Ultrahigh thermal stability and piezoelectricity of lead-free KNN-based texture piezoceramics

Formation of crystal orientation and phase boundary with MPB feature

The SEM images of the KNN-BNZ-xBKH ceramics are shown in Fig. S1. Notably, it can be seen that the textured ceramics (abbreviated as xT) exhibit typical brick-wall-like grains aligned parallel to the tape-casting plane, and the grain size is significantly larger than that of random ceramics (abbreviated as xR). The large grains of the textured ceramics (~20–30 μm) are close to the size of NN templates (Fig. S2), indicating that the NN seed templates act as a nucleation site during grain growth. In addition, as a representative component of electrical properties, the energy-dispersive spectroscopy (EDS) analysis also shows that the element distribution in 3T ceramics is uniform and there are no other impurity phases, as shown in Fig. S3. The X-ray diffraction (XRD) patterns of KNN-BNZ-xBKH ceramics for both random and textured ceramics are shown in Figs. S4a and S5. All samples exhibit typical perovskite structure and the (00l) diffraction peaks of the textured ceramics are significantly higher than those of the random ceramics, indicating strong crystal orientation along l>c. The Lotgering factor f_(00l) can be used to calculate the texturing degree of the sample, and the corresponding f_(00l) was calculated to be 98.7%, 98.3%, 98.7%, and 85.8% for the textured ceramics with x = 0.01 − 0.04, respectively³⁷. In addition, the electron backscattered diffraction (EBSD) pole diagram (Fig. 2e) and inverse pole diagram (Fig. 2f) of 3T ceramics were tested along the direction perpendicular to the casting direction, indicating the high texture degree of the ceramics. The high orientation degree is attributed to the high-quality NaNbO₃ templates and reasonably designed sintering process. The phase structure with varied BKH concentration can be determined by combining the (002)/(200) characteristic diffraction peaks, ε_r–T curves, and Raman spectrum. As shown in Fig. S4a, with the increase of x, the intensity of the (002) peak in xR ceramics is gradually exceeded by (200), and even almost disappears, indicating that a continuous phase transition occurs, e.g., a gradual increase of the T phase and a gradual decrease of the O phase. In addition, as shown in Fig. S4b, with the increase of x, although the O-T phase boundary cannot be clearly seen, it gradually shifted to low temperature, while the R-O phase boundary was significantly suppressed and disappeared at liquid nitrogen temperature. Therefore, combining the analyses of both XRD patterns and ε_r–T curves, we can deduce that the 0R ceramics belong to the single orthogonal phase, while the xR ceramics with 0.01 ≤ x ≤ 0.04 can be determined as the coexistence of O and T phases. Due to the presence of T_O-T at room temperature, the 3R ceramics have a comparable content of O and T phases, which greatly reduces the Gibbs free energy and contributes to the improvement of ferro/piezoelectrical properties. On the contrary, due to the T_O-T deviates from room temperature, 2R and 4R ceramics have more O and T phases, respectively. Hence, the room temperature ε_r of the xR ceramics first increases because it approached to the phase boundary (i.e., x = 0 − 0.03), but then decreases because of the reduced degree of phase boundary and the significantly destroyed long-range ordering at high contents of (Bi_0.5K_0.5)HfO₃ (i.e., x = 0.04)³⁸.

