In advanced semiconductor manufacturing processes, adjustment techniques that change the process recipe form run to run (R2R) are usually needed given the deterioration and random characteristics that these processes exhibits with time. In the last three years, we have developed new quality control (adjustment) algorithms for use in semiconductor manufacturing based on Optimizing adaptive Control techniques. Recent work has focused on the mathematical analysis of existing methods based on the so-called EWMA (exponentially-weighted-moving-average) statistic, a widely used family of controllers but not fully understood in practice. Both single-EWMA and double EWMA run to run controllers were studied, and their robustness and stability properties studied. This analysis includes new optimization methods for the tuning of EWMA controllers. In addition, a constrained Proportional-Integral(PI) controller has been developed for R2R control. This new controller tunes itself its parameters to achieve the minimum output variance while at the same time keeping the input variance below a specified upper bound. This is important for manufacturers as traditional PI and minimum variance controllers transfer the variability of the quality characteristic to the recipe (i.e., the input). The constrained PI controller is simpler to use than Box-Luceno's [6] constrained controllers. This paper reviews recent developments in EWMA and PI controllers for semiconductor manufacturing quality control.
展开▼
