You are on page 1of 8

Available online at www.sciencedirect.

com

ScienceDirect
Acta Materialia 80 (2014) 48–55
www.elsevier.com/locate/actamat

Relationship between electromechanical properties and phase


diagram in the Ba(Zr0.2Ti0.8)O3–x(Ba0.7Ca0.3)TiO3 lead-free
piezoceramic
Matias Acosta a, Nikola Novak a,⇑, Wook Jo b, Jürgen Rödel a
a
Institute of Materials Science, Technische Universität Darmstadt, Darmstadt 64287, Germany
b
School of Materials Science and Engineering, Ulsan National Institute of Science and Technology, Ulsan 689-798, Republic of Korea

Received 7 May 2014; received in revised form 4 July 2014; accepted 26 July 2014
Available online 24 August 2014

Abstract

The Ba(Zr0.2Ti0.8)O3–x(Ba0.7Ca0.3)TiO3 system was synthesized in a wide compositional range in order to study the relationship
between its phase diagram and electromechanical properties. Phase transitions were marked using peaks in temperature-dependent per-
mittivity, providing up to three transitions from the rhombohedral phase to an orthorhombic, tetragonal and finally cubic phase, which
meet in a region that is termed the phase convergence region in this work. In situ small and large signal electromechanical properties were
studied as a function of temperature with specific emphasis on these transitions. A small signal piezoelectric coefficient, d33, presents max-
imized values at the transition from the orthorhombic to the tetragonal phase, while a large signal piezoelectric coefficient, d 33 , does so at
both rhombohedral to orthorhombic and to tetragonal phase transitions. Maximum polarization Pmax was the only quantity determined
that had a clear maximum at the phase convergence region.
Ó 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Electroceramic; Perovskites; Ferroelectric; Piezoelectric; Lead-free piezoceramics

1. Introduction [3,4]. The lead-free (1  x)Ba(Zr0.2Ti0.8)O3 (BZT)–x


(Ba0.7Ca0.3)TiO3 (BCT), first reported by Liu and Ren [5],
Piezoelectric materials are an important class of func- is one of the key lead-free candidates. At room temperature
tional materials. Lead zirconate titanate (Pb(Zr,Ti)O3, the compositionally optimized material presents a
PZT) piezoceramic solid solutions are used in a large variety piezoelectric coefficient d33  600 pC N1 and large signal
of applications, such as actuators, transducers and sensors, d 33  1100 pm V1 at 0.5 kV mm1 [5]. These electrome-
due their superior piezoelectric properties that are adjustable chanical responses clearly qualify this material system for
for each specific application [1]. Recently, new legislation has certain applications, outperforming lead-free and even most
been passed prohibiting the use of lead in certain applica- PZT materials, albeit only within a limited temperature
tions [2]. Therefore, much effort is required to develop range [4].
lead-free piezoceramics. To date, it seems that PZT cannot Liu and Ren [5] reported the phase diagram of the BZT–
be replaced on a broad basis, but for certain specific applica- BCT system, in which they observed a “tilted morphotrop-
tions lead-free materials present promising properties, some ic phase boundary” (MPB) separating rhombohedral (R)
of which are even superior to lead-containing piezoceramics and tetragonal (T) phases. They also highlighted the pres-
ence of a tricritical point (TCP) between cubic (C)–R–T
⇑ Corresponding author. Tel.: +49 6151 16 6302. phases, similar to that found in lead-containing materials
E-mail address: Novak@ceramics.tu-darmstadt.de (N. Novak). [6]. Recently, Keeble et al. [7] and Tian et al. [8] proposed

http://dx.doi.org/10.1016/j.actamat.2014.07.058
1359-6454/Ó 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
M. Acosta et al. / Acta Materialia 80 (2014) 48–55 49

