Large piezoelectric properties induced by doping ionic pairs in BaTiO3 ceramics

Available online at www.sciencedirect.com

ScienceDirect Acta Materialia 79 (2014) 84–92 www.elsevier.com/locate/actamat

Large piezoelectric properties induced by doping ionic pairs in BaTiO3 ceramics D. Xu a, W.L. Li a,b, L.D. Wang a, W. Wang a, W.P. Cao a, W.D. Fei a,⇑ b

a School of Materials Science and Engineering, Harbin Institute of Technology, Harbin 150001, People’s Republic of China National Key Laboratory of Science and Technology on Precision Heat Processing of Metals, Harbin Institute of Technology, Harbin 150001, People’s Republic of China

Received 13 February 2014; received in revised form 15 June 2014; accepted 8 July 2014

Abstract The high piezoelectric properties of piezoelectric ceramics are typically obtained by inducing a phase transition between two ferroelectric phases, using either the morphotropic phase boundary (MPB) or the polymorphic phase transition (PPT). Here we demonstrate that neither the MPB nor the PPT is necessary to achieve high piezoelectric properties. Our results show that the optimized distribution of Li+–Al3+ pairs parallel to the [0 0 1] direction, as found in our xLiAlSiO4/BaTiO3 (xLAS/BT) lead-free piezoelectric ceramic system prepared from ordinary raw materials by conventional solid-state reaction sintering, can generate a large piezoelectric constant (d33) of 378 pC/N when x = 7.5 mol.%. The d33 value of the 7.5 mol.% LAS/BT ceramic is more than three times that of the pure BT ceramic. The distortion connected to the proposed Li+–Al3+ pairs locally creates unit cells of less than tetragonal symmetry, and these lowsymmetry cells can be responsible for the high piezoelectric response. The high-temperature stability testing reveals that these doped ceramics are usable at temperatures as high as 120 °C. This piezoelectric mechanism coming from the doping ionic pairs provides a new method to achieve large piezoelectric properties in a wide range of ABO3-type perovskite systems. Ó 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Perovskites; Piezoelectricity; Lead-free; Doping ionic pairs

1. Introduction Pb(Zr,Ti)O3 (PZT) ceramics have for many years represented one of the most commonly used piezoelectric materials and have played a vital role in various applications, such as actuators, sensors, capacitors, resonators and transducers. The exceptional piezoelectric properties of these materials are achieved by using a composition approximating Pb(Zr0.52Ti0.48)O3, which exhibits a composition-induced phase transition between two ferroelectric phases known as a morphotropic phase boundary (MPB) [1–4]. Recently, health and environmental considerations ⇑ Corresponding author. Tel./fax: +86 451 86413908.

E-mail address: [email protected] (W.D. Fei).

have prompted the extensive investigation of lead-free piezoelectric ceramics with excellent properties as replacements for toxic PZT ceramics. The excellent piezoelectric properties of PZT ceramics with MPB compositions have led to interest in a variety of binary and ternary lead-free piezoelectric ceramics with similar MPB compositions, including Na0.5Bi0.5TiO3–BaTiO3 [5–7], BiFeO3–BaTiO3 [8,9], Na0.5Bi0.5TiO3–K0.5Bi0.5TiO3–BaTiO3 [10–12], Na0.5 Bi0.5TiO3–K0.5Bi0.5TiO3–BiFeO [13] and Na0.5K0.5NbO3– LiSbO3–BiFeO3 [14]. However, the piezoelectric performance of existing lead-free ferroelectric systems still cannot compete with that of PZT, especially with regard to the performance of high-end PZT ceramics (those with a high piezoelectric constant (d33) of 600 pC/N and a high depolarization temperature (Td) of 330 °C) [15–17]. Liu and

http://dx.doi.org/10.1016/j.actamat.2014.07.023 1359-6454/Ó 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

