【Knowledge Discover】Epitaxial Commercialization: Technical Breakthroughs and Practical Challenges of HVPE and MOCVD

^{Figure 1. Manufacturing process flow of gallium oxide-based materials and devices [1]}

      Laboratory research typically focuses on parameters such as the growth rate, surface roughness, and X-ray diffraction full width at half maximum (FWHM) of individual samples. However, industrial-scale production requires addressing a broader range of challenges: Can the epitaxial layer maintain uniform thickness and doping concentration across large-area wafers? Can defect density be consistently controlled? Can the process achieve batch-to-batch reproducibility? Can the growth efficiency justify the costs of equipment, precursors, and post-processing? And can the epitaxial wafers be directly integrated into device fabrication processes?

      The competition in β-Ga₂O₃ epitaxy is gradually shifting from breakthroughs in individual performance metrics toward a comprehensive balance among growth rate, material quality, process stability, and manufacturing cost.

      Currently, HVPE and MOCVD are advancing industrialization along two distinct pathways: HVPE focuses on efficiently producing thick drift layers required for high-voltage vertical devices, while MOCVD emphasizes achieving device-grade functional structures through precise control of precursor flow, doping, and interface engineering.

      HVPE typically uses gallium halides such as GaCl as gaseous gallium sources, which react with O₂ on high-temperature substrates to form β-Ga₂O₃. Compared with metal-organic precursors, halide-based precursors feature a relatively direct reaction pathway, lower material costs, and higher mass transport efficiency, making HVPE particularly suitable for the growth of epitaxial layers with thicknesses of tens of micrometers.

      For vertical power devices rated at kilovolt levels or above, the drift layer thickness often needs to reach 10 μm or more. If the growth rate is too low, both the epitaxy time per wafer and equipment utilization costs increase rapidly. Therefore, HVPE has an inherent advantage in thick-film fabrication, with typical growth rates reaching 5–10 μm/h, and optimized processes achieving growth rates of several tens of micrometers per hour ^[2–4].

      However, industrialization does not simply mean pursuing the highest possible growth rate. In conventional GaCl–O₂ systems, increasing precursor partial pressure can enhance the growth rate, but it may also increase the risk of gas-phase parasitic reactions, leading to the formation of polycrystalline β-Ga₂O₃ particles. Once these particles deposit onto the epitaxial surface, they may trigger abnormal growth and locally degrade surface morphology ^[3,5].

      Meanwhile, large-scale epitaxial production still requires further solutions for maintaining within-wafer uniformity in film thickness, carrier concentration, and surface morphology, which remain critical challenges for HVPE industrialization.

      β-Ga₂O₃ exhibits pronounced crystal anisotropy. The atomic arrangement, surface energy, adsorption behavior, and step-flow kinetics vary significantly among different crystal planes. As a result, substrate orientation directly influences the growth rate, surface morphology, and defect evolution of epitaxial layers.

      Currently, the (001)-oriented substrate has a relatively mature foundation in large-area substrate supply and represents an important direction for β-Ga₂O₃ HVPE industrialization. However, epitaxial growth on this orientation is prone to forming stripe-like grooves extending along the [010] direction ^[6]. In addition, some defects are associated with the propagation of substrate dislocations, meaning that further improvements are still required in surface morphology and device yield.

      Meanwhile, epitaxial growth on other substrate orientations, including (-201), (100), and (011) planes, has also achieved continuous progress, providing additional pathways for optimizing β-Ga₂O₃ epitaxial quality and device performance.

Figure 2. NDIC microscopy image of an 8 μm-thick HVPE homoepitaxial layer grown on a (001)-oriented β-Ga₂O₃ substrate [6]

      According to reports, when (011)-oriented β-Ga₂O₃ was grown by HCl-based HVPE, the growth rate reached approximately 14 μm/h, which is about 60% of the growth rate achieved on (001)-oriented substrates under the same system. Meanwhile, compared with earlier Cl₂-based processes, the HCl-based approach improved the growth rate by approximately 5–7 times and significantly reduced the formation of surface polycrystalline particles ^[3].

