Abstract: The thermal stability and fatigue resistance of piezoelectric ceramics are of great importance for industrialized application. In this study, the electrical properties of (0.99–x)(K0.48Na0.52)(Nb0.975Sb0.025)O3– 0.01CaZrO3–x(Bi0.5Na0.5)HfO3 ceramics are investigated. When x = 0.03, the ceramics exhibit the optimal electrical properties at room temperature and high Curie temperature (TC = 253 ℃). In addition, the ceramic has outstanding thermal stability (d*33 ≈ 301 pm/V at 160 ℃) and fatigue resistance (variation of Pr and d*33 ~10% after 104 electrical cycles). Subsequently, the defect configuration and crystal structure of the ceramics are studied by X-ray diffraction, temperature– dielectric property curves and impedance analysis. On one hand, the doping (Bi0.5Na0.5)HfO3 makes the dielectric constant peaks flatten. On the other hand, the defect concentration and migration are obviously depressed in the doped ceramics. Both of them can enhance the piezoelectrical properties and improve the temperature and cycling reliabilities. The present study reveals that the good piezoelectric properties can be obtained in 0.96(K0.48Na0.52)(Nb0.975Sb0.025)O3–0.01CaZrO3–0.03(Bi0.5Na0.5)
Keywords: (Na,K)NbO3 (KNN)-based ceramics; thermal stability; fatigue resistance; crystal structure; defect structure
Piezoelectric materials, due to the ability of direct conversion between mechanical and electrical energy, have been widely used as fuel injection, ultrasonic motors, printing machines, actuators, and high-accuracy positioning systems . For the past decades, lead zirconate titanate (PZT) family has dominated global piezoceramic market due to their excellent electrical performance and thermal stability [2,3]. Unfortunately, PZT-based ceramics contain more than 60 wt% lead element, which have an adverse effect on human health and environment [4,5]. To circumvent above issues, exploration of lead-free piezoelectric system is imperative. Over the last few decades, potassium sodium niobate (Na,K)NbO3 (KNN) lead-free piezoelectric ceramics, as alternatives for lead-based piezoelectric ceramics, have attracted considerable attention due to their high Curie temperature (TC ≈ 420 ℃) and relatively good piezoelectric performance [6,7].
Although KNN-based materials are potential to be used in audio equipment, industrial electronics, as well as various pulsed-power applications , it should be noted that enhancing their piezoelectric properties is still the most important task now . Fortunately, in recent years, series of prominent achievements have been reported. In particular, Xu et al.  successfully fabricated non-textured KNN-based lead-free ceramics with ultrahigh piezoelectric property (d33 ≈ 570 pC/N) by construction of rhombohedral–tetragonal (R–T) phase boundary at room temperature. Besides, the superior piezoelectric properties with d33 ≈ 550 pC/N were achieved in KNN-based ceramics by adjusting phase structure and domain configurations [11,12]. However, inferior thermal stability and fatigue resistance of KNN-based ceramics tremendously restrain their industrialized application [13–15]. Some researchers widen the working temperature range of KNN-based ceramics by constructing polymorphic phase transition (PPB) [14,16–18]. Meanwhile, acceptor dopants, such as Mn and La, are able to effectively improve the thermal stability [16,19,20]. Two common strategies were applied to improve the cycling reliability until now. Firstly, PPB temperature of the piezoelectric ceramics is shifted far below/above room temperature by doping, which has been achieved in PZT  and KNN [22,23] systems. However, it inevitably results in a tremendous decrease in the piezoelectric properties. The second case is to reduce defect concentration to prevent domain pinning .
In this study, good piezoelectric response, thermal stability, and cycling reliability are achieved in (0.99–x) (K0.48Na0.52)(Nb0.975Sb0.025)O3–0.01CaZrO3–x(Bi0.5Na0.5) HfO3 piezoelectric ceramics by construction of PPB region and regulation of defect configuration. The excellent piezoelectric response (d33 = 352 pC/N, d*33 = 379 pm/V) and high TC of 253 ℃ are obtained for the piezoelectric ceramics with x = 0.03. In addition, x = 0.03 ceramic exhibits excellent thermal stability under external electric field of 3 kV/mm. Furthermore, it is demonstrated that the ferroelectric performance (remnant polarization (Pr) and normalized strain (d*33)) of x = 0.03 ceramic undergoes a slight degradation within 10% after 104 cycles under driven electric field of 4 kV/mm. This study reveals that 0.96(K0.48Na0.52) (Nb0.975Sb0.025)O3–0.01CaZrO3–0.03(Bi0.5Na0.5)HfO3
ceramics are quite promising for practical applications.