D. Xu et al. / Acta Materialia 79 (2014) 84–92

Ren [18] have reported that the Ba(Ti0.8Zr0.2)O3–x(Ba0.7Ca0.3)TiO3 (BZT–BCT) ceramic system exhibits a very large d33 value of 620 pC/N at an optimal composition. Their work involved designing an MPB material analogous to the PZT system and found that the high piezoelectricity of the BZT–BCT system originates from the compositional proximity of its MPB to the tricritical triple point (TCP) formed from combined cubic paraelectric (C), ferroelectric rhombohedral (R) and tetragonal (T) phases. Unfortunately, its relatively low Curie temperature (TC) limits the overall usefulness of the BZT–BCT system in practical applications since the operational temperature of some piezoelectric components is limited to half the TC value. More effort is therefore required to further increase the TC of the BZT–BCT system. In addition to the MPB, another means of achieving a high level of piezoelectricity is polymorphic phase transition (PPT), which is mainly found in the (K,Na)NbO3 (KNN) system [19–22]. As an example, a piezoelectric coefficient as high as 416 pC/N has been exhibited by a textured KNN-based ceramic with codopants of Li, Ta and Sb [21], while the d33 values of non-textured KNN-based ceramics are in the range of 200–300 pC/N. The enhanced piezoelectric properties evident with the KNN system are obtained owing to the coexistence of orthorhombic (O) and T ferroelectric phases when the PPT temperature (TO-T) is shifted downward to near room temperature, and a temperature-driven ferroelectric–ferroelectric phase transition such as this may provide an easy path for polarization rotation. Although high piezoelectricity has been reported in these KNN-based ceramics, the strong temperature dependence of piezoelectric properties stemming from associated PPTs means that they are unusable in devices requiring high thermal stability. Zuo et al. [19], for example, have found that the d33 values of Sb-doped KNN ceramics decrease rapidly with increasing temperature. It is therefore worth noting that good piezoelectric properties do not always equate to high performance in piezoelectric ceramics, so there is an urgent need to develop a lead-free candidate that also exhibits high performance. Barium titanate (BT) ceramic was one of the most widely used ferroelectric materials prior to the discovery of high-performance PZT. Efforts have been devoted to improving the piezoelectric properties of BT ceramic and several effective approaches are explored. The addition of some dopants into BT to replacing the A- and/or B-sites is an effective means of enhancing the ceramic’s piezoelectric properties [23–25]. In addition, the use of special fine powders and unusual sintering techniques such as spark plasma sintering (416 pC/N) [26], microwave sintering (370 pC/N) [27] and two-step sintering (460 pC/N) [28] have also resulted in a high d33. In such cases, the enhanced d33 of the BT ceramic is the result of extrinsic contributions to its polarizability, associated with submicron grain sizes and nanodomain structures. A BT ceramic with high piezoelectric properties (d33 > 300 pC/N), however, is difficult to obtain using ordinary raw materials and a conventional

85

sintering method, so, unlike PZT, BT is used primarily as a dielectric material rather than as a piezoelectric material. The reason why good piezoelectric properties cannot be obtained with ordinary BT ceramics remains a mystery. To date, the basic approach to the design and optimization of new ABO3-related piezoceramics has been to induce a phase transition between two ferroelectric phases, as exemplified by MPB and PPT. The question now arises as to whether or not there is a new high piezoelectric mechanism waiting to be found, and the answer may involve investigating piezoelectric ceramics with ABO3-type perovskite structures. Our recent work indicates that the addition of eucryptite (LiAlSiO4, LAS) can greatly enhance the piezoelectric properties of BT ceramics prepared from ordinary raw materials by conventional solid-state reaction sintering. We have found that this high piezoelectricity is dependent on a preferential distribution of Li+–Al3+ pairs in the ABO3-type lattice and is not related to the MPB and PPT effects, thus providing a new means of designing leadfree piezoelectric materials. Moreover, the high depolarization temperature of these materials has encouraged us to investigate xLAS/BT ceramic system for practical applications. 2. Experimental procedure xLAS/BT ceramics with LAS contents of 0, 4, 7.5 and 10 mol.% were prepared by conventional ceramic processing techniques. Commercial BT (99.9%, Aladdin Chemistry Co. Ltd, Shanghai, China) and LAS powders produced in our laboratory [29] were used as the starting materials. These compounds were weighed out in the desired ratios and ball-milled for 6 h by planetary milling with zirconia balls in alcohol. After ball-milling, the resulting slurries were dried at 80 °C for 12 h, ground and sieved. The mixed powders were subsequently pressed into pellets 10 mm in diameter and 1 mm thick using a few drops of 5 wt.% polyvinyl alcohol (PVA) as a binder. After burning off the PVA, the xLAS/BT pellets embedded in BT powder were sintered in covered alumina crucibles, heating at a rate of 5 °C min1 to a final temperature of 1350 °C, which was held for 2 h. The crystalline structure of each sample was analyzed using an Empyrean X-ray diffraction (XRD) system (PANalytical) while in situ XRD measurements were carried out on a Philips X’Pert diffractometer with Cu Ka radiation, operating at 40 kV and 40 mA. The morphology, grain size and local element occupancy were examined with a cold field emission scanning electron microscope (Quanta 200F) equipped for energy-dispersive spectroscopy (EDS). Platinum electrodes were evaporated onto the ceramic surfaces and annealed at 600 °C for 30 min to allow for electrical properties characterization. The temperature dependence of the dielectric constant of each sample was measured at 1 kHz across the temperature range of 25–180 °C on a Novocontrol CONCEPT40 broadband dielectric spectrometer. Polarization hysteresis and

