For the optimization of deposition condition for crack-free BTO film, we first deposit the films at different oxygen partial pressures. Fig. 2(a) shows the Raman spectra of the BTO target and BTO films grown at different oxygen partial pressures. The MgO substrate has no Raman lines in the spectral range studied [32], thus it does not interfere with the BTO film spectra. The Raman spectra of the polycrystalline bulk target consists of a broad peak centered at 260 cm−1, a sharp peak at 305 cm−1, an asymmetrical strong peak at 519 cm−1, and a broad weak peak at 720 cm−1. According to previous theoretical and experimental studies [12, 33], these Raman bands correspond to A1 (2TO), E (3TO + 2LO) + B1, A1 (3TO), and E (4LO) phonon modes, respectively. In particular, a comparison of the frequency of E phonon soft modes (3TO + 2LO, 4LO) of the film and bulk target provides valuable inference of the ferroelectricity of BTO, because the Raman bands at 305 and 717 cm−1 correspond to the spontaneous polarization (Ps) [34] of incoming light. The Raman bands for A1 (2TO), A1 (3TO), and E (4LO) phonon modes of the BTO film grown at a low oxygen partial pressure (10 mTorr) are misaligned with the peaks of the BTO target, and no sharp peak at 305 cm−1 (E (3TO + 2LO) + B1) is observed. Comparably, the 305 cm−1 peak becomes stronger with increasing oxygen partial pressure, indicating that the oxygen-rich condition enhances the ferroelectric phase at room temperature. Even though the high oxygen partial pressure during the deposition facilitates the enhancement of ferroelectricity according to the Raman spectra, the surface roughness also remarkably increases, as shown in Fig. 2(b). When the oxygen partial pressure is low (10 mTorr), the surface is ultra-smooth and no distinguishable feature is observed on the surface. However, the films grown under high oxygen partial pressure are not suitable for on-chip photonic integration because cracks are formed on the surface, leading to severe optical loss.
Instead of enhancing the ferroelectricity of BTO films by depositing at high oxygen partial pressure, post-annealing was adopted to control the ferroelectricity of the BTO films. To achieve crack-free BTO films, the BTO films for post-annealing process were deposited at low oxygen partial pressures of 10 mTorr. The ultra-smooth BTO films grown at low oxygen partial pressure were annealed in air at 750, 900, 1050, and 1150°C, which are higher than the deposition temperature. Fig. 2(c) shows the Raman spectra of the BTO target and BTO thin films annealed at different temperatures. Similar to the increase in the oxygen partial pressure during the deposition, the Raman bands of the films gradually align with the peaks of the BTO target, and the 305 cm−1 peak becomes stronger with increasing annealing temperature. The variation in the surface morphology due to the annealing was also monitored, as shown in Fig. 2(d). With the increasing annealing temperature, rectangular sharp features were formed and randomly distributed on the surface; needle-like structures were observed on the surface of the BTO films annealed at 1150°C [35]. Nevertheless, films do not suffer from cracking when the ferroelectricity of the films is enhanced by annealing.
Figure 3(a) shows the out-of-plane XRD pattern displaying the change in the crystal structure by post-annealing. The diffraction patterns were analyzed using the reference pattern of the tetragonal BaTiO3 (JCPDS no. 04-014-0448). As the annealing temperature increased, the intensity of the XRD peaks from the (001) plane increased, and some additional peaks were observed for the samples annealed at 1000 and 1150°C. On closer inspection of the (002) peak as shown in Figure 3(b), the (002) peak of the as-deposited film is located near the MgO (002) peak, and the (002) peak of the annealed films shifted from 43.6 to 45.0° with the increase in the annealing temperature. This indicates that the c-lattice parameter of the annealed film gradually decreases from 4.124 to 4.010 Å with increasing annealing temperature. Furthermore, it is clearly observed that the peaks are broaden as consisting two components which are located between (200) and (002) lattice spacings for bulk BTO. Several literatures [9, 36] have quantitatively explained that the splitting of XRD peaks can attribute to the poly-domain consisting of c-axis (002) and a-axis (200) oriented domain of crystal structure. As shown in Figure 3(b), two components were fitted the out-of-plane diffraction peak: the first one corresponding to a-axis BTO and the second one to c-axis BTO. The change of two components in XRD spectrum after annealing step demonstrate the transformation of crystal orientation from the c-axis to the a-axis oriented tetragonal structure. Although the splitting of peaks in out-of-plane XRD can sufficiently explain the transition of crystal domain of BTO film via post-annealing, it is necessary to characterize the variation of lattice strain to confirm the transformation of crystal domain via post-annealing.
For further analysis of the change of crystal domain, RSM measurements were employed as shown in Fig. 4. RSM images of as-grown and annealed BTO films provide a clearer explanation of the recrystallization of the BTO films. The RSM of the as-deposited BTO films shows that the diffraction peak position of the BTO film is located away from the relaxed line (red dashed line), and this peak migrates along the vertical line toward the red dashed line with increasing annealing temperature. This result indicates that the post-annealing affects to the change of out-of-plane lattice constant and the in-plane lattice constant is preserved during post-annealing. In other words, the as-deposited BTO film at lower oxygen partial pressure is strongly strained along the c-axis orientation and the post-annealing step helps to relax the out-of-plane strains with minor impacts on the strains along the in-plane orientation. Therefore, the transformation of crystal domain is mainly attributed to the strain relaxation in c-axis orientation.
