High growth rate metal organic chemical vapor deposition grown Ga₂O₃ (010) Schottky diodes

In recent years, there has been a growing interest in the development of ultrawide bandgap (UWBG) semiconductor materials for advanced power electronic applications.^1–6 This surge of interest is primarily driven by the potential for significantly higher switching efficiency and increased power conversion densities offered by UWBG semiconductors. Monoclinic beta-phase Ga₂O₃ (β-Ga₂O₃, hereafter referred to as Ga₂O₃) has gained substantial attention due to its exceptional characteristics, including a wide bandgap of ∼4.8 eV, predicted breakdown field of ∼8 MV/cm, and the unique advantage of cost-effective synthesis methods for free-standing native substrates.^7–11 Ga₂O₃’s experimental field strengths also make it ideal for multikilovolt (>10 kV) devices suited for medium voltage grid applications. Furthermore, Ga₂O₃ stands out among UWBG materials due to its broad range of shallow n-type dopants and scalable, melt-grown native substrates that facilitate high-quality homoepitaxy.^7,9–17 As a result, various high-performance Ga₂O₃ devices have been extensively studied, demonstrating impressive breakdown voltages and power device figures of merit.^18–23 Demonstrating superior capabilities, Ga₂O₃-based power devices suggest a potential to surpass those of traditional wide bandgap materials like silicon carbide (SiC) and gallium nitride (GaN), positioning Ga₂O₃ as a frontrunner for next-generation high-voltage power switching applications.^24,25

However, vertical device configurations are increasingly adopted for most high-voltage and high-power applications by effectively mitigating surface effects and offering numerous advantages, including chip area scalability, higher current capability, improved field management, and enhanced thermal management.^25–33 Vertical Ga₂O₃ devices encounter significant performance challenges due to the absence of shallow p-type dopants. This scarcity is attributed to the large hole effective mass and the high ionization energy required for acceptors associated with conventional doping species in Ga₂O₃.^34–41 The absence of a p–n homojunction has led to the exploration of alternative rectifying junctions, such as Schottky barriers, p–n heterojunctions, and metal–insulator–semiconductor diodes.^42–45 Among these, the vertical n-type Schottky barrier diode (SBD) stands out as the primary rectifying device for Ga₂O₃ for its high-quality interface.^45–48 Thin films of Ga₂O₃ have been grown on both native and foreign substrates using various techniques, including molecular beam epitaxy (MBE),^49,50 metal-organic chemical vapor deposition (MOCVD),^51–53 low-pressure chemical vapor deposition,^54,55 pulsed laser deposition,⁵⁶ and halide vapor phase epitaxy (HVPE).^57–60 MBE is characterized by a slow growth rate, attributed to the constrained evaporation from the metal source and desorption processes occurring under ultrahigh vacuum conditions, making the development of thick Ga₂O₃ drift layers using this technique impractical.^27,61 Thus far, HVPE has been favored for Ga₂O₃ vertical devices with thick drift layers, owing to its capability for fast growth rates of over 10 μm/h.^62,63 However, these HVPE-grown films with fast growth rates exhibit rough surface morphology with surface steps and pits,^58,64 necessitating expensive and potentially contaminating chemomechanical polishing before device manufacturing.⁶⁵ On the other hand, MOCVD is a promising technique that offers potential for high-quality Ga₂O₃ films, with a growth rate of ∼1 μm/h, smooth surface morphology, controllable doping, and the highest room (∼200 cm²/V s) and low-temperature mobility (∼10 000 cm²/V s), as well as low compensating levels (<10¹⁵ cm⁻³), which directly affect power device performance, utilizing triethylgallium (TEGa) as the precursor.^66–70 The challenge, however, is to overcome an extremely slow growth rate. In contrast, trimethylgallium (TMGa) precursor in a close-coupled showerhead reactor, achieved a rapid growth rate of up to 9.8 μm/h,⁷¹ although surface morphology and electrical transport characteristics were not thoroughly examined. Recent TMGa-based MOCVD reached 1.5 μm/h growth with promising surface quality (1.8 nm RMS).⁷⁰ This makes MOCVD a preferred growth method for Ga₂O₃ power devices. Although MOCVD growth of Ga₂O₃ using TMGa is relatively new and underexplored, a research gap exists concerning a comparative analysis of films grown with TEGa and TMGa, as well as the electrical properties of devices manufactured from these films. We present three MOCVD-grown Ga₂O₃ vertical SBDs fabricated on TEGa and TMGa-grown films.