86

D. Xu et al. / Acta Materialia 79 (2014) 84–92

strain–electric field curves were determined at different electric fields using a Radiant Technologies Precision workstation. The samples were poled at 80 °C in a silicone oil bath under a DC field of 3 kV mm1 for 30 min. The piezoelectric constant (d33) was measured employing a quasi-static piezoelectric constant testing meter (ZJ-4AN, Institute of Acoustics, Chinese Academy of Science). 3. Results and discussion 3.1. Phase and microstructure characterization Fig. 1a shows XRD patterns of xLAS/BT ceramic powders with different LAS contents measured at room temperature. The major diffraction peaks match well with the ABO3-type perovskite phases. However, small additional peaks indexed with an asterisk in the enlarged XRD patterns are observed in those ceramics with added LAS, as revealed in Fig. 1b, and their intensities increase with

increasing LAS content. The formation of secondary phases may be associated with the low solubility limit of LAS in the BT lattice. It is hard to determine the phase type for these secondary phases due to the relatively low intensities of their diffraction peaks. In order to further identify the impurity phases, surface scanning electron microscopy (SEM) micrographs of xLAS/BT ceramics with different LAS contents sintered at 1350 °C are displayed in Fig. 2. A remarkable effect of the LAS addition on the microstructure of the ceramics is obvious from Fig. 2b, c and d. Firstly, the densification of the ceramic samples is improved to some extent, with only a small number of surface pores. Secondly, the morphologies of the ceramic samples are modified by the introduction of LAS. For pure BT ceramic (Fig. 2a), inhomogeneous but mostly spherical grains are found; in contrast, a certain number of rod-shaped grains are observed in the grain boundaries, e.g. in the red circular regions in the SEM images (Fig. 2b, c and d). The exceptional rod-shaped grains are probably related to the impurity phases. Finally, the average grain size is seen to initially increase before reaching a maximum value of about 40 lm (in the case of the 4 mol.% LAS/BT ceramic), then decreases slightly with further increasing LAS content. The large grain sizes suggest that the enhanced d33 of the xLAS/BT ceramics is not the result of extrinsic contributions to the polarizability, which are typically associated with fine grains (1 lm) [30,31]. Fig. 3 shows the EDS analysis of the 7.5 mol.% LAS/BT sample. It is very hard to determine the respective Ba/Ti compositions due to their overlapping spectrum peaks. Hence, the composition table shown in the inset of Fig. 3 lists the sum of the Ba and Ti contents. The corresponding EDS analysis shown in Fig. 3a reveals a composition close to an Al:Si ratio of 1.69:1.95 for a normal spherical grain, confirming that the LAS with such a content had doped the BT lattice. The possible secondary phase was also analyzed by EDS, with the result shown in Fig. 3b. Compared with the matrix phase, the interfacial rod-shaped inclusion has higher signal intensities in Al and Si, especially in Al. This indicates that the secondary phases with an Al:Si ratio of 2 may be Al- and Li-enriched, though the Li content cannot be detected by EDS due to its low element number. It is clear that the presence of Al- and Li-enriched phases is caused by the excess of A-site ions after Al3+ and Li+ ions replacement for A-site Ba2+ ions. The identified impurities phases confirm a low solubility limit of LAS in the BT lattice, which is consistent with the XRD analysis above. 3.2. Lattice distortion and role of defects

Fig. 1. (a) XRD patterns of xLAS/BT ceramic powders measured at room temperature and (b) the enlarged views of XRD patterns in the selected 2h range showing the additional diffraction peaks on LAS doping.

As stated above, the LAS has doped the BT lattice and the precipitated phases are Al-enriched due to the low solubility limit of LAS in BT. To analyze the influence of LAS doping on the BT lattice distortion, fine scan XRD patterns obtained at room temperature were performed, and these are shown in Fig. 4a. The diffraction data confirms that