a bridging orthorhombic (O) phase between the R and T compositions were chosen deliberately, taking into account
phases. Keeble et al. [7] proposed the term “convergence representative phase diagrams proposed in the literature
region” instead of TCP, whereas Tian et al. [8] utilized [5,7] with special emphasis on the TCP [5] or phase conver-
the term “diffuse phase transition”, since they were unable gence region [7]. All the powders were added to homemade
to clearly resolve the transition region. Moreover, a third nylon milling containers with yttria stabilized zirconia balls
phase transition in the system was also outlined by pyro- and ethanol. The suspensions were then ball-milled in a
electric [9], Raman [10] and elastic studies [10], although planetary ball mill (Fritsch Pulverisette 5) for 5 h at
it was not recognized as such in those publications. Bjørne- 250 rpm and dried at 90 °C for 24 h. The powders were cal-
tun Haugen et al. [11] proposed that the third anomaly in cined at 1300 °C for 2 h with a heating rate of 5 K min1
the system might be related to the coexistence of two differ- and subsequently ball-milled for 15 h with the previously
ent symmetries and changes in the different phase fractions, described parameters. Disk-shape samples of 10 mm in
supporting the work of Ehmke et al. [12]. Hence, the phase diameter and 1 mm thick (30 mg of powder) were cold
diagram of the system remains controversial. As our dielec- isostatically pressed (KIP 100 E, Weber-Pressen) at
tric measurements point to an intermediate phase between 300 MPa for 1.5 min. The green bodies obtained were sin-
the T and R phases, we will refer to it as an orthorhombic tered in covered zirconia crucibles with atmospheric pow-
phase and term the narrow regime where the different der at 1500 °C for 2 h and a heating rate of 5 K min1.
phases coexist the phase convergence region [7]. Sintered pellets were ground and polished down to
In order to explain the outstanding electromechanical 600 lm. Electrodes were painted with a silver paste and
properties of the system, Liu and Ren [5] rationalized that burnt-in at 400 °C for 2 h. The reproducibility of represen-
the free energy of the MPB composition connected to the tative compositions was corroborated with at least two dif-
tricritical point is weakly anisotropic and thus almost inde- ferent batches. Specifically, key compositions were
pendent of polarization direction [5,13–15]. This property, produced two to four times. From each batch at least 16
accompanied by a softening of the lattice [13,16], results in samples were produced and several of them were tested,
an almost negligible energy barrier for polarization rota- allowing us to choose representative samples on which to
tion [5,17]. In other words, the enhanced response along perform further experiments.
the so-called MPB was proposed to result from phase tran- The density of all compositions was determined using
sition instabilities within a narrow temperature range. If the Archimedes method. The microstructure analysis was
this is the case, it should be expected that the anisotropy performed on polished sample surfaces. To reveal the grain
becomes negligible by approaching the TCP [5,13–15,17]. boundaries, chemical etching was performed for 90 s with
Therefore, maximized properties at the TCP are expected 50 vol.% diluted HCl and a few drops of HF. The samples
due to polarization rotation and extension originating from were imaged using an optical microscope (DMRME, Leica
the coexistence of ferroelectric and paraelectric phases [18]. Mikroskopie und Systeme GmbH). The average grain size
A recent quantitative analysis using polarization dynamics was obtained by the mean intercept length method, with a
[19] suggests that the energy barrier for domain switching is numerical multiplication factor of 1.56 [20]. For reliable
minimum at the so-called MPB. statistical information, around 200 grains of at least three
In this study we report a detailed compositional study of different representative areas of samples were analyzed.
the BZT–BCT system by means of dielectric, small signal Errors on density measurements were omitted because their
d33 and large signal d 33 properties as functions of tempera- magnitude was not statistically relevant relative to the
ture. The methodology employed allows a complete assess- reported values. Errors in the mean grain size are reported
ment of the role of the convergence region on the as standard deviations in Table 1.
electromechanical properties. Of specific interest, therefore, Temperature- and frequency-dependent real e0r and
is which – if any – of the electromechanical properties peak imaginary e00r relative permittivity were obtained on unp-
at the convergence region. The salient question then is: is oled samples for all compositions. In all cases the complex
the convergence region responsible for the outstanding relative permittivity was measured in the temperature range
electromechanical properties or is it related to the ferroelec- from 100 to 120 °C, with a heating rate of 2 K min1, and
tric/ferroelectric phase transitions in the system previously
denominated as MPB?
Table 1
Average grain size and relative density of each BZT–BCT composition.
2. Experimental
Material Average grain size (lm) Density

The ceramic powders were produced via a mixed oxide BZT–0.3BCT 28 ± 7 98


BZT–0.32BCT 20 ± 3 98
route using reagent grade oxides and carbonates (Alfa
BZT–0.35BCT 36 ± 8 94
Aesar GmbH & Co. KG). The raw chemicals BaCO3 (pur- BZT–0.37BCT 28 ± 8 97
ity 99.8%), TiO2 (purity 99.6%), CaCO3 (purity 99.5%) and BZT–0.4BCT 25 ± 5 98
ZrO2 (purity 99.5%) were mixed according to the stoichi- BZT–0.45BCT 36 ± 9 97
ometric formula (1  x)Ba(Zr0.2Ti0.8)O3–x(Ba0.7Ca0.3)TiO3 BZT–0.5BCT 35 ± 8 97
BZT–0.6BCT 26 ± 6 98
with x = 0.3, 0.32, 0.35, 0.37, 0.4, 0.45, 0.5 and 0.6. The
50 M. Acosta et al. / Acta Materialia 80 (2014) 48–55