It is noteworthy that the phase transition law of the xT ceramics is similar to that of the xR ceramics (Fig. S5), and the T_O-T of 3T is also located at room temperature (Fig. 2c, Figs. S6 and S7). However, the O-T phase boundary of xT ceramics almost disappears, in particular x = 0.03, the phase structures exhibit similarities to the MPB structure in PZT (Fig. 2c). Thus, unlike other KNN-based ceramics, the corresponding phase diagrams represented by xT ceramics can be drafted as shown in Fig. 2d. KNN-based ceramics with MPB feature will correspondingly exhibit excellent temperature stability of ferro/piezoelectrical properties. To further illustrate the newly discovered phase boundary of xT ceramics, the in-situ variable temperature XRD patterns are shown in Fig. 2a, b. From the measurements, not only the high Curie temperature of 3T ceramics can be observed (~330 °C), but the phase structure (coexistence of O and T phases) is also stable from room temperature to ~200 °C, revealing the phase boundary in 3T ceramics are similar to MPB, which is consistent with the ε_r–T curves. In addition, we also conducted XRD Rietveld refinement as shown in Fig. S8. The results indicate that there is a coexisting O-T phase structure in the temperature range of 25–200 °C, and the changes are very subtle with increasing temperature. At 25 °C, 100 °C, and 200 °C, the proportion of O phase is 68.378%, 65.206%, and 62.093%, respectively. Finally, the phase structure transforms to a cubic phase and a small amount of T phase at 350 °C. Overall, the multiphase coexistence phase boundary with MPB feature and high l> crystallographic orientation will play an important role in obtaining excellent comprehensive piezoelectricity for the 3T ceramics.

To further resolve the local structural information of the textured KNN-BNZ-BKH ceramics, aberration-corrected atomic-resolution scanning transmission electron microscopy (STEM) was employed. Fig. 3e gives a STEM annular Bright-Field (ABF) image along the [100] zone axis. For ferroelectric KNN-based materials, the polarization vectors are determined by the displacement from B-site cations (stronger intensity contrast, Nb) to the center of the four nearest neighboring A-site cations (weaker intensity contrast, Na/K) based on 2D Gaussian peak fitting. It is clearly shown that the δ_Nb−Na/K vectors are not homogeneous (as expected in normal ferroelectrics with a single phase); instead, they are heterogeneous, i.e., mainly lying along the pseudocubic axes in the left part and along the diagonals in the right part, corresponding to the T and O symmetries, respectively, as schematically shown in the inset of Fig. 3e. After peak finding with the method of 2D Gaussian peak fitting, with an accuracy ~5 pm, the δ_Nb-O displacement vector map can be obtained, as shown in Fig. 3f, g. Based on the schematic figure (the inset), T and O nanoregions can be identified, which is consistent with XRD analysis results. The dominance of large-scale long-range ordered polarization regions consisting of O and T phases can be clearly observed by combining the arrows in the zone axis along [100] zone axis, while short-range disordered multiphase nanoclusters occupy less, which is quite different from the local structure of KNN-based ceramics with R-O-T multiphase coexistence^9,17. Fig. 3n is a statistical chart of the polarization direction in quadrants. The results indicate that the O phase is the majority phase. Fig. 3f, g clearly shows the gradual polarization rotation between different states. Such nanoscale balanced multiphase coexistence may possess almost isotropic free energy and thus significantly decreased polarization anisotropy.

We also performed the calculation of the key length of both A and B sites, as shown in Fig. 3h, o, the reddish lines indicate longer bond lengths, the bluish lines indicate shorter bond lengths, and the green lines represent the average normal bond lengths. We utilized the Z-contrast feature of STEM to draw the A site intensity map, as shown in Fig. 3i, p, the reddish atoms represent heavier elements, while the bluish atoms indicate lighter elements. Fig. 3k–m present the local images of the O phase, T phase, and the boundary of the O-T phase, respectively, extracted from Fig. 3h–j. It can be observed that the O regions exhibit less heavy dopants (Bi) in A sites and weaker lattice distortion, compared with the T regions; while the O-T phase boundary regions show the most dopants and strongest lattice distortion. The local segregation of doping elements leads to the formation of the phase coexistence of O and T nanophases. As shown in Fig. 3o, p, the bond lengths and the atom intensities at B sites are relatively uniform, with fewer fluctuations, which is due to the relative differences in the atom size and atom number between B-site dopants (Zr and Hf) and Nb matrix are relatively small, compared with the A-site dopants Bi and K/Na matrix (Tables S1 and S2). The results provide insights into the relationship between A/B-site doping and the formation of the O-T phase boundary, guiding the doping behavior of KNN-based piezoelectric ceramics.