D. Xu et al. / Acta Materialia 79 (2014) 84–92

87

Fig. 2. Surface SEM micrographs of xLAS/BT ceramics: (a) x = 0 mol.%, (b) x = 4 mol.%, (c) x = 7.5 mol.% and (d) x = 10 mol.%. The red circular regions correspond to the secondary phases. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

the xLAS/BT powders exhibit P4 mm symmetry (JCPDS #83–1880), as evidenced by the relative intensity ratios of the 002/200 and 202/220 diffraction peaks and the existence of a single 111 peak at 2h  38.9°. The effect of the LAS content on the lattice parameters (a and c), as estimated from XRD full pattern fitting by HighScore Plus software, as well as the tetragonality (c/a) of the xLAS/BT ceramic powders, are shown in Fig. 4b. Both the lattice parameter c and the tetragonality are seen to decrease significantly with increasing LAS content in spite of the slight reduction in the lattice parameter a. The observed variations in lattice parameters are obviously caused by the incorporation of LAS into the BT lattice. Taking ionic valences into consideration, it is believed that Li+ and Al3+ ions, both of which have relatively small ionic radii, substitute the A-sites occupied by the Ba2+ ions of the ABO3-type perovskite structure, while Si4+ ions replace the B-sites occupied by Ti4+ [32]. In such cases, the average A-site valence will be 2 + and hence the local charge neutrality is maintained. To thoroughly understand and identify the occupation behavior of Li+ and Al3+ ions, the relative intensity ratio of the 002/200 peaks (hereafter referred to as I002/I200) was analyzed as a function of LAS content. Fig. 4c shows that the (I002/I200)x/(I002/ I200)0 value (where the superscripts 0 and x denote pure BT and LAS-doped BT ceramic powders, respectively) tends to decrease with increasing LAS content. It should be pointed out that the probability of preferred orienta-

tion- or internal stress-induced evolution can be ruled out completely, since the XRD experiments were carried out on the powders of xLAS/BT ceramics. Moreover, the observed variations in lattice parameters and I002/I200 with LAS content cannot be explained solely by the random distribution of Li+ and Al3+ ions. Assuming that Li+–Al3+ ions randomly locate at A-sites, the same trend in the variation of the diffraction intensity of the 002 and 200 peaks would be evident, but reductions in the I002/I200 intensity ratio with increasing LAS content would not occur since the same effect would apply to the structure factor (F) associated with the 002 and 200 diffractions. It is therefore necessary to provide an alternative explanation for our experimental results. Since the average of the ionic radii of Li+ and Al3+ is less than the radius of Ba2+, the substitution of either Li+ or Al3+ for Ba2+ will induce lattice distortions in the BT, reducing the XRD intensity [33]. In the case of the (0 0 2), (0 2 0) and (2 0 0) planes, the effective structure faceff eff tors, F eff 002 , F 020 and F 200 ; adjusted for lattice distortions, can be expressed as follows: F eff 200 ¼ F 200 expð

8p2 sin2 h 2 Dx Þ ¼ F 200 expðM 1 Þ k2

ð1Þ

F eff 020 ¼ F 020 expð

8p2 sin2 h 2 Dy Þ ¼ F 020 expðM 2 Þ k2

ð2Þ

F eff 002 ¼ F 002 expð

8p2 sin2 h 2 Dz Þ ¼ F 002 expðM 3 Þ k2

ð3Þ

88

D. Xu et al. / Acta Materialia 79 (2014) 84–92

Fig. 3. EDS analysis for (a) a normal spherical grain and (b) the interfacial rod-shaped inclusion in the 7.5 mol.% LAS/BT ceramic.

where F 200 , F 020 and F 002 are the average structure factors of the distortion-free crystal lattices, and Dx2 , Dy 2 and Dz2 , which represent deviations from the ideal BT lattice sites, are the mean square displacements in the [1 0 0], [0 1 0] and [0 0 1] directions, respectively. Since the diffraction intensity (I) is proportional to the square of the modulus of the structure factor (|F|2), the following expression can be obtained: I i 1 expð2M i Þ

ð4Þ

where the subscript i corresponds to the (2 0 0), (0 2 0) and (0 0 2) planes. The 200 diffraction includes contributions from the (2 0 0) and (0 2 0) planes. From the above equations, it can be concluded that lattice distortion is an important factor influencing the I002/ I200 intensity ratio. Based on the results of Fig. 4c, it is evident that the values of both M1 and M2 are much smaller than M3, thus lattice distortion has little effect on the diffraction intensity of the 200 peak but reduces the intensity of the 002 peak with increasing LAS content, eventually producing the changes in the I002/I200 ratio shown in Fig. 4c. Based on the above analysis, it can be concluded that Li+ and Al3+ ions may form Li+–Al3+ pairs within the same cell in order to minimize electrostatic energy, as indicated in Fig. 4b and c. In the case of A-site replacement, the relatively large M3 value suggests that these

