Bimodal-Structured 0.9KNbO3-0.1BaTiO3 Solid Solutions with Highly Enhanced Electrocaloric Effect at Room Temperature

0.9KNbO3-0.1BaTiO3 ceramics, with a bimodal grain size distribution and typical tetragonal perovskite structure at room temperature, were prepared by using an induced abnormal grain growth (IAGG) method at a relatively low sintering temperature. In this bimodal grain size distribution structure, the extra-large grains (~10–50 μm) were evolved from the micron-sized filler powders, and the fine grains (~0.05–0.35 μm) were derived from the sol precursor matrix. The 0.9KNbO3-0.1BaTiO3 ceramics exhibit relaxor-like behavior with a diffused phase transition near room temperature, as confirmed by the presence of the polar nanodomain regions revealed through high resolution transmission electron microscope analyses. A large room-temperature electrocaloric effect (ECE) was observed, with an adiabatic temperature drop (ΔT) of 1.5 K, an isothermal entropy change (ΔS) of 2.48 J·kg−1·K−1, and high ECE strengths of |ΔT/ΔE| = 1.50 × 10−6 K·m·V−1 and ΔS/ΔE = 2.48 × 10−6 J·m·kg−1·K−1·V−1 (directly measured at E = 1.0 MV·m−1). These greatly enhanced ECEs demonstrate that our simple IAGG method is highly appreciated for synthesizing high-performance electrocaloric materials for efficient cooling devices.


Introduction
The electrocaloric (EC) effect refers to the adiabatic temperature change in a polar material at an electric field, due to the isothermal entropy change associated with the electric-field-induced change in polarization [1][2][3]. The EC effect of ferroelectric materials has attracted continuous attention because of the potential applications in solid-state refrigeration, which is regarded as the most promising solution for cooling microelectronic devices due to the ease of miniaturization, high efficiency and low cost. Based on these considerations, a high EC performance material should possess a large isothermal entropy change (∆S) and hence large adiabatic temperature change (∆T) under a reasonable electric field (E). In other words, large EC strengths (defined by |∆T/∆E| and ∆S/∆E, where parameters T, S and E are the temperature, isothermal entropy and applied electric field, respectively) are favored. Additionally, a wide working temperature range near room temperature (RT) is favored in order to develop high performance EC cooling devices [4][5][6]. Therefore, one critical question here is how to design and develop high-performance Micron-sized KN-BT(9/1) filler powder was fabricated using KNbO 3 and BaTiO 3 as raw powders by using the conventional ceramic processing. The KNbO 3 powder was prepared from pure grade K 2 CO 3 (99.99%, Aladdin) and Nb 2 O 5 (99.95%, Alfa Aesar) powders, which were first ball-milled and calcined at 640 • C for 4 h. Then, the calcined powders were ball-milled again for 24 h. After drying at 120 • C overnight, the KNbO 3 powders were mixed thoroughly with a commercial-grade high purity nano-sized BaTiO 3 powder (D 50 = 50 nm, 99.9 wt.% purity, SAKAI Chemical Industry Co. Ltd., Osaka, Japan), in a molar ratio of 9:1, by ball-milling. The mixture was then calcined at 900 • C for 2 h, ball-milled again for 24 h, and dried at 120 • C overnight to produce the KN-BT(9/1) filler powder ( Figure S1c).
A modified Pechini method was introduced to prepare KN-BT(9/1) sol precursor. All raw materials were weighed according to the designed composition. For Nb-sources, Nb 2 O 5 powder was dissolved in hydrofluoric acid (48-51%, ACS, Alfa Aesar) at 80 • C. Then, ammonium hydroxide (28% NH 3 , Alfa Aesar) was tardily added into the solution until the pH value reached 10, followed by filtering, washing and drying of the sediment at 80 • C for 10 h. Subsequently, niobium hydroxide was formed. The Nb source was then obtained by dissolving the niobium hydroxide in a citric acid (CA) solution. For preparing the metal-CA solution (metal: K and Ba), K 2 CO 3 and BaCO 3 (99.95%, Aladdin) were dissolved directly in a CA solution. For the Ti-CA solution, tetrabutyltitanate (C 16 H 36 O 4 Ti, 96%, Alfa Aesar) with the CA solution was heated to 80 • C until the solution became transparent. Finally, all the metal sources were mixed, with a molar ratio of CA:EG (ethylene glycol) to be 1/4 and pH value to be 10. KN-BT(9/1) precursor solution was obtained by stirring the solution at 80 • C for 2 h. The precipitant was dried at 120 • C for 24 h and then calcined at 900 • C for 2 h as a matrix powder. The detailed synthesis routes are schematically shown in Figure S1c.

