FURI | Fall 2024

TCAD Simulations Aided by Physics-Informed Neural Networks

Technology-Computer-Aided-Design (TCAD) solvers are highly accurate, yet computationally expensive tools to simulate electronics device behavior. Photoconductive Semiconductor Switches (PCSS) are optoelectronic devices used in high-frequency amplification and power applications. This research project aims to compare runtime and accuracy of the TCAD solvers against the physics-informed neural network (PINN) framework by fitting PCSS device simulation data into the neural network. By implementing a PINN, this research aims to improve the efficiency of computational resources and runtime because of TCAD’s complex and tedious nature.