Engineering Fracture Mechanics 142 (2015) 170–183

Contents lists available at ScienceDirect

Engineering Fracture Mechanics

Prediction of interfacial strength and failure mechanisms

a r t i c l e i n f o a b s t r a c t

Article history: The interfacial strength and failure mechanisms of particle-reinforced metal-matrix com-
Received 31 January 2015 posites were predicted using a micromechanical model. The micromechanical model was
Received in revised form 28 May 2015 constructed based on the cohesive zone model, and the spherical particle was arranged
Accepted 1 June 2015
on the body-centered cubic distribution. The SiC particle-reinforced Al matrix composite
Available online 7 June 2015
was chosen as the model system and its interfacial properties was predicted. It can be seen
that a complete interfacial debonding all over the particle could never be reached, and the
Keywords:
plastic strain of composite when the interfacial debonding begins to appear shows an
Particle-reinforced composites
Interfacial debonding
increase trend with the increasing of interfacial strength.
Plastic deformation Ó 2015 Elsevier Ltd. All rights reserved.
Cohesive zone model
Micromechanical model

1. Introduction

Particle-reinforced metal-matrix composites, especially aluminum matrix composites, have been widely used in aero-
space and other industries as structural materials [1]. Ceramic particles, which are some of the most widely used materials
for reinforcing aluminum matrix, improves the stiffness and strength of the reinforced matrix while conserving a
quasi-isotropic nature and traditional manufacturing techniques for casting metal. In this way, the hardness can be improved
considerably with direct impact and wear behavior. A widely used particle for reinforcing aluminum matrix is silicon carbide
(SiC). Besides its density being slightly higher than the density of aluminum, it is among the widely used due to its low cost
and its wide range of available grades [2]. The SiC reinforcement promotes an increase in the elastic modulus and tensile
strength of the aluminum matrix composites, and another improvement of composites regarding the aluminum matrix is
the behavior at high temperature. In addition, the mechanical properties of particle-reinforced composites are determined
by the morphology [3,4], size [5], volume fraction [6] and distribution [7] of particles, and the materials characteristics of the
matrix and particle. Numerical simulation based on the microstructure of composites has become increasing important tool
to understand the mechanical behavior of particle-reinforced composites with the development of the finite element model
and computation power [8]. Furthermore, the numerical simulation techniques are often more effective than analytical
modeling for such multiphase materials since the complex structures of these materials are not suitable for closed-form
theoretical analysis.

⇑ Corresponding author. Tel.: +86 451 82589364.

http://dx.doi.org/10.1016/j.engfracmech.2015.06.001
0013-7944/Ó 2015 Elsevier Ltd. All rights reserved.
 Q. Meng, Z. Wang / Engineering Fracture Mechanics 142 (2015) 170–183 171

Nomenclature

A area
D scalar damage variable
D stiffness tensor
E elastic modulus
K elastic properties for the interface element.
V volume fraction of particles
n element number
r particle radius
R length of representative volume element
T thickness of the interface element
t surface traction
u displacement vector
U displacement constant
v Poisson’s ratio
r normal traction
s shear traction
d separation
/ fracture energy

The interfacial debonding plays a key role in determining the ductility and toughness of a wide variety of
particle-reinforced metal-matrix composites. Once the interfacial debonding occurs, small voids may nucleate, and subse-
quently the voids grow by plastic deformation of the surrounding matrix material until they eventually coalesce [9,10].
The complete experimental study of the effect of the material properties on the interfacial debonding process in
particle-reinforced composites is unfeasible. A predictive modeling with interfacial debonding is becoming increasingly
important as a cost-effective method for design of particle-reinforced composites. Since the particles are not uniform in size,
and are dispersed in an irregular pattern throughout the matrix, a well established approach for predicting the macroscopic
mechanical behavior of these morphologically complex three-phase systems relies on the introduction of a spatially periodic
representative volume element. The interface or cohesive models are ideal tools to study the interfacial debonding progress
in the composites. The basic idea for such models can go back to Dugdale [11] and Barenblatt [12]. These models relate trac-
tion to the relative displacement at an interface where a crack may occur and damage initiation is related to an interfacial
strength. When the area under traction–separation curve is equal to the critical fracture energy, the traction is reduced to
zero and complete crack surfaces are formed. Furthermore, various cohesive zone models have been proposed to study
the growth of crack and the debonding of interface [13,14].
A few numerical micromechanical investigations have been carried out to interface properties and interfacial debonding
in the particle-reinforced composites. Xu and Needleman [15] presented a numerical micromechanical study of interfacial
debonding in the rigid particle-reinforced crystal matrix composites using the continuum-interface model. Based on the
same interface model, Charles et al. [16] further analyzed the interface debonding in particle-reinforced metal-matrix com-
posites. Segurado and LLorca [17] studied the interface properties in a model composite made of a random distribution of
stiff spherical particles embedded in a ductile matrix based on a new cohesive crack model. Tsui et al. [18] used a
three-dimensional unit cell model to predict the debonding damage process of particle-reinforced polymer composites.
Kang et al. [19] discussed the effect of interfacial bonding state between particle and matrix on the ratchetting of SiC
particle-reinforced composites in uniaxial cyclic stressing at room temperature. Romanova et al. [20] performed a finite dif-
ference modeling of interface strength effects on the deformation of a metal-matrix composite. While the researchers have
made many efforts to study the interfacial debonding in the particle-reinforced composites, the interfacial strength and the
debonding process have not been predicted very well. Therefore, it is required to develop an effective prediction model of the
interfacial debonding in the particle-reinforced composites.
The objective of the present work is to predict the interfacial properties and the interfacial failure mechanisms of
particle-reinforced metal-matrix composites using a micromechanical model. The micromechanical model is constructed
based on the cohesive zone model (CZM) and the spherical particle is arranged on the body-centered cubic distribution.
The SiC particle-reinforced Al matrix composite is chosen as the model system and its interfacial properties is predicted.
In addition, the effect of interfacial properties on the interfacial debonding of composite is also investigated.

