Effect of Ta-doping on functional properties of K_0.51Na_0.49NbO₃

KNN based piezoelectrics have been extensively investigated after Saito et al reported a d₃₃ of 416 pC/N in the textured ceramics (K_0.44Na_0.52Li_0.04)(Nb_0.84Ta_0.10Sb_0.06)O₃ fabricated by a Reactive Templated Grain Growth Method (RTGG) [1]. Following Saito et al's work, Guo et al (2004) and (2005) reported Li and Li-Ta modified KNN respectively [2, 3] and Hollenstein et al [4] discussed Li-Ta modified KNN synthesized by solid-state method. They reported piezoelectric properties k_t = 53% and d₃₃ = 200 pm V⁻¹ for the Li-doped ceramics, and k_t = 52% and d₃₃ = 300 pm V⁻¹ for the Li-Ta-co-doped samples. Optimum properties for 0.96KNNTSx–0.04BNKZ were obtained for x = 0.04 for ceramics fabricated using conventional solid state synthesis with d₃₃ = 460pC/N [5], which is higher than those of the textured ceramics reported by Saito et al [1]. Ta⁵⁺, BNKZ, and Sb⁵⁺ were considered to be concurrently responsible for the increase in temperature of rhombohedral to orthorhombic phase transition (T_R-O) and the decrease in temperature of orthorhombic to tetragonal phase transition (T_O-T) in potassium sodium niobate ceramics [5]. All previous studies have been performed on complex compounds containing Ta. So, there was a need to investigate the individual effect of Ta-dopant in the KNN system.

Previous workers have suggested a number of possible dopants, e.g., Mn²⁺ in KNN has shown a reduction in leakage current [6] but most authors have focused on co-doping likewise Li and Ta/Sb co-doping results in a significant enhancement of piezoelectricity [2–4, 7–18], with Ta generally giving a lower d₃₃ than Sb-doped ceramics. Optimization appears to be based around lowering the T-O transition temperature which presumably facilitates greater movement of none-180° domain walls. [19, 20]. More recently, co-doped compositions based on (1−x)(K_0.50Na_0.5)NbO₃–x(Bi_0.50Na_0.50)ZrO₃ (KNN-BNZ) have been initiated through Wang et al [19]. The rationale of this study was to find out the general properties of K_.51Na_.49Nb_1−xTa_xO₃ perovskite ceramics. Consequently, Ta⁵⁺ ($T{a}_{Nb}^{x}$) isovalent-dopant was selected to incorporate into K¹⁺ efficient KNN based composition; where the potassium-rich formulation was more inclined to ferroelectricity according to the phase diagram studied by Jaffe et al [21].

Starting materials Nb₂O₅ and Ta₂O₅ were obtained from Stanford Materials Corporation with 99.999% purity, whereas K₂CO₃ and Na₂CO₃ were obtained from Fisher Scientific with a purity of 99.9%. Calcined powders of K_.51Na_.49Nb_1−xTa_xO₃ (0.05≤x≤0.30) ceramics were obtained by using solid state sintering route as explained elsewhere [22]. The calcined powders were pressed at into 10mm diameter pellets using a uniaxial press. The green bodies were then sintered in the temperature range 1165°C–1205°C for 4h.

XRD patterns of sintered pellets were obtained using a Siemens D500 diffractometer in the 2θ range of 10°–80°, using CuKα radiation. Secondary electron images (at 5 kV) were taken on an FEI SEM at a magnification of 4000X. The dielectric properties were measured using an LCR-meter (Model 4284A, Hewlett Packard). For LCR measurements, primary electrodes of Au were pasted on both flat surfaces of the pellet and fired at 800 °C. The pellets with Au-electrode were then connected between Pt secondary electrodes to form a capacitor. Samples were put in the non-inductively-wound tube furnace, with extra thermocouple near the pellet for accurate measurements of temperature. Then, LCR-meter connected to a computer through a CP-IB interface and data was measured by the software. After running the programme, capacitance and tanδ values measured automatically for total of 800 scans (scan/minute) with respect to the temperature and frequency, from room temperature to 600 °C. Capacitance was converted to relative permittivity according to the relationship, ${{\boldsymbol{\varepsilon }}}_{{\boldsymbol{r}}}=\tfrac{{\boldsymbol{C}}}{{{\boldsymbol{\varepsilon }}}_{0}}.\tfrac{{\boldsymbol{t}}}{{\boldsymbol{A}}}$ where, C = capacitance (F), ε₀ = permittivity of free space (8.85419 × 10⁻¹²F-m⁻¹), t = thickness of pellet without electrode (m) and A = area of the electrode on disc surface of the pellet (m²).

