An effective method for yield enhancement using zonal defect recognition

This paper presents a zonal defect recognition algorithm that can be applied to the problem of automated defect recognition in semiconductor manufacturing, The algorithm uses digital image processing techniques and template matching to achieve a high recognition rate by classifying zonal defects into one of three defect types: clustered defects, gross defects, or repetitive defects. The feasibility of using the algorithm for automated detection was demonstrated experimentally. Application of the algorithm to actual production lines is expected to contribute to rapid yield ramp-up.