in the frequency range from 100 Hz to 10 kHz by an properties, it can be concluded that the BZT–BCT system
Impedance Analyzer Alpha-A equipped with a cryostat was successfully synthesized.
(Novocontrol Technologies, Germany). The amplitude of
the probing ac electric signal was 1 V for all measurement 3.2. Relative permittivity
frequencies. The standard deviation of the dielectric mea-
surements was minor in comparison to the measurement Fig. 1 provides the real (e0r ) and imaginary (e00r ) parts of
error; therefore, the error on the dielectric properties is relative permittivity as functions of temperature and fre-
expressed as experimental error and is below 1%. quency. Three representative compositions were selected
The small signal piezoelectric coefficient d33 was mea- to depict these results, namely x = 0.3, 0.4 and 0.45. The
sured in situ as a function of temperature using a cus- relative permittivity can be utilized to detect structural
tom-designed device. The apparatus consists of a furnace phase transitions [23]. These phase transitions are accom-
controlled by two thermocouples and a laser vibrometer panied by a dielectric anomaly independent of the measure-
(Polytec sensor head OFV-505 and front-end VDD-E- ment frequency. Therefore, the temperature for phase
600). A pinhole was made in the middle of a thermally insu- transitions can be obtained from the local maxima of e00r
lating alumina covering lid, thereby allowing the laser to [23]. From the temperature-dependent e00r of all of the com-
pass through. Further details on the measuring machine positions investigated, a pseudo-binary phase diagram for
are presented elsewhere [21]. A heating rate of 2 K min1 the BZT–BCT system can be constructed (Fig. 2). Note
was utilized for the measurement, and points were obtained that it is consistent, within experimental error, with the
in steps of 5 °C. A sinusoidal wave with a frequency of one recently proposed by Keeble et al. [7], although both
10 kHz and an amplitude of 10 V (equivalent to a field the synthesis route and the techniques utilized to detect
strength of less than 20 V mm1) was chosen as the input phase transitions differed. Following the work by Keeble
voltage. Prior to measurement, all samples were poled at at al. [7], we term the phase between the rhombohedral
4 kV mm1 at room temperature for 10 min to ensure a and tetragonal phases the orthorhombic phase. A contour
well-poled initial state. The selection of such high fields plot of the normalized e0r as a function of temperature and
was done based on the results of Wu et al. [22]. Prior to composition is presented in Fig. 3. For comparison, the
each measurement, room-temperature d33 values were con-
trasted to values of a commercial Berlincourt meter
(YE2730, Sinocera, Inc.). 16000 x=0.3 x=0.4 x=0.45
Large signal strain and polarization as functions of tem- 12000
100 Hz
1 kHz
ε r´

perature were obtained using a commercial piezoelectric 8000


10 kHz

system TF analyzer 200 (aixACCT Systems, GmbH). A tri-


4000
angular wave with 5 Hz measurement frequency and
0
3 kV mm1 amplitude was employed in the temperature 600
range from 25 to 125 °C. The high-voltage amplitude was
εr´´

400
chosen deliberately to ensure complete saturation of the
material. All materials were tested from an initially poled 200
state after unipolar cycling at 3 kV mm1 to ensure a
0
reproducible state. For electromechanical measurements, -50 0 50 100
o
-50 0 50 100
o
-50 0
o
50 100
T ( C) T ( C) T( C)
the experimental error was considered. The error in the
large and small signal measurements is in the order of 5– Fig. 1. Relative permittivity real e0r and imaginary e00r parts of represen-
10 microstrain. The error in the polarization measurements tative compositions as functions of temperature and frequency. The peak
is in the order of 1%. Error bars were not utilized in the fig- positions indicating phase transitions are marked with dashed lines.
ures for experimental results because data markers were
selected deliberately in the same length scale as the reported
errors. 100
T R-O
90 T C
O-T

