Abstract
Many IC manufacturing processes have, by nature, Multiple Input and Multiple Output (MIMO) system. For a drifted MIMO system, the double Multivariate Exponentially Weighted Moving Average (MEWMA) feedback controller is a popular run-to-run (R2R) control scheme. Although the double MEWMA controller can guarantee a long-term stability, it usually requires a moderately large number of runs to bring the output of a process to its target. This is impractical for process with small batches. The reason is that the output deviation is usually large at the beginning of the first few runs. In order to reduce a possibly high rework rate, in this paper we introduce a variable MEWMA scheme to eliminate the initial bias as soon as possible, even though the inaccuracy of model estimation. Using the criterion of minimizing the total mean square error (TMSE), the optimal variable MEWMA controller can be obtained. An example is given to demonstrate the performance. Finally, the comparisons between variable MEWMA and double MEWMA controllers are addressed at the end of this thesis.