Acta muter. Vol. 44, No. I, pp.

173-200, 1996

Pergamon

Elsevier Science Ltd

0956-7151(95)00175-l

Copyright 6 1995 Acta Metallurgica Inc.

Printed in Great Britain. All rights reserved
1359-6454/96 $15.00 + 0.00

CRACK PROPAGATION
IN PIEZOELECTRIC
UNDER
COMBINED
MECHANICAL
AND
LOADINGS

MATERIALS
ELECTRICAL

SUSMIT KUMAR and RAJ N. SINGHJr

Department

of Materials

Science and Engineering, University of Cincinnati,

OH 45221-0012, U.S.A.

P.O. Box 210012, Cincinnati,

(Received 16 December 1994; in revised form 23 March 1995)

Abstract-A
finite element technique is used to study the stress distributions
at the crack tip of a
piezoelectric ceramic because of the mechanical and/or electrical loads. The stress distributions
at the crack
tip are found to be in conformity
with those predicted theoretically
by Sosa and Pak. For combined
mechanical and electrical loads, the stress intensity factor at the crack tip is observed to increase with an
increase in the electrical to mechanical load ratio for a negative applied electric field (electric field opposite
to the direction of poling) which is in agreement with the experimental findings of Wang and Singh. For
the positively applied electric field, compressive stresses are found to develop around the crack tip for high
electrical to mechanical load ratios. For combined mechanical and electrical loads, the stress distributions
at the crack tip under applied stress is found to be significantly different from those under applied strain.
It is shown that a negative applied electric field with a tensile strain perpendicular
to the crack surface
increases the crack propagation.

D;=D,=O

1. INTRODUCTION

The electro-mechanical
coupling in piezoelectric materials generates stresses by the application
of an
electrical field, and electrical charges by the application of a mechanical load. This makes piezoelectric
materials very useful in electromechanical
and electronic devices such as electromechanical
actuators,
sensors and transducers. In these devices, both electrical and mechanical loads are applied on the piezoelectric components
which give rise to sufficiently high
stresses which can lead to their failure. The analysis
of crack problems in piezoelectric materials is complicated in comparison
to the elasto-plastic
materials
because of the anisotropic electro-mechanical
properties and the electromechanical
coupling effect.
Recently,
several researchers
have studied the
crack problems in piezoelectric materials both analytically [l-17] and experimentally
[18-301. In the
theoretical
studies of the crack problems,
the researchers have different opinions about the boundary
conditions
at the crack surfaces. Usually, as the
dielectric constant of the air or the medium between
the crack faces is very small as compared to that of
the piezoelectric material, some researchers [2-91 have
assumed crack surfaces to be free of surface traction
and surface charge, i.e.

~To whom

all correspondence

should

(2)

where, D, is the normal electrical displacement at the

crack faces. Superscripts
+ and - denote the upper
and lower crack surfaces, respectively.
On the other hand, researchers
like Parton [l],
Zhang and Hack [lo], Hao and Shen [12], and Dunn
[ 131 have assumed that although the magnitude of the
normal electrical displacement
component
at the
crack face is very small, the electrical displacement is
continuous across the crack faces. Parton [I] used the
electrical boundary conditions as
D;=D;

(3)

++=C#-

(4)

and

where 4 is the electrical potential. Hao and Shen [12]

used the boundary condition in which the electric
permeability of air in a crack gap is considered. Apart
from equation (3), they used an additional equation
for the boundary condition at the crack faces
D,+(uHf-uug-)=c,(c#--c#J+)

(5)

where u0 is the displacement

component
normal to
the crack face and t, is the permittivity of air. If there
is no gap between the crack faces, i.e. u: - u, = 0,
then equation (5) reduces to equation (4) and for the
condition t, = 0, i.e. for a medium having zero electrical permittivity, equation (5) reduces to equation (2).

be addressed.
173

In his pioneering work. Parton [I] mathematically

analyzed the deformation
of a piezoelectric medium
under the applied far-field mechanical
stress. He
obtained a generalized Griffith formula relating the
critical crack length with an applied mechanical
stress. Deeg [2] modeled the crack with a collinear
dislocation
and charge dipole line and used the
Greens function method to study the fracture problem in piezoelectric material. Li et nl. [3] analyzed the
inhomogeneous
distributions
of internal shear stress
and electric field induced by external shear stress or
applied electric field around a Griffiths type III crack
in piezoelectric medium. They found that the applied
electric field (mechanical stress) can either weaken or
enhance the stress (electric displacement)
concentration in a piezoelectric
material. Pak [4] used a
conservation
law to derive a path-independent
integral of fracture
mechanics
for the piezoelectric
medium and evaluated this integral to obtain the
energy release rate for mode III fracture problem
under electrical
boundary
condition
defined
by
equation
(2). He obtained
solution by the semiinverse method for the case of out-of-plane
deformation and in-plane electrical loads and found the
classical mode III crack-tip singular behavior for all
the field variables. For a fixed value of the mechanical
load, he showed that the crack growth can be either
enhanced or retarded depending on the magnitude,
the direction, and the type of the electrical load. He
also showed that for certain ratios of the applied
electrical load to mechanical load. crack-arrest can be
observed. For mode III fracture problem, the crack
extension force is found to be always negative in the
absence of the mechanical
load. indicating that a
crack would not propagate
under these conditions.
Using similar electrical boundary
conditions
[i.e.
equation (2)], Sosa and Pak [5] analyzed a threedimensional
semi-infinite
crack in a transversely
isotropic piezoelectric medium by means of the eigenfunction expansions method and found the characteristic Ii,, Y singular behavior for both the stress and
electric fields in the vicinity of crack. They also
demonstrated
that coupling (or decoupling)
of the
mechanical and electrical variables is a function of the
distance to the crack tip as well as of the crack
orientation.
Sosa [6] used complex potentials formulation
to
analyze the stress concentrations
in the piezoelectric
medium in the vicinity of circular and elliptical holes.
He extended this analysis to obtain stress distributions at the crack tip in the piezoelectric medium
under mechanical and electrical loads assuming the
plane strain conditions for mode I crack loading [7].
He found that along the crack line, the normal
component
of stress is independent
of the material
properties and electrical load. Likewise, the normal
component of electrical displacement at the crack tip
along the crack line is found to be independent of the
applied stress. He showed that for combined mechanical and electrical loads, the crack might be turned

towards a &I-ectlon opposite 10 that 01the applied

field for higher electrical to mechanical load ratio\.
using the method of distributed
dislocations
and
electric dipoles. Pak [X] found similar results for the
crack tip stresses ax those ol Sosa [7]. Pak also
calculated the energy release rate for a finite crack
embedded in an infinite piezoelectric medium which
was subjected to both mechanical and electrical loads.
He showed that the electric fields generally slowed the
crack growth.
Wang [9] used the Fourier transformation
method
to analyze a flat elliptical crack embedded
in a
piezoelectric medium, and derived expressions for the
crack opening displacement.
interaction energy. and
stress intensity factors for the combined mechanical
and electrical loads. He concluded that the stress and
electric fields near the crack edge have the classical
l,:,r type of singularity, and only the component of
the electric field which is perpendicular
to the crack
surface influences the stress intensity factors.
Zhang and Hack [IO] used different
electrical
boundary conditions at the crack surfaces from those
used by Deeg [2]. Pak [4, 81 and Sosa [6.7]. They
assumed the electrical displacement
component
perpendicular to the crack surfaces to be continuous.
For mode III cracks in piezoelectric medium. they
found that the stress field at the crack tip is independent of the applied electric field under remotely
applied stress. This result is in stark contrast to the
result obtained by assuming that the electrical displacement
component
perpendicular
to the crack
surfaces is equal to zero. Under conditions
of a
remotely imposed strain, the electric field can exert a
significant influence on the crack propagation
and as
a result the electric field may promote or retard the
crack propagation
depending
on its direction.
A
negative electric load (with constant applied strain) is
found to always promote
crack growth while a
positive electric load is found to retard crack growth
at low magnitudes and promotes crack growth as its
magnitude increases [IO].
In a recent paper. Zhang [1 I] has discussed the
effects of sample width on the energy release rate and
electric boundary
conditions
along crack surfaces.
Using the electrical boundary condition at the crack
tip as defined by equation (5). Hao and Shen [12]
found that the electric permeability of air in a crack
gap leads to a value of stress intensity factor less than
that of an impermeable crack. Dunn [ 131 analyzed the
effects of crack face boundary
conditions
on the
fracture mechanics
of piezoelectric
materials.
His
results indicated that the impermeable crack assumption can lead to significant errors regarding the effects
of the electric fields on the crack propagation
based
on an energy release rate criterion. He showed, based
on an energy release rate criterion, that the electric
field does not promote crack growth, the amount that
they can retard crack growth can be significantly
overestimated
by the impermeable
assumption.
Suo
114. 151 found that an electric field applied perpen-