2. Numerical modeling

2.1. Micromechanical model

In the real particle-reinforced composites, the particle distribution is typically random in the filled matrix, although par-
ticles are expected to be distributed uniformly. For the micromechanical modeling of particle-reinforced composites, one can
172 Q. Meng, Z. Wang / Engineering Fracture Mechanics 142 (2015) 170–183

choose a sufficiently large representative volume element (RVE) to describe the microstructural features of composites as
many as possible [8,21] and then to numerically obtain the overall mechanical behavior of composites, which has a relatively
complex establishment of models and needs a great number of calculations. An alternative approach is using the periodic
representative volume element in which the behavior of composites is approximated by an idealized, staggered array of par-
ticles, and the staggered nature of the particle arrangement captures the major effects of particle–particle interactions
[22,23]. The micromechanical models obtained by the periodic representative volume element are advantageous to the gen-
eration of structured mesh and the extraction of results or data. The cohesive zone model can easily be incorporated in a
finite element code by implementing so-called interface element that must be a regular geometry element, and the shape
of the reinforced particle is often spherical in practical applications. Therefore, the spherical particles are assumed to be
packed in the matrix. The periodic distribution of particles for the modeling of composites mainly contains the single cubic
(SC), face-centered cubic (FCC) and body-centered cubic (BCC) distribution, etc. [19]. Furthermore, in order to contain a
whole particle in the model and consider the effects of particle–particle interactions, the random particle distribution is ide-
alized by arranging the particle on the body-centered cubic (BCC) lattice in this study, as shown in Fig. 1a. The representative
volume element (RVE) for the BCC particle distribution shown in Fig. 1b contains two particles, and the volume fraction of
particles can be given by

8  r 3
V¼ p ð1Þ
3 R
where r is the particle radius and R is the length of RVE.
The finite element models are generated and analyzed in ABAQUS/Standard. The finite element mesh for the RVE with
particle volume fraction 20% is shown in Fig. 2. The matrix and particles are meshed with an 8-node linear brick, reduced
integration element (C3D8R) with hourglass control and an 8-node three-dimensional cohesive element (COH3D8) is
employed between particles and matrix to simulate the interfacial debonding. For convenience of calculation, a volume
100 lm  100 lm  100 lm would yield a representative group of particles for modeling. The thickness of interface element
is 0.1 lm, which is the same for the different models. The number of elements and nodes in the RVE are approximately
164,800 and 175,000, respectively. The mesh size is about 1 lm, and the same mesh is used for the model systems of dif-
ferent particle volume fractions.
The periodic boundary condition is applied to the representative volume element [22,24] and it can be expressed by
imposing the following equations on the corresponding surfaces of the RVE

ux ¼ 0 t y ¼ t z ¼ 0 on x ¼ 0
uy ¼ 0 t x ¼ t z ¼ 0 on y ¼ 0 ð2Þ
uz ¼ 0 t x ¼ t y ¼ 0 on z ¼ 0

ux ¼ U x ty ¼ tz ¼ 0 on x ¼ R
uy ¼ U y tx ¼ tz ¼ 0 on y ¼ R ð3Þ
uz ¼ U z tx ¼ t y ¼ 0 on z ¼ R

where R is the length of RVE, ux, uy, uz are components of the displacement vector u along x-, y- and z-directions, tx, ty, tz are
components of the surface traction t along x-, y- and z-directions, Ux, Uy and Uz are the displacement constant corresponding
to the x-, y- and z-directions, respectively. The periodic of boundary conditions means that the surfaces x = 0, y = 0 and z = 0
are symmetric planes, and the surfaces x = a, y = a and z = a are maintained straight and move parallel with respect to their

(a) (b)
Fig. 1. Body-centered cubic particle distribution (a) and representative volume element (b) in particle-reinforced composites.
 Q. Meng, Z. Wang / Engineering Fracture Mechanics 142 (2015) 170–183 173

Matrix

Interface
Element

Particle
z
y
x

Fig. 2. Finite element mesh for representative volume element with the volume fraction 20%.

