Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

International Journal of Engineering Science 40 (2002) 1647–1662

www.elsevier.com/locate/ijengsci

Load transfer at imperfect interfaces––dislocation-like model


a,* b,1 b
H.Y. Yu , Y.N. Wei , F.P. Chiang
a
US Army Research Laboratory––Far East Research Office, Tokyo 106-0032, Japan
b
Department of Mechanical Engineering, State University of New York, Stony Brook, NY 11794, USA
Received 30 June 2000; accepted 9 May 2001

(Communicated by J.T. ODEN)

Abstract
A dislocation-like model describing the boundary conditions of a partially debonded interface is verified
experimentally. The effect of the imperfect interface on the load transfer is studied by photoelastically
measuring the elastic deformation in bimaterials due to an inclusion with dilatational misfit strain. The
boundary conditions to be modeled are that the radial and the tangential tractions are continuous across
the interface and the displacements may be discontinuous from one solid to another. The discontinuity of
displacement across the interface is assumed linearly proportional to the displacement at the interface of the
constituent where the stress source is. The maximum shear stress distributions measured from the iso-
chromatic fringe patterns are in good agreement with the theoretical calculations. The results show that
an imperfect interface could be viewed as a continuum entity with interface rigidity as proposed by the
dislocation-like model.
Ó 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Imperfect interfaces; Debonded interfaces; Dislocation-like model; Inclusions; Interface rigidity; Photo-
elasticity

1. Introduction

Scientific and engineering studies of the physical and mechanical properties of macroscopic
systems often require a knowledge of the interface––the boundary region––separating bulk

*
Corresponding author. Address: Army Research Office––Far East, 7-23-17 Roppongi, Minato-Ku, Tokyo 106-
0032, Japan. Tel.: +81-3-3423-1374; fax: +81-3-3423-1832.
E-mail address: yuh@arofe.army.mil (H.Y. Yu).
1
Present address: Shanghai Jiao Tong University, Shangai 200030, People’s Republic of China.

0020-7225/02/$ - see front matter Ó 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 2 0 - 7 2 2 5 ( 0 2 ) 0 0 0 2 8 - 9
1648 H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662

phases. On the macroscopic scale, interfaces are viewed as continuum entities with macroscopi-
cally defined parameters such as free energy per unit area, mobility, etc. From the point of view of
classical elasticity, an interface manifests itself through interfacial boundary conditions imposed
upon equations of elastic equilibrium. Well defined interfaces are (1) a welded interface when two
solids are perfectly bonded together (both the tractions and displacements are continuous at the
interface); (2) a sliding interface when two solids are in frictionless contact with each other (al-
lowing free tangential slip at the interface); and (3) a free surface when two solids are completely
separated from each other. A sliding interface is an ideal interface that is physically impossible.
While the boundary conditions for these interfaces are well defined, the appropriate boundary
conditions for most real interfaces lie somewhat between a welded interface and a free surface.
Some models have been introduced to describe the boundary conditions of the real or imperfect
interfaces. One is the three-phase model [1–4] that introduces an interphase or a mesophase with
elastic constants different from those on either side of it to represent an imperfect interface. This
layer has a given thickness. However, both the matrix-layer and layer-reinforcement interfaces are
assumed to be welded interfaces.
Another model is the linear spring-like model [5–15] that a thin layer of interphase material is
introduced near the interface. In the limit of vanishing layer-thickness, the interfacial tractions
become continuous, but the displacements at either side of the interface layer become discon-
tinuous, the jump in displacement being linearly proportional to the interfacial traction. The
proportional constants are interface parameters called spring-like constant. This is the same as
that the tractions exerted on the interface are assumed to vary linearly with the displacement jump
[5,15]. Although these boundary conditions have been used extensively to model imperfect in-
terface properties, it might lead to physically impossible phenomenon of interpenetration at the
interfaces as discussed briefly by Achenbach and Zhu [7].
This study is motivated by the need for a better understanding of the role of interfaces on the
elastic deformation. What effect it has on the load transfer? Here the isochromatic fringe patterns
and the maximum shear stress distributions due to an inclusion in a bimaterial with an imperfect
interface is studied by using photoelasticity. A dislocation-like model [16,17] is used to describe
the boundary conditions of imperfect interfaces. The boundary conditions are similar to the linear
spring-like model except that the jump in displacement at the interface is assumed linearly pro-
portional to the displacement at the interface of the constituent where the stress source is. Linear
elasticity is used to formulate the analytical solutions.