2 wt% Mn modified (0.99–x)(K0.48Na0.52)(Nb0.975Sb0.025) O3–0.01CaZrO3–x(Bi0.5Na0.5)HfO3 (x = 0, 0.01, 0.02, 0.03, 0.04, 0.05) (abbreviated as (0.99–x)KNNS–CZ– xBNH) ceramics were prepared using the conventional solid-state reaction method. All the raw powders of Nb2O5 (99.5%), Na2CO3 (99.8%), K2CO3 (99%), Sb2O3 (99.99%), CaCO3 (99%), ZrO2 (99%), Bi2O3(99.8%), HfO2 (99.9%), and MnO (99.9%) were purchased from Sinopharm Chemical Reagent Beijing Co., Ltd., China. All the raw materials were weighed according to nominal stoichiometric composition and then homogenized in a planetary mill for 24 h using ethyl alcohol as medium. The mixed powders were calcined twice at 850 ℃ for 4 h with intermediate milling to enhance the compositional homogenization. Subsequently, the powder mixtures were milled, dried, and sieved. The powders were then compacted into pellets with diameter of ~10 mm and thickness of ~1 mm by uniaxial pressing in a stainless-steel die using polyvinyl butyralas binder. The green pellets were sintered in air at 1040–1100 ℃ for 2 h. For electric measurement, the two main surfaces of the sintered disk samples were coated with silver paste and then heat-treated at 550 ℃ for 30 min. The disk samples were poled in silicon oil under a direct current electric field of 3 kV/mm at 70 ℃ for 30 min.
After mechanical polishing and thermally etching at 1040–1090 ℃ for 0.5 h, the microstructure of the sintered ceramics was checked by the scanning electron microscope (SEM; SU8010, Hitachi Company, Japan). Crystalline structure of the crushed sintered ceramics was determined by X-ray diffraction (XRD; D8 Advance, Bruker, Germany) with Cu Kα radiation. XRD results were analyzed using Rietveld refinements with general structure analysis system (GSAS) software. Quasistatic piezoelectric coefficient (d33) of the poled ceramics was measured using a quasistatic piezoelectric coefficient meter (ZJ-3A, Institute of Acoustics, Chinese Academy of Sciences, Beijing, China). Dielectric permittivity ( εT33 ) and dielectric loss (tanδ) at 1 kHz were measured using a capacitance meter (Agilent 4294A, Agilent, Santa Clara, USA). Planar electromechanical coupling factor (kp) was determined by an impedance analyzer (Agilent 4294A, Agilent, Santa Clara, USA). Insulation resistance (IR) under a fixed applied electric field of 1 kV/mm was measured by an insulation resistance tester (TH2683A, Tonghui, Jiangsu, China). Temperature-dependent dielectric properties at 1 kHz were measured using a capacitance meter (HP4278A, Hewlett-Packard, Santa Clara, USA) with an automated temperature controller. Ferroelectric polarization hysteresis (P–E) loops and electric-fieldinduced strain (S–E) with a frequency of 1 Hz were obtained by TF Analyzer 2000E ferroelectric measuring system (aixACCT Systems GmbH, Aachen, Germany). Impedance spectrum over frequencies from 40 Hz to 1 MHz under a root-mean-square voltage (Vrms) of 1 V AC signal was measured by dielectric impedance spectrometer (Precision impedance analyzer 4294A, Agilent, USA) equipped with a cryostat. The impedance spectra were fitted with an assumed equivalent circuit assisted by Z–View software (Version 3.0; Scribner Associates, Inc.) to evaluate the equivalent direct current (DC) resistance (R) and capacitance (C) component values.
