Lei Ge

With the increased power density of gallium nitride (GaN) high electron mobility transistors (HEMTs), effective cooling is required to eliminate the self-heating effect. Incorporating diamond into GaN HEMT is an alternative way to dissipate the heat generated from the active region. In this review, the four main approaches for the integration of diamond and GaN are briefly reviewed, including bonding the GaN wafer and diamond wafer together, depositing diamond as a heat-dissipation layer on the GaN epitaxial layer or HEMTs, and the epitaxial growth of GaN on the diamond substrate. Due to the large lattice mismatch and thermal mismatch, as well as the crystal structure differences between diamond and GaN, all above works face some problems and challenges. Moreover, the review is focused on the state-of-art of polycrystalline or nanocrystalline diamond (NCD) passivation layers on the topside of GaN HEMTs, including the nucleation and growth of the diamond on GaN HEMTs, structure and interface analysis, and thermal characterization, as well as electrical performance of GaN HEMTs after diamond film growth. Upon comparing three different nucleation methods of diamond on GaN, electrostatic seeding is the most commonly used pretreatment method to enhance the nucleation density. NCDs are usually grown at lower temperatures (600–800 °C) on GaN HEMTs, and the methods of “gate after growth” and selective area growth are emphasized. The influence of interface quality on the heat dissipation of capped diamond on GaN is analyzed. We consider that effectively reducing the thermal boundary resistance, improving the regional quality at the interface, and optimizing the stress–strain state are needed to improve the heat-spreading performance and stability of GaN HEMTs. NCD-capped GaN HEMTs exhibit more than a 20% lower operating temperature, and the current density is also improved, which shows good application potential. Furthermore, the existing problems and challenges have also been discussed. The nucleation and growth characteristics of diamond itself and the integration of diamond and GaN HEMT are discussed together, which can more completely explain the thermal diffusion effect of diamond for GaN HEMT and the corresponding technical problems.

This paper presents a fabricated solar-blind phototransistor based on hydrogen-terminated diamond. The phototransistor shows a large photocurrent and enhancement of responsivity over conventional two-terminal diamond-based photodetector. These enhancement effects are owing to the internal gain of the phototransistor. The fabricated phototransistor exhibits a high photoresponsivity (R) of 2.16 × 10⁴ A/W and a detectivity (D*) of 9.63 × 10¹¹ jones, with gate voltage (V_G) and drain voltage of approximately -1.5 V and -5 V, respectively, under 213 nm light illumination. Even at ultralow operating voltage of -0.01 V, the device records satisfactory performance with R and D* of 146.7 A/W and 6.19 × 10¹⁰ jones, respectively. By adjusting the V_G, photocurrent generation in the device can be continuously tuned from the fast photoconductive effect to the optical gating effect with high optical gain. When V_G increases from 1.4 to 2.4 V, the decay time decreases from 1512.0 to 25.5 ms. Therefore, responsivity, dark current, I_photo/I_dark, and decay time of the device can be well tuned by V_G.

β−Ga₂O₃ field−effect transistors (FETs) were fabricated with and without local thinning to change the threshold voltage. A 220 nm Ga₂O₃ layer was mechanically exfoliated from a Cr−doped gallium oxide single crystal. Approximately 45 nm Ga₂O₃ was etched by inductively coupled plasma to form the local thinning. The threshold voltage of the device with etched local thinning increased from −3 V to +7 V compared to the unetched device. The effect of the local thinning was analyzed by device simulation, confirming that the local thinning structure is an effective method to enable enhancement−mode Ga₂O₃ FETs.

A high-performance solar-blind phototransistor, which is based on hydrogen-terminated diamond was fabricated and reported. The fabricated phototransistor was based on metal–semiconductor field effect transistor architecture with a high photoresponsivity ( $2.48\times10$ ⁴ $\text{A}/\text{W}$ ), high external quantum efficiency ( $1.44\times10$ ⁵ ), and high detectivity ( $5.08\times10$ ⁹ Jones) under 213-nm light illumination (437 $\text{W}/\text{m}$ ² ). At 5240 $\text{W}/\text{m}$ ² light illumination, the change in drain current exceeds six orders of magnitude. Through transient response measurement, the rise/decay time of the phototransistor is about 88/36 ms and there is no significant persistent photoconductive effect.