2. Experimental procedure

The bimaterial specimen used to investigate the effect of imperfect interfaces on the load
transfer from one side of the interface to the other side is shown in Fig. 1. It consists of two
photoelastic epoxy resin plates; PSM-5A and PSM-1 (obtained from Measurement Group, Inc.).
To ensure a welded interface is formed on the bonding area, the plates are machined to size,
mechanically abraded by sanding and cleaned. Then, a thin layer of PC-1C resin glue (obtained
from Measurement Group, Inc.) is applied to the surface and the two specimen halves are held
together until the resin cures. The properties of these materials are list in Table 1. After room
temperature, curing for 12 h the PC-1C resin achieves the same elastic modulus as PSM-5A and
H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662 1649

Fig. 1. Schematic of disc inclusion in a bimaterial specimen.

Table 1
Material properties of bimaterial system
Properties PSM-5A PSM-1 PC-1C
Young modulus (ksi) 450 360 450
Poisson’s ratio 0.36 0.38 0.36
Material fringe value, fa (psi/in) 60 40 60

forms a welded interface. The procedure used in this study is similar to those used by Singh and
Shukla [18] to form a welded interface between photoelastic polyester resin (Homalite-100) and
aluminum. To mimic an imperfect interface, the plates are partial bonded by a procedure such
that part of the surfaces of the two solids is unbonded and the rest is a welded interface. A thin
paper is used to cover a portion of the surface of PSM-1 plate as shown in Fig. 2. Then, a thin
layer of PC-1C resin glue is applied to the surface and the two specimen halves are held together
until the resin cures. The fraction of the bonded surface area is used to quantify the extent of
bonding, c, of the interface. Let A be the total surface of the interface and Ac be the surface area
covered by paper, then

A  Ac
c¼ : ð1Þ
A

When there is no cover paper, i.e., Ac ¼ 0 and c ¼ 1, the whole interface is a welded interface and
the two plates are perfectly bonded together. If there is no plate 2 or the surface is totally covered
with paper, two plates are separated and c ¼ 0. When Ac ¼ 0:75A, then the area fraction of surface
area that are perfectly bonded together is 25% and c ¼ 0:25.
1650 H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662

Fig. 2. Schematic of bimaterial interface partially cover with paper to mimic imperfect interfaces.

As indicated in Fig. 1, a round hole with radius a is drilled in the PSM-5A plate of the bi-
material specimen. A PSM-5A disc with radius a slightly larger than a is dipped into liquid ni-
trogen such that its radius shrinks to a value slightly less than a. The disc is than inserted into the
hole at room temperature. Now the bimaterial is a stressed solid due to a disk inclusion with misfit
stain (eigenstrain)

a  a
eTij ¼ edij ; where e ¼ ; ð2Þ
a

and dij is the Kronecker delta. Then the elastic deformation in the bimaterial can be analyzed as
an inclusion problem with pure dilatational eigenstrain.
The Isochromatic fringe patterns in the specimen due to the misfit inclusion is observed using
polariscope and photographed. This is made possible by the transparent and photoelastic nature
of the specimen. Isochromatic fringe patterns are contours of constant average maximum shear
stress across the thickness of the specimen. From the stress optic law, the relationship is

Nfa
smax ¼ ; ð3Þ
2h

where N is the fringe order, h is the thickness of the specimen and fa is the material fringe value. It
should be noted that the specimen has a thickness of 7 mm and the fringe patterns are taken in a
transmission poloriscope. Thus the stress field, as it is calculated from photoelastic fringes, is the
stress field averaged over the entire specimen thickness.

