Impact of Temperature and Interface Traps on Threshold Voltage and Carrier Mobility of Hydrogen-Terminated Diamond MOSFETs
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Materials |
| Published | 2026-07-01 |
| Authors | Nuwayyir Alshammari, Mulpuri V. Rao, Qiliang Li |
| DOI | 10.3390/ma19132791 |
Calibrated TCAD simulations show that low-trap effective mobility in a hydrogen-terminated diamond p-channel MOSFET falls from 34.23 cm²/V·s at 300 K to 7.78 cm²/V·s at 600 K. Interface traps produce a sharp mobility knee between 10¹³ and 10¹⁴ cm⁻²·eV⁻¹. Rising temperature and increasing trap density also increase the magnitude of the fitted threshold-voltage shift.
Modeled device and calibration
Section titled “Modeled device and calibration”The study used two-dimensional Synopsys Sentaurus TCAD simulations of the hydrogen-terminated diamond p-channel MOSFET reported by Hirama et al. The modeled gate stack had a 17 nm Al₂O₃ dielectric and a 0.2 nm interfacial layer. Gate length was 0.4 µm, with source-gate and gate-drain spacings near 1 µm.
The source and drain used ohmic Au contacts with a terminal resistance of 2.2 Ω·mm per contact. The diamond buffer included unintentional boron doping with a 0.36 eV activation energy. The bottom surface stayed at 300 K as a thermal reference, without restricting the lattice temperature during each sweep.
Transport used a temperature-dependent mobility model combining phonon scattering, Coulomb scattering, and normal-field degradation. Shockley-Read-Hall statistics represented recombination and trapping. The threshold voltage was extracted mainly with the Y-function, smoothed near the transconductance peak. A constant-current criterion at |ID| = 1 mA/mm provided a cross-check. Gate sweeps covered VD = −12, −10, −6, −3, −2, and −1 V.
At VD = −10 V, the simulation reproduced published transfer characteristics on linear and semilogarithmic scales. The reported constant-current thresholds were 8.306 V experimentally and 8.912 V in simulation, a 0.606 V difference. Log-scale errors were 0.014 decades RMSE and 0.010 decades MAE, corresponding to about 3% and 2.3%. Simulated minimum leakage was 0.0534 mA/mm, versus 0.290 mA/mm experimentally at |VGS| ≈ 10.2 V.
Threshold voltage under drain bias
Section titled “Threshold voltage under drain bias”Across the drain-bias analysis, extracted Vth remained negative and became less negative as |VD| decreased. The drain-dependent shift grew with temperature, consistent with stronger drain-channel electrostatic coupling near turn-on. The authors caution against a simple drain-induced barrier lowering interpretation because the two-dimensional hole gas forms by surface transfer doping rather than bulk inversion.
At the representative condition Dit = 1 × 10¹³ cm⁻²·eV⁻¹ and VD = −10 V, acceptor-like traps at the Al₂O₃/diamond interface shifted the transfer curve toward higher VGS. The added states reduced drain current and degraded subthreshold slope. Occupied traps were negatively charged and partly screened the gate field. Their Gaussian energy distribution began at the valence-band edge and extended into the gap.
Temperature and interface-trap coupling
Section titled “Temperature and interface-trap coupling”The threshold study used Dit values of 1 × 10¹², 2 × 10¹², 5 × 10¹², and 1 × 10¹³ cm⁻²·eV⁻¹. The mobility study extended the range to 1 × 10¹⁶ cm⁻²·eV⁻¹. This range starts above the 3.19 × 10¹¹ cm⁻²·eV⁻¹ state density reported for an optimized H-diamond gate stack in prior work.
The fitted ΔVth became more negative as temperature rose and Dit increased. The additional shift became noticeable above about 10¹³ cm⁻²·eV⁻¹. A constrained semi-empirical fit used α₁ ≈ −4.4 × 10⁻² V and α₂ ≈ −2.3 × 10⁻¹⁵.
