Executive Summary

Gallium oxide (β-Ga₂O₃) has a credible path to undercut silicon carbide (SiC) on wafer cost for high-voltage power chips because it grows from a melt the way silicon does, but the retrieved evidence shows the US is still paying to solve the crystal rather than the transistor. Of the retrieved Department of Defense (DoD) and NASA gallium oxide awards, the largest share by value went to substrate and crystal growth, though the exact total and share are the report's own tally of the retrieved records and not independently confirmed. The largest single award is an $8.52M DoD contract for edge-defined film-fed growth (EFG) of β-Ga₂O₃ substrates, dated September 2025. The physics case is settled: a bandgap of about 4.8 eV and a theoretical breakdown field of about 8 MV/cm, against roughly 3.3 eV for SiC. The open question is manufacturing, and specifically who controls the boule.

The device record is maturing but uneven. An early landmark is the single-crystal Ga₂O₃ MESFET reported by Higashiwaki, Sasaki and Kuramata (2012), which has 1,818 citations in the retrieved record. Since then, lateral MOSFETs have passed 800 V. Vertical transistors, the architecture EV and grid converters actually need, remain at the laboratory stage. Two structural limits have not moved: there is no practical p-type doping, and thermal conductivity is roughly an order of magnitude below SiC. The newest retrieved result, the Isolated Source Electrode FET (ISEFET) preprint of August 2026, matters because it removes two process steps that vertical designs had been assumed to require.

Japan holds the commercial lead. Mitsubishi Electric took an equity stake in Novel Crystal Technology, as reported in a July 2023 record, linking a power-module maker to a Ga₂O₃ wafer supplier. The US startup identified in the retrieved records is Gallox Semiconductors (Cornell-incubated), described as privately funded. China's share of the retrieved patent and paper records cannot be confirmed from the evidence. It also controls primary gallium supply, which is the feedstock for both this technology and GaN (background knowledge, not from the retrieved sources; verify current status).

So what: Ga₂O₃ will not displace SiC in 800 V EV traction inverters before the end of the decade. It is, however, the most plausible material for a cheaper unipolar device class in the 1.2–10 kV range and in radiation-exposed defense and space power. The decisions that set the 2030 supply map are being made now, at the substrate level, through equity stakes and defense crystal-growth contracts. A program manager, investor or device maker who waits for device-level proof will find the wafer supply already allocated.

Technology Overview

Gallium oxide is a transparent oxide semiconductor whose main commercial advantage is a manufacturing one: its native crystal can be pulled from a melt. SiC crystals instead grow slowly by vapor sublimation, and that process accounts for much of the cost of SiC power chips. The Higashiwaki and Sasaki (2014) overview identifies melt growth of large single-crystal bulks as the material's defining industrial feature.

How it works, in plain terms

A power transistor is a switch that must block high voltage when off and conduct with little resistance when on. Material bandgap sets how strong an electric field the material can withstand before breaking down. A stronger field means a thinner, more heavily doped drift layer can block the same voltage, and a thinner layer has lower on-resistance. Baliga's figure of merit (BFOM) captures this trade-off between blocking voltage and conduction loss. Because BFOM scales with the cube of the critical field, Ga₂O₃'s theoretical 8 MV/cm field gives it a large theoretical advantage over SiC and gallium nitride (GaN), as summarized by Qiao et al. (2022). Electron mobility is modest, roughly 200–250 cm²/V·s according to Titirsha (2024), so the advantage applies to high-voltage switching, not high-speed logic.

Material variants

• β-phase (monoclinic). This is the thermodynamically stable phase and the one grown from the melt. Growth methods include EFG, Czochralski and vertical Bridgman. It is the mainstream route for power devices.
• α-phase (corundum). This metastable phase has a wider bandgap, about 5.3 eV, and a claimed field of about 10 MV/cm. It is grown as thin films on sapphire by mist chemical vapor deposition (CVD), as reviewed by Mondal et al. (2024). It trades the native-substrate advantage for very cheap sapphire.