Ferroelectricity and piezoelectricity

Fig. S9a demonstrates the P-E hysteresis loops of the xT ceramics tested at 30 kV/cm and 10 Hz. Although the coercive field (E_c) of xT ceramics is comparable to that of xR ceramics, all the xT ceramics exhibit more well-saturated square P-E hysteresis loops with relatively larger remnant polarization (P_r) and maximum polarization (P_max) than xR ceramics (Fig. S9b), such as, the P_r is close to 28 μC/cm² in 3T ceramics, while in 3R ceramics, the P_r is only around 21 μC/cm², indicating enhanced ferroelectricity. The crystal orientation makes the arrangement of dipoles more effective under the applied electric field, thereby improving the polarization efficiency and ferroelectricity of xT ceramics. The ferroelectricity is also influenced by the phase structure. It can be seen that the changes in P_r and P_max of the KNN-BNZ-xBKH ceramics regardless of texture and random keep increasing from x = 0.01 to 0.03 as the O-T phase boundary gradually moved towards room temperature, and then decrease at x = 0.04. Meanwhile, E_c also gradually decreases with the increase of the T phase, consistent with previously reported results³⁹. This is initially associated with a decrease in free energy due to the emergence of the phase boundary, followed by an increase in relaxor feature due to the gradual dominance of the T phase. Compared with their random counterparts, the textured ceramics achieved significant improvements in both strain and piezoelectricity, especially for the 3T ceramics, which not only possessed both high piezoelectric coefficient and high Curie temperature (d₃₃~550 pC/N, T_C~330 °C, Fig. S10b), but also observed a large strain of ~0.2% at 30 kV/cm (Fig. S10a). Such excellent comprehensive performance is highly competitive in lead-free piezoceramics (Fig. S11)^{3,5,9,25,27,34,35,40,41,42,43,44,45,46,47}. It should be noted that the random and textured KNN-BNZ-xBKH ceramics share a similar chemical composition and room temperature phase structure, and the significant difference in piezoelectricity between them mainly stems from the high crystal orientation. Therefore, the superposition of the effective arrangement of dipoles caused by the preferred crystal orientation and the low domain wall energy of the phase boundary contributes to the easy polarization rotation along different polarization states under an external electric field, thus obtaining a high piezoelectric response.

To further establish the relationship between performance and microstructure, the 3T ceramics were subjected to in-situ electric field X-ray diffraction tests, and the corresponding (200)/(002) peaks are shown in Fig. 2g and Fig. S12. As the applied electric field increases, the position of (002)/(200) peaks slightly moved to the lower 2θ due to the lattice distortion caused by electric field. More importantly, it is also observed that the peak intensity ratios of the (002)/(200) change gradually with the increase of the applied electric field, resulting in a continuous phase transition from the O-T phase to the pure O phase, in particular near the coercive field (~14 kV/cm). Fig. 2h shows the crystal structure evolution of the T-O sequential phase transition. The above results provide key evidence to reveal that efficient polarization rotation occurs inside the 3T ceramics under the action of an external electric field¹⁸. It is noteworthy that the phase transition is almost irreversible after the withdrawal of the electric field, indicating a high irreversible lattice distortion, leading to high residual polarization and excellent piezoelectricity.

The structure of domains is closely related to the electrical properties of ferroelectric materials^48,49,50, and the SEM images of acid-etched domain patterns are shown in Fig. 3a, b. Not only the morphology of the watermark domain, the T/O phase-related abundant hierarchical domain structure inside can be seen visually. The hierarchical domain architecture originates from low domain wall energy and almost disappeared polarization anisotropy, typically occurring in multiphase coexisting regions, which not only facilitates the reduction of the coercive field but also plays an important role in the construction of diffusion-type phase boundary²⁹. In addition, it can be seen from the complete SEM images that there is a significant difference in the grain size between 3R ceramics and 3T ceramics, as shown in Fig. S13. Due to the large grain size, the overall domain size in 3T ceramics is significantly larger than that in 3R ceramics, which can be further verified in TEM images (Fig. 3c, d, and Fig. S14). The reduced coercive field caused by fewer grain boundaries and the preferred crystal orientation compensates for the increased domain wall energy of the larger domains, resulting in comparable E_c in xT and xR ceramics (Fig. S9a, b), which allows the larger-sized domains in 3T ceramics to undergo efficient switching and contributes to a more saturated polarization. Therefore, the high piezoelectric response of 3T ceramics also benefits from the contribution of larger-size domains. In addition, from the element mapping images of TEM, it can be further clearly seen that the several main elements in 3T ceramics are uniformly distributed without obvious impurities (Fig. S15).

