An optimization of GaN Physics-based Compact Model Based on Manifold Mapping Algorithm

GaN physics-based compact models cannot be modelled without accurate structural parameters. In this paper, the structural parameters are optimised using a manifold mapping algorithm, with the compact model as the coarse model, the TCAD model as the fine model, and the real device as the target. When the fine model response agrees with the target response, the structural parameters are then determined. Finally, a genetic algorithm is used to make the coarse model response close to the fine model response, at which point the compact model modelling is completed.