Extreme-band-gap semiconductors with shallow dopants and mobile carriers
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | npj Computational Materials |
| Published | 2026-07-21 |
| Authors | Sieun Chae, Nocona Sanders, Kelsey Mengle, A. Wang, Xiaoli Zhang |
| DOI | 10.1038/s41524-026-02173-z |
Computational screening identified extreme-band-gap semiconductors with gaps as wide as 9.50 eV that retain shallow dopants and mobile carriers. The main candidates are rocksalt MgO, zinc-blende BeO, rutile SiO₂, and $Mg_xBe_{1-x}O$ alloys. These are calculated predictions rather than experimental demonstrations of the proposed doped devices.
Gap size alone does not define semiconducting behavior
Section titled “Gap size alone does not define semiconducting behavior”Materials with gaps above 3 eV are often treated as insulators, but the authors test whether dopants ionize and carriers remain mobile. AlN at 6.2 eV is the wide-gap semiconductor reference. Prior p-type candidates reached 4.5 eV. The new compounds exceed that limit, with zb-BeO reaching 9.50 eV. Their proposed semiconducting behavior depends on light effective masses, dielectric screening, and weak polaron binding.
The authors judge semiconducting behavior from dopant ionization and the persistence of mobile carriers. Their results reject a fixed 3 eV cutoff, but they do not claim that every material above that cutoff is a usable semiconductor.
Chemical structure controls whether wide gaps remain useful
Section titled “Chemical structure controls whether wide gaps remain useful”The authors screened oxides, nitrides, and carbides using hybrid density functional theory. HSE06 band structures used 25 to 35% Hartree-Fock exchange, selected separately for each material to approach measured gaps. Structures were relaxed until forces fell below 0.01 eV/Å. Effective masses came from nonparabolic fits near the band edges. Static dielectric constants came from density functional perturbation theory.
Lighter cations generally produced wider gaps. Large, spherical cation s orbitals created dispersive conduction bands and lighter electrons. Dense anion packing increased orbital overlap and reduced hole masses. Rocksalt and rutile structures generally packed anions more densely than wurtzite and zinc-blende structures. Main-group cation s orbitals overlapped isotropically, while anion 2p valence bands usually produced heavier holes. Mixed cation s or d character in rs-ZnO, wz-ZnO, and GaN supported lighter holes.
Conduction bands in binary compounds mainly follow the lowest unoccupied cation orbital. Main-group cations use s orbitals, heavy p-block cations use p orbitals, and transition metals and rare earths use d or f orbitals. This orbital character helps explain why light main-group compounds can retain dispersive bands at wide gaps. BeO polytypes, MgO, γ-Al₂O₃, and rutile SiO₂ combined gaps above 7 eV with electron masses below 0.7 mₑ and hole masses below 2 mₑ.
The ionization screen narrows the candidate list
Section titled “The ionization screen narrows the candidate list”The hydrogenic Bohr model converted effective masses and dielectric constants into donor and acceptor ionization energies. Ionization energy scales inversely with effective mass and with the square of dielectric screening. Predictions agreed with available measurements for known semiconductors. The authors also cite experimental confirmation of their earlier rutile GeO₂ donor prediction.
At room temperature, $k_BT$ is about 25 meV, so dopants must ionize on a comparable energy scale to contribute free carriers. The screen highlighted shallow donors below 0.21 eV in rs-ZnO, r-GeO₂, rs-MgO, r-SiO₂, and BeO polytypes. Rutile binary oxides formed an exception to the usual trend because their smaller lattices increased oxygen packing and reduced hole masses. Rocksalt BeO had a static dielectric constant of 167, but phonon calculations later showed ambient dynamic instability.
Defect relaxation determines dopant depth
Section titled “Defect relaxation determines dopant depth”Hybrid defect calculations then tested dopants under O-rich/cation-poor and cation-rich/O-poor limits. Reference defect cells contained 64 to 72 atoms, while intentional dopant cells contained 192 to 216 atoms. Each dopant was introduced in a distorted configuration and relaxed. The authors compared distorted and undistorted substitutions to identify DX centers for donors and AX centers for acceptors. Finite-supercell corrections and secondary-phase calculations constrained the defect energetics.
The calculation tracked both the defect level and the relaxed atomic state. A stable DX center can capture an added electron and keep a donor deep. A stable AX center can bind a hole and raise acceptor activation energy. Avoiding those states was separate from having favorable band alignment.
The principal calculated properties are:
| Material | Calculated gap | Dopant result | Polaron formation energy |
|---|---|---|---|
| rs-MgO | 7.47 eV | F and Al-related donors below 0.07 eV | 22 meV, electron |
| zb-BeO | 9.50 eV | F donor at 0.78 eV; no DX center | 37 meV, electron |
| r-SiO₂ | 8.87 eV | Al acceptor at 0.13 eV; no AX center | 84 meV, hole |
The defect picture is directional. rs-MgO is an n-type candidate, but Li and Na acceptors are deep at 0.61 to 0.73 eV. Rutile SiO₂ is the p-type candidate. BeO end members have acceptor energies of 1.24 to 1.65 eV, which excludes useful p-type doping. The deep F level in zb-BeO leaves impurity-band conduction as its possible transport route.
