Materials and Design 152 (2018) 54–64

Contents lists available at ScienceDirect

Materials and Design

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

Failure mode, ferroelastic behavior and toughening effect of bismuth

titanate ferroelectric ceramics under uniaxial compression load
Yu Chen a,b, Jiageng Xu c, Shaoxiong Xie d, Rui Nie b, Jing Yuan b, Qingyuan Wang a,d,e,⁎, Jianguo Zhu b,⁎⁎
a
School of Mechanical Engineering, Chengdu University, Chengdu 610106, China
b
College of Materials Science and Engineering, Sichuan University, Chengdu 610065, China
c
School of Architecture and Civil Engineering, Chengdu University, Chengdu 610106, China
d
College of Architecture and Environment, Sichuan University, Chengdu 610065, China
e
Failure Mechanics and Engineering Disaster Prevention and Mitigation Key Laboratory of Sichuan Province, Chengdu 610065, China

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Two types of microcracks (opening and

closed) initiated in the matrix of
Bi4Ti3O12 ceramic during mechanical
loading.
• The underlying micromechanism of
ferroelastic domain switching was de-
scribed by the non-180o domain wall
motion.
• The constitutive relation of ceramics
under uniaxial compression load was
obtained by a domain distribution
model.
• The expression of ferroelasticity in-
duced toughening effect (ΔГ/Г) was de-
duced from the indentation fracture
mechanics.

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

Article history: A typic BLSF ceramics: bismuth titanate (Bi4Ti3O12), whose failure mode, ferroelastic behavior and toughening effect
Received 20 January 2018 under uniaxial compression load were comprehensively assessed. Firstly, a series of crystallographic parameters in-
Received in revised form 9 April 2018 cluding lattice constants (a, b, c and v), orthorhombicity (g) and single-crystal distortion (S0lattice) were calculated
Accepted 19 April 2018
from XRD patterns. The ferroelastic behavior referred to the 90° switching of domains with compression stress
Available online 24 April 2018
was identified by the nonlinear stress-strain curve, and the underlying micromechanism was more accurately de-
Keywords:
scribed by the non-180° domain wall motion. SEM observation on the fracture surface of the broken sample reveals
Bi4Ti3O12 ceramics its failure mode in terms of microcrack initiation and propagation behaviors. Further, a simplified domain distribu-
Ferroelastic domain switching tion model was constructed for unpoled ferroelectric ceramics based on a constitutive framework, which deduced
Microcracks their constitutive relation subjected to uniaxial compression load. The ferroelasticity induced toughening was ver-
Compression load ified by indentation test, and further analyzed by a micromechanics model of crack propagation. The expression of
Toughening effect toughening effect (ΔГ/Г) was also deduced from the indentation fracture mechanics. Finally, the evolution in me-
chanical properties (including coercive stress (σc), switching strain (εzz), apparent elastic modulus (E) and compres-
sion strength (σcf)) of Bi4Ti3O12 ceramics with the doping content of W was also quantified by the compression test.
© 2018 Elsevier Ltd. All rights reserved.

⁎ Correspondence to: Q. Wang, School of Mechanical Engineering, Chengdu University, Chengdu 610106, China.
⁎⁎ Corresponding author.
E-mail addresses: wangqy@scu.edu.cn, (Q. Wang), nic0400@scu.edu.cn (J. Zhu).

