Research on parameter extraction of thin-film transistors based on swarm intelligence

Abstract

The development of integrated circuits for displays and other applications requires semiconductor device models and appropriate parameter extraction techniques to predict and understand the circuit behavior. These techniques are paramount in reducing design errors and shortening the product development cycle. This paper presents an algorithm that employed swarm intelligence in exploring an automated and accurate parameter extraction technology. First, an automatic parameter extraction of Rensselaer Polytechnic Institute (RPI) Model for polysilicon thin-film transistor (Poly-Si TFT) is achieved by genetic algorithm (GA) and particle swarm optimization (PSO) algorithm. Compared with the best solution of the GA algorithm for automatic parameter extraction, the PSO outperformed the GA. However, it still prematurely converges to the suboptimal solution henceforth cannot obtain the expected solution accuracy. Second, the mutual learning particle swarm optimization (MLPSO) algorithm is proposed that introduces the concept of “mutual learning.” The new algorithm aims to find the global optimum in getting suitable trade-off between exploration and exploitation. In addition, the MLPSO algorithm implemented the novel random initialization and fitness function in simplifying the complex manual processes and the empirical calibration, and it led to achieving automatic and accurate parameters extraction.