Statistically based feedback control of photoresist application

The authors present a statistically based algorithm for the implementation of run-by-run feedback control on a semiconductor manufacturing process. The method is based on a multivariate cumulative-sum statistical process control procedure. This procedure monitors multiple process parameters in order to identify significant deviations from a given target. Once such a deviation is detected, the algorithm uses adaptive response surface models of the process in order to change the equipment settings. This procedure has been used to control the thickness and the photoactive compound concentration of photoresist applied an oxidized silicon wafers. In this application, feedback control is based on inexpensive thickness and reflectance measurements that can be totally automated. The experiments, conducted in the Berkeley Microfabrication Laboratory, show that this method can significantly improve the photoresist application process