In this work, we propose for the first time a high-gradient (HG) stepped-Carbon (stepped-C) doped buffer design in GaN-on-GaN structures to enhance device RF performance. This design not only avoids the impact of the Fe tailing effect on the 2DEG but also effectively mitigates interface conduction losses caused by Si impurities at the regrown interface. Most importantly, the HG stepped-C buffer design significantly alleviates the trapping effects associated with high concentrations of C. The GaN-on-GaN HEMTs with HG stepped-C buffer achieved a breakdown voltage of 249 V, a peak transconductance ( g m ) of 319 mS/mm with a low current collapse (CC) of 6.1%. Collectively, these factors enabled the GaN-on-GaN HEMTs to achieve f T / f max values of 47.6 GHz/68.1GHz, with a state-of-the-art output power density ( P out ) of 15.1 W/mm and a Power-Added efficiency (PAE) of 57.2% at 2 GHz. These results demonstrate the immense potential of HG stepped-C buffer in advancing the RF performance for GaN-on-GaN HEMTs.
This study investigates the thermal dissipation ability and electrical performance of GaN-on-GaN HEMTs through a stepped-C buffer design. We analyzed the relationship between impurity (C and Fe) concentrations and the thermal conductivity of the GaN material by fitting Debye–Callaway model. A stepped-C buffer design is proposed to avoid the Fe impurity and its tailing effect on thermal conduction in GaN epitaxial layers. In addition, the high concentration of C doping is designed to suppress the epitaxial interface leakage in GaN-on-GaN structures. The transducer-less transient thermoreflectance (TL-TTR) technique revealed that the stepped-C structure significantly improves thermal conductivity of epitaxial layers compared with that of Fe/C co-doped structure. Due to the optimization of heat dissipation ability, the peak temperature of the stepped-C sample decreased by ∼30 °C compared to the Fe/C co-doped sample at PDC = 10.4 W/mm. Consequently, the GaN-on-GaN HEMTs with the stepped-C buffer achieved a record output power density (Pout) of 14.8 W/mm and a power-added efficiency (PAE) of 48.2% at 3.6 GHz, underscoring the critical role of thermal management in advancing GaN-on-GaN HEMT RF performance.
Gallium nitride (GaN)-based high electron mobility transistors (HEMTs) are pivotal for next-generation power-switching applications, but their reliability under high electric fields remains constrained by lattice mismatches and high dislocation densities in heterogeneous substrates. Herein, we systematically investigate the electrical performance and reliability of GaN-on-GaN HEMTs in comparison to conventional GaN-on-SiC HEMTs via DC characterization, reverse gate step stress, off-state drain step stress, and on-state electrical stress tests. Notably, the homogeneous epitaxial structure of GaN-on-GaN devices reduces dislocation density by 83.3% and minimizes initial tensile stress, which is obtained through HRXRD and Raman spectroscopy. The GaN-on-GaN HEMTs exhibit a record BFOM of 950 MW/cm2, enabled by a low specific on-resistance (RON-SP) of 0.6 mΩ·cm2 and a high breakdown voltage (BV) of 755 V. They withstand gate voltages up to −200 V and drain voltages beyond 200 V without significant degradation, whereas GaN-on-SiC HEMTs fail at −95 V (reverse gate stress) and 150 V (off-state drain stress). The reduced dislocation density suppresses leakage channels and defect-induced degradation, as confirmed by post-stress Schottky/transfer characteristics and Frenkel–Poole emission analysis. These findings establish GaN-on-GaN technology as a transformative solution for power electronics, offering a unique combination of high efficiency and long-term stability for demanding high-voltage applications.
15 August 2025 in IEEE Microwave and Wireless Technology Letters
This article first proposes a series-shunt compact (SSC) radio frequency (RF) switch design based on gallium nitride (GaN) high-electron-mobility transistors (HEMTs) which can significantly improve the device’s isolation (ISO) and mitigate the degradation of insertion loss (IL). The design follows a distributed architecture, introducing a shunt structure on conventional HEMTs and tightly coupling the capacitances of the series and shunt parts’ active regions. This leads to an increase in $C_{\mathbf {on}}$ during the on-state, while the depletion of the negatively biased shunt gate enhances signal shielding, avoiding the effect of increased frequency on IL degradation. Moreover, the device exhibits excellent ISO performance in the off-state. Overall, these design characteristics enable the 500 nm gate-length SSC GaN HEMT to achieve an ISO larger than 14 dB, an IL flatness of 0.38 dB across the 2–20 GHz range, and a $R_{\mathbf {on}} C_{\mathbf {off}}$ coefficient of 478 fs. Currently, the SSC HEMT boasts state-of-the-art ISO metrics in single-cell devices, outperforming even some typical circuit topologies. Additionally, the SSC HEMT achieves a maximum output power ($P_{\mathbf {max}}$ ) >37 dBm and an output third-order intercept point (OIP3) = 66 dBm at 3.6 GHz. These results demonstrate the great potential of the SSC HEMT in improving ISO and mitigating IL degradation across a broad frequency band, providing new possibilities for device selection in broadband RF circuits.
Pump–probe thermoreflectance technique encounters challenges in balancing high spatial resolution and high-throughput testing during wafer-level scanning. To address this, we introduce super-resolution reconstruction (SR), an emerging computer vision technique, into semiconductor thermophysical field characterization. By analyzing the spatial isomorphism between thermal field data and visual images, we developed a “ sparse sampling-high resolution reconstruction” framework based on a learnable SRResNet architecture, achieving 4× super-resolution reconstruction of low-resolution thermal property maps obtained through automated scanning. This method overcomes the inherent limitation of traditional point-by-point scanning measurements, which requires a time that increases with the square relationship of spatial resolution (t ∝ N2), enhancing wafer-level thermal property characterization efficiency by over 16-fold at equivalent resolution requirements. Furthermore, to resolve edge smoothing artifacts caused by SR, we developed a correction strategy incorporating transient thermoreflectance edge measurement data, significantly improving the authenticity and accuracy of thermal property distributions. This work systematically explores the application potential of SR in thermal–physical characterization, providing an efficient and rapid approach for semiconductor material analysis that effectively resolves the trade-off between testing time and spatial resolution.