Piezoelectric temperature stability

The heat generated by piezoelectric materials during service or fluctuations in external temperature can cause changes in the temperature of the material itself. Therefore, in addition to focusing on the performance of piezoelectric materials at room temperature, the temperature stability of piezoelectric performance is also a very important indicator for transducer and sensor applications, especially the small signal piezoelectric coefficient d₃₃ (i.e., direct piezoelectric effect)^51,52,53,54. It is well known that d₃₃ is proportional to ε_r·P_r, so the temperature stability of both dielectric constant and ferroelectricity are closely related to the piezoelectric temperature insensitivity. As shown in the ε_r–T curves (Fig. 2c), it can be seen that the O-T phases boundary with MPB feature makes the ε_r of 3T ceramics rise slowly between room temperature and Curie temperature, and remains almost at a horizontal line until 150 °C. In contrast, from the in-situ variable temperature P-E hysteresis loops (Fig. S16), it can be found that both the P_max and P_r decrease slowly from room temperature to 180 °C. Thus, the complementary effect between the dielectric constant and P_r may lead to the temperature stability of the piezoelectricity for the 3T ceramics. The stability of piezoelectric performance is also closely related to the internal structure of ceramics, so we conducted a detailed analysis of the poled phase structure and domain structure. Fig. 4a shows the in-situ temperature-dependent XRD of 3T ceramics after poled. As the temperature increases, the XRD diffraction peaks remain basically unchanged below 200 °C. Peak fitting was performed on the (002)/(200) peaks (Fig. S17) in order to more accurately understand the changes in phase composition. As can be seen in Fig. S17b, only the single O phase exists in the range from room temperature to 150 °C, and the T phase begins to gradually emerge only when the temperature approaches 200 °C, exhibiting excellent temperature stability of the phase structure similar to that of the pre-poled one, which greatly improves the temperature stability of P_r and the post-poled ε_r.

Moreover, we also investigated the evolution of poled domain structure through in-situ variable temperature PFM and TEM measurements. It can be seen that after poled, the micron-scale macrodomains of the 3T ceramics (Fig. S18d) undergo efficient switching and domain growth, resulting in the phase of all domains becoming almost 180° (Fig. S18a), consistent with the high P_r. Macroscopically, no significant large-scale poled domain recovery occurred when the temperature increases to 200 °C (Fig. S18a–c), which is another key factor for P_r to maintain good temperature stability. While locally, due to thermally stimulated degradation of some unstable smaller domain structures, the peak intensity of the ±180° domains gradually weakened with increasing temperature (inset of Fig. S18a–c). The in-situ variable temperature TEM was further used to investigate the evolution of poled large-sized stripe domains with temperature in 3T ceramics, as shown in Fig. 4b. It can be clearly observed that the large-sized stripe domains can be well maintained at 280 °C. The number of domains has not decreased, and their morphology remains intact. When approaching the T_C, the corresponding domain structure begins to be destroyed and disappears, resulting in depolarization. In addition to stable phase structure, the large-sized domain structure plays an important role in resisting thermal depoled, as the disturbance of larger domain requires a higher driving force. To investigate the temperature response of internal structure for textured ceramics and random multiphase ceramics with different domain sizes after poled, phase-field simulation was employed. Several effects were considered in our model: Firstly, the ferroelectric polycrystal structures with different grain sizes (KNN system) were generated, resulting in textured or random orientations at different grains. Secondly, the random fields were employed due to the presence of defects caused by doping. After the domain structure evolution is stable, a polarization electric field of up to 3 kV/mm was applied to observe the changes in domain structure with temperature. Fig. 4c₁–c₄ and d show the simulated domain structure changes of the large macrodomains for the textured multiphase ceramics and the small macrodomains for the random multiphase ceramics after poled and the corresponding temperature response. The textured ceramics with larger domain structures exhibit a more saturated polarization state after poled, which is beneficial for piezoelectricity. More importantly, its poled domain structure can remain stable at higher temperatures without being damaged (e.g., >173 °C), which is consistent with actual experimental results. Furthermore, we also found through EPR tests that in addition to more oxygen vacancy defects introduced by the increase in xBKH, the oxygen vacancy content of xT ceramics was also higher than that of xR ceramics (Fig. S19). The generation of oxygen vacancies tends to form defect dipoles in ferroelectric ceramics and will be aligned along the applied electric field, which has a pinning effect on the poled domain structure and increases the depolarization energy, thus improving the domain stability³¹, as shown in Fig. 4c₅–c₈. Therefore, comprehensive structural analysis and simulations have ascertained that in addition to the stable phase structure induced by the multiphase coexistence phase boundary with the MPB feature, the larger domain size and the pinning of defects are closely related to the temperature stability of the domain structure. The stability of phase and domain structures will be beneficial for the piezoelectric temperature stability.

