Leuven | More than two weeks ago
The electrification of the automotive and energy sectors is driving rapid innovation in power devices, with wide-bandgap (WBG) semiconductors, such as gallium nitride (GaN), leading the way. Thanks to these materials, electrical switches and converters can operate at higher switching frequencies and temperatures, greatly enhancing efficiency. However, despite their remarkable performance benefits, WBG power devices also pose several fabrication and reliability challenges. Engineered to function under extreme conditions, these semiconductor devices face stresses that can compromise their long-term reliability. Issues such as thermal management, material degradation, and high electric fields necessitate robust packaging and advanced thermal solutions to ensure the required performance. However, these very packaging and thermal solutions can complicate the process of post-mortem defect localization and analysis, making it more difficult to accurately identify and diagnose issues after device failure.
This project seeks to develop innovative methodologies for conducting failure analysis (FA) on these next-generation power devices, providing deeper insights into their underlying failure mechanisms. The application of imaging techniques for pinpointing failure locations such as lock-in thermography, photon emission microscopy, and magnetic current imaging will be thoroughly investigated and refined to improve both accuracy and detection capabilities. Furthermore, the adoption of machine learning will be explored to enhance data analysis and pattern recognition in identifying failure mechanisms. By leveraging machine learning algorithms, we aim to automate the detection of subtle failure signatures and identify trends in failure modes across large datasets. This approach will accelerate the FA process and provide predictive insights that can inform design improvements and enhance the long-term reliability of WBG semiconductor devices.
Required background: Physics, Electrical Engineering, Materials Science
Type of work: 20% Literature and technological study, 20% simulation, 60% experimental
Supervisor: Claudia Fleischmann
Daily advisor: Kristof J.P. Jacobs, Izabela Kuzma Filipek
The reference code for this position is 2025-103. Mention this reference code on your application form.