Skip to content

Tunable Transport in a Dual-Gate Diamond Valley Field-Effect Transistor

FieldValue
JournalACS Applied Electronic Materials
Published2026-07-27
AuthorsNattakarn Suntornwipat, Jan Isberg, Saman Majdi
DOI10.1021/acsaelm.6c01210

A dual-gate, two-drain diamond field-effect transistor steered valley-selective electron transport between two drains and reported that valley polarization remained largely preserved from 10 to 77 K. In the constant-bias temperature series, the measured peak arrival time increased by about 2.5, from 8.7 ns at 10 K to 21.6 ns at 77 K.

The devices used freestanding single-crystal CVD diamond plates measuring 4.5 × 4.5 mm, with thicknesses of 490 and 510 μm. The nitrogen concentration was below 0.05 ppb to limit ionized impurity scattering and maintain high carrier mobility for valley-resolved transport.

Each plate was cleaned in HNO3:H2SO4:HClO4 for 40 min at 180 °C. Low-power oxygen plasma then provided 60 s of oxygen termination. A 30 nm Al2O3 layer was deposited by ALD at 300 °C using trimethylaluminum and ozone. The layer served as both gate dielectric and surface passivation. Photolithography and hydrofluoric acid etching defined the contacts. Ti/Al layers of 20/300 nm formed the source, gate, and drain electrodes. A 10 nm semitransparent Au layer formed the back contact. The final device had one source, two independently biased gates, and two drains.

Electron-hole pairs were generated near the source with 213 nm laser pulses from a passively Q-switched DPSS laser. The pulse width was 800 ps, the repetition rate was 313 Hz, and the photon energy was 5.82 eV. The authors attenuated each pulse below 1 nJ, keeping carrier density below 10¹⁰ cm⁻³ and thermal load below 0.3 μW. Drain currents were recorded at virtual ground. A 1 GHz amplifier and a 3 GHz oscilloscope sampling at 10 GS/s captured the signals. Temperature was held within ±0.1 K.

Photoexcited electrons initially occupied all six valleys equally. The authors assigned valley populations from drain-resolved transient currents under the Shockley-Ramo theorem. Different anisotropic masses produced different trajectories and transit times, yielding spatially and temporally distinct components. The assignment used methods validated in earlier work. The paper did not report a direct valley-occupation measurement for this device.

At 77 K, the source and backplane were fixed at 7 V and 2 V. The right gate ranged from 2 to 4 V, while the left gate swept from 0 to 14 V. Two transport populations appeared. The (001) valleys formed a surface channel, whereas the (100) and (010) valleys extended farther into the bulk.

The left gate controlled penetration depth at injection. Increasing its bias pushed the deeper populations farther into the diamond and brought the (001) signal into Drain 1, then Drain 2 at higher voltage. The right gate controlled the final distribution between drains. As its bias rose, the (001) current at Drain 1 decreased and shifted toward Drain 2. The first (001) peak remained at a left-gate voltage of about 5 V regardless of right-gate bias. This separated early transport from drain-side steering. At the highest left-gate biases, slower carriers extended beyond the contact area or were collected by the backplane. The authors selected a 7 V source bias for sufficient velocity difference and limited diffusive broadening.

Backplane confinement and interface response

Section titled “Backplane confinement and interface response”

At 77 K, the backplane was swept from 2 to 3 V while the left gate swept from 0 to 14 V. The source remained at 7 V and the right gate at 3 V. Raising the backplane bias moved carriers toward the top surface and reduced their penetration depth. The Drain 1 signal intensity fell, which the authors linked tentatively to surface-roughness scattering or carrier loss near the interface. Drain 2 changed in both arrival time and curve shape. The authors described the response to simultaneous backplane and right-gate changes with a Poisson-based electrostatic model.

The authors tested two samples at 77 K with source, backplane, and right-gate voltages of 7 V, 3.4 V, and 3.4 V. The left gate again swept from 0 to 14 V. Both samples showed the same valley-steering trends, but peak intensity and voltage position differed.

Earlier vertical measurements gave room-temperature electron mobilities of 2190 cm²/V·s for Sample #1 and 2140 cm²/V·s for Sample #2. The gap was about 2.3%. The authors suggested that the threshold difference arose from interface traps, fixed charges, and surface morphology before oxide deposition. Repeating the measurement on one sample after one week left the signal profiles, shapes, and peak positions largely unchanged.

Temperature response and comparison with prior work

Section titled “Temperature response and comparison with prior work”

Under constant bias, the measured peak arrival time rose from 8.7 ns at 10 K to 21.6 ns at 77 K. The signal was sharper and larger at 10 K. It broadened and decreased slightly in amplitude at 77 K as diffusion increased. Intervalley scattering remained minimal over this range, so the authors reported that valley polarization was largely preserved during transit.