Li+–Al3+ pairs tend to adopt a preferential alignment such that most of them are parallel to the [0 0 1] direction. Fig. 4d presents a diagram of the lattice distortion induced by the preferential distribution of Li+–Al3+ pairs parallel to the [0 0 1] direction. Here the lattice distortion induced by doped Li+–Al3+ pairs with a small average ionic radius occurs primarily along the [0 0 1] direction. As can be seen in Fig. 4b, the (0 0 2) interplanar spacing (d002) is obviously decreased because of the lattice distortion induced by the preferential substitution of Li+–Al3+ pairs parallel to the [0 0 1] direction. A decrease in the (2 0 0) and (0 2 0) interplanar spacings (referred as d200 and d020, respectively), however, does not occur since the lattice distortion has minimal influence on these spacings. The value of d002 thus decreases more rapidly than the value of d200 with increasing LAS content. In addition, the rate of reduction of the (I002/I200)x/(I002/I200)0 ratio tends to slow at an LAS content above x = 7.5 mol.% (as in Fig. 4c), demonstrating that the number of [1 0 0]- and [0 1 0]-oriented Li+–Al3+ pairs begins to increase as a result of elastic energy limitations. According to the chemical formula of LiAlSiO4, however, for such a replacement of Li+–Al3+ pairs for Ba2+ and Si4+ for Ti4+, if the excess Li and Al cannot be precipitated fully in the grain boundaries, there will be either an excess of A-site ions or a deficiency of B-site ions, accompanied by the formation of oxygen vacancies to keep charge neutrality. As a consequence, the occurrence of oxygen vacancies and Si4+ replacement for Ti4+, which has a larger radius, can reduce the elastic energy induced by the lattice distortion stemming from the preferential distribution of Li+–Al3+ pairs. In particular, the elastic energy can be decreased more effectively when the smaller Si4+ ions and oxygen vacancies are distributed in the vicinity of Li+–Al3+ pairs. Furthermore, the substitution of Ti4+ with the smaller Si4+ is very beneficial for enhancing the solubility of LAS in BT. As shown in Fig. 4d, the preferential distribution of Li+–Al3+ pairs along the [0 0 1] direction may be caused by the effect of the electric field formed from Li+–Al3+ pairs upon C–T phase transition. The detailed discussions are as follows. Firstly, the paraelectric–ferroelectric phase transition of BT ceramics with ABO3-type perovskite structure is a first-order transition, which includes the nucleation and growth of the tetragonal ferroelectric phase. The c-axis orientation of the tetragonal ferroelectric phase is determined by the nucleus orientation. Secondly, the nucleation process can be adjusted by changing the electric field, and it is energetically favorable for the spontaneous polarization (PS) of ferroelectric nuclei parallel to the electric field. Finally, Li+–Al3+ pairs can create an electric dipole moment PD, the local electric fields (ED) formed by the dipole at the head and tail of the pairs are nearly parallel to PD, and the ED in the middle and at the sides around the dipole are nearly antiparallel to PD. The former ED can lead to a ferroelectric phase nucleus with PS nearly parallel to PD, whereas the latter ED can induce a

D. Xu et al. / Acta Materialia 79 (2014) 84–92

89

Fig. 4. (a) Fine-scan XRD patterns in the 2h ranges of 38.3–39.3o, 43.7–46.3 o and 65.0–66.7o, (b) lattice constants (a and c) and tetragonality (c/a) as a function of LAS content x, (c) (I(002)/I(200))x/(I(002)/I(200))0 intensity ratio and (d) schematic illustration of the preferential distribution of Li+–Al3+ pairs parallel to the [0 0 1] direction. The error bars in (b) are a few tenths of the whole.

ferroelectric phase nucleus with PS nearly antiparallel to PD. Therefore, the nucleation of the tetragonal phase in the vicinity of Li+–Al3+ dipole causes the coexistence of PS nearly parallel (PS//PD) and nearly antiparallel to PD (-PS//PD) in the xLAS/BT ceramics, which is in good agreement with the observed symmetric hysteresis loops (see the next section and Fig. 5b). 3.3. Electrical properties The temperature dependence of the dielectric constant (e), as measured at 1 kHz for xLAS/BT ceramics with compositions of x = 0, 4, 7.5 and 10 mol.%, are shown in Fig. 5a. The dielectric curves of these ceramics only exhibit one dielectric anomaly, characterized by a phase transition from the ferroelectric phase (T) to paraelectric phase (C). The O–T phase transition cannot be observed in the dielectric–temperature curves within the measuring temperature. The dielectric constant (em) at TC first increases, reaching a maximum value of 6245 at x = 7.5 mol.%, then decreases with increasing LAS content. In addition, the LAS doping has little impact on TC, as evident in Fig. 5a. Fig. 5b shows the room-temperature polarization hysteresis loops of xLAS/BT ceramics with compositions of x = 0, 4, 7.5 and 10 mol.%. A well-saturated ferroelectric loop is observed for pure BT ceramic under an electric field of 40 kV cm1. After adding LAS, good ferroelectric loops