3. Modeling

As shown in Fig. 3, the bimaterial consists of semi-infinite solid 1 (z P 0) with shear modulus l,
and Poison’s ratio m, and semi-infinite solid 2 (z 6 0) with shear modulus l0 and Poison’s ratio m0 .
In the following, symbols without prime are for solid 1 and with prime are for solid 2. For ex-
ample, ui and u0i are the elastic displacements in solid 1 and solid 2, respectively. The interfacial
defects such as dislocation, voids and microcracks might occur during the specimen preparations
or nucleated by external loading during the life of service. To simplify the argument, let us
consider two solids bonded together by diffusion bonding that is a commercial welding process
and also used extensively in metal matrix composites. Diffusion bonding is a solid-state welding
H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662 1651

Fig. 3. Schematic of bimaterial interface S with void and the stress source is a disc inclusion X with radius a and center
ð0; dÞ in solid 1.

process by which two prepared surfaces are joined at elevated temperature and under applied
pressure. The initial contact of the asperities on the prepared surfaces creates a series of voids,
which subsequently may act as a weakness unless they are removed entirely to form a welded
interface. Studies have been done on the mechanisms of diffusion bonding that are closure of
interface voids by a range of diffusion pathway or by a creep deformation [19–21]. Since the voids
are difficult to remove completely, an imperfect interface S can be modeled as a surface with voids
as schematically shown in Fig. 3. Now if a stress source is in solid 1, the displacement ui in solid 1
at the S is not transmitted fully to the adjacent point in solid 2 because the two solids are not
completely contacted with each other. This means that the displacement u0i in solid 2 at S in not
the same as ui as in the case of welded interface. These displacements should be thought of as
averaged displacement [7]. Since elastic deformation in solid 2 is caused by the deformation in
solid 1 at the interface, the relationship could be written as

u0i ¼ aij uj þ aijk uj;k þ aijk‘ uj;k‘ þ    ; ð4Þ

where a repeated subscript indicates summation over the values 1, 2, 3, and subscripts preceded by
a comma denote differentiation with respect to the Cartesian coordinates corresponding to those
subscripts. The tensor aij , aijk . . . and aijk...n are to be defined. If the assumption is made such that
aij ¼ dij ; aijk‘ and all other higher order tenser coefficients vanished, then at the interface, one has
the boundary condition

u0i  ui ¼ aijk uj;k : ð5Þ


1652 H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662

This is the linear spring-like model when aijk are constants. The displacement discontinuity in the
linear spring-like model [13] can be expressed in the forms of

u0i  ui ¼ jip rpq nq ; ð6Þ

where jip is the interface compliance and nq is the unit outward normal vector of the interface. By
using Hooke’s law and (5) and (6), one has the relationship between the third order tensor aijk in
(5) and spring-like compliance jip in (6) as

aijk ¼ jip Cpqjk nq ; ð7Þ

where Cpqik is the elastic constant of solid 1. Certain restrictions on the tensor are needed to insure
there is no unrealistic overlap of the two materials in the interfacial zone [7].
On the other hand, if one assumed the only nonzero tensor is the second order tensor aij , then
(5) becomes

u0i ¼ aij uj ; ð8Þ

which gives the displacement jump at the interface of the dislocation-like model and aij are
dimensionless parameters that represent the behavior of imperfect interfaces. If aij is constant,
(8) means that the variable discontinuity of displacement across the interface is assumed to
vary linearly with the displacement due to the stress source. For simplicity, it is further assumed
that

aij ¼ gi dij ; g1 ¼ g2 ¼ gT ; g3 ¼ gN ; ð9Þ

where gT and gN are two dimensionless parameters that describe the bonding condition in the
tangential and the normal directions of the interface, respectively. By using (8) and (9), the
boundary conditions at x3 ¼ 0 for the dislocation-like model are

r03i ¼ r3i ; u0k ¼ gT uk ; u03 ¼ gN u3 ; ð10Þ

where i ¼ 1, 2, 3 and k ¼ 1, 2. When gT ¼ gN ¼ 1, the interface is a welded interface. The two


solids are completely separated and S is a free surface when gT ¼ gN ¼ 0. Any values of either
parameter between zero and one defines an imperfect interface. Because the stress source is lo-
cated in solid 1, gT and gN are always less than or equal to one. Therefore there is no unrealistic
overlap of the two materials in the interfacial zone.
It has been shown [16,17] that for an inclusion with pure dilatation eigenstrain given by (2), the
three-dimensional elastic displacement in a bimaterials with boundary conditions described by
(10) are

