High-Quality (0001) α-Ga₂O₃ Film Grown by Mist Chemical Vapor Deposition on (0001) α-Cr₂O₃ Template

Abstract

A (0001) α-Ga₂O₃ film was grown by the mist chemical vapor deposition method on a (0001) α-Cr₂O₃ template (100 μm thick α-Cr₂O₃ layer formed on an α-Al₂O₃ substrate). Benefiting from the small a-axis lattice mismatch between α-Ga₂O₃ and α-Cr₂O₃, a high-quality α-Ga₂O₃ film with a small twist distribution, and consequently a low edge dislocation density, was coherently grown on an α-Cr₂O₃ template. The edge dislocation density of 7 × 10⁷ cm⁻², estimated from the full-width at half-maximum value of the X-ray rocking curve (XRC) in X-ray diffraction (XRD), was more than two orders of magnitude lower than that of the film grown on an α-Al₂O₃ substrate, and was almost consistent with that of the α-Cr₂O₃ template. The bright-field transmission electron microscopy (TEM) image supports the dislocation density estimated from the XRD measurements. The high-angle annular dark-field scanning TEM and inverse fast Fourier transform images indicate coherent growth, with almost no misfit dislocations generated at the α-Ga₂O₃/α-Cr₂O₃ interface.

1. Introduction

Ga₂O₃ is a promising wide-band-gap material for use in post-Si power devices. Ga₂O₃ is known to have five crystal structures: α, β, δ, ε (κ) and γ [1]. Monoclinic (β-gallia)-structured β-Ga₂O₃, with a band gap of 4.5–4.9 eV [2,3,4,5], is widely regarded as the most promising among the five polymorphs for power-device applications, as it is the most thermally stable phase [6], and bulk single crystals have been successfully grown by the floating zone [7,8], edge-defined film-fed growth [9,10], Czochralski [11], and vertical Bridgman [12] methods. Homoepitaxial growth has also been studied on a single-crystal β-Ga₂O₃ substrate [13,14,15,16].

Corundum-structured α-Ga₂O₃ has been reported to have the widest band gap (5.3–5.6 eV) among the five polymorphs [17,18]. Owing to its high Baliga’s figure of merit, along with its expected high dielectric breakdown strength and low on-resistance, α-Ga₂O₃ is attracting increasing attention as a potentially promising material, particularly when compared with β-Ga₂O₃. On the other hand, α-Ga₂O₃ is a thermally metastable phase, and bulk single crystals have not yet been fabricated. Therefore, α-Ga₂O₃ films are typically grown heteroepitaxially on α-Al₂O₃ substrates with the same crystal structure [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]. Single-crystalline (0001) α-Ga₂O₃ films have been reproducibly grown by the mist chemical vapor deposition (mist CVD) [17,18,19,20,22,23,24,26,27,31,32] and halide vapor-phase epitaxy [21,25,28,29,30] methods on (0001) α-Al₂O₃ substrates. Thanks to the establishment of stable growth, extensive research has recently been conducted in this field, leading to review papers that comprehensively summarize developments in growth, material properties, and device applications [33,34,35,36,37]. However, the fundamental problem of generating a high density of dislocations during (0001) α-Ga₂O₃ growth on (0001) α-Al₂O₃ substrates has not yet been resolved, owing to the large a-axis lattice mismatch of 4.8% between α-Ga₂O₃ and α-Al₂O₃ [19]. A high density of dislocations (in the order of 10¹⁰ cm⁻²) has been widely observed in α-Ga₂O₃ films with thicknesses of several hundreds of nanometers [19,22,24,31,32]. A reduction in the dislocation density is required for device applications.

The dislocation density can be reduced by increasing the film thickness [24,30,31]. Oshima et al. reported a low dislocation density of 1.5 × 10⁸ cm⁻² in a 200 μm thick α-Ga₂O₃ film [30]. Epitaxial lateral overgrowth (ELO) is another effective technique for reducing the dislocation density [25,29,30,38]. However, it requires an additional mask-fabrication process, and non-uniformity between the window and mask regions remains an issue. The insertion of a buffer layer also offers an effective technique for reducing dislocation density [22,23,27,32]. This approach can enable the growth of uniform, high-quality films without relying on very thick layers or complex processes such as ELO. Jinno et al. succeeded in decreasing the dislocation density from 7 × 10¹⁰ cm⁻² to 9 × 10⁸ cm⁻² in a 150 nm thick α-Ga₂O₃ film by inserting quasi-graded α-(Al,Ga)₂O₃ buffer layers [22]. As an alternative material for the buffer layer, we have focused on α-Cr₂O₃. α-Cr₂O₃ has the same corundum structure as α-Ga₂O₃ and exhibits a small a-axis lattice mismatch of 0.4% [39], which is much smaller than that for α-Al₂O₃. Stepanov et al. reported the growth of α-Ga₂O₃ on an α-Cr₂O₃ buffer layer with a thickness of 150 nm [28]. By using the α-Cr₂O₃ buffer layer, they suppressed the inclusion of the ε-Ga₂O₃ phase into the film and reduced the dislocation density from 2 × 10¹⁰ to 5 × 10⁹ cm⁻² [28]. Growth on a high-quality α-Cr₂O₃ layer is expected to further improve the crystallinity of the α-Ga₂O₃ film.

