You are on page 1of 20

European Journal of Mechanics A/Solids 25 (2006) 230–249

Crack propagation in a four-parameter piezoelectric medium


Aldino Piva a , Francesco Tornabene b , Erasmo Viola b,∗
a Department of Physics, University of Bologna, Via Irnerio 46, 40126 Bologna, Italy
b DISTART, University of Bologna, Viale Risorgimento 2, 40136 Bologna, Italy

Received 26 April 2005; accepted 16 September 2005


Available online 4 November 2005

Abstract
The simplified three-parameter formulation of a piezoelectric medium proposed by Gao et al. [Gao, H.J., Zhang, T.Y., Tong, P.,
1997. Local and global energy release rates for an electrically yielded crack in a piezoelectric ceramic. J. Mech. Phys. Solids 45,
491–510] is extended to a four-parameter modified model in order to point out the features of a steadily propagating Griffith crack. It
is assumed that the crack is electro-elastically free and the medium is subjected to a generalized electro-mechanical loading applied
at infinity. The complete solution is provided under impermeable and permeable boundary conditions and results are presented in
order to show the main dynamical features.
 2005 Elsevier SAS. All rights reserved.

Keywords: Piezoelectric material; Crack propagation; Energy release rate; Electric field; Hoop stress

1. Introduction

Due to the property of inducing an electric self-polarization when subjected to a mechanical stress or undergoing a
strain when an electric field is applied, piezoelectric materials are widely used in many areas of science and technology.
In particular, piezo-ceramic transducers play a remarkable role in medical applications as well as in the fabrication
of devices to control intelligent structural systems. However, piezo-ceramics are easily affected by fracture at all
scales of electro-mechanical loads. Thence, knowledge of the structural performance and service lifetime of devices,
is strictly related to a complete understanding of fracture processes of piezoelectric components. Much effort has been
devoted to static problems of fracture in piezoelectric materials. Theoretical results include those of Parton (1976),
Deeg (1980), McMeeking (1989), Pak (1990), Suo et al. (1992), Sosa (1992), Dunn (1994), Hao and Shen (1994),
Park and Sun (1995), Zhang and Tong (1996), Gao (2000), Zhang et al. (2002) and Zhang and Gao (2004), among
others.
In the last two quoted papers one can find a thorough review of the literature. Few researchers have been engaged
in dynamic crack problems in piezoelectric materials. However, to the best of the authors’ knowledge, they studied
mostly the Mode-III of crack propagation. See, for example, Chen and Yu (1997), Chen et al. (1998), Chen and
Karihaloo (1999), Kwon et al. (2000), Kwon and Lee (2000) and Li et al. (2000). More recently Soh et al. (2002)

* Corresponding author. Tel.: +39 51 209 3510; fax: +39 51 209 3495.
E-mail address: erasmo.viola@mail.ing.unibo.it (E. Viola).

0997-7538/$ – see front matter  2005 Elsevier SAS. All rights reserved.
doi:10.1016/j.euromechsol.2005.09.002
A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249 231

studied the generalized plane problem of an impermeable Griffith crack propagating in an anisotropic piezoelectric
medium subjected to a far-field general electromechanical loading. Using the Stroh formalism they obtained a closed
form solution to electro-elastic fields in terms of complex variables and pointed attention to crack branching by the
maximum hoop stress criterion.
The result of experience concerning the solution to static or dynamic crack problems in piezoelectric materials is
that using the correctly stated constitutive equations often increases the difficulty of finding the closed form solution
and makes the interpretation of physical results difficult. Following this point of view and aiming to capture the
fundamental features of electro-mechanical interaction, Gao et al. (1997), in studying a strip saturation model for
an impermeable static crack in a piezoelectric material, introduced a simplified constitutive equation in which only
three material constants were taken into account. In addition, they assumed a null displacement component parallel
to the crack direction. The model has been introduced in order to explain discrepancies between theoretical and
experimental results concerning the effect of electrical loading on incipient crack propagation. The above model has
been re-proposed by Sih (2002), Li (2003) and more recently by Spyropoulos (2004).
In the present paper, the simplified three-parameter formulation of a piezoelectric medium considered by the above
quoted authors is extended to a four-parameter model without the constraint of null displacement component along
the crack direction. Under this model the dynamical problem of an electro-elastically free Griffith crack is studied
with the assumption that the medium is subjected to a remote generalized electro-mechanical loading. Using an
approach related to that proposed by authors (Piva and Viola, 1988; Piva et al., 2005) in studying elastodynamic
crack problems in orthotropic media, the field quantities are obtained by complex potentials and the complete solution
is provided under impermeable and permeable boundary conditions. Finally, results are discussed in order to give
almost a qualitative understanding of the electro-mechanical interaction under dynamical conditions.

2. The constitutive model and basic equations

The general constitutive equations for a piezoelectric medium are:


σij = cij kl γkl − ekij Ek , Dk = eij k γij + εkm Em , i, j, k = 1, 2, 3, (2.1)
where σij , γij , Di and Ei are the stress tensor, the strain tensor, the electric displacement vector and the electric field
vector, respectively. cij kl , eij k and εij stand for the elastic, piezoelectric and dielectric constants, respectively. In a
Cartesian coordinate system (x1 , x2 , x3 ) the equations of motion and the quasistatic Maxwell equations are:
∂ 2 ui
σij,j = ρ , Di,i = 0, Ei = −ϕ,i (2.2)
∂t 2
where ρ is the density of the material, t is time, ui are the components of the elastic displacement and ϕ is the
electric potential. In Eq. (2.2) a dependent variable followed by a comma and a letter index denotes the derivative
of the dependent variable with respect to the variable corresponding to the letter index. For a transversely isotropic
piezoelectric medium poled along the x3 axis, Eq. (2.1) admit the following matrix form:
c11 c12 c13 0 0 0 0 0 e31
      
σ11 γ11
 22   12 c11 c13 0 0 0   γ22   0 0 e31 
σ  c   
 E1
 
     
 σ33   c13 c13 c33 0 0 0   γ33   0 0 e 33 
 =
σ   0
 −   E2  , (2.3)
 23   0 0 c44 0 0   2γ23   0 e15 0 
   