3. Results 80
T c

70
T (°C)

T
3.1. Density and microstructure 60
50
Density and mean grain size measurements are pre-
40 R O
sented in Table 1. The grain size varies non-monotonically
30
from 20 ± 3 lm at x = 0.32 to 36 ± 9 lm at x = 0.45. All
the materials present relative densities above 94%. Varia- 0.30 0.35 0.40 0.45 0.50 0.55 0.60
BZT-xBCT
tions in grain size and density are comparable to prior
reports [22]. The X-ray analysis of the samples revealed a Fig. 2. Phase diagram obtained from the imaginary part of the relative
single phase for all materials. Considering the mentioned permittivity e00r peaks at 1 kHz.
M. Acosta et al. / Acta Materialia 80 (2014) 48–55 51

Fig. 3. Contour plot of normalized real relative permittivity e0r at 1 kHz as Fig. 5. Contour plot of the small signal piezoelectric coefficient d33 as a
a function of temperature and composition. function of temperature and composition.

phase transitions previously discussed in Fig. 2 are super- temperature up to 95 °C, making this composition attrac-
imposed. In the normalized figure, local maxima appear tive for certain low-temperature applications.
insignificant. Maximum e0r values should occur, as The curves displayed in Fig. 4 were combined to pro-
observed, at temperatures slightly higher than the Tc duce a contour plot of d33 as a function of temperature
obtained from the local maxima of e00r . This is because its and composition (Fig. 5). The series of phase transitions
maximum coincides with the inflection point of e0r [24]. It identified from the dielectric properties is superimposed
is seen that the maximum value of e0r is given at around for comparison. Fig. 5 clearly reveals that the overall max-
x = 0.3 and 65 °C instead of the convergence region, which imum piezoelectric response of the BZT–BCT system
is consistent with the literature [5]. occurs along the O–T phase boundary, which is consistent
with the literature [8,27,28]. It is noted that the piezoelec-
3.3. Small signal piezoelectric properties tric coefficient exhibits local maxima at the convergence
region, but overall it is not as high as near to the O–T phase
The small signal piezoelectric coefficient d33 was mea- boundary.
sured in situ in all compositions as a function of tempera-
ture. These curves, together with the phase transitions 3.4. Large signal electromechanical response
obtained from dielectric properties, are depicted in Fig. 4.
All phase transitions in the system are accompanied by Fig. 6 provides temperature-dependent unipolar strain
local maxima in d33. Interestingly, the d33 does not present values at 3 kV mm1 for the three selected representative
the highest values for all compositions at the Tc but only compositions. It is apparent that at certain temperatures
for x < 0.4. For the range of 0.4 < x < 0.5, d33 is observed the strain output of the system is enhanced. Therefore,
to peak at the O–T phase boundary. For all compositions the d 33 was monitored as a function of temperature for
d33 retains significant values of 100 pC N1 even up to all eight compositions to facilitate comparison with the
10 °C above the Tc. Residual piezoelectricity above the Tc small signal d33. In order to ensure saturation of the
was also observed in other studies [10,25,26]. Meanwhile, BZT–BCT system, the d 33 values are presented at
on the tetragonal side (x = 0.6), the BZT–BCT system 3 kV mm1, i.e. a field strength 30 times larger than the
exhibits an almost temperature-insensitive d33 value of coercive field Ec utilized. This reflects the reduced d 33 in
300 pC N1 in the temperature range from room comparison to the values reported in the literature, which

600 x=0.3 x=0.32 x=0.35 x=0.37


d33 (pC/N)

400
200
0
600 x=0.4 x=0.45 x=0.5 x=0.6
d33 (pC/N)

400
200
0
50 100 50 100 50 100 50 100
T (°C) T (°C) T (°C) T (°C)

Fig. 4. In situ small signal piezoelectric coefficient d33 as a function of temperature. The peak positions indicating phase transitions are marked with
dashed lines.
52 M. Acosta et al. / Acta Materialia 80 (2014) 48–55

0.20 25 °C
x=0.3 x=0.4 x=0.45 35 °C
45 °C
0.15 55 °C
65 °C

Strain (% )
0.10 75 °C
85 °C

0.05

0.00
0 1 2 3 0 1 2 3 0 1 2 3
E (kV/mm) E (kV/mm) E (kV/mm)

Fig. 6. Large signal electromechanical response of representative compositions as a function of temperature measured at 5 Hz.

x=0.3 x=0.32 x=0.35 x=0.37


d33* (pm/V)

400

200

0
d33* (pm/V)

x=0.4 x=0.45 x=0.5 x=0.6


400

200

0
50 100 50 100 50 100 50 100
T (°C) T (°C) T (°C) T (°C)

Fig. 7. The large signal coefficient d 33 as a function of temperature calculated at 3 kV mm1. The peak positions indicating phase transitions are marked
with dashed lines.