KUMAR

and SINGH:

CRACK

PROPAGATION

dicular to the crack surface produces

a positive
driving force for the conducting crack and a negative
driving force for the insulating crack. Suo et al. [16]
presented
a fracture mechanics
theory to evaluate
static toughness
and fatigue crack growth in ferroelectric
ceramics
under small scale hysteresis.
Shindo
et al. [17] used Fourier
transformation
method to determine the distribution of stresses in an
infinitely long piezoelectric
strip containing
a finite
Griffith crack. Their results showed that the effect of
electroelastic
interactions
on the stress and electric
field intensity factors was more pronounced
with
decreasing ratio of the width of the strip and the
crack length.
In the last decade, some researchers have experimentally investigated
the fracture behavior of the
piezoelectric
ceramics. The anisotropic
behavior of
the piezoelectric ceramics is because of the difference
in the axis lengths in a and c directions of the crystals
in tetragonal
phase below the Curie point. The
crystals try to switch the a and c axes in order to align
their c axes along the poling direction during the
poling process which is usually done at a very high
electric field. The difference in the axis lengths in these
two directions and constraints imposed by the neighboring grains causes residual stresses in the materials
and its anisotropic
properties.
Using indentation
technique, Yamamoto
et al. [18], Cook et al. [193,
Okazaki [20] and Wang and Singh [21,22] have
investigated the anisotropic residual stresses in poled
piezoelectric materials and their effects on the crack
propagation.
Using double torsion testing technique,
McHenry
and Koepke [24] found that the crack
always turned in a direction opposite to the poling
direction
when an a.c. or d.c. field was applied
perpendicular
to the crack plane in poled specimen.
They found the crack propagation
to be always
transgranular.
Cook et al. [19], Niitsuma
and
Chubachi
[25], Pisarenko
et al. [26] and Pohanka
et al. [27] experimentally
investigated
the fracture
strength
of the piezoelectric
materials.
Pisarenko
et al. [26], Mehta and Virkar [28] and Wang and
Singh [21-231 found that the crack length was shorter
in the poling direction than in the direction perpendicular to the poling axis. Due to poling, compressive
residual stresses are developed in the materials along
the polarization
direction which should give rise to a
lower fracture toughness in the polarization
plane.
Hence this experimental
anisotropic behavior of the
fracture toughness can not be explained by the residual stresses in the materials. In order to explain
this discrepancy,
Pisarenko
et al. [26] suggested a
model based on the assumption
of a stress-induced
domain reorientation
zone near the crack tip. Similar
results for the anisotropic
fracture toughness were
reported by Wang and Singh [2 1,221 and Mehta and
Virkar [28]. Mehta and Virkar [28] concluded that
this anisotropic
fracture
toughness
behavior
of
piezoelectric material is due to the domain switching
during fracture and due to this switching, c axes of

IN PIEZOELECTRIC

MATERIALS

155

many of the domains

became orthogonal
to the
fracture surfaces. Recently, White et al. [29], Cao and
Evans [30] and Lynch et al. [3 I] have investigated the
crack growth in piezoelectric materials under cyclic
loads.
One of the authors [21-231 has recently experimentally investigated
the residual stresses and crack
growth in poled piezoelectric
materials under combined mechanical and electrical loads using an indentation technique.
The motivation
for this finite
element analysis of fracture mechanics in these materials has been to correlate the theoretical findings to
the experimental results. The experimental results will
be used as a feedback to develop a model to study the
fracture mechanics in these materials in greater detail.
This paper is the first in a series of papers to be
published on the finite element analysis of the fracture mechanics in piezoelectric materials.
In the present study, a finite element technique is
used to study the fracture mechanics in piezoelectric
materials. A double edge-notched
specimen (Fig. 1) i:s
modeled to analyze the stress distributions
at the
crack tip under combined mechanical and electrical
loads. The calculated stress distributions are found to
be in conformity with those calculated theoretically
by Sosa [7] and Pak [8]. For combined mechanical
and electrical loads, the stress intensity factor at
the crack tip is found to increase with increasing

Crack Length, a = OE
c/a = 2.1)
b/c = 2.0

Fig. I. The double edge notch (DEN) model in two dimensions. Direction shown by P is the poling direction. Due to
symmetry,
only half of the double edge notch specimen,
ABCD, has been modeled. For the mechanical
boundary
condition,
point C was fixed; line CD and line AC were
constrained
to have displacement
only along the X- and
Y-axis, respectively. For the electrical boundary condition,
the line CD was grounded.
Crack surfaces at OE were
assumed to be traction free, i.e. (roO= or0 = bg, = 0. For the
electrical boundary
conditions
at the crack surfaces, the
normal electrical displacement
was assumed to be zero.
Normal stress (normal strain) and/or electric potential were
applied on the line AB.

I 76

kUMAR

and SING11:

(-RACK

PROPAGATIOh

The constitutive
materials are

The electric field strength is related to the electrical

potential (i, by the equation

The governing

fl,,,i +.L = 0
D,,, = Y,

(10)
(11)

o,,n, = T,

(12)

D,n, = -Y,

(13)

where, n, is the unit normal vector to the boundary

surface. T, and qxare the surface traction and surface
charge, respectively, applied on the boundary surface.
Poled piezoelectric materials like barium titanate
and lead zirconate titanate (PZT) are transversely
isotropic elastic material with hexagonal symmetry of
class 6 mm. The constitutive
relations for these materials can be written as (with .yj as the poling
direction and x,-x? plane as the isotropic plane)

(6)

ek,,Ek

D, = e,,,s,, + ( ,1.Ek

(7)

where, Us, So,, Ek and D, are stress, strain, electric field

strength and electric displacement
tensors, respectively. C,,,, cd and elk, are the elastic constant tensor,
the dielectric constants
and the piezoelectric
constants, respectively, of the material. The strain tensor

Cl, CL2 Cl,

Cl2 (1, Cl3

G,

c,,

c,,

c&t

1NC, i-c,,)

1L
0

are given by

where, fj and qb are the body force and body charge.

respectively.
The boundary conditions are given by

for the piezoelectric

gI = C&L, -

e31

field equations

EQUATIONS

equations

MAIkRIALh

whcrc

electrical to mechanical
load ratio for a negative
applied electric field which is in agreement with the
experimental
findings of Wang and Singh [22,23].
For combined mechanical
and electrical loads, the
stress distributions
at the crack tip under applied
stress are found to be significantly
different from
those under applied strain. The validity of the mechanical boundary
condition
at the crack surfaces
defined by equation (1) has also been discussed.
Generally. the piezoelectric ceramics have nonlinear electromechanical
behavior even under moderate
electric field and stress. But in the present study, we
have assumed linear constitutive behavior all the way
to the crack tip. The results reported in this paper are
calculated using the electrical boundary condition as
defined by equation (2). In future, we will report the
results of the fracture mechanics analysis based on
the electrical
boundary
conditions
defined
by
equations (3) and (5).
2. CONSTITUTIVE

