Adaptive EWMA controller based on dynamic modeling in semiconductor manufacturing

Abstract

In semiconductor manufacturing, it is important to produce multiple products on the same equipment to enhance the overall equipment effectiveness so as to improve the productivity. However, the "high-mix" production is difficult to control due to the time-varying model. To address this problem, an adaptive exponentially weighted moving average (EWMA) control method of which the core content is online dynamic modeling is put forward. During dynamic modeling process, the noise disturbance can be predicted through EWMA controller; meanwhile, the process gain can be estimated by using the predicted noise disturbance. Case studies of the chemical mechanical polishing (CMP) unit operation show the efficacy and the attractiveness of the strategy.