Doctoral Dissertation
The high frequency optimization of the wide bandgap power devices
The development and commercialization of wide-bandgap power semiconductors enable new opportunities in the applications that extend the switching frequency of power conversion systems from hundreds of kilohertz to the multi-MHz range. The increasing switching frequency can reduce the energy storage requirements of the passive elements, leading to higher power densities (which reduces size) or even eliminating magnetic cores (which reduces weight) for the inductors in the power converters. Candidates like gallium nitride (GaN) and silicon carbide (SiC) devices lead this evolution with excellent electrical properties like high critical fields and thermal conductivity. Prior literature has characterized numerous commercial GaN and SiC devices and discovered significant large-signal off-state Coss charge-voltage hysteresis (namely, Coss loss). Unfortunately, the loss is not currently documented in the datasheet or modeled in manufacturers’ SPICE models, which imposes challenges on circuit designers since it significantly degrades the efficiency and performance of power converters. This thesis addresses the issues of Coss loss in GaN and SiC power devices from the characterization, loss trends, root locations and causes, and optimization guidelines using circuit techniques and TCAD simulations.
Publications
- Analytical Circuit Model for COSS Losses in SiC Junction Termination Extensions ∙ PCIM Europe ∙ 2022
- Small- and Large-Signal Dynamic Output Capacitance and Energy Loss in GaN-on-Si Power HEMTs ∙ IEEE Trans. Electron Devices ∙ 2021
- A physical investigation of large-signal dynamic output capacitance and energy loss in GaN-on-Si power HEMTs at high-frequency applications ∙ IEEE ECCE ∙ 2020
- Demonstration of GaN Impact Ionization Avalanche Transit-Time (IMPATT) Diode ∙ IEEE DRC ∙ 2020
- An Investigation into the Causes of COSS Losses in GaN-on-Si HEMTs ∙ IEEE COMPEL ∙ 2019
- Substrate Bias Effect on E-Mode GaN-on-Si HEMT Coss Losses ∙ IEEE WiPDA ∙ 2018