3 Results and discussion
Phase structure of (0.99–x)KNNS–CZ–xBNH was determined by XRD and temperature dependence of dielectric constant ( εr –T) curves. Figure 1 shows XRD patterns of the crushed ceramics. All the ceramics exhibit typical ABO3-type perovskite structure without any impurity phases. The XRD results were analyzed using Rietveld refinements with GSAS software to confirm the phase structure and lattice parameters. The refinement parameters are summarized in Table 1. All the wRp values are less than 15%, demonstrating the reliability of refinement results. With the increase of x, the phase structure transforms obviously. When x is less than 0.02, the ceramics exhibit a pure orthorhombic structure. For x = 0.03 and x = 0.04 samples, the rhombohedral and tetragonal phases coexist in the ceramics. When x increases to 0.05, the ceramics show a pure tetragonal phase. In addition, the lattice parameters are summarized in Table 1. It can be found the lattice shrinks with the increase of x for the compositions with the same crystal structure, which should be ascribed the smaller ionic radius of Bi3+ (1.39 Å) with 12 coordination number (CN) than that of K+ (1.64 Å) with 12 CN . The shrunken unit cell also confirms the diffusion of the doping (Bi0.5Na0.5)HfO3 into KNN lattice. To clearly clarify the phase structure of (0.99–x)KNNS–CZ–xBNH ceramics, the εr –T curves from –110 to 200 ℃ measured at 100 kHz were performed as shown in Fig. 2. The anomalies in the curves correspond to the orthorhombic–tetragonal phase transition (TO–T), rhombohedral–orthorhombic phase transition (TR–O), rhombohedral–tetragonal phase transition (TR–T), and tetragonal–cubic transition (or TC). TO–T and TR–O of x = 0 ceramic are at about 180 and –60 ℃, respectively. Therefore, x = 0 ceramic exhibits orthorhombic phase structure at room temperature, in accordance with the XRD results in Fig. 1. TO–T decreases and TR–O increases when x increases. For the ceramics with 0.03 ≤ x ≤ 0.04, the TO–T and TR–O merge into a single peak TR–T, indicating these ceramics located at rhombohedral–tetragonal phase boundary at room temperature . The single peak TR–T of x = 0.05 ceramic is severely suppressed as shown in Fig. 2(f), demonstrating that x = 0.05 sample possesses a tetragonal phase structure [27,28].
Fig. 1 XRD patterns and Rietveld refinement of (0.99–x)KNNS–CZ–xBNH ceramics: (a) x = 0, (b) x = 0.01, (c) x = 0.02, (d) x = 0.03, (e) x = 0.04, and (f) x = 0.05.
Fig. 2 Temperature dependence of εr measured at 100 kHz of (0.99–x)KNNS–CZ–xBNH ceramics.
Table 1 Lattice parameters and refinement parameters of the KNN-based ceramics
Figures 3(a) and 3(b) display the temperature dependence of εr and tanδ from –100 to 400 ℃, respectively. It can be found that TC decreases rapidly with the increase of doping content, in coincidence with Refs. [27,29]. It is notable that the dielectric abnormal peaks become broader as the increase of doping content, which could enhance the thermal stability of piezoelectric properties and widen the working temperature range. According to Figs. 2 and 3(a), phase diagram is depicted in Fig. 3(c). The relatively high TC of 253 ℃ is obtained in x = 0.03 sample.
Fig. 3 Temperature dependence of (a) εr and (b) tanδ measured at 100 kHz of (0.99–x)KNNS–CZ–xBNH ceramics. (c) Phase diagram of (0.99–x)KNNS–CZ–xBNH ceramics.
After mechanical polishing and thermal etching at 1040–1090 ℃ for 0.5 h, the microstructure of (0.99–x)KNNS–CZ–xBNH ceramics is detected by SEM as shown in Fig. 4. More than 100 random grains from SEM images were counted to figure out the average grain sizes, which is summarized in Table 2. All the ceramics show a relatively homogeneous grain distribution. It is noted that the doping (Bi0.5Na0.5)HfO3 can immensely suppress the grain growth, in accordance with some previous reports [29,30]. To further analyze element distribution and confirm the existence of the elements involved, element mappings of x = 0.05 ceramic are carried out as displayed in Fig. 5. All the elements (e.g., K, Na, Nb, Sb, Ca, Zr, Bi, and Hf) are found in the ceramics. The element mapping data show that all the elements are homogenously distributed in the ceramics. In addition, the densities of the ceramics are measured by Archimedes method and the relative densities are calculated and summarized in Table 2. It can be found that relative density of the doping ceramics is higher than the undoped ceramics, which can improve the electrical properties.
