Search

Due to narrower bandgap and lower critical electric field, silicon (Si) power devices have reached their limit in terms of the maximum blocking voltage capability. Exploiting this limitation, wide bandgap devices, namely silicon carbide (SiC) and gallium nitride (GaN) devices, are increasingly encroaching on the lucrative power electronics market. Unlike GaN, SiC devices can exploit most of the established fabrication techniques of Si power devices. Having substrate of the same material, vertical device structures with higher breakdown capabilities are feasible in SiC, unlike their GaN counterpart. Also, the excellent thermal conductivity of SiC, compared to GaN and Si, let SiC devices operate at higher temperatures (~ 300°C). Hence, a more compact and cost-effective power electronic system can be designed with SiC devices without cumbersome cooling requirements. Specifically, SiC power MOSFETs have started to dominate applications such as three-phase inverters, PWM rectifiers, DC-DC converters in the 1.2 kV – 3.3 kV voltage range. However, SiC power MOSFET's full potential can only be harnessed with accurate and efficient simulation tools that enable optimally designed systems without multiple cost and time-consuming prototyping. Significant work has already been done on the compact modeling of SiC power MOSFET. However, those works focus on the first quadrant behavior. The third quadrant behavior, especially the gate bias dependent body diode characteristics, needs proper modeling for synchronous rectification, freewheeling diode action, dead time optimization, and EMI analysis. Further, the existing physics-based compact models lack efficient and continuous temperature scaling, which is essential for efficient and accurate simulation of power electronic systems with significant self-heating. This dissertation reports on an accurate and efficient electro-thermal model of the SiC power MOSFET that will include the third quadrant behavior with the body diode. In modeling body diode characteristics, the gate bias dependency and reverse recovery are included for the first time. For accurate switching characteristics, gate dependency of the input capacitance has been included. Accurate Miller capacitance modeling both at very low bias and at very high bias is ensured without adding too many model parameters. Avalanche-induced breakdown characteristics are included to make the model capable of predicting Safe Operating Area (SOA). Double pulse tests (DPT) at various temperatures were performed to validate the model's switching characteristics accuracy. A buck converter was implemented with the same half-bridge configuration as the DPTs to validate the model's performance with self-heating effects. Various other power electronics topologies are simulated to validate the accuracy and efficiency of the developed model. Detailed comparison with a previous physics-based model and a vendor-provided empirical model highlights the importance of the developed model in terms of accuracy and efficiency. Lastly, an easy-to-follow parameter extraction procedure has been described to enable broader use of the model among power electronics designers.