You are on page 1of 18

Engineering Fracture Mechanics 77 (2010) 309–326

Contents lists available at ScienceDirect

Engineering Fracture Mechanics


journal homepage: www.elsevier.com/locate/engfracmech

Fracture mechanics of piezoelectric materials – Where are we right now?


Meinhard Kuna
Technische Universität Bergakademie Freiberg, Institute of Mechanics and Fluid Dynamics, Lampadiusstraße 4, D-09596 Freiberg, Germany

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

Article history: Piezoelectric and ferroelectric materials have gained extensive applications in electrome-
Received 19 December 2008 chanical devices, microelectromechanical systems and smart composite materials. In order
Received in revised form 15 March 2009 to assess the strength and durability of those materials and components, exhaustive theo-
Accepted 20 March 2009
retical and experimental investigations have been performed over the past three decades.
Available online 5 April 2009
The aim of the paper is to give a short overview and a critical discussion about the present
state in the field of piezoelectric fracture mechanics. After an introduction, linear piezo-
Keywords:
electric fracture theory is explained with emphasis to special features like anisotropy,
Linear and non-linear piezoelectric fracture
mechanics
mode mixture and electric properties of cracks. Next, suggested fracture criteria are pre-
Fracture criteria sented and contrasted with experimental observations in fracture testing. Cracks under
Ferroelectric domains static, cyclic and dynamic loading by electrical and mechanical fields are taken into
Electromechanical loading account. A great challenge is to tackle the non-linear phenomena and ferroelectric domain
Crack tip parameters switching in the fracture process zone. Finally, conclusions are drawn with respect to open
Domain switching problems and desirable future research areas. To limit the scope of the paper, fracture
behavior of interface cracks will not be addressed.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Piezoelectric materials have widespread applications in modern technical areas such as mechatronics, micro system tech-
nology or smart structures, serving as sensors, actuators or transducers [1,2]. New smart composites consist e.g. of carbon
fiber reinforced polymeric laminates with integrated piezoelectric plates or fibers. Functional ceramics as barium titanate Ba-
TiO3, lead zirconate titanate PZT and familiar materials, which possess good actuating strain (maximum 0.2%), fast response
and high stiffness are of most technical importance. However, one great problem of these materials is their inherent brittle-
p
ness (ultimate strength < 100 MPa) and low fracture toughness (0.5–2.0 MPa m). Furthermore, highly inhomogeneous and
concentrated stresses and electrical fields may occur inside the piezoelectric structures and composites due to fabrication or
operational loads. Damage emerges as a combined electromechanical phenomenon leading either to mechanical failure or
electric breakdown at critical sites as materials flaws, electrodes or interfaces, see Fig. 1. Most crucial are mechanical fatigue
and subcritical crack growth in multi-layer actuators (Fig. 2), when exposed to pure cyclic electric loading e.g. [3,4]. Severe
internal stresses may arise in smart composites at geometrical imperfections or materials inhomogeneities due to mismatch
of the thermo-electro-elastic properties. The assessment of defects like micro-cracks, voids a. o. plays an important role for
the strength and reliability of piezoelectric components under combined electrical, mechanical and thermal loading. There-
fore, much attention has been devoted to fracture mechanics of piezoelectric, ferroelectric and electrostrictive materials for
the last 25 years.
Historically, intensive research on piezoelectric fracture mechanics started in the 1980s motivated by failure in piezoelec-
tric devices. Using mostly Vickers-indentation experiments, the apparent anisotropy in fracture toughness of poled piezo-
electric ceramics and its dependence on applied electric field were observed [5–7]. Later, real fracture mechanics

E-mail address: meinhard.kuna@imfd.tu-freiberg.de

0013-7944/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfracmech.2009.03.016
310 M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326

conducting crack

electrodes

ceramic Insulating crack

Fig. 1. Failure of piezoelectric components (capacitors, actuators, and composites). Electric fields around insulating and conducting defects.

Fig. 2. Crack in a PZT multi-layer actuator (Rödel, Darmstadt).

experiments were performed by several researchers at three-point bending, compact tension and double cantilever beam
specimens under mechanical and/or electrical loading [8–15]. Recently, Schneider [16] summarized and critically discussed
the achievements in fracture testing.
On the theoretical side, the first paper on piezoelectric fracture mechanics was written by Parton in 1976 [17]. With the
pioneering works of Deeg, Sosa, Suo, Pak, Park and Sun, Xu and Rajapakse [18–25] the mathematical methods used in linear
elastic fracture mechanics (LEFM) such as complex function theory (Lekhnitskii, Stroh), integral transformation and singular
integral equations have been adopted to linear piezoelectric problems. Many two-dimensional (2D) standard crack config-
urations in infinite domains were treated successfully. A few exact solutions are available for simple three-dimensional (3D)
configurations as embedded circular or elliptic cracks [26–28]. Meanwhile, an overwhelming amount of theoretical papers
has been published giving solutions of diverse crack configurations. The theoretical fundamentals of linear piezoelectric
crack analysis are well documented in the monograph of Qin [29] and the review articles written by Zhang et al. [30], Chen
and Lu [31] and Chen and Hasebe [32] including original literature cited therein.
However, for practical applications and analysis of fracture test specimens, the bounded geometry, the complex electro-
mechanical loading conditions and materials non-linearity require the application of numerical methods as finite elements
(FEM) or boundary elements (BEM). After the first studies of Kumar and Singh [35,36], substantial progress was made in this
respect by many contributions, e.g. [37–42]. Approved FEM-techniques for LEFM such as special crack tip elements, crack
closure integrals and J-integrals have been extended to calculate the relevant fracture parameters like intensity factors
and energy release rates for quite complicated electromechanical boundary value problems with cracks. Interested readers
are referred to the review article by Kuna [34].
The major difficulty in developing a fracture theory of functional ceramics lies particularly in the physical understanding
of the involved phenomena. The essential question is, how does the electric field influence the loading situation at the crack
tip, the actual non-linear material state and the fracture toughness itself. Functional ceramics have the following features,
making a significant difference to fracture of classical structural ceramics. The major challenges in research originate from
these reasons:

 Due to polarization all material properties are generally anisotropic.


 For large amplitudes ferroelectrics exhibit a strongly non-linear behavior.
 Mechanical and electrical quantities are inherently coupled at the crack.
 There is always a mixture of (mechanical and electrical) crack opening modes.
 The electric properties of the medium inside the crack have to be accounted for.
 In many applications (actuators, composites, etc.) cracks occur at interfaces between piezoelectrics and other materials.
M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326 311

Table 1
Basic relations of electromechanical field theory.

Mechanics Electrostatics
Primary variables Displacement ui Electric potential u
Kinematical relations Strain eij ¼ 1=2ðui;j þ uj;i Þ Electric field Ek = u,k
Dual variables Stress rij Electric displacement Di
Balance equations for forces and charges in domain V rij;j þ bi ¼ qu€ i volume forces bi mass density q Di;i  x  V ¼ 0 volume charges x V
Natural boundary conditions on ST (normal vector ni) rij nj ¼ ti tractions ti D j n j ¼ x  S surface charges x
S
Essential boundary conditions on Su ui ¼ u i u¼u 

Fig. 3. Ferroelectric hysteresis of polarization and deformation (butterfly); (. . .. . ..) piezoelectric regime (Pr – remanent polarization, Ec – coercive field
strength, and er – remanent strain).

