Skip to content

Efficient and Robust p-Type Transistor Based on Ultrawide-Bandgap Semiconductor

FieldValue
JournalACS Nano
Published2026-02-17
AuthorsXing Kx, Zherui Yang, Weiyao Zhao, Yuefeng Yin, Huiping Han
DOI10.1021/acsnano.5c22605
Citations1

At room temperature, a hydrogen-terminated diamond p-channel MOSFET delivered a maximum normalized on-current of about 200 mA/mm. This occurred at Vg of −16 V and Vds of −42 V.

The transistor used a two-dimensional hole channel on a hydrogen-terminated (100) diamond surface and a 60 nm SrTiO3 membrane as its gate dielectric. SrTiO3 had a measured dielectric constant of 30.5 in a separate metal-insulator-metal device. The membrane was physically transferred by dry pickup to form a van der Waals interface. Cross-sectional TEM and EDS mapped the interface. The interface showed no reaction layer, lattice intermixing, dislocations, or chemically bonded phase within measurement resolution. The membrane could be mechanically peeled from the hydrogen-terminated surface. DFT calculated negligible interfacial charge transfer.

The SrTiO3 membrane came from PLD-grown SrTiO3/Sr3Al2O6 films on TiO2-terminated (001) SrTiO3. PLD used a 248 nm excimer laser at 2 Hz and a fluence of 1.8 J/cm². The substrate stayed at 760 °C, with oxygen pressure at 1 × 10⁻² torr. The SrTiO3 film was separated from its Sr3Al2O6 support by floating the PDMS-covered stack in deionized water overnight. Repeated lamination and peeling produced micrometer-size SrTiO3 flakes for dry pickup.

The diamond was polished, then hydrogen terminated in a 2.4 GHz microwave plasma-assisted chemical vapor deposition reactor. Photolithography and oxygen plasma isolated the active channels. Pd source and drain contacts were formed by electron-beam evaporation and lift-off. The control Al₂O₃ dielectric was formed by ALD at 150 °C. The STO membrane was transferred by the propylene carbonate dry-pickup method in a nitrogen glovebox before Pd top-gate fabrication. DC transport used two Keithley 2400 source meters and a multimeter. A Keysight B1500 and probe station measured DC and capacitance-voltage performance in atmosphere.

The initial sheet resistance of the hydrogen-terminated surface was 11 kΩ/□, measured with a four-point probe. Pre-annealing occurred in nitrogen with O₂ and H₂O each below 0.1 ppm. At 110 °C, surface conductivity became unmeasurable, then recovered to 10 kΩ/□ after several minutes in air. This recovery indicated that hydrogen termination remained intact during glovebox annealing. STO capping suppressed the air-driven recovery, while peeling the membrane left conductance unchanged.

Pre-annealing at 60 °C produced a threshold voltage of 3.6 V, and 80 °C produced 2.5 V. At 110 °C, the threshold shifted to −2.6 V, producing a normally off device. The authors attribute the change to surface acceptor desorption, which weakens valence-band bending and shifts the surface Fermi level into the gap. The normally off state was not attributed to gate-metal depletion, because Pd has a work function of about 5.6 eV. This value is similar to the H-diamond surface ionization potential.

The two-terminal mobility rose from about 230 cm²/Vs after 60 °C annealing to 565.5 cm²/Vs after 110 °C annealing. The authors propose that fewer charged acceptors reduce Coulomb scattering. DFT separately found negligible interfacial charge transfer.

The reported four-terminal device used 110 °C pre-annealing and had an approximately 30 µm channel. Width was about 2 µm, with 20 µm between inner voltage electrodes. Transfer curves were measured at Vds of −1 V, while output measurements reached Vds of −42 V. At Vg of −10 V and Vds of −6 V, the on-off ratio was about 10⁹. The extracted subthreshold swing was 70 mV/dec, close to the stated thermal limit of 60 mV/dec. Two-terminal and four-terminal field-effect mobilities were 566 and 552 cm²/Vs, respectively. The similar values were attributed to the reported room-temperature Ohmic behavior of the Pd contacts.

Interface trap density was 6.5 × 10¹¹ cm⁻² eV⁻¹, compared with 4.2 × 10¹² cm⁻² eV⁻¹ for the ALD Al₂O₃ control. Output curves showed Ohmic and saturation regions, with less than 2% saturation-current reduction at high Vds. The authors suggest a lateral gate effect as the cause. The source meter limited the measured Vds, so the paper does not report a breakdown voltage. Gate-length-normalized on-current density reached about 4000 µA/mm.