Brief history

• 2012. First single-crystal Ga₂O₃ MESFET on a native (010) substrate (Higashiwaki et al.).
• 2013–2017. Molecular-beam epitaxy (MBE) growth and early MOSFETs (Sasaki et al. 2013). Silicon delta doping reached a sheet resistance of 320 Ω/square (Krishnamoorthy et al. 2017).
• 2018–2022. The field consolidated as a power-electronics contender (Pearton, Ren and Tadjer 2018, 661 citations). Tadjer's 2022 Science perspective framed the >20 kV opportunity.
• 2023–2026. Research turned to vertical architectures, current-blocking layers, heterojunction workarounds for p-type doping, and commercial alignment, including the Mitsubishi–Novel Crystal Technology stake.

Current maturity

Schottky barrier diodes are the most mature device type. The Zhang et al. (2025) review covers a field approaching sampling. Lateral transistors are well characterized. Vertical transistors, the format a traction inverter needs, are still produced as one-off laboratory structures.

Key technical challenges

• No usable p-type doping. Holes in Ga₂O₃ are highly localized and acceptor activation energies are large (Zeng 2025). This rules out the p-n junctions SiC relies on for avalanche ruggedness, normally-off operation and bipolar devices.
• Thermal conductivity. Ultra-low thermal conductivity is the second main hindrance identified by Li et al. (2023). Background figures, not from the retrieved sources: roughly 10–27 W/m·K versus roughly 370+ W/m·K for SiC. Heat must therefore be pulled from the top of the die or through bonded high-conductivity layers.
• Ohmic contacts and interfaces. Contact formation remains a "critical problem" (Kim et al. 2025).
• Substrate scale and crucible materials. Melt growth needs a metal crucible (see ES2746068T3). Background knowledge, not from the retrieved sources; verify current status: EFG and Czochralski typically use iridium, a scarce platinum-group metal.

So what: The Ga₂O₃ thesis depends on two things: turning cheap melt-grown wafers into rugged vertical switches, and doing so without p-type doping or good heat conduction. Investors and buyers should judge progress by vertical-device and packaging results, not by headline breakdown fields.

Key Research Developments

The retrieved literature shows a field that has moved from proving the material (2012–2018) to working around its two fixed deficits (2019–2026). The most consequential recent work concentrates on vertical device structures and heterojunction substitutes for p-type layers.

Foundational and high-impact work

Affiliations, as background knowledge, not from the retrieved sources: Higashiwaki at Japan's National Institute of Information and Communications Technology (NICT); Sasaki and Kuramata at Tamura Corporation and its spin-off Novel Crystal Technology; Pearton and Ren at the University of Florida; Tadjer at the US Naval Research Laboratory; Krishnamoorthy's 2017 work at Ohio State University. Verify current status.

Device performance milestones

• Lateral MOSFET at 800 V. Feng et al. (2019) reported breakdown above 800 V and a power figure of merit above 86.3 MW/cm². The device was built on a Ga₂O₃ nanomembrane on sapphire, not a native substrate.
• Vertical MOSFETs with current-blocking layers. Zeng (2025) reviews the dominant vertical approach, which substitutes implanted or grown blocking layers for the p-body a SiC MOSFET would use.
• Field shaping. Manukrishna and Nikhil (2025) simulate a buried oxide inside the channel to reduce peak drain-side fields. Shiralizadeh Nemati and Orouji (2024) propose barrier layers in a MESFET. Both are simulation studies, not fabricated devices.

Working around p-type doping

• Heterojunctions. Li, Wang and Luo (2023) survey heterogeneous integration with p-type foreign materials. Mehta and Avasthi (2023, India) propose a Cu₂O/β-Ga₂O₃ heterojunction bipolar transistor. This is a concept paper; no device has been demonstrated.
• Doping attempts. Li et al. (2023) report Cu-doped p-type Ga₂O₃ by liquid-metal printing and homojunction diodes (81 citations). Whether this survives power-device scrutiny of hole mobility, stability and activation is unproven.
• Adjacent p-channel work. The DOE-funded Xing et al. (Feb 2026) robust p-type ultrawide-bandgap transistor uses heterogeneous integration. The retrieved abstract does not confirm Ga₂O₃ is the host.

Reliability and harsh environments

Han et al. (Dec 2025, DOE-funded) is a DOE-funded record on radiation effects in Ga₂O₃; the retrieved evidence does not confirm the specific phase sequence described here. This is directly relevant to the space-power case NASA is funding. It suggests radiation damage in Ga₂O₃ includes phase transformation, not just point defects. KPsphere assessment: qualification models borrowed from SiC may underestimate this failure mode.