Polarization versus electric field (P-E) loops were recorded using a ferroelectric-tester which comprised of an RT66A signal generator (Radiant Technology, USA) linked to a high voltage interface (HVI), power box (PB) and a high voltage amplifier (HVA, Trek). The samples were put in silicone oil between Cu electrodes. The maximum applied electric-field in air-sintered pellets was 60 kV cm⁻¹.

Poling of the thin disc pellets carried out using a high voltage power supply of SRS: Stanford Research Systems, Inc (Model PS350/5000 V–25 W)with an applied field of 40 kV cm⁻¹ at a temperature of 100 °C with ten minutes soaking time, followed by cooling to room temperature under the applied voltage. After poling, d₃₃ values measured by a Piezometer (Piezotest PM300, PiezoMeterSystem), operating at a frequency of 110Hz and a dynamic force of 0.25N.

Figure 1 shows the XRD patterns of isovalent-doped compositions K_.51Na_.49Nb_1−xTa_xO₃ (0.05 ≤ x ≤ 0.30). XRD of x = 0.00 could be compared to our previous work by Hussain et al [22]. All isovalent-doped compositions are single phase [23] within the detection limits of in-house diffractometers. This system exhibits an O-T transition at x = 0.3 as illustrated by splitting of the 2θ ∼ 45°, (220) and (002) peaks are zoomed in figure 4(b) to visualize the splitting clearly. XRD data show the structure to be tetragonal at room temperature with up to 30-mol% of Ta⁵⁺ on the B-site, but double-peaks were observed in relative permittivity plots even though XRD appeared to index based on a single perovskite phase. The simplest explanation is that Ta-doped KNN is a two-phase mix of Ta rich and Nb rich (with respect to the base composition) perovskite. The absence of a second perovskite phase in the XRD data may be explained by the identical ionic radius and thus resultant lattice parameter of Ta⁵⁺ (0.64 Å) and Nb⁵⁺ (0.69 Å) [24]. Despite their identical radius and charge, the difference in mass (Ta (180.95 amu) > Nb (92.90 amu)) [25] accounts for a difference in polarizability with (4.75 °A³) Ta⁵⁺ > Nb⁵⁺ (3.98 °A³ ) [26], hence Ta has a lower T_C than the Nb rich phase and two peaks appear in the permittivity versus temperature curves.

Zoom In Zoom Out Reset image size

In comparison to un-doped KNN [22], the grain morphology of Ta-doped KNN-51/49 remained similar in all compositions, as in figure 5. All compositions revealed the densely packed arrangement of grains, which were consistent with a relative density >90% of the theoretical density.

SEM images of Ta-doped KNN-51/49 are shown in figures 2(a)–(g). The grains were diffused together perhaps through liquid-phase-sintering [24] for low Ta-concentrations, but for x = ≥0.20, a cuboid grain morphology [25] was observed implying that the more refractory Ta₂O₅ with respect to Nb₂O₅ was favoring a solid rather than the liquid-state-reaction (figure 5). It is worth noting that the intragranular cubic pores (∼1 μm) were present in x = 0.15 and x = 0.2, as shown in figure 2(g).

Zoom In Zoom Out Reset image size

Figure 3 illustrates dielectric properties of K_0.51Na_0.49Ta_xNb_1−xO₃ (for x = 0.05, 0.10, 0.15, 0.20, 0.25, and 0.30) or K_0.51NNT at 100 kHz, and measured up to 600°C. For compositions with x > 0.05, the dielectric data revealed two peaks at the both T_T-C and T_T-O transition-temperatures respectively, as in figure 4. These double-peaks are considered to arise from two ferroelectric phases, one KNN, and other KNT-rich with respect to the base composition. There are no extra peaks found in XRD of Ta-doped KNN-51/49 formulations (figure 1), but Ta⁵⁺ and Nb⁵⁺ have the same ionic radius and differ only in mass. Hence it likely that the XRD peaks overlap for KNT and KNN based compositions with substitution occurring according to the defect equation:$T{a}_{2}{O}_{5}\to 2T{a}_{Nb}^{X}+5{O}_{O}^{X}.$

Zoom In Zoom Out Reset image size

Despite the presence of two peaks (figure 4), the T_T-O and T_T-C transition still decrease systematically with increasing x which precludes immiscibility and favors the argument that the compositions are chemically inhomogeneous as a result of incomplete inter-diffusion of the Ta and Nb species. Effectively, the presence of two phases is kinetically driven rather than thermodynamically as in the case of immiscible systems. This conclusion is further supported by the broadening of the peaks which suggests that the Ta and Nb-rich regions themselves have a distribution of Curie temperatures consistent with inhomogeneity. Converse to this, these double-double peaks could be the relaxor behaviors showing closer transitions of crystal structure, as observed and discussed by others [27–30] in the case of PIN-PMN-PT, PMN-0.28PT, PMN-32PT, and NKLNTS. Nevertheless, this type of broadening of peaks due to inhomogeneity may open new applications in microwave dielectrics or relaxor type applications in smart ceramics.

