(Invited) Ga2O3 Phase Control and Heterojunctions Using Plasma-Enhanced Atomic Layer Epitaxy

Abstract

Ga2O3 has emerged as a promising material for next generation power electronics. While β-Ga2O3 (monoclinic) is the most stable and studied of six Ga2O3 polymorphs, the slightly less energetically favorable α-, ε-, and κ-Ga2O3 phases have unique characteristics that can be exploited such as larger bandgaps, alloying for dopant control, or polarization beneficial to the formation of two-dimensional electron gas (2DEG) channels. Specifically, α-Ga2O3 (rhombohedral, corundum) has the largest bandgap of ~5.3 eV and can be alloyed with α-Al2O3 (8.8 eV) and α-In2O3 (3.7 eV) for bandgap engineering. Both ε-Ga2O3 (hexagonal, P63mc) and κ-Ga2O3 (orthorhombic, Pna21 ) phases are polar, with a predicted spontaneous polarization strength up to 10 times larger than GaN and 3 times larger than AlN. Like the III-N system, polarization induced charges can lead to higher charge densities and mobilities in 2DEGs formed at heterojunctions, which would improve the viability of Ga2O3 electronic devices. Plasma-enhanced atomic layer deposition (PEALD) is a popular, conformal, energy-enhanced synthesis method for thin films due to its many advantages, including: deposition at reduced growth temperatures, access to metastable phases, improved crystallinity, and increased growth rates. In this work, we use PEALD to produce high-quality heteroepitaxial Ga2O3 and (AlxGa1-x)2O3 (AlGO) films and investigate materials properties such as phase selectivity, ternary solubility limits, and electrical and optical performance.All Ga2O3 films were deposited in a Veeco Fiji G2 reactor equipped with a load lock and turbo pump using trimethygallium, trimethylaluminum, and O2 plasma. Initial studies on c-plane sapphire substrates at 350°C and 8 mTorr show the phase could be altered from β to α by a varying the pure O2 flow during plasma pulse from 5-40 sccm [1]. Optical emission spectroscopy indicate that the changes in the relative concentration of atomic oxygen is crucial for phase selectivity while the high ion flux to the surface can contribute to the crystallinity at low Tg [2]. To grow ε(κ)-Ga2O3 on c-plane sapphire required going to a much higher temperature (500°C), pressure (100’s mTorr), and O2 flow (100sccm) [1]. Without modifications to the current ALD system, pure ε(κ)-Ga2O3 on sapphire was not achieved under any conditions. Using optimum growth conditions for the three phases on sapphire, films were deposited on GaN and diamond to determine the effect of substrate structure. Transmission electron microscopy was conducted to determine the specific phases (β, ε, and κ) present in each case, and showed the amount of each phase varied with PEALD parameters. While films on diamond resulted in mixed β/ε(κ) phases, pure ε(κ)-phase films were attained on GaN and the strain varied with pressure and Tg. Vertical breakdown measurements were taken for both β- and ε(κ)-Ga2O3/n+ GaN substrates. Breakdown fields varied between 3.8-7.0 MV/cm dependent on the phase and strained state of the Ga2O3 films. β-Ga2O3 films showed less variability in breakdown field from device to device than ε(κ)-Ga2O3 films, but neither showed a dependence on device size.While PEALD is beneficial for depositing thin films of metastable phases, practical devices often require much thicker barrier and active layers. For this reason, we investigated integrating PEALD metastable Ga2O3 films with traditional semiconductor deposition techniques, such as molecular beam epitaxy (MBE), capable of extending these layers beyond 100 nm in thickness. The same MBE conditions were used to deposit Ga2O3 films on GaN substrates with and without PEALD ε(κ)-Ga2O3 nucleation layers. Those deposited without the PEALD metastable nucleation layer resulted in stable β-phase films, while those with nucleation layers resulted in pure ε(κ)-phase films. This shows importance of PEALD for realizing practical device structures using metastable phases.Finally, to investigate heterojunctions for 2DEG formation, AlxGa1-xO films were developed. While the full stoichiometric range could be reached using a PEALD digital alloying method, crystallinity was lost above x = 0.2 for the β phase, x = 0.35 for the ε(κ) phase, and x=0.6 for the α phase. Initial device structures will be shown in order to establish the feasibility of these films in device applications.[1] Wheeler, et al. Chem. Mater. 2020, 32, 1140-1152[2] Boris, et al. JVST A 2019, 37(6), 060909 Figure 1