were mostly obtained at around 0.5–1 kV mm1 [5]. It is


noted that the d 33 of the currently studied materials
obtained at the optimum driving field of 1 kV mm1 at
room temperature reaches 1000 ± 200 pm V1 (depend-
ing on the composition), consistent with literature values
[5]. Again, obtained d 33 values are presented in context to
the phase transitions obtained from dielectric properties
in Fig. 7. In contrast to the d33 curves (Fig. 4), the correla-
tion between the large signal d 33 and the phase transitions
is rather weak, with only smooth changes in the strain
obtained around the transitions. In all cases, a d 33 of Fig. 8. Contour plot of the large signal coefficient d 33 as a function of
200 pm V1 persists even at 10 °C above the Tc, consis- temperature and composition calculated at 3 kV mm1.
tent with the non-zero residual d33 (Fig. 4). The composi-
tion on the tetragonal side (x = 0.6) presents the highest
temperature stability; nevertheless, the temperature depen- transition (Fig. 5). As observed in the dielectric properties
dence is more pronounced than for the small signal d33. In and also in the small signal d33, the large signal d 33 is not
this case, a decay of d 33 from 350 to 250 pm V1 is maximized at the phase convergence region.
observed from room temperature up to 95 °C. Polarization values at 3 kV mm1, designated as maxi-
The d 33 values from Fig. 7 are summarized in a contour mum polarization Pmax and the remanent polarization
plot to visualize the electromechanical response of the Prem, were also monitored as functions of temperature
BZT–BCT system as a function of temperature and com- and composition. No clear trends in the Prem values could
position at high fields. This data is presented in Fig. 8, be detected, within experimental error, regardless of tem-
together with the phase diagram obtained from the dielec- perature and composition. Pmax is introduced in Fig. 9
tric properties. Interestingly, contrary to the small signal together with the phase diagram obtained from the dielec-
d33, there is a broad region in which the system presents tric properties. In contrast to the other electromechanical
a high electromechanical response. Nevertheless, the maxi- properties, Pmax remains relatively constant around all fer-
mum values still follow the phase transitions. Both phase roelectric/ferroelectric phase transitions. Nevertheless, it is
transitions from R–O and O–T present the highest d 33 val- maximized in the proximity of the convergence region.
ues. This indicates a different electromechanical output to Pmax acquires considerably high values (Pmax  13–
the small signal d33, which peaked only at the O–T phase 14 lC cm2), which are 45% higher than those at room
M. Acosta et al. / Acta Materialia 80 (2014) 48–55 53

an increase relative permittivity. Nevertheless, the


5 mol.% BCT discrepancy between the maximized proper-
ties and the triple points gives an indication that a more
complicated physical behavior may arise [5]. We are cur-
rently performing in situ field-dependent TEM to better
correlate the domain morphology near triple points to
the presented phenomenological approach. Hence, domain
wall mobility, size and morphology are all expected to play
a key role, as recently shown in the dielectric and piezoelec-
tric response of BT [32].