Ih PII;LOEL.t(TKIC

e31

e33

215
00

et5
0
0

1-l
26,
2%
.s>:
S3R

2SI?

is related to the displacement

by the equation

vector {u,,

s,, = ;($ + u,.,)

i=

1, 2, 3f

(8)

where C,, and et, are the elastic constant

tensor
and the piezoelectric
constants,
respectively,
in
the contracted
notation. The relationships
between
the contracted
and expanded
notations
of these

KUMAR and SINGH:

variables

CRACK PROPAGATION

IN PIEZOELECTRIC

MATERIALS

177

are given as

c,, = c,,,,

= C,,,,

C,, = C3333

c,, = c1122 C,, = cu33 = c,,,,

C,, = C,,,, = c,,,,

C6, = c,*,z = l/2 (C,,-C,,)

c31 = e311= e321

e 33 =

e333

e15 =

el13

e323

3. MODEL
In order to study the stress distributions
at the
crack tip in a piezoelectric material, a double edgenotched (DEN) specimen has been modeled. The
dimensions of the specimen are shown in Fig. 1. Due
to symmetry, only the right half of the model (shown
by ABCD) has been used in the finite element analysis. Direction shown by P is the poling direction. The
crack surface OE is assumed to be perpendicular
to
the poling direction. A specialized preprocessor
has
been written to divide the model into finite elements taking into consideration
the two free surfaces of the crack along OE. In order to accurately
calculate the stress distributions
at the crack tip, the
mesh density has been increased near the crack tip.

Fig 20)
Distorted Quadrilateral
Element

Fig. 2(c)
Fig. 2. (a) The finite element mesh for the model. Eightnoded quadrilateral
finite elements with 3 d.f. U, (x-displacement), uJ (y-displacement)
and 4 (electric potential) were
used in the model. The total number of nodes and elements
were 4551 and 1488, respectively. (b) The enlarged view of
the finite element mesh near the crack tip. The quadrilateral
elements at the crack tip were at an interval of 7.5. (c) The
angular distributions
of the radial and tangential
stresses
were calculated for the four node sets defined by the points
along the circular arcs OP, OQ, OR and OS. The ratios
OP/OE,
OQ/OE,
ORjOE
and OSjOE
are 1.0 x 10m5.
4.0 x 10m5, 6.0 x 10m5 and 8.0 x 10-5, respectively.
OE is
the crack length (Fig. 1).

Fig. 2(a)

The finite element mesh for the model is shown in Fig.

2(a). Eight-noded quadrilateral
finite elements with 3
d.f.--u, (x-displacement),
uY(y-displacement)
and 4
(electric potential)-are
used in the model. The total
number of nodes and elements in the finite element
analysis are 4551 and 1488, respectively.
Figure 2(b) shows the enlarged view of the finite
element mesh near the crack tip. The quadrilateral
elements at the crack tip are at an interval of 7.5 .
Sosa and Pak [5] and Wang [9] theoretically showed
that stress and electric fields at the crack tip show
classical l/Jr type singularity. In order to produce
a l/Jr type of singularity for the displacement
and
electric potential at the crack tip, the three nodes
along one of the sides of each of the quadrilateral
elements are collapsed at the crack tip and the two
adjoining mid-points are moved to the quarter point
distances [32]. The angular distributions
of the radial
and tangential components
of stresses at an interval
of 7.5 are calculated for the four node sets defined

bq points along the circular arcs OP. OQ. OR and OS

[Fig. 2(c)]. The ratios OPOE. OQ!OE. 0R:OE and
OS!OE (radial distances in terms of the crack length
OE) arc I .I) x 10 . 4.0 x IO . 6.0 x IO and
8.0x 10 '.respectively.
The side lengths (in radial
direction from the crack tip) of the elements are
about
I.25 times that of the preceding
element
(starting from the element at the crack tip) in order
to increase the accuracy of the results. Initially, we
tried a finite element mesh for which the ratios
ORjOE
and
OSjOE
are
OP,OE.
OQ/OE,
1.25 x 10 4. 5.0 x lo-. 7.5 x 10 and I.0 x IO-.
respectively. but for this mesh. the angular variations
of stress distributions
calculated for applied electric
field with no mechanical load are found to be inaccurate and different from those predicted theoretically by
Sosa [7] and Pak [8]. Although, due to symmetry.
only one quarter of the model shown in Fig. 1 needs
to be modeled, but in order to check the accuracy of
the results Tar all 0 (ranging from 0 to 360), we
modeled half of the DEN specimen (i.e. ABCD).
For the mechanical boundary conditions. point C
was fixed; lines CD and AC were constrained to have
displacement
only along the X- and Y-axis, respectively. For the electrical boundary condition, the line
CD was grounded. Crack surfaces at OE were assumed to be traction free. i.e. u,,,)= g,,) = glir = 0. For
the electrical boundary conditions at the crack surfaces, the normal electrical displacement was assumed
to be zero. Normal tensile stress 0 or tensile strain s
and/or electric potential were applied on the line AB
for creating mechanical
and/or electrical loads, respectively. The finite element package ABAQUS Version 5.3 was used to calculate the stresses, electrical
field strengths,
and electrical displacements
at the
nodes in the model.

elcctrlc i~cid I pos~t~\e cluclr-lc tick. A ncgat~\~

electric held. 13hich was opposite to the poling direction. was produced by the application of a positive
electric potential on the edge AH. Mechanical load
was produced by the application of tither ;I normal
tensile stress or a normal tensile strain on the edge AB
along the Y-direction.
The PZT-SH
material
properties.
ah given in
Table I. were used for the analysis. For various
loading conditions.
the angular distributions
of the
radial and tangential stresses were calculated for the
four node sets defined by points along the circular
arcs OP, OQ. OR and OS at an interval of 7.5 [Fig.
2(c)]. The stresses were normalized
by multiplying
them by the factor L :2(, , where Iand N are the radial
distance of the point from the crack tip and crack
length. respectively.
The angular distributions
of
stresses along the circular arc defined by the radius
OQ were always found to be different from those for
the circular arcs defined by the radii OP. OR and OS.
For low values of the electrical to mechanical load
ratio. no substantial difference in the angular distributions of the stresses was found for the circular arcs
OP. OQ and OS. For high values of the electrical to
mechanical load ratio, the angular distributions
of
the stresses differ for the different circular arcs. and
only the angular distributions
of the stresses for the
circular arc defined by the radius OP were found to
be accurate and the same as those predicted by the
theoretical analysis of Sosa [7] and Pak [8]. Hence. a
very fine mesh near a crack tip is required to accurately calculate stresses near the crack tip in a
piezoelectric
material. The angular distributions
of
stresses reported in this paper are for the circular arc
defined by the radius OP which is nearest to the crack
tip and at a distance of I.Ox 10 of the crack length.
4. I. Mrchunicul

4. RESULTS

AND DISCUSSION

The double edge-notched

specimen, as shown in
Fig. 1, was used to study the stress distributions at the
crack tip under electrical and mechanical loads. In
order to study the stress distributions
at the crack tip
due to electrical loads, negative or positive electric
potential was applied to the top edge AB (Fig. 1). As
the bottom edge CD was grounded, negative electric
potential on AB produced electric field which was
parallel to the poling direction. We call this type of
Table

I. Material properties for PZT-5H

piezoelectric ceramfc [8]

