Skip to content

Efficient and Robust p-type Transistor based on Ultra-wide-bandgap Semiconductor

FieldValue
JournalOpen MIND
Published2026-02-07
AuthorsKaijian Xing, Zherui Yang, Weiyao Zhao, Yuefeng Yin, Huiping Han
DOI10.48550/arxiv.2602.07424

At room temperature, the STO/H-diamond p-type MOSFET reached about 200 mA/mm at V_g = -16 V and V_ds = -42 V. Its subthreshold swing was about 70 mV/dec, and its on-off ratio was about 10⁹ at V_g = -10 V and V_ds = -6 V. The two-terminal and four-terminal field-effect mobilities were 566 cm²/Vs and 552 cm²/Vs, respectively.

The semiconductor channel was a two-dimensional hole layer at the surface of hydrogen-terminated (100) diamond. The reported bandgap of the hydrogenated diamond channel was 5.6 eV. Purchased (100) diamond substrates were polished before hydrogen termination. Hydrogen termination was then applied in a Seki 6500 microwave plasma-assisted CVD reactor with a 2.4 GHz chamber. Photolithography and oxygen plasma defined the active channel. Pd source and drain contacts were formed by electron beam evaporation and lift-off. The gate dielectric was a 60 nm free-standing SrTiO3 membrane.

STO films were deposited by pulsed laser deposition on TiO₂-terminated SrTiO₃ (001) substrates at 760 °C. The 248 nm excimer laser operated at 2 Hz with a fluence of 1.8 J/cm². Both Sr₃Al₂O₆ and STO growth used an oxygen partial pressure of 1 × 10⁻² torr. PDMS-assisted release in deionized water separated STO from the sacrificial Sr₃Al₂O₆ layer. Micrometer-size STO flakes were dry-transferred onto hydrogenated channels using a propylene carbonate method. The transfer followed pre-annealing in nitrogen with O₂ and H₂O each below 0.1 ppm.

Cross-sectional TEM and EDS found a sharp STO/diamond boundary without a reaction layer. TEM used 200 keV, and final FIB milling used 5 keV to reduce ion damage. No lattice intermixing, dislocations, bonded phases, interdiffusion, or contamination was detected within measurement resolution. The membrane could be peeled from diamond, supporting a weakly interacting van der Waals interface. DFT calculations found negligible interfacial charge transfer. Low-energy transfer avoids deposition processes that can strip hydrogen from the diamond surface.

Pre-annealing controlled surface acceptor desorption and therefore the transistor mode. Threshold voltage shifted from 3.6 V at 60 °C to 2.5 V at 80 °C and -2.6 V at 110 °C. The 60 °C device operated in depletion mode, while the 110 °C device was normally off. The authors propose that acceptor desorption shifts the surface Fermi level and weakens upward valence-band bending. A four-point probe measured an initial surface sheet resistance of 11 kΩ/□. The diamond surface became unmeasurable above 110 °C. Air exposure restored resistance to 10 kΩ/□ after a few minutes, indicating that hydrogen termination remained intact. The authors attribute the mobility gain from about 230 cm²/Vs at 60 °C to less Coulomb scattering by charged acceptors.

Characterization used a four-terminal device pre-annealed at 110 °C. At V_ds = -1 V, the transfer curve confirmed normally off operation. The channel was about 30 µm long and 2 µm wide, with about 20 µm between inner electrodes. The two-terminal and four-terminal mobility results agreed, which the authors attribute to excellent Pd Ohmic contacts at room temperature. The measured subthreshold swing was close to the 60 mV/dec thermal dynamic limit.

Interface trap density was 6.5 × 10¹¹ cm⁻² eV⁻¹. The control Al₂O₃/H-diamond MOSFET had 4.2 × 10¹² cm⁻² eV⁻¹. The gate leakage density was a few orders of magnitude below the 10⁻² A/cm² CMOS low-power limit. Capacitance-voltage measurements at 10 kHz showed accumulation onset near V_g = -2 V. The extracted dielectric constant was 30.7, consistent with 30.5 from a metal-insulator-metal capacitor. The 60 nm membrane produced 7.6 nm effective oxide thickness, compared with about 10 nm for the hBN reference.

The paper separately reports gate-length-normalized current density of about 4000 µA/mm. At high drain bias, output saturation fell by less than 2%. The authors hypothesize that the reduction is caused by a lateral gate effect. The source meter limited applied V_ds, so the paper does not establish a maximum rated current.

