Research on a Machine Learning Approach to Classifying Radiation Degradation on Epoxy Molding Compound

This study investigated the radiation degradation behavior of epoxy molding compounds (EMCs) for reliability studies of power semiconductor materials used in nuclear environments. EMC samples were exposed to different cumulative doses through Total Ionizing Dose (TID) evaluation and analyzed by Fourier Transform Infrared (FTIR) spectroscopy to measure chemical changes. A machine learning model was developed based on the FTIR spectral data to identify chemical changes in EMCs as a function of cumulative dose. Thermal conductivity measurements confirmed the effect of cumulative dose on thermal properties, but no changes in mechanical properties were observed.