A cascaded iterative learning motion control scheme for an ultra-precision dual-stage actuated wafer stage

This paper proposes a cascaded iterative learning control (ILC) scheme for the developed wafer stage to achieve ultra-precision motion control performance. In practical motion systems, standard iterative learning control is frequently utilized to realize high tracking performance. However, Q-filter of the standard ILC inevitably causes residual tracking error as a result of a limited cut-off frequency and its non-ideal magnitude characteristic below the cut-off frequency. Therefore, a cascaded ILC scheme is developed to further eliminate the residual tracking error. Specifically, once the tracking error converges to its limit under standard ILC, the learned feedforward signal is frozen and injected to the control loop as constant feedforward. Based on the forzen feedforward, new learning process begins. The same operations are repeated several times until no more improvement can be obtained. Comparative experiments are conducted on our developed wafer stage which consists of a short-stroke stage for fine positioning and a long-stroke stage for coarse positioning. The results show that the proposed cascaded ILC scheme can significantly improve the tracking performance - the moving average (MA) of tracking error is decreased from 14nm to 3nm and the settling time is reduced from 17ms to 4ms. The proposed precision motion control scheme is also suitable for application in practical semiconductor manufacturing equipments and robotic systems.