This work investigated the MOCVD growth of Si-doped Ga₂O₃ homoepitaxial drift layers on (010) Sn-doped Ga₂O₃ substrates using TEGa and TMGa precursors. We conducted a comprehensive analysis of the surface morphology of the grown thin film using atomic force microscopy (AFM) and field emission scanning electron microscopy (FESEM). There are limited reports on the fabrication and characterization of vertical power devices’ MOCVD-grown Ga₂O₃ films using different growth rates. The primary focus of this study is to explore an in-depth investigation into the electrical characteristics of three MOCVD-grown Ga₂O₃ Schottky diode samples, each differing in the epilayer thickness, doping levels, and growth rates. The promising Schottky diode characteristics in terms of the highest breakdown voltage, highest on-off ratios, and moderate field strength have been achieved from the highest reported Ga₂O₃ drift layer thickness grown using TMGa with a high growth rate (3 μm/h).

Three Ga₂O₃ films were grown through MOCVD on Sn-doped (010) oriented Ga₂O₃ substrates (commercially acquired from Novel Crystal Technology, Inc.), labeled S1, S2, and S3. Prior to loading into the MOCVD growth chamber, Ga₂O₃ substrates were cleaned using acetone, isopropanol, and de-ionized water. High-purity oxygen (O₂) gas was used as the oxygen precursor, and argon (Ar) was the carrier gas. N-type doping was achieved using diluted silane (SiH₄) as the dopant source.

Sample S1 was grown under previously established conditions using trimethylgallium (TMGa) as the gallium precursor,⁷² while samples S2 and S3 followed the growth conditions using triethylgallium (TEGa) as the gallium source.⁶⁹ The growth temperature for S1 (using TMGa) was maintained at 950 °C, whereas for S2 and S3 (using TEGa), it was set at 880 °C. All samples were grown at a constant pressure of 60 Torr. S1 achieved a thickness of 9.5 μm with a growth rate of 3 μm/h, while S2 and S3 reached thicknesses of 3 and 2 μm, respectively, with a growth rate of 0.65 μm/h. The targeted electron concentrations for S1, S2, and S3 were at 2 × 10¹⁶, 2 × 10¹⁶, and 7 × 10¹⁶ cm⁻³, respectively. The electron concentrations in β-Ga₂O₃ were effectively controlled by tuning the silane molar flow rate during growth. Si acts as a shallow donor with a low activation energy in β-Ga₂O₃, making the electron concentration directly proportional to the silane molar flow rate, as demonstrated in previous studies.^69,72 These electron concentrations were also verified through room temperature Hall measurements conducted on the coloaded sample grown on Fe-doped (insulating) Ga₂O₃ substrates. Please refer to Table I in this study for a comprehensive overview of the growth conditions of the three Ga₂O₃ films grown on Sn-doped Ga₂O₃ substrates.

The surface morphologies of the β-Ga₂O₃ homoepitaxial thin films were characterized using FESEM (FEI Helios 650). Figures 1(a)–1(c) show the top view FESEM images of β-Ga₂O₃ homoepitaxial films grown on (010) Sn-doped β-Ga₂O₃ substrates. Despite S1 having the thickest film of 9.5 μm, S2 exhibits a higher density of surface defects than S1. This increased defect density in S2 could be attributed to its longer growth duration, necessitated by the slower growth rate when using TEGa. In contrast, S3 demonstrates the smoothest surface due to its relatively thinner film thickness. The AFM analysis of the thin films revealed surface characteristics. Figures 2(a)–2(c) illustrate AFM images with a 5 × 5 μm² scan for all samples, where the RMS roughness values for S1, S2, and S3 thin films were 3.12, 11.4, and 10.7 nm, respectively. Despite S1 exhibiting lower micrometer-scale RMS roughness compared to other samples, it exhibits larger surface steps and pits on the macroscale. S1 has a lower pit density compared to S2 and S3, but the pits are notably large, likely due to its high growth rate. Conversely, although S2 and S3 have a higher pit density than S1, their pits are shallower, indicating a lower growth rate. S3, with a shorter growth duration, exhibits a smoother surface compared to S2.

Figure 3(a) illustrates a schematic cross section of the device structure employed for electrical testing. The fabrication process initiated with backside etching using BCl₃ reactive-ion etching, removing 1 μm thick Ga₂O₃. Subsequently, a Ti/Au Ohmic metal stack was deposited through electron beam evaporation and subjected to rapid thermal annealing in N₂ at 470 °C for 1 min Finally, the top Ni/Au Schottky contacts were defined by electron beam lithography. Following the fabrication of the devices, current density–voltage (J–V) measurements were performed at room temperature to extract the fundamental Schottky diode properties using the HP 4155B semiconductor parameter analyzer. Furthermore, a standard reverse-biased capacitance–voltage (C–V) measurement on the Schottky contacts was performed using an Agilent 4294A precision impedance analyzer. A room temperature reverse breakdown measurement of the diodes was subsequently performed with the samples submerged in Fluorinert FC-40 dielectric liquid.

