(K_0.5Na_0.5)NbO₃-based ceramic powders for lead-free piezoelectric applications are usually prepared by solid state synthesis. This generally involves wet ball milling in an organic liquid such as ethanol, which is a drawback for industrial production. In the present work, a method for the preparation of NaNbO₃ nanopowders is modified to produce a simple method of preparing (K_0.5Na_0.5)NbO₃ nanopowders by an aqueous route. K₂CO₃, Na₂CO₃ and NH₄[NbO(C₂O₄)₂].xH₂O are mixed in water and dried to form a gel, which is then calcined at temperatures between 400 ~ 650 °C for 1 h. The uncalcined gel and calcination products are analysed using Thermogravimetric Analysis/Differential Thermal Analysis, X-ray Diffraction, Fourier Transform Infrared Spectroscopy, Scanning Electron Microscopy, Transmission Electron Microscopy and Raman Scattering. A single phase tetragonal (K_0.5Na_0.5)NbO₃ nanopowder with a particle size of ~ 30 nm can be produced after calcination at 650 °C for 1 h.

Keywords: lead-free piezoelectric, (K_0.5Na_0.5)NbO₃, nanopowder, microstructure

Materials based on (K_0.5Na_0.5)NbO₃ (KNN) are leading contenders to replace lead-based piezoelectric ceramics such as Pb(Zr,Ti)O₃ and Pb(Mn_1/3Nb_2/3)O₃-PbTiO₃ [1-7]. KNN-based powders are most commonly prepared by solid-state synthesis, in which alkali carbonates and Nb₂O₅ are mixed together by wet ball milling, followed by calcination at temperatures of 750 ~ 950 °C [3, 8-10]. Due to the hygroscopic and water-soluble nature of K₂CO₃ and Na₂CO₃, ethanol or acetone are usually used as the milling liquid [11, 12]. While use of organic solvents is acceptable in the laboratory, the use of organic solvents on an industrial scale increases production costs, as the solvents must be handled, recycled or disposed of according to local health and safety and environmental regulations [12]. In addition, KNN powders prepared by solid-state synthesis can suffer from chemical inhomogeneity which can cause variations in the electrical properties [10, 13, 14].
As a result, numerous studies have been carried out on the production of KNN powders by water-based methods such as the Pechini method, sol-gel combustion method, hydrothermal method, mechanochemical activation and spray-drying [15-31]. Besides avoiding the use of organic solvents, these processes have potential advantages such as lower synthesis or calcination temperatures, reduced powder size, narrow particle size distribution, improved chemical homogeneity and the possibility of forming nanostructures of different shapes such as nanowires, nanorods, nanocubes and microfingers.
The methods mentioned above can prepare KNN powders but often require specialized equipment or use complex chemical processes or corrosive chemicals. A simpler method which uses easily-handled chemicals would be beneficial. Su et al. described the synthesis of NaNbO₃ powders by simply mixing Na₂CO₃ and ammonium niobium oxalate, NH₄[NbO(C₂O₄)₂(H₂O)₂]      (H₂O)₃, in water to form a slurry, followed by drying and calcination at different temperatures [32]. A single phase NaNbO₃ nanopowder could be formed at a calcination temperature as low as 400 °C. In the present work, their method is modified to produce nanopowders of (K_0.5Na_0.5)NbO₃.