 E3

     
 σ13   0 0 0 0 c44 0   2γ13   e15 0 0 
σ12 0 0 0 0 0 c66 2γ12 0 0 0
 
γ11
γ 
  22  
D1 0 0 0 0 e15 0  ε11 0 0 E1
    

 γ33 
 D2  =  0 0 0 e15 0 0 2γ 
+ 0 ε11 0   E2  . (2.4)
 23 
D3 e31 e31 e33 0 0 0   0 0 ε33 E3
 2γ13 
2γ12
232 A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249

In what follows the three-parameter simplified model introduced by Gao et al. (1997) is extended to a four-parameter
model which leads to the following simplifications of Eqs. (2.3) and (2.4)
m ∗ ∗ 0 0 0 0 0
      
σ11 γ11 e
σ   ∗ m ∗ 0 0 0 γ   0 0 e 
 22     22  
 E1
 
     
 σ33   ∗ ∗ m 0 0 0   γ33   0 0 −e 
 σ   0 0 0 n 0 0   2γ   0 −e 0  E2 ,
 =  +   (2.5)
 23     23  
 E3

     
 σ13   0 0 0 0 n 0   2γ13   −e 0 0 
σ12 0 0 0 0 0 ∗ 2γ12 0 0 0
 
γ11
γ 
  22  
D1 0 0 0 0 e 0  ε 0 0 E1
    

 γ33 
 D2  =  0 0 0 e 0 0 2γ  + 0
  ε 0   E2  (2.6)
 23 
D3 −e −e e 0 0 0   0 0 ε E3
 2γ13 
2γ12
where a ∗ means that the corresponding elastic constant does not enter into the model and m, n, e and ε are inde-
pendent constants. The introduction of the additional parameter n into Eq. (2.5) allows to take into account of elastic
anisotropy of the piezoelectric medium. In the three-parameter formulation all stress components depend only on the
same multiplicative constant m.
By focusing attention on plane strain problems in the x1 –x3 plane and avoiding any constraint on the displacement
components, the following generalized strain and stress vectors can be introduced:
Γ (1) = (u1,1 , u3,1 , ϕ,1 )T , Γ (2) = (u1,3 , u3,3 , ϕ,3 )T , σ 1 = (σ11 , σ13 , D1 )T , σ 2 = (σ13 , σ33 , D3 )T . (2.7)
The corresponding constitutive equations become:
σ 1 = AΓ (1) + BΓ (2) ,
(2.8)
σ 2 = BT Γ (1) + CΓ (2)
where:
m 0 0 0 0 −e n 0 0
     

A= 0 n
 e , B = n 0 0 , C= 0
 m e . (2.9)
0 e −ε e 0 0 0 e −ε
By substituting Eqs. (2.8) into Eqs. (2.2)1,2 , introducing the Galilean transformation:
x = x1 − ct, y = x3 , (2.10)
and renaming the displacement components such that u1 → u, u3 → v, it leads to the following system:
(A − aρc2 )Γ (1) T (1) (2)
,x + (B + B )Γ ,y + CΓ ,y = 0, a = diag(1, 1, 0). (2.11)

By introducing the vector Φ = (u,x , u,y , v,x , v,y )T , the system (2.11) can be transformed into:

Φ ,x + DΦ ,y = 0,
(2.12)
∇ 2 (ϕ − γ v) = 0
where γ = e/ε and
0 α α 0
 
 −1 0 0 0
D= . (2.13)
 
 α1 0 0 β
0 0 −1 0
A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249 233

Fig. 1. (λ(1) )2 and (λ(2) )2 vs. M2 .

The entries of the above matrix are:


n m
 n
α= 2
, β= , α1 = (2.14)
m(1 − M1 ) n̄(1 − M22 ) n̄(1 − M22 )
in which:
ρc2 ρc2
M12 = , M22 = , n̄ = n + γ e, m  = m + γ e. (2.15)
m n̄
The Mach numbers M1 and M2 are related to velocities:

m n̄
v1 = , v2 = . (2.16)
ρ ρ
The characteristic equation of the matrix D is:
λ4 + 2a1 λ2 + a2 = 0 (2.17)
with:
2a1 = α − αα1 + β, a2 = αβ. (2.18)
The analysis will be performed for the PZT-4 piezoceramic material and the following averaged values as introduced
by Sih (2002) will be used:
m = 6.93 × 1010 N/m2 , e = 13.64 C/m2 , ε = 57.4 × 10−10 C/Vm.
In addition, the value of the elastic constant C44 of the material, i.e. n = 2.56 × 1010 N/m2 , will be assumed for the
parameter n.
It may be shown that the matrix D has two pairs of imaginary eigenvalues in the range 0  c < v2 , two real
eigenvalues and a couple of imaginary eigenvalues for v2 < c < v1 and two pairs of real eigenvalues for c > v1 . The
elastodynamic crack problem will be studied in the subsonic regime 0  c < v2 where the matrix D has two pairs of
imaginary eigenvalues, as is indicated in Fig. 1, which are:
(1) (2)
λ± = ±ip1 , λ± = ±ip2 (2.19)
where:

1/2
1/2
p1 = a1 − a12 − a2 , p2 = a1 + a12 − a2 (2.20)
are positive constants for the above range of velocity.
234 A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249

It should be noted that the dynamical extension of the three-parameter model gives v1 ≃ v2 and two couples of
complex eigenvalues in the whole range 0  M2 < 1.
(1) (2)
The eigenvectors corresponding to λ+ and to λ+ are:
2 T
(j ) (j ) (j )
pj α pj α
f = g + ih = ,i , −ipj , 1 , j = 1, 2. (2.21)
α − pj2 α − pj2

The real basis h(1) , g(1) , h(2) , g(2) gives the matrix:
0 −p1 0 0
 
p
 1 0 0 0 
S = T−1 DT =  (2.22)


 0 0 0 −p2 
0 0 p2 0
where T = (h(1) , g(1) , h(2) , g(2) ). Using the transformation (2.22) into Eq. ( 2.12 )1 leads to the following Cauchy–
Riemann system:
Ψ ,x + SΨ ,y = 0 (2.23)
where Ψ is a four-dimensional column vector field defined by:
Φ = TΨ . (2.24)
The system (2.23) justifies the introduction of complex potentials:
y
Ω1 (z1 ) = Ψ1 (x, y1 ) + iΨ2 (x, y1 ), z1 = x + iy1 , y1 = ,
p1
y (2.25)
Ω2 (z2 ) = Ψ3 (x, y2 ) + iΨ4 (x, y2 ), z2 = x + iy2 , y2 =
p2
so that from Eq. (2.24) the components of the vector Φ can be represented in terms of potentials in Eqs. (2.25), as
follows:
2
(k)

Φj = Im fj Ωk (zk ), j = 1, . . . , 4. (2.26)
k=1
Making use of Eq. (2.12)2 , which specifies that the function ϕ − γ v can be written as the real (or imaginary) part of
an analytic function, one obtains the derivatives of the electric potential:
ϕ,x = γ Φ3 + Re Ω3 (z3 ),
(2.27)
ϕ,y = γ Φ4 − Im Ω3 (z3 )
in which z3 ≡ z = x + iy.

