Abstract
High electron-mobility transistors (HEMTs) have emerged as an attractive alternative for high-efficiency power systems, due to their good material properties to perform at high voltages, temperatures, and frequencies. For that reason, design optimization of HEMTs becomes imperative to ensure the quality and capability of the device when in service. There have been models derived from experimentation to guide the design of HEMTs. Nonetheless, due to its expensive manufacturing process, the relationship of the temperature channel with respect to the design parameters has not been investigated thoroughly. This paper presents a multiphysics finite element (FE) simulation to predict the HEMT’s device maximum channel temperature when varying different design parameters. Furthermore, Gaussian Process (GP) based surrogate model was developed using the simulation results as the training database with adaptive sampling techniques for the optimization process. The proposed high-fidelity surrogate model effectively predicts the channel temperature of the HEMT device and enables an optimum search over the design space.
