Composition and Temperature Dependence of Structure and Piezoelectricity in (1−x)(K1−yNay)NbO3-x(Bi1/2Na1/2)ZrO3Lead-Free Ceramics

Lead-free piezoceramics with the composition (1 – x)(K 1 – y Na y ) NbO 3 -x(Bi 1/2 Na 1/2 )ZrO 3 (KNyN-xBNZ) were prepared using a conventional solid-state route. X-ray diﬀraction, Raman spectroscopy, and dielectric measurements as a function of temperature indicated the coexistence of rhombohedral (R) and tetragonal (T) phase, typical of a morphotropic phase boundary (MPB) as the BNZ concentration increased and by adjusting the K/Na ratio. High remnant polarization ( P r = 24 l C/cm 2 ), piezoelectric coeﬃcient ( d 33 = 320 pC/N), eﬀective piezocoeﬃ- cient ( d " 33 = 420 pm/V), coupling coeﬃcient ( k p = 48%), and high strain ( S = 0.168%) were obtained at room temperature, but signiﬁcant deterioration of P r , d " 33 , and k p were observed by increasing from room temperature to 160 ° C (17.5 l C/cm 2 , 338 pm/V, and 32%, respectively) associated with a transition to a purely T phase. Despite these compositions showing pro-mise for room-temperature applications, the deterioration in properties as a function of increasing temperature poses chal-lenges for device design and remains to be resolved.

Recently, PZT-like morphotropic phase boundary (MPBs) (R-T) have been reported in KNN-based ceramics by integrating compositions which were previously used to optimize the O-T and R-O phase boundaries. [39][40][41][42][43][44][45][46][47][48][49][50] In 2011, Zuo et al. reported R-T phase coexistence in (Li,Ta,Sb) and BaZrO 3 multimodified KNN lead-free ceramics and obtained a high d 33 = 365 pC/N. [39][40][41] For similar compositions, Wu et al. obtained d 33 $ 425 pC/N. 42 But more recently, Sb together with (Bi,M)NO 3 (M=Na, K, Li, Ag, N=Zr, Hf, Sn) have been used to enhance d 33 and electromechanical strain (S) with d 33 and S improved to >400 pC/N and 0.46% at 3 kV/ cm, respectively. [43][44][45][46][47][48][49] Despite promising d 33 and S values there are still a number of concerns. For Sb doping, T C is reported to decrease alarmingly and de-poling may jeopardize some applications. More importantly, however, to our knowledge, the temperature dependence of the piezoelectric properties is rarely reported. It is likely that the supposed MPB in many KNN-based compositions is not temperature independent and therefore phase coexistence and high d 33 /S are not maintained during high-temperature applications or at high field cyclic loading in which temperature is likely to increase. Given the above concerns, this contribution presents the composition and temperature dependence of the piezoelectric properties (1Àx)(K 1Ày Na y )NbO 3 -x(Bi 1/2 Na 1/2 ) ZrO 3 (KNyN-xBNZ) lead-free ceramics in which LiSbO 3 dopants have been excluded to maintain a high T C .
