Simultaneously Enhancing Thermal Stable Dielectric Property and Piezoelectric Response in Lead-Free LiNbO3-Modified (K0.5Na0.5)NbO3- (BaNi0.5Nb0.5O3) System

Perovskite ferroelectic oxides with simultaneous highly thermal stable dielectric property and piezoelectric response are promising candidate for advanced energy, dielectric, and smart devices. The (1-x)[0.98[(K0.5Na0.5)NbO3]-0.02(BaNi0.5Nb0.5O3)]-xLiNbO3 (abbreviated as (1-x)KNBNNO-xLiNbO3; x = 0.00, 0.02, 0.04, 0.06, 0.08) lead-free multifunction ferroelectric ceramic is synthesized by solid-state reaction method. XRD analysis reveals that the samples exhibit perovskite structure with 0 ≤ x ≤ 0.06, and the second phase K3Li2Nb5O15 appears at x = 0.08. The scanning electron microscopy image show that the grain size of ceramics increases from 0.65 to 3.58 μm with LiNbO3 content increasing. Meanwhile, the Curie temperature (TC) shifts to a higher temperature (~ 427°C for x = 0.06). A high dielectric thermal stability of Δε/ε40°C ≤ ±10%, with a high dielectric permittivity (~1,400), is achieved at x = 0.06 over a wide temperature range of ~40–348°C with d33 of ~160 pC·N−1, and a remnant polarization (Pr) of 20.5 μC·cm−2. This work shows that this multifunction material could be applied in sensor to efficiently covert both solar and kinetic energies into electricity over a wide temperature range.

However, lead-based ceramics have urgent toxicity for the environment and/or human development due to their intrinsic volatility under extreme conditions (Hong et al., 2016;Zheng et al., 2018). The emerging of (K 0.5 Na 0.5 )NbO 3 (KNN)based lead-free system would be a remarkable alternative for these drawbacks due to its high-performance piezoelectric, ferroelectric properties, and high curie temperature (Du et al., 2007b;Ding et al., 2018;Zhao et al., 2019). Furthermore, the piezoelectricity and temperature stability substantially determine the macroscopical performance of ferroelectric materials (Zhao et al., 2007;Sun et al., 2016;Yan et al., 2017Yan et al., , 2018c. According to the structure-activity relationship, chemical modification is an effective method to enhance the temperature stability and piezoelectric properties of KNN-based ceramics, including the formation of solid solutions with other compounds, such as LiNbO 3 (Wang and Li, 2010a;Long et al., 2016), LiTaO 3 (Chang et al., 2007;Zhao et al., 2007), LiSbO 3 , BaTiO 3 , Bi 0.5 Na 0.5 TiO 3 (Zuo et al., 2007;Liu et al., 2014Liu et al., , 2017, or a combination of multiple additives (Mazhao et al., 2015). Among them, the lithium niobate (LiNbO 3 ), not composing of the expensive Ta and toxic Sb elements, could simultaneously increase the piezoelectric coefficient (d 33 ) and the Curie temperature (T C ) instead of sacrificing T C for a higher d 33 . Moreover, the Li ion is beneficial to reduce the sintering temperature of ceramics (Wang and Li, 2010a). Unfortunately, few studies have focus to the thermal stability of KNN-based functional ceramics, which is the pivotal parameter for practical device working. Therefore, it is of great interest to design KNN-based multifunction materials with both improved piezoelectricity properties and favored temperature stability. The KNBNNO system is a kind of excellent multifunctional ferroelectric material with narrower bandgap (1.6 eV), large piezoelectric coefficient (100 pC·N −1 ), large pyroelectric coefficient (130 µC·K −1 ·m −2 ) but smaller remanent polarization (11.3 µC·cm −2 ). The synergetic enhancement of as-mentioned piezoelectricity and thermal stability would make this system to be a paradigm for smart platform applications (Bai et al., 2017).
In this work, (1-x)[0.98[(K 0.5 Na 0.5 )NbO 3 ]-0.02(BaNi 0.5 Nb 0.5 O 3 )]-xLiNbO 3 (abbreviated as (1-x)KNBNNO-xLiNbO 3 ; x = 0.00, 0.02, 0.04, 0.06, 0.08) ceramics were prepared by solidstate reaction method through A-site substitution by Li + . The results show that both piezoelectricity and thermal stability of KNBNNO-LiNbO 3 system can be enhanced simultaneously. The underlying physical picture regarding these properties is the appearing of morphotropic phase boundary (MPB) between orthorhombic and tetragonal phases by LiNbO 3 doping. Furthermore, the phase structure, surface morphology, and electric properties (such as dielectric, ferroelectric and piezoelectric properties) were studied. This work would promote the development of lead-free multifunctional ferroelectrics over a broad temperature range.