3. Representations of field quantities

Under the assumption that the piezoelectric medium is subjected only to the uniform electro-mechanical loading
at infinity:

σ 1 (∞) = (σxx ∞
, σxy , Dx∞ )T , ∞
σ 2 (∞) = (σxy ∞
, σyy , Dy∞ )T (3.1)
the vector Ω(z) = (Ω1 (z1 ), Ω2 (z2 ), Ω3 (z3 ))T can be represented as:
Ω(z) = Ω 0 + Λ(z) (3.2)
where Ω 0 is a constant vector and Λ(z) is an analytic vector vanishing at infinity.
Eqs. (2.26), (2.27) and (3.2) allow us to represent vectors Γ (1) and Γ (2) defined by Eqs. (2.7)1,2 in the following
matrix form:
Γ (1) = (Φ1 , Φ3 , ϕ,x )T = Γ∞
(1)
Γ (2) = (Φ2 , Φ4 , ϕ,y )T = Γ∞
(2)
   
+ Im EΛ(z) , + Im FΛ(z) (3.3)
A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249 235

where:
(1) (2)
f1 f1 0
 

 (1) (2) i
E =  f3 f3 0, F = E diag , k = 1, 2, 3 (p3 = 1) (3.4)

pk
(1) (2)
γf3 γf3 i
and
Γ (1)
∞ = Im[EΩ 0 ], Γ (2)
∞ = Im[FΩ 0 ]. (3.5)
It should be noted that, assuming a null rotation at infinity, Eqs. (3.5) may be rewritten as follows:
Γ (1) ∞ ∞ ∞
∞ = (γ11 , γ12 , −Ex ),
(2)
Γ∞ ∞
= (γ12 ∞
, γ22 , −Ey∞ ) (3.6)
where the elastic deformations and the electric field at infinity can be obtained from the uniform field equation (3.1)
through the constitutive equations, Eqs. (2.8) ( by using the appropriate notation).
The matrix forms of the stress vectors defined by Eqs. (2.7)3,4 are obtained as:
σ 1 = (σxx , σxy , Dx )T = σ ∞ σ 2 = (σxy , σyy , Dy )T = σ ∞
   
1 + Im GΛ(z) , 2 + Im HΛ(z) (3.7)
where:
mf1(1) − eγf4(1) mf1(2) − eγf4(2) e nf2(1) + n̄f3(1) nf2(2) + n̄f3(2) ie
   

G =  nf2(1) + n̄f3(1) (2)


nf2 + n̄f3
(2)
ie  , H= m
f4
(1)
m
f4
(2)
−e  (3.8)
   
(1) (2) (1) (2)
ef2 ef2 −iε −ef1 −ef1 ε
and
σ∞
1 = σ 1 (∞) = Im[GΩ 0 ], σ∞
2 = σ 2 (∞) = Im[HΩ 0 ]. (3.9)
From Eq. (3.2) the primitive of the vector Ω(z) takes on the form:
ω(z) = diag(zk )Ω 0 + λ(z) (3.10)
in which λ(z) is the primitive of Λ(z). The displacement vector defined as U = (u, v, ϕ)T can be obtained by direct
integration of Eq. (3.3)1 which gives the following representation:
U = U∞ + Im Eλ(z) , U∞ = Im E diag(zk )Ω 0 .
   
(3.11)

4. The elastodynamic crack problem. Preliminaries

Let an electro-mechanically free Griffith crack of length 2l propagate at constant velocity c < v2 along the positive
x1 -axis in the piezoelectric medium which is subjected to the remote load stated by Eqs. (3.6). The plane of reference
is identified by the coordinate system (x, y) centered in the middle of the crack (Fig. 2).
It is well known that one of the main subjects for debate in studying fracture behaviour of piezoelectric media is
the electric boundary conditions on crack surfaces. A lot of papers discussed about this topic and correspondingly
about the modeling of a crack in a piezoelectric medium. The models which receive the attention of researchers are:

(i) The impermeable crack model in which it is assumed that the permittivity inside the crack is negligible (zero)
with respect to that of the surrounding medium (Deeg, 1980; Pak, 1990; Suo et al., 1992; Sosa, 1992; Park and
Sun, 1995; Chen and Yu, 1997; Chen et al., 1998; Soh et al., 2002).
(ii) The permeable crack model in which it is assumed that the crack is transparent to the electric field. For this model
the crack behaves as if the permittivity in its interior is infinite (Parton, 1976; McMeeking, 1989; Dunn, 1994;
Kwon et al., 2000; Li et al., 2000).
(iii) The “exact crack model” in which it is assumed a finite permittivity inside the crack, and in addition the crack-
surfaces separation is taken into account (Hao and Shen, 1994; Zhang and Tong, 1996; Zhang and Gao, 2004).
236 A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249

Fig. 2. The propagating Griffith crack.

More recently, Landis (2004) has introduced a new crack model based on the so-called “energetically consistent
boundary conditions” in order to make clear why the exact crack model leads to discrepancies between the local and
global energy release rates (McMeeking, 2004).
In what follows, both the impermeable and permeable crack models still implemented in a lot of previous studies
will be carried out.
For both models the continuity condition of the generalized traction:
1
σ2 = σ∞ 

2 + HΛ(z) − HΛ(z) (4.1)
2i
is required along the x-axis which leads to:
σ+ −
2 (x, 0) = σ 2 (x, 0), |x| < ∞. (4.2)
Substituting Eq. (4.1) into Eq. (4.2) produces:
Λ̄(x) + = HΛ(x) + H Λ̄(x) − ,
   
HΛ(x) + H |x| < ∞ (4.3)

in which the definition: Λ̄(z) = Λ(z̄) was used. Considering that Λ(∞) = 0, Eq. (4.3) leads to:
Λ̄(x) = −HΛ(x),
H |x| < ∞. (4.4)