The density of the sintered samples was measured by the Archimedes method. The relative density of all studied samples was higher than 95%. The phase structure of the sintered samples was studied using a Bruker D2 Phaser Xray powder diffraction (XRD) and in situ XRD performed for selected samples using a Siemens D5000 HTXRD in the temperature range 30°C-350°C. Ceramic morphology and microstructure was examined using an FEI Inspect F scanning electron microscope (SEM). Raman spectra were acquired from À150°C to 350°C using a Renishaw inVia Raman microscope. For electrical tests, sintered samples were electroded using fire-on silver paste, followed by the samples being poled in silicon oil at room temperature with an applied electric field of 30-40 kV/cm. Piezoelectric coefficient (d 33 ) was measured using a Piezotest PM300 d 33 meter. Polarization hysteresis and strain-electric field behavior were determined using an aixACCT TF 2000 ferroelectric tester at a frequency of 1 Hz from room temperature to 160°C. The displacement data were synchronously captured by a laser interferometer. The temperature-dependent dielectric permittivity from 25°C to 600°C was measured using an Agilent 4184A multifrequency precision LCR meter. The planar electromechanical coupling factor (k p ) was determined from the resonance and antiresonance frequencies, which were measured using an Agilent 4294A Impedance/Gain-phase analyzer according to IEEE standards on piezoelectricity. 51,52

III. Results and Discussion
(1) Compositional Evolution of Structure-Property Relations The room-temperature XRD patterns of KNyN-xBNZ in the 2h range of 20°-70°are shown in Fig. 1. All peaks   could be attributed to perovskite phases, indicating that a stable solid solution was formed in the studied range. Diffraction peaks shifted to a lower and higher diffraction angle with increasing BNZ and Na concentration, respectively, consistent with differences in relative ionic radius of the dopant/substituent ions with the matrix composition.
Splitting of (200) peak at~2h = 45°was used to determine the likely symmetry of compositions. In the expanded XRD patterns [ Fig. 1(a)], O phase was dominant in KN0.6N-xBNZ with x ≤ 0.02, but R, O and T phases coexisted for compositions of 0.03 ≤ x ≤ 0.04. As x increased further, ceramics with 0.045 ≤ x ≤ 0.05 showed mixed R and T     Journal of the American Ceramic Society-Wang et al.
phases. For x > 0.05, a single peak emerged, suggesting that R phase dominated. For compositions in which x was maintained at 0.045 (KNyN-0.045BNZ), the splitting of (200)/(002) peaks gradually disappeared as y increased, suggesting the coexistence of R and T for a wide variation in Na:K ratio, Fig. 1(b). To confirm the phase evolution in KNyN-xBNZ as a function of composition, simulations were performed, Fig. 1(c-e), which are consistent with experimental data and also the interpretation by Wang et al. [43][44][45][46][47][48][49][50] Rietveld refinement analysis of the MPB composition KN0.6N-0.045BNZ was also performed using the GSAS+EXPGUI package, 53,54 where a two phases refinement (R3c + P4 mm) was used, as shown in Fig. 2. A good agreement between the observed and calculated patterns was obtained with R p = 6.2%, R wp = 9.7% and chi-squared (v 2 ) = 1.88, indicating the coexistence of R (R3c, 32.5%) and T (P4 mm, 67.5%) phases in the MPB composition.
To further confirm phase evolution as a function of composition, room-temperature Raman spectra of KNyN-xBNZ ceramics were obtained as shown in Fig. 3. Among the full Raman active mode of KNN, m 1 , m 2 , and m 3 are stretching modes, m 5 and m 6 are bending modes of the NbO 6 octahedra. 55 The strongest peak around 600 cm À1 is related to the m 1 mode; whereas the left and right weaker peaks are assigned to the m 2 and m 3 mode, respectively, as indicated by the simulated data shown in the inset of Fig. 3(a). With increasing BNZ content, the m 1 and m 2 peaks were found to shift to lower wave number as shown in Fig. 3(a). On the other hand, the m 1 and m 2 peaks shift to higher wave number with an increase in Na content as shown in Fig. 3(b). The peak shift to a lower/higher frequency is due to an increase/ decrease in binding strength caused by the expansion/shrinkage of the distance between Nb 5+ and its coordinated oxygen, resulting in the corresponding variation in crystal cell volume. 55,56 However, the clear but gradual symmetry changes observed by XRD were not apparent in the Raman spectra, as evidenced by the absence of the appearance of new symmetry related modes. Raman, however, is not only sensitive to the macroscopic symmetry but is influenced strongly by local distortions to the lattice. It is postulated that within the KNyN-xBNZ system there are compositional regions dominated by "pseudosymmetry" in a manner discussed for La-doped BiFeO 3 compositions by Khesro et al. 57 and Bi 1/2 (K 1Àx Na x ) 1/2 TiO 3 by Levin and co-workers. 58,59 In these compositions, there are local perturbations in the average macroscopic symmetry more commonly associated with adjacent phases within the phase diagram, for example, short-range antipolar order in the paraelectric phase for Ladoped BiFeO 3 . We propose a similar model for KNyN-xBNZ based on the absence of new symmetry modes in Raman and the gradual changes in peak splitting in XRD but extensive transmission electron microscopy (TEM) work is required before this can be proved conclusively.
