Enhancing growth rate in homoepitaxial growth of β-Ga₂O₃ with flat surface via hydrochloric acid addition in mist CVD

Gallium oxide (Ga₂O₃) is a wide-bandgap oxide semiconductor, boasting a bandgap of ∼4.9 eV,^1,2 making it a promising material for power device applications. Ga₂O₃ is characterized by five polymorphs: α, β, γ, δ, and κ (ε) phases. Notably, the monoclinic β-Ga₂O₃ stands out with an estimated critical breakdown field of 8 MV/cm, surpassing those of typical wide-bandgap semiconductors, such as 4H–SiC (2.5 MV/cm) and GaN (3.3 MV/cm). This characteristic enables higher breakdown voltages and enhanced power efficiency in devices.^3–5 As the β-phase is the most thermodynamically stable at room temperature and pressure, diverse techniques have been explored to grow single-crystalline bulks from melts. These include the Czochralski (CZ) method,^6–9 the floating zone (FZ) technique,^10–13 the edge-defined film-fed growth (EFG),^14–16 and the vertical Bridgman (VB) method.¹⁷ These methods represent a distinctive feature not commonly observed in other wide-bandgap semiconductors. The growth rate of β-Ga₂O₃ using these melt growth techniques is higher than that of other wide-bandgap semiconductor materials, including SiC and GaN, suggesting a potential for lower manufacturing costs and the feasibility of mass production in the future.

In addition to bulk crystal growth, the development of homoepitaxial growth approaches is crucial for β-Ga₂O₃ device applications. Commercially available β-Ga₂O₃ substrates come in four crystal orientations: (100), (010), (001), and (−201). Homoepitaxial growth methods on these substrates encompass a range of techniques, such as halide vapor phase deposition,^18–20 low-pressure chemical vapor deposition (LPCVD),^21,22 molecular beam epitaxy (MBE),^23–25 mist chemical vapor deposition (CVD),^26,27 metal-organic chemical vapor deposition (MOCVD),^28–31 and pulsed laser deposition.^32,33 Among these, our group has been focusing on developing mist CVD for the deposition of oxide thin films. The use of precursor solutions in mist CVD allows impurities or alloying elements, if soluble, to be easily incorporated as doping or alloy materials. We have successfully demonstrated fluorine (F) doping in α-Ga₂O₃,³⁴ a challenging achievement with other deposition methods. Additionally, our recent work has led to controlled carrier concentrations and electrical conductivity by doping germanium (Ge) in β-Ga₂O₃.³⁵ The mist CVD technique has also facilitated the alloying of α-, β-, and γ-(Al_xGa_1−x)₂O₃^36–38 and α-, κ (ε)-(In_xGa_1−x)₂O₃^36,39 thin films.

At present, Novel Crystal Technology, Inc. is offering 4-in. wafers that feature (001)-oriented β-Ga₂O₃ substrates commercially. Numerous reports have highlighted the application of (001) β-Ga₂O₃ substrates in power devices. A variety of power devices, including Schottky barrier diodes (SBDs),^40,41 MOS transistors (MOSFETs),⁴² FinFETs,^43,44 and current aperture vertical electron transistors (CAVETs),⁴⁵ have been successfully fabricated and demonstrated. A key factor in enhancing the performance of these devices is the improvement of surface and interface planarity. For example, in SBDs, the surface morphology, particularly the presence of surface defects, significantly influences the contact quality. Therefore, achieving a flat surface with a low surface defect density is essential for optimal Schottky contact. Step bunching leading to surface roughness is a major challenge in the homoepitaxial growth of (001) β-Ga₂O₃.³¹ While hydride vapor phase epitaxy (HVPE) has reported high growth rates of 5 µm/h, the surface exhibited roughness visible under optical microscopy.¹⁸ The MOCVD study has reported using In as a surfactant to suppress step bunching.³¹ LPCVD has achieved growth rates of 1.2 µm/h with relatively high flatness, although step bunching was still observed.⁴⁶ While the homoepitaxial growth of (001) β-Ga₂O₃ using mist CVD has not been previously reported, studies on other thin films have shown that improving the surface flatness of thin films generally involves diluting the precursor concentrations in an aqueous solution,^47,48 which results in an enhanced oxygen-to-metal ratio. However, diluting the concentration leads to a decrease in the growth rate due to the reduced supply of metal precursor. In MOCVD, the addition of HCl has been reported to control the crystal polymorphs of Ga₂O₃. Mist CVD studies have indicated that the addition of HCl improves the crystallinity, surface roughness, and growth rate of α-Ga₂O₃. However, there is limited reporting on the addition of HCl for the homoepitaxial growth of β-Ga₂O₃ in CVD. Investigating the influence of HCl addition on the homoepitaxial growth of β-Ga₂O₃ through mist CVD is crucial and represents a significant contribution to the field of Ga₂O₃ growth. In this study, we successfully demonstrated the homoepitaxial growth of (001) β-Ga₂O₃ using mist CVD. Furthermore, to improve the growth rate at higher growth temperature, we proposed adding hydrochloric acid to the precursor solution. HCl in hydrochloric acid has been reported to significantly influence Ga₂O₃ growth in CVD processes. This approach allowed us to form flat-surfaced (001) β-Ga₂O₃ thin films at lower precursor concentrations, while simultaneously increasing the growth rate.