In this study, we report the growth of a high-quality (0001) α-Ga₂O₃ film with a small twist distribution and low edge-dislocation density by the mist CVD method on a (0001) α-Cr₂O₃ template (100 μm thick α-Cr₂O₃ layer formed on an α-Al₂O₃ substrate), fabricated by NGK INSULATORS, LTD. The a-axis and c-axis lattice constants of the α-Cr₂O₃ template used in this study were 4.949 and 13.590 Å, respectively. The estimated screw and edge dislocation densities of the α-Cr₂O₃ layer were 3 × 10⁶ and 7 × 10⁷ cm⁻², respectively, as is discussed later. On this α-Cr₂O₃ template, we successfully grew a coherently strained α-Ga₂O₃ film with a low dislocation density, significantly lower than previously reported values [19,22,24,27,28,30,31,32] and comparable to that of the template. These results were compared with those obtained for an α-Ga₂O₃ film grown on a (0001) α-Al₂O₃ substrate.

2. Experimental Procedure

An α-Ga₂O₃ film was grown on either a (0001) α-Cr₂O₃ template (7 × 7 mm²) or a (0001) α-Al₂O₃ substrate (10 × 10 mm²) using a mist CVD system, which is schematically shown in ref. [17]. The source solution was prepared using 0.05 mol·L⁻¹ of Gallium (III) acetylacetonate (Ga(C₅H₇O₂)₃), 0.09 mol·L⁻¹ of hydrochloric acid (HCl), and deionized water. The source solution was atomized using an ultrasonic transducer at 2.4 MHz, and the formed aerosols were transferred to a substrate set in a quartz tube using oxygen carrier gas at a flow rate of 3.5 L·min⁻¹. Before growth, the α-Cr₂O₃ template and α-Al₂O₃ substrate were sequentially cleaned ultrasonically in acetone (C₃H₆O), isopropyl alcohol (C₃H₈O), and deionized water for 5 min each. α-Ga₂O₃ films with thicknesses of 600–700 nm were grown at 460 °C for 1 h. Structural evaluations were performed using a tabletop scanning electron microscope (SEM; Miniscope TM3000, Hitachi High-Tech, Tokyo, Japan), X-ray diffraction (XRD; X’Pert PRO, Malvern PANalytical, Malvern, United Kingdom), and a transmission electron microscope (TEM; JEM-ARM300F2, JEOL, Tokyo, Japan). TEM lamellae were prepared using the focused ion beam (FIB) method (JIB-4000, JEOL, and Helios Nanolab 660, FEI, Waltham, MA, USA).

3. Results and Discussion

Figure 1 shows SEM images of Ga₂O₃ films grown on an α-Cr₂O₃ template and an α-Al₂O₃ substrate. In both cases, the film surfaces are relatively smooth and flat, indicating uniform film growth. Notably, the grain size of the Ga₂O₃ film grown on the α-Cr₂O₃ template is larger than that of the film grown on the α-Al₂O₃ substrate.

Figure 2 shows the results of the XRD θ-2θ measurements comparing Ga₂O₃ films grown on an α-Cr₂O₃ template and α-Al₂O₃ substrate. In addition to the peaks originating from the template or substrate, the main peak for (0006) α-Ga₂O₃ was observed in the XRD patterns of both samples. Although the samples also exhibited a smaller (0004) ε-Ga₂O₃ peak, its intensity was significantly suppressed in the sample grown on the α-Cr₂O₃ template. This result is in agreement with that of a previous report [28]. Furthermore, in the wider-range θ-2θ XRD scan (2θ = 10–70°), no additional diffraction peaks other than those described above were observed. As can be seen in Figure 2, the (0006) peak of α-Ga₂O₃ film grown on the α-Cr₂O₃ template shifted to a lower angle relative to that of the film grown on the α-Al₂O₃ substrate, whose peak position is close to the theoretical value. Using the asymmetric XRD measurements, the a-axis and c-axis lattice constants of α-Ga₂O₃ grown on the α-Cr₂O₃ template were found to be 4.948 Å and 13.476 Å, while the a-axis and c-axis lattice constants of the α-Ga₂O₃ grown on an α-Al₂O₃ substrate were 4.993 Å and 13.431 Å. In the sample grown on the α-Cr₂O₃ template, the a-axis lattice constant was smaller than that of α-Ga₂O₃ grown on an α-Al₂O₃ substrate and was nearly identical to the a-axis lattice constant of the α-Cr₂O₃ template itself, as presented in the Introduction. By contrast, the c-axis lattice constant was larger than that of α-Ga₂O₃ grown on an α-Al₂O₃ substrate. The latter is close to the theoretical value [40]. This clearly indicates that the α-Ga₂O₃ film grew coherently under strain on the α-Cr₂O₃ template. It should also be noted that the sample grown on the α-Cr₂O₃ template exhibited a slight shoulder peak at a higher angle, indicating the presence of a small amount of relaxed α-Ga₂O₃.

