Cornell Develops Oxide Semiconductor With a Bandgap Above 7 eV and Builds Working Transistors

In power grids and electric-vehicle inverters, a key challenge is to withstand ever-higher voltages while reducing the energy lost as heat. The electric field a semiconductor can endure before dielectric breakdown is closely tied to its bandgap, the energy needed for an electron to move from the valence band to the conduction band. In general, the wider the bandgap, the higher the field a material can withstand. But the material also becomes more like an insulator, and it becomes harder to supply, through doping, the electrons needed to carry current. This physical trade-off has long constrained the development of higher-performance semiconductor materials.

A research team centered at Cornell University has announced results that push past this limit. The team developed a silicon-doped aluminum gallium oxide thin film with a corundum structure that has a bandgap above 7.0 electron volts (eV) while still allowing electrical conduction to be controlled. They also built a diode and a field-effect transistor from the material and demonstrated that they work. The results were published in the journal Nature on October 7, 2026, as the peer-reviewed paper "A 7-eV bandgap semiconductor based on silicon-doped α-(AlxGa1−x)2O3," and Cornell announced them on November 3.

The team defines the region where the bandgap exceeds 6.0 eV as the "colossal-bandgap" regime. The study experimentally shows that electron conduction can be controlled, and active devices operated, in semiconductors in this regime. However, the fact that the prototype devices work does not mean they can immediately replace the silicon carbide (SiC) and gallium nitride (GaN) now in commercial use. Below, we look at what the result means for materials science and at the challenges that remain before it can be applied to practical power semiconductors.

{
  "type": "bar",
  "title": "主要半導体材料のバンドギャップ比較",
  "unit": "eV",
  "categories": ["Si", "4H-SiC", "GaN", "β-Ga2O3", "α-Ga2O3", "ダイヤモンド", "c-BN", "α-(Al,Ga)2O3"],
  "series": [
    { "name": "バンドギャップ", "values": [1.1, 3.2, 3.4, 4.8, 5.4, 5.5, 6.4, 7.0] }
  ],
  "caption": "文献値および今回の研究で測定されたチャネル層の代表値。新材料のバンドギャップは7.0 eVを超える。",
  "source": "Nature (2026) 論文および付録資料"
}

As this comparison shows, the newly developed film sits in the region of especially wide bandgaps among known semiconductor materials.

Measurements Above 7 eV Break Through the Insulator Barrier

In a semiconductor's energy structure, the forbidden gap that separates the valence band from the conduction band (the bandgap, \(E_g\)) is a fundamental property that governs a material's electrical and optical behavior. The bandgap of ordinary silicon (Si) is about 1.1 eV, while that of 4H-SiC, increasingly adopted for high-voltage power semiconductors, is about 3.2 eV and that of GaN is about 3.4 eV. The wider a material's bandgap, the higher the critical electric field strength at which avalanche breakdown occurs inside the crystal. This makes it potentially possible to design devices that withstand high voltages with thinner semiconductor layers.

But the wider the bandgap, the harder it is to secure the electrons needed to carry current. Even if donor impurities that supply electrons are introduced into the crystal, if their energy level lies far below the bottom of the conduction band, room-temperature thermal energy has difficulty promoting electrons into it. Another problem is "compensation," in which unintended impurities and point defects cancel out the donors' effect. For this reason, in crystalline materials with bandgaps above 6.0 eV, it has been extremely difficult to obtain sufficient conductivity through doping, and in many cases they have behaved essentially as insulators.

According to the paper, cubic boron nitride (c-BN, about 6.4 eV) has been cited as a conventional semiconductor with an especially wide bandgap. The team focused on α-(AlₓGa₁₋ₓ)₂O₃ (alpha-phase aluminum gallium oxide), an alloy combining aluminum oxide (sapphire, about 8.6 eV) and gallium oxide (α-Ga₂O₃, about 5.4 eV). The bandgap of this material can be tuned by changing the ratio of aluminum to gallium. The team added silicon while keeping the crystal structure in the α phase, producing a thin film capable of electron conduction.