K₂CO₃ (Alfa Aesar, 99%), Na₂CO₃ (Acros Organics, 99.5%) and ammonium niobium oxalate NH₄[NbO   (C₂O₄)₂].xH₂O (Aldrich, 99.99%) were used as starting materials. The mass% of Nb₂O₅ in the NH₄[NbO(C₂O₄)₂]. xH₂O powder was determined by Thermogravimetric Analysis/Differential Thermal Analysis (TG/DTA, DTG-60, Shimadzu, Kyoto, Japan). The K₂CO₃ and Na₂CO₃ powders were dried at 250 °C for 2 h before weighing to remove adsorbed water. The NH₄[NbO(C₂O₄)₂]· xH₂O powder was not dried. Stoichiometric amounts of the starting materials (enough to make 5 g of (K_0.5Na_0.5)NbO₃ powder) were weighed and dissolved in 100 ml of deionized water. The solution was stirred using a digital hotplate/magnetic stirrer (MaXtir 500, Daihan, Wonju, Korea) at a temperature of 65 °C and a rotation speed of 250 rpm for 1 h. The solution temperature was controlled using a temperature probe inserted in the solution. The temperature was then increased to 95 °C and the solution evaporated until ~25 vol% remained. The remaining solution was heated at 200 °C for 3 h. The solution transformed into a white porous gel. After cooling, the gel was crushed into powder in an agate mortar and pestle.
Thermogravimetric Analysis/Differential Thermal Analysis (TG/DTA, DTG-60, Shimadzu, Kyoto, Japan) was carried out on the NH₄[NbO(C₂O₄)₂]·xH₂O powder and gel powder in the temperature range 25 ~ 1000 °C in air at a heating rate of 10 °C·min^-1. Samples of gel powder were calcined at temperatures of 400, 450, 500, 550, 600 and 650 °C for 1 h in air with heating and cooling rates of 5 °C·min^-1. X-ray Diffraction (XRD) was carried out on uncalcined and calcined samples with Cu Kα radiation (XRD XPert PRO, PANalytical, The Netherlands) with 2θ range = 20° ~ 80°, scan speed = 3°·min^-1 and 0.02° step size. Fourier Transform Infrared spectroscopy was carried out on the starting materials, uncalcined gel powder and calcined samples (FTIR, Spectrum 400, Perkin Elmer, Waltham, MA). Powder samples were mixed with KBr and pressed into pellets. Scanning Electron Microscopy (SEM, Hitachi S-4700, Tokyo, Japan) was carried out on uncalcined and calcined samples. Raman scattering (LabRam HR800 UV Raman microscope, Horiba Jobin-Yvon, France), with a 515 nm diode laser and output power of 10 mW, was carried out on a powder sample calcined at 650 °C for 1 h. Spectra were recorded in back scattering geometry over a range of 50 ~ 1,000 cm^-1 with a resolution of ~ 0.5 cm^-1.
In order to study the structure in more detail, XRD (D8 Advance, Bruker, Billerica, MA) with Cu Kα radiation, 2θ range = 20° - 80°, scan speed = 0.3°·min^-1 and 0.0062° step size was carried out on a powder sample calcined at 650 °C for 1 h. A powder sample calcined at 650 °C for 1 h was also studied using Transmission Electron Microscopy (TEM, FEI TECNAI F 20, FEI Company, Hillsboro, OR) operated at an accelerating voltage of 200 kV.