Preparation of Bimodal Structured KN-BT(9/1) Ceramics
The IAGG method is schematically shown in Figure S1c. The micron-sized KN-BT(9/1) filler powder and the nano-sized KN-BT(9/1) matrix powder were first mixed thoroughly by ball balling in ethanol for 4 h. After drying at 120 • C, the mixture was then uniaxially pressed into green pellets with a diameter of 10 mm and a thickness of about 1 mm. Finally, the green pellets were sintered at 1000-1050 • C for 2 h in air with a heating rate of 2 • C·min −1 . For comparison, the filler powder and the matrix powder were also used separately to prepare KN-BT(9/1) ceramics by similar procedures. All the samples were cooled by natural cooling in the furnace.  (2001)). Microstructures of the ceramic samples were observed by using scanning electron microscopy (SEM), equipped with energy dispersive X-ray spectroscopy (EDS) (JSM-6335F, JEOL Japan Electronics Co., Ltd., Kyoto, Japan) at 30 kV. The morphology, microstructures and polar nanodomain regions (PNRs) were observed using a high-resolution transmission electron microscope (HRTEM, Tecnai G2 F20 S-Twin, FEI, Hillsboro, OR, USA), acceleration voltages 200 kV, spot size 2.

Measurement of Specific Heat Capacity
The specific heat capacity of the samples was measured by using the Mettler Toledo DSC3 instrument according to the Sapphire method. The heat flow was measured directly in the temperature range of −50-200 • C such that the specific heat capacity can be given by where C p,sam and C p,sap , Φ p,sam and Φ p,sap , m sam and m sap , are the specific heat (J/K·g), heat flow (W·g −1 ), and mass of the sintered KN-BT(9/1) bulk sample at 1050 • C and standard sapphire as reference, respectively.

Characterization of Dielectric and Ferroelectric Properties
To measure dielectric and ferroelectric properties of the ceramics, two sides of the disc samples were coated with Ag paste fired at 600 • C for 30 min as electrodes. Temperaturedependent dielectric characteristics were measured over 1-100 kHz by using a dielectric analyzer (TZDM-RT-800, Harbin Julang Technology Co., Ltd., Harbin, China) over −20-500 • C, at a rate of 1 K·min −1 . Ferroelectric hysteresis (P-E) loops were recorded by using a modified Sawyer-Tower circuit method operated at a frequency of 10 Hz, over the temperature range from RT to 52 • C using the power supply Trek Model 610C.

Measurement of Electrocaloric Effect
Direct ECE measurement was carried out in this study. To directly measure the ECE signals, a pulsed electric field was applied to the sample with a thermocouple (Precision Fine Wire Thermocouple, Omega Engineering, Inc., Norwalk, CT, USA) attached directly to record the temperature variation as shown in Figure 1. the ceramic samples were observed by using scanning electron microscopy (SEM), equipped with energy dispersive X-ray spectroscopy (EDS) (JSM-6335F, JEOL Japan Electronics Co., Ltd., Kyoto, Japan) at 30 kV. The morphology, microstructures and polar nanodomain regions (PNRs) were observed using a high-resolution transmission electron microscope (HRTEM, Tecnai G2 F20 S-Twin, FEI, Hillsboro, OR, USA), acceleration voltages 200 kV, spot size 2.

Measurement of Specific Heat Capacity
The specific heat capacity of the samples was measured by using the Mettler Toledo DSC3 instrument according to the Sapphire method. The heat flow was measured directly in the temperature range of −50-200 °C such that the specific heat capacity can be given by where Cp,sam and Cp,sap, Φp,sam and Φp,sap, msam and msap, are the specific heat (J/K·g), heat flow (W·g −1 ), and mass of the sintered KN-BT(9/1) bulk sample at 1050 °C and standard sapphire as reference, respectively.

Characterization of Dielectric and Ferroelectric Properties
To measure dielectric and ferroelectric properties of the ceramics, two sides of the disc samples were coated with Ag paste fired at 600 °C for 30 min as electrodes. Temperature-dependent dielectric characteristics were measured over 1-100 kHz by using a dielectric analyzer (TZDM-RT-800, Harbin Julang Technology Co., Ltd., Harbin, China) over −20-500 °C, at a rate of 1 K·min −1 . Ferroelectric hysteresis (P-E) loops were recorded by using a modified Sawyer-Tower circuit method operated at a frequency of 10 Hz, over the temperature range from RT to 52 °C using the power supply Trek Model 610C.