ðj  7Þe
ui ¼ fU1;i þ ½c1 þ ð1  jdi3 Þc2 U2;i þ c2 x3 U2;3i g; ð11Þ
4pð1 þ jÞ
H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662 1653

and
ðj  7Þle
u0i ¼ f½c0 þ ð1  j0 di3 Þc02 U1;i þ c02 x3 U1;3i g; ð12Þ
4pð1 þ jÞl0 1
for x3 P 0 and x3 6 0, respectively, where i ¼ 1, 2, 3, subscripts preceded by a comma denote
differentiation with respect to the Cartesian coordinates corresponding to those subscripts,

ðj  2Þ½ðl  lT ÞðlN þ j0 lÞ þ ðl  lN ÞðlT þ j0 lÞ þ ðj þ 1ÞðlN  lT Þl


c1 ¼ ;
ðl þ jlT ÞðlN þ j0 lÞ þ ðl þ jlN ÞðlT þ j0 lÞ
2½ðl  lT ÞðlN þ j0 lÞ þ ðl  lN ÞðlT þ j0 lÞ
c2 ¼ ;
ðl þ jlT ÞðlN þ j0 lÞ þ ðl þ jlN ÞðlT þ j0 lÞ
2ðj þ 1Þ½lN ðlN þ j0 lÞ þ ðlN  lT Þl ð13Þ
c01 ¼ ;
ðl þ jlT ÞðlN þ j0 lÞ þ ðl þ jlN ÞðlT þ j0 lÞ
2ðj þ 1ÞðlT  lN Þl
c02 ¼ ;
ðl þ jlT ÞðlN þ j0 lÞ þ ðl þ jlN ÞðlT þ j0 lÞ
lT ¼ gT l0 ; lN ¼ gN l0 ; j ¼ 3  4m; j0 ¼ 3  4m0 ;

and
Z
dx01 dx02 dx03
Uk ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; ðk ¼ 1; 2Þ: ð14Þ
X ðx1  x01 Þ2 þ ðx2  x02 Þ2 þ ½x3 þ ð1Þk x03 2

Uk is the Newtonian potential due to the inclusion X with unit mass density when k ¼ 1 and it is
the Newtonian potential due to the mirror image of the inclusion when k ¼ 2. The parameters lT
and lN defined in (13) are the interfacial rigidities. The stresses are

ðj  3Þl
rij ¼  um;m dij þ lðui;j þ uj;i Þ; ð15Þ
j1
for solid 1 and

ðj0  3Þl0 0
r0ij ¼  u dij þ l0 ðu0i;j þ u0j;i Þ; ð16Þ
j0  1 m;m
for solid 2. It should be noted that there are some misprints in the results given by Yu [17]. The
correct results are given by (11)–(16).
The stress inside the inclusion is

rXij ¼ rij  rTij ; ð17Þ

where rTij is the uniform stress derived from the eigenstrain eTij by using Hooke’s law, thus

j7
rTij ¼  ledij : ð18Þ
j1
1654 H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662

Since the specimen in the present study is in the shape of plate with finite thickness, the de-
formation due to the inclusion is neither the plane strain deformation nor the plane stress de-
formation but somewhat in between. Therefore, one can use plane strain and plane stress results
as upper and lower bonds to model the deformation. For plane strain and plane stress defor-
mations in the x2 –x3 plane, the solution is obtained by integrating (14) with respect to x01 from 1
to 1, thus u1 ¼ u01 ¼ 0, and u2 , u3 , u02 and u03 are given by (11) and (12) where Uk is the Newtonian
potential due to a cylinder with radius a and its mirror image. The stress field in solid 1 is
   
ðj  7Þle 5j
r22 ¼  U1;33 þ c1 þ c2 U2;33 þ c2 x3 U2;333 þ b;
2pð1 þ jÞ 2
   
ðj  7Þle j1 3j
r33 ¼ U1;33 þ c1  c2 U2;33 þ c2 x3 U2;333 þ b; ð19Þ
2pð1 þ jÞ 2 1þj
   