original shapes under loading. We mainly study the uniaxial tensile behavior of particle-reinforced composites in this paper.
The tensile load is subjected to the RVE in the z-direction while there are no tractions applied in the x- and y-directions. The
boundary conditions can be expressed as
R
rx dA ¼ 0 on x ¼ R
RA
r dA ¼ 0 on y ¼ R
A y
ð4Þ
U z ¼ u on z ¼ R
where A is the current surface area of the RVE and u is the applied displacement load. The average applied stress to the com-
posites in the z-direction is calculated as
R Pn
A
rz dA i¼1 rz Ai
r z ¼ ¼ ð5Þ
A A
where rz is the stress in an element with area Ai, Ai is the area of the element on the face with area A under consideration and
n is the total number of such elements. The Young’s modulus is calculated from the average stress
r z
E¼ ð6Þ
ez
where the applied strain is calculated as the ratio of the prescribed displacement to the length of the RVE.

2.2. Cohesive zone model

The zero-thickness interface element with 8-node is shown in Fig. 3a. The constitutive equation of zero-thickness inter-
face elements is established in terms of relative displacements and tractions across the interface. The traction stress vector r
consists of three components rn, ss, st which represent the normal and two shear tractions, respectively. The local coordinate
vector n is the through-the-thickness direction and corresponds to mode I failure while the vectors denoted by s and t span
the midplane of the interface and correspond to mode II and III failure. The corresponding separations are denoted by dn, ds
and dt. The strain components can be defined as
dn ds dt
en ¼ ; es ¼ ; et ¼ ð7Þ
T T T
where T is the constitutive thickness of the interface element. Then the elastic constitutive relations can be written as
8 9 2 38 9
< rn >
> = K nn 0 0 < en >
> =
r¼ ss ¼ 6
4 0 K ss
7
0 5 es ¼ K e ð8Þ
>
: > >
: >
st ; 0 0 K tt et ;
174 Q. Meng, Z. Wang / Engineering Fracture Mechanics 142 (2015) 170–183

8 7
3
4 n
t

s
5 6

1 2
(a)
) Traction
t
0
( s0 ,
0
n

(1 D) K

0 max f Separation
m m m

(b)
Fig. 3. An 8-node three-dimensional cohesive element (a) and its traction–separation law with linear softening (b).

where uncoupled behavior between the normal and shear components is assumed so that the off-diagonal terms in the elas-
ticity matrix to zero. The terms Knn, Kss and Ktt are the elastic properties of the interface element in the normal and two shear
directions, respectively.
Failure mechanism for the interface element involves three aspects: the damage initiation, the damage evolution and the
element deletion reaching a completely damage state. Damage initiation refers to the beginning of degradation of the
response of a material point. The process of degradation begins when the stresses or strains satisfy certain damage initiation
criteria. Here, the quadratic stress failure criterion [25] is considered for the initiation of damage. Damage is assumed to ini-
tiate when a quadratic interaction function involving the nominal stress ratios reaches a value of one. This criterion can be
represented as
 2  2  2
rn ss st
þ þ ¼1 ð9Þ
r0n s0s s0t
where r0n ; s0s and s0t represent the peak values of the nominal stress when the deformation is either purely normal to the
interface or purely in the first or the second shear direction, respectively.
The damage evolution law describes the rate at which the material stiffness is degraded once the corresponding initiation
criterion is reached. Assume that a scalar damage variable D represents the overall damage in the material and captures the
combined effects of all the active mechanisms. If damage evolution is modeled, the value of D monotonically evolves from 0
to 1 upon further loading after the initiation of damage. The stress components of the traction–separation model are affected
by the damage follow as

ð1  DÞr
n r n P 0
rn ¼ ð8Þ
r n otherwise

ss ¼ ð1  DÞss ð9Þ

st ¼ ð1  DÞst ð10Þ
where r  n; s
s and s
t are the stress components predicted by the elastic traction–separation behavior for the current strains
without damage. To describe the evolution of damage under a combination of normal and shear deformation across the
interface, it is useful to introduce an effective displacement dm defined as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
dm ¼ d2n þ d2s þ d2t ð11Þ
 Q. Meng, Z. Wang / Engineering Fracture Mechanics 142 (2015) 170–183 175

The linear form of damage evolution law based on effective displacement [26] is chosen for computational convenience as
shown in Fig. 3b. The evolution of the damage variable D can be expressed as

dmf dmax  d0m
D¼ m  ð12Þ
dm dmax  d0m
max f

where d0m and dmf denote the effective displacement at the initiation of damage and complete failure, respectively, and dmax
m
refers to the maximum value of the effective displacement attained during the loading history. The fracture energy / is the
area enclosed under the traction–separation displacement curve
1 0 f
/¼ rd ð13Þ
2 n m
As the damage evolution continues, once the overall damage variable reaches the upper bound, the cohesive elements are
deleted. The element deletion approach [27] is often appropriate for modeling complete fracture of the bond and separation
of components. Once deletion, cohesive elements offer no resistance to subsequent penetration of the components.

