Statistical models for microwave GaN HEMTs

Statistical model of semiconductor field effect device is important for yield analysis. In this paper, the large signal statistical characterization of GaN HEMTs is modeled based on empirical equivalent circuit model. First, the parameters of small signal and large signal equivalent circuit model are extracted by in-house extraction program based on 10 batches GaN HEMTs. Then, the statistical model is fulfilled by combining principal component analysis, factor analysis, and multiple regressions modeling techniques. Finally, the statistical model is implemented in Agilent-ADS and validated by comparison between Monte Carlo simulation and measurement results. Furthermore, two S-band GaN HEMT power amplifiers are designed by using the established statistical model for demonstration purpose. The results show that good accuracy has been achieved between measurement results and Monte Carlo simulation results. The proposed method will be useful for GaN device yield optimization.