5. The impermeable crack

For a crack impermeable to the electric field, the boundary condition on the crack faces is:
σ+ −
2 (x, 0) = σ 2 (x, 0) = 0, |x| < l. (5.1)
By using Eq. (4.4) the above condition leads to the Hilbert problem:
Λ+ (x) + Λ− (x) = −2iH−1 σ ∞
2 , |x| < l (5.2)
which admits the following solution:
Λ(z) = i diag g(zk ) − 1 H−1 σ ∞
   
2 (5.3)
A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249 237

where g(zk ) = zk (zk2 − l 2 )−1/2 , k = 1, 2, 3. Substituting Eq. (5.3) into Eq. (4.1) leads to the generalized stress vector
σ 2 in the following form:
σ 2 = Re H diag g(zk ) H−1 σ ∞
   
2 . (5.4)
In particular, Eq. (5.4) indicates that ahead of the crack tip x = l one has:
σ 2 (x, 0) = x(x 2 − l 2 )−1/2 σ ∞
2 , x > l. (5.5)
The generalized stress vector σ 1 given by Eq. (3.7)1 becomes:
σ1 = σ∞ −1 ∞
   −1  ∞
1 − Re(GH )σ 2 + Re G diag g(zk ) H σ2 . (5.6)
The generalized vector intensity factor at the tip x = l defined as:

K(l) = lim 2π(x − l) σ 2 (x, 0) (5.7)
x→l+

is obtained from Eq. (5.5) as follows:



K(l) = πl σ ∞ 2 . (5.8)
The primitive λ(z) of Λ(z) is:
λ(z) = i diag h(zk ) − zk H−1 σ ∞
 
2 (5.9)
in which h(zk ) = (zk2 − l 2 )1/2 . Then, the generalized displacement, Eq. (3.11), becomes:
U = Γ (1) (2)
   −1  ∞
∞ x + Γ ∞ y + Re E diag h(zk ) − zk H σ2 . (5.10)
With respect to polar coordinates (r, θ ) centered at the tip x = l, the asymptotic dominant terms of the stress vector,
Eqs. (5.4) and (5.6), as well as of the displacement vector, Eq. (5.10), are:
   
exp(−iθk /2) −1 K(l)
σ 2 (r, θ ) = Re H diag √ H √ , (5.11)
ck (θ ) 2πr
   
exp(−iθk /2) −1 K(l)
σ 1 (r, θ ) = Re G diag √ H √ , (5.12)
ck (θ ) 2πr
   
θk −1 2r
U(r, θ ) = Re E diag ck (θ ) exp i H K(l) (5.13)
2 π
with the following notations:
sin2 (θ ) 1/2
 
2 −1 tg θ
ck (θ ) = cos (θ ) + , θk = tg . (5.14)
pk2 pk
Using Eq. (2.17) into Eqs. (3.4) and (3.8) one obtains:
K(l)
σ 2 (r, θ ) = L(θ ) √ , (5.15)
R(M2 ) 2πr
K(l)
σ 1 (r, θ ) = M(θ ) √ , (5.16)
R(M2 ) 2πr

K(l) 2r
U(r, θ ) = N(θ ) (5.17)
R(M2 ) π
where the entries of matrices L(θ ), M(θ ) and N(θ ) are given in Appendix A. The function R(M2 ) has the form:
R(M2 ) = p1 (n̄p12 − αγ e) ε m(α − p22 ) − αe2 p22 − p2 (n̄p22 − αγ e) ε
m(α − p12 ) − αe2 p12
   

+ α 2 e2 m
(p12 − p22 ). (5.18)
In Fig. 3 which represents the dimensionless real function R(M2 )/me2 it is shown that the function vanishes at
M2∗ ≈ 0.8 which defines the upper limit of the crack velocity in the subsonic regime, as will be pointed out in Section 7.
238 A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249

Fig. 3. Dimensionless function R(M2 )/me2 vs. M2 .

The dynamic local energy release rate may be calculated as the generalized specific flux into the crack tip (see for
example Freund, 1972) which may be written as:
π T 
G= σ 2 (r, 0)U(r, π) . (5.19)
2
By using Eqs. (5.11) and (5.13) the result is:
1
G = KT (l)[Q]K(l) (5.20)
2
with:
N(π)
Q = Re[iEH−1 ] = . (5.21)
R(M2 )

6. The permeable crack

For a crack permeable to the electric field the boundary conditions are:
+ − + −
σyy (x, 0) = σyy (x, 0) = 0, σxy (x, 0) = σxy (x, 0) = 0, |x| < l, (6.1)
Dy+ (x, 0) = Dy− (x, 0), Ex+ (x, 0) = Ex− (x, 0), |x| < ∞. (6.2)
The starting point to solve the problem is Eq. (4.1) evaluated on the crack surfaces which gives:
Λ+ (x) + Λ− (x) = 2iH−1 σ ± ∞
 
2 (x) − σ 2 , |x| < l. (6.3)
The boundary conditions Eqs. (6.1) and (6.2)1 allow us to write:
c T
σ± c
2 (x) = (0, 0, Dy ) = Dy k, |x| < l (6.4)
where k = (0, 0, 1)T
and Dyc ≡ Dy±
is the unknown component of the electric displacement on the crack surface. By
substituting Eq. (6.4) into Eq. (6.3) leads to the Hilbert problem:
Λ+ (x) + Λ− (x) = 2iH−1 j, |x| < l (6.5)
in which j = Dyc k − σ ∞
2 .
The solution to the above problem is:

Λ(z) = i 1 − diag g(zk ) H−1 j


  
(6.6)
where the function g(zk ) is that introduced in Eq. (5.3). Substituting Eq. (6.6) into Eq. (4.1) produces:
σ 2 = Dyc k − Re H diag g(zk ) H−1 j.
   