Parameter
C,,
(I
(22

12.6 x IO N/m
5.5 x ION/II?
12.3 x 10Nhn2

('I,

11.1x 10'"N:im~

( J1
<I,
(3;
115

3.53 x 1ONh
-6.5 C/m
23.3 C/m
17.0 Cm

(1,

(17

151X IO C/Vm
130 x IO C,Vm

lad

In order to study the stress distributions

under pure
mechanical load, a tensile stress D of 100 N/m was
applied along the Y-direction or along the poling
direction and perpendicular
to the crack surface. For
plane strain conditions,
the resulting angular distributions of the radial and tangential components
of
the normalized stresses cr,,,?,grr and grOare shown in
Fig. 3. The stresses are found to be symmetric about
the crack face. These stresses are normalized by the
factor $\I:. where r and u are the radial distance of
the point from the crack tip and crack length. respectively. These stresses are similar to those calculated
theoretically by Sosa for PZT-4 (Fig. 2 in Ref [7]) and
Pak for PZT-5H (Fig. 7(a) in Ref. [S]) except a
normalization
factor due to the finite size of the DEN
specimen. Sosas [7] and Paks [8] calculations
are
valid for a finite size crack in an infinite piezoelectric
medium. The normalization
factor for the DEN
model shown in Fig. 1 is K,/Jn. where K,is the mode
I stress intensity factor of the DEN specimen for a
homogeneous isotropic elastic material. For the DEN
specimen shown in Fig. 1, K, 2 1.47 (see Appendix).

KUMAR and SINGH:

CRACK PROPAGATION

IN PIEZOELECTRIC

MATERIALS

179

1.0

0.5

0.0

Angle
Fig. 3. Angular variation of the stresses eoO, err and em at the crack tip for applied tensile stress under
plane strain condition. r, a and angle 0 are defined in Fig. 1. Stress (Twas applied normal to the line AB
in the positive Y-direction. The stress distributions shown are similar to those calculated theoretically by
Sosa [7] (Fig. 2) and Pak [8] [Fig. 7(a)], except a normalization factor due to the finite size of the specimen.
The normalization

factor in this case is K,/,/K where K, is the mode I stress intensity

specimen with a homogeneous
isotropic elastic property.

A +ve stress is tensile and a -ve stress is compressive throughout

this paper. The hoop or the tangenfor the crack
tial stress a,, which is responsible
propagation
in a brittle piezoelectric material is maximum at 8 = O, showing that the crack has a tendency
to propagate in a straight line under a purely mechanical load. Also, as the effective (normalized) maximum hoop stress aoO is equal to mode I stress
intensity factor, the electromechanical
coupling in
piezoelectric
material has no effect on the crack
propagation
under pure mechanical loads.
Figure 4 shows the angular distributions
of the
radial and tangential components
of the normalized
stresses coo, or, and crR under plane strain conditions
for the applied normal tensile strain in the Y-direction. The calculated stresses were normalized by the
equivalent stress (C,, x s) corresponding
to the applied strains s. Comparison of the stress distributions
in Figs 3 and 4 show that due to the electromechanical coupling effect, the stresses at the crack tip due
to the applied strain are slightly lower than those due
to the equivalent applied stress.
Figure 5 shows the angular variations of the electrical displacement
components
D, and D, at the crack
tip due to a tensile stress e for plane strain conditions.

factor

of the DEN

These distributions
are similar to those calculated
theoretically by Sosa for PZT-4 (Fig. 3 in Ref [7]) and
Pak for PZT-5H (Fig. 8(a) in Ref. [S]) except for a
normalization
factor due to the finite size of the DEN
specimen. The normalization
factor in this case is the
stress intensity factor K,/JTCfor mode I fracture of
the DEN specimen with a homogeneous
isotropic
elastic property. The maximum value of the hoop
component of the electrical displacement D, occurs at
about 0 % 110 where the tangential component D, is
zero. The substantial difference in the angular variations of the electrical displacement
components
D,
and D, at the crack tip has been found between
applied stress and applied strain.
4.2. Electrical load
In order to study the stress distributions under pure
electrical load, an electric potential of - 100 V was
applied on the face AB and the lower face CD was
grounded (Fig. 1). Hence, the applied electric field
was in the same direction as the direction of poling.
Figure 6 shows the angular distributions
of the radial
and tangential stresses e,,, or, and crrUunder plane
strain condition. The stress distributions
are found to
be symmetric about the crack surface. These stresses

I so

KUMAR

and SIIGGH:

CRACK

PROPAGA1103

IN PlCZOELECfRIC

MAIERIALS

Angle
Fig. 4. Angular variation of the stresses oRU, (r,, and or0 at the crack tip for applied tensile strain under
plane strain condition. I, a and angle 0 are defined in Fig. I. Normal strain along Y-direction was applied.

The calculated stresses were normalized by the equivalent stress 0 ( = C,, x s) corresponding to the applied
strain s.
are normalized by the factor bx$ x lo*, where D is
the equivalent electrical displacement
corresponding
to the applied electric field E, i.e. D = t,) x E. The
stress distributions
shown in Fig. 6 are similar to
those calculated
theoretically
by Sosa for PZT4
(Fig. 4 in Ref. [7]) and Pak for PZT-5H (Fig. 7(b) in
Ref. [8]) except for a normalization
factor due to the
finite size of the DEN specimen.
For a positive
applied electric field, the hoop stresses developed are
compressive for all 8. For a negative applied electric
field, the stresses are the inverse or negative of the
stress values shown in Fig. 6 and the hoop stresses
developed
in this case are tensile for all 8. One
important
point to note here is that the maximum
hoop stress is at an angle which is about B % 82
where the shear stress, CT,(,
= 0. Hence, a negative
applied electric field (i.e. electric field applied opposite
to the poling direction), creates a tensile hoop stress
which is maximum at an angle 82 from the crack
plane. This will tend to propagate the crack obliquely
in that direction.
Figure 7 shows the angular variation of the electrical displacement
components
D, and D, at the crack
tip for a positive applied electric field E under plane
strain condition.
The electrical displacement
components are normalized
by the factor iJ$ where,
D = tj, x E. The distributions
shown in Fig. 7 are

similar
PZT4
8(b) in
to the

to those calculated theoretically

by Sosa for
(Fig. 5 in Ref [7]) and Pak for PZT-5H (Fig.
Ref. [S]) except for a normalization
factor due
finite size of the DEN specimen.

4.3. Combined mechanical and electrical loads

In order to study the stress distributions
at the
crack tip due to the combined mechanical and electrical loads, positive or negative potential was applied
at the top edge AB (Fig. 1) as the electrical load and
a normal tensile stress (cr.) or a normal tensile strain
(s) was applied along Y-direction on the edge AB as
the mechanical load. Four cases of combined mechanical and electrical loads were considered.
Case I

E/o = 0.021 Vm/N

Case II

E/o = 0.769 Vm/N

Case III :

E/u = 3.846Vm/N

Case IV :

E/cr = 7.692Vm/N.

In all four cases, the applied normal tensile stress

u was kept constant at 100 N/m2 and the magnitude
of the electric field was varied. In the case of applied
strain, the normal tensile strain was kept constant at
100/C,, The proportion
of the electrical load, in the
combined load, increases from case I to IV. These

KUMAR and SINGH:

CRACK PROPAGATION

values of E/o are equivalent to the values of D,/a,,

used in Paks paper (Fig. 9 in Ref. [S]) to study the
effects of the combined loadings on the stress distributions at the crack tip in PZT-SH. For each of the
four cases, the effects of the direction of the electric
field (positive electric field and negative electric field)
on the results are also studied. When normal tensile
strain (s) is applied as mechanical load, the equivalent
stress (C,, x s) corresponding
to the applied normal
strain is used to normalize the calculated stresses. In
subsequent figures, the stress distributions
under the
plane strain conditions are reported.
4.3.1. Applied stress. Figure 8 shows the angular
distributions
of the hoop stresses at the crack tip
because of the combined mechanical and electrical
loads under positive applied electric fields. Normal
tensile stresses are applied as the mechanical loads.
The hoop stresses go0 shown in Fig. 8 are of importance because these are responsible
for the crack
propagation.
These distributions
(Fig. 8) are similar
to those in Fig. 6 of Ref. [7] except a normalization
factor due to the finite size of the DEN specimen.
Hence, for a positive applied electric field, the crack
opening stress is maximum at tI = 0 and the crack
will have a tendency to propagate in a straight line.