Fig. 9. Contour plot of Pmax as a function of temperature and compo- 4.3. Small signal piezoelectric properties
sition calculated at 3 kV mm1.
For BT, it was shown that reversible and irreversible
domain wall movement contribute to the small signal prop-
temperature (Pmax  8–9 lC cm2). Therefore, the peak of erties at fields above 5 V mm1 and at 20 Hz [33], giving
Pmax does not coincide with the response of either dielectric rise to a Rayleigh behavior. Regardless of the fact that
or large and small signal electromechanical properties. BZT–BCT is a very soft system [16], we assume, as a first
approximation, a negligible contribution of reversible and
4. Discussion irreversible domain wall movement with our chosen exper-
imental conditions of 0.2 V mm1 and 10 kHz. There-
4.1. Density and microstructure fore, extrinsic contributions to d33 will not be considered
in this section. The Rayleigh analysis of the system is
Based on the results from Hao et al. [26], the grain size beyond the scope of the present study, but should be car-
and density values (higher than 20 lm and 94%, respec- ried out for better clarification of the Rayleigh threshold
tively) of the samples under the current investigation are field of the system.
large enough to correlate the measured functional proper- For further insight, we ignore the polycrystalline and
ties with the composition and temperature (i.e. the presence multi-domain nature of the system and instead consider a
of phase transitions). Therefore, both grain size and density Landau phenomenological approach [14,15,34,35] to
are considered as secondary parameters influencing the describe the d33, as previously proposed for the BZT–
electromechanical properties of the system under the stud- BCT system [5,30]. For a material with a high-temperature
ied synthesis conditions. cubic phase, the intrinsic electric-field-induced strain can be
obtained with Eq. (1)
4.2. Dielectric properties S ij ¼ Qijkl  P k  P l ð1Þ
As previously mentioned, the local maxima of the e00r where Qijkl is the electrostrictive coefficient and P k;l the
were attributed to the phase transitions of the system, spontaneous polarization. Based on the general definition
@S
which is in good agreement with the phase diagram from of the small signal piezoelectric coefficient d mij ¼ @Eijm , the
Keeble et al. [7]. Contrary to the frequency-dependent local piezoelectric coefficient can be deduced, as demonstrated
maxima commonly observed in relaxors [29], it was in Eq. (1), by ignoring the temperature dependence of the
observed that the local maxima did not present frequency electrostrictive coefficient Qijkl [36]:
dispersion consistent with the dielectric response of classi-  
cal ferroelectrics [1]. @P k @P l
d mij ¼ Qijkl  P l  þ Pk 
The highest values of relative permittivity do not occur @Em @Em
at the convergence region but at x = 0.3 and 65 °C, consis-
¼ Qijkl  ðP l  ekm þ P k  elm Þ ð2Þ
tent with the literature [5]. The domain morphology of
BZT–BCT near x = 0.5 has already been revealed by where ekm;lm is the relative permittivity. From Eq. (2) the
Gao et al. [30] using TEM. They observed that a hierarchi- piezoelectric coefficient d mij is proportional to the electro-
cal nanodomain structure arises around the O–T phase strictive coefficient Qijkl, the spontaneous polarization P l;k
boundary. These results could be rationalized within a clas- and the relative permittivity ekm;lm . We stress that this result
sical domain theory, in which domain size is proportional is independent of the symmetry of the material, the crystal-
to the domain wall energy, leading to minimized domains lographic direction and the flattening or isotropization of
near phase transitions. It is further consistent with theoret- the free energy. Since the electrostrictive coefficients Qijkl
ical predictions [31] indicating that domain miniaturization are not known for BZT–BCT and are normally considered
is also expected near phase instabilities in general. This to be temperature independent [34], we ignore their contri-
leads to an increased domain wall density and therefore bution to temperature dependence. The maxima observed
54 M. Acosta et al. / Acta Materialia 80 (2014) 48–55