Fig. 4 SEM surface images of (0.99–x)KNNS–CZ–xBNH ceramics: (a) x = 0, (b) x = 0.01, (c) x = 0.02, (d) x = 0.03, (e) x = 0.04, and (f) x = 0.05.
Fig. 5 Element mappings of the rectangular region in Fig. 4(f).
Table 2 Electrical properties and physical properties of (0.99–x)KNNS–CZ–xBNH ceramics
The defect configuration has a great influence on the structure and properties of the KNN-based ceramics . To explore the defect configuration of (0.99–x) KNNS–CZ–xBNH ceramics, impedance spectroscopy analysis is performed in frequency range from 40 Hz to 1 MHz at different temperatures. Nyquist plots of all samples at a series of temperatures (450–650 ℃) are exhibited in Fig. 6, where Z‘ and Z’‘ refer to the real and imaginary parts of the complex impedance (Z*), respectively. It is evident that only one semicircle is observed in the Z'–Z'' plots. Accordingly, an equivalent electrical circuit comprising one RC circuit (R is parallel to C) is used to fit the experimental impedance spectra . All the complex plane plots are not perfect semicircles, which are slightly depressed instead of being centered on the abscissa axis (known as Debye’s model). Therefore, a constant phase element, CPE, replaces the standard capacitance element in practice to provide a better fit to the experimental data . The capacitance value of CPE is calculated by Q and relaxation distribution parameter (n) with respect to the following equation :
C = ( R1-nQ)1/n (1)
Fig. 6 Impedance spectra at a series of temperatures for (0.99–x)KNNS–CZ–xBNH ceramics: (a) x = 0, (b) x = 0.01, (c) x = 0.02, (d) x = 0.03, (e) x = 0.04, and (f) x = 0.05.
The fitting R, Q, n, and C parameters are listed in Table 3. It is noted that the capacitance values are 10–10–10–9 F, which is consistent with the bulk ferroelectric ceramics . It demonstrates that the proposed equivalent electrical circuit and fitting process are reasonable in this study. The extracted direct current (DC) resistance can be used to evaluate the defect concentration. It can be found that the DC R of x = 0.03 ceramic is much smaller than the other sample, which implies that the defect concentration of x = 0.03 ceramic is relatively lower than the other samples. According to Rietveld refinements, the doping Bi and Hf elements would substitute into A-site and B-site, respectively. The reactions are expressed by the following equations:
Table 3 Fitting parameters obtained from impedance spectroscopy
Besides, the oxygen ions in the lattice ( O×O ) would turn into oxygen gas, accompanied by the formation of an ionized oxygen vacancy ( V••O ) and two electrons (e' ) when the perovskite-type piezoelectric ceramics are sintered at high temperature. The reaction can be given by
It has been verified that the dominant defect is oxygen vacancy in the KNN-based ceramics [36,37]. In the doped ceramics, the defect diploes of (2 V'A-V••O ) and ( 2Hf'Nb – V••O ) form, restraining the oxygen vacancies migration. Therefore, with the increase of doping (Bi0.5Na0.5)HfO3, the mobile oxygen vacancy concentration decreases for 0 < x < 0.03. With the further increase of doping content, the content of these defect diploe increases and reaches saturation point, leading to the separation of defect diploes. Hence, the oxygen vacancy concentration increases when x > 0.03.
In order to analyse the defect migration of (0.99–x) KNNS–CZ–xBNH ceramics, the activation energy for conduction is calculated by Arrhenius equation as follows:
where conductivity (σ) is calculated by DC R, σ0 is pre-exponential term, kB is Boltzmann constant, and T is measuring temperature. Ea is activation energy for conduction, which is the formation energy of the charge carriers and migration/hopping energy of charge carriers over a long distance [32,38,39]. Ea values are calculated and shown in Fig. 7. All the coefficient of determinations R2 are close to 1, indicating the linear fitting is reliable. It can be seen that Ea firstly increases and then decreases, which reaches the maximum for x = 0.03 ceramic. Therefore, the defect mobility becomes difficult in x = 0.03 ceramic. The low Ea of highly doped (Bi0.5Na0.5)HfO3 ceramics (x = 0.05 sample) should be ascribed to the high defect concentration in the ceramics. The low defect concentration and migration would enhance the electrical properties of x = 0.03 ceramic.