Figure 3 shows the (0006) and (10$\overline{1}$4) X-ray rocking-curve (XRC) profiles of α-Ga₂O₃ films grown on an α-Cr₂O₃ template and an α-Al₂O₃ substrate. Enlarged views of the (10$\overline{1}$4) XRC profiles are shown in the inset. The red and blue curves correspond to the α-Ga₂O₃ films grown on the α-Cr₂O₃ template and the α-Al₂O₃ substrate, respectively. For the (0006) symmetric reflection, the rocking-curve width of the film grown on the α-Cr₂O₃ template is slightly larger than that of the film grown on the α-Al₂O₃ substrate. In contrast, for the (10$\overline{1}$4) asymmetric reflection, the rocking-curve width is clearly narrower for the film grown on the α-Cr₂O₃ template.

Table 1. Measured (0006) and (10$\overline{1}$4) XRC FWHM values and estimated dislocation densities of α-Ga₂O₃ films grown on an α-Cr₂O₃ template and α-Al₂O₃ substrate. The dislocation density was calculated using Equations (1) and (2) using extrapolated (10$\overline{1}$0) XRC FWHM derived from the FWHM values of multiple diffraction planes. Values in parentheses correspond to those of the α-Cr₂O₃ template and are shown as reference values.

	XRC FWHM (Arcsec.)		Estimated Dislocation Density (cm−2)
	0006	10$\overline{1}$4	Screw	Edge
α-Ga2O3 film on α-Cr2O3 template	105	135	3 × 106	7 × 107
	(95)	(117)	(3 × 106)	(6 × 107)
α-Ga2O3 film on α-Al2O3 substrate	51	3016	8 × 105	3 × 1010

shows the (0006) and (10$\overline{1}$4) X-ray rocking-curve (XRC) full-width at half-maximum (FWHM) values of the α-Ga₂O₃ films grown on the α-Cr₂O₃ template and α-Al₂O₃ substrate. The values shown in parentheses are for (0006) and (10$\overline{1}$4) XRC FWHM values of the α-Cr₂O₃ template itself. As summarized in Table 1, the (0006) XRC FWHM values are 105 and 51 arcsec for the α-Ga₂O₃ films grown on the α-Cr₂O₃ template and the α-Al₂O₃ substrate, respectively. The corresponding (10$\overline{1}$4) XRC FWHM values are 135 and 3016 arcsec, respectively. The small twist distribution, reflected in the narrow (10$\overline{1}$4) XRC FWHM value, is attributed to the small lattice mismatch between α-Ga₂O₃ and α-Cr₂O₃. Furthermore, the XRC FWHM values of the α-Ga₂O₃ film on the α-Cr₂O₃ template are close to those of the α-Cr₂O₃ template itself.

The table also shows the screw dislocation density (${\mathit{\rho }}_{\mathit{screw}}$) and edge dislocation density (${\mathit{\rho }}_{\mathit{edge}}$) estimated from the XRC FWHM values using the following equation [24,27,31]:

$${\mathit{\rho }}_{\mathit{screw}/\mathit{edge}}={\displaystyle \frac{{{\mathit{\alpha }}^2}_{\mathit{tilt}/\mathit{twist}}}{2\mathit{\pi }·\mathit{ln}2·{\mathit{b}}^2}}$$

(1)

where $\mathit{b}$ is the length of each Burgers vector and α is the angle of the tilt/twist distribution. The Burgers vector $b={\displaystyle \frac{1}{3}}$<2$\overline{1}\overline{1}$0> was used for calculating tilt distribution, whereas $b=$<0001> was used for twist dislocations. The edge dislocation density was evaluated from the estimated FWHM of the (10$\overline{1}$0) reflection, which was obtained by extrapolation using Equation (2) from the FWHM values measured for multiple diffraction planes, including the (0001) and (10$\overline{1}$4) reflections [31].