As a result, the silicon-doped α-(AlₓGa₁₋ₓ)₂O₃ films showed a bandgap above 7.0 eV. The paper also reports that, compared only against materials with bandgaps above 6.0 eV, room-temperature conductivity exceeded previously reported values by more than 100 million times, or eight orders of magnitude. However, the paper's abstract gives only the relative improvement factor; it does not state absolute conductivity values for either the comparison materials or the new material.

It is important to note that this "100 million times" figure does not mean an improvement in conductivity over silicon or GaN. In earlier materials above 6 eV, very few electrons were available at room temperature, and almost no current could flow. The result shows that even such extreme materials can attain conductivity sufficient to serve as the channel of an electronic device.

Suboxide Molecular-Beam Epitaxy and Multilayer Growth Techniques

How was the team able to make current flow in an oxide alloy that had previously behaved almost like an insulator? Doctoral student Jacob Steele and colleagues, who handled film growth, used a crystal growth method called suboxide molecular-beam epitaxy. In this technique, source materials are supplied as beams in an ultra-high vacuum, and crystals are grown while the film's composition and thickness are precisely controlled.

The central challenge was to widen the bandgap beyond 7 eV by increasing the aluminum fraction while still creating a state in which the added silicon supplies electrons. When crystals are grown at high temperature, compensation by defects and thermodynamic constraints can prevent enough usable electrons from being produced. By carefully controlling the growth temperature and incorporating silicon into the crystal lattice at a relatively low temperature, the team succeeded in securing electrons that carry current even at room temperature.

That said, the entire film was not grown at a single temperature. According to the paper's Supplementary Information, device fabrication used multi-step temperature control combining high-, medium-, and low-temperature growth steps. In the representative evaluation sample "Sample A1," an aluminum-rich buffer layer was formed on the substrate, and a silicon-doped \(\alpha\text{-(Al}_{0.51}\text{Ga}_{0.49}\text{)}_2\text{O}_3\) channel layer was grown on top of it. In other words, not every layer in the device has the same bandgap; the device has a multilayer structure designed with crystal growth and electrical properties in mind.

Resistance in injecting electrons from the electrodes was also a major technical challenge. In wide-bandgap semiconductors, the Schottky barrier formed at the interface with a metal electrode tends to be high, making it hard for electrons to flow from the electrode into the semiconductor. To address this, the team combined, directly beneath the transistor electrodes, regrown heavily doped n-type (n+) regions with reduced aluminum content for higher conductivity, and α-Ga₂O₃ contact layers. Dedicated structures were needed not only for the channel material but also for the connections to the electrodes.

To measure the bandgap, the team used spectroscopic ellipsometry, which derives a material's optical constants from its reflection characteristics. Taking the crystal's optical anisotropy into account, they acquired data using combinations of three azimuthal angles and incidence angles of 50, 60, and 70 degrees, and performed fitting analysis. Electrical properties were evaluated by Hall effect measurements, and secondary ion mass spectrometry (SIMS), atom probe tomography (APT), and transmission electron microscopy (TEM) were also used. These analyses examined the silicon concentration distribution inside the film, the atomic arrangement of the crystal lattice, and local crystal defects.

The "AlphaFET" Prototype Transistor: Results and Limits

What is especially important about this study is that the team did not merely measure the material's conductivity but built prototype active devices that can actually control current. Under the guidance of Professors Huili Grace Xing and Debdeep Jena, doctoral student Debaditya Bhattacharya, a co-first author, and colleagues handled device design, fabrication, and electrical measurement.

The team fabricated a Schottky diode called "AlphaDiode" and a field-effect transistor called "AlphaFET," both using the colossal-bandgap material. According to Cornell's announcement, in the AlphaFET, applying voltage to a capacitively coupled metal gate electrode modulated the drain current by more than four orders of magnitude, or more than 10,000 times. However, the university's materials give only the current modulation ratio; they do not state the specific on- and off-state currents or measurement conditions such as the applied voltages.