(6.7)
A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249 239

In particular, Eq. (6.7) evaluated ahead of the crack tip x = l gives:


x
σ 2 (x, 0) = Dyc k + √ (σ ∞ c
2 − Dy k), x > l. (6.8)
2
x −l 2

The generalized vectors σ 1 and U given by Eqs. (3.7)1 and (3.11)1 respectively, become:
σ1 = σ∞ −1
   −1 
1 + Re[GH ]j − Re G diag g(zk ) H j, (6.9)
(1) (2)
   −1 
U = Γ ∞ x + Γ ∞ y + Re E diag zk − h(zk ) H j. (6.10)
In order to obtain Dyc one can start from the boundary condition Eq. (6.2)2 which can be rewritten as follows:

[Γ (1) ]3 (x, 0) = [Γ (1) ]+


3 (x) − [Γ
(1) +
]3 (x) = 0, |x| < ∞ (6.11)
where [·]3 stands for the third row of the term inside brackets. From Eqs. (3.3)1 , (4.4) and (5.21) the boundary
condition Eq. (6.11) leads to the condition:
N(π) 3 (HΛ+ )(x) = N(π) 3 (HΛ− )(x), |x| < ∞
   
(6.12)
which produces:
 
N(π) 3 (HΛ)(x) = 0, |x| < ∞. (6.13)
Thence, along the crack faces one has the following relations:
3
 3

N3k (π)[H]k Λ+ (x) = 0, N3k (π)[H]k Λ− (x) = 0, |x| < l (6.14)
k=1 k=1
or equivalently:
3

N3k (π)[H]k Λ+ (x) + Λ− (x) = 0,
 
|x| < l. (6.15)
k=1
After substituting Eq. (6.5) into Eq. (6.15) it leads to:
N33 (π)(Dyc − Dy∞ ) − N31 (π)σxy
∞ ∞
− N32 (π)σyy = 0, |x| < l. (6.16)
By noting that N31 (π) = 0, the required component of the electric displacement on the crack surfaces can be deter-
mined as:
Dyc = Dy∞ + Γ (M2 )σyy

(6.17)
with Γ (M2 ) = N32 (π)/N33 (π).
The vector intensity factor at the tip x = l, defined by Eq. (5.7), is obtained from Eq. (6.8) in the form:

K(l) = πl(σxy ∞ ∞
, σyy , Dy∞ − Dyc )T (6.18)
with Dyc given by Eq. (6.17).
The asymptotic dominant terms at the tip x = l of the generalized stress and displacement vectors as well as the
energy release rate are formally the same as those obtained in Section 5 with K(l) given by Eq. (6.18).

7. Discussion of results

In this section, numerical results are represented and discussed by focusing the attention on a crack propagating
∞ , E ∞ ), thence assuming σ ∞ = σ ∞ = D ∞ = 0. The
under the action of the far-field electro-mechanical loading (σyy y xx xy x
use of Ey∞ instead of Dy∞ is with a view to experiments where it is easier to impose an electric field in the medium
than an electric displacement field.
For the present case, the relation between the two far-field electric loadings has been obtained from the following
equation:
e
Dy = ε1 Ey + (σyy − σxx ) (7.1)
m
240 A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249

produced through Eqs. (2.5) and (2.6), with ε1 = (2e2 + mε)/m.


First of all, the combined effects of both the far-field loading and crack velocity on the energy release rate, under
impermeable and permeable conditions, will be analyzed.
For an impermeable crack Eqs. (5.20) and (5.21) lead to G∗ = 2G/ lπ = G∗M + G∗E , where G∗M and G∗E are the
normalized mechanical and electrical components of the energy release rate respectively, obtained as follows:
 
G∗M = N22 ∗ (π) + e N ∗ (π) (σ ∞ )2 + ε N ∗ (π)σ ∞ E ∞ ,
yy 1 23 yy y
m 23
    (7.2)
∗ e ∗ e ∗ ∞ 2 ∗ 2e ∗ ∞ ∞ 2 ∗ ∞ 2
GE = N (π) + N33 (π) (σyy ) + ε1 N32 (π) + N33 (π) σyy Ey + ε1 N33 (π)(Ey )
m 32 m m
with the position that an asterisked entry of a matrix stands for the ratio between the entry and function R(M2 ), given
by Eq. (5.18). For the permeable crack the same quantity takes the simple form:
 ∞ 2
G∗p = N22
 ∗ ∗
(π) − Γ (M2 )N23 (π) (σyy ) (7.3)
which is independent of electric loading.
Figs. 4 and 5 represent the behaviours of the dimensionless quantities mG∗ /(σyy ∞ )2 and mG∗ /(σ ∞ )2 vs. M for
p yy 2
∞ ∞
σyy = 10 MPa and 20 MPa, respectively and Ey = ±5 kV/cm. It is shown that both quantities are positive and
increasing functions of crack velocity up to the critical value defined by M2∗ ≈ 0.8 (formerly identified in Fig. 3)
and negative functions for greater values of M2 . In what follows, it will be assumed that 0  c < c∗ = M2∗ v2 is the
effective subsonic range of crack velocity, which corresponds to positive values of the energy release rate.
Moreover, in the same figures it appears that for the impermeable case and for both values of the far-field me-
chanical load, a positive electric loading increases the dimensionless energy release rate more than a negative one,
particularly as the critical velocity is approached.
For both the stress levels and low crack velocities, the permeable energy release rate is nearly coincident with the
impermeable one corresponding to the positive electrical loading.
Figs. 6 and 7 show the normalized energy release rate G∗ as a function of the applied electric field when σyy ∞=

10 MPa and 20 MPa, respectively, for various values of M2 . It should be noted that an increase of the range of electric
loading enabling a positive energy release rate corresponds to an increase of crack velocity as well as of mechanical
loading.
It may be inferred that for a sufficient high stress level the mechanical component of the energy release rate
overrides the negative contribution due to the correspondent electrical component.

∞ )2 and mG∗ /(σ ∞ )2 vs. M for σ ∞ = 10 MPa and E ∞ = ±5 kV/cm.


Fig. 4. Dimensionless energy release rates mG∗ /(σyy p yy 2 yy y
A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249 241

∞ )2 and mG∗ /(σ ∞ )2 vs. M for σ ∞ = 20 MPa and E ∞ = ±5 kV/cm.


Fig. 5. Dimensionless energy release rates mG∗ /(σyy p yy 2 yy y

Fig. 6. Normalized energy release rate G∗ as a function of electric loading for σyy
∞ = 10 MPa and various values of M .
2

It is worthwhile noting that, at constant crack velocity, the energy release rate increases up to its maximum as
the electric loading increases from zero to some positive value which depends on both the stress level and crack
velocity. As the electric loading exceeds this value or increases negatively the trend is reversed. From above illustrated
features one may infer that, for the chosen loading, impermeable crack propagation is promoted or inhibited by
the electric field, depending on its dynamical state. This result seems in accordance with that pointed out by Chen
and Karihaloo (1999) for a crack propagating in a piezoelectric medium
√ under Mode III conditions. By combining
Eqs. (2.6) and (5.15) the normalized near-tip electric field Ey∗ = 2x/ l Ey ahead of the impermeable crack tip is
obtained as follows:
   
∗ e ∗ ∞ e ∗ e ∗ ∞
Ey = 1 + M13 (0) Ey + M12 (0) + M13 (0) σyy . (7.4)
m mε1 m
242 A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249

Fig. 7. Normalized energy release rate G∗ as a function of electric loading for σyy
∞ = 20 MPa and various values of M .
2

∞ = 10 MPa and various values of electric loading.


