The effects of Ba²⁺ content on depolarization temperature and pyroelectric properties of lead-free 0.94Na_0.5Bi_0.5TiO₃–0.06Ba_1+xTiO₃ ceramics

A solid state synthesis route was used to prepare NBT–0.06B_1+xT composition. The raw materials used in this project were powders, bismuth oxide (Bi₂O₃, 99.999 %), sodium carbonate (Na₂CO₃, 99.5 %), barium carbonate (BaCO₃, 99.98 %), and Titanium dioxide (TiO₂, >99.8 %). All chemicals were purchased from Sigma-Aldrich, UK.

The amounts of powders were calculated according to the chemical formula of 0.94Na_0.5Bi_0.5TiO₃–0.06Ba_1+xTiO₃ with x = 0.00, 0.01, 0.02, and 0.03. The raw materials were ball-milled in acetone for 24 h. in order to mix and mill the powders. The resultant slurries were dried at 50 °C for 8 h. The dried powder cakes were ground in a mortar for 10 min. and sieved through a 250 μm mesh in order to aid the calcination. The powders were calcined at 850 °C for 180 min at a heating ramp rate of 1 °C per min and cooling rate 5 °C per min in a furnace. After calcination, powders were re-milled for 24 h. in acetone, and 2 % of poly vinyl alcohol (PVA) was added as an organic binder to the dried powders to enhance the mechanical strength of the particles. After that the powders were dried in oven at 80 °C until fully dried. The dried powders were ground and sieved and subsequently pressed into green pellets with a diameter of 10 mm under an uniaxial compaction with load ~78 MPs for 5 min at RT. The pellets were sintered at temperatures up to 1150 °C in closed crucibles in order to minimize the loss of volatile Na⁺ and Bi³⁺. The pellets were manually wet polished on both sides using silicon carbide paper to improve uniformity and dried at 100 °C overnight to remove the moisture. Silver conductive paint (RS limited) was used to electrode the pellets, and electrical poling at 5.5–6 kV/mm for 10 min at RT in silicone oil was carried out using a Keithley (6517 Electrometer/high resistance) dc power supply.

SEM (FEI XL30 SFEG) was used to look at the surface morphology of the sintered samples and X-Ray Diffraction (XRD) (Siemens Ltd Model: D500) was used to investigate microstructure. As part of the electrical characterisation of the materials, dielectric measurements were made using an impedance analyser (Wayne kerr Electronics Ltd. Model 3245 and Hewlett Packard HP4092A). Dielectric data was collected over a temperature range from RT to 150 °C using a temperature controlled hotplate in the frequency range of 0.1–10 kHz. Pyroelectric measurements were made using the Byer-Roundy [16] method at the temperature between 20 and 90 °C (equipment up limitation) under vacuum and the pyroelectric current was collected by a Kiethley electrometer (Model 6217). Dielectric and pyroelectric data was then used to study the phase changes such as depolarization temperatures (T _d ), and to determine figure of merit values F _i , F _v and F _D .

Sample compositions are listed in Table 1.

Figure 1 shows the XRD patterns of 0.94NBT–0.06B_1+xT (NBT–0.06B_1+xT) ceramic powders calcined at 850 °C for 3 h. Figure 1a showed that all ceramic samples have a pure perovskite structure (ABO₃) and no second phase was observed. The absence of other phase indicates that NBT–0.06B_1+xT lattices have either absorbed the extra amount of Barium (Ba²⁺), and formed the NBT–0.06B_1+xT ceramic solid solutions [17, 18] or the amount of the second phase is too small to be tested by XRD. Figure 1 shows, in all the samples, a splitting in the peak [111] into [003] and [021] at 39.0°–41.0°, indicating the existence of rhombohedral phase (Fig. 1b) and a splitting in the peak [200] into [200] and [002] at 45.0°–48.0°, indicating the existence of tetragonal phase (Fig. 1c). The coexistence of both rhombohedral and tetragonal phases in NBT–0.06B_1+xT ceramics proves that the sample structures are at morphotropic phase boundary (MPB), which was consistent with what was reported in literature [12, 14, 19]. Figure 1c shows that the splitting in peak [200] became quite pronounced for the sample at x = 0.02 as shown in Fig. 1b. It seems that the composition of NBT–0.06B_1+xTi with x = 0–0.03 is still within MPB.

The measurement of the lattice parameters (a, b and c) reveals that the extra Ba content has slightly changed a, b and c from a = b = 5.488 Å, c = 13.505 Å to a = b = 5.493 Å, c = 13.505 Å for Samples A to D respectively (Table 1). Sample C presents the biggest lattice parameters (a = b = 5.494 Å, c = 13.505 Å).