The academic significance of this result should be separated from any assessment of practical performance. What has been directly demonstrated is the principle that even in a material with a bandgap on the order of 7 eV, electrons can be supplied by impurity doping and current can be controlled using the field effect. It shows that materials in a regime long thought to be close to insulating could potentially be used in active electronic devices.

On the other hand, the high power density, fast switching performance, energy conversion efficiency, and long-term reliability required of commercial power semiconductors have not been demonstrated. The prototype devices are tiny research test devices, and factors that limit performance remain, including parasitic resistance of the electrodes, thermal properties, and the magnitude of current that can actually be passed. Current modulation characteristics confirmed in prototype devices on the scale of a few micrometers cannot be directly linked to the performance of commercial high-voltage devices or large power modules. In addition, the results are the research team's own report published in a peer-reviewed journal, and independent replication by third parties and verification of manufacturing yield remain challenges for the future.

Nevertheless, the result is considered to have substantially changed the research and development challenge. The biggest hurdle had been whether controllable electron conduction could be produced at all in materials with bandgaps above 6 eV. With that principle now demonstrated, researchers can turn to more concrete device-engineering challenges, such as reducing crystal defects, lowering the resistance of electrodes and interfaces, and establishing practical device structures.

The Gap Between Theoretical Baliga Figure of Merit and Measured Performance

A metric often used to assess the potential performance of ultra-wide-bandgap semiconductors is Baliga's figure of merit (BFOM). It is a theoretical index that evaluates the trade-off between on-resistance and breakdown voltage in power semiconductors, mainly assuming low-frequency operation. Using the material's permittivity (\(\epsilon\)), electron mobility (\(\mu\)), and critical electric field strength (\(E_c\)), it is expressed by the following relation.

\[ \mathrm{BFOM}\propto\epsilon\mu E_c^3 \]

Because the critical electric field strength depends strongly on the bandgap, in approximations that include other material properties BFOM is said to be roughly proportional to the bandgap to the 5.5th power (\(E_g^{5.5}\)). In other words, widening the bandgap could in theory greatly improve the performance of power semiconductors. However, this relationship is for evaluating a material's potential performance and does not directly represent actual device performance.

The paper's Supplementary Information (Supplementary Table 1) summarizes the properties of various semiconductor materials and their theoretical relative BFOM, with silicon set to 1.

The relative BFOM values in the table, ranging from tens of thousands to millions of times, are theoretical indices that assume ideal material properties. They are not comparisons of the conversion efficiency or performance of actual devices, nor do they directly show which material is better. In particular, the 6.5-million-times figure for sapphire (\(\alpha\text{-Al}_2\text{O}_3\)) is merely a reference value calculated from theoretical critical field strength and similar quantities. This performance has not been demonstrated in any conductive transistor made from sapphire.

For the newly developed \(\alpha\text{-(Al}_x\text{Ga}_{1-x}\text{)}_2\text{O}_3\) as well, the critical field strength, permittivity, and upper limit of electron mobility for each composition are not yet sufficiently understood. For that reason, it is not currently possible to calculate a definitive BFOM or to state numerically a theoretical performance advantage over SiC and GaN. A wide bandgap certainly leads to great potential, but how much of that potential can be realized in actual devices is a separate question.

On the manufacturing side, however, there is a notable advantage. The film growth and device prototyping in this work used sapphire (single-crystal aluminum oxide) as the substrate. For ultra-wide-bandgap materials such as diamond and cubic boron nitride, it is difficult to produce large-area, high-quality single-crystal wafers at low cost. Sapphire, by contrast, has long been used to make blue LEDs and others, and wafers on the scale of 6 to 8 inches in diameter are mass-produced. The ability to use existing substrate manufacturing technology is an advantageous condition for future scale-up.