Fig. 8. Normalized electric field Ey∗ ahead of the crack tip vs. M2 for σyy

For the permeable crack the same quantity is given by:


 
∗ 1 e ∗ ∗
 ∞
(Ey )p = M (0) − Γ (M2 )M13 (0) − 1 − Γ (M2 ) σyy . (7.5)
ε1 m 12
Figs. 8 and 9 show the variation of above fields vs. M2 under various values of the electric loading, when σyy ∞ =

10 MPa and σyy = 20 MPa, respectively. As can be seen, at low velocities the electric field for the impermeable
crack is nearly independent of the electric loading and is indistinguishable from the permeable one. The effects of the
applied electric field on the impermeable crack are significant at high crack velocities. In particular, it can be seen that
the near tip electric field increases as the electric loading increases negatively and otherwise decreases. Moreover, by
comparing the above figures, it should be noted that an increase of the tensile loading leads to a positive electric field
irrespective of the sign of the electric loading.
A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249 243

∞ = 20 MPa and various values of electric loading.


Fig. 9. Normalized electric field Ey∗ ahead of the crack tip vs. M2 for σyy

∞ = 10 MPa, 20 MPa and E ∞ = ±4 kV/cm.


Fig. 10. Dimensionless electric field on the permeable crack surface vs. M2 for σyy y

As concerns the permeable crack, the electric field on the crack surface is obtained in the following form:
 
1 e
Eyc = Ey∞ + + Γ (M2 ) σyy ∞
. (7.6)
ε1 m
Fig. 10 represents the dimensionless field Eyc /Ey∞ against M2 under mixed loading conditions which are specified in
the same figure. It can be seen that at low crack velocity the above field is nearly independent of loading conditions.
When the crack velocity increases the electric field increases positively with the mechanical loading for a negative
electric loading and, in the same way, decreases for a positive one.
Among the features which can be interesting in the topic of crack propagation in piezoelectric media are the effects
of electro-mechanical loadings on the√near-tip dynamic hoop stress.
The normalized hoop stress σθ∗ = 2r/ lσθ given by:
σθ∗ = σxx

sin2 θ + σyy

cos2 θ − σxy

sin 2θ (7.7)
244 A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249

∞ = 5 MPa, E ∞ = 0 and various values of M .


Fig. 11. Angular variation of normalized hoop stress for σyy y 2

∞ = 5 MPa, E ∞ = 5 kV/cm and various values of M .


Fig. 12. Angular variation of normalized hoop stress for σyy y 2

is obtained through Eqs. (5.15) and (5.16) which produce the following normalized near-tip stress components for the
impermeable crack:
 
∗ ∗ e ∗ ∞ ∗
σxx = M12 (θ ) + M13 (θ ) σyy + ε1 M13 (θ )Ey∞ ,
m
 
∗ ∗ e ∗ ∞
σyy = L22 (θ ) + L23 (θ ) σyy + ε1 L∗23 (θ )Ey∞ , (7.8)
m
 
∗ e
σxy = L∗12 (θ ) + L∗13 (θ ) σyy∞
+ ε1 L∗13 (θ )Ey∞ .
m
Fig. 11 represents the near-tip normalized hoop stress σθ∗ against the polar angle θ for various values of M2 , when a
∞ = 5 MPa is applied. Assuming (but not claiming) the validity of the maximum hoop stress
pure tensile loading σyy
criterion for crack branching, it appears that the crack tends to deviate from its original direction as the crack velocity
increases.
In Fig. 12 the same quantity is represented when a pure electric loading Ey∞ = 5 kV/cm is applied. Under this
condition, it seems that at low velocity the crack maintains its original direction and tends to branch at higher velocity.
A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249 245

∞ = 5 MPa, E ∞ = −5 kV/cm and various values of M .


Fig. 13. Angular variation of normalized hoop stress for σyy y 2

∞ = 10 MPa, M = 0.5 and various values of electric loading.


Fig. 14. Angular variation of normalized hoop stress for σyy 2

The effect of crack velocity when a pure electric loading Ey∞ = −5 kV/cm is applied is shown in Fig. 13. For this
case it seems that crack deviation occurs for all crack velocities.
In order to clarify the influence of electric loading on the hoop stress behaviour and in particular on crack branching,
Figs. 14 and 15 represent the normalized hoop stress for M2 = 0.5 and M2 = 0.7, respectively, when a tensile stress
∞ = 10 MPa is applied. The chosen values of electric loading are in accordance with positive values of the energy
σyy
release rate conforming to Fig. 7. From Fig. 14 it can be inferred that at low crack velocity the effect of electric loading
is negligible and the applied tensile stress is responsible for the crack deviation.
Nearly the same behaviour is shown in Fig. 15 where it can be seen that at high velocity crack branching is almost
independent of electric loading whose solely effect is to change the maximum of the hoop stress. This result is not in
agreement with that obtained by Soh et al. (2002), who have shown that a positive electric loading may promote crack
branching for high crack velocity.
This discrepancy is, almost certainly, due to the simplified model introduced in the present paper with respect to
that employed in the above quoted one.
The dimensionless hoop stress (σθ∗ /σyy ∞ ) obtained under permeable conditions, has the following expression:
p
246 A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249

∞ = 10 MPa, M = 0.7 and various values of electric loading.


Fig. 15. Angular variation of normalized hoop stress for σyy 2

Fig. 16. Angular variation of dimensionless hoop stress (σθ∗ /σyy


∞ ) for various values of M .
p 2

σθ∗

(θ ) sin2 θ + L∗22 (θ ) − Γ (M2 )L∗23 (θ ) cos2 θ
 ∗ ∗
  

= M12 (θ ) − Γ (M2 )M13
σyy
− L∗12 (θ ) − Γ (M2 )L∗13 (θ ) sin 2θ
 
(7.9)
which is independent of the electric loading.
Fig. 16 displays the dependence of this quantity on the polar angle for various values of M2 .
It is observed that a small crack deviation may occur as the crack velocity increases.

