A non-local model for hot-electron injection in the passivation layer of AlGaN/GaN HEMTs

This thesis presents the development and implementation of a non-local hot-electron injection model for AlGaN/GaN high electron mobility transistors (HEMTs), aimed at estimating the current density injected into the passivation layer. The high-energy carriers generated during hard-switching events can be captured by traps present in the passivation layer, leading to dynamic effects such as the degradation of the ON-state resistance. The proposed approach exploits the hot-electron distribution function extracted from single-particle Monte Carlo simulation results and adopts a non-local formulation to account for carrier transport effects. The model is implemented as a Physical Model Interface (PMI) within the Synopsys© Sentaurus TCAD framework, enabling efficient simulation of long-term trapping phenomena with good computational cost. The implemented model is presented in the second chapter and validated through comparison with both a Python implementation of the same equations and Monte Carlo simulation results, showing good agreement in terms of magnitude and spatial localization of the injected current. TCAD simulations leveraging the implemented PMI are then used in the third and fourth chapters to investigate the impact of hot-electron trapping and detrapping on device degradation and recovery under different semi-ON stress conditions. The influence of device geometry, electric field distribution, and trap energy distributions is analyzed, highlighting the key mechanisms responsible for dynamic degradation and incomplete current recovery.