https://doi.org/10.1016/j.matdes.2018.04.055
0264-1275/© 2018 Elsevier Ltd. All rights reserved.
 Y. Chen et al. / Materials and Design 152 (2018) 54–64
1. Introduction
Ferroelectric materials are one important category of functional mate-
rials with the piezoelectric effect which enables the conversion between
mechanical energy and electrical energy, which have been commonly
used as sensitive element for many electric devices including piezoelectric
sensors, electromechanical transducer, non-volatile memories, etc. [1]. It
is the essential attribute of ferroelectrics that the domain state can be
modified by electrical or mechanical loadings of sufficient magnitude,
which gives rise to the reorientation of polarization/strain along external
loadings. Specifically, domains can be switched 180° or 90° from the orig-
inal direction by an electric field, which is called “ferroelectricity (or ferro-
electric behavior)”, and domains can be switched only 90° by a
mechanical stress, which is so-called “ferroelasticity (or ferroelastic be-
havior)” [2]. In most piezoelectric sensors and actuators, ferroelectric ma-
terials are prone to fatigue due to cyclic electrical or mechanical loadings.
The fatigue manifests its effect as a reduction in domain switching or do-
main wall movement and subsequent premature failure of devices [3,4].
Recent research reported a discussion of switching kinetics for ferroelec-
trics, demonstrating experimentally that nucleation can be the more im-
portant step, rather than domain wall forward-growth or sideways
velocities [5].
Ferroelastic domain textures generated by an interchanging of the
crystallographic degenerate and preferred axes result in a macroscopic
strain component [6], while both lattice strain and ferroelastic strain con-
tribute to the large signal macroscopic strain response during an applied
bipolar electric field [7]. The spontaneous strain generated by domain
switching can be simulated directly by dislocations, Xie et al. [8] extended
the dislocation simulation method of domain switching toughening to pi-
ezoelectric coupling field. 90° ferroelectric/ferroelastic domain reorienta-
tion was observed in a cantilever comprised of a 500 nm thick lead
zirconate titanate (Pb(Zr, Ti)O3, PZT) film on a 3 μm thick elastic layer
composite of SiO2 and Si3N4 [9]. For both electrical and mechanical
switching, ferroelastic switching is found to occur most readily at the
highly active needle points in ferroelastic domains [10]. The dynamic
properties of domain reorientation under mechanical cyclic loading in
bulk PZT ceramics are studied by Pojprapai et al. [11]. A distinctive
ferroelastic creep was observed at 293 K during the mechanical deforma-
tion of lanthanum strontium cobalt ferrite under uniaxial compression
[12]. Comparison of neutron diffraction and Raman spectroscopic studies
of the ferroelastic behavior of ceria-stabilized zirconia at elevated temper-
atures was conducted by Bolon et al. [13]. On the other hand, the reorien-
tation of ferroelastic domains around the crack tip is also proposed as a
toughening mechanism for ferroelectric ceramics [14–17]. Moreover,
the domain switching of ferroelectric ceramics is also found to be related
with the crystalline structure of materials. Li et al. [18] used a combined
theoretical and experimental approach to establish a relation between
crystallographic symmetry and the ability of a ferroelectric polycrystalline
ceramics to switch. Fu et al. [19] investigated the domain switching and
lattice strains in (Ka, Na)NbO3-based lead-free ceramics across
orthorhombic-tetragonal phase boundary.
Bismuth titanate, Bi4Ti3O12 (BIT), was identified as a typical bismuth
layer-structured ferroelectrics (BLSFs or Aurivillius phase) with high
Curie temperature (Tc = 675 °C) and large spontaneous polarization
(Ps (a-axis) = 49 μC/cm2) [20]. Many works have been devoted to the
research on the structural, optical, dielectric and electrical properties
of BIT [21–23]. However, many important mechanical problems
seem to be less discussed for BLSFs. In our previous works, W/Cr co-
doped Bi4Ti3O12 ceramics have been developed to achieve both good
electrical and mechanical properties, aiming at the high-temperature
piezoelectric application [24–27]. In this work, some mechanical
problems with respected to ferroelastic domain switching were
elaborately investigated for this kind of ceramics under uniaxial com-
pression load, with laying stress on both underlying mechanism and
theoretical models of the microcrack propagation and the ferroelasticity
toughening.
55
2. Experimental procedure
2.1. Preparation of ceramics
W/Cr co-doped Bi4Ti3O12 ceramics with a chemical composition of
Bi4Ti3−xWxO12+x + 0.2 wt% Cr2O3 (x = 0, 0.025, 0.05, 0.075 and 0.1, ab-
breviated them as BITWx-0.2Cr) were fabricated by a traditional solid
reaction process, according to the process technology described in our
previous works [24,25]. In addition, pure Bi4Ti3O12 (BIT) ceramics
were also fabricated by the same process in this work. All these BIT-
based ceramics were made into cylinders with a size of φ 10 mm × h
10 mm, as well as fine polished before the mechanical testing.
2.2. Characterization of ceramics
2.2.1. Microstructural characterization
The phase structure of ceramics was characterized by a powder X-
ray diffractometer (XRD; DX2700, Dandong, China) employing Cu-Kα
radiation (λ = 1.5418 Å). The lattice strain and lattice spacing were
determined by a single peak fitting approach using a specific code
(Oringin 9.1 ware, OriginLab, USA) according to Gaussian functions.
The microstructural morphology of ceramics was observed by a field-
emission scanning electron microscopy (FE-SEM; JSM-7001F, JEOL,
Japan).
2.2.2. Uniaxial compression test
We focus on the uniaxial compression stress states since ferroelec-
tric ceramics can be realized comparably easily for a brittle material
with low tensile strength, thus the uniaxial compression test was per-
formed on a universal testing machine (8501, INSTRON, UK). A com-
pression preload of 5 MPa was applied for the specimen fixation. The
loading mode was in the displacement control at a stroke rate of
1 mm/min. The macroscopic strain of specimen was measured using
an extensometer with a gauge length of 5 mm mounted to its middle,
while the applied load of machine was automatically recorded by an in-
ternal force sensor. The failure stress referring to the compression fail-
ure of materials is defined as their compression strength (σcf), which
is calculated according to the following equation:
P
σ cf ¼ ð1Þ
S
where P is the crushing load (N) and S is the cross-sectional area of the
specimen (mm2).
2.2.3. Vickers indentation test
Because only small samples could be obtained for ferroelectric ce-
ramics using this fabrication method, the fracture toughness was esti-
mated by microindentation techniques using a Vickers indenter
(AKASHI AVK-A, Kanagawa, Japan). The fracture toughness with the
form of the work of fracture (Г), which is related to the energy release
rate, is preferred over the fracture toughness KIC (=ГE)1/2 because it is
the more relevant parameter and independent of the elastic modulus
(E), and Vickers hardness (Hv) could be calculated by two equations as
follows,
2
2 d
Γ ¼ 2ξ P ð2Þ
c3
P
H v ¼ 0:464 2
ð3Þ
d
where P is the indentation load, d is the half-length diagonal of the
Vickers indentation, c is the half-length of the penny-shaped crack and
ξ (0.016) is the non-dimensional coefficient. In this experiment, the in-
dentation load of 19.6 N was applied and the holding time was 15 s.
Cracks resulting from the Vickers indentation were measured by SEM
56 Y. Chen et al. / Materials and Design 152 (2018) 54–64
immediately to avoid slow crack growth after printing, started by the
stress field that acts upon loading. The crack extent was measured
with an image analysis software (Leica Application Suite EZ; Leica
Microsystems Ltd, Germany).
3. Results and discussions
3.1. Crystallographic structures of BIT-based ceramics
A room-temperature XRD pattern, associated with a cross-section
SEM image was derived from the BIT ceramic, which are shown in
Fig. 1 (the data of BITWx-0.2Cr ceramics are similar to this and have
been shown in [24]). In XRD pattern, all sets of reflections could be
well indexed in the parent compound of Bi4Ti3O12 (JCPDS card No. 72-
1019) with an pseudo-orthorhombic structure and a space group of
B2cb (41), and there is no obvious second phase existing in the matrix.
As observed from the SEM image that, a dense microstructure con-
structed by many randomly orientated plate-like grains has been devel-
oped in the BIT ceramic, even, a typic layer structure can be seen from
the fractured surface of several coarse grains indistinctly. Furthermore,
the (1 1 7) plane shows the highest diffracted intensity, which agrees
with the fact that the randomly orientated BLSFs usually reflect the
strongest diffraction in their (1 1 2m + 1) crystallographic planes