Fig. 7 Arrhenius plots of conductivity versus 1000/T for (0.99–x)KNNS–CZ–xBNH ceramics: (a) x = 0, (b) x = 0.01, (c) x = 0.02, (d) x = 0.03, (e) x = 0.04, and (f) x = 0.05.
The electrical properties of (0.99–x)KNNS–CZ–xBNH ceramics are summarized in Table 2. Dielectric constant ɛr increases monotonously with the increase of (Bi0.5Na0.5)HfO3 doping content, which is consistent with Ref. . Polarization–electric field (P–E) hysteresis loops are displayed in Figs. 8(a) and 8(b). It can be seen that all the ceramics exhibit the well-saturated P–E hysteresis loops, indicating the good ferroelectricity. It can be observed that the spontaneous polarization (Pr) reaches the highest value near the PPB region (0.03 ≤ x ≤ 0.04), which should be ascribed to the abundant polarization direction and low free energy [41,42]. In addition, composition dependence of d33 and ɛr × Pr are shown in Fig. 8(c). In ferroelectric materials, the piezoelectric coefficients were proportional to the product of polarization and permittivity. The longitudinal piezoelectric coefficient d33 can be expressed as follows :
Fig. 8 (a, b) P–E hysteresis loops of (0.99–x)KNNS–CZ–xBNH ceramics. (c) d33 and ɛr × Pr values as a function of x. (d, e) Unipolar S–E curves of (0.99–x)KNNS–CZ–xBNH ceramics. (f) (d*33 and Smax values as a function of x.
where ε33 and P3 are the relative permittivity and polarization along the polar axis, ε0= 8.854 × 10–12 F/m is the permittivity of the vacuum, and Q11 is the electrostrictive coefficient of the paraelectric phase, which typically varies between 0.05 and 0.1 m4/C2 for different materials. In this case, Q11 could be constant because of the low concentration of doped elements. For poled piezoceramics, Pr and ɛr were equal to P3 and ɛ33, respectively. Therefore, d33 is in proportion to ɛr × Pr. It can be found that d33 shows a similar tendency with ɛr × Pr, which both reach the highest values near the PPB region (0.03 ≤ x ≤ 0.04), Besides, unipolar electric field–induced strain (S–E) curves for all samples are shown in Figs. 8(d) and 8(e). The electric field-induced maximum strain (Smax) and d*33 (d*33 = Smax/Emax) is plotted in Fig. 8(f). It can be found that Smax and d*33 reach the maximum in x = 0.03 ceramics. The enhanced Pr and low defect concentration should be responsible for the high piezoelectric properties for x = 0.03 sample . As summarized in Table 2, the other piezoelectric properties, such as electro-mechanical coupling factor kp, exhibit the similar trend with d33. The IR reaches maximum and dielectric loss tanδ is lowest for x = 0.03 sample, in accordance with the DC R obtained from impedance analysis. It should be ascribed to the low defect concentration and mobility. The optimal electrical properties (d33 = 352 pC/N; d*33= 379 pm/V; kp = 40.9%; ɛr = 2016; tanδ = 0.019; IR = 4.82 × 1011 Ω·cm) are obtained in x = 0.03 ceramic.