A TG/DTA trace of the precursor NH₄[NbO(C₂O₄)₂]·xH₂O powder is shown in Fig. 1. The DTA trace contains endothermic peaks at 129 °C, 183 °C and 258 °C (with a shoulder at 275 °C), and an exothermic peak at 596 °C (with a shoulder at 579 °C). The peak at 129 °C corresponds to the loss of water of hydration and is accompanied by a loss in weight of ~ 10%. The endothermic peaks at 183 °C and 258 °C correspond to the release of NH₃ and the decomposition of the organic material in NH₄[NbO(C₂O₄)₂]·xH₂O as it transforms to amorphous niobic acid Nb₂O₅·nH₂O [32-35]. These peaks are accompanied by a weight loss of ~ 55%. The exothermic peak at 596 °C corresponds to the crystallisation of amorphous niobic acid into a hexagonal or orthorhombic phase [33, 34, 36, 37]. The small accompanying weight loss of ~ 4% may be due to the release of CO₂ caused by oxidation of residual carbon [33]. A total weight loss of 70.3% was measured between 25 ~ 1,000 °C.
A TG/DTA trace of the uncalcined gel is shown in Fig. 2. The sample has a broad endothermic peak at 100 °C, accompanied by a weight loss of ~ 3.8% which corresponds to the loss of water of hydration. A large endothermic peak at 257 °C followed by a smaller peak at 361 °C corresponds to the decomposition of the gel into K₂CO₃, Na₂CO₃ and Nb₂O₅·nH₂O [27, 32]. These peaks are accompanied by a weight loss of ~ 47.8%. Exothermic peaks at 490 °C and 535 °C correspond to the decomposition of the K₂CO₃ and Na₂CO₃ and the reaction to form (K_0.5Na_0.5)NbO₃ [27, 38, 39]. These peaks are accompanied by a weight loss of ~ 2.5%. A further small weight loss may be due to the decomposition of unreacted K₂CO₃ and Na₂CO₃. Total weight loss is 57.3%.
Fig. 3(a) shows an XRD pattern of the uncalcined gel. The gel has an amorphous structure. An FTIR spectrum of the precursor NH₄[NbO(C₂O₄)₂].xH₂O powder is shown in Fig. 3(b). The main infrared bands are summarized in Table 1 [33, 34, 40]. An FTIR spectrum of the uncalcined gel is shown in Fig. 3(c). The main infrared bands are summarized in Table 2 [33, 34, 41]. Shifts in the peak positions and the absence of the main K₂CO₃ and Na₂CO₃ peaks at 1,450 cm^-1 indicate that NH₄[NbO(C₂O₄)₂]·xH₂O has reacted with K₂CO₃ and Na₂CO₃ to form a gel [41]. However, the appearance of new peaks at 1,259 and 907 cm^-1 due to the ligand exchange reaction between H₂O and NH₃ was not observed [32]. In addition to the bands from NH₄[NbO(C₂O₄)₂].xH₂O, bands from the carbonate groups of K₂CO₃ and Na₂CO₃ are present at 703 and 860 cm^-1, indicating that not all of the K₂CO₃ and Na₂CO₃ has reacted with the NH₄[NbO(C₂O₄)₂].xH₂O.
XRD patterns of the gel calcined at temperatures between 400 ~ 650 °C for 1 h are shown in Fig. 4. Also included are the patterns for ICDD card #77-0038 [(Na_0.35K_0.65)NbO₃, monoclinic, space group Pm(6)], ICDD card # 74-2449 [NaNbO₃, monoclinic, space group P2/m(10)] and ICDD card #76-0977 [K₄Nb₆O₁₇, orthorhombic, space group P21nb(33)]. After calcination at 400 °C, the sample appears to be still amorphous. As the calcination temperature increases to 450 °C, very broad peaks corresponding to (K_0.5Na_0.5)NbO₃, NaNbO₃ and K₄Nb₆O₁₇ form. As the calcination temperature increases from 450 to 500 °C, the peaks become narrower and better defined. The peaks corresponding to NaNbO₃ shift to smaller values of 2θ and merge with the peaks corresponding to (K_0.5Na_0.5)NbO₃. As the calcination temperature increases from 450 °C, the peaks correspond-     ing to K₄Nb₆O₁₇ gradually decrease in intensity, finally disappearing at 600 °C. Some unreacted Na₂CO₃ is also present in the samples calcined at 500 °C and 550 °C (ICDD card # 86-0309).
The crystallite size of the sample calcined at 650 °C for 1 h was estimated using Scherrer’s formula:
    (1)
where τ = crystallite size, λ = X-ray wavelength (0.154184 nm for Cu Kα), β = FWHM (in radians) and θ = Bragg angle. Using the (200) peak with θ = 11.16°, the crystallite size was estimated to be 20 nm.
FTIR spectra of the gel calcined at temperatures between 400 °C - 650 °C for 1 h are shown in Fig. 5. FTIR spectra of the K₂CO₃ and Na₂CO₃ precursor powders are also included. A comparison of the spectra of the uncalcined gel and the sample calcined at 400 °C clearly show that the gel has decomposed. The peaks at 520 and 850 cm^-1 in the sample calcined at 400 °C are due to Nb-O vibrations [33, 42]. The peak at 1,330 cm^-1 in the sample calcined at 400 °C may belong to a CH₃ bending mode [16] or to an oxalate group such as Na₂C₂O₄ or K₂C₂O₄ [43, 44]. It should be noted that the gel calcined at 400 °C contains partially decomposed organic material and is black in colour. The peaks at 1,330 cm^-1, 1,540 cm^-1 and 1,590 cm^-1 may also belong to a carbonato complex [45, 46]. The band at 1,055 cm^-1 may correspond to a