2. Fundamental equations of electromechanics

The basic field variables and governing equations for a coupled electromechanical boundary value problem are summa-
rized in Table 1 (for details see [43,44]). We consider the interaction of anisotropic elasticity and static electric fields. The
fundamental physical laws are the mechanical equilibrium equations (Cauchy) and Gauss’ law of electrostatics, providing
together with the kinematical relations a system of partial differential equations for the field variables ui and u. The set
of equations is completed by natural and essential boundary conditions prescribed on the parts ST and Su of the surface S.
Most functional ceramics as BaTiO3 or PZT are polycrystalline ferroelectric materials. Below the Curie-temperature, the
crystal unit cells possess an electric dipole moment – the spontaneous polarization vector P0. Inside a grain, unit cells with
the same polarization form clusters – the ferroelectric domains. For most applications, these ferroelectric ceramics are poled
by means of a high electric field above the Curie-temperature, i.e. a macroscopically uniform alignment of the domains is
frozen. This configuration can be described by a linear piezoelectric electromechanical behavior. In piezoelectrics, the electrical
and mechanical fields are coupled by the constitutive equations (1), describing the actuatoric (left) and sensoric (right) effect.
@H @H
rij ¼ cijkl ekl  ekij Ek ¼ ; Dn ¼ enkl ekl þ jnm Em ¼  ð1Þ
@ eij @En
The materials law (1) can be deduced from the corresponding thermodynamic state potential – the electric enthalpy density
H – by differentiation with respect to the state variables.
1 1
Hðeik ; Ei Þ ¼ cijkl eij ekl  eikl Ei ekl  jij Ei Ej ð2Þ
2 2
Here cijkl, ekij, jij are the tensors of elastic, piezoelectric and dielectric constants. In general, the material behavior is aniso-
tropic. Poled ceramics obey transversal isotropy with respect to the poling direction. The majority of piezoelectric sensors
and actuators in use are operated at this poled state, which corresponds to the saturated points at the ends of the hysteresis
branches (dotted red lines) in Fig. 3, i.e. within a small signal range.
If however, the applied electric field changes or reverses its direction, the domains rearrange according to the field,
which leads to the polarization-field hysteresis and an associated strain-electric field ‘‘butterfly”-hysteresis, see Fig. 3.
The change of polarization can also be caused by mechanical stresses – which is called ferroelastic behavior. Piezoelectric
and ferroelectric components operated under large signal ranges will experience a continuous local change in polarization,
following the non-linear hysteresis laws. The same may happen at highly concentrated electric and/or mechanical fields
around crack tips, leading to a non-linear behavior in a restricted region, comparable with the plastic zone in ductile
312 M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326

materials. A phenomenological description of the hysteresis on a macro level is easy for 1D behavior. The complete 3D ten-
sorial constitutive relations are still subject of current research (see review [45]). For further physical details about ferro-
electric/piezoelectric materials we refer to [46,47].

3. Linear piezoelectric fracture mechanics (LPFM)

3.1. Concept of intensity factors

The asymptotic behavior of all electromechanical fields shows an identical form, if the crack tip is approached (r ? 0) in
2D or at a point of the crack front in 3D, see Fig. 4. These near tip fields are given in polar coordinates (r, h) by Eqs. (3) and
pffiffiffi
(4). The mechanical stresses and the electric displacements Eq. (3) behave singular as 1= r,whereas the electric potential
pffiffiffi N
and the mechanical displacements in Eq. (4) show a parabolic shape  r . The angular functions fijN ; g Ni ; di ; and v N depend
only on the material constants. The coefficients KI, KII and KIII are the well-known mechanical stress intensity factors (SIF),
which are complemented by the new forth electric displacement intensity factor KIV (EDIF). The fracture parameter KIV char-
acterizes the singularity of the electric displacement field at a crack tip, i.e. the electric loading of the material. Since the
electric field is a vector quantity (instead of tensors in mechanics), no ‘‘descriptive opening mode IV” is associated with it.
From Eqs. (3) and (4) the mutual interdependence between mechanical and electrical quantities at the crack tip can be
recognized.

1 h i
rij ðr; hÞ ¼ pffiffiffiffiffiffiffiffiffi K I fijI ðhÞ þ K II fijII ðhÞ þ K III fijIII ðhÞ þ K IV fijIV ðhÞ
2pr
ð3Þ
1 h i
Dj ðr; hÞ ¼ pffiffiffiffiffiffiffiffiffi K I g Ij ðhÞ þ K II g IIj ðhÞ þ K III g III IV
j ðhÞ þ K IV g j ðhÞ
2pr
rffiffiffiffiffi
2rh I II III IV
i
ui ðr; hÞ ¼ K I di ðhÞ þ K II di ðhÞ þ K III di ðhÞ þ K IV di ðhÞ
p
rffiffiffiffiffi ð4Þ
2r  
uðr; hÞ ¼ K I v ðhÞ þ K II v ðhÞ þ K III v ðhÞ þ K IV v ðhÞ
I II III IV
p
The angular functions have the following mathematical structure, where the index N = {I, II, III, IV} comprises all four fracture
modes:
( ) ( )
X
4
M ia NaN pa X
4
M ia N a N
fi1N
¼ Re pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; fi2N
¼ Re pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a¼1 cos h þ pa sin h a¼1 cos h þ pa sin h
( ) ( ) ð5Þ
X4
M4a NaN pa X 4
M 4a N a N
N N
g1 ¼  Re pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; g2 ¼ Re pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

a¼1 cos h þ pa sin h a¼1 cos h þ pa sin h

X
4  qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi X
4  qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
N
di ¼ Re Aia NaN cos h þ pa sin h ; vN ¼ Re A4a NaN cos h þ pa sin h ð6Þ
a¼1 a¼1

Similar as in anisotropic elasticity, the solution of the basic partial differential equations is obtained by complex function
theory, expressed in the coordinates za = x1 + pax2. Hereby, pa(a = 1, 2, . . . , 4) are the eigenvalues of the characteristic

Fig. 4. Singular fields at the crack tip.


M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326 313

algebraic system Eq. (7), formed by the material constants. The matrices NaN and MMa are determined from the correspond-
ing eigenfunctions Aka, spanning the possible space of solutions. For the detailed derivation we refer to [19–23].
     

ci1k1 ei11 ci2k1 þ ci1k2 ei21 þ ei12 ci2k2 ei22 Ak


þ pþ p2 ¼0 ð7Þ
e1k1 j11 e2k1 þ e1k2 j12  j21 e2k2 j22 A4

ðci2k1 þ ci2k2 pa ÞAka þðe1i2 þ e2i2 pa ÞA4a


N 1
aN ¼ M Ma ¼ ð8Þ
ðe2k1 þ e2k2 pa ÞAka ðj21 þ j22 pa ÞA4a

3.2. Energy release rate ERR

The total change DP of potential electrical and mechanical energy in a structure, when the crack grows by a length Da, is
called the energy release rate G.

DP dP
G ¼  lim ¼ ¼ GIm þ GIIm þ GIII
m þ Ge ð9Þ
Da!0 Da da

This expression contains the common terms of mechanical energy release rates for each opening mode plus an electric part
Ge. Analogous to Irwin’s crack closure integral, G equals the work performed by the stresses and electric displacements (be-
fore crack advance) at the ligament Da with the jumps of displacements Dui and of the electric potential Du over the crack
faces during a virtual crack opening process. Because of reversibility this energy release rate is identical with the so-called
crack closure integral:
Z Da
dP 1
G¼ ¼ lim fr2i ðsÞDui ðDa  sÞ þ D2 ðsÞDuðDa  sÞgds ð10Þ
da Da!0 2Da 0

The relation between ERR and the intensity factors K = {KII, KI, KIII, KIV}T is given by [20,23]:

1 T X
4
G¼ K YðC; e; jÞK; Y ¼ ½Y NM  ¼  IfAMa NaN g ð11Þ
2 a¼1

The generalized Irwin-matrix Y depends on the elastic, piezoelectric and dielectric material constants and the relative ori-
entation of the crack with respect to the polarization axis.

3.3. Electromechanical J-integral

The generalization of the classical J-integral by Rice and Cherepanov to cracks in electromechanical media is the so-called
J em
k -integral vector, which is the energetic (configurational) force associated with a virtual displacement of the crack tip. It is
defined as a closed contour integral around the crack tip, if Ce is shrunk towards r ? 0, see Fig. 5.
Z Z
J em
k ¼ lim Q jk nj dC ¼ lim ðHdkj  rij ui;k  Dj u;k Þnj dC ð12Þ
Ce!0 Ce Ce!0 Ce
Q jk ¼ Hdjk  rij ui;k  Dj u;k

In the spirit of Eshelby’s concept, Qjk is called the piezoelectric energy momentum tensor. The J em
k -integral was first extended to
general coupled mechanical and electromagnetic field problems by Cherepanov [48] and later by Pak and Herrmann [49,22],
McMeeking [50–52], Chen and Lu [31] and Dascalu and Maugin [53]. For k = 1 the energy release rate J em 1 ¼ G is recovered.

Fig. 5. Electromechanical J-integral in 2D.


