Abstract
Run-to-run (R2R) control allows modification of a product recipe between machine runs in order to reduce the process variation and quality loss to empower intelligent manufacturing for semiconductor industry. Since process parameters are defined off-line and updated passively after discovering excursions and yield loss. Process drift detection or defect diagnoses and their prevention are increasingly crucial as wafer fabrication reaching nano technology nodes. Focusing on real settings, this study aims to propose a framework that employs the concept of Industry 3.5 as a hybrid strategy. In particular, parameter’s self-tuning mechanism is developed to analyze data of R2R control for early detection of the parameters and thus recommend adjusted parameters to prevent the process drift and vibration for “dynamic rolling optimization”. A decision support system embedded the developed solutions is developed and validated in a leading semiconductor manufacturing company in Taiwan.