ðj  7Þle 3j
r13 ¼ r12 ¼ 0; r23 ¼ U1;32 þ c1 þ c2 U2;32 þ c2 x3 U2;332 ;
2pð1 þ jÞ 2

where

b ¼ 0; ð20Þ

for point outside the inclusion and1

j7
b¼ le; ð21Þ
1j

for point inside the inclusion. The stress field in solid 2 is


  
ðj  7Þle 5  j0 0
r022¼ 0
c1 þ 0
c2 U1;33 þ c2 x3 U1;333 ;
2pð1 þ jÞ 2
  
0 ðj  7Þle 0 j0  1 0 0
r33 ¼ c1  c2 U1;33 þ c2 x3 U1;333 ; ð22Þ
2pð1 þ jÞ 2
  
0 0 0 ðj  7Þle 0 3  j0 0 0
r13 ¼ r12 ¼ 0; r23 ¼ c1 þ c2 U1;32 þ c2 x3 U1;332 :
2pð1 þ jÞ 2

For plane stress deformation, the j and j0 in (19), (21) and (22) are

3m 3  m0
j¼ ; j0 ¼ ; ð23Þ
1þm 1 þ m0
and

r11 ¼ r011 ¼ 0: ð24Þ


H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662 1655

For plane strain deformation, the j and j0 in (19), (21) and (22) are

j ¼ 3  4m; j0 ¼ 3  4m0 ; ð25Þ

and

r11 ¼ tðr22 þ r33 Þ; r011 ¼ t0 ðr022 þ r033 Þ: ð26Þ

Eq. (26) is due to the requirement that the longitudinal tensile strain e11 ¼ e011 ¼ 0. The max-
imum shear stress at any point of the x2 –x3 plane in a condition of plane stress or plane strain is
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 r  r 2  0 2
22 33 0 r22  r033 2
smax ¼ 2
þ r23 ; smax ¼ þ r023 ; ð27Þ
2 2

for plate 1 and plate 2, respectively. The functions Uk are obtained from the ellipsoidal inclusion
solution by letting their longest axes approach infinite [22], thus

U1 ¼ pa2 log½x22 þ ðx3  dÞ2 ; ð28Þ

for point outside X, and

U1 ¼ p½x22 þ ðx3  dÞ2 ; ð29Þ

for point inside X. The potential due to the mirror image of X is

U2 ¼ pa2 log½x22 þ ðx3 þ dÞ2 : ð30Þ

The result of the dislocation-like model is equivalent to the introduction of two effective in-
terfacial moduli lT and lN . This is consistent with the conclusion ‘‘A range of bond strength
between the two crystals is likewise cared for in terms of an interfacial rigidity modulus l.’’ that
Van der Merwe made [23] by studying the strain energy of an interface. In his studies [24,25], the
imperfect interface is assumed to be (1) a boundary caused by a difference of atomic spacing, (2) a
twist boundary, or (3) a symmetrical tilt boundary. His calculation is based on the assumptions
introduced by Peierls [26] and Nabarro [27] in dealing with a single dislocation.

4. Results and discussion

Typical isochromatic fringes of bimaterials observed are shown in Figs. 4a, 5a and 6a for c ¼ 1,
0.5 and 0, respectively. By using (27), the maximum shear stress contours are calculated in units of
le=2p. Figs. 4b, 5b and 6b are the plane stress deformation contours for gT ¼ gN ¼ 1, 0.5 and 0,
respectively. Figs. 4c, 5c and 6c are the plane strain deformation contours for gT ¼ gN ¼ 1, 0.5
and 0, respectively. Closed similarity between the pattern of the isochromatic fringe and calcu-
lated contours are observed. In the calculation, the two parameters gT and gN are assumed the
same and are equal to c. This is because the photoelastic polyester resin used to bond the two
1656 H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662

Fig. 4. (a) Dark field isochromatic fringe for welded interface (c ¼ 1), (b) calculated smax contours for plane stress
deformation where gT ¼ gN ¼ 1. The stress is in units of le=2p, (c) calculated smax contours for plane strain deformation
where gT ¼ gN ¼ 1. The stress is in units of le=2p.

plates together is an isotropic material. It should be noted that in the photoelastic study the
material fringe values of the two materials are not the same.
The specimen for the photoelastic measurements is made of five different c values 1, 0.75, 0.5,
0.25 and 0 where c is defined in (1). Figs. 7 and 8 show the maximum shear stress distribution
along the interface of plate 1 for the conditions two plates completely separated (the interface is
a free surface, c ¼ gT ¼ gN ¼ 0) and welded interface (c ¼ gT ¼ gN ¼ 1), respectively. When c
H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662 1657