Still, the relatively easy availability of inexpensive sapphire substrates does not guarantee low overall manufacturing cost for the semiconductor device. This technology requires growing oxide thin films with lattice constants different from sapphire's under multiple temperature conditions, and also forming regrown layers for electrode contact. Challenges remain in such complex epitaxial growth processes, including growth rate, crystal quality, and manufacturing yield. What matters is whether the advantage of substrate availability can be tied to a semiconductor manufacturing process capable of mass production.

Crystal Defects and Electrical Properties: Challenges Before Practical Use

Several problems remain to be solved before the result can lead to practical devices such as power semiconductors and optical communications. In Cornell's announcement, Professor David Anthony Muller and colleagues, who handled microstructural analysis by electron microscopy, reported finding many crystal defects inside the fabricated films. Senior author Professor Darrell Schlom has also compared the material's current state to that of other semiconductors in their early stages of development, explaining that defects must be controlled and improved to unlock its true performance.

The paper's Extended Data Fig. 1 shows misfit dislocations in the AlphaFET's layer stack. These are disruptions in crystal arrangement that arise because adjacent crystal layers have different lattice constants. Such dislocations act as scattering sources that impede electron movement and, depending on conditions, can also become paths for leakage current. Reducing defect density by adjusting film thickness, composition, and growth temperature is therefore essential for improving device performance and reliability.

The crystal quality problem also shows up in the electron mobility measurements. According to Hall effect measurements of "Sample A27" in the Supplementary Information, the electron mobility at room temperature (300 K) was only \(2\pm1\ \mathrm{cm^2/(V\cdot s)}\). This is extremely low compared with the several hundred \(\mathrm{cm^2/(V\cdot s)}\) obtained in commercial SiC and roughly 1,000 to 2,000 \(\mathrm{cm^2/(V\cdot s)}\) in GaN. With low electron mobility, electrons move less readily under the same electric field, and resistance to current flow increases. Thus, beyond the potential for high breakdown voltage offered by the wide bandgap, whether sufficient current can be passed with low loss is also an important evaluation criterion.

Temperature dependence also presents challenges. According to the Supplementary Information, when the sample was cooled from 300 K to 200 K, sheet resistance increased about tenfold and electron concentration fell at the same time. However, the document gives only an approximate factor for the change in resistance, not specific sheet-resistance values at each temperature. The team links this phenomenon to "carrier freeze-out," in which, as temperature falls, electrons remain on donor levels and no longer contribute to conduction. Because practical power semiconductors must operate stably over a wide range of temperatures, electrical properties from low to high temperatures need to be examined in more detail.

As Professor Debdeep Jena states in Cornell's announcement, further research and development is essential to raise these devices to practical performance. The university cites as future applications aerospace and ground-transport propulsion systems, ultra-high-voltage power supplies for data centers, quantum cascade lasers for communications, and vacuum ultraviolet optical devices. These, however, are future application concepts based on the new material's physical properties, and practical use has not been achieved at this point.

To judge whether this material becomes a leading candidate for next-generation power semiconductors, three main developments will need to be confirmed. First, how far can the lattice mismatch with the sapphire substrate be relaxed, and the densities of threading and misfit dislocations reduced? Second, how far can contact resistance be lowered by improving ohmic contacts to the electrodes, and electron mobility, now only a few \(\mathrm{cm^2/(V\cdot s)}\), be raised? Third, can uniform films be formed on large-area wafers, and can breakdown voltage, power density, fast switching performance, thermal properties, and long-term reliability be verified at practical device sizes? It is also essential to confirm, through replication by third parties, the reproducibility of the reported material properties and device operation.

The demonstration of transistor operation in a semiconductor with a bandgap above 7 eV is an important advance in solid-state physics and materials engineering. It shows that a material in which even passing current was once difficult could be used as an active electronic device. Whether that potential can be put to use in real power infrastructure, however, depends on reducing crystal defects, improving electrode structures, establishing manufacturing processes, and verifying performance under practical conditions. The result is not a finished commercial power semiconductor but a starting point for full-scale device development in a new materials regime that had been hard to use until now.