2 0:25 (b).

predicted by the GSCM and TOA for the composite with m1 =m2 10 (Fig. 6(a)). For the
other two composites, the present predictions are also almost indistinguishable from those
of the GSCM (Figs. 6(b) and (c)). In Figs. 7(a) and (b), we compare the effective shear
moduli predicted by the present model with other methods for two composites containing
soft spheres and voids, respectively. It is seen that the present predictions are practically
identical to those of the GSCM, the TOA and the numerical results (for composite
containing voids). Moreover, it is found that the effective bulk and shear moduli obtained
by substituting the strain concentration tensor in Eq. (74) into Eq. (66) are identical to

in Eq. (67). This feature demonstrates that the present model is
those obtained by using T
also self-consistent.
8. Conclusions
The Eshelby and stress concentration tensors are derived for a spherical inhomogeneity
with a graded shell embedded in an innite elastic matrix. The general Eshelby tensors are
then specialized to inhomogeneous inclusions in nite domains under xed displacement or
traction-free boundary conditions. These tensors are very useful for solving many
problems in mechanics and materials science, e.g. the strain elds in coreshell
nanoparticles which have a novel composite structure and myriad application in many
elds. Finally, a micromechanical scheme is proposed to predict the effective moduli of
composites containing spherical particles. The main advantage of this scheme is that,
whilst its predictions of the effective moduli are almost identical to those of the classical
generalized self-consistent method (GSCM) and the third-order approximation (TOA), the
resulting expressions are simple and concise.
Acknowledgements
This work is supported by the National Natural Science Foundation of China under
Grant nos. 10525209 and 10372004. The authors thank Professor J. LLorca and
Dr J. Segurado for kindly providing their computational results. The authors also thank
Professor B. L. Karihaloo of Cardiff University and Professor Gengkai Hu of Beijing
Institute of Technology for helpful comments and discussions.

ARTICLE IN PRESS
1424

H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

References
Abe, M., Suwa, T., 2004. Surface plasma resonance and magneto-optical enhancement in composites containing
multicore-shell structured nanoparticles. Phys. Rev. B 70, 235103-1235103-15.
Aboudi, J., 1991. Mechanics of Composite MaterialsA Unied Micromechanical Approach. Elsevier,
Amsterdam.
Benveniste, Y., 1987. A new approach to the application of MoriTanakas theory in composite materials. Mech.
Mater. 6, 147157.
Bornert, M., Stolz, C., Zaoui, A., 1996. Morphologically representative pattern-based bounding in elasticity.
J. Mech. Phys. Solids 44, 307331.
Brongersma, M.L., 2003. Nanoshells: gifts in a gold wrapper. Nature Mater. 2, 296297.
Christensen, R.M., Lo, K.H., 1979. Solutions for effective shear properties in three phase sphere and cylinder
models. J. Mech. Phys. Solids 27, 315330.
Dabbousi, B.O., Rodriguez-Viejo, J., Mikulec, F.V., Heine, J.R., Mattoussi, H., Ober, R., Jensen, K.F., Bawendi,
M.G., 1997. (CdSe)ZnS coreshell quantum dots: synthesis and characterization of a size series of highly
luminescent nanocrystallites. J. Phys. Chem. B 101, 94639475.
Danek, M., Jensen, K.F., Murray, C.B., Bawendi, M.G., 1996. Synthesis of luminescent thin-lm CdSe/ZnSe
quantum dot composites using CdSe quantum dots passivated with an overlayer of ZnSe. Chem. Mater. 8,
173180.
Ding, K., Weng, G.J., 1998. The inuence of moduli slope of a linearly graded matrix on the bulk moduli of some
particle- and ber-reinforced composites. J. Elasticity 53, 122.
Duan, H.L., Wang, J., Huang, Z.P., Luo, Z.Y., 2005. Stress concentration tensors of inhomogeneities with
interface effects. Mech. Mater. 37, 723736.
Eshelby, J.D., 1957. The determination of the elastic eld of an ellipsoidal inclusion and related problems. Proc.
R. Soc. Lond. A 241, 376396.
Eshelby, J.D., 1959. The elastic eld outside an ellipsoidal inclusion. Proc. R. Soc. Lond. A 252, 561569.
Freund, L.B., Johnson, H.T., 2001. Inuence of strain on functional characteristics of nanoelectronic devices.
J. Mech. Phys. Solids 49, 19251935.
Goncharenko, A.V., 2004. Optical properties of coreshell particle composites. I. Linear response. Chem. Phys.
Lett. 386, 2531.
Gosling, T.J., Willis, J.R., 1995. Mechanical stability and electronic properties of buried strained quantum wire
arrays. J. Appl. Phys. 77, 56015610.
Hashin, Z., Shtrikman, S.A., 1963. A variational approach to the theory of the elastic behavior of multiphase
materials. J. Mech. Phys. Solids 11, 127140.
He, L.X., Bester, G., Zunger, A., 2004. Strain-induced interfacial hole localization in self-assembled quantum
dots: compressive InAs/GaAs versus tensile InAs/InSb. Phys. Rev. B 70, 235316-1235316-9.
Herve, E., Zaoui, A., 1993. N-layered inclusion-based micromechanical modelling. Int. J. Eng. Sci. 31, 110.
Hill, R., 1963. Elastic properties of reinforced solids: some theoretical principles. J. Mech. Phys. Solids 11,
357372.
Huang, Y., Hu, K.X., Wei, X., Chandra, A., 1994. A generalized self-consistent mechanics method for compositematerials with multiphase inclusions. J. Mech. Phys. Solids 42, 491504.
Itskevich, I.E., Lyapin, S.G., Troyan, I.A., Klipstein, P.C., Eaves, L., Main, P.C., Henini, M., 1998. Energy levels
in self-assembled InAs/GaAs quantum dots above the pressure-induced Gamma-X crossover. Phys. Rev. B 58,
R4250R4253.
Kim, H., Achermann, M., Balet, L.P., Hollingsworth, J.A., Klimov, V.I., 2005. Synthesis and characterization of
Co/CdSe coreshell nanocomposites: bifunctional magnetic-optical nanocrystals. J. Am. Chem. Soc. 127,
544546.
Lauhon, L.J., Gudiksen, M.S., Wang, D., Lieber, C.M., 2002. Epitaxial coreshell and core-multishell nanowire
heterostructures. Nature 420, 5761.
Li, S., Sauer, R., Wang, G., 2005. A circular inclusion in a nite domain I. The Dirichlet-Eshelby problem. Acta
Mech. 179, 6790.
Little, R.B., El-Sayed, M.A., Bryant, G.W., Burke, S., 2001. Formation of quantum-dot quantum-well
heteronanostructures with large lattice mismatch: ZnS/CdS/ZnS. J. Chem. Phys. 114, 18131822.
Luo, H.A., Weng, G.J., 1987. On Eshelbys inclusion problem in a three-phase spherically concentric solid, and a
modication of MoriTanakas method. Mech. Mater. 6, 347361.

