Enhancing β-Ga₂O₃ with hetero-integration

Thanks to the introduction of a far higher thermal conductivity and p-type doping, better devices are realised when pairing β-Ga₂O₃ with SiC or diamond.

BY ARPIT NANDI, ADITYA BHAT, INDRANEEL SANYAL, SAI CHARAN VANJARI, JAMES POMEROY, MATTHEW SMITH AND MARTIN KUBALL FROM THE UNIVERSITY OF BRISTOL

Climate change and unpredictible extreme weather are impacting human life more frequently than ever. In just the last few months many in central Europe have suffered from such events, with severe flooding occuring at the same time that more than 5,000 firefighters were fighting wildfires in northern Portugal. These events underscore the urgent need to decarbonise the energy sector, an endeavour where power electronics can play a crucial role.

Within the power electronics portfolio, different materials are seeing deployment at different voltages. For low-to-medium-voltage applications, it is the silicon-based devices that dominate, due to their cost-effective manufacturing processes. But in the mid-to-higher voltage range, GaN and SiC have gained traction, with GaN-on-silicon benefiting from silicon-style manufacturing. And there is also Ga₂O₃ to consider – it has emerged as a highly promising material for power electronic devices, due to its large bandgap (4.8 eV), its tolerance for high electric fields (8 MV cm^-1), and the promise of cheaper production than the more established SiC.

Figure 1.(a) and (b) Ga₂O₃ MOCVD, and (c) clean room facility at the University of Bristol.

Researchers working with this ultra-wide bandgap oxide have enjoyed tremendous success over the last decade or so, with interest ignited by the first report of a Schottky barrier diode in 2013. Spurred on by this triumph, alongside the ease of n-type doping and the availability of melt-grown substrates, these pioneers are now gaining further encouragement as 6-inch substrates appear on the horizon, as well as a push from multiple material vendors across the globe. There are now producers of Ga₂O₃ substrates in Japan, the US, Germany, South Korea and China. Based on all this promise, commercialisation of Ga₂O₃ looks inevitable for high-voltage applications.

Those developing Ga₂O₃ material and devices include the UK government-funded Innovation and Knowledge Centre REWIRE, led by our team at the University of Bristol. While we acknowledge that today’s Ga₂O₃ devices exhibit excellent performance, with breakdown voltages that can exceed 8 kV, there are drawbacks that we are starting to address. Significant concerns preventing this material from harnessing its full potential include an on-state currents that’s relatively low, and a device reliability that still needs to be established and proven.

Figure 2. (a) Cross-sectional TEM image, low magnification plan view of Ga₂O₃ thin film grown on diamond. (b) The rocking curve illustrates grain size increase after incorporating an improvised seeding layer, reflecting a decrease in FWHM.

Behind these limitations are: a lack of useable p-type doping for Ga₂O₃, primarily due to the flatness of the valence bands; and a low thermal conductivity. But there are ways to mitigate this through the heterogeneous integration of Ga₂O₃ with SiC and diamond – they are materials with high thermal conductivity and p-type doping. Combining Ga₂O₃ with other materials is an increasingly popular approach to tackling the critical technological bottlenecks. For example, Ga₂O₃ MOSFETs attached to diamond through mechanical exfoliation and bonding are showing good promise, as are devices featuring composite wafer fabrication with SiC. Still, concerns remain related to scalability, impairing manufacturing. Note that in addition to Ga₂O₃ and its alloys, REWIRE is exploring, amongst others, SiC, AlGaN, diamond and BN device technologies, in a team with partners at Warwick University and Cambridge University.

Powered by state-of-the-art growth and clean room facilities at the University of Bristol, we are approaching heterogeneous integration of Ga₂O_{3 i}n various ways. Our goals include: optimising Ga₂O₃ heteroepitaxy on SiC and diamond substrates, the latter in collaboration with Element Six Technologies and Orbray; and in partnership with Srabanti Chowdhury’s Wide Bandgap Lab at Stanford University, exploring thin p-type diamond overgrowth on Ga₂O₃.

Pairing Ga₂O₃ with diamond…
Pairing Ga₂O₃ with diamond is very attractive. With a bandgap of 5.4 eV, p-type conductivity, a thermal conductivity of up to 2000 W m^-1 K^-1, and a predicted critical electric field strength of 10 MV cm^-1 – that’s even higher than that of Ga₂O₃ –diamond compliments and enhances the effectiveness of Ga₂O₃ when its adequately integrated.

Our initial work on this front has involved the growth of Ga₂O₃ on single-crystalline diamond substrates. For this effort, we faced two underlying challenges: prevention of oxidation of the diamond surface at high temperatures during initial growth stages; and overcoming the challenge of realising good nucleation, which is plagued by the high surface energy differences between Ga₂O₃ and diamond.

To address these concerns, we have pursued a two-step process, beginning with the low-temperature growth of a thin layer of Ga₂O₃ that protects the diamond surface, followed by high-temperature growth of an epitaxial layer. Our detailed analysis of the microstructures and grains produced during this process has revealed two-grain variants aligned to [110] diamond and its perpendicular direction. Due to a peculiar asymmetric hexagonal (-201) face and mirror symmetry, each set has its own four equivalent subvariants (for more details see Nandi et al. Crystal Growth & Design 23 8290 (2023)).