Zoom In Zoom Out Reset image size

Figures (a)–(f) illustrates the hysteresis loops for Ta-doped KNN-51/49 compositions with 0.05 ≤ x ≤ 0.3. All samples show symmetrical loops, but as the Ta concentration increased, P_r decreased for the same applied field. E_c broadly remained the same for all formulations (10 kV cm⁻¹) at the maximum applied field in each composition as in figure 5. Figure (b) is extracted from figure 6(a) where P_r and E_c versus composition are plotted at the 40 kV cm⁻¹ field. From this data of the 40 kV cm⁻¹ field, it was attributed that P_r and E_c were decreased from x = 0 to x = 0.30 on average (figure 6). Undoped KNN has the maximum peak values of remanent polarisation and coercive field, but then decreased at x = 0.15 and slightly increased up till x = 0.30. These anomalous changes are principal because not all samples saturate at 40 kV cm⁻¹. These loops were observed similar types as Yunwei Sheng et al [31] reported in the case of KNN-LTS-xCu.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Single doping of Ta⁵⁺ in KNN showed little improvement in direct-piezoelectricity, but showed electro-trictive behaviours at higher percentages which can be attributed from its results of LCR-peaks broadening and also infer from their PE-loops with higher slope (from P_r to P_s as in x = 0.15, 0.20, 0.25). In addition to this, it can be suggested to obtain good values of inverse piezoelectricity (d₃₃*) from these systems. The challenging work here is to prepare the Ta-doping compositions homogenously mixed into KNN system and optimization of its sintering (T_s), as it is a refractory so T_s goes higher at higher addition of Ta-doping, to consolidate its particles to get good bulk density. Nonetheless, the bulk density has directly relationship to enhance piezoactivity. M Matsubara et al [32, 33] reported to improve piezoelectric properties of Ta-doped KNN with addition of sintering adds in their two studies (KCT & CuO). Nam-Bin Do et al [34] in 2011, investigated that effect of Ta transformed the tetragonal and rhombohedral ferroelectric phases into pseudocubic paraelectric phase in the case of Bi_0.5 (Na_0.82K_0.18)_0.5TiO₃ Furthermore, they observed excellent ratio of S_max/E_max at an applied electric field of 6 kV mm⁻¹. The trend is that electrostrictive behavior is promoted with addition of Ta. d₃₃ values for Ta-doped KNN-51/49 are presented in Figure 7. These data are consistent with their respective PE loops (Figures (a)–(f)) with compositions showing in some cases marginally higher values than undoped KNN (10%Ta reveals a maximum value of 135pC/N). However, the dielectric data unambiguously illustrates that these compositions are heterogeneous and or having relaxor behaviors. Relaxor type ceramics usually showe higher inverse piezoelectricity [35]. Hence the real trends in d₃₃ might be obscured until further studies are carried out on compositions that have to be undergone either longer sintering times or multiple calcination to homogenize the B-site ion distribution.

Zoom In Zoom Out Reset image size

Overall, the use of a single dopant of isovalent (Ta⁵⁺) had a little positive impact on the dielectric, piezoelectric, and ferroelectric properties of KNN. Maximum d₃₃ = 135pC/N was obtained at x = 0.10. Lower dielectric losses in certain compositions (e.g., x = 0.05, x = 0.20) with a bit higher ε_r could be suitable for MLCCs applications. Interestingly, double peaks obtained at both transitional points which could further tuned for new applications. Broadening in TC peaks of LCR data in some compounds were also observed which showed good agreement with ferroelectric loops where the P_r was much lower than the Ps; this might be suggested to add more doping to explore microwave-dielectric properties. Despite this, these kinds of relaxor ceramics could show higher inverse piezoelectricity. For isovalent dopants such as Ta, the only positive potential impact is a broadening of the Curie maxima, leading to suggestions that these compositions may have practical applications in temperature-stable-capacitors and microwave dielectrics in some formulations respectively, although the cost of Ta may prove prohibitive.

FH acknowledges NED University of Engineering and Technology for funding support. AK is thankful to Abdul Wali Khan University Mardan for a PhD studentship. All authors acknowledge financial support from the Sustainability and Substitution of Functional Materials and Devices EPSRC grant (EP/L017563/1).