IN PIEZOELECTRIC

MATERIALS

181

The electric field has no effect on the magnitude of the

maximum
stress responsible
for the crack propagation. Also, with the increase in the proportion
of
electrical load (for cases III and IV), compressive
stresses are developed for some values of 0 although
the remotely applied stress is tensile.
Figure 9 shows the angular distributions
of the
hoop stresses at the crack tip due to the combined
mechanical and electrical loads for a negative applied
electric field. Normal tensile stresses are applied as
the mechanical loads. The distributions
of the hoop
stress co0 shown in Fig. 9 are similar to those in Fig. 7
of Ref. [7] and in Fig. 9 of Ref. [8] except for a
normalization
factor due to the finite size of the DEN
specimen. As shown in Fig. 9, for lower values of
electrical loads (keeping mechanical stress constant)
(cases I and II), the hoop stress is maximum at 13= O,
and the crack will have a tendency to propagate in a
straight line. For higher values of the electrical loads
(cases III and IV), the hoop stress is no longer
maximum at 0 = 0 and so the crack will tend to turn
away from its path. With the increase in the electrical
load, keeping mechanical stress constant, the magnitude of the maximum hoop stress increases and the
angle 0 for maximum
hoop stress also increases.

4.0

-D

9
___fJ___D
r

-2.0

-4.0

I
0

I
60

I
la

I
180

I
240

I
300

Angle
Fig. 5. Angular variation of the electrical displacement components
D, and D, at the crack tip for applied
tensile stress under plane strain condition.
r, a and angle 0 are defined in Fig. 1. Stress cr was applied
normal to the line AB in the positive Y-direction. These electric displacement
distributions
are similar to
those calculated theoretically by Sosa [7] (Fig. 3) and Pak [8] [Fig. 8(a)], except for a normalization
factor
due to the finite size of the specimen. The normalization
factor in this case is K,/ Jn where K, is the mode
I stress intensity factor of the DEN specimen with homogeneous
isotropic elastic property.

I
360

I81

KUMAR

and SINGH:

CRACK

PROPAGATION

Hence, for larger electric fields, the crack will deviate

more and more from its straight line path. This
behavior of the hoop stress distribution
for higher
electrical loads can be explained with the help of
Fig. 6. For a negative applied electric held, the
magnitudes of the hoop stress will be negative of the
stresses shown in Fig. 6 and the hoop stress is
maximum
for 0 # 0. McHenry
and Koepke [24]
experimentally
observed the crack skewing or turning
in the piezoelectric
material under applied electric
loads.
In general, for low (positive) electrical to mechanical load ratios (cases I and II), an increase in the
applied stresses has not much effect on the stress
distributions
at the crack tip. Whereas,
for high
electrical to mechanical load ratios (cases III and IV),
an increase in the negative applied electric fields,
keeping
mechanical
load
constant.
drastically
changes the stress distributions
at the crack tip and
also it increases the tendency for the cracks to deviate
from its path and propagate at an oblique angle. But,
if we consider the energy release rate criteria for the
crack propagation
(taking into account both the
electrostatic
and strain energies) under combined
electrical and mechanical
loads, then the energy

IN Ilt.%OELE~

fRIC MA I FRIAI.>

rcleasc rates arc maximum lot- if = 0 IOI- hoth posi

tive and negative applied electrical tields and hence
the crack will have a tendency to propagate
in a
straight line [33]. Wang and Singh [3;?. 231 have
experimentally
observed similar phenomenon
for the
PZT. They used Vickers indentation
technique to
create cracks normal to the poling direction in PZT
and measured the length of the indentation cracks for
various magnitudes
of both negative and positive
applied electric tields for a constant indentation load.
They observed that cracks introduced by indentation
propagate less under a positive applied electric field.
whereas the crack propagation
is enhanced under a
negative applied electric field [Fig. lO(a. b)]. In both
cases.

the cracks

were

found

to propagate

in straight

They also reported that such an effect becomes

more profound with increasing electric field strength
and decreasing mechanical loading. Same specimens,
one each of PZT-EC-64 and PZT-EC-65. were used
in Wang and Singhs studies reported in Ref. [23].
The crack lengths under pure mechanical loads (i.e.
with zero electric held) shown in Fig. lO(a. b) for
PZT-EC-65
were different
because of using two
different specimens (of the same material) in Wang
and Singhs studies reported in Ref. [22].
lines.

0.2

---O---

rr

.0-0 9

0.1

0.0

-0.1

-0.2

-0.3

-0.4
0

60

I.20

180

240

300

Angle
Fig. 6. Angular variation of the stresses ooO, u,, and urO at the crack tip for a positive applied electric field
under plane strain condition. r, a and angle 6 are defined in Fig. 1. The stresses were normalized by the
equivalent electrical displacement D ( =cxj x E) corresponding
to the applied electric field E. The direction
of the applied electric field E was the same as the direction of poling. The stress distributions
shown are
similar to those calculated
theoretically
by Sosa [7] (Fig. 4) and Pak [S] [Fig. 7(b)], except for a
normalization
factor due to the finite size of the specimen.

360

KUMAR

2.0

and SINGH:

CRACK

PROPAGATION

IN PIEZOELECTRIC

MATERIALS

183

1.0

0.0

-1.0

Angle
Fig. 7. Angular variation of the electrical displacement components
D, and D, at the crack tip for positive
applied electric field under plane strain condition. r, a and angle f3 are defined in Fig. 1. The electrical
displacement
components
were normalized
by the equivalent
electrical displacement
D ( =E~~x E)
corresponding
to the applied electric field E. The direction of the applied electric field E was the same
as the direction of poling. These electric displacement
distributions
are similar to those calculated
theoretically
by Sosa [7] (Fig. 5) and Pak [8] [Fig. 8(b)], except for a normalization
factor due to the finite
size of the specimen.

4.3.2. Applied strain. Figure 11 (a-c) shows the

angular distributions
of the radial and tangential
stresses at the crack tip because of the combined
mechanical and electrical loads for positive applied
electric fields. A normal tensile strain was applied as
the mechanical
load. The calculated
stresses were
normalized by dividing the stresses by the equivalent
stress (C,, x s) corresponding
to the applied strain, s.
For lower values of the electric field (case I), the hoop
stresses are tensile and the angular distributions
of
the stresses are similar to those under applied pure
mechanical stress (i.e. Fig. 3). But, with the increase
in the electrical field (cases II-IV), the angular distributions of the stresses change drastically and also the
stresses developed at the crack tip are compressive in
nature, although a tensile normal strain perpendicular to the crack surface is remotely applied on the
specimen.
Figure 12 (as) shows the angular distributions
of
the radial and tangential
stresses at the crack tip
because of the combined mechanical
and electrical
loads for negative applied electric fields. Normal
tensile strains are applied as the mechanical loads. As
shown in Fig. 12 (ax), an increase in the negative
applied electric field with constant strain or displace-

ment drastically
increases the magnitudes
of the
stresses developed at the crack tip. As the hoop stress
is always maximum at 0 = O, the crack will always
propagate along its straight line path under negative
electric field and applied strain. Hence, an increase in
the negative applied electric field with a constant
normal tensile strain will cause an increase in the
crack propagation
rate. Zhang and Hack [lo] have
analytically
found similar behavior for Mode III
crack propagation
under combined loading with applied strain or applied stress. They have found that in
the case of remotely applied stress, the stress field at
the crack tip depends only on the remote mechanical
loading and the remote electric field has no effect on
the crack propagation.
But, under conditions
of a
remotely imposed strain, the electric field can exert a
significant influence on the crack propagation.
These
observations
are reasonable considering that an imposed strain situation may be similar to a constrained
case whereas the constant stress situation may be an
unconstrained
case.
For an applied electric field of 1 MV/m magnitude
which is the typical value of the coercive field for the
piezoelectric
materials, the corresponding
values of
the applied stresses for cases I-IV are 47.62, 1.30,0.26