[28–30], which are closely related to their layer structure along the c-
axis. In addition, many reflections with crystallographic index of (h k
l)/(k h l) such as (2 0 0)/(0 2 0) are observed to split into two peaks,
which is due to the typical orthorhombic structure with a slight differ-
ence between the lattice constants of a and b. Here, orthorhombicity
(g) is used for evaluating the orthorhombic degree of the structure,
which could be calculated by the following formula,
2ða−bÞ
g¼
aþb
On the other hand, for ferroelectrics, domain switching can yield a
maximum degree of orientation that corresponds to the maximum
number of possible directions of ferroelastic structural distortions rela-
tive to a prototype unit cell. The single-crystal distortion (S0lattice) is de-
fined as the difference of the ferroelastic unique axis (herein referred to
as [h k l⁎]) relative to the crystallographic degenerate axis. BIT
Fig. 1. XRD patterns of the Bi4Ti3O12 ceramic registered at room temperature (the inserted
SEM image describes the microstructure of its polished and hot-etched cross section, the
gray vertical lines at the bottom of the figure represent the peak positions and relative
intensities derived from the JCPDS card.)
crystallizes into an orthorhombic system below the Curie temperature,
with two orthogonal single-crystal distortion directions relative to the
prototype tetragonal phase, and [0 1 0] is defined as [h k l⁎] [6]. S0lattice
of BIT could be thus calculated by the following formula (a N b),
a−b
S0lattice ¼ ð5Þ
b
According to the Rietveld 0refinement for the diffraction pattern, the
cell parameters of Bi4Ti3O12 ceramic calculated are basically in agree-
ment with the values determined by Ivanov et al. [31]. All these crystal-
lographic parameters of BIT-based ceramics are listed in Table 1. As can
be seen, a remarkable decrease of g (6.82 → 4.99) reveals a less ortho-
rhombic distortion in BIT after the incorporation of Cr2O3 (x = 0),
which may be due to the released rotation state of oxygen octahedra
by the substitution of Cr3+ for Ti4+. Except for an ephemeral increase
(4.99 → 5.53) of g in the introduction of less W6+ (x = 0.025), this factor
seems to decrease with increasing the doping content of W6+ (x N
0.025) linearly, which may be due to the gradually released tilting
state of oxygen octahedra by the substitution of W6+ for Ti4+. In addi-
tion, it is also found that after introducing Cr2O3 into BIT, both (2 2
0) and (1 1 15) reflection shift to a lower 2θ angle, which could be relat-
ed to the increase of lattice parameters according to the Bragg equation
and the formula about interplanar spacing. Especially, (1 1 15) reflection
is observed to disappear at x = 0.1, which may be caused by the struc-
ture anomaly (possibly the second phase like Bi6Ti3WO18 or Bi6Ti5WO22
[32]) of BITW0.1-0.2Cr with a high W content. Based on the analysis
above, for the ion-doped BIT system, the structure with a higher
orthorhombicity (g) tends to contain a larger single-crystal distortion
(S0lattice), which could be correlated with the larger displacements of
cationic polarization within a more distorted structure.
ð4Þ
3.2. Stress-strain response of the unpoled BIT ceramic
Fig. 2 shows the stress-strain response of the unpoled BIT ceramic
subjected to uniaxial compression load. It can be seen from
Fig. 2(a) that, the total stress-strain curve includes three segments
with different slope, which indicates a typical nonlinear deformation
process of the ceramic. Specifically, when a sufficiently small loading
was exerted on the ceramic from the starting point O, lattice ions will
be moved only slightly from their equilibrium positions and the ceramic
begin to elastically deform (Eini = 44 GPa) in this process (OA stage).
After the load stress exceeds the minimum coercive stress (σc, min =
45 MPa) of ferroelastic domains at the point A, their switching process
will be initiated, leading the ceramic into a ferroelastic deformation pro-
cess, where certain ions reach to new equilibrium positions such that
the longer c-axis of domains orientates to a direction perpendicular to
the compression load (90° switching). In this process (AB stage), both
the switching strain (εzz) of those reorientated ferroelastic domains
due to the interchanging of ferroelastic axis and the lattice strain (εlat)
of several reconstituted lattice planes due to the d-spacing change of
bismuth layer structure, as well as a slight elastic strain (εela) resulted
from those unmoved domains whose longer c-axis are originally per-
pendicular to the compression stress, contribute to the macroscopic
total strain of ceramic (Δεtol = εmax − εmin = εela + εzz + εlat =
0.60%). Both domain switching and lattice reconstitution are inelastic
and irreversible, their summations constitute the main part of remanent
strain (εrem = 0.65%) which was evaluated by the intercept of the load-
ing curve of BC. Here, the remanent strain represents the macroscopic
average of the spontaneous strains of unit cells when the domain struc-
ture is basically preserved (after backswitching) after unloading, be-
cause the loading curve of BC derived from the failure stress to the
switching-performed stress, thus the remanent strain evaluated by
this curve approximates to its actual value which should be determined
by the unloading curve [12]. For ferroelectric ceramics, the driving force
for domain switching is the minimization of the elastic free energy
 Y. Chen et al. / Materials and Design 152 (2018) 54–64
Table 1
Crystallographic parameters of BIT-based ceramics.
BIT BITW0-0.2Cr
Crystal system Orthorhombic
Space group B2cb (41)
a (Å) 5.4457 5.4699
b (Å) 5.4087 5.4427
c (Å) 32.8378 32.9354
V (Å) 967 982
g (×10−3) 6.82 4.99
S0lattice (×10−3) 6.84 5.00
Note: Cell parameters of BITWx-0.2Cr ceramics were derived from our previous works [24].
within the frontal zone [33]. Because the volume change of unit cell dur-
ing domain switching is zero thus only shear stresses will initiate the
domain switching. As a result, if the matrix is not able to accommodate
the twinning shear induced by the ferroelastic domain switching, many
holes may form at the blocked domain walls and then develop into
microcracks.
On the other hand, during ferroelastic domain switching, the appar-
ent elastic modulus of materials tends to be gradually improved by the
compressive stress due to the anisotropic elastic properties of noncubic
unit cells such as Bi4Ti3O12 [34]. After most of possible (or achievable)
domains were switched at point B (σc, max = 73 MPa), the ceramic
seems to react more elastically (Eld = 70 GPa) to the applied stress
until to the compression failure at point C (σcf = 281 MPa; εtol =
1.06%), since the displacement linearly increases with the load again.
In this process (BC stage), with increasing the load, microcracks start
closing and the apparent elastic modulus increases (E = 44 GPa
→ 70 GPa); if further increasing the load, damage takes over, which
leads the ceramic to cracking finally. The underlying micromechanism
can be further stated as follows [35]: since the cracks perpendicular to
the applied stress tend to close, with consequent stiffening of the
body, while those parallel to the stress tend to open, with damage intro-
duced to the material. Additionally, two decoupled strain-stress curves,
which could reflect the elastic deformation resulted from the reversible
ionic balance exercise (black line) and the inelastic deformation caused
by the ferroelastic domain switching and lattice plane reconstitution
(blue one), respectively, are depicted with the total deformation curve
(red one) in Fig. 2(b).
In order to identify the failure mode of ceramic, SEM observation
was conducted by focusing on the fracture surface of the broken sample.
In Fig. 3, the ceramic behaves in the mode of transgranular fracture and
a direct evidence of microcrack creation can be identified by those clear-
ly visible microcracks. The very thin crack, such as the one indicated by
the red arrow, may have nucleated upon cooling from the Curie
Fig. 2. Stress-strain response of the unpoled Bi4Ti3O12 ceramic subjected to uniaxial compression load, (a) total stress-strain curve and (b) decoupled strain-stress curves.
57
BITW0.025-0.2Cr BITW0.05-0.2Cr BITW0.075-02Cr BITW0.1-0.2Cr
5.4578 5.4485 5.4464 5.4567
5.4277 5.4199 5.4215 5.4400
32.9155 32.8742 32.8320 32.9274
975 971 969 979
5.53 5.26 4.58 3.07
5.55 5.28 4.59 3.07
temperature due to the release of grain microstresses. These thermal
microcracks seem to remain closed in the process of loading since per-
pendicular to the applied stress. And the large opening microcracks,
such as the two indicated by the blue arrow, may be initiated at some
pore points of high tensile stress or evolved from those blocked
ferroelastic domains in switching during the mechanical loadings. As
parallel to the applied stress, these mechanical microcracks tended to
open, some of them further developed into macroscopic flaws, as exhib-
