Skip to content

Two‐Dimensional Carrier Gas at Diamond and Cubic Boron Nitride (111) Interface - Role of Interfacial Bonding and Defects

FieldValue
JournalAdvanced Electronic Materials
Published2026-10-27
AuthorsShikha Saini, Konstantin Klyukin, M. Liu, Mark J. Polking, Bilge Yildiz
DOI10.1002/aelm.70623

Hybrid DFT predicts that interfacial termination selects the carrier type at a diamond/cubic boron nitride (111) interface. The same interface can host either a two-dimensional hole gas or a two-dimensional electron gas. The highest calculated density is 1.33 × 10¹⁵ cm⁻² for a defect-free three-bond interface, nearly three times the 4.44 × 10¹⁴ cm⁻² of the corresponding one-bond configuration. HSE06 results attribute this difference to the number of polarity-mismatched interfacial bonds.

Electron counting gives the carrier origin. Bulk sp3 C-C and B-N bonds each contain two electrons. An abrupt C-B bond contains 1.75 electrons, so it is deficient by 0.25 electron per bond. An abrupt C-N bond carries the same excess in the opposite direction.

The calculated electronic states match this imbalance. C-B interfaces have acceptor-like unoccupied states near the Fermi level and form a two-dimensional hole gas. C-N interfaces have donor-like occupied states near the Fermi level and form a two-dimensional electron gas. The authors therefore tie carrier type to interfacial polarity rather than chemical doping.

Four bonding configurations define the interface

Section titled “Four bonding configurations define the interface”

The models cover B-terminated and N-terminated c-BN surfaces. Each termination has two local bonding environments. A B atom can bond to one C atom and three N atoms, labeled 1C-B-3N, or to three C atoms and one N atom, labeled 3C-B-1N. An N atom can bond to one C atom and three B atoms, labeled 1C-N-3B, or to three C atoms and one B atom, labeled 3C-N-1B.

The in-plane lattice constants are averaged to reduce the approximately 1.35% mismatch between diamond and c-BN. The authors newly identified the two three-C configurations. Their larger number of C-B or C-N bonds explains their higher carrier density.

The authors ran spin-polarized density functional theory in VASP through ASE, using the projector-augmented wave method. PBE relaxation used a 500 eV cutoff, a 7 × 7 × 1 k-point grid, a 0.01 eV/Å force limit, and a 10⁻⁶ eV energy limit. HSE06 electronic calculations used an exact-exchange fraction of 0.32 and a 450 eV cutoff. They used 6 × 6 × 1 and 3 × 3 × 1 grids for the 1 × 1 and 2 × 2 superlattices. Tetrahedron smearing was used for projected density of states.

The model checks against known bulk values. Calculated lattice constants were 3.57 Å for diamond and 3.61 Å for c-BN. The corresponding experimental values were 3.567 Å and 3.615 Å. HSE06 band gaps were 5.61 eV and 6.12 eV, close to experimental ranges of 5.46 to 5.60 eV and 6.10 to 6.40 eV.

Hydrogen and fractionally charged pseudo-hydrogen atoms passivate the vacuum-side surfaces. A dipole correction removes spurious vacuum fields and aligns the electrostatic potential. Antisite defects were placed in a 2 × 2 interfacial supercell. One antisite among 16 sites represents 6.25% atomic intermixing, with an areal density of 1.11 × 10¹⁴ cm⁻².

Carrier density follows the number of polar bonds

Section titled “Carrier density follows the number of polar bonds”

Projected density of states and integrated charge densities give the following defect-free results.

InterfaceInterfacial bond environmentCarrier gasCarrier density, cm⁻²
1C-B-3NOne C-B bondTwo-dimensional hole gas4.44 × 10¹⁴
3C-B-1NThree C-B bondsTwo-dimensional hole gas1.33 × 10¹⁵
1C-N-3BOne C-N bondTwo-dimensional electron gas4.44 × 10¹⁴
3C-N-1BThree C-N bondsTwo-dimensional electron gas1.33 × 10¹⁵

The integrated counts agree with electron counting. The 3C-B-1N cell contains 12 C-B bonds and about three holes. The 1C-B-3N cell contains 4 C-B bonds and about one hole. The C-N interfaces show the reverse electron counts. Carrier density approaches zero away from the interface.

The two-dimensional character is supported by spatial confinement. The three-C interfaces confine carriers within about 1.2 nm. The one-C interfaces extend to about 1.8 nm. Calculated interfacial effective masses are 0.41 to 1.42 m0 for holes and 0.32 to 3.00 m0 for electrons.

Adhesion and band offsets determine confinement

Section titled “Adhesion and band offsets determine confinement”

More positive adhesion energy means stronger binding in this work. HSE06 gives 1.25 eV/Ų for 3C-B-1N and 1.09 eV/Ų for 3C-N-1B. The one-C values are 0.67 eV/Ų for 1C-B-3N and 0.51 eV/Ų for 1C-N-3B. Thus, the three-C interfaces are more stable within each termination. The C-B interfaces are also more stable than their C-N counterparts.