Four other devices showed a mean two-terminal mobility of 530 ± 30 cm²/Vs. The mobility relative standard deviation was about 6%. Mean subthreshold swing was 72 ± 4 mV/dec, and mean threshold voltage was −2.2 ± 0.4 V. These results show reproducible transfer characteristics and dielectric behavior across the small device set.

Capacitance-voltage measurements at 10 kHz showed the onset of hole accumulation at Vg of −2 V. The dielectric constant was 30.7 from the FET capacitor and 30.5 from the metal-insulator-metal reference. The 60 nm SrTiO3 membrane had an effective oxide thickness of 7.6 nm, compared with about 10 nm for hBN. Gate leakage density was a few orders of magnitude below the cited CMOS low-power limit of 10⁻² A/cm².

On cooling, the STO device current fell by less than one order of magnitude and became nearly temperature independent below 20 K. An Al₂O₃ control reached its off state at about 40 K under Vg of −5 V and Vds of −2 V. At Vg of −5 V, channel sheet resistance increased from about 8 kΩ/□ at 300 K to 56 kΩ/□ at 10 K. After threshold-voltage normalization, resistance changed from about 8 kΩ/□ at 300 K to 40 kΩ/□ at 10 K. The logarithmic temperature dependence was attributed to hole-hole interactions within the two-dimensional hole gas. Output curves at 1.8 K indicated Schottky Pd contacts, although carriers still tunneled through the barrier.

High-temperature measurements reached 400 K, which was the equipment limit. Threshold voltage shifted from −2.4 V to −2.8 V over that range. At Vg of −4 V, on-current fell from 3.5 × 10⁻⁵ A to 2 × 10⁻⁵ A, while mobility decreased by less than 10%. The device retained p-type behavior after two months in air without passivation. Two-terminal mobility fell about 9% from 565 to 510 cm²/Vs. Subthreshold swing increased from 70 to 88 mV/dec, while threshold voltage remained nearly unchanged.

The reported 566 cm²/Vs mobility exceeds the paper’s stated 1 to 30 cm²/Vs range for p-type materials. The authors place the device among the highest field-effect mobilities reported for hydrogen-terminated diamond FETs. They describe the 70 mV/dec subthreshold swing as the lowest reported value in their literature comparison. The gate-length-normalized current was also the highest in their comparison with reported p-type devices.

The authors report better device performance with STO than with amorphous Ga₂O₃. They also report a higher dielectric constant and lower subthreshold swing than with hBN. The Ga₂O₃ dielectric was formed by a liquid metal method. The authors attribute the lower Ga₂O₃ performance to numerous defect traps in the dielectric and at the interface.

The selected SrTiO3 came from micrometer-size flakes, so the demonstrated process was not wafer-scale. Wafer-scale dielectric transfer was cited as a possible route to larger devices, not demonstrated here. At 1.8 K, the contacts were Schottky, and the authors suggest a charge-transfer electrode could improve them. The authors identify surface roughness, membrane quality, and fabrication optimization as possible routes toward higher mobility.

The results make surface acceptor control and dielectric contact quality central variables for p-channel design. Hydrogen termination, pre-anneal temperature, and dielectric transfer must be considered together. The authors propose power electronics, UV optoelectronics, and harsh-environment electronics as target applications, rather than demonstrated systems. Any scale-up must preserve the hydrogen-terminated interface and control surface acceptors during annealing.

The p-channel MOSFET uses a hydrogen-terminated (100) diamond channel with Pd contacts. 6CCVD’s diamond materials catalog lists single-crystal diamond grown by MPCVD, plates and wafers up to 125 mm, and metallization with Pd. These entries match the single-crystal base material and Pd contact choices, while the engineering support link provides custom fabrication support for device-specific preparation steps.

View Original Abstract

The p -type transistor is an indispensable component of semiconductor technology, enabling a 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 the 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 ultrawide-bandgap semiconductor and a high- κ 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- κ (30.5) SrTiO 3 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 /(V·s) to 572 cm 2 /(V·s)), and high on-off ratio (∼10 9 ). Furthermore, tuning the 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, ultraviolet (UV) optoelectronics, and harsh-environment electronic applications.