Wafer-Scale Boron Carbon Nitride Delivers the Missing p-Type Semiconductor for 2D Chips

For more than two decades, engineers have dreamed of shrinking transistors into the ultimate limit of thinness: a single atomic layer. Two-dimensional materials such as molybdenum disulfide and tungsten diselenide have shown that electronics can work with channels just three atoms thick, promising denser, more energy-efficient chips than silicon can deliver. Yet one stubborn asymmetry has held the field back. While n-type 2D semiconductors, which carry electrons, have steadily improved, their p-type counterparts, which carry holes, have lagged far behind. Because complementary metal-oxide-semiconductor (CMOS) logic requires both carrier types working in tandem, the absence of a high-performance p-type 2D material has been the single most conspicuous gap in the periodic table of next-generation electronics. A new study published in Nature now claims to close that gap with an unexpected candidate: boron carbon nitride, a tunable alloy of three abundant elements grown as a pristine monolayer across an entire wafer.

The research, led by Chien-Chih Tseng, Chang-Hsun Huang and Jui-Cheng Kao under the direction of Vincent Tung at the University of Tokyo, together with collaborators across Taiwan and Japan, reports the epitaxial growth of monolayer boron carbon nitride (BCN) with a field-effect hole mobility of 100 square centimeters per volt-second. Transistors built from the material switch with an on-off ratio of 10 to the eighth power, deliver on-currents exceeding 0.9 milliamperes per micrometer, and operate at a threshold voltage of minus 0.45 volts. Those numbers, achieved across wafer-scale arrays rather than on isolated champion devices, surpass the performance of every p-type 2D semiconductor reported to date, according to the authors. The work was published on 30 September 2026 and represents a collaboration spanning the University of Tokyo, National Taiwan University, National Yang-Ming Chiao Tung University, Academia Sinica, the National Institute for Materials Science, and Chang Gung University.

The difficulty of making a good p-type 2D semiconductor is rooted in fundamental physics. In many atomically thin materials, the valence band states that transport holes are strongly localized around particular atomic orbitals, which makes holes scatter easily off defects and phonons. Many candidate materials also exhibit a natural tendency toward electron doping, so their surfaces accumulate unwanted negative charge that degrades hole transport. Compounding these intrinsic problems is a synthetic one: to be useful in a fab, a 2D semiconductor must be grown as a continuous, single-crystal film over an entire wafer, with uniform composition and no grain boundaries that would scatter carriers or cause device-to-device variability. Achieving that uniformity for a ternary compound, in which three different elements must arrive at the growing surface in exactly the right proportions, is far harder than growing a binary crystal such as hexagonal boron nitride.

The team’s central insight concerns a problem they describe as a spatial and temporal mismatch in the delivery of boron, carbon and nitrogen atoms. The standard precursor chemistry for boron nitride growth relies on ammonia borane, a molecule that decomposes when heated. When carbon-bearing precursors are added to the mix to make BCN, the different decomposition pathways of the constituent molecules release boron, carbon and nitrogen species at different times and in different ratios. The growing lattice therefore receives an uneven supply of atoms, producing composition fluctuations, defects and discontinuous crystallization that wreck the electronic quality of the film. Rather than accepting this mismatch, the researchers engineered it away by carefully controlling the dehydrogenation and surface reaction pathways of two related precursors: monomethyl ammonia borane and ammonia borane itself.

Monomethyl ammonia borane carries a methyl group that serves as a built-in carbon source, and by tuning the balance between the two precursors the team synchronized the release of all three elements so that they arrived at the substrate together. Density functional theory calculations, performed with the VASP code, mapped the reaction landscape and identified the decomposition intermediates that would either help or hinder uniform growth. The calculations guided the choice of growth conditions, ensuring that carbon atoms and carbon dimers incorporated into the lattice at the right moment rather than segregating into graphene-like domains. The result is a film in which carbon substitutes primarily for nitrogen sites within a continuously crystallized boron nitride lattice, with local distortions around the carbon atoms but no breakdown of long-range order.

