Figures 2(a)-(c) show the XRD patterns of the unsintered composites and the top and bottom surfaces of the sintered sample, respectively. All observed peaks of the unsintered composites belong to Bi12TiO20 and BaTiO3, and no impurity phase was found. For the top and bottom surfaces of the sintered sample, the Bi12TiO20 and BaTiO3 phases were still present. Except for a small (006) diffraction peak of Al2O3, no other new crystalline phases were detected. Compared with BaTiO3, the relative intensity of Bi12TiO20 on the top and bottom surfaces of the sintered ceramics decreased, indicating that the volume fraction of the Bi12TiO20 crystal phase decreased. Since no diffusion halo was found under the diffraction peak, the content of the amorphous phase was limited [19, 20]. There was no significant difference in the relative intensity of Bi12TiO20 and BaTiO3 phases between the top and bottom surfaces of the sintered sample, so the volume fraction of Bi12TiO20 and BaTiO3 crystal phases of the top and bottom surfaces had little difference. For a better comparison, detailed XRD locations of the unsintered composites and sintered sample are given in Fig. 2 (d). After sintering, the XRD peaks of Bi12TiO20 and BaTiO3 on the top surface of sintered samples obviously moved to a small direction of 2θ, indicating the slight lattice expansion for crystalline Bi12TiO20 and BaTiO3 phases [21, 22]. However, the peak shift and lattice expansion were not obvious for the bottom surface. In this case, the lattice expansion may be caused by the residual stress formed during sintering [22].
XRD results showed that sintered sample was mainly composed of crystalline BaTiO3, crystalline Bi12TiO20, and amorphous Bi12TiO20. Figure 3(a) shows a typical SEM image of the sintered sample. It mainly included two distinct forms: small grains which were easy to be identified with grain size of about several hundred nanometers; large size matters that have an irregular shape with the size of about several to a dozen micrometers. According to the EDS spectrum shown in Fig. 3 (b), the small grains corresponded to crystalline BaTiO3, and the large-sized materials corresponded to Bi12TiO20 crystal phase with amorphous Bi12TiO20 phase distributing on the surface. Because the samples were sintered near the melting temperature of Bi12TiO20, a part of Bi12TiO20 was liquid during sintering process. During the cooling process, the fine BaTiO3 grains may be wrapped in the irregular Bi12TiO20 phase.
The DSC curves of the sintered sample and pure Bi12TiO20 ceramic were shown in Fig. 4. For the pure Bi12TiO20 ceramic, a strong endothermic peak was present around 860°C, and the calculated enthalpy (∆H1) was 282.5 kJ mol− 1. This endothermic peak is the melting peak of Bi12TiO20 crystalline phase. In comparison, a smaller endothermic peak at approximately 836°C was present for the sintered sample. The calculated enthalpy (∆H2) was approximately 137.5 kJ mol− 1. This only emerging endothermic peak is also the melting peak of crystalline Bi12TiO20. The melting point of crystalline Bi12TiO20 decreased because of the expansion of lattice [23]. Assuming that Bi12TiO20 phase in sintered sample was 100% crystalline, the content of crystalline Bi12TiO20 phase can be calculated by the ratio of ∆H2/∆H1 [24]. According to calculation, about 48% of Bi12TiO20 was crystalline. Therefore, we can further estimate an approximately 20% volume fraction of amorphous Bi12TiO20 in the sintered sample. Although the sintered sample contained no oriented grains, it can present relatively large piezoelectricity without undergoing an electric poling course. Most of the piezoelectric strain constants (d33) were in the range of 13−15pC/N, and the maximum value was increased to 20pC/N. In addition, the symbol of d33 remained almost unchanged over the entire surface. In comparison to Bi12TiO20-based ceramics reported in the past, the piezoelectric constants were greatly improved [18].
Figure 5(a) shows the temperature dependence of dielectric constant and electromechanical coupling coefficient k (determined by resonance method). The results showed that the dielectric properties of sintered samples were similar to those of pure BaTiO3 ceramics [25, 26]. The dielectric curve had a maximum at approximately 120°C. This temperature corresponded to the Curie temperature of crystalline BaTiO3. There was no macroscopic ferroelectricity in BaTiO3 phase, above this temperature [25]. At room temperature, the dielectric constant of the sample was only 280, which was far lower than samples sintered on Al2O3 ceramics (about 600). It was observed that the k value had little change between − 50°C and 350°C, and there was no decreasing trend even though the temperature was much higher than the Curie temperature of BaTiO3. Therefore, this unusual piezoelectricity appears to have no relationship with crystalline BaTiO3. Figure 5(b) shows the piezoelectric resonance frequencies of BaTiO3 ceramics and the sintered sample as functions of temperature, and the illustration shows the details of piezoelectric resonance frequencies of the sintered sample at 400°C. The experimental results showed that the piezoelectric resonance signal of BaTiO3 ceramic can be observed at about 180°C. Below this temperature, the tendency of the piezoelectric resonance frequency of the sintered sample was similar to that of BaTiO3 ceramics. While, above this temperature, the resonance frequency changed slowly with the increase of temperature. The resonance frequency curve can be fitted into a straight line from 250 to 600°C, and the calculated temperature coefficient was about 100×10− 6/°C. This is an fascinating feature for Bi12TiO20-basded ceramics.