We utilized mist CVD for the homoepitaxial growth of β-Ga₂O₃ thin films on Sn-doped (001) β-Ga₂O₃ substrates provided by Novel Crystal Technology, Inc. The mist CVD setup consists of a precursor supplying unit and a growth unit. In the precursor supplying unit, a solution containing the precursor is atomized by 2.4 MHz ultrasonic transducers, creating a mist that floats in an atomizer bottle. The growth unit comprises a heated quartz tube where the substrate is placed. The floating mist is transferred by carrier gas into the quartz tube, where a thin film forms on the heated substrate through a chemical reaction. For the mist CVD process, we prepared a GaCl₃ aqueous solution as the precursor, with concentrations ranging from 0.2 to 0.5M. To investigate the impact of HCl, we added hydrochloric acid (35%–37%) to the starting solution at ratios of 4.76 and 9.09 vol. %. N₂ was utilized as the carrier gas to transport the mist at a flow rate of 5 l/min. In our experiments, GaCl₃ was dissolved in deionized water (H₂O) and N₂ was used as the carrier gas. Therefore, H₂O is the only oxygen source in the growth of the β-Ga₂O₃ thin film. The growth temperatures were set at 800 and 900 °C, and the growth duration was 2 h. The crystalline structures of the homoepitaxial β-Ga₂O₃ thin films were analyzed using x-ray diffraction (XRD, Bruker D8 Discover) 2θ–ω scans. Film thickness measurements were conducted with a stylus surface profiler (Bruker Dektak XT-3). A mask was applied to the substrate, and the stylus was scanned across the gap between the non-deposited and deposited areas. Atomic force microscopy (AFM, SII NanoNavi) was utilized to examine the surface morphology.

First, we examine the impact of solution conditions, including precursor concentration and the addition of HCl, on the growth rate in mist CVD. Solution conditions are significant parameters affecting thin film growth in this process. Figure 1(a) represents the dependency of β-Ga₂O₃ homoepitaxial thin film thickness on solution concentration at different growth temperatures: 800 and 900 °C. An increase in solution concentration led to thicker, thin films. This can be attributed to the nearly constant mist supply, which, despite varying concentrations, results in a higher quantity of precursor at increased concentrations. Regarding growth temperature, an increase in temperature caused a decrease in thin film thickness. This phenomenon can be explained by the dynamics of the CVD process at high temperatures. In high-temperature CVD, rapid chemical reactions contribute to the loss of mist precursors due to film deposition on the walls of the quartz tube. Additionally, enhanced desorption of adatoms from the substrate surface leads to a decrease in the number of atoms incorporated into the substrate. Figure 1(b) shows the impact of added hydrochloric acid concentration in the solution on the thickness of β-Ga₂O₃ homoepitaxial thin films grown at 900 °C using 0.2M GaCl₃ aqueous solution. The film thickness increased linearly with the volume fraction of hydrochloric acid in the solution. During deposition at 900 °C, the growth rate increased by ∼4.8 times when the amount of hydrochloric acid added to the precursor solution was increased to 9.09 vol. %. In mist CVD, the addition of HCl derived from hydrochloric acid has been reported to increase the growth rate of α-Ga₂O₃.^49,50 Our results for (001) β-Ga₂O₃ align with these findings. Furthermore, in MOCVD for Ga₂O₃, it has been reported that the addition of HCl enhances the growth rate.⁵¹ This suggests that the presence of HCl in the vapor phase reaction significantly impacts the growth of Ga₂O₃ in the CVD process. Although our experiments did not involve excessive hydrochloric acid addition, it is worth noting that excessive HCl addition has been reported to reduce the growth rate in both mist CVD⁵⁰ and MOCVD.⁵¹ We hypothesize that in our experiments, an excessive addition of hydrochloric acid would likely lead to saturation or a decrease in the growth rate, rather than a continuous increase.