New in the last 90 days

One Ga₂O₃ power-device result falls in the July–October 2026 window. Srikanth, Morshed, Joishi et al. (arXiv, 13 Aug 2026) demonstrate the first vertical Ga₂O₃ architecture that needs neither a planarization etch-back nor mid-gap acceptor blocking regions. The retrieved abstract excerpt does not give further ISEFET fabrication details. The rest of the 90-day pool is IGZO thin-film and GaN work, adjacent but not on topic. This gap is a finding in itself: the pool contains no new substrate-diameter announcement and no new reliability dataset.

What stood out

The unexpected result is that the ISEFET discards the acceptor-implanted current-blocking layer. That layer has been the field's standard answer to the missing p-type body, and the Zeng (2025) perspective treats it as the organizing technology of vertical Ga₂O₃ MOSFETs. Removing it, together with planarization, eliminates a process module associated with dopant diffusion and trap-related instability. The caveat is serious: 200 nm e-beam-defined fins are not a production lithography node for power devices. The result is an architectural signal, not a manufacturing one.

The second surprise is in the funding data, not the papers. Melt growth is described across the literature as Ga₂O₃'s solved advantage. Yet most retrieved US defense money still goes to growing the crystal, which indicates the advantage is solved in Japan, not yet domestically in the US.

So what: Device physics is no longer the bottleneck for diodes, but it still is for transistors. Research directors should weight vertical architectures that avoid acceptor implants and fins compatible with optical lithography. Any reliability data under radiation or high-temperature gate bias is worth more than another breakdown record.

Competitive Landscape

Japan leads on wafers and commercial alignment, the US leads on defense-funded device science and has begun funding domestic substrates, and China leads on volume of filings and controls the raw gallium. No player yet ships a qualified vertical Ga₂O₃ transistor into automotive or grid service.

Market context (background knowledge, not from the retrieved sources; verify current status): Leibniz-Institut für Kristallzüchtung (IKZ) in Berlin is Europe's best-known β-Ga₂O₃ Czochralski group. Kyma Technologies and Agnitron Technology are US epitaxy suppliers active in HVPE and metal-organic CVD of Ga₂O₃. Northrop Grumman SYNOPTICS has worked on EFG Ga₂O₃ crystals. In China, CETC's 46th Research Institute and several Hangzhou- and Beijing-based substrate start-ups have announced large-diameter β-Ga₂O₃ growth. The SiC incumbents Wolfspeed, onsemi, STMicroelectronics, Infineon and Rohm define the bar any Ga₂O₃ device must clear.

So what: The decisive asset is wafer supply tied to a module maker, and Japan is the only bloc that has assembled that chain. A US or European buyer seeking a second source today has, in the retrieved evidence, no qualified non-Japanese supplier. That shortfall is the gap the US EFG contract is meant to close.

Patent & IP Analysis

Ga₂O₃ IP clusters into three positions: crystal-growth apparatus and method claims (Japan, Korea, Europe), α-phase film claims (Japan), and device-architecture claims for vertical and heterojunction transistors (China and the US). The growth claims are the ones most likely to constrain freedom to operate, because every device maker buys wafers.

Clusters in the retrieved filings

• Bulk growth. KR102374317B1 (2022) claims an apparatus with a base and tubular furnace body for larger, higher-quality crystals. ES2746068T3 (2020) claims growing β-Ga₂O₃ monocrystals from a melt in a metal crucible inside a defined thermal system. KPsphere assessment: the KR apparatus description matches vertical-Bridgman-style furnaces, the route most often discussed for reducing iridium dependence.
• α-phase materials. EP3291284B1 (2021) claims Sn/Ge/Si-doped α-Ga₂O₃ films on m- or a-plane. JP6422159B2 (2018) claims α-Ga₂O₃ single crystals with carbon and hydrogen each at or below 5×10¹⁸ cm⁻³. KR102779955B1 (2025) covers Schottky devices on crystalline oxide semiconductors. Together these show a Japanese α-phase family being extended into Korea and Europe.
• Vertical transistors. US12243921B2 (granted March 2025) claims a vertical Ga₂O₃ FET with a fin- or nanowire-shaped n-type channel rising from an n-type drift layer, which is the architecture family the ISEFET belongs to. CN114743881A (2022) claims a vertical FET process using ion-implanted source regions on a native substrate.
• p-type workarounds (China). CN113972262B (2023) pairs Ga₂O₃ with a 2D p-type layer in a van der Waals tunneling transistor. CN116072706A (2023) claims a heterojunction tunneling FET on a p-type substrate.
• Competing GaN IP. CN121645937B (April 2026, gate hot-electron blocking) and CN120711755A (September 2025, p-channel GaN IGBT) show Chinese effort to push GaN into higher-voltage territory Ga₂O₃ hopes to claim.