2.3. Materials properties

A SiC particle-reinforced Al matrix composite [28] is chosen as a model system. The particles behave as elastic, isotropic
solids characterized by their elastic modulus Ep = 410 GPa and Poisson’s ratio vp = 0.19. The matrix is characterized as an
isotropically hardening elastic–plastic material following the incremental J2 theory of plasticity [29], and the total strain
is given by the elastic components ee and plastic strain components epl

e ¼ ee þ epl ð14Þ
and
r ¼ D : ee ð15Þ
where D is a stiffness tensor which depends on the Young’s modulus, the Poisson’s ratio and the tangential modulus of the
matrix material. The yield condition is taken as the von Mises yield surface
 
f ¼ req  ry epleq ¼ 0 ð16Þ

and
rffiffiffiffiffiffiffiffiffiffiffiffiffi
3
req ¼ S:S ð17Þ
2
 
where ry epleq is the yield stress, S is the deviatoric stress tensor, and epleq is the equivalent plastic strain, and it is defined
incrementally by the relation
Z rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3 pl pl
epleq ¼ e : e dt ð18Þ
2
The stress–strain curve of Al matrix shown in Fig. 4 is extracted from the experimental results by Chawla and Shen [1] and
the elastic constants of the Al matrix are Em = 75 GPa and vm = 0.33. The stress–strain data are input to define the material
hardening behavior in the simulation, and a digitized version of the yield stress ry versus equivalent plastic strain epl eq for the
Al matrix is given in Table 1. Furthermore, the interfacial properties, including the elastic properties, the traction stress com-
ponents (i.e. the interfacial strength) and the effective failure displacement or the fracture energy (which are equivalent by
Eq. (13)), will be predicted.

3. Results and discussion

3.1. Determination of interfacial properties

The interfacial properties and theirs determination have been always a field of major interest in composite materials,
since the interfacial strength between the matrix and the reinforcing phase plays a key role in the damage of composite
materials. Unfortunately, the determination of the interfacial properties by experiment is very difficult and the obtained
results are often doubtable [30]. Here we will perform a prediction of the interfacial properties in particle-reinforced com-
posites by a micromechanical model.
The interfacial properties can be determined by a simple graphical interpretation of tensile stress–strain curves for
particle-reinforced composites, and the similar approach has been adopt by Li et al. [31]. The tensile behavior of a SiC/Al
composite is obtained from experiment by Chawla and Shen [1], and the tensile stress–strain curves for the proposed
176 Q. Meng, Z. Wang / Engineering Fracture Mechanics 142 (2015) 170–183

Fig. 4. The tensile stress–strain curve of Al matrix [1].

Table 1
Yield stress versus equivalent plastic strain for the Al matrix.

Yield stress (MPa) 426.00 454.85 470.07 481.53 489.23 490.48 492.99 495.50 500.51 503.20 506.96 508.21
Equivalent plastic strain 0 0.0003 0.0006 0.0008 0.0015 0.0029 0.0040 0.0058 0.0071 0.0089 0.0108 0.0128
Yield stress (MPa) 513.23 517.17 520.93 524.69 528.45 532.39 534.89 537.40 542.41 545.10 546.35 547.61
Equivalent plastic strain 0.0143 0.0176 0.0198 0.0218 0.0241 0.0268 0.0291 0.0317 0.0342 0.0361 0.0382 0.0418

micromechanical model are obtained using the definition of the applied stress and strain from Eqs. (5) and (6). It is notewor-
thy to mention that the interfacial stiffness defines the elastic response of interface elements when the damage does not take
place. The interfacial stiffness mainly depends on the elastic behavior of composites, and the interfacial stiffness can be
determined by the comparison of the elastic portion of numerical and experimental stress–strain curves. For the
particle-reinforced composites under tensile loads in the experiment [1], after the damage, such as matrix cracking, interface
debonding and particle fracture appears, the trend of stress–strain curves begin to fall sharply. Accordingly, when the inter-
face dobonding occurs in the simulation, the stress–strain curves also show a decrease trend, but the trend is relatively flat.
Therefore, by running successive numerical simulations with the known interfacial stiffness and different values of interfa-
cial strength and failure displacement, numerical predictions of the declining portion of the stress–strain curve can be com-
pared to the experimental stress–strain curves. The best fit between the numerical predictions and the experimental results
gives the characteristic properties of the interface. The interface between particle and matrix is considered as an isotropic
solid, and the interface properties of composites with different volume fractions of particles are the same. The following
interfacial properties are determined in the simulation: the interfacial stiffness Knn = Kss = Ktt = 15 GPa, the interfacial
strength r0n ¼ s0s ¼ s0t ¼ 705 MPa, and the interfacial effective failure displacement dmf ¼ 0:35 lm. According to the Eq.
(13), the fracture energy of interfacial element is / ¼ 123:4 J/m2. Fig. 5 shows the tensile stress–strain curves of a SiC
particle-reinforced Al matrix composite with 20% and 30% volume fraction of particles obtained by the proposed microme-
chanical model and experiment [1]. It is found that the predicted results of the RVE show a reasonable agreement with the
experimental data. The maximum difference between predicted values and experimental results is limited to 3%, and it can
be acceptable to the engineering application. This difference between simulation and experiment is associated with the dis-
tribution of the particle, and this also implies that the assumption made in modeling the SiC/Al composite is valid and can be
extended for the analysis of particle-reinforced composites.
It can be observed that higher Young’s modulus, yield strength and tensile strength of composites are obtained with the
increase of particle volume fraction while the ductility of material decreases gradually. As the volume fraction of particle
increases, the more loads are transferred to the particles, which lead to a higher yield strength and tensile strength of com-
posite. This means that the increasing of particle volume fraction is advantageous to the work hardening of composites.
Furthermore, it is also found that the threshold strain when the trend of stress–strain curves begin to decline decreases grad-
ually with the increase of particle contents. It suggests that the debonding initiation or void nucleation in the composite with
the high particle volume fraction occurs earlier than that of lower particle contents. It is also noted that for the stress–strain
curves obtained by experiment, the stress has a sharp decline when it reaches up to the tensile strength of composite with
the increasing of the strain for the real composite due to the interfacial debonding accompanied by particle fracture and
matrix cracking in the damage process, while the decrease trend of the stress is relatively flat for the stress–strain curves
obtained by the micromechanical model because it only considers the interfacial debonding.
 Q. Meng, Z. Wang / Engineering Fracture Mechanics 142 (2015) 170–183 177

