An Efficient Automatic Design Method for LDMOS Devices Based on Particle Swarm Optimization

In this study, an efficient automatic design method is proposed for laterally-diffused metal-oxide semiconductor (LDMOS) devices. This method integrates Particle Swarm Optimization (PSO) and Deep Neural Network (DNN) model to automatically design device structure parameters that meet target requirements in a multi-dimensional design space. To comprehensively evaluate the performance of this method, the performance of integrating PSO algorithm is compared with that of Bayesian Optimization (BO) algorithm. The results show that PSO algorithm performs excellently in device design and takes much less computing resource and run time than the BO algorithm. This proposed method provides an efficient and intelligent new approach for LDMOS devices design, laying the foundation for the design and process development of power devices.