Comparing the ionic radii of the A-site cations of Na⁺ (1.39 Å), Bi³⁺ (1.31 Å) and Ba²⁺(1.61 Å) (coordination No. 12) according to Hume-Rothery rules, Ba²⁺ in NBT–0.06B_1+xTi could replace Na⁺ but seems to be difficult to replace Bi³⁺ because the radius difference between Bi³⁺ and Ba²⁺ is more than 15 % [19]. Yoon et al. [19] reported that the c/a ratio increased with adding more extra Ba content and reached the maximum at x = 0.02. Thus the increase in Ba²⁺ content would lead to the improvement of the tetragonality of BT in NBT–0.06B_1+xT by compensating for the loss in Ba during the sintering stage. However, any extra Ba²⁺ content more than x = 0.02 would not go to increase the c/a ratio because the c/a ratio was saturated at 0.02, and therefore consequently the extra Ba would precipitate into a secondary phase [19], but in this study the second phase was not identified up to x = 0.03, probably due to the amount of the secondary phase being too small to be tested by XRD.

Figure 2 shows the surface morphology of the NBT–0.06B_1+xT ceramics by SEM. The average grain size was reduced from 1.7 µm at x = 0.00 to 1.61, 1.31, 1.52 at x = 0.01, 0.02 and 0.03, respectively. Increasing Ba²⁺ led to a sharp reduction in the grain size and reached the lowest value at x = 0.02. In general, Ba²⁺ behaves as an inhibitor of the grain growth in NBT system. In lead-based ceramics, such as PZT and PLZT (La-containing PZT), similar effect was also observed when lead was substituted with barium [17, 19, 20]. All the samples showed a density 94.63 ± 0.333 g/cm³ as shown in Table 1. Density measurements revealed that Sample A, B and D at x = 0.0, 0.01 and 0.03 have a similar density value; however, Sample C, at x = 0.02 has a higher density.

Figure 3 shows the temperature dependence of the dielectric properties of poled NBT–0.06B_1+xT from RT up to 150 °C at different frequencies (1, 10 and 100 kHz). The relative dielectric permittivity (ɛ_r) and the dielectric loss tangent (tanδ) show a strong dependence upon Ba content and temperature. The ɛ _r increased with temperature for all samples; however, the ɛ_r showed increasing at x = 0.01, and 0.02 at room temperature and this can be related to extra Ba²⁺. The extra Ba at x = 0.01 compensated the deficiency in Ba in BT side and rectified the tetragonality in NBT. The excess Ba up to 0.02 can then replace cations in the A-site of the NBT–0.06B_1+xT perovskite structure, so Ba²⁺ will replace Na⁺ according to the radius matching rules and the donor behaviour of Ba will make the domain wall motion easier resulting in the increase of the dielectric properties of NBT–0.06B_1+0.02T [12].

The ɛ_r values at x = 0.01 and 0.02 were close to 1500 for both compositions whereas the ɛ_r values at x = 0.0 and 0.03 at RT were near 1225. However, at T _d , and 150 °C, the ɛ _r for the samples at x = 0.01 and 0.02 increased to around 2000 and 2150 (T _d ), and above for both compositions 2500 (150 °C) respectively. However, the ɛ _r values for the samples at x = 0.00 and 0.03 were relatively low, which are around 2250.

Depolarization temperature (T _d ) is a parameter that has not clearly been defined [10], but the European standard on piezoelectric properties has defined it as the reduction of remnant polarization because of temperature and other influences [21]. However, there are other definitions of T _d such as the steep reduction of remnant polarization [7], or the phase transition temperature from ferroelectric (FE) to anti-ferroelectric (AFE) or to relaxor anti-ferroelectric or to relaxor [8, 22‐25].

In this study, T _d was identified from the measurement of the dielectric loss (tanδ) versus temperature. Figure 3 shows that the T _d of the NBT–0.06B_1+0.00T was at ~125 °C and decreased to ~85 °C at x = 0.01 and 0.02, and then rises to ~115 °C at x = 0.03. The reduction in T _d is probably due to the formation of A-site vacancies with increasing Ba content in NBT–0.06B_1+xT structures. In ferroelectric NBT–BT perovskite structure, the ferroelectric domain stability is affected by the coupling between ferroelectric (BO₆) and A-site cations, and the stability of the ferroelectric domains will be reduced when vacancies are generated between the octahedral (BO₆) ferroelectric and A-site cations. In this study, Ba²⁺ content at x = 0.01–0.02 created A-site vacancies so the stability of the ferroelectric domain reduced as a result inducing the phase change from ferroelectric (FE) to relaxor anti-ferroelectric (relaxor AFE) and fast reduction by 40 °C in depolarization temperature (T _d ) [14, 18, 21]. With a further increase of Ba²⁺ to x = 0.03, which is more than Ba²⁺ saturation content (x = 0.02), the T _d was decreased by only about 10 °C, because there were less vacancies in A-site due to minimal (Ba²⁺/Na⁺) replacement and more ferroelectric domain wall stability [9, 14, 26]. The value of the T _d of NBT–0.06BT in this work is close to that reported in the literature (~123 °C) but the depolarization temperature of the NBT–0.06B_1+xT, (0.01 ≤ x ≤ 0.02) is lower than 123 °C [27]. Table 1 lists the values of relative dielectric permittivity (ε _r ) and dielectric loss (tanδ) at RT and depolarization temperatures (T _d ) at 1 kHz achieved in this work.