To investigate the crystal structure of β-Ga₂O₃ thin films on (001) β-Ga₂O₃ substrates using mist CVD, XRD 2θ–ω scans were performed. Figures 2(a)–2(c) display XRD 2θ–ω scans ranging from 10° to 50° for β-Ga₂O₃ homoepitaxial thin films grown with GaCl₃ precursor solution concentrations of 0.2, 0.35, and 0.5M at growth temperatures of 800 and 900 °C. Figure 2(d) represents XRD 2θ–ω scans over the same range for β-Ga₂O₃ homoepitaxial thin films grown at 900 °C, with the addition of 4.76 and 9.09 vol. % hydrochloric acid to a 0.2M concentration of the GaCl₃ precursor solution. The patterns of (00l) homoepitaxial β-Ga₂O₃ were clearly observed in all cases, irrespective of the solution concentration, growth temperature, and the amount of HCl added to the precursor, without any other diffraction peaks derived from different planes or polymorphs. This confirms the successful growth of homoepitaxial (001) β-Ga₂O₃ thin films by mist CVD. Furthermore, the XRD 2θ–ω scan of the reference bare (001) β-Ga₂O₃ substrate is presented in Fig. 1(e). Figure 2(f) shows a comparison of the XRD 2θ–ω scans in the range of 30.5°–33° for the bare β-Ga₂O₃ substrate and the homoepitaxial film grown under optimal conditions (0.2M precursor concentration with 9.09 vol. % HCl addition). As evident from Fig. 2(f), the peaks of the substrate and the homoepitaxial film almost perfectly overlap, with no observable peak shifting or broadening of the FWHM. This indicates that the addition of HCl does not significantly alter the lattice parameters or induce deterioration of the crystalline quality. While there appears to be a slight broadening at the base of the peaks for the bare substrate compared to the homoepitaxial film, this difference is minimal considering the low-intensity regions being examined. Additionally, Figs. 2(g) and 2(h) depict the rocking curve FWHM when varying the growth temperature and hydrochloric acid concentration. As shown in Figs. 2(g) and 2(h), all values are approximately in the range of 60–80 arcsec, nearly equivalent to the FWHM from the substrates. This suggests that the crystallinity, which can be evaluated from the rocking curve, largely remains unchanged.