Measurement of Electrocaloric Effect
Direct ECE measurement was carried out in this study. To directly measure the ECE signals, a pulsed electric field was applied to the sample with a thermocouple (Precision Fine Wire Thermocouple, Omega Engineering, Inc., Norwalk, CT, USA) attached directly to record the temperature variation as shown in Figure 1. For measurement, the sample was hung in air through two wires to avoid any heat dissipations. The thermocouple that directly touched one surface of the sample (ground side) was connected to an oscilloscope (Teledyne LeCroy WaveSurfer 3024 Oscilloscope 200 MHz) to record the temperature, power supply: Trek Model 610C. The measurement was carried out by (i) manually applying an electric field to the sample with a positive peak appearing on the oscilloscope, (ii) waiting for the heat peak to completely pass and temperature curve to become constant, then manually removing the electric field, and (iii) showing a cooling peak. No constant pulse wide of the electric field was set. Based on the dimension of the samples, the period was about 2 s. The final cooling performance was For measurement, the sample was hung in air through two wires to avoid any heat dissipations. The thermocouple that directly touched one surface of the sample (ground side) was connected to an oscilloscope (Teledyne LeCroy WaveSurfer 3024 Oscilloscope 200 MHz) to record the temperature, power supply: Trek Model 610C. The measurement was carried out by (i) manually applying an electric field to the sample with a positive peak appearing on the oscilloscope, (ii) waiting for the heat peak to completely pass and temperature curve to become constant, then manually removing the electric field, and (iii) showing a cooling peak. No constant pulse wide of the electric field was set. Based on the dimension of the samples, the period was about 2 s. The final cooling performance was obtained by calculating the temperature difference between the initial temperature and maximum value of the cooling peak. The precision and validity of thermocouple were checked as follows. Firstly, a heat plate with a known temperature, which was confirmed by using another infrared (IR) thermometer, was used to check the precision of the thermocouple, where the temperature read from the thermocouple should be in perfect agreement with that read from the IR thermometer. Secondly, before the ECE of the ceramic samples was calculated, a Teflon plate with the same Au electrode coated on each side was used to check the validity. As shown in the schematic diagram, a Teflon plate was used to replace the ceramic samples, and a similar electric field was applied. No temperature changes can be read from the thermocouple. Therefore, the observed temperature changes should be the real temperature changes of ceramics.   obtained by calculating the temperature difference between the initial temperature and maximum value of the cooling peak. The precision and validity of thermocouple were checked as follows. Firstly, a heat plate with a known temperature, which was confirmed by using another infrared (IR) thermometer, was used to check the precision of the thermocouple, where the temperature read from the thermocouple should be in perfect agreement with that read from the IR thermometer. Secondly, before the ECE of the ceramic samples was calculated, a Teflon plate with the same Au electrode coated on each side was used to check the validity. As shown in the schematic diagram, a Teflon plate was used to replace the ceramic samples, and a similar electric field was applied. No temperature changes can be read from the thermocouple. Therefore, the observed temperature changes should be the real temperature changes of ceramics. Figure 2a shows XRD patterns of the commercial-grade nano-sized BaTiO3 powder, micron-sized KNbO3 powder (calcined at 640 °C ), the micron-sized KN-BT(9/1) filler powder and the nano-sized KN-BT(9/1) matrix powder calcined at 900 °C respectively, together with the bimodal structured KN-BT(9/1) ceramics sintered at 1050 °C using the IAGG method. The TEM (transmission electron microscope) images of the filler and matrix of KN-BT(9/1) powders calcined at 900 °C are shown in Figure 2b,c. As shown in Figure 2a, all the samples exhibit a typical perovskite structure, while the commercial nano-sized BaTiO3 powder exhibits a cubic perovskite structure (PDF: 31-174) [15]. As is well known, for perovskite solid solutions, the orthorhombic and tetragonal structures experience different lattice distortions with respect to the cubic structure. The orthorhombic phase (O-phase) and tetragonal phase (T-phase) can be identified according to the peak splitting, with (220)O/(002)O and (200)T/(002)T peaks for the O-phase and T-phase, respectively. Therefore, the calcined KNbO3 powder exhibits the O-phase due to the (220)O/(002)O splitting with higher left peak than the right one (PDF: 32-0822) [16]. On the other hand, the (IAGG-prepared) ceramic sintered at 1050 °C possesses a tetragonal phase, as evidenced by the splitting of the diffraction peaks (200)/(002) located near 44-47° (inset of Figure S2a), which is in agreement with the phase transition diagram of the KN-BT system [11]. Moreover, the Rietveld refinement of the XRD pattern is shown in Figure S2b, including Figure S2c showing the corresponding crystal structure As shown in Figure 2a, all the samples exhibit a typical perovskite structure, while the commercial nano-sized BaTiO 3 powder exhibits a cubic perovskite structure (PDF: 31-174) [15]. As is well known, for perovskite solid solutions, the orthorhombic and tetragonal structures experience different lattice distortions with respect to the cubic structure. The orthorhombic phase (O-phase) and tetragonal phase (T-phase) can be identified according to the peak splitting, with (220) O /(002) O and (200) T /(002) T peaks for the O-phase and T-phase, respectively. Therefore, the calcined KNbO 3 powder exhibits the O-phase due to the (220) O /(002) O splitting with higher left peak than the right one (PDF: 32-0822) [16]. On the other hand, the (IAGG-prepared) ceramic sintered at 1050 • C possesses a tetragonal phase, as evidenced by the splitting of the diffraction peaks (200)/(002) located near 44-47 • (inset of Figure S2a), which is in agreement with the phase transition diagram of the KN-BT system [11]. Moreover, the Rietveld refinement of the XRD pattern is shown in Figure S2b, including Figure S2c showing the corresponding crystal structure information using the Rietveld method respectively. By comparison, Figure S3a,b shows the Rietveld refinement of the parent XRD patterns, i.e., micron-sized pure KNbO 3 powders calcined at 640 • C, and commercial-grade nano-sized pure BaTiO 3 powders, including the corresponding Rietveld refinement information in Figure S3c,d respectively.