The Bi and Ba elements of the top and bottom surfaces of the sintered sample were quantitatively estimated by XPS. By using C1s signal (284.6eV) as a reference, the high-resolution spectra of the elements determined in the measurement scanning were corrected [27]. The percentages of Bi and Ba elements were calculated by corresponding software. The calculation results showed that the Bi/Ba ratio for the top surface was 1.27, while for the bottom surface was 1.94. Because some small BaTiO3 grains may be wrapped in Bi12TiO20, the calculated Bi/Ba ratio may have errors. However, the difference of the Bi/Ba ratio between the top and bottom surfaces indicated that the Bi content of the bottom surface was higher than that of the top surface. The results of XRD showed that the contents of crystalline Bi12TiO20 and BaTiO3 phase had very little difference. The difference of Bi/Ba ratio according to XPS spectra indicates the different contents of amorphous Bi12TiO20 phase. As shown in Fig. 6, the Ba 3d spectra of the top and bottom surfaces were the same, but for Bi 4f, the peak positions in the bottom surface shifted to the higher binding energies than that obtained in the top surface. The peak shift showed that the content of the amorphous Bi12TiO20 phase in the bottom surface was higher than that in the top surface. This result further proved the inhomogeneity of the content of Bi12TiO20 amorphous phase [12, 27, 28]. Since the amorphous content is limit, the inhomogeneous distribution of amorphous Bi12TiO20 along thickness direction is the main reason for the very low dielectric constant of the sintered sample.
In general, repeated annealing is an effective way of releasing elastic stress [29–31]. The piezoelectric constant of the sintered sample remained unchanged for repeated as well as long time annealing. Figure 7(a) shows detailed XRD comparisons of the unsintered composites and the annealed sample. The XRD peak position of annealed sample was similar to that of unsintered composites, which confirmed the release of the elastic residual stress. Thus, the elastic deformation of crystalline phase should not be the main contributor of this abnormal piezoelectricity. Figure 7(b) shows the Raman spectra of Bi12TiO20 standard sample and the annealed sample. The Raman peaks of the annealed sample can be divided into two groups: p1-p10 which is related to Bi12TiO20 [31, 32] and p11-p13 which is related to BaTiO3[33] (Bi12TiO20 and BaTiO3 peaks completely overlapped at approximately 720cm− 1). The vibration modes of Bi12TiO20 include the TiO4 tetrahedron mode which appeared above 700cm− 1 and the BiO5 polyhedron mode at low frequency. In order to study the detailed contrast of Raman spectra, an extended view near p1, p2, p10/p13 peaks is shown in the illustration. Compared with the Raman peak of Bi12TiO20 standard sample, p10/p13 peak had no obvious change. This suggests that the TiO4 tetrahedra remained unchanged. However, the p1 and p2 peaks in the top and bottom surfaces of the annealed sample moved almost the same degree towards lower wavenumber direction. As a reflection of Bi-O bond vibrations of BiO5 polyhedra, the movement of these peaks indicates the asymmetric elongation of the Bi-O bond and further confirms the expanded distortion of the BiO5 polyhedra. Combining with the XRD spectra, the expansion was related with amorphous structure. By comparing the Raman spectra near the top and bottom surfaces, the BiO5 polyhedra distorted at approximately the same degree. The uniform deformation degree of BiO5 polyhedra along thickness direction may result in the enhancement of piezoelectric constant.
The difference of the preparation process between the sintered sample and the previously reported Bi12TiO20-based ceramics is only the substrate used during sintering process. Since all Bi12TiO20-based ceramics were sintered near the melting point of Bi12TiO20, Bi12TiO20 phase was partially liquid. The liquid Bi12TiO20 phase tended to move downward under the action of gravity. The movement of liquid Bi12TiO20 resulted in a small composition gradient and a certain orientation along the thickness direction. During natural cooling process, the crystalline BaTiO3 phase and Al2O3 single crystal substrate could inhibit the recrystallization of Bi12TiO20, thus result in the forming and maintaining of the gradient of Bi12TiO20 amorphous content and the orientation of distorted BiO5 polyhedra. Since the temperature gradient is rather small, the alignment of distorted BiO5 polyhedron might be lead by the movement of Bi12TiO20 liquid phase [34]. As the substrate was Al2O3 ceramics that had defects, the liquid Bi12TiO20 phase will gradually fill the apparent pores of the substrate, thus the composition gradient was greatly reduced.