Fig. 5. Comparison of predictions results for stress–strain curves of composite under tensile loads with experimental data [1]. The experimental results
were obtained by Chawla and Shen [1] for SiC/Al composite with varying volume fraction at a constant particle size of 5 lm.

3.2. Analysis of interfacial failure mechanisms

We have predicted the interfacial properties based on the micromechanics model in the above section. The purpose of
study in this part is to reveal micromechanisms of the interfacial failure in the particle-reinforced metal-matrix composites.
Since the change of volume fraction did not change the main damage phenomena observed in the particle-reinforced
composites [32], we take the composites with 20% volume fraction of particles for example to demonstrate the interfacial
failure mechanisms of the particle-reinforced metal-matrix composite. Fig. 6 shows the contour plot of the Mises stress
for the RVE drawn from the mid cross-section and scalar stiffness degradation variable of interfacial element at various
strains, where the size of the void formed is represented by the size of the element deletion of interface element. It can
be observed that when the tensile loads are applied in the composites, the particle and matrix will results in the different
deformation due to the difference of the material properties and the stress concentration will occur in the interface between
particle and matrix (Fig. 6a). Then the debonding damage will happen when the level of the stress concentration reaches up
to the interfacial strength between particle and matrix. The debonding first appears at the end of particles orientated parallel
to the stress axis near the side of the loading (Fig. 6b). After initiation, the interfacial debonding between particle and matrix
expands progressively. As the deformation increases, new debonding will appear on particle already debonded at the oppo-
site pole (Fig. 6c). With the continued increase of the tensile deformation, the size of the interfacial debonding, i.e. the void
formed, is found to increase at both sides of the particle. In Fig. 6d, the debonding damage appears in the horizontal direction
while in the vertical direction there is the strong compression effect between particle and matrix. In addition, in the damage
region a small portion of element in the matrix and particle material have lost the bearing capacity and are already unload-
ing, which can be the regard as the failure region. As the debonding process continues, the pressure region decreases
gradually whereas the failure elements in the interface increase significantly. We can further predict that the matrix damage
or fracture will appear along the overlap region between the failure region and the pressure region in the numerical model.
The experimental results [10] has also shown that as the load increases, the small voids may nucleate and the impingement
of the voids nucleated at neighbor particles or the localization of the plastic deformation in the intervoid matrix further leads
to matrix fracture or damage, which are mainly oriented with an angle roughly equal to 45° from the tensile direction.
However, it is also noticed that a complete interfacial debonding all over the particle could never be reached in the simula-
tion. This result also confirms observations made by experimental investigations [9,32]. Actually, the internal stresses in the
particle and matrix are differently modified by the interfacial debonding and the interfacial debonding relaxes progressively
these stresses. When the interfacial debonding reaches a certain level, the stress of the model decreases gradually as the
applied strain increases, as shown in Fig. 5. Moreover, the plastic deformation of the matrix around the particle leads to
the formation of a void at the poles [9], as shown in Fig. 7. It is also shown by experimental study [9] that the extent of inter-
facial damage would generally depend on the particle characteristics such as the size, shape, composition, and volume frac-
tion and the fabricating process of the composites. Fig. 8 shows contour plot of the Mises stress for the RVE viewed from a
side paralleled the loading direction and scalar stiffness degradation variable (SDEG) of interfacial element. It is observed
that the interfacial debonding between matrix and particles also appears at the corners of the RVE. These damages would
have an effect on the decreasing portion of stress–strain curves obtained from the RVE, and by this way the RVE in the paper
captures the effect of particle–particle interactions.
178 Q. Meng, Z. Wang / Engineering Fracture Mechanics 142 (2015) 170–183

(a)

(b)

(c)

Pressure
region

Failure region

(d)
Fig. 6. Contour plot of the Mises stress for the particle-reinforced composite with 20% volume fraction of particles and scalar stiffness degradation variable
(SDEG) of interfacial element at various strains: (a) 1.5%; (b) 2.5%; (c) 2.94%; (d) 3.5%. The loading axis is horizontal and the load is applied on the right side
of model.
 Q. Meng, Z. Wang / Engineering Fracture Mechanics 142 (2015) 170–183 179

Fig. 7. Experimental characterization of the interfacial debonding in particle-reinforced composite [9].