Fig. 5a presents the in-situ temperature-dependent d₃₃ of the xT ceramics (x = 0.01 − 0.04). Typically, when a PPT-type O-T phase boundary exists above room temperature, the d₃₃ of KNN-based ceramics will shake up at T_O-T and then drop with increasing temperature, showing extreme sensitivity to temperature³⁵. Whereas, in xT ceramics, the temperature sensitivity of the piezoelectric coefficient d₃₃ decreases as the diffusion of the O-T phase boundary increases with composition. For example, when 0.01 ≤ x ≤ 0.02, the O-T phase boundaries are already very diffuse, effectively improving the temperature stability, which can be confirmed by the in-situ temperature-dependent d₃₃ value of xR and xT in Fig. 5a, b. When the T_O-T is further shifted to room temperature or even below room temperature, the phase boundary characteristics are even close to the MPB structure. Thus, for 3T ceramics, its high d₃₃ remains almost unchanged (fluctuating by only 1.2%) from room temperature to 150 °C. Even in the range of 25–250 °C, its d₃₃ decreases by only 10%, and significant changes occur only when T_C is reached. In random-oriented ceramics, the corresponding components ceramics are of poor temperature stability due to insufficient diffusion, as shown in Fig. 5b. Therefore, the new phase boundary with MPB feature obtains an unprecedented break through in temperature stability for high piezoelectric coefficient KNN-based ceramics (e.g., d₃₃ > 400 pC/N) and also highly competitive concerning other studied Pb/Pb-free ceramics (Fig. 5c–e)^{25,27,31,55,56,57,58}. The collaborative optimization of crystal orientation and design of new phase boundary involved in this work provides an important advance for optimizing the comprehensive piezoelectricity of KNN-based ceramics, especially temperature stability.

In summary, ultra-high temperature stability and piezoelectric coefficient were achieved in the 3T ceramics. Both the T/O phase local distortion associated with the O-T phase boundary in the atomic scale and the correlation between A/B-site doping and the formation of the O-T phase boundary can be confirmed by STEM. While XRD as well as EBSD can further resolve the macro-scale crystal orientation. Benefiting from the O-T multiphase coexistence at room temperature and high orientation, the material undergoes irreversible electric field-induced efficient large-size macro-domain switching and phase transitions, which exhibit high residual polarization as well as a significantly improved d₃₃. In addition, the synergistic effect between the ultra-high stability of phase structures and domain structures through mimicking PZT’s MPB structure is the reason why the 3T ceramics break through the inherent drawbacks of temperature instability. The intrinsic reason is related to the induction of larger-size hierarchical domain structure and the formation of point defect during the grain template growth method (TGG), which greatly improves the depolarization energy of the KNN-BNZ-xBKH textured ceramics. Finally, the excellent thermal stability, with a change rate of less than 1.2% in the temperature range of 25–150 °C and less than 10% in the temperature range of 25–250 °C, coupled with high piezoelectric coefficient (d₃₃)~550 pC/N and high Curie temperature (T_C)~330 °C, making the 3T ceramics a promising candidate for future lead-free piezoelectric ceramics applications.