Wafer Defect Classification Based on DCNN Model

Abstract

Wafer defect classification is essential in semiconductor manufacturing for fast response of equipment and process stability monitoring, it is also critical for product yield management. Manual defect classification is time-consuming and prone to errors. This study presents an automatic defect classification (ADC) method based on a deep convolution neutral network (DCNN) model. The trained model has proven itself to be able to achieve defect classification performance sufficiently good to serve in the Fab.