Fig. 8. Contour plot of the Mises stress for the RVE viewed from a side paralleled the loading direction (a) and scalar stiffness degradation variable (SDEG) of
interfacial element (b).

3.3. Effect of interfacial properties

The onset of damage in the interface between particle and matrix is mainly characterized by three parameters, the inter-
facial stiffness K = Knn = Kss = Ktt, the interfacial strength r0 ¼ r0n ¼ s0s ¼ s0t and the fracture energy /, and their effect on the
composite behavior is studied in the RVE model with 20% volume fraction of particles.
180 Q. Meng, Z. Wang / Engineering Fracture Mechanics 142 (2015) 170–183

Since the interface element only has the elastic and damage behavior in our simulation, the interfacial stiffness has a
small effect on the plasticity and damage behavior of composite. For convenience, we only test the tensile behavior of com-
posite under the no interface damage condition. The tensile stress–strain curve of composite for different interfacial stiffness
is shown in Fig. 9. It can be seen that variations in the interfacial stiffness do not result in a significant difference of stress–
strain behavior in the composite. The Young’s modulus of composite obtained by Eq. (6) at different interfacial stiffness in
the simulation and the experimental value of Young’s modulus of composite with 20% volume fraction of particles [33] are
shown in Table 2. With the increase of interfacial stiffness, the Young’s modulus shows an increase trend, and is gradually
close to the limiting value. The value of Young’s modulus in the composite with the interfacial stiffness K = 15 GPa is close to
the experimental result and this is why we consider the interfacial stiffness K = 15 GPa in the above simulation. Moreover, it
must be pointed out that there are two hypotheses concerning the interfacial structure between the particle and matrix: (1)
there exist an interface layer between matrix and particle; (2) there is nothing at the interface, i.e. the interface is formed
simply by two surfaces of particle and matrix. The reason for no clearly reveal of the interfacial structure is not only the dif-
ferent crystallographic structure and chemical bonding of both particle and matrix, but also the way of production of the
composite [34]. Interfacial bonds may be formed by mutual dissolution or reaction of particles and matrix. The reaction
phenomena are very detrimental to the composite as they bring about a decrease of the mechanical properties [35]. If the
interface layer or the interfacial reaction product, such as Al4C3 [36,37] and MgAl2O4 [38,39], is existed, the thickness and
stiffness of the interface element in the RVE become an important problem in the modeling process. When the interface layer
is not existed or there is no chemical reaction product at the interface derived from diffusion bonding [40], the stiffness of the
interface element in the RVE should be as small as possible as but not be zero, and the interfacial thickness should be close to
or equal to zero as far as possible. The effect of interfacial microstructure on the mechanical behavior of composites will be
analyzed by the micromechanical model in further research.
The stress–strain curves of composite with 20% particle volume fraction at different interfacial strength are shown in
Fig. 10 when the interfacial stiffness K = 15 GPa and the interfacial failure displacement dmf ¼ 0:35 lm. It can be observed that
with the increase of interfacial strength, the stress and strain of composite when the trend of stress–strain curves begin to
decline increase gradually. When the interfacial strength is smaller than 600 MPa (Fig. 10a), the interfacial debonding occurs
within the elastic range of deformation. If the interfacial strength is greater than 600 MPa (Fig. 10b), the interfacial debond-
ing appears in the range of plastic deformation. Similarly, it can be seen from Fig. 5 that the interfacial debonding occurs
within the range of plastic deformation of composite in the experiment. As the interfacial strength increases, the plastic
strain of composite when the interfacial debonding occurs shows an increase trend. The higher the interfacial strength,
the quicker increase rate the plastic strain has when the interfacial debonding appears in the composite. Moreover,
According to the Fig. 5 and Eq. (13), it can be seen that the fracture energy of interfacial debonding is determined by the
interfacial strength and the interfacial failure displacement. When one of them is changed, the variety of the fracture energy
also appears. In the above simulation, we have study effect of the different interfacial strength, and the variety of the

Fig. 9. The stress–strain curves of composite with 20% particle volume fraction at different interfacial stiffness (GPa).

Table 2
Young’s modulus of particle-reinforced composites with 20% volume fraction of particles at different interfacial stiffness.

Interfacial stiffness (GPa) 200 150 100 50 15 Experiment [33]

Fig. 10. The stress–strain of composite with 20% particle volume fraction for different interfacial strength (MPa): (a) lower interfacial strength; (b) higher
interfacial strength.