Calculated mobilities test carrier transport
Section titled “Calculated mobilities test carrier transport”Transport calculations used G0W0 quasiparticle energies, electron-phonon coupling, and an iterative Boltzmann solution. Phonons and electron-phonon matrix elements came from density functional perturbation theory before interpolation onto fine grids. The transport calculation used states within 0.5 eV of the conduction band minimum, while carrier lifetimes used states within 0.25 eV.
At 300 K, calculated electron mobility was 141 cm²/V·s for rs-MgO and 391 cm²/V·s for zb-BeO. The rs-MgO value is comparable to β-Ga₂O₃ below 200 cm²/V·s and to $Al_xGa_{1-x}N$ at 136 to 200 cm²/V·s for x = 0.4 to 0.8. Rutile SiO₂ hole mobility is 100 cm²/V·s along a and 143 cm²/V·s along c.
The large polarons summarized in the table are described by the authors as effectively free carriers. P-type rs-MgO behaves differently. Its small hole polaron has a formation energy of 370 meV, but hopping barriers of 90 meV give about 0.25 cm²/V·s at room temperature. The calculated dissociation density is about 4 × 10²⁰ cm⁻³, which requires large supercells for accurate modeling.
Mg alloying makes BeO donors shallower
Section titled “Mg alloying makes BeO donors shallower”Mg alloying makes the F donor shallower in both BeO structures. In wurtzite $Mg_xBe_{1-x}O$, the F transition crosses the conduction band at 26% Mg, where the calculated gap is 8.33 eV. In zinc-blende $Mg_xBe_{1-x}O$, the crossing occurs at 15% Mg and a gap of 8.51 eV. A direct calculation for wz-Mg₀.₅Be₀.₅O supported the interpolation. The result gives the largest-gap material a donor pathway without making its F level shallow in pure zb-BeO.
Stability and experimental readiness remain uneven
Section titled “Stability and experimental readiness remain uneven”Single-crystal rs-MgO has been produced industrially by Czochralski growth, but n-type operation has not been demonstrated. The authors suggest compensation by Mg vacancies and hydrogen interstitials as the likely cause. Rutile SiO₂ can be synthesized by high-pressure, high-temperature methods, while $Mg_xBe_{1-x}O$ has only been made over limited compositions. Most other candidates remain at the synthesis-prediction stage.
Rocksalt BeO has imaginary phonon frequencies at ambient conditions. The authors report the rocksalt $Mg_xBe_{1-x}O$ alloy as stable when x is greater than 0.83. Wurtzite and zinc-blende $Mg_xBe_{1-x}O$ differ by only 0.26 meV/f.u. at x = 0 and less than 40 meV/f.u. at x = 0.25. The authors therefore suggest epitaxial templates or seed crystals for zinc-blende synthesis. They also note beryllium’s strict regulation because of toxicity. Alloy disorder effects on defects remain for future work.
Implications for diamond material selection
Section titled “Implications for diamond material selection”For diamond device engineers, the result changes the screening criteria. A wide gap alone does not establish useful conduction. Carrier effective mass, dopant ionization, dielectric screening, and polaron binding must be checked together. The paper’s rutile SiO₂ model predicts a lower acceptor ionization energy than diamond and cubic boron nitride, which offers a direct comparison between candidate wide-gap hosts.
The authors identify power electronics and vacuum-ultraviolet emission below 200 nm as target uses for extreme-gap materials. Those uses depend on realizing stable phases, controlling dopant incorporation, and measuring carrier transport. For diamond material selection, band gap is only one property. Ionized dopants and mobile carriers must also be demonstrated.
View Original Abstract
Abstract The conventional distinction between semiconductors and insulators is often based on the magnitude of the band gap, with materials exhibiting gaps wider than 3 eV typically classified as insulators. However, the emergence of ultra-wide-band-gap (UWBG) semiconductors such as AlGaN, diamond, BN, and β-Ga 2 O 3 challenges this paradigm for materials classification and raises fundamental questions about the upper bound of band gaps compatible with semiconducting behavior. Here we develop a computational-discovery strategy to identify semiconductors with band gaps exceeding that of AlN (6.2 eV), while retaining essential semiconducting properties such as shallow dopants and mobile charge carriers. Consistent with prior observations on individual materials, we find more generally that materials composed of light elements in densely packed crystal structures exhibit wide band gaps and light carrier effective masses that enable shallow dopants, high mobility, and weak polaron binding. By applying the hydrogenic Bohr model and first-principles defect calculations - validated against available experimental data - to screen for materials with shallow dopants, we identify dopable compounds with gaps as wide as 9.5 eV that nonetheless host mobile charge carriers. Our findings demonstrate that semiconducting behavior persists even at extreme band gaps, far beyond conventional upper bounds traditionally associated with semiconductor materials.
Boron Doped Single Crystal Diamond Plate
View Specs →
Free Standing Boron Doped Diamond Pcd Film
View Specs →