C-OH stretching vibration or to the ν₁ vibration of the carbonate ion [16, 45, 47]. The ν₁ vibration is usually infrared inactive, but becomes active when the coordination symmetry of the carbonate ion is lowered i.e. the crystal structure of the carbonate is degraded. These peaks therefore indicate the decomposition of the gel and the formation of alkali carbonates. As the calcination temperature increases to 500 °C, these peaks change position and gradually disappear, while peaks associated with Na₂CO₃ (1,455 cm^-1) and K₂CO₃ (1,455 cm^-1, 1,365 cm^-1, 1,630 cm^-1) appear. These peaks belong to unreacted alkali carbonates. As the calcination temperature increases further, the peaks associated with the alkali carbonates also decrease, finally disappearing in the sample calcined at 650 °C. The colour of the calcined samples also becomes lighter with calcination temperature, finally becoming white at 650 °C. Changes in the position and width of the peaks at 520 and 850 cm^-1 as the calcination temperature increases indicate the formation of (K_0.5Na_0.5)NbO₃.
An SEM micrograph of the uncalcined gel is shown in Fig. 6(a). The gel has a honeycomb appearance, which may be caused by the release of NH₃ and CO₂ during the ion exchange reaction between the starting materials. A micrograph taken at higher magnification shows that the gel appears to lack any polycrystalline structure [Fig. 6(b)]. An SEM micrograph of the gel after calcination at 650 °C for 1 h shows that the calcined gel maintains the honeycomb structure [Fig. 6(c)]. The samples calcined at lower temperatures look similar.
High magnification SEM micrographs of the gel after calcination at different temperatures for 1 h are shown in Fig. 7. All the samples show a polycrystalline microstructure consisting of nano-sized grains, even the sample calcined at 400 °C. The grain size coarsens slightly as calcination temperature increases, but even after calcination at 650 °C, the mean grain size is only ~ 30 nm. This agrees quite well with the estimation of crystallite size using Scherrer’s formula. The samples calcined at temperatures ≥ 500 °C show some areas with abnormal coarsening, with some grains up to 100 nm in diameter. The nano-sized grains have partially sintered together to form agglomerates, even after calcination at the low temperature of 400 °C. The small grain radius and high surface area provides a high driving force for sintering, causing the grains to partially sinter together during the calcination process [48, 49]. The agglomerate size may be reduced by ball milling [50] or by cold isostatic pressing of the powder to break the agglomerates [51].
A slow-scan XRD pattern of a gel sample calcined at 650 °C for 1 h is shown in Fig. 8. The pattern can be indexed with ICDD card# 71-0948 (KNbO₃, tetragonal, space group P4mm). The unit cell parameters were refined as a = 3.9528 Å, c = 4.0033 Å. (K_0.5Na_0.5)NbO₃ usually has an orthorhombic unit cell at room temperature. The change in symmetry to tetragonal is caused by the very small grain size [52]. A small peak at 28.5° (marked with an asterisk) may be due to unreacted K₂CO₃ (ICDD card # 87-0730) or to K₃Nb₇O₁₉ (ICDD card # 84-0812).
TEM micrographs of a gel sample calcined at 650 °C for 1 h are shown in Fig. 9(a) and Fig. 9(d). The grains are cubic in shape, as is typical for (K_0.5Na_0.5)NbO₃ [10, 53] and vary in size from 20-200 nm. It can clearly be seen that the nano-sized grains have sintered together to form agglomerates. Fig. 9(b) shows a high-resolution TEM image of the grain marked by an arrow in Fig. 9(a). The crystal structure of the grain can clearly be seen. Fig. 9(c) shows a selected area diffraction pattern of a different grain. The pattern was indexed using a tetragonal unit cell. EDS point analysis of the three grains marked in Fig. 9(d) is shown in Table 3. The chemical compositions of the grains are close to the nominal composition for (K_0.5Na_0.5)NbO₃, indicating good chemical homogeneity of the powder.
A Raman spectrum of a gel sample calcined at 650 °C for 1 h is shown in Fig. 10. The spectrum shows features typical of (K_0.5Na_0.5)NbO₃ [54, 55]. The peak at 85 cm^-1 and the shoulder at 190 cm^-1 are assigned to translational modes of Na⁺/K⁺ cations and rotations of the NbO₆ octahedra. The peak at 260 cm^-1 is assigned to the ν₅ NbO₆ bending mode. The shoulder at 560 cm^-1 and the peak at 615 cm^-1 are assigned to the ν₂ and ν₁ NbO₆ stretching modes respectively. At 860 cm^-1, there is a mode assigned to a combination tone of the ν₁ and ν₅ modes. The overlap of the ν₁ and ν₂ modes may indicate the tetragonal phase of (K_0.5Na_0.5)NbO₃ [56].
According to Su et al., the following ion exchange reaction takes place between alkali carbonates and ammonium niobium oxalate [32]:

2NH₄[NbO(C₂O₄)₂(H₂O)₂](H₂O)₃ + A₂CO₃ →
2A[NbO(C₂O₄)₂(H₂O)₂](H₂O)₃ + 2NH₃ + CO₂ + H₂O
(A = K, Na)                                                     (2)
 
This is then followed by a ligand exchange reaction:
 
A[NbO(C₂O₄)₂(H₂O)₂](H₂O)₃ + 2NH₃ →
A[NbO(C₂O₄)₂(NH₃)₂](H₂O)₃ + 2H₂O                (3)
 
Only half of the oxalate will undergo the ligand exchange reaction due to an insufficient amount of ammonia. Upon heating, the gel decomposes into K₂CO₃, Na₂CO₃ and Nb₂O₅·nH₂O according to the following reactions:
 
2A[NbO(C₂O₄)₂(H₂O)₂](H₂O)₃ + 2O₂ → A₂CO₃
+ Nb₂O₅·nH₂O + 7CO₂ + (10 - n)H₂O                (4)
 
2A[NbO(C₂O₄)₂(NH₃)₂](H₂O)₃ + 2O₂ → A₂CO₃
+ Nb₂O₅·nH₂O + 7CO₂ + 4NH₃ + (6 - n)H₂O      (5)
 
Because the alkali ions and niobium ions are mixed in the gel on a molecular scale, the gel decomposes to give a finely interspersed mixture of alkali carbonates and Nb₂O₅·nH₂O. This reduces the diffusion distances needed for subsequent chemical reactions, allowing the (K_0.5Na_0.5)NbO₃ phase to be formed at lower calcination temperatures than those achieved by conventional solid state processing. The higher reactivity of Nb₂O₅·nH₂O compared to Nb₂O₅ is also beneficial. In the present work, single phase (K_0.5Na_0.5)NbO₃ nanopowders could be formed at calcination temperatures comparable to those of other aqueous preparation methods, but with a much simpler process [16, 18, 26].
In their work, Su et al. were able to prepare single phase NaNbO₃ powders using calcination temperatures as low as 400 °C [32]. However, in the present work a calcination temperature of 600 °C was necessary to obtain a single phase powder. Furthermore, the (K_0.5Na_0.5)NbO₃ phase appears to develop via intermediate NaNbO₃ and K₄Nb₆O₁₇ phases. It should be noted that, as well as the NaNbO₃ phase, there are other (K_1-xNa_x)NbO₃ phases with very similar diffraction patterns to NaNbO₃ and (K_0.5Na_0.5)NbO₃ (ICDD card # 74-2024 for (K_0.02Na_0.98)NbO₃ and ICDD card #s 74-2025 and 77-0037 for (K_0.1Na_0.9)NbO₃) and it is difficult to assign particular diffraction patterns to the XRD traces as KNbO₃ and NaNbO₃ form a complete range of solid solutions [57]. On calcination at 400 and 450 °C, the gel powder probably forms a range of sodium-rich (K_1-xNa_x)NbO₃ phases along with K₄Nb₆O₁₇ and as the calcination temperature increases, the overall composition moves towards (K_0.5Na_0.5)NbO₃. The exact nature of the sample after calcination at 400 °C is not known. From the XRD pattern (Fig. 4) it appears amorphous, but from the SEM micrograph it appears polycrystalline (Fig. 7). The sample may be partially crystalline.
Malic et al. studied the synthesis of (K_0.5Na_0.5)NbO₃ from alkali carbonates and Nb₂O₅ by the use of diffusion couples of K₂CO₃/Nb₂O₅, Na₂CO₃/Nb₂O₅ and (K₂CO₃+Na₂CO₃)/Nb₂O₅ [58]. They found that the perovskite phase formed via intermediate phases such as Na₂Nb₄O₁₁, K₆Nb_10.88O₃₀, K₄Nb₆O₁₇ and (K,Na)₂Nb₄O₁₁. They also found that the reaction rate for formation of (K_0.5Na_0.5)NbO₃ was determined by the diffusion of the more slowly moving K⁺ ions and that the reaction rate constant for the reaction in the K₂CO₃/Nb₂O₅ and (K₂CO₃+Na₂CO₃)/Nb₂O₅ diffusion couples was an order of magnitude slower than that of the Na₂CO₃/Nb₂O₅ diffusion couple.