ited in the inserted figure which shows the appearances of the broken
sample. Here: thermal microcracks closure, with consequent stiffening
of the body, and mechanical microcracks opening, with damage intro-
duced to the material. Those two processes are considered to act in com-
petition at lower loads, which may be responsible for the nonlinear step
(as marked by the green circle) at point B in Fig. 2(a).
Further, single peak fitting is an additional method by which to ex-
tract the lattice strain, however, its actual value should be given by an
average of all lattice strains parallel to the compressive stress, weighted
by their relative presence [36]. For Bi4Ti3O12 with the ferroelastic unique
axis of [0 1 0], the interplanar spacing change of (0 2 0) plane could be
considered as an estimate of bulk-averaged lattice strain of materials
approximately parallel to the compression axis, thus the subtraction of
the lattice strain from the total irreversible strain can yield the
ferroelastic switching strain for the ferroelectric ceramics subjected to
the compressive stress. For the BIT ceramic after compression at σc,
max, the lattice strain can be evaluated as 0.44% (see Section 3.3), and
the total irreversible strain (0.55%) was calculated by the macroscopic
total strain (0.60%) subtracting the elastic strain ((σc, max − σc, min) /
Eave = 0.05%), thus the ferroelastic switching strain was determined
as 0.11% finally. According to the volume-weighted model proposed in
[6], the calculated value of switching strain approximates 3/4 of its sat-
urated value (εzz = 0.212 · S0lattice = 0.15%) for the BIT ceramic. In fact,
the realistically or attainable achievable domain switching textures are
a fraction of this saturation texture because of the adverse
58 Y. Chen et al. / Materials and Design 152 (2018) 54–64
Fig. 3. SEM image focusing on the fracture surface of the Bi4Ti3O12 ceramic broken by the
compression load.
microstructural interactions in real ceramics (such as interactions be-
tween neighboring grains and domain-wall pinning due to point de-
fects, etc.), as well as the number of possible ferroelastic structural
distortion directions relative to the prototype phase.
3.3. Ferroelastic behavior and constitutive equation of unpoled ferroelectric
ceramics
For most of perovskite-type ferroelectrics, when they are cooled
through the paraelectric-ferroelectric phase transition temperature,
the combination of electric and elastic boundary conditions loading on
the crystal usually leads to a complex domain structure with many
90° walls (to minimize the elastic energy) and 180° walls (to reduce
the depolarizing electric fields). 90° walls are both ferroelectric and
ferroelastic as they differ both in orientation of the spontaneous polari-
zation vector (Ps) and the spontaneous strain tensor (xs). In a single
crystal of Bi4Ti3O12, there has been reported that the ferroelectric do-
main switching with electric fields along the c axis involves a large
change in the extinction angle with illumination along b [37], and
these ferroelastic domain walls are established after the rearrangement
of non-ferroelastic 180° domains accompanied by Ps(c) [38]. As a result,
for the polycrystalline ceramics of Bi4Ti3O12, their ferroelastic domain
switching are more accurately described by non-180° (mainly 90°) do-
main wall motion, whose micromechanism could be described by Fig. 4.
Here, in order to simplify the problem of ferroelastic domain
switching in polycrystalline Bi4Ti3O12 ceramics, as well as in view of
the comparability of tetragonal phase with orthorhombic phase in the
interchange between long axis (c) and short axis (a/b), we assumed
the orthorhombic Bi4Ti3O12 as the tetragonal phase in the later discus-
sion and analysis with respect to its ferroelastic behaviors. The fresh-
sintered ferroelectric ceramics could be considered as the thermally
depoled piezoceramics with randomly orientated ferroelectric/
ferroelastic domains (i.e. initially isotropic), which could be described
by the state at point O. Starting at point A, a mechanical stress cannot
trigger a unique switching direction for the spontaneous strain of a
unit cell like that, all the c-axis will prefer a position close to a plane per-
pendicular to the compression stress, the distribution of these c-axis do-
mains within this plane will be random. It is worthy to noting that
domain orientations with the c-axis in the range 0° b α b 30° from the
compression axis are easier to switch than those with the c-axis at an
angle of 30° b α b 60° [39], and the domains with a smaller coercive
stress are earlier to switch than those with a larger coercive stress. As
a result, in the cross section parallel to the compression stress, succes-
sional 90° switching of ferroelastic domains is practically in promoting
the gradual shift of 90° domain walls between c-axis domains and a/b-
axis ones situated at this plane. The corresponding bulk contributions
to the driving force are caused by variations of the ferroelectric/
ferroelastic anisotropy energy [40]. Ending at point B, the domain
state induced by strong uniaxial compression stresses is transversely
isotropic without macroscopic polarization, those domains switched
left an irreversible deformation to the material, since the inelastic strain
reaches to the maximum value.
The switching process of ferroelastic domain yields a maximum de-
gree of orientation that corresponds to the maximum number of possi-
ble directions of ferroelastic structural distortions relative to a prototype
unit cell. In order to estimate the fraction of ferroelastic domains
switched, using Kamlah-Jiang's constitutive framework for reference
[41], a simplified domain distribution model was built for unpoled fer-
roelectric ceramics in Fig. 5.
Here, the direction of uniaxial compression load is assumed to coin-
cide with the z-axis, a region within a grain of the ferroelectric polycrys-
tal where the c-axis of the unit cells (spontaneous strain axis) have the
same orientation to the loading direction is called a c-axis domain, all
the domains within the cones of 45° angle with the z-axis being the
cone axis are considered to be c-axis domain for simplicity. The unpoled
ferroelectric ceramics are in the thermally depoled reference state, by
the cutoff of the spherical surface by these cones, the fraction of c-axis
domains is assumed to be 1/3, if we consider that the state of relative
polarization of domains will have no influence on the remanent distor-
tion of the lattice, and there is a simple linear relation between the accu-
mulation of irreversible strain with the fraction of switched domains,
the constitutive equation of an unpoled ferroelectric ceramic subjected
to a uniaxial compression load could be expressed as follows,
σ
S¼ þ Ssat ð1−3βÞ ð6Þ
E
here, β represents the fraction of c-axis domains (1/3 ≥ β ≥ 0), which is
decreasing with increasing load, Ssat is the maximum value of the mac-
roscopic remanent strain of the ceramic which is assumed for a domain
state of highest order with respect to a certain axis (i.e. β = 0), S is the
total strain, σ is the applied load and E is the elastic modulus. Now, we
considered a point close to the failure point C where most of possible
(or achievable) domains have been switched by the stress, after
substituting the corresponding parameters derived from the stress-
strain curve into the Eq. (6), the fraction of c-axis domains (β) which
still stay in the cones of 45° angle is 4.66%, thus the corresponding
switching fraction of c-axis domains is 86.01% (=1 − 3β), as for
whole domains, the switching fraction is 28.67% (=1/3 − β), which in-
dicates that nearly all of c-axis domains switched to the direction per-
pendicular to the compression stress.
On the other hand, the ferroelastic (or 90°) domain switching caused
by the mechanical loading has been determined previously by quantify-
ing the intensity changes of specific peaks in a diffraction pattern, e.g.
the “c”(0 0 l) and “a” (h 0 0) peaks in tetragonal perovskite ceramics
[42,43]. The orthorhombic Bi4Ti3O12 is also of perovskite units, Fig. 6
shows XRD patterns of the Bi4Ti3O12 ceramic before and after compres-
sion at σc, max (73 MPa). As can be seen from the enlarged drawing on
the left side, after the ceramic was compressed, the intensity of the
peak corresponding to (0 0 l) planes decreases, indicating the grains
with c-axis oriented along with the pressing direction have been
switched by the compress stress and added to the low degree of (0 0
l) favored orientation observed [44]. Further, to obtain a more quantita-
tive analysis of the switching behavior and domain texture of BIT, the in-
tensities measured in the diffraction experiment were used to calculate
a density value which describes domain preferences in the unit multiple
of a random distribution (MRD) and which is expressed through the
 Y. Chen et al. / Materials and Design 152 (2018) 54–64 59