The temperature series used a 5.6 V source bias, a 300 μm transport distance, and an average field of around 187 V/cm. At that lower lateral field, the (001) and (100)/(010) responses merged into one centroid. The 2.5 arrival-time factor therefore describes a mixed valley ensemble, not two separately resolved peaks.

Temperature-series metricLateral temperature seriesPrior vertical time-of-flight study
Travel distance300 μm510 μm
Average electric fieldaround 187 V/cm135 V/cm
Fast-population drift timemerged into one centroid12 to 18 ns, factor 1.5
Slow-population drift timemerged into one centroid28 to 62 ns, factor 2.2
Peak arrival-time change8.7 to 21.6 ns, factor 2.5Fast and slow values reported separately

The prior vertical study used a 510 μm travel distance and 135 V/cm field over a similar temperature range. Its fast and slow drift times changed by factors of 1.5 and 2.2, respectively. The lateral device had a shorter distance and higher field, but its factor of 2.5 was slightly above the prior slow-carrier factor of 2.2. The authors suggested that the difference could reflect additional scattering sensitivity at the diamond/dielectric interface.

Diamond has six equivalent conduction band valleys along the 〈100〉 directions. Their longitudinal effective mass is about 1.56 times the free-electron mass, compared with 0.28 for the transverse mass. The longitudinal value is roughly 5.5 times larger. This anisotropy changes drift velocity, vertical diffusion, and collection under the gate fields.

The intervalley scattering time is about 300 ns at 77 K, while intravalley acoustic scattering takes about 1 ps. Orthogonal-valley scattering requires a 110 meV phonon, and opposite-valley scattering requires 165 meV. Both energies exceed the thermal energy of about 6.6 meV at 77 K. The authors therefore treated phonon occupation and intervalley transitions as largely frozen out. They state that transit times much shorter than 300 ns preserve the valley-dependent transport signatures.

The measurements covered 10 to 77 K, and no room-temperature operation of this lateral transistor was demonstrated. Only two samples were tested, and their peak positions differed despite similar trends. The authors treated the threshold difference as an interface effect rather than a change in valley-steering behavior. Extra current features at higher temperatures suggest increased intervalley coupling, which could affect valley lifetime and device performance.

Device design must account for dielectric interface conditions as well as bulk valley transport. The data support independent injection-depth control, drain-side steering, and a lateral field strong enough to separate arrival times. The paper recommends low-temperature operation for applications requiring long valley coherence times. The authors propose these properties for energy-efficient electronic and hybrid quantum devices, but the paper does not report a logic, memory, or power-efficiency benchmark.

The relevant material is a freestanding, ultralow-nitrogen single-crystal diamond plate with a controlled dielectric interface and metal contacts. The 490 μm plate thickness used in the study fits 6CCVD’s listed 0.1 to 500 μm range for single-crystal MPCVD plates, and its diamond materials catalog lists Ti and Au among metallization options. Process-specific plate and contact-stack questions can be directed through engineering support.

View Original Abstract

Abstract Device stability is essential for quantum information technologies, where reliable control of electronic states is crucial. Diamond valleytronics offers a promising platform by exploiting the valley degree of freedom to store and manipulate information. In this work, we demonstrate a diamond-based valley transistor with a dual-gate, two-drain architecture that enables tunable valley-polarized transport via gate voltage modulation. By leveraging the significant effective-mass anisotropy of diamond’s conduction band valleys, this architecture provides control over spatial distribution and transit times. We further demonstrate that valley-polarized transport in diamond is remarkably robust against thermal fluctuations over millimeter-scale transport distances. These results demonstrate the resilience of valley states and highlight diamond’s potential for energy-efficient valleytronic devices in next-generation quantum and high-power electronics.

  1. 2015 - Ultrafast Multi-Level Logic Gates with Spin-Valley Coupled Polarization Anisotropy in Monolayer MoS2 [Crossref]
  2. 2018 - Valleytronics: Opportunities, Challenges, and Paths Forward [Crossref]
  3. 2025 - Van Der Waals Valleytronics [Crossref]
  4. 2020 - Room-Temperature Valleytronic Transistor [Crossref]
  5. 2019 - Valley Optomechanics in a Monolayer Semiconductor [Crossref]
  6. 2023 - Giant and Controllable Valley Currents in Graphene by Double Pumped THz Light [Crossref]
  7. 2025 - Creation and Control of Valley Currents in Graphene by Few Cycle Light Pulses [Crossref]
  8. 2012 - Valley-Based Field-Effect Transistors in Graphene [Crossref]
  9. 2022 - Room-Temperature Valley Transistors for Low-Power Neuromorphic Computing [Crossref]
  10. 2006 - Valley Polarization in Si(100) at Zero Magnetic Field [Crossref]