314 M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326

Using the Gauss integral theorem, the J em


k -integral can be converted into an integral along any outer contour C, its en-
closed domain A and the crack faces C+ and C as shown in Fig. 5. All divergence terms of the energy momentum tensor
are involved in the area-integral to achieve path independence.
Z Z Z
J em Q kj nj dC   u  qu
½b € i ui;k þ x
 V Ek þ H;k jexp dA þ ½Hd2k  t i ui;k  x
 S Ek dC
k ¼ i i;k ð13Þ
C A Cþ þC

This extended form [54] can be used to determine the electromechanical energy release rate of a crack in a complex piezo-
electric component under quite general electrical x  S and mechanical t i loading at the crack faces as well as body forces b,
i
inertia forces qu and charges xV in the volume A. It allows treating continuous spatial variations of the piezoelectric, elastic
€ 
and dielectric properties by the explicit derivative H;k jexp , which may occur in functionally graded materials or via non-linear
effects.
To derive the Jem-integral in the 3D case, one can assume a virtual extension Dlk(s) at each position s of the crack front,
directed normal (x1-axis) and lying in the crack plane (x1, x3), see Fig. 6. Then, an analogous expressions to Eq. (13) can be
derived, if Ce is replaced by a closed surface Se around a given segment Ds of the crack front and if instead of A the enclosed
volume V is considered in Fig. 6. The averaged released energy DP with respect to the area of crack growth DA is the sum of
all 2D terms G ¼ J em
1 along the crack front segment Ds, which is read as:
Z " R R
 #
em DP 1  V ðHdkj  rij ui;k þ Dj Ek Þqk;j dV þ SþC þSC ½Hnk  ti ui;k  x
 S Ek qk dS
J ðsÞ ¼  ¼ J em
k ðsÞDlk ds ¼ R ð14Þ
DA Ds DA  ½bi ui;k  qu
€i ui;k  x
 V u;k qk dV
V

The value J em ðsÞ in Eq. (14) has an average meaning for the crack extension along Ds and should be attributed to a mean posi-
tion s. Eq. (14) sets up another convenient formulation to compute the Jem-integral within the framework of FEM, the so-
called equivalent domain integral [40,54] known from LEFM. Here, the arbitrary smooth weighting function qk ðxÞ ¼ qDlk ðsÞ
varies from 1 at the inner surface Se down to zero at the outer surface S.
Finally, one important variant of the J-integral is introduced, the so-called interaction integral or M-integral. It is particu-
larly useful to separate the individual terms of J em
1 ¼ G correlated with each opening mode {I, II, III, IV} in case of mixed mode
loading situation. This technique is indispensable for piezoelectric fracture, since even in simple cases a combination of mode
I and IV appears, see [31,38,57,70]. Eq. (15) represents the interaction energy between the numerical solution of the actual
boundary value problem (Index 1) and an auxiliary known analytic solution (Index 2) for the same crack configuration, for
which usually the asymptotic field Eq. (4) is used.
emð1;2Þ
M ¼ J1
Z nh i o
ð1Þ ð2Þ ð2Þ ð1Þ ð1Þ ð2Þ ð2Þ ð1Þ
¼ Dð2Þ ð1Þ ð2Þ ð1Þ € ð1Þ ð2Þ
m Em  rmn emn d1j þ rij ui;1 þ rij ui;1 þ Dj u;1 þ Dj u;1 q;j þ qui ui;1 q dA
A
Z
þ  ð1Þ
x ð2Þ ð1Þ ð2Þ
S u;1  t i ui;1 qdC ð15Þ
Cþ þC

The electromechanical J-integral can also be determined experimentally from measured quantities as force F, displacement
u, electric charge Q and voltage U on the specimen during fracture tests, see [56,16,89].

3.4. Electric boundary conditions

One important question arises, if cracks in piezoelectric materials are investigated: how does the medium inside the crack
influence the behavior? Its dielectric property affects the electromechanical solution around the crack tip. Even air has a non-
zero relative permittivity of jr  1. Moreover, the electric field can easily penetrate into the thin crack slit, as sophisticated
measurements show [64]. Therefore, an insulating dielectric medium inside a crack must be considered as semi-permeable
for the electric field. Several analytical solutions dealt with this coupled field problem as e.g. McMeeking [51], Zhang

Fig. 6. Virtual crack extension DA along a crack front and J-integral domain.
M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326 315

Fig. 7. Electric field lines at a crack tip: impermeable (full lines) and permeable (dashed lines), conducting (full lines) and limited permeable (dashed lines).

et al. [30] and Gao and Fan [58] for the 2D Griffith crack. Depending on the ratio b between the dielectric permittivities of the
medium jC and that of the piezoelectric solid j, two extreme cases can be fixed:
(a) Impermeable crack: physically, this crack model means that the upper and lower crack faces (index +,) are free of
charges, i.e. the normal component D2 of the electric displacement vector vanishes (Fig. 7):

Du ¼ uþ  u – 0; Dþ2 ¼ xþS ¼ D2 ¼ xS ¼ 0 ð16Þ

It is valid in the limiting case b = jC/j ? 0. Then the field lines circumvent the crack, which gives in most relevant loading
cases the highest crack tip concentration KIV. A representative value for air in ceramics is b  0.001.
(b) Permeable conditions are attained at jC/j ? 1, i.e. the crack is fully penetrated by the electric field. The boundary con-
ditions are characterized by equal potentials at both crack faces. This condition also occurs, if the crack is closed.

uþ ¼ u ; Dþ2 ¼ D2 – 0: ð17Þ

(c) Semi-permeable crack: this is the most realistic picture of a medium with limited permittivity. In fact, we have the prob-
lem of a deformable medium inside the crack, coupled along the crack faces with the solid ceramic by electromechanical
interface conditions. According to the model of Hao and Shen [59] (earlier by [43]), the opposite crack faces are considered
as a set of parallel capacitors flushed by a vertical electric field E2, see Fig. 8. Making a body cut, the occurring electric dis-
placement component DC2 at the capacitor plates can be applied on the crack faces as distributed electrical charge density xcS .
This way, the capacitor analogy gives an implicit relation between the local displacement gap Du(x1) = d of the crack and the
charge density xcS as expressed by Eq. (18). For a Griffith crack, the model is exact and inside the crack a constant E2 field is
generated. In all other cases, the solution has to be found by an iterative numerical algorithm incorporated into a FEM-anal-
ysis [50,61–63]. Corresponding modifications of the J-integral are discussed in [55].

Duðx1 Þ ðDC2 Þ2
Dþ2 ¼ D2 ¼ DC2 ¼ xCS ¼ jC ; rC22 ¼ ð18Þ
Duðx1 Þ 2jC
Recently, Landis [63] suggested so-called ‘‘energetically consistent boundary conditions”, which resolve the deficiency of the
capacitor model that global and local energy release rates are not identical [60]. In addition to the above charges, this leads to
tractions rC22 on the crack faces (18), induced by the electric field inside the crack medium, too. Based on electrodynamical
Maxwell stresses at interfaces, Ricoeur [67] derived a more general relation for crack face tractions, yielding under simplified
assumptions to the same expression as Eq. (18).
(d) Conducting crack: if the crack interior is filled by a conducting medium, e.g. an electrolyte fluid or moist air, no electric
field can be built up inside. Therefore, the electric crack face boundary conditions are
uþ ¼ u ; E1 ¼ E3 ¼ 0; E2 – 0 ð19Þ

Fig. 8. The Iterative Capacitor Analogy (ICA).


316 M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326

The same comes true, if the crack faces are covered by partially debonded metallization, which can happen in laminates and
multi-layer devices. Another extreme case is an embedded electrode as shown in Fig. 1, though there is no opening displace-
ment [65,66]. Nevertheless, the electric field in the ceramic material becomes singular ahead of the crack tip and is quanti-
fied by the electric field intensity factor (EFIF):
pffiffiffiffiffiffiffiffiffi
K E ¼ lim 2pr E2 ðr; h ¼ 0Þ ð20Þ
r!0

One can summarize: the theoretical fundamentals and the solution techniques to analyze cracks in LPFM are well developed
and nearly completed. There exists an overwhelming amount of theoretical papers on sophisticated piezoelectric crack
configurations.

