Classification of Killer and Non-Killer Silicon Carbide Epitaxial Defects and Accurate Prediction of Device Yield

摘要

The identification and classification of all defects present in Silicon Carbide device material is very important for the successful proliferation of SiC power devices. The various classes of defects affect not only yield, but can also cause reliability issues. In this work, we present the identification of silicon carbide epitaxial defects and their classifications into various defect types. Further these are categorized into killer and non killer defects and verified by electrical tests on commercial diode wafers. The sources of these classified defects are identified and a clustering model is built, which is verified by comparing yield on wafers fabricated with different die sizes.