Physical Mechanisms Affecting Hot Carrier-Induced semi-ON State Degradation in Gallium Nitride HEMTs

Abstract

Gallium Nitride or GaN-based high electron mobility transistors (HEMTs) is currently the most promising device technology in several key military and civilian applications due to excellent high-power as well as high-frequency performance. Even though the performance figures are outstanding, GaN-based HEMTs are not as mature as some competing technologies, which means that establishing the reliability of the technology is important to enable use in critical applications. The objective of this research is to understand the physical mechanisms affecting the reliability of GaN HEMTs at moderate drain biases (typically Vds < 30 V in the devices considered here). The degradation in device performance is believed to be due to the formation or modification of charged defects near the interface by hydrogen depassivation processes (due to electron-activated hydrogen removal) from energetic carriers. A rate-equation describing the defect generation process is formulated based on this assumption. A combination of ensemble Monte-Carlo (EMC) simulation statistics, ab-initio density functional theory (DFT) calculations, and accelerated stress experiments is used to relate the candidate defects to the overall degradation behavior (Vt and gm). The focus of this work is on the ‘semi-ON’ mode of transistor operation in which the degradation is usually observed to be at its highest. This semi-ON state is reasonably close to the biasing region of class-AB high power amplifiers, which are popular because of the combination of high efficiency and low distortion that is associated with this configuration. The carrier-energy distributions are obtained using an EMC simulator that was developed specifically for III-V HFETs. The rate equation is used to model the degradation at different operating conditions as well as longer stress times from the result of one short duration stress test, by utilizing the carrier-energy distribution obtained from EMC simulations for one baseline condition. This work also attempts to identify the spatial location of these defects, and how this impacts the Vt shift and gm degradation of the devices.