from the piezoelectric coefficients in Fig. 4 at the first phase energy barrier minimum for switching from around
transition (namely R–C, R–O, O–T or T–C) are due to a x = 0.4 to x = 0.5 at room temperature. This is due to a
softening of the lattice [16], which is also accompanied by balance between the activation barrier per unit volume
local maxima in the e0r and thus in d33, as predicted in and the critical volume of nucleating domains, which is a
Eq. (2). For compositions with more than one phase tran- parameter governed by the domain morphology and size.
sition in Fig. 4, we observed that either the second phase Fig. 8 verifies that a large contribution of the switching
transition (O–C, O–T or T–C) or third phase transition around the ferroelectric phase transitions should originate
(T–C) present only local maxima of d33, indicating that from non-180° domain wall motion, leading to high d 33 val-
polarization extension [13] may not play a major role in ues. Moreover, the broad minimum for switching, rational-
the electromechanical properties of the system. This has ized as a switching energy barrier, also occurs at higher
recently been verified by in situ high-energy X-ray diffrac- temperatures around both phase transitions, as depicted
tion, comparing lattice extension and domain switching from the maximized d 33 . The broad switching minimum
as functions of the electric field [37]. Since maximized prop- can be attributed to the hierarchical domain structure
erties are not observed at high-temperature phase transi- observed by Gao et al. [30] due to the proximity to the
tions, a continuous loss of spontaneous polarization of phase transitions, which aids non-180° domain wall switch-
the system should be responsible for the non-maximized ing [37]. On the other hand, the d33 was found to peak only
properties in Fig. 5 based on Eq. (2). This phenomenon at the O–T phase transition. This leads to a complex sce-
was demonstrated, specifically at compositions next to nario in which small signal and large signal properties are
the phase transitions, by Benabdallah et al. [9]. Interest- not fully coupled. Although switching should also be aided
ingly, Benabdallah et al. [9] observed a continuous decay at the convergence region and Tc due to both hierarchical
of the spontaneous polarization Ps with increasing temper- domain structures and the minimum switching energy bar-
ature. Hence, the convergence region is not able to provide rier, high d 33 values are not observed – probably due to the
maximized small signal piezoelectric coefficients due to the small contribution to the strain per switching event. This
already low spontaneous polarization. could be rationalized based on the results by Keeble
Instead, the maximum values of d33 are observed at and et al. [7], in which local distortions from a cubic symmetry
slightly above the O–T phase boundary (Fig. 5). The ability are very small at such temperatures. This strain per switch-
of the system to retain the spontaneous polarization is ing event can be quantified using a figure of non-cubic dis-
therefore critical in determining the high d33 values. It tortion for the tetragonal symmetry as c/a  1. For the
remains an open question for future investigations why tetragonal side of the BZT–BCT phase diagram, an
the R–O phase transition does not yield a high piezoelectric increased switching contribution was found with decreased
coefficient d33. Since the composition with x = 0.6 is com- distance to ferroelectric/ferroelectric phase transitions [37].
positionally away from all phase transitions, it presents a In general, there is a maximum in d33 as well as in d 33 for an
much more stable spontaneous polarization, which is intermediate non-cubic distortion [40]. A high distortion
reflected in its almost temperature-insensitive small signal provides a high strain per switching event, but only few
d33 (Fig. 4). domains switch, whereas a low distortion provides many
switching events with small contributions per domain wall
4.4. Large signal electromechanical response [40].
Fig. 9 demonstrates that the maximum polarization
The large signal strain measurements and calculation of Pmax peaks at around the Tc and presents maximized values
d 33 were done deliberately at very high fields (more than at the convergence region. Since high strain is not involved
one order of magnitude higher than the Ec for all compo- in this composition and temperature regime (Fig. 8), we
sitions) in order to avoid unsaturated loops. Since intrinsic attribute the high Pmax values mostly to 180° domain
contributions to strain are known from the previous sec- switching, which is known to be an extrinsic contribution
tion and non-saturation can be discarded, so d 33 can be uti- to polarization that does not involve remanence in the
lized to study the large signal properties and the extrinsic strain [1]. This contribution to Pmax is more substantial
contributions on them. It is already known that domain than that of non-180° domain switching, since we do not
wall contributions can account for more than 50% of the observe maximized values of Pmax along the phase
total strain output in d 33 , due to non-180° domain wall transitions.
switching [38]. Figs. 8 and 9 reveal that non-zero values of d 33 and Pmax
The domain size, morphology, mobility and activation occur even at 10 °C above the Tc. Hence, a considerable
barrier all play a major role in determining the ease of amount of 180° domain switching may give rise to the high
switching and thus the electromechanical output of BZT– values of Pmax around the convergence region and to a les-
BCT [19,30]. Moreover, all these contributions are ser amount of non-180°, giving rise to non-zero electrome-
expected to be highly dependent on the crystal structure chanical output [41]. Moreover, it has already been verified
[10,19,30]. As previously shown [19] based on an inhomo- in the BZT–BCT system that an electric field induces phase
geneous field mechanism model [39], there is a broad changes and an increased Tc [42].
M. Acosta et al. / Acta Materialia 80 (2014) 48–55 55