The calculation of the figure of merits (FOMs) is based on the values of pyroelectric coefficient (p), relative dielectric permittivity (ɛ _r), the dielectric loss tangent (tanδ) and the specific heat (\(C_{v} )\). There are several numbers of FOMs which are more appropriate to certain pyroelectric materials applications [6, 27] and are derived from these application requirements.

Pyroelectric coefficient (p) and the FOMs were analysed by using the following equations: [5, 15, 28] where T is absolute temperature, t the time, I _p the pyroelectric current, F _i the high current (i) detectivity figure of merit, F _v the high voltage (V) detectivity figure of merit, F _D high detectivity figure of merit, and \(C_{v}\) the specific heat (2.8 JK⁻¹ cm⁻³, quoted fm [29]).

The pyroelectric current of NBT–0.06B_1+xT, (0.00 ≤ x ≤ 0.03) ceramics was measured from RT to T _d or 90 °C that is our equipment upper limit.

Table 2 shows the pyroelectric coefficient \((p\)) and the figures of merit (FOMs) values of the NBT–0.06B_1+xT, (0.00 ≤ x ≤ 0.03) at RT, T _d or 90 °C. The p values varied with the Ba²⁺ content, and increased from 2.9 × 10⁻⁴ (C m⁻² °C⁻¹) to 3.54 × 10⁻⁴ (C m⁻² °C⁻¹) when x increased from 0.00 to 0.02 at RT. All the samples showed a great increase in p when the measurement temperature increased from 20 °C to T _d or 90 °C. This great increase in p at T _d or 90 °C was caused by the increase of I _p induced by the reorientation of ferroelectric domain and phase transition in MPB. When the temperature rises from RT to T _d , the phase transition from ferroelectric (FR) to relaxor antiferroelectric (relaxor AFR) started, and thus a sharp change in polarization resulted in a significant increase in p at T _d [15].

The T _d for Sample B (x = 0.01) and C (x = 0.02) is below 90 ^°C and for Sample A (x = 0.00) and D (x = 0.03) above 90 °C, which explained why Sample B and C had a greater increase in p than Sample A and D. According to Anton et al. [7], the T _d determination depends on the highest value of the dielectric loss tangent, which may consistently generate higher T _d than other methods such as the thermally stimulated depolarization current; therefore the T _d for Sample A and D might actually be less than ~125 and ~110 °C, as a result that the observed increase in p for Sample A and D could be related to the beginning of the phase transition in these samples.

The increase of Ba content up to x = 0.02 reduces the distortion in the perovskite structure due to the replacement of Na⁺(1.39 Å) by Ba²⁺ (1.61 Å), which can shift the T _d to lower temperature [28]. Further increase in Ba content to x = 0.03 would lead to the appearance of a secondary phase or the replacement of Na⁺. Thus, the depolarization temperature at x = 0.03 shows a slight increase than that at x = 0.01–0.02 [14, 28, 30]. Table 2 also lists the p value (4.14 × 10⁻⁴ C m⁻² °C⁻¹) of a typical PZT ceramic measured at RT as a reference. The prepared NBT–0.06B_1+xT at x = 0.01–0.03 in this research show the comparable pyroelectric coefficients (3.30–3.5 × 10⁻⁴ C m⁻² °C⁻¹) [5, 6, 9].

The values of F _i , F _v and F _D of the NBT–0.06B_1+xT samples (0.00 ≤ x ≤ 0.03) at RT, T _d or 90 °C are also shown in Table 2. The F _i value increased from 1.04 × 10⁻¹⁰ (m/V) (x = 0.00) to 1.24 × 10⁻¹⁰ (m/V) (x = 0.02) at RT and from 1.24 × 10⁻¹⁰ (m/V) (x = 0.02) to 260 × 10⁻¹⁰(m/V) (x = 0.02) at T _d (85 °C). The F _i of Sample C (x = 0.02) at RT is very close to that for PZT [1.415 × 10⁻¹⁰ (m/V)] [3]. However, at T _d (85 °C) this sample (x = 0.02) shows a much higher p value [260 × 10⁻¹⁰ (m/V) than the best value (~7.33 × 10⁻¹⁰(m/V)] at T _d (50 °C) reported by Gue et al. [15].

The F _v values at RT and T _d or 90 °C increased from 0.99 × 10⁻² (m² C⁻¹) at x = 0.00 to 1.05 × 10⁻² (m² C⁻¹) at x = 0.03. All these F _v values exceed the value for PZT [5, 15, 29, 30]. F _v values increased at T _d from 0.95 × 10⁻² to 164 × 10⁻² m²/C for Sample C (x = 0.02).

The values of F _D at RT, T _d or 90 °C are shown in Table 2. The F _D values increased with extra Ba²⁺ content at RT from 8.6 × 10⁻⁶ Pa^−1/2 (x = 0.00) to 9.2 × 10⁻⁶ Pa^−1/2 (x = 0.03). Giant values for F _D achieved at T _d for Samples B and C were 674 and 915(×10⁻⁶ Pa^−1/2) respectively. F _D for PZT [5, 15, 29, 30] at RT is 9.01 × 10⁻⁶ Pa^−1/2 and Sample D showed a higher F _D value than PZT (Table 2).