of the resulting xLAS/BT ceramics, which do not exhibit any obvious signs of leakage, are observed under a higher electric field of 50 kV cm1. The offset behavior in the coercive field reflecting the level of internal bias, which generally occurs in acceptor-doped ferroelectrics [16,34–36], is not observed in the polarization vs. electric field curves of xLAS/BT ceramics measured at room temperature. This can be attributed to the coexistence of PS//PD and -PS// PD in the xLAS/BT ceramics, and PD is distributed randomly in space. Consequently, the total net electric field from PD is zero, giving rise to the relatively symmetric and no internal-bias P–E loops. The situation is different from that in the aged acceptor-doped ferroelectrics with a severely constricted hysteresis loop and a notable bias field, where there is a defect dipole moment PD oriented along the PS (along the crystallographic c-axis), as noted in the previous reports [34–36]. The polarization hysteresis loop of the pure BT ceramic exhibits typical ferroelectric behavior, with a remnant polarization (2Pr) of 25 lC cm2, while the xLAS/BT ceramics show a relatively low 2Pr value of 17 lC cm2 at x = 4 and x = 7.5 mol.% (Fig. 5c). As the LAS content increases to 10 mol.%, the hysteresis loop becomes more slanted in comparison with those obtained from the 4 and 7.5 mol.% samples, accompanied by a further decrease in 2Pr to approximately 13 lC cm2 at x = 10 mol.% (Fig. 5c). The reduced 2Pr values can be accounted for

90

D. Xu et al. / Acta Materialia 79 (2014) 84–92

Fig. 5. (a) Temperature dependence of the dielectric constant at 1 kHz for xLAS/BT ceramics, in which the highest permittivity peak appears for the composition of x = 7.5 mol.%, (b) hysteresis loops of xLAS/BT ceramics measured at room temperature and (c) 2Pr values with different LAS contents.

by the decreased tetragonality (c/a) (Fig. 4b) [37,38] and Li+–Al3+ pairs oriented antiparallel to c-axis because the orientation of these dipole moments cannot be changed by an external electric field. Fig. 6a presents unipolar strain (S)–electric field (E) curves obtained for poled xLAS/BT ceramics with various LAS contents at room temperature. In the case of the xLAS/BT ceramics, a curve close to saturation at E P 3.5 kV mm1 and characteristic nonlinearity of unipolar strain are observed, compared with the results obtained for the pure BT ceramic. The maximum converse piezoelectric coefficient (dS/dE), calculated from the slope of the unipolar S–E plot, achieves an optimum value of 540 pm V1 at x = 7.5 mol.%, comparable to the reported high d33 values of the BT ceramics prepared by microwave sintering [27] and two-step sintering [28]. For comparison purposes, the maximum converse piezoelectric coefficient and the piezoelectric constant measured using a quasi-static d33 meter are shown together in Fig. 6b. When increasing the LAS content, the d33 initially increases, reaches its highest value of 378 pC/N at the composition of x = 7.5 mol.%, then declines at x = 10 mol.%. The d33 value (378 pC/N) associated with the 7.5 mol.% LAS/BT is very difficult to achieve for BT-based ceramics prepared from ordinary raw materials by a conventional sintering method, and is much higher than the values exhibited by existing lead-free piezoelectrics with TC P 120 °C, which primarily show d33 values of 200 pC/N

[16,17]. In addition, the high d33 value of the 7.5 mol.% LAS/BT is on a par with that of 300 pC/N reported for a non-textured KNN-based ceramic [21]. This is indeed an exciting result, but the question still remains as to why these xLAS/BT ceramics have such high piezoelectric constants compared with the pure BT ceramic. To clarify the origin of the high piezoelectricity in BT ceramics doped with Li+–Al3+ pairs, a possible piezoelectric mechanism is proposed here. The Li+ and Al3+ ions, substituting Ba2+ sites, locate at the interstices formed from four immediately neighbouring BO26 octahedras. The replacements of Li+ and Al3+ with small radii for the large Ba2+ will lead to two important consequences. On the one hand, a large lattice distortion can be generated in the vicinity of Li+–Al3+ pairs, as revealed in Fig. 4d, and this distortion can induce a deviation of spontaneous polarization direction from the c-axis near Li+–Al3+ pairs. Most of all, the distortion connected to the Li+–Al3+ dipoles locally creates unit cells of less than tetragonal symmetry, and these low-symmetry cells can be responsible for the high piezoelectric response, in analogy to the local monoclinic phase at the MPB in PZT [39]. On the other hand, because their radii are smaller than that of Ba2+, Li+ and Al3+ ions, with their relatively high mobility, can respond immediately to the external fields, which contributes to the piezoelectric properties of xLAS/BT ceramics. The piezoelectric mechanism proposed in the present work is different from any of the effects previously noted