Phase Composition and Microstructure of the Bimodal Structured Ceramics
As observed in Figure 2b,c, the micron-sized KN-BT(9/1) powder, as the filler, consists of the irregular crystalline particles with grain sizes of about 200-500 nm, while the nano-sized KN-BT(9/1) powders, as the matrix, are characterized by uniform rectangle crystalline particles with an average grain size of about 100 nm.
Additionally, Figure S5 shows XRD patterns of the KN-BT(9/1) bulk ceramics prepared by using the conventional ceramic processing (Figure S1a), the sol-gel technique ( Figure S1b) and IAGG method sintering at 1050 • C. All the samples exhibit the singlephase perovskite structure. Figure 3 shows the SEM images and grain size distribution profiles of the KN-BT(9/1) ceramics fabricated using the IAGG method. As shown in Figure 3c,d, the average grain size and size distribution of the bimodal structure of the KN-BT(9/1) bulk ceramics sintered at 1000 and 1050 • C were evaluated by using the software equipped with the SEM equipment with both surface and cross-sectional SEM images at different magnifications. Clearly, the bimodal structure is demonstrated by the two well-separated distribution peaks. For the sample sintered at 1000 • C, only a small number of grains are larger than 1.0 µm in size, surrounded by nano-sized grains of~0.1 µm. When the sintering temperature was increased to 1050 • C, the relatively large grains began to be exaggerated and elongated. Eventually, ultra-large grains with sizes of~10-50 µm were formed. Therefore, our IAGG method is an effective way to develop bimodal grain-size distribution, with a small number of coarse grains uniformly distributed in the fine-grained matrix.
information using the Rietveld method respectively. By comparison, Figure S3a,b shows the Rietveld refinement of the parent XRD patterns, i.e., micron-sized pure KNbO3 powders calcined at 640 °C , and commercial-grade nano-sized pure BaTiO3 powders, including the corresponding Rietveld refinement information in Figure S3c,d respectively.
As observed in Figure 2b,c, the micron-sized KN-BT(9/1) powder, as the filler, consists of the irregular crystalline particles with grain sizes of about 200-500 nm, while the nano-sized KN-BT(9/1) powders, as the matrix, are characterized by uniform rectangle crystalline particles with an average grain size of about 100 nm.
Additionally, Figure S5 shows XRD patterns of the KN-BT(9/1) bulk ceramics prepared by using the conventional ceramic processing (Figure S1a), the sol-gel technique ( Figure S1b) and IAGG method sintering at 1050 °C . All the samples exhibit the singlephase perovskite structure. Figure 3 shows the SEM images and grain size distribution profiles of the KN-BT(9/1) ceramics fabricated using the IAGG method. As shown in Figure 3c,d, the average grain size and size distribution of the bimodal structure of the KN-BT(9/1) bulk ceramics sintered at 1000 and 1050 °C were evaluated by using the software equipped with the SEM equipment with both surface and cross-sectional SEM images at different magnifications. Clearly, the bimodal structure is demonstrated by the two well-separated distribution peaks. For the sample sintered at 1000 °C , only a small number of grains are larger than 1.0 μm in size, surrounded by nano-sized grains of ~0.1 μm. When the sintering temperature was increased to 1050 °C, the relatively large grains began to be exaggerated and elongated. Eventually, ultra-large grains with sizes of ~10-50 μm were formed. Therefore, our IAGG method is an effective way to develop bimodal grain-size distribution, with a small number of coarse grains uniformly distributed in the fine-grained matrix.  Here, it is very important to ensure the composition homogeneity in the coarse and fine grains. To demonstrate this, Figure 3e shows the cross-sectional surface image with high amplification. The EDS spectra of the coarse and fine grains are plotted in Figure 3f, and the automated element identification for the EDS spectra evaluation is shown in Figure S6 respectively. Indeed, the stoichiometry of the large grains is very close to that of the fine-sized grains within the measuring uncertainties. Therefore, it is confirmed that the IAGG method is powerful for obtaining such a bimodal structure while maintaining composition homogeneity over the whole samples. In other words, the IAGG method did not trigger the uneven distribution of K + and Nb 5+ in the two types of grains. Figure 4 shows HRTEM images and the corresponding SAED (selected area electron diffraction) patterns of the coarse and fine grains. As shown in Figure 4b,c,e,f, on the one hand, the almost identical lattice planes demonstrated homogenous structure of the samples. On the other hand, the indexed KN-BT(9/1) grains exhibit the characteristic T-phase structure (PDF #71-0945). Therefore, it is concluded that the bimodal structured KN-BT(9/1) bulk ceramics have a typical T-phase as mentioned above.