Fig. 11. The schematic traction–separation laws of the interface element for the different fracture energy (J/m2) obtained by Eq. (13).
182 Q. Meng, Z. Wang / Engineering Fracture Mechanics 142 (2015) 170–183

Fig. 12. The stress–strain curves of composite for different interfacial strength (MPa) and the corresponding fracture energy (J/m2).

interfacial strength also results in the change of the fracture energy. The schematic traction–separation laws of the interface
element for the different fracture energy obtained by Eq. (13) are shown in Fig. 11 when the interfacial failure displacement
dmf ¼ 0:35 lm. The stress–strain curves of composite for different interfacial strength and the corresponding fracture energy
are shown in Fig. 12. However, when the effect of interfacial failure displacement is studied at the constant interfacial
strength, the model program is not convergent and it is difficult to obtain well simulation results. Therefore, we would
not analyze the effect of interfacial failure displacement. It must be pointed out that the interfacial strength and the fracture
energy play a key role in the interfacial debonding of particle-reinforced metal-matrix composite.

4. Conclusions

We have analyzed the interfacial failure mechanisms in particle-reinforced metal-matrix composites using a microme-
chanical model proposed and predicted the interfacial properties of a SiC particle-reinforced Al matrix composite: the inter-
facial stiffness K = 15 GPa, the interfacial strength r0 = 705 MPa, and the fracture energy / ¼ 123:4 J/m2. It is found that
Young’s modulus, yield strength and tensile strength increase gradually while the ductility of material decreases gradually
with the increase of particle volume fraction.
It can be seen that the particle and matrix will result in the different deformation under loading due to the difference of
the material properties. The interfacial debonding between particle and matrix will appear when the stress concentration
level reaches up to the interfacial strength. With the increase of the tensile deformation, the size of the void formed gradually
increases at both sides of the particle. However, a complete interfacial debonding all over the particle could never be reached.
Finally, the matrix damage will appear along the overlap region between the failure region and the pressure region.
Effect of interfacial properties on the tensile behavior of composite is also studied. It is found that variations in the
interfacial stiffness do not result in a significant difference of stress–strain behavior in the composite. When the interfacial
strength is greater than 600 MPa, the interfacial debonding occurs within plastic range of deformation. If the interfacial
strength is smaller than 600 MPa, the interfacial debonding occurs within elastic range of deformation. As the
interfacial strength increases, the plastic strain of composite when the interfacial debonding occurs shows an increase trend.
The interfacial strength and the fracture energy play a key role in the interfacial debonding of particle reinforced
metal-matrix composite.

Acknowledgements

This work is supported by the National Natural Science Foundation of China (Nos. 11272096 and 11472086).

References

[1] Chawla N, Shen YL. Mechanical behavior of particle reinforced metal matrix composites. Adv Eng Mater 2001;3(6):357–70.
[2] Torralba JM, Da Costa CE, Velasco F. P/M aluminum matrix composites: an overview. J Mater Process Technol 2003;133(1–2):203–6.
[3] Watt DF, Xu XQ, Lloyd DJ. Effects of particle morphology and spacing on the strain fields in a plastically deforming matrix. Acta Mater
1996;44(2):789–99.
[4] Qin SY, Chen CR, Zhang GD, Wang WL, Wang ZG. The effect of particle shape on ductility of SiCp reinforced 6061 Al matrix composites. Mater Sci Eng A
1999;272(2):363–70.
[5] Yan YW, Geng L, Li AB. Experimental and numerical studies of the effect of particle size on the deformation behavior of the metal matrix composites.
Mater Sci Eng A 2007;448(1–2):315–25.
[6] Kiser MT, Zok FW, Wilkinson DS. Plastic flow and fracture of a particulate metal matrix composite. Acta Mater 1996;44(9):3465–76.
 Q. Meng, Z. Wang / Engineering Fracture Mechanics 142 (2015) 170–183 183