Thermal and cycling reliabilities of (0.99–x)KNNS– CZ–xBNH ceramics are very important for industrialized application. Figures 9(a) and 9(b) exhibit the temperature dependence of unipolar S–E curves of x = 0 and x = 0.03 ceramics measured at 4kV/mm, respectively. d*33 of x = 0 and x = 0.03 ceramics are depicted in Fig. 9. It can be observed that d*33 reduces sharply from 20 (163 pm/V) to 160 ℃ (65 pm/V), indicating the thermal stability of x = 0 ceramic is inferior. The increasing unipolar Smax from 160 to 200 ℃ is attributed to the coexistence of orthorhombic and tetragonal as shown in Fig. 3(a). For x = 0.03 ceramic, d*33 reduces slowly from 20 to 120 ℃ and shows excellent thermal stability. When the measurement temperature reaches 120 and 160 ℃, the normalized d*33 can attain 95% and 79%, respectively. The excellent thermal stability of x = 0.03 ceramic should be attributed to the depressed rhombohedral and tetragonal phase boundary near room temperature [14,44,45]. Figures 10 and 11 show the evolution of P–E hysteresis loops and unipolar S–E curves of x = 0 and x = 0.03 ceramics with electrical cycling up to 104 cycles under a fixed driven electric field of 3 kV/mm. An obvious degradation of Pr and d*33 can be observed with the increase of cycle for x = 0 ceramic. The cycling results in the reduction in Pr and d*33 by about 20% after 104 cycles. Nevertheless, Pr and d*33 hardly degrade up to 103 cycles for x = 0.03 ceramic. After 104 cycles, only approximately 10% loss of Pr and d*33 occurs. The fatigue resistance of (Bi0.5Na0.5)HfO3 doped ceramics is much better than that of undoped ceramics, which should be ascribed to the low defect concentration [18,46].
Fig. 9 Temperature dependence of unipolar S–E curves at a fixed electric field of 4 kV/mm for (a) x = 0 and (b) x = 0.03 ceramics.
Fig. 10 P–E hysteresis loops measured after 100, 101, 102, 103, and 104 fatigue cycles of (a1–e1) x = 0 and (a2–e2) x = 0.03
ceramics. Pr as a function of fatigue cycle for (f1) x = 0 and (f2) x = 0.03 ceramics.
Fig. 11 Unipolar S–E curves measured after 100, 101, 102, 103, and 104 fatigue cycles of (a1–e1) x = 0 and (a2–e2) x = 0.03
ceramics. d*33 values as a function of fatigue cycle for (f1) x = 0 and (f2) x = 0.03 ceramics.
In summary, the piezoelectric response, thermal stability, and fatigue resistance of (0.99–x)KNNS–CZ–xBNH piezoelectric ceramics are systematically investigated in the study. x = 0 ceramic exhibits a pure orthorhombic structure at room temperature and high defect concentration, which results in the inferior electrical properties, thermal stability, and cycling reliability. The doping (Bi0.5Na0.5)HfO3 makes the dielectric constant peaks flatten, which can enhance the temperature and cycling reliability of (0.99–x)KNNS–CZ–xBNH ceramics. When x = 0.03, the ceramics exhibit the optimal electrical properties (d33 = 352 pC/N; d*33 = 379 pm/V; kp = 40.9%; ɛr = 2016; tanδ = 0.019; IR = 4.82 × 1011 Ω·cm) and high TC of 253 ℃. In addition, x = 0.03 ceramic exhibits excellent thermal stability (d*33 ≈ 301 pm/V at 160 ℃) and fatigue resistance (variation of Pr and d*33~10% after 104 electrical cycles). The study demonstrates that (0.99–x)KNNS–CZ–xBNH piezoelectric ceramics are expected to replace part of lead-based piezoelectric ceramics in the future.
 Rödel J, Jo W, Seifert KTP, et al. Perspective on the development of lead-free piezoceramics. J Am Ceram Soc2009, 92: 1153–1177.
 Jo W, Dittmer R, Acosta M, et al. Giant electric-fieldinduced strains in lead-free ceramics for actuator applications–status and perspective. J Electroceram 2012, 29: 71–93.
 Moure C, Peña O. Recent advances in perovskites: Processingand properties. Prog Solid State Chem 2015, 43: 123–148.
 Saito Y, Takao H, Tani T, et al. Lead-free piezoceramics. Nature 2004, 432: 84–87.
 Wei HG, Wang H, Xia YJ, et al. An overview of lead-free piezoelectric materials and devices. J Mater Chem C 2018, 6: 12446–12467.
 Li JF, Wang K, Zhu FY, et al. (K,Na)NbO3-based lead-free piezoceramics: Fundamental aspects, processing technologies, and remaining challenges. J Am Ceram Soc 2013, 96: 3677–3696.
 Peng XY, Zhang BP, Zhu LF, et al. Multi-phase structure and electrical properties of Bi0.5Li0.5ZrO3 doping K0.48Na0.56NbO3 lead-free piezoelectric ceramics. J Adv Ceram 2018, 7: 79–87.