The reduced fit had an RMSE of 0.04661 V and an MAE of 0.03953 V. The full fit gave 0.04648 V and 0.03969 V. The third-order term improved RMSE by only 1.3 × 10⁻⁴ V while slightly worsening MAE, so the authors removed it.
The authors interpret the coupled response through temperature-dependent trap occupancy, interfacial charge redistribution, and changes in two-dimensional hole concentration. They use the fit for interpolation and design comparison, not as a microscopic trap model. The extracted dVth/dT rose monotonically with Dit.
Mobility response to traps and temperature
Section titled “Mobility response to traps and temperature”Field-effect mobility was extracted from low-drain-bias transfer curves and evaluated within VG = −4 to −2 V. The effective-mobility comparison used VD = −0.1 V. A median value reduced sensitivity to numerical fluctuations. The gm/ID method creates a narrow peak just below Vth. The authors attribute this peak to rapid channel formation, not a physical mobility increase.
The fitted response separates the low-trap plateau μ₀, the transition density D₀, and the roll-off exponent β. D₀ was assigned where the smoothed effective mobility reached 0.5μ₀.
| Temperature | Low-trap mobility μ₀ (cm²/V·s) | D₀ (cm⁻²·eV⁻¹) | β |
|---|---|---|---|
| 300 K | 34.23 | 4.17 × 10¹³ | 1.31 |
| 400 K | 20.98 | 3.76 × 10¹³ | 1.31 |
| 500 K | 12.02 | 4.23 × 10¹³ | 1.30 |
| 600 K | 7.78 | 4.23 × 10¹³ | 1.30 |
At each temperature, μeff stayed nearly constant at low Dit before a clear knee appeared between 10¹³ and 10¹⁴ cm⁻²·eV⁻¹. The D₀ values stayed close to 4 × 10¹³ cm⁻²·eV⁻¹, while β stayed near 1.3. Thus, temperature mainly lowered the mobility level instead of changing the fitted curve shape.
At 300 K and VD = −6 V, on-state current remained close across Dit values from 10¹² to 10¹⁵ cm⁻²·eV⁻¹. The main change was a turn-off shift. At fixed Dit = 10¹² cm⁻²·eV⁻¹, increasing temperature reduced on-state current and degraded subthreshold slope.
Gate length and earlier temperature studies
Section titled “Gate length and earlier temperature studies”Under the selected bias and temperature conditions, the trap density associated with maximum |ΔVth| changed with gate length. It rose from about 10¹³ cm⁻²·eV⁻¹ at LG = 0.05 µm to 10¹⁴ at 0.10 µm. It reached about 10¹⁵ at 0.20 µm and remained near 10¹⁵ at 0.40 µm.
The authors relate the short-channel response to stronger electrostatic coupling and localized interface charge near gate edges. Longer channels require more traps to produce a similar maximum shift. The apparent saturation over the longer gate lengths indicates that additional gate-length scaling has a weaker influence in that range.
Earlier temperature studies demonstrated H-diamond operation over broader or comparable ranges. Kawarada et al. reported C-H MOSFETs from room temperature to 400 °C and from 10 K to 673 K. Peterson et al. examined two-dimensional hole gas transport from 25 K to 700 K, while Oi et al. studied a vertical two-dimensional hole gas MOSFET up to 300 °C. This work instead varies temperature and Dit independently in a calibrated model, then quantifies Vth, dVth/dT, and effective mobility together.
Limits and design implications
Section titled “Limits and design implications”All temperature and trap results are TCAD predictions, not new measurements. The model was calibrated against one published device and its transfer characteristics. Its traps use an idealized Gaussian energy distribution and a 0.2 nm interfacial layer. Practical devices may have more complex energy and spatial trap distributions that depend on gate material and stress history. Self-heating during high-power operation was not included explicitly.
