In the fabrication process of wide bandgap gallium nitride high electron mobility transistors (GaN HEMT), due to the process error, process fluctuation, material defects, and other factors, wide bandgap semiconductor transistors produced in different batches will also have dispersed fluctuation of electrical characteristics under these special working conditions, which results in deviations between ideal software simulation parameters based on device model and the actual measurement parameters. Some statistical analysis methods like principal component analysis, factor analysis, and multivariate statistical regression in this article are introduced to process S-parameter which can be measured from different batches of CGH40010GaN HEMT and statistical model parameters can be got from them. Error of S-parameter comparison are less than 10% between calculation results of established small-signal statistical equivalent circuit model (ECM) and S-parameter sample measurement results. Monte Carlo simulation based on statistical model can be well adopted to predict dispersed fluctuation of electrical characteristics for different samples and which can be used as the basis of large-signal ECM statistical model.

No abstract