The SEM images for the surface of KNyN-xBNZ ceramics (not shown) revealed that the grain size of KN0.6N-xBNZ decreased with increasing BNZ content, from~5 lm for x = 0 and to~1 lm for x = 0.08 but the grain size of KNyN-0.045BNZ was insensitive to the K/Na ratio.
The high electric field bipolar polarization hysteresis and strain loops for KNyN-xBNZ as a function of BNZ and Na content are shown in Fig. 4, from which the remnant polarization (P r ), coercive field (E C ) and average electric field induced strain (S) as a function of BNZ and Na content can be obtained, Fig. 5. The normalized strain coefficient d Ã 33 , representing the average strain per unit of electric field, is calculated by, d Ã 33 ¼ S max =E max , where E max is the maximum electric field value, and S max is the average value of the corresponding maximum strain. As shown in Fig. 5(a and b), with increasing BNZ and Na concentration in KNyN-xBNZ, P r increased and then decreased after reaching a peak value, at which point the lowest E C was obtained. The highest value of P r = 24 lC/cm 2 and lowest value of E C = 10.5 kV/cm was achieved for x = 0.045, y = 0.6 at the MPB. 60,61 The S and d Ã 33 values increased with increasing BNZ and Na content, reaching maximum values of 0.163%, 408 pm/V at x = 0.045, y = 0.6 and 0.168%, 420 pm/V at x = 0.045, y = 0.56, respectively, above which they reduced, Fig. 5(c  and d). d Ã 33 is dominated by extrinsic effects, mainly domain wall motion, which the coexistence of R and T phases encourages at the MPB. [61][62][63] The dielectric and piezoelectric properties of KNyN-xBNZ as a function of BNZ and Na content are shown in Fig. 6. The room-temperature permittivity (e r ) of KN0.6N-xBNZ increases with BNZ concentration, whereas it initially increases and then decreases as a function Na/K ratio, Fig. 6(a and b). tand initially decreased with BNZ and Na concentration but then became stable for a broad range of compositions, Fig. 6(a and b). d 33 and k p both increased significantly with increasing BNZ and Na concentration as shown in Fig. 6(c and d), reaching a maxima of 310 pC/N and 48% for x = 0.045, y = 0.06, and 320 pC/N and 50% for x = 0.045, y = 0.056, respectively, and then decreasing for BNZ > 0.05 and Na > 0.6. As with d Ã 33 , d 33 is optimized at the MPB largely through extrinsic contributions. The facile rotation of the polarization vector between states of similar free energy facilitated by phase coexistence, allows the domain walls to displace reversibly, resulting in a strong extrinsic contribution to the piezoelectric properties. 61,64 (2) Evolution of Structure-Property Relations as a Function of Temperature The temperature dependence of dielectric permittivity and loss for KNyN-xBNZ with different BNZ concentration and K/Na ratio is given in Fig. 7. It is well-known that KNN exhibits two dielectric peaks above room temperature, corresponding to the ferroelectric O-T phase transition at~176°C (T O-T ) and the T-cubic (C) phase transition at~381°C (Curie temperature, T C ). With increasing BNZ concentration, the maximum e r at T C (e max ) increased and T C decreased continuously, meanwhile, T O-T decreased and merged with the ferroelectric R-O phase transition (T R-O ), resulting in coexistence of R and T phases at room temperature, Fig. 7(a,c), which is consistent with the XRD results (Fig. 1) and previous reports. [43][44][45][46][47][48][49][50] However, T C remained at~310°C and T R-T increased with increasing Na concentration, suggesting the K/Na ratio had little effect on the ferroelectricparaelectric phase transition of KNyN-xBNZ and that its major influence is on lower temperature transitions. We note also that broad relaxor-like peaks are observed in all compositions if the BNZ concentrations exceeds, x = 0.08. Relaxor behavior, however, is not the focus of this study and we note only that KNyN-xBNZ follows trends typically observed in ferroelectric to relaxor compositional transitions with a transformation from a clearly defined macroscopic symmetry  Journal of the American Ceramic Society-Wang et al.