Materials
KNBNNO-LiNbO 3 lead-free ceramics were synthesized by solidstate reaction method using the analysis reagent of oxides and carbonates: Na 2 CO 3 (99.8%), K 2 CO 3 (99.0%), Nb 2 O 5 (99.5%), Li 2 CO 3 (98.0%), BaCO 3 (99.0%), and NiO (98.0%). The powders were mixed by stoichiometric ratio, then ball-milled in ethanol with stabilized zirconia media for 12 h. After drying at 90 • C, the powder was calcined at 850 • C for 3 h using alumina crucibles before milled for 12 h. The 6 wt.% PVA was mixed into the powders after drying and sieving, and the pellets were pressed at 200 MPa. Finally, the pressed disks were sintered between 1,040 • C and 1,180 • C for 2 h. In order to weaken the evaporation of alkalis during sintering, the disks were embedded in precursor powders.

Methods
The bulk densities were obtained by Archimedes' Method. The phase structure of samples was measured using X-ray diffraction spectrometry (XRD, Shimadzu, LabX XRD-600) with Cu-Kα radiation (λ = 1.54056 Å). The morphology and grain size were investigated using scanning electron microscopy (SEM, JEOL, JSM EMP-800 Tokyo). The remaining disks were polished and coated with high temperature sliver electrodes for electrical property analysis. The dielectric property was measured at 10 kHz using an LCR meter (Agilent, Agilent 4980A, Santa Clara) from 40 • C to 480 • C, while the ferroelectric property was measured using a ferroelectric analyzer (axiACCT, TF-2000, Germany). The piezoelectric coefficient d 33 was measured by Berlincourt-type quasi-static meter, before that, the samples were poled at 150 • C for 30 min in silicon oil, then keep pressure and cool to room temperature for testing, the poling electric field was 40 kV/cm.