4. Fracture criteria for static electromechanical loading and experimental observations

4.1. Fracture criterion for a Griffith’s crack

A fracture criterion for crack initiation should be formulated in the general form:

^
BðK I ; K II ; K IV Þ ¼ B GIm ; GIIm ; Ge ¼ BC Pri ; Ei ; Da; . . . ð21Þ

The left hand side quantifies the driving force available for crack propagation, while the right hand side represents the mate-
rials resistance. On both sides of the criterion, now the electric field comes into play. On the applied loading side, electric
energy terms and intensity factors EDIF KIV are well defined. But the essential question is, how fracture toughness is con-
cerned? May fracture still be treated as a pure mechanical failure phenomenon or not?
For a basic understanding we consider the solution for an electromechanically loaded Griffith’s crack, see Fig. 9. The ana-
lytical solution for the SIF KI and the EDIF KIV is read:
pffiffiffiffiffiffi
KI ¼ r1
22  r22
C
pa in any case ðwith crack face tractionsÞ
1 pffiffiffiffiffiffi
K IV ¼ D2 pa impermeable DC2 ¼ rC22 ¼ 0
C pffiffiffiffiffiffi Y r1 þY ðD1 DC Þ ð22Þ
K IV ¼ ðD1
2  D2 Þ pa semi-permeable DC2 ¼ jC Y 42r122þY 44 D21 D2C
22 22 24 ð 2 2Þ

K IV ¼ f ðK I Þ permeable ðonly via piezoelectric effectÞ


In this special 2D case, when the crack lies perpendicular to the polarization, Eq. (11) for the ERR G can be expressed as fol-
lows by means of the two K-factors, where cT, e and j are material constants:
! !
1 K 2I K I K IV 1 K I K IV K 2IV
G ¼ Gm þ Ge ¼ þ þ  ð23Þ
2 cT e 2 e j
The electric contribution has a negative sign, i.e. it reduces the available total energy! If we formulate a fracture criterion
based either on G or on Gm, the influence of the applied field E1 2 (or KIV) can be found out as follows. The magnitudes of both
intensity factors are determined via Eq. (22) according to the crack face boundary conditions. In the impermeable case, the
behavior is completely symmetric with respect to the sign of E1 2 and gives the strongest electrical effect. Contrary, for per-
meable conditions (KIV  0), G is nearly independent on E1 2 . The semi-permeable solutions lie within these limits as can be
seen from Fig. 10 for a crack filled with air (jr = 1), silicon oil (jr = 2.5) or nearly permeable medium (jr = 20). Due to the

Fig. 9. Griffith crack with perpendicular poling and electric field.


M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326 317

Fig. 10. Energy release rates versus electric field for various relative permittivities jr of the Griffith’s crack in PZT 5H at constant remote stress r1
22 .

non-linear influence of crack face charges and tractions, the behavior of GðE1 2 Þ becomes asymmetric and the maximum is
shifted to higher values. Without electric field, all curves meet at the same point. For comparison, the pure mechanical part
Gm of ERR Eq. (23) is depicted as dashed lines.

4.2. Crack growth resistance curves

Since indentation fracture experiments suffer from complicated loading conditions, undefined crack geometries and
residual stresses, experiments with fracture test specimens are favored [8–15]. dos Santos e Lucato et al. [8] used CT-spec-
imens to investigate the fracture toughness (R-curves) of PZT PIC 151 by exclusively mechanical loading. The specimens were
poled in three different directions with respect to the crack, see Fig. 11. Their results show that poling along the crack front
leads to the highest toughening, whereas samples poled perpendicular to the crack orientation revealed lowest toughness
compared to the non-poled state. This behavior can be explained by domain switching in the process zone, which absorbs
more energy, if their orientation has to be turned. This is confirmed by Schneider [16], who annealed the specimens above
the Curie-temperature, whereby the toughening increase is completely lost. Also X-ray analyses [5,16] gave evidence of
stress-induced ‘self-adjusted’ domain reorientation at the growing crack.

Fig. 11. R-curves for PZT PIC151 specimens: poled parallel (A), perpendicular (C) and transversal (B) to the crack direction. Reference curve unpoled state
(X) [8].
318 M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326

Fracture experiments with DCB-specimens of BaTiO3 have been performed by Förderreuther et al. [15] allowing to inves-
tigate the influence of electric fields on the fracture toughness during stable crack growth. Whilst the crack is driven
mechanically by a wedge, an electric field was temporarily applied perpendicular to the crack. When the field was switched
on, a noticeable retardation of crack speed was observed. The appropriate fracture parameters were calculated by FEM-anal-
ysis for this experiment [85,67]. A high jump of KIV occurred during field action. According to (23), the total electromechan-
^
ical energy supply G ¼ ¼ Gem is reduced, whereas the pure mechanical contribution Gm is enhanced, as shown in Fig. 12. A
distinct increase of KI was determined, which has to be in equilibrium with the fracture toughness KIC(E, KIV). Thus, the exper-
imental findings can only be explained, if a toughening effect by the electric field (or KIV) is supposed due to domain
switching.

4.3. Fracture experiments at insulating cracks in CT-specimens

Park and Sun [9] carried out static fracture experiments with CT-specimens of PZT4 under defined electric fields, applied
via top and bottom electrodes, see Fig. 13. To ensure insulating crack face conditions, the specimen was immersed into sil-
icon oil. Fig. 14 gives the obtained loads at fracture, which show an asymmetry: a positive electric loading parallel to the
initial polarization enhances crack initiation, but a negative electric field impedes crack growth. The mechanical and total
energy release rates Gm and G were computed by means of FEM [9] as function of the electric field E, supporting the trend
expected from Eq. (23). The experimental failure load could not be described by a criterion, formulated with the total energy
release rate, since the electric field effect is too strong and symmetrical, see Fig. 14. However, the mechanical energy release
rate alone gave satisfactory results, Eq. (24).
^ ^
GIm ðF; EÞ ¼ GM
I ¼ GIc ¼ 2:34 N=m; GðF; EÞ ¼ GI ¼ GIc ? ð24Þ
Therefore, the authors concluded that the pure mechanical part Gm is a suitable fracture criterion. However, fracture tests of
other researchers [10,11] found no difference if the direction of the electric field was changed? It is also questionable,
whether the electrical and mechanical energy densities stored in a piezoelectric component may be physically treated as
independent.

Fig. 12. Mechanical and total energy release rate during crack growth [85].

Fig. 13. CT-specimen with mechanical and electric loading device. Poling and electric field are perpendicular to the crack [9].
M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326 319

Fig. 14. Critical fracture loads versus applied electric field obtained at CT-specimens of PZT4 [9].

Fig. 15. Poled SENB-specimen under four-point bending with a conducting crack [12].

4.4. Fracture experiments at conducting cracks in SENB-specimens

The critical fracture load of conducting cracks under simultaneously applied electric fields was measured by Heyer et al.
[12] in PZT PIC 151. Fig. 15 sketches the experimental set-up. Single edge notched beam specimens (SENB) with different
crack lengths a were tested in four-point bending. The poling direction and the applied electric field E are parallel to the
crack, which has been made conductive by infiltrating an electrolyte. The measured critical fracture loads Fc(E, a) depend
on the field strength E and the initial crack size a. In order to convert the external mechanical F and electrical loads E into
the KI and electrical intensity factors KE (see Eq. (20)), detailed FEM-analysis were performed yielding the geometry func-
tions. The measured critical fracture points were plotted in the ðK I ; K E Þ-space and fitted by a polynomial failure curve
f(KI, KE) = 0 (dashed line in Fig. 16). Next, the authors postulated a fracture criterion on the basis of the energy release rate
Eq. (23), which has for a conducting crack the form [12]:
1
G ¼ GIm þ Ge ¼ Y 22 K 2I þ 2Y 24 K I K E þ Y 44 K 2E ¼ GIc ¼ 12 J=m2 ð25Þ
2
Eq. (25) is plotted in Fig. 16 as solid line. It matches the failure curve not completely, which was explained by shielding ef-
fects due to domain switching.
In conclusion we can state: up to now, the experimental observations known from literature do not allow a unique under-
standing and are often hard to compare. They are sometimes difficult to interpret and give even contradictory results. The
reason lies in quite diverse ceramic compositions used (soft or hard), in dissimilar test geometries and electromechanical
loading conditions, in unlike magnitudes of the electric field compared to coercive strength and various other factors. Fur-
thermore, the measurement of electrical quantities (stored energy, charge flow) is quite complicated. Therefore, at present
exists no common sense (not to mention any standard) how to measure fracture toughness KIc under monotonic electrome-
chanical loading conditions in piezoelectric materials.