8. Conclusions

In this paper a simplified constitutive equation for piezoelectric material has been addressed within the framework
of a plane problem of a steadily propagating Griffith crack. The mechanical and electric fields under generalized
electro-mechanical loading were formulated by complex functions for both the impermeable and permeable boundary
conditions.
Although the adopted model of piezoelectric material seems primitive it leads, at least from a qualitative point of
view, to some results illustrating the main dynamical features which may be a trend for more general studies.
A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249 247

Some main conclusions are summarized as follows:

(a) The effective range of crack velocity in the subsonic regime is 0  c < c∗ ≈ 0.8v2 where v2 is the lowest wave
speed which enters in the chosen model of piezoelectric medium. This range is independent of far-field loading
as well as of electrical boundary conditions.
The analysis performed for generalized Mode-I loading conditions (σyy ∞ , E ∞ ) has shown that under impermeable
y
boundary conditions:
(b) For a sufficiently high stress level the energy release rate is a positive increasing function of crack velocity in the
above mentioned range. At constant crack velocity, an increase of the energy release rate may occur depending
on a positive electric load level as well as of the stress level. A negative increase of electric loading leads to a
decrease of the energy release rate.
(c) The near tip electric field is an increasing function of crack velocity and this trend is promoted by a negative
increase of the electric loading and retarded by a positive one.
(d) On the basis of maximum hoop stress criterion it seems that under mixed loading conditions crack branching is
nearly independent of the electric loading as well as of crack velocity.
For a permeable crack the most significant results are:
(e) The energy release is independent of the electric loading and is a positive increasing function of crack velocity in
the same range as in (a).
(f) The electric field on the crack surface is a positive increasing function of crack velocity for a negative electric
loading and decreasing function for a positive one.
(g) Crack branching is promoted by an increase of crack velocity.

Acknowledgements

This research was supported by the Italian Ministry for University and Scientific, Technological Research MIUR
(40% and 60%). The research topic is one of the subjects of the Centre of Study and Research for the Identification of
Materials and Structures (CIMEST)—“M. Capurso”.

Appendix A

By introducing the following constants and notations:


pk (n̄pk2 − γ eα) pk2 eα cos θk /2 sin θk /2
ak = , bk = , C(θk ) = √ , S(θk ) = √ , k = 1, 2,
α − pk2 α − pk2 ck (θ ) ck (θ ) (A.1)
C(θ ) = cos θ/2, S(θ ) = sin θ/2
the entries of matrix L(θ ) are:
   
L11 = − a1 h11 C(θ1 ) + a2 h21 C(θ2 ) + eh31 C(θ ) , L12 = a1 h12 S(θ1 ) + a2 h22 S(θ2 ) + eh32 S(θ ) ,
     
L13 = a1 h13 S(θ1 ) + a2 h23 S(θ2 ) + eh33 S(θ ) , L21 = m h11 S(θ1 ) + h21 S(θ2 ) − eh31 S(θ ) ,
       
L22 = m  h12 C(θ1 ) + h22 C(θ2 ) − eh32 C(θ ) , L23 = m h13 C(θ1 ) + h23 C(θ2 ) − eh33 C(θ ) , (A.2)
   
L31 = − b1 h11 S(θ1 ) + b2 h21 S(θ2 ) − εh31 S(θ ) , L32 = − b1 h12 C(θ1 ) + b2 h22 C(θ2 ) − εh32 C(θ ) ,
 
L33 = − b1 h13 C(θ1 ) + b2 h23 C(θ2 ) − εh33 C(θ )
in which:
h11 = e2 αp22 (α − p12 ) − m
ε(α − p12 )(α − p22 ), h21 = m ε(α − p12 )(α − p22 ) − e2 αp12 (α − p22 ),
h12 = − εp2 (n̄p22 − eγ α) + e2 p22 α (α − p12 ), h22 = εp1 (n̄p12 − eγ α) + e2 p12 α (α − p22 ),
   

(α − p22 ) + p2 (n̄p22 − eγ α) (α − p12 ), h23 = e m (α − p12 ) + p1 (n̄p12 − eγ α) (α − p22 ), (A.3)
   
h13 = −e m
eα p12 h12 p22 h22

m

h31 = m
(h11 + h21 ), h32 = + , h33 = (h13 + h23 ).
ε α − p12 α − p22 e
248 A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249

For the entries of matrix M(θ ) the following contractions are considered:
mαpk2


mk = − eγ , nk = pk − n̄ , k = 1, 2. (A.4)
α − pk2 α − pk2
Thence:
M11 = m1 h11 S(θ1 ) + m2 h21 S(θ2 ) + eh31 S(θ ), M12 = m1 h12 C(θ1 ) + m2 h22 C(θ2 ) + eh32 C(θ ),
 
M13 = m1 h13 C(θ1 ) + m2 h23 C(θ2 ) + eh33 C(θ ), M21 = − n1 h11 C(θ1 ) + n2 h21 C(θ2 ) + eh31 C(θ ) ,
M22 = n1 h12 S(θ1 ) + n2 h22 S(θ2 ) + eh32 S(θ ), M23 = n1 h13 S(θ1 ) + n2 h23 S(θ2 ) + eh33 S(θ ),

b1 b2

b1 b2 (A.5)
M31 = − h11 C(θ1 ) + h21 C(θ2 ) − εh31 C(θ ) , M32 = h12 S(θ1 ) + h22 S(θ2 ) − εh32 S(θ ),
p1 p2 p1 p2
b1 b2
M33 = h13 S(θ1 ) + h23 S(θ2 ) − εh33 S(θ ).
p1 p2
The entries of matrix N(θ ) are:
1  1 
N11 = − b1 h11 c1 (θ )S(θ1 ) + b2 h21 c2 (θ )S(θ2 ) , N12 = b1 h12 c1 (θ )C(θ1 ) + b2 h22 c2 (θ )C(θ2 ) ,
e e
1  (A.6)
N13 = b1 h13 c1 (θ )C(θ1 ) + b2 h23 c2 (θ )C(θ2 ) , N21 = p1 h11 c1 (θ )C(θ1 ) + p2 h21 c2 (θ )C(θ2 ),
e
N22 = p1 h12 c1 (θ )S(θ1 ) + p2 h22 c2 (θ )S(θ2 ), N23 = p1 h13 c1 (θ )S(θ1 ) + p2 h23 c2 (θ )S(θ2 ),
N31 = γ N21 − h31 C(θ ), N32 = γ N22 − h32 S(θ ), N33 = γ N23 − h33 S(θ ).
Appendix B