Before we measure the ferroelectric characteristic of BTO films by C-V measurement, CMP was performed to polish the textured surface of the annealed BTO film. Before and after CMP, the surface morphology and RMS roughness of the BTO thin films were investigated by AFM. The obtained RMS values of as-deposited BTO thin films were 0.24 (before CMP) and 0.12 nm (after CMP). Even if the RMS values of the as-deposited BTO films were low, post-annealing was essential to achieve strong ferroelectricity in the in-plane direction. As shown in Fig. 5(a), a textured surface was observable on the annealed BTO thin film, and its RMS value increased significantly up to 0.71 nm. The RMS values of the BTO films annealed at 900 and 1000°C are maintained at 0.34 and 0.23 nm after the CMP process(RMS roughness before CMP process: 1.4 nm and 0.71 nm, respectively). To further investigate the surface uniformity and overall textures of the films, cross-sectional SEM analyses of the polished BTO films were performed for each annealed film (Fig. 5(b)). The BTO thin films annealed at 900 and 1000°C showed a smooth, uniform, and texture-free surface after the CMP, as shown in Fig. 5(b), whereas the textured surface with sharp nanoscale features of the BTO film annealed at 1150°C were not accurately eliminated by the CMP. A further intensive CMP, such as a two-step method using different sizes of colloidal-silica for coarse and fine polishing, may be suitable to clearly eliminate nanoscale features. To minimize the thickness effect of ferroelectricity in C-V characterization between the BTO films after CMP, two-step CMP does not performed in this experiment. Additionally, the XRD spectra obtained after CMP (Fig. 5(c)) demonstrates that additional XRD peaks induced by the annealing originate from surface micro-texturing because all extra peaks disappear after the polishing.
For C–V measurements, electrodes were patterned on the film surfaces as shown in Fig. 6(a). The 2-µm-spaced channel between two electrodes is designed to align with BTO crystalline axis. The pattern of electrode consists of four channels with the angle of 0, 15, 30, and 45 ° with the respect to the [100] direction of BTO film. To determine the crystalline axis of BTO film, we performed Raman measurement while rotating polarization of incident light as shown in Fig. 6(b). The minimum intensity of Raman spectra was measured when an electric field of incident light is aligned along the [100] direction of MgO substrate.
Figure 6(c) shows the C–V characteristics of the BTO thin films annealed up to 1150°C, in which the electrodes are aligned along the MgO [100] direction at a frequency of 10 kHz. The BTO film without annealing showed no signs of ferroelectric characteristics. When a low DC field was applied to ferroelectric films, the dielectric constant was affected by the polarization increase owing to ferroelectric domain switching. Otherwise, most of the domains were already aligned by a high field, and thus, the dielectric constant was small [37, 38]. The BTO films annealed at temperatures greater than 900°C exhibited clear butterfly loops. The samples annealed at higher temperatures exhibited larger capacitances and lower coercive voltages, indicating enhanced crystallinity at higher annealing temperatures as observed in XRD spectrum (Fig. 3). In addition to the C–V characteristics of the annealed films, the angular dependence of ferroelectricity due to the tensorial nature of electro-optic effects [9] were measured, as shown in Figure 6(d). To confirm the angular dependence, the electrodes were patterned and aligned at angles of 0, 15, 30, and 45° on the BTO film annealed at 1150°C, and the maximum capacitance and minimum coercive voltage were measured at an angle of 45° and vice versa at 0°. This result shows the presence of randomly distributed a-axis-oriented tetragonal domains resulting from the post-annealing. The estimated coercive voltage extracted from the sample aligned at 45° was Ec = 7.25 × 105 V/m, which is much smaller than the previously reported coercive voltage of the BTO films deposited by PLD method [39].
To summarize the experimental results, the calculated relative permittivity and lattice constants of each film are listed in Table 1. To examine the dielectric properties in the in-plane direction, the relative permittivity of the BTO films was calculated from its capacitance measured with coplanar electrodes [40]. As mentioned earlier, for the C–V curve, the relative permittivity of the annealed films was over 415, which is higher than that of the as-deposited film. Moreover, as the annealing temperature increased from 750 to 1150°C, the relative permittivity of the films also increased from 415 to 1082. This improvement in relative permittivity in the MgO [100] direction indicates that the crystal structure and ferroelectricity in the in-plane direction improved. Additionally, the relative permittivity of the as-deposited film is similar to that of the film reported by PLD; however, the films annealed at high temperatures have a corresponding or higher relative permittivity than the values reported for other BTO films.
Table 1. Comparison of calculated relative permittivity and lattice constants with barium titanate (BTO) thin films annealed at 750, 900, 1000, 1150 °C and the values reported for other BTO.