5. Dynamic electrical and/or mechanical loading

Many piezoelectric devices are exposed to pulse-like or high frequency electric loading, acting with light speed much fas-
ter than mechanical impacts. Such severe electrically induced shocks play an important role and are a new challenge as well.
Despite of this, the dynamic fracture mechanics of piezoelectric structures under transient electrical and mechanical loading
320 M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326

Fig. 16. Comparison between experimental (———) and theoretically predicted (–––) fracture curves according to Eq. (25) for Gc = 12 J/m2 for a conducting
crack under four-point bending [12].

has been rarely investigated. Due to mathematical obstacles, most of the dynamic semi-analytical solution methods are re-
stricted to anti-plane fracture. Using integral transformations, Shindo et al. [68] and Wang and Yu [69] examined the re-
sponse of cracks in piezoelectrics under mode I impact loading. Recently, numerical methods like finite elements (FEM)
[70,81] and boundary elements (BEM) [71,72] have been presented.
Enderlein et al. [70] developed a finite element code based on an explicit time integration for transient loading of 2D and
3D piezoelectric structures. To account for the semi-permeability crack face conditions, the interior of the crack was virtually
meshed with one-dimensional capacitor elements, see Eq. (18). In order to calculate mechanical and electric intensity fac-
tors, the interaction integral equation (15) was extended for transient electromechanical problems. Interpenetration of crack
faces during oscillations is avoided by a contact algorithm. The features are demonstrated by a 2D example dealing with a
stationary central crack in a rectangular plate under impact load by either pure remote stress r0(t) or electric displacement
D0(t). The applied loads follow a ramp function within the time interval of s ¼ 106 s up to the maximum values
 0 ¼ 5:0 103 C=m2 or r
D  0 ¼ 5:0 MPa. The material properties are BaTiO3, poling direction and loading are parallel, as in
Fig. 9. The calculated stress intensity factor KI is plotted as function of time t for both load cases in Fig. 17. Under mechanical
loading, the dynamic stress intensity factor overshoots the corresponding static value (by a factor of 2.4), which is known
from LEFM. Concerning the electrical loading, a high KI-value is found as well, which is contrary to the related static problem,
where KI = 0. This surprising effect can be explained as follows: if the electrical loading is switched on, the electric field will
propagate through the whole body with speed of light. The induced mechanical fields propagate as waves with sound speed

Fig. 17. SIF versus time for mechanical or electrical impact.


M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326 321

starting from the load free edges of the plate. When the generated stress waves reach the crack, a KI – 0 raises up and falls
down with the declining impulse. For both loading types, there is no significant difference between the impermeable and the
permeable crack.
One can conclude: dynamic loading of stationary piezoelectric cracks reveals several interesting new effects. Unfortu-
nately, no fracture experiments have been published concerning impact loading of piezoelectric ceramics.

6. Non-linear piezoelectric fracture mechanics (NPFM)

6.1. Macroscopic description

Many experimental findings and microstructural diagnostics give evidence that at the crack tips in piezoelectric materials
non-linear effects occur, which are caused by the high electromechanical field concentration. In a macroscopic treatment,
this is reflected by the polarization and strain-electric field hysteresis loops of Fig. 3. The first idea was, to adopt the concepts
of elastic–plastic fracture mechanics. So one can estimate the size of the ‘‘electrically yielded” zone in analogy to Irwin’s plas-
tic zone model by means of the coercive strength EC:
 2  2
1 KE ^ 1 K IV
ry ¼ ¼ ð26Þ
2p EC 2p jEC
As long as the radius ry is much smaller than the size RK of that region, where the linear asymptotic near field of Eq. (4) dom-
inates at the crack tip, the loading is controlled by the intensity factors KN(N = I, II, III, IV). That means, all non-linear and
micromechanical phenomena can be attributed to the materials toughness quantity BC in Eq. (21). In this case we speak
about small scale switching zones, which is illustrated in Fig. 18.
For large scale switching, when ry >> RK or even reaches the specimen dimensions, adequate solutions are still missing.
Analogous to the Dugdale model, Gao et al. [73] and Wang [74] developed an electric strip saturation model, whereby
the ceramic is treated as a mechanically brittle but electrically ductile material. However, there are some important aspects
distinguishing the non-linearity of ferroelectric materials from the one of ductile metals:

 The non-linear effect is always anisotropic and mainly triggered by another (? electric) physical field.
 The non-linearity is confined to the region of the hysteresis loop. At larger amplitudes of field and strain, a saturation or
‘‘locking” into a somehow linear behavior happens.
 Unlike plastic deformation by dislocation kinetics, the physical mechanism is a spontaneous phase transformation – the
so-called domain switching – induced either by mechanical or electrical fields.

From these arguments it becomes clear, why up to now no characteristic non-linear asymptotic field like the HRR-singu-
larity in ductile fracture could be found for ferroelectrics and why the classical J-concept of elastic–plastic fracture mechan-
ics could not be copied.

6.2. Micromechanical domain switching models

However, the major obstacle seems to be that the domain switching processes considerably affect the materials fracture
resistance itself, which is obviously one reason for the discrepancies between LPFM-criteria and experimental observations.
Therefore, a multi-scale modeling is necessary to make progress in NPFM, as was pointed out by Fulton and Gao [87,88]. In-
side a grain, ferroelectric domains are arranged according to the crystallographic orientations with a specific spontaneous

Fig. 18. Domain switching zone (2) and fracture process zone (1) inside the near tip field (3).
322 M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326

Fig. 19. Elementary switching steps by 90° and 180°.

polarization vector P0. There are various species of domains possible, which are separated by domain walls of 90° or 180°.
During poling the domains realign themselves in the direction of the applied external electric field, which is called domain
switching. Thus a resulting global polarization vector is developed, forming the macroscopic hysteresis, as illustrated in Fig. 3.
The elementary switching step of a domain in the 2D plane perpendicular to a crack is explained in Fig. 19. The c-axes of a
cell can rotate by an angle u = ±90° or 180°, i.e. three switching variants exist. During switching not only the local state of
polarization is rotated, but also the strain state is changed spontaneously, because the axes a and c of the tetragonal unit cell
differ for BaTiO3 by: eD = (c  a)/a0  0.005. This is described by the polarization switch vector DPi and the inelastic devia-
toric switching strain tensor Deij. An energetic switching criterion was suggested in [75] to determine whether switching
occurs or not:

rij Deij þ Ei DPi


2EC P0 ð27Þ
Assuming as first approximation that stresses rij and electric fields Ei remain unchanged during switching, the left hand side
of Eq. (27) represents the specific work dissipated during switching. The threshold value on the right hand side represents
approximately half of the area of a polarization hysteresis.
At present intensive research is invested to develop multi-scale constitutive material laws for ferroelectrics, based on
micromechanical simulation of domain processes at various levels of complexity [76–80]. Only some first attempts were
made to apply such laws to the fracture process zone [81].
On the other hand, much simpler domain switching models have been proposed to estimate crack tip process zones in
ferroelectrics, which can successfully predict possible toughening or softening effects [82–84]. As an example, the model
of Ricoeur and Kuna [67,85] is presented in more detail. If small scale switching is presumed, the mechanical and electrical
fields around the crack tip are determined by the KN-factors and by Eq. (4). This is indicated by contour (3) in Fig. 18. The size
and shape of the zone (2) around the crack tip, where switching is to occur, can be calculated by inserting these fields into
the switching criterion Eq. (27). The idea is illustrated in Fig. 18. Now, the explained switching kinematics of Fig. 19 is applied
to a crack under constant mode I mechanical load (KI  KIc) and varying electric loads, quantified by E1 2 or KIV. Fig. 20 shows
the calculated switching zones. In case of parallel poling the switching zone is much larger, since the high normal stresses in
front of the crack activate a ±90° reorientation. The absolute size of 45 lm corresponds quite well with experimental obser-
vations in BaTi03.
In the next step, the inhomogeneous polarization strain state generated inside (2) is regarded as a residual strain field
affecting the crack tip loading K tip
N , see Fig. 18. For this purpose, the concept of transformation toughening in structural