RESULTS AND DISCUSSION
The XRD patterns of KNBNNO-xLiNbO 3 solid solution with 0 ≤ x ≤ 0.08 at room temperature are shown in Figure 1A. Only pure perovskite structure is formed at 0 ≤ x ≤ 0.06, and the original diffraction peak moves toward lower angles upon the increasing of x. This characteristic could be attributed to the K + (1.33 Å) and Na + (0.97 Å) ions are substituted by Li + (0.90 Å) ions at the A-site of perovskite structure (Wang and Li, 2007). However, the second phase of tetragonal tungsten bronze K 3 Li 2 Nb 5 O 15 (ICDD: 52-0157) appears when x = 0.08. This phenomenon results from the limited solid solubility of Li into the A-sites of KNBNNO, and the different space group between KNN (Amm2) and LiNbO 3 (R3c) (Yan et al., 2004).
The relatively integrated intensity of the enlarged XRD patterns between 44 • and 47 • were fitted using the Lorentz function (see Figure 1B). It deserves to be noted that the theoretical ratio of I(200) / I(002) is 1:2 and 2:1 for an orthorhombic and tetragonal phase, respectively (Liu et al., 2012). Therefore, the pure KNBNNO sample is orthorhombic phase according to the fitting data (Cheng et al., 2012). However, the intensity ratio of I(200) / I(002) gradually raises with the increasing of doping content. It implies that the obtained phase is combined with orthorhombic and tetragonal phases when x = 0.02-0.06 (Liu et al., 2012). This is because the integrated intensity of these two peaks approaches to ∼ 1 (Zhang et al., 2006). In this regard, a MPB region has been established in the range of x = 0.02-0.06, which would significantly affect the piezoelectric and dielectric properties (Saito and Takao, 2006). The discrepancy of (002) peaks at x = 0.06 might be due to the mild texturing that formed on the surface of solid solution, as proved in Figure 2C. When x = 0.08, the tetragonal phase is appeared with an intensity ratio of ∼ 2:1. Figure 2 shows the morphology of the (1-x)KNNBNNO-xLiNbO 3 solid solution with various x values sintered at their optimized temperatures. It demonstrates most of the regular rectangular grains surrounded by tiny grains. The average grain size increases from ∼ 0.65 µm (x = 0.00) to ∼ 3.58 µm (x = 0.08) with increasing LiNbO 3 content. Besides, the high relative density (>96%) has been obtained. Specifically, the densities of as-synthesized ceramics first increase from 4.35 g·cm −3 (x = 0.00) to 4.46 g·cm −3 (x = 0.06), and then decreased to 4.33 g·cm −3 at x = 0.08. The substitution of Li ions would promote the formation of a transient liquid phase during the sintering process, which could benefit grain growth and compact structure . The reason for the decrease in density may be due to the appearance of a low-density second phase K 3 Li 2 Nb 5 O 15 (theoretical density: 4.376 g·cm −3 ).
The effect of temperature on the dielectric properties at 10 kHz for unpoled (1-x)KNBNNO-xLiNbO 3 ceramics with 0 ≤ x ≤ 0.08 is shown in Figure 3a. Two dielectric peaks are observed from 40 • C to 480 • C, corresponding to the phase transition temperatures, from the orthorhombic phase to the tetragonal phase (marked as T O−T ) and from the tetragonal phase to the cubic phase (marked as T C ), respectively (Zhang et al., 2006;Hao et al., 2015). Meanwhile, the inverse dielectric constant (1/ε) as a function of temperature at 10 kHz has  Figure 3b. The dielectric permittivity above the curie point follows the Curie-Weiss law, as shown in Equation (1).
where T 0 is the Curie-Weiss temperature and C is the Curie Weiss constant. And the Curie point T C can be got from the curve (Bokov and Ye, 2012). The T O−T also corresponds with another peak of the curve. The phase transition temperature of pure KNBNNO (x = 0.00) ceramics can be observed at T C = 375.5 • C and T O−T = 148.5 • C, respectively. Interestingly, the ceramics at x = 0.06 shows the lowest T O−T value, being near to room temperature, and a highest T C , close to 427 • C. Overall, the T O−T decreases largely from 148.5 • C to 57 • C, while the T C increases from 375.5 • C to 423 • C with the increasing LiNbO 3 content (0 ≤x ≤ 0.06). It is likely due to the smaller radius of the Li ion than that of the K and Na ions, consequentially, the tolerance factor of the perovskite structure decreases upon increasing x. Therefore, the tetragonal distortion of the sample should be enhanced, where T C shifts to a higher temperature, while T O−T moves to a lower temperature, as shown in Figure 3c (Yan et al., 2018a,b). Furthermore, both the high T C of LiNbO 3 (∼1,200 • C) and the large crystal anisotropy could also induce this effect (Acosta et al., 2017). Figure 3d illustrates the thermal stability of the relative permittivity ( ε/ε 40 • C ) between 40 • C and 400 • C for the (1-x)KNBNNO-xLiNbO 3 (x = 0.00-0.08) ceramics at   10 kHz. The doping of LiNbO 3 could reduce the fluctuation, namely, enhancing the thermal stability of the (1-x)KNBNNO-xLiNbO 3 solid solution ( ε/ε 40 • C ≤ ±10%). As a result, the 0.94KNBNNO-0.06LiNbO 3 composition exhibits large relative permittivity (∼1,400), optimum thermal stability ( ε/ε 40 • C ≤ ±10%) at 40-348 • C, and low dielectric loss (<2.5%). It is well established that this composition will be a promising candidate for advanced smart devices operating at a broad temperature range.
The piezoelectric properties for the (1-x)KNBNNO-xLiNbO 3 (x = 0.00-0.08) ceramics are shown in Figure 4. The piezoelectric coefficient of the samples increase to an optimal value (∼160 pC·N −1 ) at x = 0.06, and then it drops to 147 pC·N −1 at x = 0.08. The pure KNBNNO composition also exhibits similar piezoelectric property compared with the existing results (Wang et al., 2013). The piezoelectric properties of the samples could be influenced by the existing of MPB (between orthorhombic and tetragonal). The emerging of more polarization vectors than that of single-phase favors more dipole reorientations, which largely improves the piezoelectric property Xu et al., 2016;Acosta et al., 2017;Lv et al., 2019). Moreover, the piezoelectric property might be affected, to some extent, by the dense structure and grain size (Song et al., 2007). Figure 5A illustrates the polarization-electric field hysteresis loops of the (1-x)KNBNNO-xLiNbO 3 (x = 0.00-0.08) ceramics at 10 Hz under room temperature. These samples exhibit typical P-E hysteresis loops. Figure 5B shows that the P r value first slightly increases during x = 0.00-0.06 and then decreases to 11.18 µC·cm −1 at x = 0.08. The obtained maximum value of 20.5 µC·cm −2 appeared at x = 0.06. The E c value decreases from 20 kV·cm −1 (x = 0.00) to 15 kV·cm −1 (x = 0.08). That can be ascribed to more polarization states caused by the coexistence of O-T phase at MPB, where domain wall movement would be easier and coercive field would be lower (Du et al., 2007a;Wang and Li, 2010b). On the other hand, when the grain size of samples increases with the improving of LiNbO 3 content, the density of grain boundary would decrease. Accordingly, the switching of the domain will much easier, leading to the reduction of the coercive field and increasing of remnant polarization. The piezoelectric and dielectric properties have a similar physical mechanism.

CONCLUSIONS
(1-x)KNBNNO-xLiNbO 3 ceramics (x = 0-0.08) were successfully synthesized via the traditional solid-state reaction method. The pure perovskite structure was obtained at 0≤x≤ 0.06. However, the second phase of K 3 Li 2 Nb 5 O 15 with tetragonal tungsten bronze structure was formed due to the solubility limit of Li-ions in the A-sites of solid solution (x = 0.08). There exist excellent piezoelectric and dielectric temperature stability properties with a high Curie temperature at 0.02 ≤ x ≤ 0.06, owing to the formation of MPB between the orthorhombic and tetragonal phases. 0.94KNBNNO-0.06LiNbO 3 ceramics exhibits excellent electrical performance of ε r ∼1400 ( ε/ε 40 • C ≤ ±10%), tanδ < 0.02 at 40-348 • C, and d 33 of ∼160 pC N −1 , P r of 20.5 µC cm −2 at room temperature. It is uncovered that this material could be used in energy, sensor, and smart devices, operating over a wide temperature range.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

AUTHOR CONTRIBUTIONS
HJ and WL conceived and designed the experiments. HJ, HM, and ZC performed the experiments. LX, ZD, LP, and JX analyzed the data. HJ and LX wrote and revised the paper.