Fig. 4. Micromechanism of the ferroelastic behavior of unpoled ferroelectric ceramics.

following equation [45], shown in the enlarged drawing on the right side. According to their rel-

 ative intensity variation, MRD of the BIT ceramic can be identified as
I 0012 =IR0012 0.4033, which is closed to that of unpoled PZT ceramics after cyclic com-
MRD ¼

 ð7Þ pression at 140 MPa/1 Hz/100 times [46]. In addition, the interplanar
I0012 =IR0012 þ 2  I 200 =IR200 spacing of (0 2 0) planes has a decrease after compression, indicating
that the compression stress applied along the [0 1 0] direction of some
where I0012 and I200 are, respectively, the integrated intensities of the (0 unit cells has caused lattice strain for the structure. Here, the diffraction
0 12) and (2 0 0) peaks after poling or mechanical loading. IR0012 and data obtained from the X-ray diffractometer were used to ascertain lat-
IR200 are the intensities obtained from the sample with random domain tice strain, which is calculated from the difference between the lattice
texture (i.e. before compression). The evolution in the diffraction pat- spacing before and after loading, expressed as [11]:
tern of (0 0 12)/(2 0 0)/(0 2 0) peaks before and after compression is

0
dhkl −dhkl
εhkl ¼ 0
ð8Þ
dhkl

Fig. 5. Simplified domain distribution model of unpoled ferroelectric ceramics based on a Fig. 6. XRD patterns of the Bi4Ti3O12 ceramic before and after compression at σc, max
constitutive framework. (73 MPa).
60 Y. Chen et al. / Materials and Design 152 (2018) 54–64
where εhkl is the lattice strain (denoted by εlattice in this work), and dhkl
and d0hkl are the lattice spacings before and after loading. Therefore, the
value of lattice strain can be determined as 0.44% for the BIT ceramic
after compression.
3.4. Toughening effect resulted from ferroelastic behavior
Under the application of a Vickers indenter, the indentations and
cracks produced in the BIT ceramic before and after compression are
shown in Fig. 7. SEM observation reveals that a symmetric rhombic in-
dentation has been introduced into the ceramic, and the resulting cracks
tend to propagate along the diagonal direction of indentation. However,
most of cracks seem to be shorter in the ceramic after compression, and
some of them have even disappeared. According to Eq. (2), the fracture
toughness (Г) calculated from the length of cracks and indentations was
determined as 32.21 J/m2 and 36.68 J/m2 for the BIT ceramic before and
after compression, respectively, which presents an obvious toughening
effect (ΔГ = 4.47 J/m2). On the other hand, the Vickers hardness (Hv)
was found to be 3.81 GPa and 3.49 GPa for the BIT ceramic before and
after compression, respectively. It is well known that the hardness rep-
resents the ability of materials to resist the plastic deformation. The pro-
cess zone of ferroelastic domain switching is more suitable for the
movement of lattice dislocations due to the residual strain existing
there, which tends to decrease the hardness of the ceramic. Moreover,
after compression, the crack extension seems to be more anisotropic
in view of their uneven length, the anisotropy of the indentation crack
length and corresponding apparent fracture toughness could be related
with the interaction of domain switching and residual strain, which has
been observed in perovskite membrane materials [47].
Here, this toughening effect is derived from the ferroelastic domain
switching, whose micromechanism can be illustrated by Fig. 8. As de-
scribed by this micromechanics model for the propagation of indenta-
tion cracks within ferroelectric ceramics. When ferroelectric ceramics
were compressed by a loading stress exceeding their coercive stress,
ferroelastic domain (90°) switching has occurred (domains only orien-
tated to the direction perpendicular to the compression load, which
seems like those colorized bloc), the frontal region involved with
Fig. 7. SEM images of indentations and cracks produced in the Bi4Ti3O12 ceramic by a Vickers indenter. (a), (b) and (c) were derived from the upper surface of the sample before
compression; (d), (e) and (f) were derived from the lower surface of the sample after compression (73 MPa). Three testpoints were used for the average measurements.
domain switching is referred to as the process zone. In the contacting
process of sharp indenter with the material surface, the maximum ten-
sile stress occurs at the elastic/plastic interface, with fall-off within the
plastic zone to a negative value at the indenter/specimen contact and
within the surrounding elastic region to zero remote from the contact
[48], which initiates the nucleation of crack at the vertex angle of inden-
tation. When cracks begin to propagate along the diagonal line of inden-
tation and grow into the process zone, those domains previously located
in the process zone at the crack tip are now located in the crack wake. As
the polar rotation of ferroelastic domains switched by 90° has caused a
strain mismatch of up to 1% (0.12% for BIT in this case) there, compres-
sion stresses will be created perpendicular to the crack plane because of
the preferred orientation of the longer axis of ferroelastic domains in
this direction. When the crack extends, these compression stresses
tend to act in the crack wake and cause shielding of the crack tip from
the applied load as in the case of the second phase toughening the ma-
trix [49]. In addition to compression stresses, tensile stresses are present
parallel to the crack plane. However, the tensile stresses parallel to the
crack plane are, in a first approximation, not of importance for the stress
intensity factor at the crack tip. Moreover, TEM observation from C.
Mercer et al. [50]. has presented compelling evidence for the irrevers-
ible deformation of ferroelastic domains in a process zone on either
side of a crack formed by indentation. The ferroelasticity induced tough-
ening is supposed to have a quadratic relationship with the volume per-
cent of T′ phase within YSZ ceramics, which is also identified by the
Vickers indentation test [51].
Fracture toughness, defined as the resistance to fast crack propaga-
tion at a critical stress level (KIC), is widely used for mechanical charac-
terization of brittle ceramics. In view of the absorption of fracture
energy during ferroelastic domain switching, the crack tip process
zone is conducive to resisting the crack extension, resulting in an elevat-
ed fracture toughness for materials, thus the elevation of the steady-
state toughness Г (J/m2) would scale as [51]:
ΔΓ ¼ 2  f  σ c  εc  h ð9Þ
in which σc is the coercive stress related to switching of the orientation
 Y. Chen et al. / Materials and Design 152 (2018) 54–64 61

Fig. 8. Micromechanics model for the propagation of indentation cracks within ferroelectric ceramics.

of ferroelastic domains, εc is the associated switching strain, σc.εc is the
dissipated energy per unit volume associated with ferroelastic toughen-
ing, h is the width of the processing zone (it is assumed to remain con-
stant in front of the crack tip during crack propagation), and f is the
volume fraction of domains that switch within the zone. Here, Mehta
and Virkar [52] used a similar approach for estimating the size of do-
main switching process zone with that of the crack tip plastic zone as
follows,
 2
K IC
h ≈ 0:08
σc
And for the brittle materials such as ferroelectric ceramics under the
application of a sharp indenter, another fracture toughness with the
form of KIC could be calculated by the following formula based on the
criterion of fracture mechanics [53],