KUMAR and SINGH:

184

CRACK PROPAGATION

and 0.13 MI%, respectively. Although. the equivalent

applied strain would be low in magnitude for cases III
and TV, the high stress intensity factors as shown in
Figs 1 l(b. c) and 12(b, c) can result in a very high
stresses at the crack tip.
4.4. PI~micuI interpretation
The deformed meshes of the DEN specimen for
different types of combined electrical and mechanical
loads are shown in Figs 13 (a-d) and 14 (a-d).
Figures 13 and 14 are for the cases I and IV,
respectively. of the combined electrical and mechanical loads. The displacements
shown in the figures are
scaled to produce the deformed
shape. The scale
factor is one-tenth of the maximum model dimension.
For comparison,
the difference between the Y-displacement of the node at right hand side of the upper
crack surface and the Y-displacement
of the node at
right hand side of the lower crack surface, Cc,is shown
in Fig. 13(a). The crack opening d is 3.4252 x 10m9
whereas the width of the specimen is 1.O. Due to the
scaling of the displacements.
the nodes near the crack
tip in Fig. 14(c) are shown outside the left hand
boundary of the DEN model which is not really the
case.

1% PlEZOEL~.CTRIC MA IERIALS

The deformation
01 the DEN specmien
undci
different types of combined electrical and mechanical
loads (Figs 13 and 14) can be explained with the help
of Fig. ISa-c). In Fig. 15(a). OE and OE are the
upper and lower crack surfaces, respectively.
The
distance EE
in the undeformed mesh is zero. Due
to the increase in the electric flux density in the
untracked region in the neighborhood
of the crack
tip, the untracked region has a tendency to elongate
(contract)
more in the Y-direction
relative to the
crack region (where the electric flux density is almost
zero) for a positive (negative) applied electric field.
This results in compressive
stress in the untracked
region and tensile stress in the cracked region for the
positive applied electric field. The reverse stresses will
develop under a negative electric field. In general, the
crack will experience
a compressive
stress and a
tensile stress ahead of the crack for the positive and
negative electric fields, respectively. Hence. a positive
electric field will tend to slow the crack growth and
a negative electric field will tend to enhance the crack
growth.
For case I (low electrical to mechanical load ratio),
the tensile stress is high enough to overcome the
compressive
effect of the positive applied electric

--o-

0.021 vmm

---o---O.769
1.0

VrnlN

-A-

3.846 VndN

---A---

7.692 Vm/N

0.5
go
D
&it a
L

0.0

-0.5

60

l20

180

240

300

Angle
Fig. 8. Angular variation of the hoop stress uBBat the crack tip for different electrical to mechanical load
ratios (E/a) for plane strain conditions. r, a and angle 0 are defined in Fig. 1. A tensile stress a was applied
normal to the line AB in the positive Y-direction. The direction of the applied electric field E was the
same as the direction of poling. The plotted hoop stresses uoOare similar to those calculated theoretically
by Sosa [7] (Fig. 6) except for a normalization factor due to the finite size of the specimen.

30

KUMAR

and SINGH:

CRACK

PROPAGATION

IN PIEZOELECTRIC

MATERIALS

185

E/O

2.0

0.021 vtnm

---n---O.769VmlN

-0.5

60

120

180

240

3cO

360

Angle
Fig. 9. Angular variation of the hoop stress bgHat the crack tip for different electrical to mechanical load
ratios (E/u) for plane strain conditions. r, a and angle 0 are defined in Fig. 1. A tensile stress 0 was applied
normal to the line AB in the positive Y-direction. The direction of the applied electric field E was opposite
to the direction of poling. The plotted hoop stresses ooR are similar to those calculated theoretically
by
Sosa [7] (Fig. 7) and Pak [8] (Fig. 9), except for a normalization
factor due to the finite size of the specimen.

field, hence there is crack opening in all four types

of the electrical (positive and negative) and mechanical (stress and strain) loads [Fig. 13(a-d)]. But,
for case IV (high electrical to mechanical load ratio),
crack closure occurs in the case of a positive applied
electric field although
a remote tensile strain is
applied.
A comparison
of Fig. 13(a and b) shows that for
the same applied tensile stress, the crack opens more
under a negative applied electric field than under a
positive applied field. This can be clearly explained
with the help of Fig. lS(b and c). Figure 13(c and d)
shows the similar behavior for a remotely applied
tensile strain. Comparison
of Figs 13(a) and 14(a)
shows that the crack closure occurs with an increase
in the positive applied electric field under a constant
tensile stress. For a given tensile strain, the crack
opening increases with an increase in negative applied
electric field [Figs 13(d) and 14(d)].
For case IV (high electrical to mechanical
load
ratio), a compressive
stress field (in Y-direction)
is
developed near the crack tip in the untracked region
[F in Fig. 15(a)] when a remote positive electric field
is applied, and for this reason the effect of a remotely
applied tensile stress is reduced
near the crack
(Fig. 8). But when the polarity of the electric field is

reversed [Fig. 14(b)], there is a tensile stress near the

crack tip in the untracked region and the effect of the
remotely applied tensile stress is enhanced near the
crack tip (Fig. 9). In this case, there is an overlap of
the crack surfaces, i.e. the upper crack surface OE+
[Fig. 15(a)] goes under the lower crack surface OEdue to the local compressive stress field in the cracked
regions G and H near the crack tip. The upper
diagram in Fig. 14(b) clearly shows the effects of
tensile and compressive stress fields in the untracked
and cracked regions, respectively. The deformed line
marked by two arrows, was parallel to the crack in
the undeformed mesh. With an increase in the electrical to mechanical load ratio, the upper (lower) crack
surface will dip (move up) more and more (i.e. the
angle between the crack surfaces and the horizontal
direction or the line 6) = O, will increase with an
increase in the electrical to mechanical load ratio) and
this might be the reason for the maximum hoop
stresses to be at an oblique angle as shown in Fig. 9.
It is also apparent from Fig. 9 that the angle 0 for the
maximum hoop stress increases with an increase in
the electrical to mechanical load ratio.
For a remote tensile strain and a high positive
applied electric field (case IV), the untracked region
[F in Fig. 15(a)] tries to expand in the vertical

IX6

KUMAR

and SINGH:

(RACK

PROPAGATION

direction: but as the expansion in the vertical direction is restricted (due to a fixed applied strain in that
direction). the region F tries to expand in the horizontal direction due to Poisson effect and it forces the
untracked
regions G and H to expand more in the
vertical direction. As the untracked regions G and H
have free boundaries
OE and OE , respectively,
hence it results in an overlap of the upper and lower
crack surfaces and the crack tip is pushed to the left
[Fig. 14(c)]. The effect of these phenomena
results in
an increase in compressive stress field at the crack tip
[Fig. 1 I(c)]. Therefore, the mechanical boundary condition defined by equation (1) (i.e. boundary free of
surface traction) is not valid for the last two conditions of the combined electrical and mechanical
loads [i.e. (1) stress and negative electric field; and (2)
strain and positive electric field] when the applied
electric field is very high as compared to the mechanical load. For these two cases, the mechanical boundary conditions for the crack surfaces need to be modified and the following equation for the displacement
field at the crack surface should be used (Fig. 1)
u,(x,b-)=u,(.x,b-),

x>(c

-a)

(16)

where, bm and b+ denote the lower and upper crack

surfaces, respectively, and u,.(x, y) is the Y-displacement at (x, x). The effects of the modified mechanical