Here, it is very important to ensure the composition homogeneity in the coarse and fine grains. To demonstrate this, Figure 3e shows the cross-sectional surface image with high amplification. The EDS spectra of the coarse and fine grains are plotted in Figure 3f, and the automated element identification for the EDS spectra evaluation is shown in Figure S6 respectively. Indeed, the stoichiometry of the large grains is very close to that of the fine-sized grains within the measuring uncertainties. Therefore, it is confirmed that the IAGG method is powerful for obtaining such a bimodal structure while maintaining composition homogeneity over the whole samples. In other words, the IAGG method did not trigger the uneven distribution of K + and Nb 5+ in the two types of grains. Figure 4 shows HRTEM images and the corresponding SAED (selected area electron diffraction) patterns of the coarse and fine grains. As shown in Figure 4b,c,e,f, on the one hand, the almost identical lattice planes demonstrated homogenous structure of the samples. On the other hand, the indexed KN-BT(9/1) grains exhibit the characteristic T-phase structure (PDF #71-0945). Therefore, it is concluded that the bimodal structured KN-BT(9/1) bulk ceramics have a typical T-phase as mentioned above. For comparison, surface morphologies of the KN-BT(9/1) bulk ceramics fabricated using the conventional ceramic processing and sol-gel technique are shown in Figure  S7a,b, together with the corresponding grain size distribution profiles (c,d). As expected, these two KN-BT(9/1) bulk samples with unimodal structures, with grain sizes of about 250 nm and 300-400 nm, respectively. No abnormal grain growth (AGG) phenomena are observed in the samples sintered at 1050 °C .
Grain growth behavior of the KN-BT(9/1) ceramics prepared by using the IAGG method can be understood with the explanation of Kingery and Kanget et al. [17,18]. Due to the difference in free-energy across a curved grain boundary, the irregular micron-sized KN-BT(9/1) filler powders with large curvature, underwent exaggerated growth, acting as a "seed" to consume the neighboring nano-sized ones in the matrix. Therefore, in a given polycrystalline system, the grain growth behavior is governed by the maximum driving force (Δgmax) relative to the critical driving force (Δgc), showing the mixed controlling growth behavior. Although the grain size could be increased by sintering at very high temperatures or for very longer times in theory, no grain growth with specific morphologies and crystal orientations of large grains occur when Δ gmax is smaller than Δ gc. For comparison, surface morphologies of the KN-BT(9/1) bulk ceramics fabricated using the conventional ceramic processing and sol-gel technique are shown in Figure S7a,b, together with the corresponding grain size distribution profiles (c,d). As expected, these two KN-BT(9/1) bulk samples with unimodal structures, with grain sizes of about 250 nm and 300-400 nm, respectively. No abnormal grain growth (AGG) phenomena are observed in the samples sintered at 1050 • C.