K IC ¼ 0:016ðE=HÞ1=2 P=C 3=2
3.5. Mechanical properties of unpoled BITWx-0.2Cr ceramics
Fig. 9 shows the stress-strain response of unpoled BITWx-0.2Cr ce-
ramics subjected to uniaxial compression load. Clearly, the compression
behavior of BITWx-0.2Cr ceramics are similar to that of BIT ceramic, also
including the inelastic deformation caused by the ferroelastic domain
switching. However, some specific difference can be observed from
their stress-strain curves, as indicated by those saffron circles. Such as,
BITW0-0.2Cr has the highest compression strength while BITW0.025-
0.2Cr received the lowest one, BITW0.05-0.2Cr has a smaller ferroelastic
ð10Þ deformation while BITW0.075-0.2Cr possesses a larger one. The inserted
image gives out the change in the apparent elastic modulus (Eini and Eld)
of each sample during the initial process and loading process. As can be
seen, all the others except for BITW0.025-0.2Cr present an increased elas-
tic modulus in the loading process, which is similar to the case of BIT.
The unexpected increase in the elastic modulus may be attributed to
the very small ratio of microcracks (perpendicular to the applied stress)
which tend to close during loading. Therefore, the premature failure of
ð11Þ BITW0.025-0.2Cr is considered to be dominated by the plenty of

Here, P is the load, E is the Elastic modulus, Hv is the hardness, and C

is the length of the crack measured from the centre of indent. According
to the expression of fracture toughness Г from Eq. (2), we can deduce
out the relationship between the two kinds of fracture toughness as fol-
lows,

ΓE
K IC 2 ¼ ð12Þ
0:928

Ultimately, substituting Eqs. (10) and (12) into Eq. (9), we can get
the expression of toughening effect (ΔГ/Г) with respect to the
ferroelastic domain switching as follows,