As seen in Table II, there is a trend observed in ideality factors among the three Ga₂O₃ diode samples (S1, S2, and S3), which can be attributed to variations in surface quality and doping concentrations. The high ideality factor represents the prevalence of nonideal effects, notably the spatial inhomogeneity of Schottky barriers, likely arising from surface roughness at the metal–semiconductor interfaces. The diodes exhibit distinct characteristics, with S1 displaying the highest surface roughness on the macroscale and an ideality factor of 2.17, which suggests increased recombination centers associated with surface defects, possibly due to its macroscale roughness and highest growth rate. The higher η value in S1 indicates the presence of nonideal effects, including thermionic field emission, trap-assisted tunneling currents, and the quality of metal adhesion and interfaces. In contrast, S2, characterized by a lower growth rate and thickness, demonstrates improved device quality, reflected in a lower ideality factor of 1.73. Notably, S3, distinguished by its smooth surface and a higher doping concentration of 7 × 10¹⁶ cm⁻³, exhibits the most favorable performance with the lowest ideality factor of 1.31, suggesting enhanced carrier recombination characteristics. The measured Schottky barrier heights for S1, S2, and S3 from the J–V curve are 1.72, 1.18, and 1.0 eV, respectively. Thus, a smoother surface and higher doping concentration are observed to correlate with a reduced Schottky barrier height.⁵⁵ Therefore, S3, with the smoothest surface and the highest doping concentration, exhibits the lowest Schottky barrier height of 1.0 eV. S1, characterized by higher surface roughness and a lower doping concentration, shows the highest barrier height of 1.72 eV. S2, positioned between S1 and S3, demonstrates an intermediate barrier height of 1.18 eV. The calculated Schottky barrier heights agree with the values reported in the literature.^77–79 Nevertheless, the minimum differential specific on-resistance (R_on,sp) for the S1, S2, and S3 diodes have been extracted as 2.36, 1.02, and 0.417 mΩ cm⁻², respectively [Fig. 3(d)]. The S3 diode is the most favorable among the three samples. This can be attributed to the combination of a thinner epilayer, smoother surface, and higher doping concentration in S3. Among the examined four parameters—surface roughness, growth rate, epilayer thickness, and doping concentration—it appears that both epilayer thickness and doping concentration play pivotal roles in influencing specific on-resistance. Specifically, S1 (thicker epilayer, lower doping) demonstrates the highest R_on,sp of 2.36 mΩ cm⁻². The slope of the J–V plot [Fig. 3(b)] for S2 begins to decrease beyond approximately 3 V, indicating an increase in the specific on-resistance for S2 after this voltage, as shown in Fig. 3(d).

Figure 4(a) shows the room temperature reverse bias C–V plots of the three samples, which indicated a complete depletion of the drift layers of the three samples. The 1 MHz C–V measurements revealed a net doping concentration of 2.02 × 10¹⁶, 2.82 × 10¹⁶, and 6.08 × 10¹⁶ cm⁻³ in the drift layers of the S1, S2, and S3 Schottky diodes, respectively (supplementary material).⁸² These numbers match very well with the target electron concentrations mentioned in Table I. The S1 sample features a drift layer that is highly compensated, containing very few electrons (supplementary material).⁸² The doping profile obtained from the CV plot exhibits high error bars. Therefore, the higher turn-on voltage and increased parasitic resistance may likely result from the combined impact of space charge limited transport and Schottky transport.

The 1/C²–V analysis, as seen in Fig. 4(b), also provided SBHs of S1, S2, and S3 diodes as 3.13, 1.80, and 1.66 eV, respectively, which were calculated from the extracted built-in voltage (V_bi) by accounting for the Fermi level position relative to the conduction band (E_C–E_F).^48,80 The barrier heights measured from the C–V characteristics of the samples are reasonable but slightly larger than the barrier heights measured from the J–V characteristics. However, both J–V and C–V measurements consistently exhibit that S1 has the highest SBH, while S3 has the lowest.