(x = 0) to pseudosymmetry (x > 0.08) that is associated with loss of long range ferroelectric order through chemical disorder that induces local competition/frustration between phases of similar free energy.
To clarify the structural evolution of KNyN-xBNZ as a function of temperature, in situ temperature dependence of Raman and XRD measurements for the selected MPB composition KN0.6N-0.045BNZ were performed, the results from which are plotted in Figs. 8 and 9. As shown in Fig. 8, the general trend is that the m 1 peak shifted to lower wave number and broadens, accompanied by a gradual fading of the m 2 peak. More specifically, the m 1 peak [ Fig. 8(c)] revealed a gradual shift in frequency (wave number) up to 70°C-100°C after which there was a sharp change in frequency, coincident with the broad peaks in e r at the same temperature range in Fig. 7.
In situ high-temperature XRD data for KN0.6N-0.045BNZ is shown in Fig. 9 which focused on evolution of splitting of the {200} peaks. Phase coexistence was apparent at room temperature as shown by broad multiple peaks that gradually disappeared to become a single peak at high temperatures in the cubic phase. To confirm the phase evolution as a function of temperature, simulations were performed, Fig. 9(c-k). As temperature increased, the coexistence of R-T phases remained up to 70°C-100°C, after which the phase structure total transformed to a T and then C phase above T C , consistent with experimental e r and Raman spectroscopy data as function of temperature.
The in situ temperature dependence of high electric field bipolar polarization hysteresis and strain loops for two MPB compositions (KN0.49N-0.045BNZ and KN0.6N-0.045BNZ) are shown in Fig. 10, from which the P r and d Ã 33 as a function of temperature were obtained, Fig. 11. As temperature increased, P r and d Ã 33 for both MPB compositions decreased linearly, coincident with a similar temperature dependence of k p , as shown in the inset of Fig. 11. The piezoelectric effect in ferroelectric ceramics is attributed to both intrinsic (lattice deformation) and extrinsic (domain wall motion) contributions. Normally, P r reflects intrinsic contribution, and d Ã 33 and k p contain both components. 40,48 However, detailed knowledge of the domain morphology is required to fully appreciate how the extrinsic contributions is likely to vary as a function of temperature and composition and extensive transmission electron or piezoforce microscopy is required. Nonetheless, the data presented in this contribution clearly illustrates that KNN-BNZ suffers from extreme temperature  dependence which may inhibit its usage in high drive applications and high-temperature applications despite the attractive values of k p and d 33 at room temperature.

IV. Conclusion
In this work, KNN based lead-free ceramics with the composition (1Àx)(K 1Ày Na y )NbO 3 -x(Bi 1/2 Na 1/2 )ZrO 3 were successfully prepared by a conventional solid-state route. With the increase in BNZ content, the phase structure of KNN gradually transformed from O to mixed R-T and then R phase, indicative of the construction of an R-T based MPB with 0.045 ≤ BNZ ≤ 0.05 and 0.4 ≤ Na ≤ 0.7. Optimum ferroelectric, piezoelectric, and strain properties were obtained for the MPB compositions with P r , d 33 , k p , S, and d Ã 33 of 24 lC/cm 2 , 320 pC/N, 48%, 0.168%, and 420 pm/V, respectively. However, despite these attractive room-temperature values at the MPB, there is a dramatic decrease in properties with temperature indicating the MPB is not temperature independent unlike in PZT. Hence there are concerns about the viability of these compositions for high-temperature or high field applications.