      The exploration of different substrate orientations is not solely aimed at achieving higher growth rates. More importantly, it seeks to establish the optimal relationship among crystal orientation, defect propagation behavior, and device structures, providing a pathway toward high-quality and scalable β-Ga₂O₃ epitaxial manufacturing.

      During HVPE epitaxial growth, parameters including growth temperature cdeposi.

      Growth t determines the fundathree-dim. In contrast, excessivelysubstrate surface decomposition, interface instability, and increased defect formation. Therefore, temperature optimization requires a careful balance among gas-phase reaction efficiency, surface diffusion, and high-temperature desorption processes.

Figure 3. NDIC surface morphologies of (001)-oriented β-Ga₂O₃ HVPE homoepitaxial layers grown at different temperatures (left: 800 °C; right: 1000 °C) ^[7]

      The VI/III ratio and precursor supply jointly determine the reaction driving force, local supersaturation, and mass transport processes in the GaCl–O₂ system ^[8,9]. Insufficient oxygen supply may prevent complete oxidation of the gallium precursor, resulting in an oxygen-deficient growth environment and increasing the risk of non-stoichiometric defects and related defect formation. Increasing the O₂/GaCl ratio helps maintain sufficient oxidation conditions; however, an excessively high oxidant partial pressure, especially under conditions of high GaCl partial pressure, elevated reaction pressure, or prolonged gas residence time, may enhance gas-phase precursor reactions, thereby affecting the effective transport and utilization efficiency of gallium species toward the substrate.

      On the other hand, increasing the GaCl partial pressure within an appropriate range can enhance the growth rate. However, excessive reactant concentration may promote the formation of gas-phase particles and their transport toward the growth surface, leading to surface defects and morphological degradation. Therefore, the VI/III ratio is not simply a parameter where “higher is better”; its optimal range must be carefully coordinated with growth temperature, pressure, precursor partial pressures, total flow rate, and reactor design.

      Reaction pressure further regulates precursor residence time, mass transport efficiency, and surface growth modes. According to reported studies, during (001)-oriented β-Ga₂O₃ homoepitaxial growth, reducing the pressure from 6.06 × 10⁴ Pa to 1.01 × 10⁴ Pa decreased the growth rate from 10.1 μm/h to 2.1 μm/h. The authors attributed this reduction to the decreased partial pressures of source gases such as GaCl and O₂ under lower pressure conditions, which reduced the effective reactant flux participating in surface reactions per unit time, thereby lowering the epitaxial growth rate.

      However, moderately reducing the pressure can also suppress gas-phase parasitic reactions and particle formation, improving precursor transport toward the substrate surface. This promotes a transition in growth mode from three-dimensional island growth to two-dimensional step-flow growth, resulting in more continuous and refined surface striations. Under a pressure of 2.02 × 10⁴ Pa, the epitaxial layer achieved an X-ray diffraction rocking curve full width at half maximum (FWHM) of 30.2″, approaching that of the substrate (28.9″).

      Therefore, although lowering pressure may sacrifice part of the growth rate due to reduced precursor partial pressure, it can effectively suppress gas-phase pre-reactions and particle defects, while improving the surface morphology and crystalline quality of β-Ga₂O₃ epitaxial layers.

Figure 4. X-ray diffraction (XRD) rocking curves of the (001) planes of β-Ga₂O₃ substrates and epitaxial films grown under different pressures ^[10]

      The gas flow field and reactor structure determine whether the optimized growth conditions can be uniformly achieved across large-area wafers. Even when the central region of a wafer maintains appropriate temperature, VI/III ratio, and pressure conditions, significant variations in boundary layer thickness, precursor concentration, or gas residence time across different regions of the wafer can still lead to non-uniformity in film thickness, surface morphology, and doping concentration.

