I. Introduction
Owning to their characteristics of fast switching speed, low switching loss, and high operating temperature, SiC MOSFETs have gained popularity in numerous industrial applications, like transportation electrification and renewable energy systems [1]–[7]. Lots of research efforts have been made on the power converter design to achieve high efficiency and high power density [8]. To fully utilize the potential of SiC MOSFET, and accurate model is required to predict the device performance before fabrication. According to [9], the SiC MOSFET model can be categorized into five groups: 1) behavioral model; 2) semi-physics based model; 3) physic based model; 4) semi numerical based model; and 5) numerical model. The most commonly used models are behavioral model and physic based model. Compared with behavioral models, physics based models require users to be familiar with the device whole fabrication process and structure information, which makes these models less convenient and more complicate [10]. On the other hand, behavioral models use simple I-V and C-V curves provided by manufacture datasheet or curve tracer to extract the parameters required for the model. In [11], the SiC MOSFET is modelled as one drain-to-source resistance and two constant parasitic capacitances. But the non-linearity of parasitic capacitances with device drain-to-source voltage is not considered and the drain-to-source capacitance is omitted in the model. To improve the model accuracy, in [12], all the parasitics and their non-linearity are considered to construct an accurate model to estimate the SiC MOSFET switching loss. To reduce the model complexity and get closed-form solution, some assumptions and simplifications are made to achieve a trade-off between the model accuracy and complexity. Massive computations are required, which is not convenient. In [10], the SiC MOSFET model is built by extracting the parameters from static characterization results and the model is implemented in Saber. The model shows that the simulation results match the experimental results and can provide high accuracy to evaluate the device performance.