Yang et al. in their synthesis of (K_0.5Na_0.5)NbO₃ powders by the Polymerized Complex method found that (K_0.5Na_0.5)NbO₃ formed via intermediate NaNbO₃ and K₂Nb₈O₂₁ phases [15]. Similar results were found in the synthesis of KNN powders by the hydrothermal method [17, 21] Toyama et al. in their study of the synthesis of (K_1-xNa_x)NbO₃ nanoparticles using a supercritical water flow system also found that KNbO₃ could form via an intermediate K₄Nb₆O₁₇ phase [24]. NaNbO₃, on the other hand, formed directly and (K_1-xNa_x)NbO₃ nanoparticles always had a higher Na/K molar ratio than the molar ratio of the starting precursors, indicating that Na⁺ ions have a higher reactivity than K⁺ ions. This difference in diffusion rates and reactivity of the alkali ions would explain why intermediate phases form and why a higher calcination temperature then that of Su et al. is needed to obtain a single phase (K_0.5Na_0.5)NbO₃ powder in the present work. Sodium-rich (K_1-xNa_x)NbO₃ phases form first due to the higher diffusion rate and reactivity of the Na⁺ ions. These phases then change to the (K_0.5Na_0.5)NbO₃ composition at higher calcination temperatures due to the increase in diffusion rate and reactivity of K⁺ ions.

Based on the work of Su et al., a facile aqueous method for the production of (K_0.5Na_0.5)NbO₃ nanopowders from alkali carbonates and ammonium niobium oxalate has been developed. The alkali carbonates and ammonium niobium oxalate are dissolved and mixed in water and an ion exchange reaction takes place between the ammonium niobium oxalate and alkali carbonates. The water is then evaporated, leaving behind a gel which is then calcined. From the TG/DTA, XRD and FTIR results, the calcination reactions begin at temperatures between 450-500 °C, but a calcination temperature of 650 °C is needed to form a single phase (K_0.5Na_0.5)NbO₃ powder. The (K_0.5Na_0.5)NbO₃ phase appears to develop via intermediate sodium-rich (K_1-xNa_x)NbO₃ and K₄Nb₆O₁₇ phases. This is probably due to the difference in diffusion rate and reactivity between Na⁺ and K⁺ ions. A single phase (K_0.5Na_0.5)NbO₃ powder with tetragonal structure and a mean grain size of ~ 30 nm could be prepared at a calcination temperature of 650 °C. This method allows the formation of (K_0.5Na_0.5)NbO₃ nano powder at temperatures 100 ~ 300 °C lower than that of conventional solid state reaction routes and without the need for ball milling the starting materials using organic solvents.

This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Education under Grant number: 2015R1D1A1A01057060. The authors would like to thank Kyeong-Kap Jeong and Jeong-Chi Sang (Centre for Research Facilities, Chonnam National University) for operating the XRD and FTIR, Hey-Jeong Kim (Centre for Development of Fine Chemicals, Chonnam National University) for operating the SEM and Dr. Sang-Hun Jeong and Kang-Min Lee (Korea Basic Science Institute, Gwangju centre) for carrying out the Raman scattering and slow-scan XRD experiments respectively.