HSE06 produces type-I alignment for 1C-B-3N, 3C-B-1N, and 3C-N-1B. It produces type-II alignment for 1C-N-3B. The large positive valence-band offset of the C-B interfaces supports strong hole confinement on the diamond side. The 1C-N-3B interface has a 0.23 eV conduction-band offset and confines electrons on the c-BN side. The 3C-N-1B offset is −0.14 eV, which gives weak electron confinement and is detrimental to device performance. PBE labels all four interfaces type-I, but the authors attribute that result to underestimated band gaps.

Antisite defects trade carrier density for stability

Section titled “Antisite defects trade carrier density for stability”

The C-B interfaces favor the n-type CB and NC defects under N-rich conditions. The authors state that negative formation energies allow these defects to form spontaneously during growth, with concentrations controlled by growth conditions. The BC antisite has higher formation energy and introduces mid-gap states at the C-B interfaces.

The 1C-N-3B interface favors the p-type BC and CN defects under both B-rich and N-rich conditions. The NC defect has high formation energy and causes structural distortion and mid-gap states at this interface. Antisites cause large distortion and high formation energies in the 3C-N-1B interface.

Thermodynamically favored defects reduce the carrier density. The hole density of 1C-B-3N falls from 4.44 × 10¹⁴ to 3.32 × 10¹⁴ cm⁻². The hole density of 3C-B-1N falls from 1.33 × 10¹⁵ to 1.22 × 10¹⁵ cm⁻². The electron density of 1C-N-3B also falls from 4.44 × 10¹⁴ to 3.32 × 10¹⁴ cm⁻². The authors summarize these reductions as about 25% for the one-C interfaces and about 8% for the three-C interfaces. The remaining density stays on the order of 10¹⁴ to 10¹⁵ cm⁻².

Benchmarks, limits, and device implications

Section titled “Benchmarks, limits, and device implications”

The reported density extends the range reported for earlier diamond/c-BN interfaces.

Diamond/c-BN interface studyCarrier density, cm⁻²Exchange-correlation functional
(111), prior study [71]2.03 × 10¹⁴PBE
(111), prior study [45]6.57 × 10¹³ to 2.24 × 10¹⁴PBE
(111), prior study [31]4.17 × 10¹⁴HSE06
(111), present work3.32 × 10¹⁴ to 1.33 × 10¹⁵HSE06

The one-C densities exceed the calculated 6.8 × 10¹³ cm⁻² of GaN/AlN and the experimental 10¹¹ to 10¹³ cm⁻² range reported for wide-bandgap heterostructures. They remain comparable to the roughly 10¹⁴ cm⁻² reported for AlN and ScAlN interfaces. The three-C configurations reach the 10¹⁵ cm⁻² order, while prior diamond/c-BN (100) calculations reached 1.6 × 10¹⁴ to 4.1 × 10¹⁴ cm⁻².

The evidence remains computational. The work does not synthesize the interfaces or measure transport, mobility, or breakdown strength. Calculating absolute stability for the polar interfaces also requires surface energies and pseudo-hydrogen chemical potentials that the authors leave for future work. Growth remains difficult because cubic boron nitride competes with hexagonal and wurtzite phases. Stacking faults can also create rotated domains and wurtzite material near the interface.

For device engineering, the calculations support termination control as a route to either p-type or n-type conduction. The authors infer that current-carrying capacity rises with carrier density, while on-resistance falls with it. They propose that higher density could reduce ohmic dissipation and thermal-runaway risk in high-power and RF devices. Experimental growth must still control the three-C or one-C termination and limit defect states.

For c-BN-on-diamond (111) interface development, the relevant catalog material is single-crystal diamond grown by MPCVD, supplied as plates and wafers. The paper’s model needs a defined (111) diamond surface and controlled c-BN bonding, so orientation and surface preparation require confirmation, while c-BN growth is outside the stated catalog. 6CCVD supplies the diamond material through its diamond materials catalog, and offers custom fabrication and engineering support.

View Original Abstract

ABSTRACT Diamond and cubic boron nitride (c‐BN) are promising ultrawide‐bandgap semiconductors for next‐generation high‐power and high‐frequency devices. Enabling these devices needs better understanding of interface electronic structure. Here, we investigate the effects of interfacial bonding and antisite point defects on the formation of two‐dimensional (2D) carrier gases at (111) diamond/c‐BN interfaces using hybrid density functional theory. c‐BN can be B‐ or N‐terminated. For each termination type, we identify two distinct interfacial bonding configurations, where a B (or N) atom is bonded to either one or three C atoms, resulting in four bonding types. Our results show that electron‐deficient C−B and electron‐rich C−N interfacial bonds result in 2D hole gas and 2D electron gas, respectively. Configurations with B (or N) bonded to three C atoms exhibit three times higher carrier density (1.3310 ) than those bonded to one C atom (4.4410 ). Furthermore, ‐type and ‐type antisite defects can stabilize C−B and C−N bonded interfaces, respectively, while reducing carrier density. Nevertheless, the carrier density remains high () even with these defects. This work may guide the engineering of diamond/c‐BN interfaces with tunable carrier densities for ‐ or ‐type conduction.