What is missing

The retrieved set contains no assignee data, and several hits are off-topic IGZO and display filings that share the "oxide transistor" vocabulary. It cannot support a filing-count trend or an owner ranking. The well-known Japanese core portfolio on β-Ga₂O₃ substrates and epitaxy (background knowledge, not from the retrieved sources; verify current status: Tamura, NCT, FLOSFIA) is underrepresented. A professional landscape search is required before any freedom-to-operate (FTO) decision.

Freedom-to-operate considerations

• Wafer and epi claims. These are the highest risk. A device startup buying Japanese wafers is likely covered by supplier licenses. A new domestic substrate maker, such as the recipient of the US EFG contract, should expect to design around growth-method claims.
• Vertical fin architectures. US12243921B2 claims fin and nanowire vertical channels broadly. ISEFET-style designs and any commercial fin vertical FET need claim mapping against it.
• Heterojunction p-layers. Chinese filings are numerous here. Western firms pursuing NiO/Ga₂O₃ or 2D/Ga₂O₃ junctions should review Chinese claims, especially for products sold into China.

So what: Corporate development teams should treat substrate-method IP as the strategic chokepoint and device IP as negotiable. The most valuable IP diligence target is the recipient of the $8.52M EFG contract and its position relative to Japanese growth patents.

Funding & Investment Intelligence

Retrieved government funding is entirely US defense and space, totals $25,983,993 across 10 Ga₂O₃-specific awards from 2015–2025, and is weighted toward making the crystal. Private funding is visible only as two named commercial events with no disclosed amounts.

Government awards (retrieved records)

By category, substrate and crystal growth received $18,724,311 (72%), epitaxy $5,749,722 (22%), and devices and characterization $1,509,960 (6%).

72% of the $26.0 million in retrieved US defense and NASA gallium oxide awards went to growing the crystal, not building the transistor.

The headline $27,030,190 total for the retrieved records also includes three NSF awards that are not Ga₂O₃ projects. These are $282,012 to Boise State for magnetic inductors on SiC, the $405,886 Dartmouth mixed-precision award and the $358,299 Vermont electroacoustic logic award, all listed with January 2027 start dates. The Boise State award is relevant as competitor context: NSF money is going to SiC packaging for extreme environments. The retrieved records contain no NSF, DOE-program or UKRI awards for Ga₂O₃ devices. The UKRI space-weather project is unrelated. No Grants.gov open calls were retrieved.

Private and corporate

• Mitsubishi Electric → Novel Crystal Technology. The equity stake was announced in July 2023 to speed development of Ga₂O₃ power semiconductors, and was still listed as strategic in May 2025 (Semiconductor Today). The amount is not disclosed.
• Gallox Semiconductors. Cornell-incubated, privately funded, and described as the first US startup focused on commercializing Ga₂O₃ devices (36Kr, Aug 2025). Round size and investors are not disclosed.
• SEC filing mention. A Ramaco Resources 10-K/A exhibit (July 2026) surfaced in the search. It is a mining company disclosure, plausibly in a critical-minerals context. It is evidence that upstream resource firms discuss the area, not a funding event.

Market context (background knowledge, not from the retrieved sources; verify current status): FLOSFIA in Japan has raised multiple venture and corporate rounds, with Japanese strategic investors among its backers. Novel Crystal Technology has drawn Japanese government development support. Chinese Ga₂O₃ substrate start-ups have reported angel and Series A rounds from domestic funds. This report does not state amounts for any of these.