Fig. 20. Domain switching zones for parallel (full lines) and perpendicular (dashed line) initial poling to the crack.
M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326 323

Fig. 21. Apparent increase of fracture toughness due to domain switching zones for various applied electric fields.

ceramics [86] was transferred to piezoelectrics. Using the piezoelectric law, the tractions ti can be obtained, acting along the
switching zone boundary (2). By means of crack weight functions hiN(x) for piezoelectric materials [90], the influence of
these tractions on the stress intensity factors DKN at the real crack tip can be computed as follows:
Z
DK N ¼ t i ðxÞhiN ðxÞds ð28Þ
ð2Þ

The total field intensity factors K tip 1


N are a superposition of DKN and the contributions from the remote loads K N . So, the DKN
supply an additional loading or unloading of the direct crack tip region (1) in Fig. 18, where the real fracture of the ceramics
occurs. Restricting further more to mode (I, IV)-loading, a brittle fracture criterion can be defined, based entirely on the
mechanical stresses or the stress intensity factor KI:

K tip tip 1
I ¼ K Ic ¼ K I þ DK I ð29Þ
This is justified, since the fracture toughness K tip
Ic represents an inherent parameter for cleavage fracture of the material under
any circumstances. According to this criterion, the influence of electric fields and domain switching is accounted for in terms
of DKI on the right hand side of Eq. (29):
tip

K1 1 1 1
I ¼ K IC  DK I ¼ K IC Da; w; K I ; K IV ð30Þ
Depending on the sign of DKI, it reduces or enhances the apparent fracture load measured in an experiment. From the results,
a screening or amplifying effect is found. Therefore, it can also be interpreted as a microstructural ‘‘switching induced” con-
tribution to the fracture toughness. Fig. 21 shows the influence of the magnitude of an applied electric field E1 2 on the sim-
ulated crack resistance curves of BaTiO3. Compared to earlier approaches [82,100], this micromechanical model accounts for
the complete piezoelectric field coupling and weight functions.

7. Cyclic electrical and/or mechanical loading

In many technical applications, e.g. as multi-layer actuators for fuel injection pistons, the piezoelectric ceramics are ex-
posed to cyclic mechanical, electrical or combined electromechanical loading. The problem of mechanical fatigue and sub-
sequent fatigue crack growth, induced by alternating electric fields was recognized already by Uchino [4] and Cao and Evans
[7]. During the last decade, among others the research groups around Lynch, Yang, Fang, Rödel have performed particular

Fig. 22. DCB-specimen for cyclic electric field with mechanical preload (thickness b = 1, 5 mm).
324 M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326

0
10 Applied Field Amplitude
[MV/m]:

0.8 1.4

da/dN [mm/cycle]
-1 0.9 1.5
10 1.0 1.7
1.1 1.8
1.2 1.9
1.3
-2
10 Fit according to
Paris Law

-3
10 -4 -3 -2 -1 0
10 10 10 10 10
-3/2
Δ K IV [C m ]

Fig. 23. Double-logarithmic plot of crack growth rate versus cyclic EDIF for different amplitudes. Fitted by a Paris law Eq. (31).

experimental and theoretical studies [91–100]. A comprehensive review was given by Yang et al. [33] with forthcoming
literature. It is experimentally evident [92] and generally accepted that the failure mechanism in electrically cycled ferro-
electrics results from domain switching, too. According to [96,97] crack growth starts at that point of the load cycle, where
the highest reverse poling of domains occurs. The associated cyclic inelastic deformations in the fracture process zone cause
high local stress peaks, i.e. the induced stress intensity factor exceeds the fracture toughness periodically. It was observed
that the amplitude of the applied electric field has to reach approximately the coercive field strength EC to activate this crack
propagation mechanism. First attempts to explain and to simulate fatigue mechanisms by domain processes were done in
[95].
In the following some results of combined experimental–numerical investigations of fatigue crack growth in PZT PIC151
ceramics under cyclic electric loading plus static mechanical preloads are reported [98,99]. The fatigue tests were performed
on poled DCB-specimens with a sharp starting notch, see Fig. 22, which provided the necessary conditions for a quantitative
fracture mechanical evaluation. In the experiments, the crack propagation rates were measured for different combinations of
mechanical and electrical loads. The crack growth rate increases with the electric field amplitude, more precisely with its
surplus to the coercive strength EC. An additional mechanical force supports accelerated crack growth.
To determine the dependency of both intensity factors KI and KIV on applied loads F and E for this specimen type, a numer-
ical analysis of the coupled electromechanical field problem was carried out [99]. KIV was found to be independent on crack
length, which is an advantage of this test geometry. Below an applied field Eclo ¼ 0:92EC , the crack remains closed and KIV is
zero. This represents a threshold value for an electrostatic crack closure effect. Using the derived correlation KIV = KIV(KI, Eclo),
it could be shown that the crack growth rate is particularly a function of the amplitude of EDIF DKIV. Therefore, a quantitative
description was suggested by means of a power law as the Paris-equation in classical fatigue:
da
¼ A DK m
IV with A ¼ 106:33 and m ¼ 2:48 ð31Þ
dN
This relation agrees well (fitted line) with all experimental data plotted in Fig. 23 for the whole range of applied field ampli-
tudes (symbols).

8. Conclusions and outlook

It can be summarized that the theoretical fundamentals of linear piezoelectric fracture mechanics (LPFM) have been suc-
cessfully established to date. Important aspects as the electromechanical crack tip fields, the relevant fracture parameters,
the role of electric crack face boundary conditions and mode mixture have been recognized and solved. Meanwhile, also
numerical methods as FEM and BEM are available to analyze cracks in complicated real piezoelectric structures exposed
to complex electromechanical loading situations.
In contrast to that, the achievements in experimental fracture investigations are relatively poor. Because of many influ-
encing factors, the obtained results are difficult to interpret and sometimes contradictory. Up to now it seems that no uni-
versal fracture criterion was found, which can explain all observed linear piezoelectric fracture phenomena and
measurement data on a unique common basis. Thus, approved concepts for practical assessment of fracture in piezoelectric
components are still missing! Therefore, further research has to be focused to close the gap between theoretical predictions and
experimental observations in piezoelectric fracture mechanics. This requires much more experimental work under well-de-
fined boundary conditions and a careful numerical analysis of the electromechanical loading situation at the crack.
M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326 325

On the other hand, many fracture phenomena under large scale switching (operation of components at large alternating
signals) are much less understood. At present, non-linear piezoelectric fracture mechanics (NPFM) is still under research and
not any convincing fracture mechanical concept is developed. One great challenge consists in finding the asymptotic fields at
the crack tip, which requires the knowledge of appropriate non-linear constitutive laws, either on a phenomenological or
better on a micromechanical basis. A key issue is to elucidate the important influence of domain switching processes on
the fracture behavior.
Concerning fatigue crack growth under combined electromechanical loading, there are first promising theoretical and
experimental investigations. Again, the key to further improvements lies in the understanding of domain switching behavior
in the process zone during cyclic fully reverse loading. Another open question is the choice of the criterion, which determines
the direction of fatigue crack propagation under mixed electromechanical situations.
The majority of research up to now has been devoted to cracks of macroscopic size in order to achieve basic understanding
and to transfer classical fracture mechanics principles to piezoelectric ceramics. Contrary to that, the real failure mechanisms
in piezoelectric components are controlled by defects of microscopic size. To make better predictions of practical concern as
residual life time and reliability, more emphasis should be drawn to the early stages of micro-crack formation and damage.

Acknowledgements

Many thanks are given to my co-workers Dr. A. Ricoeur, Dr. M. Enderlein, Dr. M. Abendroth and K. Wippler for the very
fruitful cooperation. I gratefully acknowledge the German Research Foundation (DFG), which supported this research by sev-
eral projects.