 Prateek, Thakur VK, Gupta RK. Recent progress on ferroelectric polymer-based nanocomposites for high energy density capacitors: Synthesis, dielectric properties, and future aspects. Chem Rev 2016, 116: 4260–4317.
 Wang K, Li JF. (K,Na)NbO3-based lead-free piezoceramics: Phase transition, sintering and property enhancement. J Adv Ceram 2012, 1: 24–37.
 Xu K, Li J, Lv X, et al. Superior piezoelectric properties in potassium-sodium niobate lead-free ceramics. Adv Mater2016, 28: 8519–8523.
 Yao WZ, Zhang JL, Zhou CM, et al. Giant piezoelectricity, rhombohedral-orthorhombic-tetragonal phase coexistence and domain configurations of (K, Na)(Nb, Sb)O3–BiFeO3–(Bi, Na)ZrO3 ceramics. J Eur Ceram Soc 2020, 40: 1223–1231.
 Zhang JL, Sun X, Su WB, et al. Superior piezoelectricity and rhombohedral-orthorhombic-tetragonal phase coexistence of (1–x)(K,Na)(Nb,Sb)O3–x(Bi,Na)HfO3 ceramics. Scr Mater2020, 176: 108–111.
 Zhang SJ, Xia R, Shrout TR. Lead-free piezoelectric ceramics vs. PZT? J Electroceram 2007, 19: 251–257.
 Quan Y, Ren W, Niu G, et al. Large piezoelectric strain with superior thermal stability and excellent fatigue resistance of lead-free potassium sodium niobate-based grain orientation-controlled ceramics. ACS Appl Mater Interfaces 2018, 10: 10220–10226.
 Koruza J, Bell AJ, Frömling T, et al. Requirements for the transfer of lead-free piezoceramics into application. J Materiomics 2018, 4: 13–26.
 Fang MX, Rajput S, Dai ZH, et al. Understanding the mechanism of thermal-stable high-performance piezoelectricity. Acta Mater 2019, 169: 155–161.
 Hinterstein M, Hoelzel M, Rouquette J, et al. Interplay of strain mechanisms in morphotropic piezoceramics. Acta Mater 2015, 94: 319–327.
 Li P, Chen XQ, Wang FF, et al. Microscopic insight into electric fatigue resistance and thermally stable piezoelectric properties of (K,Na)NbO3-based ceramics. ACS Appl Mater Interfaces 2018, 10: 28772–28779.
 Lv X, Wu JG, Zhu JG, et al. Temperature stability and electrical properties in La-doped KNN-based ceramics. J Am Ceram Soc 2018, 101: 4084–4094.
 Zhang MH, Wang K, Du YJ, et al. High and temperatureinsensitive piezoelectric strain in alkali niobate lead-free perovskite. J Am Chem Soc 2017, 139: 3889–3895.
 Jiang QY, Subbarao EC, Cross LE. Effect of composition and temperature on electric fatigue of La-doped lead zirconate titanate ceramics. J Appl Phys 1994, 75: 7433–7443.
 Zhang SJ, Xia R, Hao H, et al. Mitigation of thermal and fatigue behavior in K0.5Na0.5NbO3-based lead free piezoc-eramics. Appl Phys Lett 2008, 92: 152904.
 Wang K, Yao FZ, Jo W, et al. Temperature-insensitive (K, Na)NbO3-based lead-free piezoactuator ceramics. Adv Funct Mater 2013, 23: 4079–4086.
 Yao FZ, Patterson EA, Wang K, et al. Enhanced bipolar fatigue resistance in CaZrO3-modified (K, Na)NbO3 leadfree piezoceramics. Appl Phys Lett 2014, 104: 242912.
 Shannon RD. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Cryst Sect A 1976, 32: 751–767.
 Tao H, Wu JG, Xiao DQ, et al. High strain in (K,Na)NbO3-based lead-free piezoceramics. ACS Appl Mater Interfaces2014, 6: 20358–20364.
 Yang WW, Li P, Wu SH, et al. Coexistence of excellent piezoelectric performance and thermal stability in KNNbased lead-free piezoelectric ceramics. Ceram Int 2020, 46: 1390–1395.