ΔΓ 1
¼ 0:172  f  εc  E  ð13Þ
Γ σc

for the BIT ceramic, substituting its corresponding parameters (f =

86.01%; σc, average = 59 MPa; εc = 0.11%; Eaverage = 59 GPa) derived
from its stress-strain curve (Fig. 2(a)) into Eq. (13), the toughening ef-
fect ΔГ/Г can be calculated to reach at 16.55%, which is slightly higher Fig. 9. Stress-strain response of unpoled BITWx-0.2Cr ceramics subjected to uniaxial
than its experimental data of 13.88%. compression load.
62 Y. Chen et al. / Materials and Design 152 (2018) 54–64
microcracks (parallel to the applied stress) which tend to open during
loading.
According to the classic Griffith theory [54], the strength of a materi-
al can be described as follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffi
2Eγ f 0
σ f −σ i ¼
πc
where σf − σi stands for the actual fracture strength (σf is the fracture
strength without internal stress and σi is the part contributed on frac-
ture strength by internal stress). c is the length of the critical crack, γf0
is the fracture surface energy without internal stress and E is the elastic
modulus. Based on this theory, both the lower elastic modulus due to
few “closed microcracks”, and more internal stress centralized by
many “opening microcracks”, might contribute to the much lower
strength of BITW0.025-0.2Cr as compared with others.
In order to quantify the evolution of mechanical properties for
unpoled BITWx-0.2Cr ceramics, eight mechanical parameters related
to the ferroelastic behavior were extracted from their stress-strain
curves, which have been depicted as a function of the doping content
of W (x) in Fig. 10. As can be seen from this complex pattern that, firstly,
the variation of coercive stress (average) with x is just opposite to that of
orthorhombicity (Fig. 10(a)), which shows that ferroelastic domains
with a higher structural orthorhombicity tend to be switched by a
smaller compression stress. And then, the varying trend of switching
strain with x seems to agree with that of single-crystal distortion well
(Fig. 10(b)), which has been decided by their tie-in equation of εzz =
0.212 · S0lattice based on a volume-weighted average of the single-
crystal distortions of each domain over the entire orientation space. Fur-
ther, more switched domains certainly contribute to more irreversible
deformation for ferroelectric ceramics, which has been presented by
the similar trend of switching fraction to remanent strain in Fig. 10(c).
Moreover, the remanent strain resulted from partial ferroelastic domain
switching and several lattice-plane reconstitution contribute about 60%
to the total strain of these BIT-based ferroelectric ceramic, which can be
Fig. 10. Mechanical properties of unpoled BITWx-0.2Cr ceramics as a function of the doping content of W (x), (a) σc vs g; (b) εzz vs S0lattice; (c) f vs εrem; (d) E vs σcf.
observed from their stress-strain curves. Finally, the brittle fracture usu-
ally occurs in ferroelectric ceramics during their elastic compression, the
failure stress is mainly contingent on their elastic modulus, with the ad-
dition of partial influence exerted by the nonlinear ferroelastic behavior
[55]. This mechanism could be reflected by the consistent varying trend
between the two parameters when x is below 0.05, and the slight devi-
ð14Þ ation when x is above 0.05 (both BITW0.075-0.2Cr and BITW0.1-02Cr
have a larger switching strain), as shown in Fig. 10(d). In addition, as
marked by the red rectangular frame, BITW0.05-0.2Cr with a medium
W content possesses both good microcosmic switching properties
(small coercive stress and large switching strain), and good macroscopic
mechanical properties (high elastic modulus and compression
strength).
Fig. 11 gives out both testing fracture toughness (Г) and calculated
toughening effect (ΔГ/Г) for BITWx-0.2Cr ceramics with different W
content (x). As can be seen, BITW0.05-0.2Cr gains the highest fracture
toughness of 58.21 J/m2, as well as expected to be of a larger toughening
effect (13.54%) than the others in the case of compression. A higher frac-
ture toughness may be contributed by a larger grain size for the quasi-
brittle fracture of ceramics, according to the prediction of ceramic frac-
ture with normal distribution pertinent to grain size [56,57]. Here,
BITW0.05-0.2Cr indeed has the largest grain size among BITWx-0.2Cr,
which has been reported in one of our previous works [27]. Moreover,
the reduced piezo-response is also expected relative to the bulk of the
grain, due to the restricted domain-wall motion at grain boundaries.
Therefore, such ferroelectric ceramics with coarse grains tend to present
a higher piezoelectric ability after poling, which also agrees in the
highest d33 ~ 28 pC/N of BITW0.05-0.2Cr ceramics as found by us in [25].
4. Conclusion
For a typic number of bismuth layer-structured ferroelectric (BLSFs)
ceramics: bismuth titanate (Bi4Ti3O12, BIT), its ferroelastic behavior sub-
jected to uniaxial compression was detected by the nonlinear variation
of the stress-strain curve. The switching strain could approximate 3/4 of
its saturated value while the switching fraction of c-axis domains was
 Y. Chen et al. / Materials and Design 152 (2018) 54–64
Fig. 11. Testing fracture toughness (Г) and calculated toughening effect (ΔГ/Г) of unpoled
BITWx-0.2Cr ceramics with different W content (x).
28.67%. SEM observation on the fracture surface reveals that it behaved
in the mode of transgranular fracture and two types of microcracks
(opening and closed) initiated in the matrix during mechanical loading.
The ferroelastic domain switching was also been demonstrated by the
diffraction evolution of (0 0 12)/(2 0 0)/(0 2 0) peaks after compression.
The ferroelastic induced toughening effect (ΔГ/Г) reached to 16.55%.
Among W/Cr co-doped Bi4Ti3O12 (BITWx-0.2Cr) ceramics, BITW0.025-
0.2Cr gained the smallest coercive stress (σc = 56 MPa) and lowest
compression strength (σcf = 134 MPa) while BITW0.05-0.2Cr possessed
the highest fracture toughness (Г = 58.21 J/m2) and a significant poten-
tial toughening effect (ΔГ/Г = 13.54%). This research could not only
help us to further understand the correlation between macroscopic
properties and ferroelastic domain switching of BLSFs, but also effec-
tively promote the possible application of BITWx-0.2Cr ceramics in
high-temperature piezoelectric devices.
Data availability
The datasets generated during the current study are available from
the corresponding author on reasonable request.
Acknowledgement
This work was supported by the Applied Basic Research Program
from Sichuan Province (2017JY0091), National Natural Science Founda-
tion of China (Grant No. 11702037, 11572057 and 51332003), China
Postdoctoral Science Foundation Funded Project (2017M623025), Spe-
cial Funding for Post Doctoral Research Projects from Sichuan Province
((2017, presided over by Yu Chen).
References
[1] M. Kobune, T. Kuriyama, R. Furotani, et al., Ferroelectric materials and their applica-
tions, Jpn. J. Appl. Phys. 54 (10S) (2015).
[2] B. Kaltenbacher, P. Krejčí, A thermodynamically consistent phenomenological model for
ferroelectric and ferroelastic hysteresis, J. Appl. Math. Mech. 96 (7) (2016) 874–891.
[3] D. Lupascu, J. Rödel, Fatigue in bulk lead zirconate titanate actuator materials, Adv.
Eng. Mater. 7 (10) (2005) 882–898.
[4] N. Chaiyo, D.P. Cann, N. Vittayakorn, Lead-free (Ba, Ca)(Ti, Zr)O3 ceramics within the
polymorphic phase region exhibiting large, fatigue-free piezoelectric strains, Mater.
Des. 133 (2017) 109–121.
[5] J.F. Scott, A review of ferroelectric switching, Ferroelectrics 503 (1) (2016) 117–132.
[6] J.L. Jones, M. Hoffman, K.J. Bowman, Saturated domain switching textures and
strains in ferroelastic ceramics, J. Appl. Phys. 98 (2) (2005), 024115. .
[7] M.C. Ehmke, N.H. Khansur, J.E. Daniels, et al., Resolving structural contributions to
the electric-field-induced strain in lead-free (1 − x)Ba(Zr0.2Ti0.8)O3-
x(Ba0.7Ca0.3)TiO3, piezoceramics, Acta Mater. 66 (3) (2014) 340–348.
63
[8] C. Xie, Q.H. Fang, Y.W. Liu, et al., Dislocation simulation of domain switching tough-
ening in ferroelectric ceramics, Int. J. Solids Struct. 50 (9) (2013) 1325–1331.
[9] G. Esteves, C.M. Fancher, M. Wallace, et al., In situ X-ray diffraction of lead zirconate
titanate piezoMEMS cantilever during actuation, Mater. Des. 111 (2016) 429–434.
[10] P. Gao, J. Britson, C.T. Nelson, et al., Ferroelastic domain switching dynamics under
