Nonlinear Run-to-Run Controller for Semiconductor Manufacturing C. Zhang / Advisor: J. S. Baras -- Systems Engineering and Integration Laboratory

Future Work Run-to-run control is a generic methodology in control of semiconductor manufacturing processes. In semiconductor manufacturing, if the perturbations are small enough, they can be compensated successfully using the exponential weighted moving average (EWMA) method. But, unfortunately, this is not always the case. For example, many plasma processes have been shown to exhibit small to large nonlinearity in behavior. Furthermore, the photoresist process requires dynamic process models too. Thus it is necessary to develop nonlinear algorithms to solve this type of problems. We developed the setvalued run-to-run controller (SVA) to control nonlinear processes. The SVA combines the set-valued method with the ellipsoid algorithm and tracks the process model in each run for multi-targets. The ellipsoid algorithm provides an approximate minimum bound on the parameter sets. Identifying the process model within the parameter set makes the identification very efficient. Different from other conventional RtR controllers, it is applicable for various types of noises and disturbances, which are present in real industrial processes. Furthermore, it is robust to model errors and sensor errors. It can be used for both linear and nonlinear processes.