Detection of Defects on SiC Substrate by SEM and Classification Using Deep Learning

Abstract

In recent years, next generation power semiconductor devices using semiconductors with large band gap such as SiC (Silicon Carbide) attract attention. It is very important to detect crystal defects, surface processing defects including polishing, defects contained in the SiC substrate, defects included in the epitaxial growth film, defects caused by the device forming process, and so on. This is because elucidating the cause of the detected defect and investigating the influence on device quality and reliability lead to development of a better manufacturing method. Recently, observation with a low energy scanning electron microscope (LE-SEM) which is more accurate than C-DIC and PL has been put to practical use. As a result, crystal information of just below the outermost surface can also be obtained. However, since image processing techniques targeting SEM images of SiC substrates have not existed so far, it has not been possible to efficiently and automatically extract defects from enormous amounts of data. In this paper, we propose a method for detecting defects on SiC substrate by SEM and classifying them using deep learning.