ARTICLE IN PRESS
H.L. Duan et al. / J. Mech. Phys. Solids 54 (2006) 14011425

1425

Lutz, M.P., Zimmerman, R.W., 1996. Effect of the interphase zone on the bulk modulus of a particulate
composite. J. Appl. Mech. 63, 855861.
Ma, H.L., Hu, G.K., Huang, Z.P., 2004. A micromechanical method for particulate composites with nite particle
concentration. Mech. Mater. 36, 359368.
Milton, G.W., 2002. The Theory of Composites. Cambridge University Press, Cambridge.
Mori, T., Tanaka, K., 1973. Average stress in matrix and average elastic energy of materials with mistting
inclusions. Acta Metal. 21, 571574.
Nemat-Nasser, S., Hori, M., 1999. Micromechanics: Overall Properties of Heterogeneous Materials, second ed.
Elsevier, Amsterdam.
Ostoja-Starzewski, M., Jasiuk, I., Wang, W., Alzebdeh, K., 1996. Composites with functionally graded interfaces:
meso-continuum concept and effective transverse conductivity. Acta Mater. 44, 20572066.
Peng, X.G., Schlamp, M.C., Kadavanich, A.V., Alivisatos, A.P., 1997. Epitaxial growth of highly luminescent
CdSe/CdS core/shell nanocrystals with photostability and electronic accessibility. J. Am. Chem. Soc. 119,
70197029.
Perez-Conde, J., Bhattacharjee, A.K., 2003. Electronic structure and optical properties of ZnS/CdS
nanoheterostructures. Phys. Rev. B 67, 235303-1235303-4.
Ponte Castaneda, P., Suquet, P., 1998. Nonlinear composites. Adv. Appl. Mech. 34, 171302.
Rockenberger, J., Troger, L., Rogach, A.L., Tischer, M., Grundmann, M., Eychmuller, A., Weller, H., 1998. The
contribution of particle core and surface to strain, disorder and vibrations in thiolcapped CdTe nanocrystals.
J. Chem. Phys. 108, 78077815.
Segurado, J., LLorca, J., 2002. A numerical approximation to the elastic properties of sphere-reinforced
composites. J. Mech. Phys. Solids 50, 21072121.
Talbot, D.R.S., Willis, J.R., 1985. Variational principles for inhomogeneous nonlinear media. IMA J. Appl.
Math. 35, 3954.
Theocaris, P.S., 1987. The Mesophase Concept in Composites. Springer, Berlin.
Torquato, S., 1998. Effective stiffness tensor of composite media: II. Applications to isotropic dispersions.
J. Mech. Phys. Solids 46, 14111440.
Torquato, S., 2002. Random Heterogeneous Materials: Microstructure and Macroscopic Properties. Springer,
New York.
Tzika, P.A., Boyce, M.C., Parks, D.M., 2000. Micromechanics of deformation in particle-toughened polyamides.
J. Mech. Phys. Solids 48, 18931929.
Walpole, L.J., 1981. Elastic behaviour of composite materials: theoretical foundations. Adv. Appl. Mech. 21,
169242.
Wang, W., Jasiuk, I., 1998. Effective elastic constants of particulate composites with inhomogeneous interphases.
J. Compos. Mater. 32, 13911424.
Wang, G., Li, S., Sauer, R., 2005. A circular inclusion in a nite domain II. The Neumann-Eshelby problem. Acta
Mech. 179, 91110.
Weng, G.J., 2003. Effective bulk moduli of two functionally graded composites. Acta Mech. 166, 5767.
Williamson, A.J., Zunger, A., 1999. InAs quantum dots: predicted electronic structure of free-standing versus
GaAs-embedded structures. Phys. Rev. B 59, 1581915824.
Willis, J.R., 2000. The overall response of nonlinear composite media. Eur. J. Mech. A/Solids 19, S165S184.
Wylie, C.R., Barrett, L.C., 1982. Advanced Engineering Mathematics, second ed. McGraw-Hill, New York.
Zheng, Q.S., Du, D.X., 2001. An explicit and universally applicable estimate for the effective properties of
multiphase composites which accounts for inclusion distribution. J. Mech. Phys. Solids 49, 27652788.
Zhou, H.S., Honma, I., Haus, J.W., Sasabe, H., Komiyama, H., 1996. Synthesis and optical properties of coated
nanoparticle composites. J. Luminescence 70, 2134.