Domain Engineering in Bulk Ferroelectric Ceramics via Mesoscopic Chemical Inhomogeneity

Abstract Domain engineering in ferroelectrics endows flexibility for different functional applications. Whereas the domain engineering strategy for single crystals and thin films is diverse, there is only a limited number of strategies for bulk ceramics. Here, a domain engineering strategy for achieving a compact domain architecture with increased domain‐wall density in (K,Na)NbO3 (KNN)‐based ferroelectric ceramics via mesoscopic chemical inhomogeneity (MCI) is developed. The MCI‐induced interfaces can effectively hinder domain continuity and modify the domain configuration. Besides, the MCI effect also results in diffused phase transitions, which is beneficial for achieving enhanced thermal stability. Modulation of chemical inhomogeneity demonstrates great potential for engineering desirable domain configuration and properties in ferroelectric ceramics. Additionally, the MCI can be easily controlled by regulating the processing condition during solid‐state synthesis, which is advantageous to industrial production.

Please note that the assignment of primary phase and secondary phase is interchangeable, but for the sake of simplicity, the phase which possesses reflections of higher intensity is labeled as primary phase. Peaks (1) and (3) (020) reflections of secondary phase. For powders calcined at 850 ℃, the amount of secondary phase increases significantly with Ta addition. However, for powders calcined at 950 ℃, no significant secondary phase is observed. Instead, the peaks of primary phase seem to shift towards each other, indicating the interdiffusion of primary phase and secondary phase. From Figure S2, we find that the chemical inhomogeneity in KNN-15Ta@850 is almost everywhere, compared to little but concentrated chemical inhomogeneity in KNN-15Ta@950. From Figure S2b, the distribution of Nb seems to compensate with that of Ta. The domain wall is semi-quantitatively evaluated from the edge detection of phase images. The edge detection was done by using Canny algorithm via programming software MATLAB. Even though the edge detection is visually satisfactory, the signal of grain boundaries can also be included along with the domain wall as the grain boundaries cannot be extracted accurately from the phase image. The domain wall density is estimated from the ratio of edge area (white pixels) over the total area (black and white pixels). Domain-wall densities of KNN-15Ta@850 (ratio=0.179) is roughly 10% higher than KNN-15Ta@950 (ratio=0.161).     Figure S7. Overlapping SAED patterns of "core" (red spots) and "shell" (blue spots) region mentioned in Figure 5. The tilting direction of the "core" SAED pattern with respect to the "shell" SAED pattern was marked with black curved arrows. (a) EDS analysis and (b) bright-field TEM image of MCI region. SAED patterns of (c) "core" and (d) "shell" regions. The letters "C" and "S" represent "core" and "shell" respectively.
Reflections of superlattice can be found in the SAED pattern of "core", as shown in Figure S8c. Superlattice-reflection-filtered dark-field TEM is presented in Figure S9.  The quasi-static d 33 measured at room temperature is found to increase as a function of Ta addition, as shown in Figure S10e. It is worth noting that most d 33 properties of samples sintered from the powder calcined at 850 ℃ are higher than those calcined at 950 ℃. The enhancement might originate from the facilitated extrinsic contribution due to the modified domain configuration (i.e., increased domain-wall density). However, since d 33 can be affected by many factors, including porosity, defects, etc., conservatively, we should not link the promoted piezoelectricity with the engineered domain configuration herein, but recommend further investigation.  The first-order nature of phase transition can be inferred from the difference between Curie-Weiss temperature (θ) and Curie Temperature (T C ), i.e., the greater the difference, the stronger the first-order nature. From Figure S12, it is observed that KNN-15Ta@850 shows a stronger first-order nature than KNN-15Ta@950.   Figure S14. Fitting of 200 C reflections of KNN-15Ta@850 using 2 peak functions.
From Figure S13, it is observed that the diffraction peak of KNN-15Ta@850 is broader than that of the KNN-15Ta@950, as supported by the analysis of FWHM in Table S2.
From Figure S14, we find that fitting using 2 peak functions shows better agreement with the experimental diffraction pattern. It is observed that the dielectric permittivity decreases with increasing frequency, which is likely due to the diminishing space-charge polarization mechanism at high frequency.
No frequency dispersion in a typical relaxor, i.e., the shift of T C as a function of increasing frequency, is observed for both samples.