Verifying that atomic structure demanded some of the most advanced microscopy available. Aowen Li, Ryo Ishikawa and Naoya Shibata at the University of Tokyo performed high-resolution scanning transmission electron microscopy on the monolayer film, resolving individual atomic columns and confirming that carbon atoms and carbon dimers occupy nitrogen sites throughout the lattice. Complementary multi-scale characterizations, including Raman spectroscopy, X-ray photoelectron spectroscopy, Kelvin probe force microscopy and scanning tunneling spectroscopy, established the film’s composition, work function and electronic structure across the wafer. The measurements converged on a bandgap of 1.90 electronvolts, a value comfortably above the thermal energy that causes leakage in transistors, and one that places BCN squarely in the regime useful for low-power logic.

The electrical characterization is where the work becomes genuinely striking. The team fabricated arrays of p-channel field-effect transistors across the wafer, using a 20-nanometer hafnium dioxide dielectric as the gate insulator. Because the devices were measured in large numbers rather than as cherry-picked specimens, the reported figures carry statistical weight. A hole mobility of 100 square centimeters per volt-second may sound modest compared with the thousands achieved in the best silicon devices, but for a monolayer p-type 2D semiconductor it is a benchmark-setting result, and the on-current above 0.9 milliamperes per micrometer demonstrates that the material can deliver the drive current that real circuits demand. The 10-to-the-eighth on-off ratio means the transistor switches cleanly between conducting and insulating states, and the near-zero threshold voltage of minus 0.45 volts indicates that gates of practical strength can turn the devices on and off efficiently.

The broader significance lies in what BCN enables for three-dimensional monolithic integration. As silicon scaling approaches its physical limits, chipmakers are exploring stacking active layers vertically, growing one layer of 2D transistors directly on top of another to multiply transistor density without shrinking lateral dimensions. That vision requires 2D semiconductors that can be grown at temperatures compatible with the layers beneath them, with wafer-scale uniformity and both carrier polarities available. Previous work has demonstrated growth-based monolithic integration of single-crystal 2D semiconductors and complementary 2D circuits built from separate materials, but a single, scalable p-type platform grown epitaxially across a wafer has been missing. BCN, grown from simple molecular precursors on a crystalline substrate, fits the requirements in a way that exfoliated flakes or transferred films never could.

There is also a pleasing historical symmetry in the result. Boron carbon nitride was first proposed as a graphite-like material in 1987, and researchers have synthesized B-C-N nanotubes and hybrid graphene-boron nitride domains since the 1990s. Carbon impurities in hexagonal boron nitride have even been identified as the source of single-photon emission, a reminder that the BCN system has long been studied at the boundaries of materials science. What distinguishes the new work is control: by solving the precursor chemistry rather than fighting it, the team converted a notoriously inhomogeneous alloy into a device-grade semiconductor. The DFT calculations, experimental growth, atomic-resolution imaging and device engineering were all executed within a single coordinated effort, a model of how modern materials discovery increasingly proceeds from first-principles prediction straight to wafer-scale demonstration.

Challenges remain before BCN transistors appear in commercial chips. The growth is epitaxial, meaning it relies on a crystalline substrate that templates the film’s orientation, and transferring large-area 2D films without introducing damage remains an unsolved industrial problem. Contact engineering, dielectric integration and long-term stability under bias and heat will all need further refinement. The authors declare no competing interests, and the data and calculation inputs are available from the corresponding authors, which should help other groups reproduce and extend the results. But the conceptual barrier has fallen: a p-type 2D semiconductor with wafer-scale uniformity, a useful bandgap and benchmark hole transport now exists. If the result replicates across laboratories, the missing half of the 2D CMOS palette may finally be in place, and the path toward atomically thin, vertically stacked logic circuits becomes considerably more concrete.

Subject of Research: Wafer-scale epitaxial growth of p-type boron carbon nitride for two-dimensional CMOS electronics

Article Title: Wafer-scale epitaxy growth of high-mobility p-type boron carbon nitride

Article References: Wafer-scale epitaxy growth of high-mobility p-type boron carbon nitride. (n.d.). https://doi.org/10.1038/s41586-026-11047-9

Image Credits: AI Generated

DOI: 10.1038/s41586-026-11047-9

Keywords: boron carbon nitride, two-dimensional materials, p-type semiconductor, field-effect transistors, CMOS technology, epitaxial growth, chemical vapor deposition, hole mobility, bandgap engineering, monolithic 3D integration, scanning transmission electron microscopy, density functional theory