D. Xu et al. / Acta Materialia 79 (2014) 84–92

Fig. 6. (a) Unipolar strain–electric field curves measured on poled xLAS/ BT ceramics, and (b) piezoelectric constant (d33) and maximum converse piezoelectric coefficient (dS/dE) with different LAS contents.

in the PZT and BZT–BCT systems exhibiting high piezoelectricity. The piezoelectric mechanism of the xLAS/BT ceramics is different from that of the BZT–BCT system, as reported by Liu and Ren [18], the decreased energy barrier is realized by a TCP in the BZT–BCT system and only those ceramics with compositions near the MPB show large piezoelectric coefficients. This mechanism is also different from the high piezoelectricity of the PZT system associated with the MPB effect. Our work indicates that, if a suitable preferential distribution of ion pairs is designed (such as occurred with Li+ and Al3+ in our case), lead-free ceramic systems prepared from ordinary raw materials by a conventional sintering method have a high probability of exhibiting excellent piezoelectric properties.

91

the temperature dependence of d33 measured at room temperature after annealing the poled samples at various temperatures for 30 min. It is evident that, while the d33 value of the pure BT ceramic begins to decrease remarkably above 50 °C, this is not the case for the xLAS/BT ceramics. As the temperature increases, the d33 values of these ceramics remain almost unchanged and then drop dramatically above a critical temperature (Td) of approximately 120 °C, which is very close to the Curie temperature of these ceramics. Thus the piezoelectric temperature stability of the BT ceramic can be greatly improved by adding a small amount of LAS. More importantly, the large temperature dependence and considerable degradation typically exhibited by KNN-based ceramics are not observed in our xLAS/BT ceramics, as shown in Fig. 7. Their hightemperature stability demonstrates that xLAS/BT ceramics may be used at temperatures as high as 120 °C, a range which is very hard for pure BT ceramics to reach. Recent studies on lead-free piezoelectric ceramics have primarily involved three ABO3-type perovskite systems: KNN-based, NBT-based and BT-based. In the case of KNN-based systems, although non-textured KNN ceramics have a relatively high d33 of 200–300 pC/N, their most significant disadvantage is the large temperature dependence of their piezoelectric properties, such that degradation occurs through thermal cycling between the two distinct ferroelectric domain states. In the case of NBTbased systems, the inherently low d33 of 200 pC/N is only halfway to the desired piezoelectric value, even though they have a moderate depolarization temperature in the range of 100–200 °C [16]. In the case of BT-based systems, much effort is still required to enhance the low TC so as to improve their overall usefulness. In fact, in most cases, high values of d33 and Td appear to be mutually exclusive when dealing with lead-free piezoelectric ceramics such as KNN-, NBT- and BT-based systems. From the above analysis, the current piezoelectric properties of xLAS/BT ceramics are

3.4. Piezoelectric thermal stability With regard to the development of piezoelectric ceramics, one of the most serious challenges is to improve their stability and reliability. To allow piezoelectric ceramics to function properly under high-temperature conditions, minimal piezoelectric temperature dependence is vitally important if these devices are to have practical applications. Fig. 7 shows the temperature dependence of the normalized d33 values of poled xLAS/BT ceramics measured ex situ. The depolarization temperature (Td) was determined from

Fig. 7. Temperature dependence of the normalized d33 values of the poled xLAS/BT ceramics measured ex situ. Each datum point in Fig. 7 is the average value of several tests.

92

D. Xu et al. / Acta Materialia 79 (2014) 84–92

superior to those of non-textured KNN-based ceramics. Furthermore, the d33 value (378 pC/N) of the 7.5 mol.% LAS/BT ceramic is obviously higher than the values obtained with NBT-based ceramics with similar depolarization temperatures (Td  120 °C). These results indicate that there is a real possibility that the xLAS/BT system may represent a good lead-free piezoelectric ceramic material. 4. Conclusions xLAS/BT lead-free piezoelectric ceramics were prepared using ordinary raw materials and a conventional sintering method. The 7.5 mol.% LAS/BT ceramic exhibits a very high piezoelectric constant (d33) of 378 pC/N, a converse piezoelectric coefficient (dS/dE) of 540 pm V1 and a depolarization temperature (Td) of 120 °C. The high piezoelectricity of these LAS-doped BT ceramics primarily originates from the unit cells of less than tetragonal symmetry created by the local distortion connected to the preferential [0 0 1]-distributed Li+–Al3+ pairs. These properties indicate that this system represents a promising lead-free piezoelectric candidate material worthy of further study. Acknowledgements This work was financially supported by the National Nature Science Foundation of China (Grant No. 11272102). D.X. and W.L.L. contributed equally to this work. References [1] [2] [3] [4]