Finally, we discuss the surface morphologies of homoepitaxial β-Ga₂O₃ thin films as measured by AFM. An atomically flat surface is crucial in these films, as it impacts factors such as the reduction of surface defects and the formation of heterostructures. Figure 3(i) shows an AFM image of the bare (001) β-Ga₂O₃ substrate, which exhibits a smooth step-terrace morphology. Figures 3(a)–3(f) display AFM images of homoepitaxial β-Ga₂O₃ thin films grown at different solution concentrations and growth temperatures. The β-Ga₂O₃ thin films grown at 800 °C exhibited a morphology where step bunching became increasingly pronounced as the solution concentration was increased from 0.2 to 0.5M. At 800 °C, the root-mean-square (rms) roughness, calculated from AFM images, increased with higher concentrations due to the more pronounced effects of step bunching. In contrast, at a growth temperature of 900 °C, the solution concentration had almost no effect. Multi-step arrays, resembling stripe-like surface morphologies with atomically flat surfaces and suppressed step bunching, were observed at all solution concentrations. This suggests that growth temperature significantly influences the surface morphology of the thin films. Factors that might contribute to the appearance of step bunching on the surface of the grown β-Ga₂O₃ thin films include the growth rate and the Ga concentration in the precursor solution under our experimental conditions. When growth conditions are adjusted to increase the growth rate of β-Ga₂O₃ thin films, step bunching tends to become more pronounced.^21,52 In homoepitaxial growth using (010) β-Ga₂O₃ substrates by the LPCVD method, surface morphology exhibited step bunching and higher growth rates at lower temperatures, across a range of growth temperatures from 780 to 950 °C.²¹ In our experiments, GaCl₃ was dissolved in deionized water (H₂O), and N₂ was used as the carrier gas. As a result, the sole source of oxygen in the growth of β-Ga₂O₃ thin films was H₂O. The ratio of O to Ga, corresponding to the IV/III ratio, could be varied by adjusting the ratio of GaCl₃ to H₂O. For instance, a 0.5M precursor solution is 2.5 times more concentrated than a 0.2M solution, resulting in a lower ratio of O to Ga in the 0.5M solution. Therefore, a decrease in oxygen within the reaction species and a decrease in the IV/III ratio could be one potential cause of the observed step bunching. Consideration of the improvement of surface flatness at high temperatures shows that the mobility of the adatom increases with increasing temperature, while the residence time of the adatom on the surface decreases. This results in a reduced adatom diffusion length, allowing adatoms to be readily incorporated at step edges, maintaining a uniform step spacing and leading to a smoother surface morphology. Conversely, at lower temperatures, the longer residence times lead to longer diffusion lengths, increasing the likelihood of adatoms being incorporated at unintended steps or islands. This non-uniform step propagation results in step bunching and a rougher surface. We consider that the above mechanisms are at work in this experiment as well. Figures 3(g) and 3(h) display AFM images of homoepitaxial β-Ga₂O₃ thin films grown at 900 °C with the addition of hydrochloric acid to the precursor solution. A striped morphology was observed in β-Ga₂O₃ thin films grown with added HCl. The rms roughness of all homoepitaxial films remained nearly the same, irrespective of the amount of hydrochloric acid added. As shown in Fig. 1(b), the growth rate increased with the addition of HCl, while the surface remained flat. Comparing the homoepitaxial β-Ga₂O₃ thin film grown at 800 °C with a 0.5M GaCl₃ precursor solution [Fig. 3(c)] to that grown at 900 °C with a 0.2M GaCl₃ precursor containing 9.09 vol. % hydrochloric acid [Fig. 3(h)], the film prepared with a high-concentration HCl-containing precursor exhibited superior surface flatness. The rms roughness improved by ∼1/7, despite the films having nearly identical thicknesses. We believe that this improvement is attributed to the growth temperature. When grown at 800 °C, although the growth rate exhibited high, the lower growth temperature led to a reduction in the surface diffusion rate of adatoms, resulting in the occurrence of step bunching. On the other hand, when grown at 900 °C, the surface diffusion rate of adatoms increased, preventing the step bunching. Despite the lower growth rate at 900 °C, the addition of HCl enhanced the growth rate, leading to a flat surface at the same film thickness. Similar enhancement in growth rates of Ga₂O₃ with the addition of HCl in mist CVD has been reposted, aligning with our findings. Considering the results from MOCVD, HVPE, and other mist CVD processes, HCl appears to play a crucial role in the CVD process, particularly in those including Cl. We believe that our findings significantly contribute to the crystal growth and device application of Ga₂O₃.

In this study, we demonstrated the growth of homoepitaxial thin films on (001) β-Ga₂O₃ substrates with high growth rates and flat surfaces at high temperatures by adding HCl to the precursor solution using the mist CVD method. Increasing the solution concentration at 800 and 900 °C results in thicker films, attributed to a constant mist supply. At 900 °C, the addition of HCl linearly increases film thickness, leading to a 4.8-fold rise in the growth rate at 9.09 vol. %. This suggests that HCl significantly influences Ga₂O₃ growth in the CVD process, with the potential for saturation or decrease in growth rates with excessive HCl. The XRD 2θ–ω (00l) homoepitaxial patterns of β-Ga₂O₃ were consistently visible, regardless of solution concentration, growth temperature, or the amount of HCl added to the precursor. This confirms the successful growth of homoepitaxial (001) β-Ga₂O₃ thin films by mist CVD. The AFM analysis of homoepitaxial β-Ga₂O₃ thin films highlights the significance of an atomically flat surface. While films grown at 800 °C exhibited pronounced step bunching with increasing precursor concentration, those at 900 °C showed multi-step arrays, indicating the substantial influence of growth temperature on surface morphology. Despite increased growth rates with higher HCl concentrations, the surface was plane. The superior surface flatness achieved with a high concentration of HCl-containing precursor and at higher growth temperature indicates the positive impact of HCl and a higher growth temperature in minimizing step bunching with an increased growth rate during the mist CVD growth process. The involvement of HCl, which includes chlorine, is deemed vital in the CVD process. While we have successfully improved growth rates through HCl addition, further enhancements are necessary. For instance, MOCVD studies have used In as a surfactant to suppress step bunching and improve surface flatness. Given that In can also be added in mist-CVD, there is potential to maintain flatness at higher growth rates by utilizing In as a surfactant. We believe that our HCl addition technique for improving growth rates will significantly contribute to achieving high-speed growth in such future investigations.

This work was supported by the JST FOREST Program (Grant No. JPMJFR222M, Japan) and by JSPS KAKENHI under Grant No. JP23K22797.