Grain growth behavior of the KN-BT(9/1) ceramics prepared by using the IAGG method can be understood with the explanation of Kingery and Kanget et al. [17,18]. Due to the difference in free-energy across a curved grain boundary, the irregular micron-sized KN-BT(9/1) filler powders with large curvature, underwent exaggerated growth, acting as a "seed" to consume the neighboring nano-sized ones in the matrix. Therefore, in a given polycrystalline system, the grain growth behavior is governed by the maximum driving force (∆g max ) relative to the critical driving force (∆g c ), showing the mixed controlling growth behavior. Although the grain size could be increased by sintering at very high temperatures or for very longer times in theory, no grain growth with specific morphologies and crystal orientations of large grains occur when ∆g max is smaller than ∆g c . Therefore, our IAGG method in this study is an effective way to develop ceramics with extraordinarily oriented large grains at relatively low sintering temperatures when using the solid-state reaction process. It seems that the filler grains are "cloned", while the gel matrix is just like a "nutrient source or reservoir" to breed the fillers to grow. This method is simple, reproducible and low cost, which can be easily extended many other ferroelectric perovskite materials. Figure 5 shows dielectric properties dependent of the temperature (ε r (T)) at different frequencies, and the microstructure of the KN-BT(9/1) bulk ceramics sintered at 1050 • C. As shown in Figure 5a, all the ε r (T) curves exhibit a broad peak centered at about −20~100 • C, indicating that diffused phase transition is present in the KN-BT(9/1) bulk ceramic [11,19]. At RT, the relative permittivity (ε r ) and loss tangent are about 792 and 0.067 at 1 kHz, respectively. The diffused behavior is further confirmed by the dielectric loss curves. Figure 5a, all the εr (T) curves exhibit a broad peak centered at about −20 °C , indicating that diffused phase transition is present in the KN-BT(9/1) bulk cer [11,19]. At RT, the relative permittivity ( ) and loss tangent are about 792 and 0.067 kHz, respectively. The diffused behavior is further confirmed by the dielectric loss cu For relaxor ferroelectrics, the reciprocal of relative permittivity as a function of perature, follows the Uchino and Nomura function, a modified Curie-Weiss law, w is expressed as [20]

As shown in
where is the Curie constant and is the diffusion coefficient ranging from 1 (an normal ferroelectric) to 2 (an ideal relaxor ferroelectric); and are the maxim relative permittivity and corresponding temperature at a fixed frequency, respecti The slope of the fitting curves is used to determine the value in the Figure 5b. The v is = 1.59 at 100 kHz, confirming the relaxor-like ferroelectric behavior of the KN-BT bulk sample.  For relaxor ferroelectrics, the reciprocal of relative permittivity as a function of temperature, follows the Uchino and Nomura function, a modified Curie-Weiss law, which is expressed as [20] 1 where C is the Curie constant and γ is the diffusion coefficient ranging from 1 (an ideal normal ferroelectric) to 2 (an ideal relaxor ferroelectric); ε m and T m are the maximum relative permittivity and corresponding temperature at a fixed frequency, respectively. The slope of the fitting curves is used to determine the γ value in the Figure 5b. The value is γ = 1.59 at 100 kHz, confirming the relaxor-like ferroelectric behavior of the KN-BT(9/1) bulk sample. The ferroelectric nanodomains were observed through HRTEM images, as shown in Figure 5c, displaying grains with sizes of 2-10 nm randomly in the nondomain matrix. The presence of the PNRs provides strong evidence of the diffusion phase during the phase transition in the KN-BT(9/1) bulk sample. The diffused phase transition can be ascribed to the partial breaking of the ferroelectric long-range ordering by the coupled substitutions of Ba 2+ and Ti 4+ ions for K + and Nb 5+ ions with different sizes and charges, respectively, due to the simultaneous occupation of the six-coordination site by Ti 4+ and Nb 5+ [11,21,22]. Moreover, it is believed that the presence of the fine-size grains in the bimodal structured sample is responsible for additional decrease in the T m [23].
As illustrated in Figure 5d, no obvious anomaly is observed in the specific heat capacity. The specific heat capacity value is increased with increasing temperature, with a room temperature value of 0.50 J·K −1 ·g −1 . As shown in Figure S8, a weak anomaly is present on the heat capacity curve, which is similar to the observation of Pb-free relaxor Ba(Ti 0.65 Zr 0.35 )O 3 ceramics [24]. Figure 6a shows P-E hysteresis loops of the KN-BT(9/1) bulk sample sintered at 1050 • C measured at RT and 52 • C at 10 Hz. Figure 6b illustrates P-E curves measured at different electrical fields, while the curve of the remanent polarization versus the electric field at RT is shown as the inset in Figure 6b.
The ferroelectric nanodomains were observed through HRTEM images, as shown Figure 5c, displaying grains with sizes of 2-10 nm randomly in the nondomain mat The presence of the PNRs provides strong evidence of the diffusion phase during phase transition in the KN-BT(9/1) bulk sample. The diffused phase transition can be cribed to the partial breaking of the ferroelectric long-range ordering by the coupled s stitutions of Ba 2+ and Ti 4+ ions for K + and Nb 5+ ions with different sizes and charges, spectively, due to the simultaneous occupation of the six-coordination site by Ti 4+ and N [11,21,22].