At 300 K, the authors place a practical reference below about 5 × 10¹³ cm⁻²·eV⁻¹ for Dit. This limit helps keep effective mobility near its phonon-limited plateau and avoids stronger nonlinear Vth shifts. When Dit approaches D₀, effective mobility falls to roughly half its low-trap value.
Between 300 K and 600 K, the acceptable Dit range decreases by about one order of magnitude. The falling low-trap baseline leaves less mobility margin for Coulomb scattering. Reduced |VD|, suitable gate length, and suitable gate-drain spacing can lessen drain-channel coupling. The paper gives no bulk diamond acceptance criterion. Its actionable material variable is the interface-trap density at the hydrogen-terminated diamond and Al₂O₃ boundary.
Materials and support
Section titled “Materials and support”For this MOSFET application, the modeled structure requires a hydrogen-terminated diamond channel, a 17 nm Al₂O₃ gate dielectric, and Au source and drain contacts. 6CCVD’s diamond materials catalog lists MPCVD-grown single-crystal and polycrystalline diamond and metallization with Au, although the paper does not specify the channel’s crystal form or a hydrogen-termination process. The interface and metallization requirements should be reviewed through engineering support.
View Original Abstract
We present a Sentaurus TCAD model calibrated for the hydrogen-terminated diamond p-channel MOSFET reported by Hirama et al. and validated using the published transfer characteristics. We used the calibrated model to analyze threshold voltage (Vth) and effective mobility (μeff) over a temperature range from 300 to 600 K and an interface-trap density range from 1012 to 1016 cm−2·eV−1. The analysis shows three main trends: (i) a drain-bias-dependent shift in Vth that becomes stronger at higher temperatures; (ii) a threshold shift toward more negative values with increasing temperature and interface-trap density, showing a coupled thermal trap effect; and (iii) a mobility response that stays near the phonon-limited regime at low trap densities but degrades rapidly as trap density increases. We extracted compact empirical expressions for ΔVth (T, Dit) and μeff (T, Dit) that reproduce the simulated trends with good accuracy for device-level comparison. The results show that interface-trap control is important to achieve thermally stable, high-performance diamond MOSFETs.
Boron Doped Single Crystal Diamond Plate
View Specs →
Electronic Grade Intrinsic Diamond Epitaxial Wafer H Terminated Superior 100 Mosfet Ready Substrate
View Specs →
High Concentration Quantum Grade Single Crystal Diamond
View Specs →References we cover
Section titled “References we cover”- Enhancement of drain soft-breakdown strength to 1.1 MV/cm for hydrogen-terminated diamond MOSFETs by mitigating hydrogen-induced defects
- Hydrogen-terminated single crystal diamond MOSFET with the dielectric of Ga2O3
- 2.1 W/mm Output Power Density at 10 GHz for H-Terminated Diamond MOSFETs With (111)-Oriented Surface
- Hydrogen-Terminated Single Crystal Diamond MOSFET with a Bilayer Dielectric of Gd2O3/Al2O3
- An enhanced two-dimensional hole gas (2DHG) C-H diamond with positive surface charge model for advanced normally-off MOSFET devices
Original source
Section titled “Original source”References
Section titled “References”- 2002 - AlGaN/GaN HEMTs-An overview of device operation and applications [Crossref]
- 2018 - Guest Editorial: The dawn of gallium oxide microelectronics [Crossref]
- 2018 - A review of Ga2O3 materials, processing, and devices [Crossref]
- 2002 - High Carrier Mobility in Single-Crystal Plasma-Deposited Diamond [Crossref]
- 1996 - Hydrogen-terminated diamond surfaces and interfaces [Crossref]
- 2001 - High-frequency performance of diamond field-effect transistor [Crossref]
- 2017 - Logic Circuits With Hydrogenated Diamond Field-Effect Transistors [Crossref]
- 2020 - Analysis of mobility-limiting mechanisms of the two-dimensional hole gas on hydrogen-terminated diamond [Crossref]
- 2016 - Inversion channel diamond metal-oxide-semiconductor field-effect transistor with normally off characteristics [Crossref]