References

[1] Gabbert U, Tzou H. Smart structures and structonic systems. In: Proceedings of IUTAM-Symposium Magdeburg 2000. Dordrecht: Springer; 2001.
[2] Yang W. Mechanics and reliability of actuating materials. In: Proceedings of IUTAM-Symposium Beijing 2004. Dordrecht: Springer; 2006.
[3] Winzer SR, Shankar N, Ritter AP. Designing co-fired multi-layer electrostrictive actuators for reliability. J Am Ceram Soc 1989;72:2246–57.
[4] Uchino K. Materials issues in design and performance of piezoelectric actuators: an overview. Acta Mater 1998;48:3745–53.
[5] Mehta K, Virkar AV. Fracture mechanisms in ferroelectric–ferroelastic lead zirconate titanate (Zr:Ti = 0.54:0.46) ceramics. J Am Ceram Soc
1990;73:567–74.
[6] Schneider GA, Heyer V. Influence of the electric field on Vickers indentation crack growth in BaTiO3. J Eur Ceram Soc 1999;19:1299–306.
[7] Cao HC, Evans AG. Electric-field-induced fatigue crack growth in piezoelectrics. J Am Ceram Soc 1994;77:1783–6.
[8] dos Santos e Lucato SL, Lupascu DC, Rödel J. Effect of poling direction on R-curve behavior in lead zirconate titanate. J Am Ceram Soc 2000;83:424–6.
[9] Park SB, Sun CT. Fracture criteria for piezoelectric ceramics. J Am Ceram Soc 1995;78:1475–80.
[10] Fu R, Zhang TY. Effects of an electric field on the facture toughness of poled lead zirconate titanate ceramics. J Am Ceram Soc 2000;83:1215–8.
[11] Wang H, Singh RN. Crack propagation in piezoelectric ceramics: effect of applied electric fields. J Appl Phys 1997;81:7471–9.
[12] Heyer V, Schneider GA, Balke H, Drescher J, Bahr HA. A fracture criterion for conducting cracks in homogeneously poled piezoelectric PZT-PIC151
ceramics. Acta Mater 1998;46:6615–22.
[13] Zhang TY, Zhao M, Guoning L. Failure behavior and failure criterion of conductive cracks (deep notches) in piezoelectric ceramics. I. The charge free
zone model. II. Experimental verification. Acta Mater 2004;52:2013–35.
[14] Lynch CS. Fracture of ferroelectric and relaxor electro-ceramics. Acta Mater 1998;46:599–608.
[15] Förderreuther A, Thurn G, Zimmermann A, Aldinger F. R-curve effect, influence of electric field and process zone in BaTiO3 ceramics. J Eur Ceram Soc
2002;22:2023–31.
[16] Schneider GA. Influence of electric field and mechanical stresses on the fracture of ferroelectrics. Annu Rev Mater Res 2007;37:491–538.
[17] Parton VZ. Fracture mechanics of piezoelectric materials. Acta Astron 1976;3:671–83.
[18] Deeg WF. Analysis of dislocation, cracks, and inclusion problems in piezoelectric solids. PhD Thesis. Stanford, CA: Stanford University; 1980.
[19] Sosa H, Pak YE. Three-dimensional eigenfunction analysis of a crack in a piezoelectric material. Int J Solid Struct 1990;26:1–15.
[20] Suo Z, Kuo CM, Barnett DM, Willis JR. Fracture mechanics for piezoelectric ceramics. J Mech Phys Solid 1992;40:739–65.
[21] Sosa H. On the fracture mechanics of piezoelectric solids. Int J Solid Struct 1992;29:2613–22.
[22] Pak YE. Linear electro-elastic fracture mechanics of piezoelectric materials. Int J Fract 1992;54:79–100.
[23] Park SB, Sun CT. Fracture criteria for piezoelectric ceramics. Int J Fract 1995;70:203–16.
[24] Xu X-L, Rajapakse RKND. Analytical solution for an arbitrarily oriented void/crack and fracture of piezoceramics. Acta Mater 1999;47:1735–47.
[25] Xu X-L, Rajapakse RKND. A theoretical study of branched cracks in piezoelectrics. Acta Mater 2000;48:1865–82.
[26] Chen WQ, Shioya T. Fundamental solution for a penny-shaped crack in a piezoelectric medium. J Mech Phys Solid 1999;47:1459–75.
[27] Wang B. Three-dimensional analysis of a flat ellipsoidal inclusion in a piezoelectric material. Int J Solid Struct 1992;30:781–91.
[28] Wang ZK, Huang SH. Fields near elliptical crack in piezoelectric ceramics. Engng Fract Mech 1995;51:447–56.
[29] Qin QH. Fracture mechanics of piezoelectric materials. Southampton: WIT Press; 2001.
[30] Zhang TY, Zhao MH, Tong P. Fracture of piezoelectric ceramics. Adv Appl Mech 2002;38:147–289.
[31] Chen YH, Lu TJ. Cracks and fracture in piezoelectrics. Adv Appl Mech 2003;39:121–215.
[32] Chen YH, Hasebe N. Current understanding on fracture behavior of ferroelectric/piezoelectric materials. J Intell Mater Syst Struct 2005;16:673–87.
[33] Yang W, Fang F, Fang DN. Fracture and fatigue of ferroelectrics. In: Milne I, Ritchie RO, Karihaloo B, editors. Comprehensive Structural
Integrity. Amsterdam: Elsevier Pergamon; 2003. p. 645–86.
[34] Kuna M. Finite element analyses of cracks in piezoelectric structures: a survey. Arch Appl Mech 2006;76:725–45.
[35] Kumar S, Singh RN. Crack propagation in piezoelectric materials under combined mechanical and electric loadings. Acta Mater 1996;44:173–200.
[36] Kumar S, Singh RN. Energy release rate and crack propagation in piezoelectric materials, Part II, combined mechanical and electric loads. Acta Mater
1997;45:849–68.
[37] Kuna M. Finite element analyses of crack problems in piezoelectric structures. Comput Mater Sci 1998;13:67–80.
[38] Banks-Sills L, Motola Y, Shemesh L. The M-integral for calculating intensity factors of an impermeable crack in a piezoelectric material. Engng Fract
Mech 2008;75:901–25.
[39] Shang F, Kuna M, Abendroth M. Finite element analyses of three-dimensional crack problems in piezoelectric structures. Engng Fract Mech
2003;70:143–60.
[40] Kemmer G. Berechnung von elektromechanischen Intensitätsparametern bei Rissen in Piezokeramiken. Dissertation, TU Dresden, 2000. VDI-Verlag
Düsseldorf, Reihe 18, Nr. 261.
326 M. Kuna / Engineering Fracture Mechanics 77 (2010) 309–326