Fig. 5. (a) Dark field isochromatic fringe for imperfect interface with c ¼ 0:5, (b) calculated smax contours for plane
stress deformation where gT ¼ gN ¼ 0:5. The stress is in units of le=2p, (c) calculated smax contours for plane strain
deformation where gT ¼ gN ¼ 0:5. The stress is in units of le=2p.

equals zero and one, the results are for two well-defined interfaces. The solutions for these two
interfaces are obtained without using any assumption. Their results could be used as calibration.
Since the absolute stress values were not available, comparison was made equating the measured
maximum shear stress smax at point x2 ¼ a at the interface to the average value of the plane strain
and plane stress deformations at the same point. This point was chosen mainly because it is a good
fit for plate 1 with a free surface (Fig. 7). By using x2 ¼ a as a calibrating point, it also leads to a
good fit for another well defined interface (a welded interface) as shown in Fig. 8.
1658 H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662

Fig. 6. (a) Dark field isochromatic fringe for debonded interface (c ¼ 0), (b) calculated smax contours for plane stress
deformation where gT ¼ gN ¼ 0. The stress is in units of le=2p, (c) calculated smax contours for plane strain deformation
where gT ¼ gN ¼ 0. The stress is in units of le=2p.

The comparisons for imperfect interfaces are shown in Figs. 9–11 for the conditions of
25% bond (c ¼ gT ¼ gN ¼ 0:25), 50% bond (c ¼ gT ¼ gN ¼ 0:5) and 75% bond (c ¼ gT ¼
gN ¼ 0:75), respectively. The experimental results agree well the theoretical calculations ob-
tained by using dislocation-like model. Therefore, on the macroscopic scale, imperfect (par-
tially debonded) interfaces could be viewed as continuum entities with macroscopically defined
interfacial parameters gi to describe the effect on the load transfer from one material to an-
other.
H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662 1659

Fig. 7. Maximum shear stress in plate 1 along the interface for c ¼ gT ¼ gN ¼ 0.

Fig. 8. Maximum shear stress in plate 1 along the interface for c ¼ gT ¼ gN ¼ 1.

5. Summary

Photoelasticity is used to investigate the elastic deformation due to an inclusion in a bimaterial


with imperfect interface. Two photoelastic epoxy resin plates bonded together to form a bima-
terial. The imperfect interface is characterized by the fraction of surface area that bonded
1660 H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662

Fig. 9. Maximum shear stress in plate 1 along the interface for c ¼ gT ¼ gN ¼ 0:25.

Fig. 10. Maximum shear stress in plate 1 along the interface for c ¼ gT ¼ gN ¼ 0:5.

together. The maximum shear stress is measured from the dark field isochromatic fringe. The
elastic deformation due to the inclusion with pure dilatational eigenstrain is calculated by using
a dislocation-like model to describe the boundary conditions of the imperfect interface. The
boundary condition is that the interfacial tractions are continuous but the interfacial displace-
ments are discontinuous. The displacements at one side of the interface are linearly proportional
to the displacements at the other side where the elastic singularity is located. The proportionality
H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662 1661

Fig. 11. Maximum shear stress in plate 1 along the interface for c ¼ gT ¼ gN ¼ 0:75.

constants are gT for the tangential interfacial displacement and gN for the radial interfacial
displacement. The theoretical calculation agrees well with the experimental measurements. This
indicates that the dislocation-like model could be used to investigate imperfect interface prop-
erties.

Acknowledgements

The support of the Air Force Office of Scientific Research for this investigation is gratefully
acknowledged. One of the authors (H.Y.Y.) is grateful for the support from the Office of Naval
Research through the Naval Research Laboratory.

