Reliability Assessment and Health Diagnostic Methods for SiC MOSFET Devices
Time: Fri 2026-02-20 10.00
Location: Kollegiesalen, Brinellvägen 8, Stockholm
Language: English
Subject area: Electrical Engineering
Doctoral student: Bhanu Pratap Singh , Elkraftteknik
Opponent: Professor Francesco Iannuzzo, Politecnico di Torino, Torino, Italy
Supervisor: Associate professor Staffan Norrga, Elkraftteknik; Professor Hans-Peter Nee, Elkraftteknik
QC 20260123
Abstract
The transition toward high-efficiency electrified systems has accelerated the adoption of SiC MOSFET devices, whose performance benefits are often limited by package-related reliability challenges. This thesis investigates these challenges through two complementary research directions. The first focuses on the thermo-mechanical reliability of conventional, single-sided cooled (SSC), and double-sided cooled (DSC) SiC MOSFET packaging structures using finite-element modeling (FEM) in COMSOL Multiphysics. The impact of die placement, advanced interconnection technologies, solder and Ag-sinter materials, and Cu–Mo composite spacers is analyzed to understand temperature distribution, viscoplastic strain accumulation, and solder-layer lifetime under various power-cycling conditions. The results highlight important design trade-offs and identify advanced packaging configurations and materials that improve both thermal and mechanical performance.
The second part of this thesis develops experimental health-diagnostic methods using degradation data obtained from the power-cycling test (PCT) setup. Commercially available TO-247-3 packaged SiC MOSFET devices were degraded using inverse-mode and forward-mode PCTs, enabling a detailed investigation of body-diode forward-voltage reduction, package-related degradation, and ON-state resistance (RdsON) drift in SiC MOSFETs. A compensated RdsON-based diagnostic method is introduced and experimentally validated for the reliable detection of package-related degradation. Additionally, a diagnostic technique for early bond wire failure detection is proposed and experimentally validated.