[41] Groh U, Kuna M. Efficient boundary element analysis of cracks in 2D piezoelectric structures. Int J Solid Struct 2004;42:2399–416.
[42] Sanz JA, Ariza MP, Dominguez J. Three-dimensional BEM for piezoelectric fracture analysis. Engng Anal Bound Elem 2005;26:586–96.
[43] Parton VZ, Kudryavtsev BA. Electromagnetoelasticity. New York: Gordon and Breach; 1988.
[44] Eringen AC, Maugin GA. Electrodynamics of Continua I: foundations and solids media. Springer; 1990.
[45] Kamlah M. Ferroelectric and ferroelastic piezoceramics – modeling of electromechanical hysteresis phenomena. Continuum Mech Thermodyn
2001;13:219–68.
[46] Ikeda T. Fundamentals of piezoelectricity. Oxford University Press; 1990.
[47] Pohanka RC, Smith PL. Electronic ceramics. Properties. Devices and applications. In: Levinson LM, editor. Recent advances in piezoelectric ceramics,
electronic ceramics. New York: Marcel Dekker; 1988 [chapter 2].
[48] Cherepanov GP. Invariant G-integrals and their application in mechanics. Prikl Mat Mech 1977;41:399–412 [in Russian].
[49] Pak YE, Herrmann G. Crack extension force in a dielectric medium. Int J Engng Sci 1986;24:1375–88.
[50] McMeeking RM. A J-integral for the analysis of electrically induced mechanical stress at cracks in elastic dielectrics. Int J Engng Sci 1990;28:605–13.
[51] McMeeking RM. Crack tip energy release rate for a piezoelectric compact tension specimen. Engng Fract Mech 1999;64:217–44.
[52] McMeeking RM. Towards a fracture mechanics for brittle piezoelectric and dielectric materials. Int J Fract 2001;108:25–41.
[53] Dascalu C, Maugin GA. On the energy of electroelastic fracture. Z Angew Math Phys 1995;46:355–65.
[54] Abendroth M, Groh U, Kuna M, Ricoeur A. Finite element computation of the electromechanical J-integral for 2D and 3D crack analysis. Int J Fract
2002;114:359–78.
[55] Ricoeur A, Enderlein M, Kuna M. Calculation of the J-integral for limited permeable cracks in piezoelectrics. Arch Appl Mech 2005;74:536–49.
[56] Zhang TY. J-integral measurement for piezoelectric materials. Int J Fract 1994;68:R33–40.
[57] Chen YH, Lu TJ. Recent developments and applications of invariant integrals. Appl Mech Review 2003;56:515–52.
[58] Gao CF, Fan JX. Exact solution for the plane problem in piezoelectric materials with an elliptic hole or a crack. Int J Solid Struct 1999;36:2527–40.
[59] Hao T-H, Shen Z-Y. A new electric boundary condition of electric fracture mechanics and its application. Engng Fract Mech 1994;47:793–802.
[60] McMeeking RM. The energy release rate for a Griffith crack in a piezoelectric material. Engng Fract Mech 2004;71:1149–63.
[61] Gruebner O, Kamlah M, Munz D. Finite element analysis of cracks in piezoelectric materials taking into account the permittivity of the crack medium.
Engng Fract Mech 2003;70:1399–413.
[62] Wippler K, Ricoeur A, Kuna M. Towards the computation of electrically permeable cracks in piezoelectrics. Engng Fract Mech 2004;71:2567–87.
[63] Landis CM. Energetically consistent boundary conditions for electromechanical fracture. Int J Solid Struct 2004;41:6291–315.
[64] Schneider GA, Felten F, McMeeking RM. The electrical potential difference across cracks in PZT measured by Kelvin Probe Microscopy and the
implications for fracture. Acta Mater 2003;51:2235–41.
[65] Ru CQ, Mao XJ. Conducting cracks in a piezoelectric ceramic of limited electrical polarization. Mech Phys Solid 1999;47:2125–46.
[66] Ru CQ. Exact solution for finite electrode layers embedded at the interface of two piezoelectric half-planes. J Mech Phys Solid 2000;48:693–708.
[67] Ricoeur A. Theoretische Untersuchungen zum Bruchverhalten ferroelektrischer Keramiken bei elektromechanischer Beanspruchung, Habilitation. TU
Bergakademie Freiberg; 2007.
[68] Shindo Y, Narita F, Ozawa E. Impact response of a finite crack in an orthotropic piezoelectric ceramic. Acta Mech 1999;137:99–107.
[69] Wang X, Yu S. Transient response of a crack in piezoelectric strip subjected to the mechanical and electrical impacts: mode-I problem. Mech Mater
2001;33:11–20.
[70] Enderlein M, Ricoeur A, Kuna M. Finite element techniques for dynamic crack analysis in piezoelectrics. Int J Fract 2005;134:191–208.
[71] Garcia-Sanchez F, Zhang Ch, Sladek J, Sladek V. 2D transient dynamic crack analysis in piezoelectric solids by BEM. Comput Mater Sci
2007;39:179–86.
[72] Saez A, Garcia-Sanchez F. Dominguez, Hypersingular BEM for dynamic fracture in 2D piezoelectric solids. J Comput Meth Appl Mech Engng
2006;196:235–46.
[73] Gao H, Zhang TY, Tong P. Local and global energy release rates for an electrically yielded crack in a piezoelectric ceramic. J Mech Phys Solid
1997;45:451–91.
[74] Wang TC. Analysis of strip electric saturation model for crack problem in piezoelectric materials. Int J Solid Struct 2000;37:6031–49.
[75] Hwang SC, Huber JE, McMeeking RM, Fleck NA. The simulation of switching in polycrystalline ferroelectric ceramics. J Appl Phys 1998;84:1530–40.
[76] Kessler H, Balke HJ. On the local and average energy release in polarization switching phenomena. Mech Phys Solid 2001;49:953–78.
[77] Lynch CS. On the development of multiaxial phenomenological constitutive laws for ferroelectric ceramics. J Intell Mater Syst Struct 1998;9:555–63.
[78] Landis CM. Fully coupled, multi-axial, symmetric constitutive laws for polycrystalline ferroelectric ceramics. J Mech Phys Solid 2002;50:127–52.
[79] Li F, Fang D. Simulations of domain switching in ferroelectrics by a three-dimensional finite element model. Mech Mater 2004;36:959–73.
[80] Kamlah M, Liskowsky AC, McMeeking RM, Balke H. Finite element simulation of a polycrystalline ferroelectric based on a multidomain single crystal
switching model. Int J Solid Struct 2005;42:2949–64.
[81] Enderlein M. Finite-Elemente-Verfahren zur bruchmechanischen Analyse von Rissen in piezoelektrischen Strukturen bei transienter
elektromechanischer Belastung. PhD Thesis. TU Bergakademie Freiberg; 2007.
[82] Zhu T, Yang W. Toughness variation of ferroelectrics by polarization switch under non-uniform electric field. Acta Mater 1997;45:4695–702.
[83] Rajapakse RKND, Zeng X. Toughening of conducting cracks due to domain switching. Acta Mater 2001;49:877–85.
[84] Ricoeur A, Kuna M. A micromechanical model for the fracture process zone in ferroelectrics. Comput Mater Sci 2003;27:235–49.
[85] Ricoeur A, Kuna M. Influence of electric fields on the fracture of ferroelectric ceramics. J Eur Ceram Soc 2003;23:1313–28.
[86] McMeeking RM, Evans AG. Mechanics of transformation toughening in brittle materials. J Am Ceram Soc 1982;65:242–6.
[87] Fulton CC, Gao H. Effect of local polarization switching on piezoelectric fracture. J Mech Phys Solid 2001;49:927–52.
[88] Fulton CC, Gao H. Microstructural modeling of ferroelectric fracture. Acta Mater 2001;49:2039–54.
[89] Jelitto H, Felten F, Häusler C, Kessler H, Balke H, Schneider GA. Measurement of energy release rates for cracks in PZT under electromechanical loads.
J Europ Ceram Soc 2005;25:2817–20.
[90] McMeeking R, Ricoeur A. The weight function for cracks in piezoelectrics. Int J Solid Struct 2003;40:6143–62.
[91] Lynch CS, Yang W, Collier L, Suo Z, McMeeking RM. Electric field induced cracking in ferroelectric ceramics. Ferroelectrics 1995;166:11–30.
[92] Tan X, Xu Z, Shang JK. Direct observations of electric field-induced domain boundary cracking in (0 0 1) oriented piezoelectric Pb(Mg1/3Nb2/3)O3–
PbTiO3 single crystal. Appl Phys Lett 2000;77:1529–31.
[93] Jiang LZ, Sun CT. Crack growth behavior in piezoelectrics under cyclic loads. Ferroelectrics 1999;233:211–23.
[94] Liu B, Fang DN, Hwang K-C. Electric-field-induced fatigue crack growth in ferroelectric ceramics. Mater Lett 2002;54:442–6.
[95] Mao GZ, Fang DN. Fatigue crack growth induced by domain switching under electromechanical load in ferroelectrics. Theor Appl Fract Mech
2004;41:115–23.
[96] Fang DN, Liu B, Sun CT. Fatigue crack growth in ferroelctric ceramics driven by alternating electric fields. J Am Ceram Soc 2004;87:840–6.
[97] Westram I, Oates WS, Lupascu DC, Rödel J, Lynch CS. Mechanism of electric fatigue crack growth in lead zirconate titanate. Acta Mater
2007;55:301–12.
[98] Westram I, Ricoeur A, Emrich A, Rödel J, Kuna M. Fatigue crack growth law for ferroelectrics under cyclic electrical and combined electromechanical
loading. J Eur Ceram Soc 2007;27:2485–94.
[99] Ricoeur A, Enderlein M, Kuna M. Fatigue crack growth in ferroelectrics under cyclic electrical loading. In: Armstrong WD, editor. Smart structures and
materials 2006, Active materials: behavior and mechanics (Proc SPIE), vol. 6170. p. 0C1–12.
[100] Zhu T, Yang W. Fatigue crack growth in ferroelectrics driven by cyclic electric loading. J Mech Phys Solid 1999;47:81–97.

You might also like