      Therefore, for industrial-scale β-Ga₂O₃ epitaxy, further optimization of the reactor design, including the distribution of temperature fields, concentration fields, and gas flow fields, is essential to minimize variations between the wafer center and edge, as well as improve batch-to-batch reproducibility.

      Traditional Cl₂-HVPE typically utilizes the reaction between Cl₂ and liquid Ga to generate GaCl, which is then transported to the growth zone and reacts with O₂ to form β-Ga₂O₃ [3,7]. This approach features high reaction activity and relatively low material costs, making it suitable for rapid growth of thick epitaxial layers. However, under high precursor partial pressure conditions, GaCl and O₂ may undergo parasitic gas-phase reactions before reaching the substrate, generating polycrystalline Ga₂O₃ particles. Once deposited onto the epitaxial surface, these particles can interfere with normal single-crystal growth, resulting in polycrystalline particles, local protrusions, or abnormal growth regions, thereby degrading epitaxial crystal quality ^[3,11].

      HCl-based HVPE generates GaCl through the reaction between HCl and metallic Ga, with H₂ produced as a by-product. In this system, additional HCl can be introduced into the growth zone to regulate the balance between growth and etching. By selectively etching parasitic Ga₂O₃ deposits, the process can reduce the formation of gas-phase particles and unwanted deposition. However, the HCl-based route remains a chlorine-containing chemical system accompanied by hydrogen-containing by-products. Issues including chlorine incorporation, Si contamination from quartz reactor components, corrosion resistance of equipment, and exhaust gas treatment still require careful control through optimization of growth temperature, gas flow, and reactor structure ^[3,12,13].

      Another emerging route is trihalide vapor phase epitaxy (THVPE), which directly uses GaCl₃ as the gallium precursor. Unlike Cl₂- or HCl-based HVPE, THVPE does not rely on continuous in-reactor chlorination of metallic Ga. Instead, pre-generated GaCl₃ is directly delivered to the growth zone, simplifying the metal source reaction process and potentially reducing source-zone particle formation and precursor supply fluctuations. Previous studies have demonstrated that THVPE can achieve growth rates of several tens of micrometers per hour, highlighting its potential for the fabrication of thick drift layers required for high-voltage power devices ^[14].

Figure 5. Device structures and performance of β-Ga₂O₃ devices based on HVPE epitaxial layers ^[16]

Figure 6. Relationship between C, H, and Si incorporation concentrations in β-Ga₂O₃ epitaxial layers and TMGa molar flow rate ^[20].

Figure 7. Relationship between room-temperature Hall mobility and electron concentration in (010)-oriented β-Ga₂O₃ epitaxial layers grown at high rates using TMGa ^[19].

Figure 8. (a) Schematic structure of β-Ga₂O₃ multilayer SIMS samples consisting of alternating Si-doped layers and UID layers grown under different oxygen source flow conditions; (b) SIMS depth profiles of C, H, and Si concentrations in the samples ^[19].

      Unintentional doping in MOCVD-grown β-Ga₂O₃ epitaxial layers is mainly associated with H, C, and Si impurities. Hydrogen (H) may act as a shallow donor contributing to electrical conduction, but it can also form complexes with carbon (C), leading to compensation effects. Interstitial H has been reported to exhibit shallow donor behavior, while the electrical role of C strongly depends on its lattice site occupation ^[21].

      Further TMGa-MOCVD studies have shown that H–C-related complexes can passivate certain C-related compensation effects, rather than simply introducing additional compensation centers ^[22]. Therefore, the carrier concentration obtained from room-temperature Hall measurements only represents the net result of the combined effects of donors, acceptors, and trap states, and cannot be directly interpreted as a measure of material purity.

A comprehensive evaluation of background impurities and their electrical impacts requires combined analysis using techniques such as secondary ion mass spectrometry (SIMS), temperature-dependent Hall measurements, deep-level transient spectroscopy (DLTS), and deep-level optical spectroscopy (DLOS).

Figure 9. Device structure and electrical performance of β-Ga₂O₃ Schottky barrier diodes (SBDs) ^[23].