EQUIPMENT HEALTH MONITORING AND ITS RELATIONSHIP TO WAFER TOX MEASUREMENTS ON A PLASMA ETCH TOOL

摘要

An Equipment Health Monitoring and Advanced Fault Detection (EHM&AFD) system is applied to a dry etch process to determine if changes in the equipment data can be used to predict corresponding changes in the wafer quality measured by oxide thickness (Tox). A single metric derived from multivariate predictions for real-time equipment data was somewhat correlated with Tox measurements. By using the appropriate group of sensors in the multivariate models, the resulting metric (equipment health) showed the possibility to be used for picking out bad wafers on line. Using an empirically set control limit for health, the technique was able to predict with about 80% accuracy which wafers had oxide thickness outside acceptable tolerance limits.