The following symbols are used in this paper:

σij stress tensor


γij strain tensor
Di electric displacements vector
Ei electric field vector
cij kl elastic constants
eij k piezoelectric constants
εij dielectric constants
x1 , x2 , x3 Cartesian coordinates
ρ density of the material
t time
ui , u, v elastic displacement components
ϕ electric potential
Γ (1) , Γ (2) generalized strain
M1 , M2 Mach numbers
c crack velocity
v1 , v2 velocities
(1) (2)
λ± , λ± imaginary eigenvalues
Ω1 (z1 ), Ω2 (z2 ) complex potentials
σ 1 (∞), σ 2 (∞) uniform electro-mechanical loading at infinity
Ex∞ , Ey∞ electric field at infinity
U generalized displacement vector
2l length of crack
r, θ polar coordinates
G dynamic local energy release rate
K(l) vector intensity factor
G∗M normalized mechanical component of the energy release rate
A. Piva et al. / European Journal of Mechanics A/Solids 25 (2006) 230–249 249

G∗E normalized electrical component of the energy release rate


G∗ normalized total energy release rate
σθ∗ normalized hoop stress
∗ , σ ∗ , σ ∗ normalized near-tip stress components
σxx xy yy

References

Chen, Z.T., Karihaloo, B.L., Yu, S.W., 1998. A Griffith crack moving along the interface of two dissimilar piezoelectric materials. Int. J. Fracture 91,
197–203.
Chen, Z.T., Karihaloo, B.L., 1999. Dynamic response of a cracked piezoelectric ceramic under arbitrary electro-mechanical impact. Int. J. Solids
Struct. 36, 5125–5133.
Chen, Z.T., Yu, S.W., 1997. Anti-plane Yofee crack problem in piezoelectric materials. Int. J. Fracture 84, L41–L45.
Deeg, W.F., 1980. The analysis of dislocation, crack and inclusion problems in piezoelectric solids. Ph.D. Thesis, Stanford University.
Dunn, M.N., 1994. The effect of crack face boundary conditions on the fracture mechanics. Engrg. Fract. Mech. 48, 25–39.
Freund, L.B., 1972. Energy flux into the tip of an extending crack in an elastic solid. J. Elasticity 2, 341–349.
Gao, C.F., 2000. Further study of the generalized 2D problem of an elliptic hole or a crack in piezoelectric media. Mech. Res. Commun. 27,
429–434.
Gao, H.J., Zhang, T.Y., Tong, P., 1997. Local and global energy release rates for an electrically yielded crack in a piezoelectric ceramic. J. Mech.
Phys. Solids 45, 491–510.
Hao, T.-H., Shen, Z.-Y., 1994. A new electric boundary condition of electric fracture mechanics and its applications. Engrg. Fract. Mech. 47,
793–802.
Kwon, J.H., Lee, K.Y., 2000. Moving interfacial crack between piezoelectric ceramic and elastic layers. Eur. J. Mech. 19, 979–987.
Kwon, J.H., Lee, K.Y., Kwon, S.M., 2000. Moving crack in a piezoelectric ceramic strip under anti-plane shear loading. Mech. Res. Commun. 27,
327–332.
Landis, C.M., 2004. Energetically consistent boundary conditions for electromechanical fracture. Int. J. Solids Struct. 41, 6291–6315.
Li, S., 2003. On saturation-strip model of a permeable crack in a piezoelectric ceramic. Acta Mech. 165, 47–71.
Li, X.F., Fan, T.Y., Wu, X.F., 2000. A moving Mode-III crack at the interface between two dissimilar piezoelectric materials. Int. J. Engrg. Sci. 38,
1219–1234.
McMeeking, R.M., 1989. Electrostrictive forces near crack-like flaws. J. Appl. Math. Phys. 40, 615–627.
McMeeking, R.M., 2004. The energy release rate for a Griffith crack in a piezoelectric material. Engrg. Fracture Mech. 71, 1149–1163.
Pak, Y.E., 1990. Crack extension force in a piezoelectric material. J. Appl. Mech. 57, 647–653.
Park, S., Sun, C.T., 1995. Fracture criteria for piezoelectric ceramics. J. Amer. Ceram. Soc. 78, 1475–1480.
Parton, Y.E., 1976. Fracture mechanics of piezoelectric materials. Acta Astronaut. 3, 671–683.
Piva, A., Viola, E., 1988. Crack propagation in an orthotropic medium. Engrg. Fract. Mech. 29, 535–548.
Piva, A., Viola, E., Tornabene, F., 2005. Crack propagation in an orthotropic medium with coupled elastodynamic properties. Mech. Res. Comm. 32,
153–159.
Sih, G.C., 2002. A field model interpretation of crack initiation and growth behavior in ferroelectric ceramics: change of poling direction and
boundary condition. J. Theor. Appl. Fract. Mech. 38, 1–14.
Soh, A.K., Liu, J.X., Lee, K.L., Fang, D.N., 2002. On a moving Griffith crack in anisotropic piezoelectric solids. Arch. Appl. Mech. 72, 458–469.
Sosa, H.A., 1992. On the fracture mechanics of piezoelectric solids. Int. J. Solids Struct. 29, 2613–2622.
Spyropoulos, C.P., 2004. Energy release rate and path independent integral study for piezoelectric material with crack. Int. J. Solids Struct. 41,
907–921.
Suo, Z., Kuo, C.M., Barnett, D.M., Willis, J.R., 1992. Fracture mechanics for piezoelectric ceramics. J. Mech. Phys. Solids 40, 739–765.
Zhang, T.Y., Gao, C.F., 2004. Fracture behaviors of piezoelectric materials. J. Theor. Appl. Fract. Mech. 41, 339–379.
Zhang, T.Y., Tong, P., 1996. Fracture mechanics for a mode III crack in a piezoelectric material. Int. J. Solids Struct. 33, 343–359.
Zhang, T.Y., Zhao, M.H., Tong, P., 2002. Fracture of piezoelectric ceramics. Adv. Appl. Mech. 38, 148–289.

You might also like