References

[1] L.J. Broutman, B.D. Agarwal, A theoretical study of the effect of an interfacial layer on the properties of
composites, Poly. Sci. Eng. 14 (1974) 581–588.
[2] P.S. Theocaris, T.P. Philippidis, Theoretical evaluation of the extent of the mesophase in particulate and fibrous
composites, J. Reinforced Plastic Compos. 4 (1985) 173–185.
[3] F.H.J. Maurer, R. Simha, R.K. Jain, On the elastic moduli of particulate composites interlayer versus molecular
model, in: Proceedings of 1st International Conference on Composite Interface, Cleveland, Ohio, 1986, pp. 367–
375.
[4] N.J. Pagano, G.P. Tandon, Modeling of imperfect bonding in fiber reinforced brittle matrix composites, Mech.
Mater. 9 (1990) 49–64.
[5] F. Lene, D. Leguillon, Homogenized constitutive law for a partially cohesive composite materials, Int. J. Solid.
Struct. 18 (1982) 443–458.
[6] Y. Benveniste, J. Aboudi, A continuum model for fiber reinforced materials with debonding, Int. J. Solid. Struct. 20
(1984) 935–951.
[7] J.D. Achenbach, H. Zhu, Effect of interface zone on mechanical behavior and failure of fiber-reinforced
composites, J. Mech. Phys. Solid. 37 (1989) 381–393.
1662 H.Y. Yu et al. / International Journal of Engineering Science 40 (2002) 1647–1662

[8] D.A. Kouris, T. Mura, The elastic field of a hemispherical inhomogeneity at the free surface of an elastic half space,
J. Mech. Phys. Solid. 37 (1989) 365–379.
[9] Z. Hashin, Thermoelastic properties of fiber composites with imperfect interface, Mech. Mater. 8 (1990) 333–348.
[10] Z. Hashin, The spherical inclusion with imperfect interface, J. Appl. Mech. 58 (1991) 444–449.
[11] A. Levy, The debonding of elastic inclusions and inhomogeneities, J. Mech. Phys. Solids 39 (1991) 477–505.
[12] I. Jasiuk, J. Chen, M.F. Thorpe, Elastic moduli of composites with sliding inclusions, J. Mech. Phys. Solid. 40
(1992) 373–391.
[13] T. Qu, The effect of slightly weaken interfaces on the overall elastic properties of composite materials, Mech.
Mater. 14 (1993) 269–281.
[14] Z.Q. Cheng, A.K. Jemah, F.W. Williams, Theory for multilayered anisotropic plates with weakened interface, J.
Appl. Mech. 63 (1996) 1019–1026.
[15] L.E. Shilkrot, D.J. Srolovitz, Elastic analysis of finite stiffness bimaterial interfaces: application to dislocation-
interface interactions, Acta Mater. 46 (1998) 3063–3075.
[16] H.Y. Yu, S.C. Sanday, Centers of dilatation and inclusions with dilatational eigenstrain in bimaterials with
imperfect interfaces, Composites ’95: Recent Advances in Japan and United States, in: B. Ara (Ed.), Proc. Japan–
US CCM-VII, Kyoto, Japan, 1995, pp. 191–200.
[17] H.Y. Yu, A new dislocation-like model for imperfect interfaces and their effect on load transfer, Composites 29A
(1998) 1057–1062.
[18] R.P. Singh, A. Shukla, Subsonic and intersonic crack growth along a bimaterial interface, J. Appl. Mech. 63 (1996)
919–924.
[19] W.H. King, W.A. Owczarski, Diffusion welding of commercially pure titanium, Weld J. Res. Suppl. 46 (1967) 289–
298.
[20] A. Hill, E.R. Wallach, Modeling solid-state diffusion bonding, Acta Metall. 37 (1989) 2425–2437.
[21] B. Derby, C.D. Qin, Metal–ceramic interfaces: sources and sinks for mass transfer, Acta Metall. Mater. Suppl. 40
(1992) 533–558.
[22] W.D. MacMillan, in: The Theory of the Potential, fourth ed., McGraw-Hill, New York, 1930, p. 72.
[23] J.H. Van der Merwe, On the structure of epitaxial bicrystals, in: J.J. Burke, N.L. Reed, V. Weiss (Eds.), Surface
and Interfaces I: Chemical and Physical Characteristics, Syracuse University Press, Syracuse, 1967, pp. 361–381.
[24] J.H. Van der Merwe, On the structure of epitaxial bicrystals, Proc. Roy. Soc. Lond. A 63 (1950) 616–637.
[25] J.H. Van der Merwe, Crystal interfaces. Part I. Semi-infinite crystals, J. Appl. Phys. 34 (1963) 117–122.
[26] R.E. Peierls, The size of a dislocation, Proc. Phys. Soc. 52 (1940) 34–37.
[27] F.R.N. Nabarro, Dislocations in a simple cubic lattice, Proc. Phys. Soc. 59 (1947) 256–272.

You might also like