 Lv X, Li ZY, Wu JG, et al. Enhanced piezoelectric properties in potassium-sodium niobate-based ternary ceramics. Mater Des 2016, 109: 609–614.
 Tao H, Wu JG, Zheng T, et al. New (1– x)K0.45Na0.55Nb0.96 Sb0.04O3–xBi0.5Na0.5HfO3 lead-free ceramics: Phase boundary and their electrical properties. J Appl Phys 2015, 118: 044102.
 Tao H, Wu JG. Giant piezoelectric effect and high strain response in (1–x)(K0.45Na0.55)(Nb1Sb)O3–xBi0.5Na0.5Zr1–HfO3 lead-free ceramics. J Eur Ceram Soc 2016, 36: 1605–1612.
 Cui B, Werner P, Ma TP, et al. Direct imaging of structural changes induced by ionic liquid gating leading to engineered three-dimensional meso-structures. Nat Commun 2018, 9: 3055.
 Wang XZ, Huan Y, Wang ZX, et al. Electrical conduction and dielectric relaxation mechanisms in the KNN-based ceramics. J Appl Phys 2019, 126: 104101.
 Nobre MAL, Lanfredi S. Ferroelectric state analysis in grain boundary of Na0.85Li0.15NbO3 ceramic. J Appl Phys2003, 93: 5557–5562.
 Yoon SH, Randall CA, Hur KH. Effect of acceptor (Mg) concentration on the resistance degradation behavior in acceptor (Mg)-doped BaTiO3 Bulk ceramics: I. impedance analysis. J Am Ceram Soc 2009, 92: 1758–1765.
 Irvine JTS, Sinclair DC, West AR. Electroceramics: characterization by impedance spectroscopy. Adv Mater1990, 2: 132–138.
 Huan Y, Wang XH, Wei T, et al. Defect control for enhanced piezoelectric properties in SnO2 and ZrO2 co-modified KNN ceramics fired under reducing atmosphere. J Eur Ceram Soc 2017, 37: 2057–2065.
 Huan Y, Wang XH, Wei T, et al. Defect engineering of high-performance potassium sodium niobate piezoelectric ceramics sintered in reducing atmosphere. J Am Ceram Soc2017, 100: 2024–2033.
 Molak A, Ksepko E, Gruszka I, et al. Electric permittivity and conductivity of (Na0.5Pb0.5)(Mn0.5Nb0.5)O3 ceramics. Solid State Ionics 2005, 176: 1439–1447.
 Rafiq MA, Tkach A, Costa ME, et al. Defects and charge transport in Mn-doped K0.5Na0.5NbO3 ceramics. Phys Chem Chem Phys 2015, 17: 24403–24411.
 Lv X, Wu JG, Xiao DQ, et al. Electric field-induced phase transitions and composition-driven nanodomains in rhombohedral-tetragonal potassium-sodium niobate-based ceramics. Acta Mater 2017, 140: 79–86.
 Huan Y, Wang XH, Li LT. Displacement of Ta-O bonds near polymorphic phase transition in Li-, Ta-, and Sb-modified (K,Na)NbO3 ceramics. Appl Phys Lett 2014, 104: 242905.
 Huan Y, Wei T, Wang ZX, et al. Polarization switching and rotation in KNN-based lead-free piezoelectric ceramics near the polymorphic phase boundary. J Eur Ceram Soc2019, 39: 1002–1010.
 Wang K, Li JF. Domain engineering of lead-free Li-modified (K,Na)NbO3 polycrystals with highly enhanced piezoelectricity. Adv Funct Mater 2010, 20: 1924–1929.
 Zhang MH, Wang K, Zhou JS, et al. Thermally stable piezoelectric properties of (K,Na)NbO3-based lead-free perovskite with rhombohedral-tetragonal coexisting phase. Acta Mater 2017, 122: 344–351.
 Zhen YC, Cen ZY, Chen LL, et al. The effect of microstructure on piezoelectric properties and temperature stability for MnO doped KNN-based ceramics sintered in different atmospheres. J Alloys Compd 2018, 752: 206–212.
 Kim JH, Kim DH, Lee TH, et al. Large electrostrain in K(Nb1–xMnx)O3 lead-free piezoelectric ceramics. J Am Ceram Soc 2016, 99: 4031–4038.