Advanced nano-scale process control using the back-propagation network and dual filter exponentially weighted moving average

Nonlinear dual filter exponentially weighted moving average (NDEWMA) control methodology is proposed in this paper. NDEWMA control method is designed for the photolithography process. It is assumed that the photolithography process has a nonlinear relationship between recipe and methodology and has a disturbance with relatively very small drift rate compared to white noise. The method is comprised of dual exponentially weighted moving average (EWMA) filter and back propagation network (BPN). Dual EWMA filter is used for reducing the effect of white noise and detecting the exact process changes. BPN is applied to increase the prediction accuracy when the process model has a nonlinear relationship between control parameter (recipe) and methodology. To evaluate the proposed control method, simulation based experiment has been performed with two alternative control methodologies for same condition.