Moreover, it is believed that the presence of the fine-size grains in the bimodal str tured sample is responsible for additional decrease in the [23]. As illustrated in Figure 5d, no obvious anomaly is observed in the specific heat pacity. The specific heat capacity value is increased with increasing temperature, wit room temperature value of 0.50 J·K −1 ·g −1 . As shown in Figure S8, a weak anomaly is pres on the heat capacity curve, which is similar to the observation of Pb-free rela Ba(Ti0.65Zr0.35)O3 ceramics [24]. Figure 6a shows P-E hysteresis loops of the KN-BT(9/1) bulk sample sintered at 1 °C measured at RT and 52 °C at 10 Hz. Figure 6b illustrates P-E curves measured at ferent electrical fields, while the curve of the remanent polarization versus the elec field at RT is shown as the inset in Figure 6b. The sample has a ferroelectric nature, whereas the slim P-E hysteresis loops sugg that the ceramics have low hysteresis losses. The P-E loops at RT and 52 °C were nea the same, indicating that the presence of ferroelectricity in nature can be retained ov relatively broad temperature range. Additionally, as seen in Figure 6b, the P-E loop electric field dependent, with the remanent polarization to be increased almost linea with the increasing electric field. At 2.5 MV·m −1 and RT, the values of the Pr and the co cive field (EC) are 0.675 μC·cm −2 and 0.23 MV·m −1 , respectively. By comparison, the loops of the KN-BT(9/1) bulk samples sintered at 1050 °C using the conventional cera processing and sol-gel technique are shown in Figure S9. Obviously, lossy hysteresis lo are present, indicating higher conductive behavior in a unimodal structure. Therefore, bimodal structured bulk ceramic displays a relatively good ferroelectric property. It be inferred that the extra-large grains ensure the ferroelectricity, while the fine grains sure high density to suppress the tunneling current [25]. The sample has a ferroelectric nature, whereas the slim P-E hysteresis loops suggest that the ceramics have low hysteresis losses. The P-E loops at RT and 52 • C were nearly the same, indicating that the presence of ferroelectricity in nature can be retained over a relatively broad temperature range. Additionally, as seen in Figure 6b, the P-E loop is electric field dependent, with the remanent polarization to be increased almost linearly with the increasing electric field. At 2.5 MV·m −1 and RT, the values of the Pr and the coercive field (E C ) are 0.675 µC·cm −2 and 0.23 MV·m −1 , respectively. By comparison, the P-E loops of the KN-BT(9/1) bulk samples sintered at 1050 • C using the conventional ceramic processing and sol-gel technique are shown in Figure S9. Obviously, lossy hysteresis loops are present, indicating higher conductive behavior in a unimodal structure. Therefore, our bimodal structured bulk ceramic displays a relatively good ferroelectric property. It can be inferred that the extra-large grains ensure the ferroelectricity, while the fine grains ensure high density to suppress the tunneling current [25].

Electrocaloric Effect
ECE adiabatic ∆T refers to the temperature drop induced after removing the electric field. The typical thickness of the KN-BT(9/1) bulk sample used in the ECE measurement was 0.264 mm with an electrode diameter of 6 mm. The specific isothermal entropy change, ∆S, is calculated with ∆S = c ∆T/T, where c is the specific heat of the ceramic sample [26]. Figure 7a shows the directly recorded ECE signal of the bulk sample at RT at 1 MV·m −1 , where the temperature is demonstrated to rise and drop as the field is applied and removed. The ∆T and ∆S at RT at different electric fields are presented in Figure 7b. The ratios of ∆T ∆E and ∆S/∆E (or ∆Q/∆E, where ∆Q = T∆S) are used to express the electrocaloric coefficients (ECE strengths). It is found that the KN-BT(9/1) ceramics have high ECE at E = 1 MV·m −1 , corresponding to ∆T = −1.5 K and ∆S = 2.48 J·kg −1 ·K −1 . Accordingly, ∆T ∆E = 1.50 × 10 −6 K·m·V −1 and ∆S/∆E = 2.48 × 10 −6 J·m·kg −1 ·K −1 ·V −1 were obtained at RT. As discussed above, the strongly widened phase transition temperature near RT is responsible for the giant ECE response over a relatively broad temperature range [4,5].
was 0.264 mm with an electrode diameter of 6 mm. The specific isothermal entropy chan ∆S, is calculated with ∆S = c ∆ / , where c is the specific heat of the ceramic sample [ Figure 7a shows the directly recorded ECE signal of the bulk sample at RT at 1 MV·m where the temperature is demonstrated to rise and drop as the field is applied and moved. The ∆T and ∆S at RT at different electric fields are presented in Figure 7b. T ratios of | ∆ ∆ ⁄ | and ∆ /∆ (or Δ /∆ , where ∆ = ∆ ) are used to express the e trocaloric coefficients (ECE strengths). It is found that the KN-BT(9/1) ceramics have h ECE at E = 1 MV·m −1 , corresponding to ∆T = −1.5 K and ∆S = 2.48 J·kg −1 ·K −1 . According | ∆ ∆ ⁄ | = 1.50 × 10 −6 K·m·V −1 and ∆S/∆ = 2.48 × 10 −6 J·m·kg −1 ·K −1 ·V −1 were obtained at As discussed above, the strongly widened phase transition temperature near RT is resp sible for the giant ECE response over a relatively broad temperature range [4,5]. Additionally, as shown in Table 1, compared with the ferroelectric ceramics repor in Refs. [8,9,[25][26][27][28][29][30][31], our bimodal structured KN-BT(9/1) bulk ceramics shows a fairly h ECE coefficient (strength), which is close to the ECE of single crystal BaTiO3 at 10 °C . sides the diffused phase transition temperature near RT, the coarse grains (10-50 µ m the bimodal structure should enhance the dielectric and ferroelectric properties of sample, resulting in high ECE strength at relatively low electric fields [23]. At the sa time, the finer grains derived from the matrix play a crucial role in forming a dense crostructure and inhomogeneous dielectric properties, leading to high entropy [3].  