$${\mathit{\beta }}_{h\mathit{kil}}^2={\left({\mathit{\beta }}_{\mathit{tilt}}\mathit{cos}\mathit{\chi }\right)}^2+{\left({\mathit{\beta }}_{\mathit{twist}}\mathit{sin}\mathit{\chi }\right)}^2.$$

(2)

Here, $\mathit{\beta }$ denotes the FWHM of each reflection, and $\mathsf{\chi }$ is the angle between the (0001) plane and the corresponding diffraction plane. The film grown on the α-Cr₂O₃ template had an estimated screw dislocation density of 3 × 10⁶ cm⁻² and edge dislocation density of 7 × 10⁷ cm⁻², while the sample grown on the α-Al₂O₃ substrate had an estimated screw dislocation density of 8 × 10⁵ cm⁻² and edge dislocation density of 3 × 10¹⁰ cm⁻². The estimated dislocation densities in the sample grown on the α-Al₂O₃ substrate were in agreement with those reported previously [19,22,24,27,31,32], whereas the estimated edge dislocation density of the α-Ga₂O₃ film grown on the α-Cr₂O₃ template was much smaller (i.e., by a factor of ~500). The estimated screw and edge dislocation densities of the α-Ga₂O₃ film grown on the α-Cr₂O₃ template were nearly the same as those of the α-Cr₂O₃ template, as can be seen in Table 1.

Figure 4 shows bright-field TEM images, captured at an acceleration voltage of 300 kV of the samples grown on the α-Cr₂O₃ template and α-Al₂O₃ substrate. As previously reported [19,22,24,27,32], a high density of dislocations in the order of 10¹⁰/cm² was observed in the α-Ga₂O₃ film grown on the α-Al₂O₃ substrate, as shown in Figure 4b. In contrast, dislocations were rarely observed in the α-Ga₂O₃ film grown on the α-Cr₂O₃ templates, as shown in Figure 4a. The lamella used in this measurement was 500 nm thick, and there was a region more than a 2 μm wide in which no dislocations were found. The dislocation density could therefore be estimated to be below 10⁸ cm⁻², which is consistent with the value obtained from the XRD measurements (Table 1). It should be noted that a planar-like defect can be seen on the right side of the image in the sample grown on the α-Cr₂O₃ template. The presence of this defect may be related to the presence of a small amount of relaxed α-Ga₂O₃, as mentioned in Figure 2. The details of this defect, including its possible association with the relaxed phase, are currently under investigation.

Figure 5a shows a high-angle annular dark-field scanning TEM (HAADF-STEM) image of the α-Ga₂O₃/α-Cr₂O₃ interface, and Figure 5b,c present an HAADF-STEM image and an inverse fast Fourier transform (FFT) image showing the $(\overline{3}300)$ plane, respectively, taken from the region indicated by the red square in Figure 5a. The cation lattices were perfectly matched between α-Ga₂O₃ and α-Cr₂O₃, and no variation in the number of atomic planes was observed across the interface. These results indicated that misfit dislocations were not generated at the α-Ga₂O₃/α-Cr₂O₃ interface in this region.

Thus, we succeeded in growing a high-quality α-Ga₂O₃ film with a dislocation density in the order of 10⁷ cm⁻², which is comparable to that of the α-Cr₂O₃ template. The dislocation density obtained in this study was much lower than that previously reported for α-Ga₂O₃ grown on α-Cr₂O₃ [28], indicating that the crystallinity of the α-Ga₂O₃ film strongly depends on that of the underlying α-Cr₂O₃. The realization of a higher-quality α-Cr₂O₃ underlayer, including bulk substrates, is therefore expected to enable further improvements in the α-Ga₂O₃ film quality.

4. Conclusions

In summary, we investigated the growth of α-Ga₂O₃ on the α-Cr₂O₃ template using mist CVD. The α-Ga₂O₃ film was coherently grown on the α-Cr₂O₃ template. Compared with the film grown on the α-Al₂O₃ substrate, the (10$\overline{1}$4) XRC FWHM value was significantly smaller. The screw and edge dislocation densities in the film, estimated from the XRC FWHM values, were 3 × 10⁶ cm⁻² and 7 × 10⁷ cm⁻², respectively. These values were almost the same as those of the α-Cr₂O₃ template used in this study. The HAADF-STEM and FFT images indicate that almost no misfit dislocations were generated at the α-Ga₂O₃/α-Cr₂O₃ interface. Thus, we succeeded in growing high-quality α-Ga₂O₃ films with low dislocation densities on an α-Cr₂O₃ template.