Comparison with dielectric and device benchmarks

Section titled “Comparison with dielectric and device benchmarks”

The authors identify limited p-type dopants and activation energies above 1 eV as barriers to p-type wide-bandgap transistors. They also cite typical hole mobilities of 1 to 30 cm²/Vs as a constraint on high-power and high-speed operation. The authors place the device among the highest field-effect mobilities reported for hydrogen-terminated diamond FETs. Their comparison assigns it the highest gate-length-normalized current density among the plotted p-type devices.

The Al₂O₃/H-diamond control provides a matched comparison within the study. STO had lower interface trap density and retained measurable conductance as the control approached its off state. Physical-transfer Ga₂O₃/H-diamond and hBN/H-diamond devices provide closer dielectric comparisons. STO had a higher dielectric constant and lower subthreshold swing than the hBN/H-diamond device. The authors report a substantial overall improvement over physical-transfer Ga₂O₃/H-diamond. They attribute the Ga₂O₃ result to defect traps in the amorphous dielectric and at its interface. The authors describe the device as superior to the plotted p-type SiC, GaN, and two-dimensional transistor results.

STO/H-diamond transfer curves decreased moderately with cooling, while control Al₂O₃/H-diamond curves became unmeasurable at low temperature. The control reached its off state near 40 K at V_g = -5 V and V_ds = -2 V. The STO device current fell by less than one order of magnitude and became nearly temperature independent below 20 K. At V_g = -5 V, sheet resistance rose from about 8 kΩ/□ at 300 K to 56 kΩ/□ at 10 K. Correcting the gate bias relative to V_th gave an increase from about 8 kΩ/□ to about 40 kΩ/□. The authors attribute this change to hole-hole interactions, with logarithmic temperature dependence consistent with a two-dimensional Fermi liquid. Pd contacts remained Schottky at 1.8 K, and the authors suggest charge-transfer electrodes could improve injection.

The device remained functional to 400 K, with this upper limit set by measurement equipment. Between 300 K and 400 K, threshold voltage shifted from -2.4 V to -2.8 V. At V_g = -4 V, on-current fell from 3.5 × 10⁻⁵ A to 2 × 10⁻⁵ A. Two-terminal mobility decreased by less than 10%. After two months in air without passivation, mobility fell by 9% to 510 cm²/Vs. Subthreshold swing rose to 88 mV/dec, while threshold voltage remained nearly unchanged. The authors attribute both changes to minor changes at the STO/H-diamond interface.

Four additional devices made with the same process had comparable transfer and output characteristics. Mean two-terminal mobility was about 530 ± 30 cm²/Vs, with a relative standard deviation near 6%. Subthreshold swing averaged 72 ± 4 mV/dec, and threshold voltage averaged -2.2 ± 0.4 V. These results indicate repeatability across the primary device and four additional devices, although the measured set was small.

Further work may improve STO surface quality, diamond roughness, and fabrication control to approach the theoretical mobility. The membrane preparation produced micrometer-size flakes, so wafer-scale integration was not demonstrated. The authors cite recent wafer-scale dielectric transfer as a possible route to larger devices. Low-temperature contact injection, hole-hole interactions, and lateral gate effects remain issues in the reported data.

Surface termination and acceptor control therefore govern channel mode and mobility in these measurements. STO/H-diamond is a candidate when low-temperature conductance and high current density are required. The reported voltage and temperature limits are test limits, not established breakdown or reliability ratings.

The device calls for a polished, hydrogen-terminated (100) diamond substrate with Pd source, drain, and gate contacts. 6CCVD’s diamond materials catalog lists MPCVD-grown single-crystal diamond plates and wafers up to 125 mm, Pd metallization, and polishing below Ra 1 nm. The engineering support page can define substrate thickness, hydrogen termination, surface finish, and contact geometry for a scaled prototype.

View Original Abstract

The p-type transistor is an indispensable component of semiconductor technology, enabling complementary operation with n-channel transistors for computation, storage, and communication. Achieving both high robustness and high efficiency is highly desirable but challenging for p-type transistors due to limited semiconductors with reliable hole transport and their high activation energies. Here, we achieved a robust yet efficient p-type transistor by heterogeneously integrating an ultra-wide-bandgap semiconductor and a high-k dielectric layer through van der Waals integration. The p-type transistor employs a two-dimensional hole channel on hydrogenated diamond (bandgap 5.6 eV) combined with a high-k (30.5) SrTiO3 perovskite membrane. At room temperature, the transistor exhibits stable operation with a high on-current (~200 mA/mm), low subthreshold swing (70 mV/dec), high hole mobility (566 cm^2/Vs to 572 cm^2/Vs) and high on-off ratio (~10^9). Furthermore, tuning annealing temperature allows operation in either enhancement or depletion mode. The robust p-type transistor with high efficiency holds great potential for future power electronics, UV optoelectronics, and harsh-environment electronic applications.