5. Conclusions [13] Damjanovic D. Appl Phys Lett 2010;97:062906.


[14] Haun MJ, Furman E, Jang SJ, Cross LE. Ferroelectrics 1989;99:63.
[15] Haun MJ, Furman E, McKinstry HA, Cross LE. Ferroelectrics
It is suggested that the existence of a triple point or a 1989;99:27.
general phase convergence region is not the cause of the [16] Xue D, Zhou Y, Bao H, Zhou C, Gao J, Ren X. J Appl Phys
high electromechanical properties in the BZT–BCT system. 2011;109:054110.
Instead, the highest small and large signal piezoelectric [17] Kutnjak Z, Petzelt J, Blinc R. Nature 2006;441:956.
coefficients occur at ferroelectric–ferroelectric phase transi- [18] Porta M, Lookman T. Phys Rev B 2011;83:174108.
[19] Zhukov S, Genenko YA, Acosta M, Humburg H, Jo W, Rödel J, von
tions and are rather small at the phase convergence region. Seggern H. Appl Phys Lett 2013;103:152904.
This behavior is attributed to the low spontaneous polari- [20] Mendelson MI. J Am Ceram Soc 1969;52:443.
zation related to the small non-cubic distortion at the phase [21] Leist T, Chen J, Jo W, Aulbach E, Suffner J, Rödel J, et al. J Am
convergence region. The phase convergence region leads Ceram Soc 2012;95:711.
only to a maximum in only one property, polarization. [22] Wu J, Xiao D, Wu W, Chen Q, Zhu J, Yang Z, et al. J Eur Ceram Soc
2012;32:891.
This is suggested to be due to high 180° domain switching [23] Anton EM, Jo W, Damjanovic D, Rödel J. J Appl Phys
and an increase in transition temperature with electric field. 2011;110:094108.
[24] Bhaumik I, Singh G, Ganesamoorthy S, Bhatt R, Karnal AK, Tiwari
Acknowledgment VS, et al. J Cryst Growth 2013;375:20.
[25] Su S, Zuo R, Lu S, Xu Z, Wang X, Li L. Curr Appl Phys
2011;11:S120.
This work was supported by the AdRIA Hesse state [26] Hao J, Bai W, Li W, Zhai J, Randall C. J Am Ceram Soc
center for Adaptronics. 2012;95:1998.
[27] Wu J, Habibul A, Cheng X, Wang X, Zhang B. Mater Res Bull
References 2013;48:4411.
[28] Li W, Xu Z, Chu R, Fu P, Zang G, Zhai J. J Am Ceram Soc
[1] Jaffe B, Cook WR, Jaffe H. Piezoelectric ceramics. London, New 2011;94:4131.
York: Academic Press; 1971. [29] Cross LE. Ferroelectrics 1987;76:241.
[2] EU-Directive 2011/65/EU: Restriction of the use of certain hazardous [30] Gao J, Xue D, Wang Y, Wang D, Zhang L, Wu H, et al. Appl Phys
substances in electrical and electronic equipment (RoHS). Off J Eur Lett 2011;99:092901.
Union 2011;L 174:88. [31] Rossetti GA, Khachaturyan AG. Appl Phys Lett 2007;91:072909.
[3] Jo W, Dittmer R, Acosta M, Zang J, Groh C, Sapper E, et al. J [32] Ghosh D, Sakata A, Carter J, Thomas PA, Han H, Nino JC, et al.
Electroceram 2012;29:71. Adv Funct Mater 2014;24:885.
[4] Rödel J, Jo W, Seifert KTP, Anton E-M, Granzow T, Damjanovic D. [33] Hall DA. Ferroelectrics 1999;223:319.
J Am Ceram Soc 2009;92:1153. [34] Lines ME, Glass AM. Principles and applications of ferroelectrics
[5] Liu W, Ren X. Phys Rev Lett 2009;103:257602. and related materials. Oxford: Clarendon Press; 1979.
[6] Cox DE, Noheda B, Shirane G, Uesu Y, Fujishiro K, Yamada Y. [35] Haun MJ, Furman E, Jang SJ, Cross LE. Ferroelectrics 1989;99:13.
Appl Phys Lett 2001;79:400. [36] Zhang QM, Wang H, Kim N, Cross LE. J Appl Phys 1994;75:454.
[7] Keeble DS, Benabdallah F, Thomas PA, Maglione M, Kreisel J. Appl [37] Tutuncu G, Li B, Bowman K, Jones JL. J Appl Phys
Phys Lett 2013;102:092903. 2014;115:144104.
[8] Tian Y, Wei L, Chao X, Liu Z, Yang Z, Damjanovic D. J Am Ceram [38] Damjanovic D. The science of hysteresis. New York: Elsevier; 2006
Soc 2012:496. [chapter 4].
[9] Benabdallah F, Simon A, Khemakhem H, Elissalde C, Maglione M. J [39] Genenko YA, Zhukov S, Yampolskii SV, Schütrumpf J, Dittmer R,
Appl Phys 2011;109:124116. Jo W, et al. Adv Funct Mater 2012;22:2058.
[10] Damjanovic D, Biancoli A, Batooli L, Vahabzadeh A, Trodahl J. [40] Leist T, Granzow T, Jo W, Rödel J. J Appl Phys 2010;108:014103.
Appl Phys Lett 2012;100:192907. [41] McQuarrie M. J Am Ceram Soc 1956;39:54.
[11] Bjørnetun Haugen A, Forrester JS, Damjanovic D, Li B, Bowman [42] Singh G, Bhaumik I, Ganesamoorthy S, Bhatt R, Karnal AK, Tiwari
KJ, Jones JL. J Appl Phys 2013;113:014103. VS, et al. Appl Phys Lett 2013;102:082902.
[12] Ehmke MC, Ehrlich SN, Blendell JE, Bowman KJ. J Appl Phys
2012;111:124110.

You might also like