A Deep Cutoff Capacitance Model for GaN Switch HEMTs

An accurate large-signal model of switch transistors is important for high-power capacity switch design. Conventional nonlinear capacitance models mainly focus on applications in power amplifiers (PAs) and low-noise amplifiers (LNAs). However, for a switch, the nonlinearity in the off state is crucial for insertion loss and output power evaluation and needs to be further investigated. In this article, a deep cutoff capacitance model is proposed based on the quasi-physical zone division (QPZD) model. First, the nonlinearity of switch transistors in the deep cutoff state, mainly induced by the bias dependence of the fringing capacitance (FC), is investigated. Based on the findings, an improved physics-based capacitance model that considers the bias-dependent FC is deduced to realize characterization across the entire region (from deep cutoff to the on state). The proposed capacitance model is embedded into the QPZD large-signal model for validation. A double-layer Shockley–Read–Hall (SRH) trapping model is further used for accurate modeling of both buffer-related traps and surface-related traps. The validation results indicate that the proposed model shows an average accuracy improvement of more than 10% for the S-parameters, 20% for the power, and 40% for the gain. The proposed model can be applied to high-power capacity switch design.