electrical and mechanical excitations, Nat. Commun. 5 (2014) 3801.
[11] S. Pojprapai, Z. Luo, B. Clausen, et al., Dynamic processes of domain switching in lead
zirconate titanate under cyclic mechanical loading by in situ neutron diffraction,
Acta Mater. 58 (6) (2010) 1897–1908.
[12] W. Araki, J. Malzbender, Ferroelastic deformation of La0.58Sr0.4Co0.2Fe0.8O3−δ under
uniaxial compressive loading, J. Eur. Ceram. Soc. 33 (4) (2013) 805–812.
[13] A.M. Bolon, T.A. Sisneros, A.B. Schubert, et al., Comparison of neutron diffraction and
Raman spectroscopic studies of the ferroelastic behavior of ceria-stabilized zirconia
at elevated temperatures, J. Eur. Ceram. Soc. 35 (2) (2015) 623–629.
[14] A.V. Virkar, J.F. Jue, P. Smith, et al., The role of ferroelasticity in toughening of brittle
materials, Phase Transit. 35 (1) (2016) 27–46.
[15] S.M. Denkhaus, M. Vögler, N. Novak, et al., Short crack fracture toughness in (1
− x)(Na1/2Bi1/2)TiO3–xBaTiO3 relaxor ferroelectrics, J. Am. Ceram. Soc. 100 (2017).
[16] M. Vögler, J.E. Daniels, K.G. Webber, et al., Absence of toughening behavior in
0.94(Na1/2Bi1/2) TiO3-0.06 BaTiO3 relaxor ceramic, Scr. Mater. 136 (2017) 115–119.
[17] Y. Wang, H. Yang, R. Liu, et al., Ferroelastic domain switching toughening in spark
plasma sintered t′-yttria stabilized zirconia/La2Zr2O7 composite ceramics, Ceram.
Int. 43 (15) (2017) 13020–13024.
[18] J.Y. Li, R.C. Rogan, E. Üstündag, et al., Domain switching in polycrystalline ferroelec-
tric ceramics, Nat. Mater. 4 (10) (2005) 776–781.
[19] J. Fu, R. Zuo, Y. Xu, et al., Investigations of domain switching and lattice strains in
(Na, K) NbO3-based lead-free ceramics across orthorhombic-tetragonal phase
boundary, J. Eur. Ceram. Soc. 37 (3) (2017) 975–983.
[20] T. Jardiel, A.C. Caballero, M. Villegas, Aurivillius ceramics: Bi4Ti3O12-based piezoelec-
trics, J. Ceram. Soc. Jpn. 116 (1352) (2008) 511–518.
[21] Y.J. Wong, J. Hassan, S.K. Chen, et al., Combined effects of thermal treatment and Er-
substitution on phase formation, microstructure, and dielectric responses of
Bi4Ti3O12 Aurivillius ceramics, J. Alloys Compd. 723 (2017) 567–579.
[22] A. Bettaibi, R. Jemai, M.A. Wederni, et al., Effect of erbium concentration on the
structural, optical and electrical properties of a Bi4Ti3O12 system, RSC Adv. 7 (36)
(2017) 22578–22586.
[23] F. Rehman, L. Wang, H.B. Jin, et al., Effect of Fe/Ta doping on structural, dielectric,
and electrical properties of Bi4Ti2.5Fe0.25Ta0.25O12 ceramics, J. Am. Ceram. Soc. (2)
(2017) 602–611.
[24] Y. Chen, Z. Pen, Q. Wang, et al., Crystalline structure, ferroelectric properties, and
electrical conduction characteristics of W/Cr co-doped Bi4Ti3O12 ceramics, J. Alloys
Compd. 612 (2014) 120–125.
[25] Y. Chen, D. Liang, Q. Wang, et al., Microstructures, dielectric, and piezoelectric prop-
erties of W/Cr co-doped Bi4Ti3O12 ceramics, J. Appl. Phys. 116 (7) (2014), 074108. .
[26] Y. Chen, C. Miao, S. Xie, et al., Fracture behaviors and ferroelastic deformation in W/
Cr co-doped Bi4Ti3O12 ceramics, J. Am. Ceram. Soc. 99 (6) (2016) 2103–2109.
[27] Y. Chen, C. Miao, S. Xie, et al., Microstructural evolutions, elastic properties and me-
chanical behaviors of W/Cr Co-doped Bi4Ti3O12 ceramics, Mater. Des. 90 (2016)
628–634.
[28] I.A. Md, M.A. Gafur, S.M. Islam, Sintering characteristics of La/Nd doped Bi4Ti3O12
bismuth titanate ceramics, Sci. Sinter. 47 (2015) (2015) 175–186.
[29] S. Kumar, K.B.R. Varma, Dielectric relaxation in bismuth layer-structured BaBi4Ti4O15
ferroelectric ceramics, Curr. Appl. Phys. 11 (2) (2011) 203–210.
[30] P. Fang, Z. Xi, W. Long, X. Li, J. Li, Structure and electrical properties of SrBi2Nb2O9-
based ferroelectric ceramics with lithium and cerium modification, J. Alloys
Compd. 575 (2013) 61–64.
[31] S.A. Ivanov, T. Sarkar, E.A. Fortalnova, et al., Composition dependence of the multi-
functional properties of Nd-doped Bi4Ti3O12 ceramics, J. Mater. Sci. Mater. Electron.
28 (11) (2017) 7692–7707.
[32] T. Jardiel, M. Villegas, A. Caballero, et al., Solid-state compatibility in the system
Bi2O3–TiO2–Bi2WO6, J. Am. Ceram. Soc. 91 (1) (2008) 278–282.
[33] G.G. Pisarenko, V.M. Chushko, S.P. Kovalev, Anisotropy of fracture toughness of pie-
zoelectric ceramics, J. Am. Ceram. Soc. 68 (5) (1985) 259–265.
[34] K.G. Webber, E. Aulbach, T. Key, et al., Temperature-dependent ferroelastic
switching of soft lead zirconate titanate, Acta Mater. 57 (15) (2009)
4614–4623.
[35] I. Pozdnyakova, G. Bruno, A.M. Efremov, et al., Stress-dependent elastic properties of
porous microcracked ceramics, Adv. Eng. Mater. 11 (12) (2010) 1023–1029.
[36] M.R. Daymond, The determination of a continuum mechanics equivalent elastic
strain from the analysis of multiple diffraction peaks, J. Appl. Phys. 96 (8) (2004)
4263–4272.
[37] S.E. Cummins, L.E. Cross, Electrical and optical properties of ferroelectric Bi4Ti3O12
single crystals, J. Appl. Phys. 39 (5) (1968) 2268–2274.
[38] Y. Kitanaka, S. Katayama, Y. Noguchi, et al., Electric-field-stabilized ferroelastic do-
main walls in monoclinic Bi4Ti3O12 crystals, Jpn. J. Appl. Phys. 46 (10) (2007)
7028–7030.
[39] J.L. Jones, M. Hoffman, S.C. Vogel, Ferroelastic domain switching in lead zirconate ti-
tanate measured by in situ neutron diffraction, Mech. Mater. 39 (4) (2007)
283–290.
[40] H. Kessler, H. Balke, A continuum analysis of the driving force of ferroelectric/
ferroelastic domain wall motions, J. Mech. Phys. Solids 54 (1) (2006)
113–127.
[41] M. Kamlah, Q. Jiang, A constitutive model for ferroelectric PZT ceramics under uni-
axial loading, Smart Mater. Struct. 8 (4) (1999) 441.
[42] J.L. Jones, M. Hoffman, S.C. Vogel, Domain switching under cyclic mechanical loading
in lead zirconate titanate, J. Am. Ceram. Soc. 89 (11) (2006) 3567–3569.
64 Y. Chen et al. / Materials and Design 152 (2018) 54–64
[43] S. Pojprapai, J.L. Jones, T. Vodenitcharova, et al., Investigation of the domain
switching zone near a crack tip in pre-poled lead zirconate titanate ceramic via in
situ X-ray diffraction, Scr. Mater. 64 (1) (2011) 1–4.
[44] A.A. Azlan, W. Krengvirat, A.F.M. Noor, et al., Sintering and characterization of rare
earth doped bismuth titanate ceramics prepared by soft combustion synthesis,
Sintering of Ceramics - New Emerging Techniques, InTech, 2012.
[45] J.L. Jones, E.B. Slamovich, K.J. Bowman, Domain texture distributions in tetragonal
lead zirconate titanate by x-ray and neutron diffraction, J. Appl. Phys. 97 (3)
(2005) 1194.
[46] S. Pojprapai, J.L. Jones, A.J. Studer, et al., Ferroelastic domain switching fatigue in lead
zirconate titanate ceramics, Acta Mater. 56 (7) (2008) 1577–1587.
[47] B.X. Huang, J. Malzbender, The effect of an oxygen partial pressure gradient on the
mechanical behavior of perovskite membrane materials, J. Eur. Ceram. Soc. 34 (7)
(2014) 1777–1782.
[48] R. Hill, The Mathematical Theory of Plasticity, Clarendon Press, Oxford, 1950.
[49] F. Meschke, A. Kolleck, G.A. Schneider, R-curve behaviour of BaTiO3 due to stress-
induced ferroelastic domain switching, J. Eur. Ceram. Soc. 17 (9) (1997) 1143–1149.
[50] C. Mercer, J.R. Williams, D.R. Clarke, et al., On a ferroelastic mechanism governing
the toughness of metastable tetragonal-prime (t′) yttria-stabilized zirconia, Proc.
R. Soc. Lond. A Math. Phys. Eng. Sci. 463 (2081) (2007) 1393–1408.
[51] X. Ren, W. Pan, Mechanical properties of high-temperature-degraded yttria-
stabilized zirconia, Acta Mater. 69 (2014) 397–406.
[52] Karun Mehta, Anil V. Virkar, Fracture mechanisms in ferroelectric-ferroelastic lead
zirconate titanate (Zr:Ti = 0.54: 0.46) ceramics, J. Am. Ceram. Soc. 73 (3) (1990)
567–574.
[53] G.R. Anstis, P. Chantikul, B.R. Lawn, et al., A critical evaluation of indentation tech-
niques for measuring fracture toughness. I.—direct crack measurements, J. Am.
Ceram. Soc. 64 (9) (1981) 533–538.
[54] B. Lawn, Fracture of Brittle Solids, Cambridge University Press, 1993.
[55] H. Cao, A.G. Evans, Nonlinear deformation of ferroelectric ceramics, J. Am. Ceram.
Soc. 76 (4) (1993) 890–896.
[56] C. Zhang, X. Hu, T. Sercombe, et al., Prediction of ceramic fracture with normal dis-
tribution pertinent to grain size, Acta Mater. 145 (2018) 41–48.
[57] Bibi Malmal Moshtaghioun, et al., Grain size dependence of hardness and fracture
toughness in pure near fully-dense boron carbide ceramics, J. Eur. Ceram. Soc. 36
(7) (2016) 1829–1834.