Noheda B, Gonzalo JA. Phys Rev B 2000;61:8687. Asada T, Koyama Y. Phys Rev B 2007;75:214111. Schhierholz R, Fuess H. Phys Rev B 2011;84:064122. Cordero F, Trequattrini F, Craciun F, Galassi C. Phys Rev B 2013;87:094108. [5] Babu JB. Appl Phys Lett 2007;90:102901. [6] Cernea M, Vasile BS, Capiani C, Ioncea A, Galassi C. J Eur Ceram Soc 2012;32:133. [7] Cernea M, Galassi C, Vasile BS, Capiani C, Berbecaru C, Pintilie I, et al. J Eur Ceram Soc 2012;32:2389.

[8] Yang HB, Zhou CR, Liu XY, Zhou Q, Chen GH, Li WZ, et al. J Eur Ceram Soc 2013;33:1177. [9] Zhou CR, Feteira A, Shan X, Yang HB, Zhou Q, Cheng J, et al. Appl Phys Lett 2012;101:032901. [10] Li YM, Chen W, Xu Q, Zhou J, Gu XY, Fang SQ. Mater Chem Phys 2005;943:28. [11] Dai YJ, Zhang SJ, Shrout TR, Zhang XW. J Am Ceram Soc 2010;93:1108. [12] Shieh J, Wu KC, Chen CS. Acta Mater 2007;55:3081. [13] Zhou CR, Liu XY, Li WZ, Yuan CL. Mater Chem Phys 2009;114:832. [14] Jiang MH, Liu XY, Chen GH. Scr Mater 2009;60:909. [15] Liu W, Xu J, Wang YZ, Xu H, Xi XQ, Yang JL. J Am Ceram Soc 2013;96:1827. [16] Shrout TR, Zhang SJ. J Electroceram 2007;19:111. [17] Takenaka T, Nagata H. J Eur Ceram Soc 2005;25:2693. [18] Liu WF, Ren XB. Phys Rev Lett 2009;103:257602. [19] Zuo RZ, Fu J, Lv DY, Liu Y. J Am Ceram Soc 2010;93:2783. [20] Shen ZY, Li YM, Jiang L, Li RR, Wang ZM, Hong Y, et al. J Mater Sci: Mater Electron 2011;22:1071. [21] Saito Y, Takao H, Tani T, Nonoyama T, Tajatiri K, Homma T, et al. Nature 2004;432:84. [22] Wang K, Li JF. Adv Funct Mater 2010;20:1924. [23] Ang C, Yu Z, Jing Z, Guo RY, Bhalla AS, Cross LE. Appl Phys Lett 2002;80:3424. [24] Sharma S, Kumar P, Palei P. Ceram Int 2012;38:5597. [25] Yang WG, Zhang BP, Ma N, Zhao L. J Eur Ceram Soc 2012;32:899. [26] Shen ZY, Li JF. J Ceram Soc Jpn 2010;118:940. [27] Takahashi H, Numamoto Y, Tani J, Tsurekawa S. Jpn J Appl Phys 2008;47:8468. [28] Moon SM, Wang XH, Cho NH. J Ceram Soc Jpn 2009;117:729. [29] Wang LD, Fei WD, Hu M, Jiang LS, Yao CK. Mater Lett 2002;53:20. [30] Zhao Z, Buscaglia V, Viviani M, Buscaglia MT, Mitoseriu L, Testino A, et al. Phys Rev B 2004;70:024107. [31] Ghosh D, Sakata A, Carter J, Thomas PA, Han H, Nino JC, et al. Adv Funct Mater 2013. http://dx.doi.org/10.1002/adfm.201301913. [32] Shannon RD. Acta Cryst 1976;A32:751. [33] Wang YM. X-ray Diffraction of noncrystal and crystal with defects. 1st ed. Beijing: Science Press; 1988 [in Chinese]. [34] Ren XB. Nature Mater 2004;3:91. [35] Zhang LX, Erdem E, Ren XB, Eichel R-A. Appl Phys Lett 2008;93:202901. [36] Robels U, Arlt G. J Appl Phys 1993;73:3454. [37] Ullah A, Ahn CW, Hussain A, Lee SY, Kim IW. J Am Ceram Soc 2011;94:3915. [38] Fan LL, Chen J, Li S, Kang HJ, Liu LJ, Fang L, et al. Appl Phys Lett 2013;102:022905. [39] Guo R, Cross LE, Park SE, Noheda B, Cox DE, Shirane G. Phys Rev Lett 2000;84:5423.