The reverse J–V characteristics of S1, S2, and S3 diodes are illustrated in Fig. 5. The destructive breakdown voltages are observed in the Schottky diodes as −510 V for S1, −270 V for S2, and −180 V for S3. Calculated field strengths correspondingly measure 0.54, 0.9, and 0.9 MV/cm for S1, S2, and S3, respectively. Differences in the drift layer thickness, doping concentrations, barrier heights, surface roughness, and overall device quality could influence the variation in the breakdown voltages of the three diode samples. Specifically, S1, with the thickest drift layer, exhibits a lower electric field strength and a higher breakdown voltage, while S3, with the thinnest layer, shows the opposite trend. S2 and S3 exhibit an identical calculated field strength of 0.9 MV/cm despite variations in breakdown voltage. This is likely due to the combination of a thinner drift layer and higher doping concentration in S2 and S3, leading to a more concentrated electric field. Surface roughness indirectly impacts semiconductor device breakdown voltage; improving smoothness, such as through chemical mechanical polishing, is anticipated to notably enhance reverse characteristics. However, the epilayer thickness, doping concentrations, and overall device quality are more direct determinants of the observed trend in destructive breakdown voltages. It is evident that the MOCVD-grown 9.5 μm thick Ga₂O₃ film (S1), achieved with the highest growth rate (3 μm/h) using TMGa, exhibits promising performance, in terms of the highest breakdown voltage, highest on-off ratios, and moderate field strength. The demonstrated potential of the high growth rate MOCVD technique to grow thick Ga₂O₃ drift layers underscores its promise for advancing high-voltage device applications. A comparative analysis of the MOCVD-grown Ga₂O₃ homoepitaxial thin films from this study as compared to data reported in the recent literature has been presented in Table III.

In summary, three different Si-doped homoepitaxial Ga₂O₃ thin films were grown on Sn-doped (010) Ga₂O₃ substrates via MOCVD using both TEGa and TMGa as Ga precursors. The epitaxial growth of Ga₂O₃ material tailored for high-performance electronic devices is achieved with an impressive epilayer thickness of 9.5 μm, accomplished at a notably high growth rate of 3 μm/h using TMGa in an MOCVD reactor. This is a significant progress, surpassing prior reports in terms of both thickness and the growth rate in the context of electronic device-grade Ga₂O₃ material growth. Schottky diodes using the three MOCVD-grown films have also been demonstrated. The Schottky diode fabricated from the 2 μm thick Ga₂O₃ drift layer (S3) grown using TEGa precursor exhibits the highest forward current densities (4000 A/cm² at 3.5 V), lowest specific on-resistance (0.417 mΩ cm⁻²), moderate On-off ratio (>10⁸), and lowest ideality factor (1.31), indicating the promise of using TEGa as the gallium precursor for MOCVD growth for decent drift layer thicknesses. On the contrary, the Schottky diode fabricated from 9.5 μm thick Ga₂O₃ drift layer (S1), grown using TMGa precursor, shows the lowest leakage current (8.32 × 10⁻⁸A/cm² at −2 V), highest on-off ratio (>10⁹), and highest reverse breakdown voltage of −510 V without any electric field management, indicating the viability of using TMGa as gallium precursor for achieving high growth rate of Ga₂O₃ epilayer in MOCVD growth technique for vertical power electronic devices. The specific on-resistance of the samples also increases with the increasing growth rate. For all the samples, the average doping concentrations, measured from the C–V characteristics, match the target electron concentration. Overall, while MOCVD-grown thin films with TEGa, particularly at 2 and 3 μm, exhibit superior electrical performance concerning forward current density, on-off ratio, and specific on-resistance, TMGa-grown thick films at 9 μm also show promising electrical properties, especially in terms of breakdown voltage, reverse breakdown voltage, and on-off ratio. This establishes the viability of TMGa in thick film growth suitable for electronic devices in the MOCVD technique. The observed trends and characteristics in thick epilayer growth, especially at a high growth rate, position it as a valuable subject for further in-depth investigation and detailed exploration in Ga₂O₃ semiconductor device research. Achieving further advancements involves optimization of growth conditions to enhance surface smoothness. The findings underscore the potential of employing the MOCVD growth technique with a high growth rate for the deposition of thick Ga₂O₃ layers on native substrates. This holds significant promise, especially in developing vertical power devices, as this study demonstrated.

We acknowledge the support from AFOSR (Air Force Office of Scientific Research) under Award No. FA9550-18-1-0479 (Program Manager: Ali Sayir), from National Science Foundation (NSF) under Award Nos. ECCS 2019749 and 2231026, and II–VI Foundation Block Gift Program. This work used the electron beam lithography system acquired through NSF MRI Award No. ECCS 1919798.

The authors have no conflicts to disclose.

Sudipto Saha: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Methodology (equal); Validation (equal); Writing – original draft (equal); Writing – review & editing (equal). Lingyu Meng: Data curation (equal); Formal analysis (equal); Validation (equal); Writing – original draft (supporting); Writing – review & editing (supporting). Dong Su Yu: Data curation (supporting); Formal analysis (supporting); Methodology (supporting); Validation (supporting); Writing – original draft (equal). A. F. M. Anhar Uddin Bhuiyan: Data curation (supporting); Formal analysis (supporting); Methodology (supporting). Hongping Zhao: Conceptualization (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Resources (equal); Supervision (lead); Writing – original draft (equal). Uttam Singisetti: Conceptualization (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (equal); Methodology (equal); Resources (equal); Supervision (equal); Writing – original draft (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.