Gallium oxide grown on cheap silicon wafers for the first time

Silicon carbide and gallium nitride get most of the attention in the power semiconductor world. But gallium oxide has been quietly gaining ground, and a suite of six results from Nagoya University may have just shortened its path to commercial viability considerably.

The headline achievement: growing crystalline gallium oxide (Ga2O3) on standard two-inch silicon wafers for the first time. This matters because native gallium oxide substrates are expensive, and silicon's superior thermal conductivity helps offset one of gallium oxide's known weaknesses - it does not shed heat well. If the material can be manufactured on silicon, the cost equation changes dramatically.

A new oxygen source changes the game

At the core of all six advances is a newly developed High-Density Oxygen Radical Source (HD-ORS). The device uses an ozone-oxygen mixed gas to double the density of atomic oxygen available during thin-film growth compared to conventional sources.

Why does oxygen density matter? During growth, gallium suboxide needs to react with oxygen to form the desired Ga2O3. If there is not enough atomic oxygen at the surface, a volatile byproduct escapes instead, capping how fast the film can grow. By doubling the available oxygen, the HD-ORS pushes that speed limit higher while also being compatible with both molecular beam epitaxy (MBE) - a precise vacuum-based crystal growth technique - and physical vapor deposition (PVD), a higher-throughput method better suited to factory floors.

Six results spanning the full process stack

The Nagoya team, working with university spinout NU-Rei Co., Ltd., presented the results at the spring meeting of the Japan Society of Applied Physics in March 2026. Together, the six advances cover the full sequence of steps needed to turn gallium oxide from a laboratory curiosity into a manufacturable material:

• HD-ORS development: The new oxygen source itself, establishing the foundation for all subsequent growth work.
• High-speed MBE homoepitaxial growth: Using the HD-ORS, the team grew beta-Ga2O3 on tin-doped Ga2O3 substrates at 300 degrees Celsius and a rate of 1 micrometer per hour. The low growth temperature reduces thermal stress and broadens compatibility with other device components.
• High-speed PVD homoepitaxial growth: Applying the same oxygen source to PVD achieved stable films at rates exceeding 1 micrometer per hour - approaching ten times the speed of conventional MBE, pointing toward industrial-scale production.
• Silicon substrate pretreatment: A critical preparatory step combining wet chemical cleaning with controlled adsorption of a single atomic layer of gallium onto the silicon surface, preventing re-oxidation during heating.
• First heteroepitaxial growth on silicon: Building on the pretreatment, the team grew Ga2O3 on Si(100) wafers, with heat treatment confirming single-crystal formation.
• p-type formation via NiO diffusion layers: Gallium-based semiconductors are notoriously difficult to dope into p-type form, which is necessary for building pn junctions - the basic building blocks of power devices. Using nickel ion implantation followed by annealing, the team created a graded nickel oxide diffusion layer with p-type characteristics, achieving twice the current density of a standard nickel Schottky diode.

Why gallium oxide is worth the effort

Gallium oxide's appeal lies in its ultra-wide bandgap - roughly 4.8 electron volts for the beta phase, compared to 3.3 for gallium nitride and 3.2 for silicon carbide. A wider bandgap translates directly to higher breakdown voltages, which means devices that can handle more power in smaller packages. The raw materials are also more abundant and cheaper than those required for SiC or GaN substrates.

The material's Achilles heel has been thermal conductivity. Gallium oxide conducts heat poorly, which limits how much power a device can dissipate. Growing it on silicon - which conducts heat roughly four times better - is one way to mitigate this. But achieving high-quality crystalline growth on a structurally different substrate (heteroepitaxy) is inherently difficult, which is why the Nagoya result is significant.

From lab to factory, cautiously

These results build on a related advance in gallium oxide p-type control reported by the same group in September 2025, and are being commercialized through NU-Rei. The target applications include electric vehicles, power conversion systems, and space electronics - all domains where high voltage handling, radiation tolerance, and efficiency matter.

Still, there is a long road between conference presentations and commercial devices. The heteroepitaxial films on silicon need extensive characterization for defect density and electrical performance. The p-type formation via NiO diffusion is promising but requires further optimization for device-level integration. And scaling from two-inch wafers to the larger formats demanded by semiconductor fabs remains an engineering challenge.

What the Nagoya team has demonstrated is not a finished product. It is a complete process stack - from substrate preparation through crystal growth to junction formation - that works. For a material that has spent years in the shadow of its better-established competitors, that is a substantial step forward.