Cornell has made aluminium gallium oxide conduct at record bandgaps and built test devices. Sapphire offers a manufacturing foothold, but defects and low mobility keep commercial power electronics distant.
The useful clue in Cornell University’s latest semiconductor work sits in a table labelled A27. It is one of 27 thin-film samples whose growth temperatures and material flows are listed in the study’s supplementary information. At room temperature, that film carried only a modest electron mobility, a reminder that the story is as much about a difficult manufacturing problem as a record.
Jacob Steele grew the films in Ithaca, New York, using suboxide molecular-beam epitaxy. Debaditya Bhattacharya then patterned and measured devices under Huili Grace Xing and Debdeep Jena. Their material, silicon-doped alpha-phase aluminium gallium oxide, has a bandgap exceeding 7 electron volts in some films. A Schottky diode and a field-effect transistor called AlphaFET showed that this unusually insulating territory can still support controllable electronic devices.
An electron volt here measures the energy separating electronic bands, not seven volts of output or a seven-volt power rating. Wider bandgaps generally allow stronger electric fields before breakdown, which could eventually mean smaller, more efficient power conversion. Cornell reported room-temperature conductivity more than 100 million times higher than previous reports for materials above 6eV. The transistor showed more than four orders of magnitude of gate-controlled current modulation.
The record is not a product specification
Those results do not make a commercial 7eV transistor. The highest-bandgap films and the particular device channel should not be treated as interchangeable, and the team’s mobility figures remain far below those of mature power semiconductors. David Muller and Naomi Pieczulewski found numerous crystallographic defects. Contact resistance, repeatable growth, heat management and demonstrated breakdown behaviour under useful operating conditions are still engineering work, not solved specifications.
The choice of substrate is nevertheless commercially interesting. The films sit on sapphire, already produced in large sizes for other industries, rather than a scarce specialist crystal. That removes one obstacle, but not the cost and throughput of molecular-beam epitaxy or the difficulty of reducing defects across a wafer. The paper was published in Nature on 7 October; Cornell’s announcement followed on 3 November, so this is fresh institutional coverage of month-old research.
Who has exposure
Today’s power-chip suppliers have little reason to treat this as an immediate substitute. Infineon, onsemi and Wolfspeed have substantial silicon-carbide franchises, while Navitas and others sell gallium-nitride devices. In a long-term scenario where Cornell’s material achieves reliable high-voltage switching with competitive yields, those portfolios would face a new materials platform. Their manufacturing, packaging and customer qualifications would remain valuable, however, and could just as easily help an incumbent adopt it.
A nearer-term beneficiary would be research and semiconductor equipment suppliers if laboratories order more epitaxy, characterisation and fabrication capacity. That is a scenario, not an announced order. No company funding round or commercial licence tied to this paper was disclosed in the sources reviewed.
The intellectual property is more concrete. The Nature paper says Steele, Bhattacharya, Xing, Jena and senior author Darrell Schlom filed US provisional application 63/786,244 covering conductive doped films and device structures. Schlom also holds granted US patent 11,462,402 for suboxide molecular-beam epitaxy and related structures. A provisional filing is not an issued patent or evidence of a product business.
The money behind the science is mainly public. The acknowledgements list Air Force Office of Scientific Research and Air Force Research Laboratory support through award FA9550-18-1-0529, National Science Foundation facilities and PARADIM cooperative agreement DMR-2039380, alongside other university and government grants. That backing matters because improving an entirely new semiconductor can take years before a device maker can justify a production programme.
The next meaningful milestones are narrower than the launch language suggests: lower defect densities, better electron transport, reproducible diodes and transistors with credible voltage and efficiency measurements, then larger wafers. Only after those results would it be sensible to compare the material with qualified silicon-carbide or gallium-nitride components in vehicles, aircraft, grids or data centres. For now, Cornell has opened a materials route, not shortened the qualification cycle.