[7] Segurado J, LLorca J. Computational micromechanics of composites: the effect of particle spatial distribution. Mech Mater 2006;38(8–10):873–83.
[8] Chawla N, Sidhu RS, Ganesh VV. Three-dimensional visualization and microstructure-based modeling of deformation in particle-reinforced
composites. Acta Mater 2006;54(6):1541–8.
[9] Kanetake N, Nomura M, Choh T. Continuous observation of microstructural degradation during tensile loading of particle reinforced aluminum matrix
composites. Mater Sci Technol 1995;11(12):1246–52.
[10] Babout L, Maire E, Buffière JY, Fougères R. Characterization by X-ray computed tomography of decohesion, porosity growth and coalescence in model
metal matrix composites. Acta Mater 2001;49(11):2055–63.
[11] Dugdale DS. Yielding of steel sheets containing slits. J Mech Phys Solids 1960;8(2):100–4.
[12] Barenblatt GI. The mathematical theory of equilibrium cracks in brittle failure. Adv Appl Mech 1962;7:55–129.
[13] Elices M, Guinea GV, Gómez J, Planas J. The cohesive zone model: advantages, limitations and challenges. Eng Fract Mech 2002;69(2):137–63.
[14] Park K, Paulino GH. Cohesive zone models: a critical review of traction–separation relationships across fracture surfaces. Appl Mech Rev
2011;64(6):060802.
[15] Xu XP, Needleman A. Void nucleation by inclusion debonding in a crystal matrix. Model Simul Mater Sci Eng 1993;1(2):111–32.
[16] Charles Y, Estevez R, Bréchet Y, Maire E. Modelling the competition between interface debonding and particle fracture using a plastic strain dependent
cohesive zone. Eng Fract Mech 2010;77(4):705–18.
[17] Segurado J, LLorca J. A computational micromechanics study of the effect of interface decohesion on the mechanical behavior of composites. Acta Mater
2005;53(18):4931–42.
[18] Tsui CP, Chen DZ, Tang CY, Uskokovic PS, Fan JP, Xie XL. Prediction for debonding damage process and effective elastic properties of glass-bead-filled
modified polyphenylene oxide. Compos Sci Technol 2006;66(11–12):1521–31.
[19] Kang GZ, Shao XJ, Guo SJ. Effect of interfacial bonding on uniaxial ratchetting of SiCP/6061Al composites: finite element analysis with 2-D and 3-D unit
cells. Mater Sci Eng A 2008;487(1–2):431–44.
[20] Romanova VA, Balokhonov RR, Schmauder S. The influence of the reinforcing particle shape and interface strength on the fracture behavior of a metal
matrix composite. Acta Mater 2009;57(1):97–107.
[21] Mishnaevsky Jr L, Dong M, Hönle S, Schmauder S. Computational mesomechanics of particle-reinforced composites. Comput Mater Sci 1999;16(1–
4):133–43.
[22] Chen XH, Mai YW. Micromechanics of rubber-toughened polymers. J Mater Sci 1998;33(14):3529–39.
[23] Danielsson M, Parks DM, Boyce MC. Three-dimensional micromechanical modeling of voided polymeric materials. J Mech Phys Solids
2002;50(2):351–79.
[24] González C, Segurado J, LLorca J. Numerical simulation of elasto-plastic deformation of composites: evolution of stress microfields and implications for
homogenization models. J Mech Phys Solids 2004;52(7):1573–93.
[25] Brewer JC, Lagace PA. Quadratic stress criterion for initiation of delamination. J Compos Mater 1988;22(12):1141–55.
[26] Camanho PP, Davila CG, De Moura MF. Numerical simulation of mixed-mode progressive delamination in composite materials. J Compos Mater
2003;37(16):1415–37.
[27] Wulf J, Schmauder S, Fischmeister HF. Finite element modelling of crack propagation in ductile fracture. Comput Mater Sci 1993;1(3):297–301.
[28] Chawla N, Habel U, Shen YL, Andres C, Jones JW, Allison JE. The effect of matrix microstructure on the tensile and fatigue behavior of SiC particle-
reinforced 2080 Al matrix composites. Metall Mater Trans A 2000;31(2):531–40.
[29] Davis LC. Flow rule for the plastic deformation of particulate metal-matrix composites. Comput Mater Sci 1996;6(4):310–8.
[30] Hobbiebrunken T, Hojo M, Adachi T, De Jong C, Fiedler B. Evaluation of interfacial strength in CF/epoxies using FEM and in-situ experiments.
Composites Part A 2006;37(12):2248–56.
[31] Li S, Thouless MD, Waas AM, Schroeder JA, Zavattieri PD. Use of mode-I cohesive-zone models to describe the fracture of an adhesively-bonded
polymer-matrix composite. Compos Sci Technol 2005;65(2):281–93.
[32] Babout L, Maire E, Fougères R. Damage initiation in model metallic materials: X-ray tomography and modelling. Acta Mater 2004;52(8):2475–87.
[33] Chawla N, Andres C, Jones JW, Allison JE. Effect of SiC volume fraction and particle size on the fatigue resistance of a 2080 Al/SiCp composite. Metall
Mater Trans A 1998;29(11):2843–54.
[34] Romero JC, Wang L, Arsenault RJ. Interfacial structure of a SiC/Al composite. Mater Sci Eng A 1996;212(1):1–5.
[35] Shorowordi KM, Laoui T, Haseeb ASMA, Celis JP, Froyen L. Microstructure and interface characteristics of B4C, SiC and Al2O3 reinforced Al matrix
composites: a comparative study. J Mater Process Technol 2003;142(3):738–43.
[36] Lloyd DJ. The solidification microstructure of particulate reinforced aluminium/SiC composites. Compos Sci Technol 1989;35(2):159–79.
[37] Tham LM, Gupta M, Cheng L. Effect of limited matrix-reinforcement interfacial reaction on enhancing the mechanical properties of aluminium-silicon
carbide composites. Acta Mater 2001;49(16):3243–53.
[38] Lee JC, Ahn JP, Shi ZL, Kim YM, Lee HI. Modification of the interface in SiC/Al composites. Metall Mater Trans A 2000;31(9):2361–8.
[39] Shi ZL, Yang JM, Lee JC, Zhang D, Lee HI, Wu RJ. The interfacial characterization of oxidized SiC(p)/2014 Al composites. Mater Sci Eng A 2001;303(1–
2):46–53.
[40] Cheng NP, Zeng SM, Liu ZY. Preparation, microstructures and deformation behavior of SiCp/6066Al composites produced by PM route. J Mater Process
Technol 2008;202(1–3):27–40.