II\. P1EZOEL~.CTR1C MA TF.RIALS

boundary condition on the ~trcs~ distribution

:II ~hc
crack tip will be reported somcwherc else [%I]
For case IV (i.c. for high clcctrical 10 mechamcal
load ratio) under an applied strain. as the negative
applied electric field contracts the length in the \,ertical direction.
a tensile stress is developed
in the
untracked region because of the applied tensile strain.
But, in the presence of a negative applied electric
field, the untracked
regions (G and H) also try to
contract in the vertical direction and as they have fret
boundaries
[OE and OE . respectively.
in Fig.
15(a)], the edges BE+ and DE are pulled towards B
and D, respectively.
in the vertical direction [Fig.
14(d)]. The combined effect of these two phenomena
is an increase in the tensile stress field near the crack
tip [Fig. 12(c)].
Generally. the piezoelectric ceramics have nonlinear electro-mechanical
behavior even under moderate
electric field and stress. The fields at the crack tip are
high enough to rotate the polar direction. The crack
tip stresses reported in this paper do not consider this
nonlinear electro-mechanical
effect of piezoelectric
ceramics. Also, we have analyzed the stresses developed at the crack tip due to applied mechanical
and/or electrical loads from a macroscopic
point of
view. In order to accurately predict the stresses at the
crack tip in a piezoelectric
material.
the residual

-e-

PZT-EC-64

L231

- - -A- - - PZT-EC-65 [23 I

-O-

Electric

PZT-EC-65

Field (V/mm)

Fig. IO(a) (See cup/ion

oppoyirc).

[221

_-_.:
.._...1

KUMAR and SINGH:

lb)
190

CRACK PROPAGATION

IN PIEZOELECTRIC

.-

170

.-

.- A

/O-

150

PZT-EC-64

1231

- - -A- - - PZT-EC-65

1231

PZT-EC-65

[221

-O-

Ho-

MATERIALS

--

187

--O

Electric

Field W/mm)

Fig. 10. Effect of the direction and magnitude of electric field (with a constant indentation load) on the
crack propagation in poled PZT [22,23]. Vickers indentation technique was used to create cracks which
were perpendicular to the direction of poling. Indentation load was fixed at 1.2 kg. (a) Positive applied
electric field; and (b) negative applied electric field. The direction of positive applied electric field was the
same as the direction of poling.

stresses, which are developed because of the poling

process, needs to be considered in the analysis. These
residual stresses are anisotropic
in nature [21,22].
Further research needs to be done to consider the
effects of mechanical
and electrical loads on the
changes in the domain orientations
at the crack
surfaces as reported experimentally
by Mehta and
Virkar [26], because the change in domain orientations will cause stresses at the crack tip which can
not be accounted for by the macroscopic
analysis.
Some of these studies are currently underway.
5. CONCLUSIONS
The two-dimensional
finite element technique was
used to study the crack propagation
in piezoelectric
materials. A double edge-notched
specimen was modeled to determine
the radial and tangential
stress
distributions
at the crack tip under the combined
mechanical
and electrical loads. The crack surface
was assumed to be perpendicular
to the poling direction. Stresses and electric fields were applied along the
poling direction. The calculated stress distributions

were in conformity
with those predicted
by the
theoretical analyses of Sosa [7] and Pak [8]. It was
observed that the electromechanical
coupling had no
effect on the crack propagation
under pure mechanical loads. Under applied electric field with no mechanical loading, the hoop stress was maximum
at
about 82 from the plane directly ahead of the crack
surface. For combined
mechanical
and electrical
loads, compressive
stresses were found to develop
around the crack tip for high electrical to mechanical
load ratios for positive applied electric fields. The
stress intensity factor at the crack tip was found to
increase with an increase in the electrical to mechan
ical load ratio for a negative applied electric field
which was in agreement with the experimental
find-,
ings of Wang and Singh [22,23]. For high electrical
to mechanical
load ratios in the case of negative
applied electric fields, the crack tended to deviate
from its path and propagated in an oblique direction
when the maximum stress criteria was used for the
propagation
of crack.
For combined loads, the stress distributions
at the
crack tip under applied stress were significantly differ-

Inn

KUMAR

and SINGH:

CRACK

PROPAGATION

IN PIEZOELECTRIC

ML\IERIALh

E / o = 0.021 Vm/N

-0

nn

----G---o
-A-

rr
(T

rn

Angle

0))

E I o = 0.769 Vm/N

-25

I
0

I
60

I
120

180

Angle
Fig. 1I(a and b) (See caption opposite).

I
240

I
300

360

CRACK PROPAGATION

KUMAR and SINGH:

IN PIEZOELECTRIC

189

MATERIALS

E / CT= 7.692 Vm/N

-150

-200

-250

120

60

l&I

I
360

300

240

Angle
Fig. 11 Angular variations of the stresses uge, r~l, and or0at the crack tipfor dzfleerentelectrical to mechanical
load ratios (E/a) for plane strain conditions. Normal tensile strain along Y-direction was applied. The
calculated stress was normalized by the equivalent stress CJ(= C,, x s) corresponding to the applied strain
s. r, a and angle 0 are defined in Fig. 1. The direction of the applied electric field Ewas same as the direction
of poling. (a) E/a = 0.021 Vm/N; (b) E/a = 0.769 Vm/N; and (c) E/u = 7.692 Vm/N.
1.5

VmlN

E/0=0.021

n-0

,? ?

??
,Ii

1.0

CJ

rr

-A-

(J
I%

0.0

4
tA

-0.5
0

I
60

I
120

I
180

Angle
Fig. 12(a) (See caption overleaf).

I
240

I
300

I
360

190

(h)

KUMAR
pfJ

and SINGH:

(RACK

PROPAGATION

-10

IN PIEZOEL.E(TRIC

MA IERIALS

E / cs = 0.769 VmlN

60

180

120

240

300

360

Angle

250

E I o = 7.692 Vm/N

150

50

-A-

-50

-100

I
60

120

180

Angle
Fig. 12(b and c) (See

caprion

opposite).

I
240

I
30

1
360

KUMAR and SINGH:

CRACK PROPAGATION

IN PIEZOELECTRIC

MATERIALS

191

(4

Fig. 13(a) (See caption on page 194).

ent from those under applied strain. A negative

applied electric field with a tensile strain propagated
the crack very rapidly. For high electric fields, the
analysis of the deformed
mesh showed that the
mechanical boundary condition at the crack surfaces

defined by equation (1) (i.e. crack surfaces to be free

of surface traction) was not valid for the two cases of
the combined
electrical and mechanical
loads: (i)
applied stress and negative applied electric field; and
(ii) applied strain and positive applied electric field.

Fig. 12. Angular variations of the stresses ergs, o,, and o,,) at the crack tip for different electrical to
mechanical load ratios (E/a) for plane strain conditions. Normal tensile strain along the Y-direction was
applied. The calculated stress was normalized by the equivalent stress o (= C,, x S) corresponding
to the
applied strain S. r, a and angle 0 are defined in Fig. 1. The direction of the applied electric field E was
opposite
to the direction
of poling.
(a) E/o = 0.021 Vm/N;
(b) E/a = 0.769 Vm/N;
and (c)
E/o = 7.692 Im/N.

KUMAR

and SINGH:

CRACK

PROPAGATION

IN PIEZOELECTRIC

Fig. 13(b) (See caption on page 194).

MATERIALS

KUMAR and SINGH:

CRACK PROPAGATION

IN PIEZOELECTRIC

MATERIALS

(cl

Fig. 13(c) (See caption overleaf).