Trajectory

US federal investment in Ga₂O₃ is accelerating in ticket size while narrowing toward the substrate. Annual retrieved awards were $3.54M in 2015, $0.75M in 2019 and $2.28M in 2020. Spending then clustered at $1.51M in 2022, $5.0M in 2023 and $4.38M in 2024, followed by a single $8.52M award in late 2025. That last award is the largest in the set, and it is a production-oriented substrate contract rather than an SBIR. Private funding cannot be trended from two undated-amount records. A meaningful global total cannot be constructed from these records, and this report does not estimate one.

So what: The US has moved from exploratory SBIRs to a production-scale bet on a domestic wafer, which signals defense intent to avoid sole reliance on Japanese supply. Investors should note the funding gap that follows. Device and module programs are small (about 6% of awards), so the next government dollars are likely to flow downstream. Companies positioned to receive US-grown wafers stand to benefit.

Private Sector & Startup Watch

The commercial field is thin and Japan-weighted. The retrieved sources identify two commercial actors: one corporate–supplier equity link and one US startup. Neither is accompanied by a disclosed financing figure. A field with this few visible private rounds is early, and that is the relevant fact for an investor.

Startups to Watch

The retrieved USASpending rows do not name recipients. Linking specific SBIR awards to Kyma, Agnitron or others requires checking the award pages and should not be assumed.

Acquisitions and corporate moves

No acquisitions appear in the retrieved records. The Mitsubishi–NCT stake is the template for the field: a power-module incumbent secures wafer access without a full acquisition. Market context (background knowledge, not from the retrieved sources; verify current status): Japanese industrial groups, including Tamura and suppliers of crystal-growth furnaces, sit around NCT and FLOSFIA. No Western power-semiconductor major, such as Infineon, STMicroelectronics, onsemi or Wolfspeed, has a publicly visible Ga₂O₃ equity position comparable to Mitsubishi's.

So what: A Western power major that wants an option on Ga₂O₃ has few targets. Gallox and the recipient of the US EFG substrate contract are the obvious candidates. Early-stage investors face the reverse problem: the exit buyers are the SiC incumbents, who have every incentive to delay a cheaper rival until their SiC fabs are depreciated.

Geographic & Geopolitical Analysis

Ga₂O₃ is a three-way contest. Japan holds the wafer, China holds the gallium and a large patent and paper base, and the US holds defense demand and device science. Europe contributes crystal-growth science but has no visible commercial champion.

The retrieved records break down as CN 15, JP 11, US 10, KR 4, IN 2, DE, EP and ES 1 each, and 21 unknown. This reflects database coverage, which is weighted toward US and European sources, not national research share. China's lead in the retrieved count despite that weighting is still notable.

By bloc, based on named institutions and programs

• Japan. This is the origin of the field, through the Higashiwaki/Sasaki/Kuramata line. Japan holds the only retrieved corporate equity link (Mitsubishi–NCT) and the α-phase patent family (JP6422159B2, EP3291284B1).
• United States. All retrieved government funding is US: Air Force, Navy and NASA. Research comes from Florida/NRL (Pearton et al.), Ohio State (Krishnamoorthy et al.), the ISEFET group and George Mason (Maimon and Li). Gallox is the startup.
• China. Retrieved work includes vertical FET process patents, heterojunction and 2D p-type hybrid patents, a 2019 800 V MOSFET (Feng et al.), liquid-metal p-doping (Li et al.) and radiation phase-transformation work co-authored with Shandong University and US partners (Han et al.).
• Europe. The bulk melt-growth patent (ES2746068T3) is the only retrieved European IP signal. There is no EU funding record in the pool, which is a real gap.
• India and Korea. India contributes concept-level work (Mehta and Avasthi, IISc; Kachhawa et al.). Korea appears as a growth-equipment and patent filing jurisdiction.

Export controls and feedstock (background knowledge, not from the retrieved sources; verify current status)

• In August 2022, the US Bureau of Industry and Security (BIS) placed Ga₂O₃ and diamond substrates under multilateral (Wassenaar) export controls as ultrawide-bandgap materials. Ga₂O₃ is therefore already formally treated as strategic.
• China imposed export licensing on gallium and germanium from August 2023 and moved to bar gallium exports to the US in December 2024. China produces the large majority of the world's primary gallium.
• Iridium, used in EFG and Czochralski crucibles, is mined predominantly in South Africa. This is a second, less-discussed chokepoint.