Additionally, as shown in Table 1, compared with the ferroelectric ceramics reported in refs. [8,9,[25][26][27][28][29][30][31], our bimodal structured KN-BT(9/1) bulk ceramics shows a fairly high ECE coefficient (strength), which is close to the ECE of single crystal BaTiO 3 at 10 • C. Besides the diffused phase transition temperature near RT, the coarse grains (10-50 µm) in the bimodal structure should enhance the dielectric and ferroelectric properties of the sample, resulting in high ECE strength at relatively low electric fields [23]. At the same time, the finer grains derived from the matrix play a crucial role in forming a dense microstructure and inhomogeneous dielectric properties, leading to high entropy [3]. However, the unimodal structured KN-BT(9/1) made by using the conventional ceramic processing and sol-gel technique cannot be used to measure the ECEs, apparently, the novel bimodal structured KN-BT(9/1) bulk ceramics by the IAGG method can overcome the shortage of the unimodal structured samples. Additionally, the preparation of the bimodal structured KN-BT(9/1) ceramics using the IAGG method is highly compatible with the conventional ceramic process, giving them potential as micro-refrigerators to be used for cooling the microelectronic devices near RT.

Conclusions
Bimodal structured KN-BT(9/1) bulk ceramics with a tetragonal phase at RT and a diffused phase transition were prepared successfully by using the IAGG method at a relatively low sintering temperature of 1050 • C. In this bimodal structure, the exaggeratedly large grains were evolved from the micron-sized KN-BT(9/1) filler powders, while the fine grains were originated from the KN-BT(9/1) sol precursor matrix. As compared with the unimodal structured counterpart, the bimodal structured KN-BT(9/1) bulk ceramics display a high electrocaloric performance, giving a large ECE-induced adiabatic temperature drop of 1.5 K and a large EC coefficient of 2.48 × 10 −6 J·m·kg −1 ·K −1 ·V −1 at RT, which is advantageous to the design of cooling devices. It is believed that the coarse grains engender the high ferroelectricity and ECE strengths, while the fine grains are responsible for the decreased maximum temperature, and the enhanced density. Our IAGG method is simple, reproducible and cost effective, which can be easily extended to other ferroelectric perovskite materials.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/nano12152674/s1, Figure S1: Synthesis routes using (a) the conventional ceramic processing, (b) Sol-gel technique using the modified Pechini method, and (c) The induced abnormal grain growth method (IAGG). Figure S2: (a) XRD patterns of the bimodal structure KN-BT(9/1) bulk ceramics sintered at 1050 • C using IAGG method. (b) The Rietveld refinement of XRD pattern using the GSAS refinement software, including (c) corresponding crystal structure information through the Rietveld method Inset shows Zoom-in view of 44-47 • . Figure S3: The Rietveld refinement of the parent XRD patterns: (a) micron-sized KNbO 3 powders calcined at 640 • C, (b) commercial-grade nano-sized BaTiO 3 powders. (c,d) Corresponding crystal structure information through the Rietveld method using the GSAS refinement software respectively. Figure S4: (a) XRD patterns of the commercial nano-sized BaTiO 3 (BT) powder, micron-sized KNbO 3 (KN) and KN-BT(9/1) filler powders using the conventional ceramic processing, together with a bimodal structure KN-BT(9/1) bulk ceramics sintered at 1050 • C using IAGG method. (b) Zoom-in view of 44-47 • . The dash line and arrow are drawn to guide eyes. Figure S5: XRD patterns of the KN-BT (9/1) bulk ceramics prepared by using the conventional ceramic processing (i.e., solid-state reaction), sol-gel technique and IAGG method at 1050 • C respectively. Figure S6: The automated element identification for EDS spectra evaluation for (a) Coarse grain. (b) Fine grain respectively. Figure S7: SEM images of the KN-BT(9/1) ceramics prepared at 1050 • C by using (a) sol-gel technique and (b) the conventional solid-state processing, (c) and (d) are corresponding to (a) and (b) respectively. Figure S8: (a,b) The measured specific heat capacity as a function of temperature, together with the enlarge segment between 10-50 • C. (c) The loss tangent dependent of the temperature at different frequency. The red dash circle was drawn to guide eyes. Figure S9: Room-temperature P-E loops of the KN-BT(9/1) ceramics sintered at 1050 • C using (a) the conventional ceramic processing, (b) the sol-gel technique. The corresponding enlarged segments of loops are shown in the Figure S9c and Figure S9d respectively. The red dash circles were drawn to guide eyes.