Multi-Scale Electrothermal Co-Simulation of GaN HEMTs at the Device and Packaging Levels

Gallium Nitride High Electron Mobility Transistors (GaN HEMTs) are critical for high-power and high-frequency applications but face challenges from self-heating and inadequate thermal management, which degrade performance and reliability. Existing tools like Technology Computer Aided Design (TCAD) and finite element analysis (FEA) lack integrated electrothermal coupling capabilities. This study proposes a multi-scale electrothermal co-simulation framework to bridge device-level electrical behavior and packaging-level thermal dynamics. The methodology integrates 2D TCAD simulations for localized Joule heating in the two-dimensional electron gas (2DEG) region with 3D FEA modeling to evaluate heat dissipation across packaging structures. An iterative loop couples the electrical power density from TCAD into FEA as averaged heat sources, updating boundary conditions until thermal convergence is achieved. Key innovations include a deep neural network (DNN) to resolve highly localized heat distributions from TCAD data and an equivalent thermal conductivity model for computational efficiency. The multi-scale approach successfully captured electrothermal interactions, enabling optimized packaging designs and material adjustments to balance heat generation and power efficiency. This work demonstrates the necessity of a coupled simulation for accurate performance prediction and thermal management in GaN HEMTs, offering a pathway to enhance reliability in high-power applications.