Talent flow

Co-authorship in the retrieved set links Chinese and US institutions in DOE-funded work (Han et al.; Xing et al., with Macau, Monash and NTU authors). KPsphere assessment: given BIS controls on the substrate, US-funded collaborations with Chinese co-authors on Ga₂O₃ materials science are likely to face growing research-security review.

So what: A Western Ga₂O₃ supply chain needs three independent inputs: non-Chinese gallium, a non-Japanese wafer second source, and a reduced-iridium growth method. Today the retrieved evidence shows funding only for the second. National-security advisors should treat gallium recovery and crucible-free growth as part of the same problem.

Regulatory & Policy Landscape

No regulation targets Ga₂O₃ devices specifically. Adoption will be governed by export controls on the substrate, feedstock controls on gallium, and automotive and grid qualification standards written for silicon, SiC and GaN.

• Export controls. The US/Wassenaar substrate controls (background knowledge, verify current status) add licensing friction to cross-border wafer trade and to joint development with non-allied partners.
• Feedstock policy. Chinese gallium restrictions (background knowledge, verify current status) raise input-cost uncertainty. Any Western gallium-recovery capacity from alumina refining or zinc processing becomes a policy lever.
• Standards. JEDEC's JC-70 wide-bandgap committee has guidance for GaN and SiC. Automotive discretes qualify under AEC-Q101, and grid equipment under IEC and IEEE converter standards. Background knowledge, verify current status: no Ga₂O₃-specific JEDEC reliability guideline is known to this analysis. Writing one would be a clear acceleration signal, because test methods for threshold drift and dynamic on-resistance differ by material.
• Industrial policy. US CHIPS-era funding and DoD Defense Production Act-style authorities could support domestic substrate capacity. The $8.52M EFG contract is consistent with that intent. No pending Ga₂O₃-specific legislation appears in the retrieved records.
• Ethics and dual use. The material's stated applications include pulsed power for avionics and electric ships (Pearton et al.) and microwave electronics (SBIR II, RF epitaxy). That puts it squarely within defense-relevant electronics.

So what: Policy is more likely to delay commercial adoption, through export friction and gallium price shocks, than to accelerate it. The exception would be a dedicated qualification standard or a domestic-sourcing preference in defense power procurement, either of which would pull demand forward.

Risk Assessment

The dominant risks are technical (thermal management and the absence of p-type doping) and supply-side (gallium and iridium). Demand risk is moderate because SiC is improving and getting cheaper at the same time.

So what: Any investment or program should be conditioned on thermal-packaging and ruggedness data, not on material figures of merit. Supply-chain mitigation, meaning gallium recovery and crucible alternatives, deserves funding in parallel, not after device success.

Strategic Recommendations

For investors

Ga₂O₃ is investable at seed and Series A for substrate and epitaxy plays with defense anchor customers. It is not yet investable at growth stage for device companies. The highest risk-adjusted position is in wafer supply, consistent with where the US government is putting about 72% of its retrieved dollars. Diligence questions for Gallox or any device startup: wafer source and its export-control status; thermal path, whether top-side or bonded to a high-conductivity substrate; avalanche and short-circuit data; and IP position against fin-vertical claims such as US12243921B2. Plan exits around strategic buyers: SiC majors and Japanese module houses.

For government

• Fund downstream. Pair the EFG substrate contract with a device and packaging program that guarantees offtake of domestic wafers. Devices received only about 6% of retrieved awards.
• Fund the chokepoints. Support iridium-free melt growth and domestic gallium recovery as named priorities.
• Radiation qualification. Commission a Ga₂O₃-specific radiation-effects campaign that builds on Han et al. before space-power insertion (NASA SBIR lineage).
• Allied coordination. Use the US–Japan relationship to secure second-source wafer access from NCT while domestic capacity matures.

For industry

Partner rather than build. SiC device makers should take a minority position or a joint development agreement with a wafer supplier within 12–24 months, following the Mitsubishi–NCT template. Building in-house crystal growth is justified