A parameter extraction method based on Particle Swarm Optimization

Abstract

Model parameter extraction plays an important role in bridging semiconductor manufacturing and integrated circuit design. Most commercial extraction and optimization tools have a default extraction and optimization procedure. PSO (Particle Swarm Optimization) is a global random searching algorithm with swarm intelligence, having the strong searching capability on nonlinear problems [1-2]. Compared with Genetic Algorithm, PSO does not require crossover and mutation operations. It can be easily understood and implemented, and has high velocity of convergence as well. Thus, it can find the optimal solution quickly during the parameter extraction [3-4]. In this work, a parameter extraction and optimization strategy using PSO is presented for SOI MOSFETs based on the BSIM SOI 3.1 model which is developed by the BSIM group of UC Berkeley [5]. The global optimal strategy and standard PSO algorithm are implemented to fulfill the extraction of direct-current parameters, which are related to gate voltage and drain voltage.