193

KUMAR

and SIN<iH:

Fig. 13. The deformed

(E/g = 0.021 Vm/N).
and negative applied
applied tensile strain
view of the deformed
perpendicular
to the
magnification
factor

CRACK

PROPAGATlON

IN PIL%OELtCTRI~

MA I EKlALS

mesh of the DEN model for the case 1 of combined mechanical and electrical loads
(a) Applied tensile stress and positive applied electric field; (b) applied tensile stress
electric field; (c) applied tensile strain and positive applied electric field; and (d)
and negative applied electric field. The upper diagram in each case is the enlarged
mesh near the crack tip. The direction of positive applied electric field was
crack surface and was the same as the direction of poling. See text about the
of the displacement
shown. The relative displacement
d shown in Fig. 13 (a) is
3.4252 x 10M9 whereas the width of the specimen is 1.0.

KUMAR and SINGH:

CRACK PROPAGATION

IN PIEZOELECTRIC

(4

Fig. 14(a) (See caption on page 198)

MATERIALS

195

I%

KUMAR

and SINGH:

(RACK

PROPAGATION

IN PIEZOELECTRIC

(b)

Fig. 14(b) (See caption on page 198).

MATERIALS

KUMAR

and SINGH:

CRACK

PROPAGATION

IN PIEZOELECTRIC

(c)

Fig. 14(c) (See cupfion overleaf).

MATERIALS

197

198

KUMAR

CRACK

and SlhGH:

PROPAGATION

RIG MA-1 ERIALS

IN PIEZOELEC-I

(4

w///11111\\\u

L_._L_I__J

Fig. 14. The deformed mesh of the DEN model for the case IV of combined mechanical and electrical
loads (E/c = 7.692 Vm/N). (a) Applied stress and positive applied electric field; (b) applied stress and
negative applied electric field; (c) applied strain and positive applied electric field; and (d) applied strain
and negative applied electric field. The upper diagram in each case is the enlarged view of the deformed
mesh near the crack tip. The direction of positive applied electric field which was perpendicular
to the
crack surface was the same as the direction of poling. See text about the magnification
factor of the
displacement
shown.

KUMAR

and SINGH:

CRACK

PROPAGATION

IN PIEZOELECTRIC

MATERIALS

(4

I
u mxked

Region
I
I
F I

G
E+

0
E-

I
I
1 H
I
I

I
I
C

Lower Cracked

Region
D

Compressive

Compressive
stress

r-

stress
Tensile
Stress
. f

Untracked
Region

Cracked
Region

Crack Tip 1
(b) For Positive Applied
Electric Field

Tensile
Stress

r-

--I

.+

Cracked
Region

Untracked
Region

/c

-1

- Tip
Crack

(c) For Negative Applied

Electric Field

Fig. 15. Physical interpretation

of the deformation
of the DEN specimen under combined mechanical and
electrical loads, (a) DEN model. Stresses near the crack tip (b) for positive applied electric field and (c)
for negative applied electric field. Solid and dashed horizontal
rules denote the free and constrained
elongations,
respectively, in the Y-direction.

199

200

KUMAR

and SINGH:

CRACK

PROPAGATION

Ac,kno~rk~,d~cm~nt.s-Computational
assistance of the Cray
Y-MP8/864 computer at the Ohio Super Computer Center,
Columbus, Ohio, is gratefully acknowledged.
We would also
like to acknowledge
helpful discussions on this subject with
Dr Eric Leung of Hibbitt, Karlsson and Sorensen Inc. This
project was supported by the National Science Foundation
through a grant No. DMR-92021 Il.
REFERENCES
1. V. 2. Parton, Acta Astronuutica 3, 671 (1976).
2. W. F. Deee. Ph. D. thesis. Stanford Universitv. Stanford, Calif&ia
(1980).
3. S. Li. W. Cao and L. E. Cross, Mater. Lett. 10, 219
(1990).
4. Y. E. Pak, J. appl. Mech. 57, 647 (1990).
5. H. A. Sosa and Y. E. Pak. Int. J. Solids Strut. 26, 1
(1990).
6. H. Sosa, Int. J. Solids Struct. 28, 491 (1991).
7. H. Sosa, ht. J. Solids Strut-t. 29, 2613 (1992).
8. Y. E. Pak, ht. J. Fract. 54, 79 (1992).
9. B. Wang, Int. J. Engng Sci. 6, 781 (1992).
10. T. Y. Zhang and J. E. Hack, J. appl. Phys. 71, 5865
(1992).
11. T. Y. Zhang, Int. J. Fracr. 66, R33 (1994).
12. T. H. Hao and Z. Y. Shen, Engng Fract. Mech. 47,793
(1994).
13. M. L. Dunn, Engng Fract. Mech. 48, 25 (1994).
14. Z. Suo, in Smart Structures and Materials (edited by
A. V. Srinivasan),
pp. 14. ASME, New York (1991).
15. Z. Sue. J. Mech. Phys. Solids 41, 1155 (1993).
16. Z. Suo. C. M. Kuo. D. M. Barnett and J. R. Willis.
J. Mech. Phys. Solids 40, 739 (1992)
17. Y. Shindo, E. Ozawa and J. P. Nowacki, Appl. Electra.
Mater. 1, 77 (1990).
18. Y. Yamamoto,
H. Igarashi and K. Okazaki, Ferroelectries 50, 273 (1983).
19. R. F. Cook, S. W. Freiman, B. R. Lawn and R. C.
Pohanka,
Ferroelectrics 50, 267 (1983).
20. K. Okazaki, Ceram. BUN. 63, 1150 (1984).
21. H. Wang and R. N. Singh, to be published in the J. Am.
Ceram. Sot.
22. H. Wang and R. N. Singh, to be published in the J. Am.
Chem.

IN PIEZOELLCTRIC

Mc~~ir24. K. D. McHenry and B. G. Koepke. in Fra~~rurc,

cmic.~ ofCeramtc.s (edited by R C. Bradt, D. P. Hasselman and F. F. Lange). Vo 5~ on. 337 352. Plenum
Press, New York (I ski).
.
Ferrorlecrrtc~sSO, 179
25. H. Niitsuma and N. Chubachi.
(1983).
V. M. Chushko and S. P. Kovale\.
26. G. G. Pisarenko.
J. Am. Crram. Sot. 68, 259 (1985).
27. R. C. Pohanka, P. L. Smith and J. Pasternak, Ferroeiet rrics 50, 285 (1983).
28. K. Mehta and A. V. Virkar. J. Am. Ceram. Sot. 73, 567
(1990).
29. G. S. White. A. S. Raynes, M. D. Vaudin and S. W.
Freiman. J. Am. Ceram. Sot. 77, 2603 (1994).
30. H. Cao and A. G. Evans, J. Am. Ceram. Sot. 77, 1783
(1994).
31. C. S. Lynch, L. Chen, W. Wang, Z. Suo and R. M.
McMeeking, in Adaptive Structures and Materiuls Systems (edited by G. P. Carmen and T. E. Garcia),
Vol. 35, ASME AS (1994).
32. T. L. Anderson, Fracture Mechanics Fundamentals and
Applications. CRC Press, Boston (1991).
33. S. Kumar and R. N. Singh, Ceramics under Combined
Electrical and Mechanical Loads, to be published.
34. S. Kumar and R. N. Singh, Effects of the Mechanical
Boundary Condition at the Crack Surfaces on the Stress
Distribution at the Crack Tip in Piezoelectric Materials.
to be published.

APPENDIX

Calculation of the Mode I Stress Intensity

for the DEN Specimen

and

R. N. Singh,

Ceram. Trans. 43, 277

Factor K,

Assuming
the DEN specimen having a homogeneous
isotropic property of Youngs modulus E and Poisson ratio
v, we calculated the J-integral of the model under a normal
tensile stress o using ABAQUS version 5.3. Plane strain
condition
was assumed. For plane strain conditions,
the
relationship
between K, (for unit applied stress) and J is
given by

Sot.

23. H. Wang
(1994).

MATERIALS

K,=!