Abstract
The RCA cleaning method is the industry standard way to clean silicon wafers, where the temperature control of the cleaning solutions is important for a stable cleaning performance but it involves difficulties such that many of the, cleaning solutions show nonlinear and time-varying exothermic chemical reactions, the system has long and fluctuating time lags owing that the cleaning system has special equipments for dealing with corrosive cleaning solutions, and so on. Here, we first present a thermal model of the system, where by means of the DSC (differential scanning calorimetry) method we analyze exothermic chemical reactions of the cleaning solutions such as SPM (sulfuric acid/hydrogen peroxide mixture), APM (ammonia/hydrogen peroxide mixture) and HPM (hydrochloric acid/hydrogen peroxide mixture). To control the temperature of the solutions, we use adaptive predictive controller, where the adaptive method is used to cope with the nonlinear and time-varying exothermic reactions, and the predictive method is for overcoming the problems on the time lags. Further, the objective trajectory is designed for limiting the overshoot and eliminating the steady state error, and a new sampling method called virtual sampling method is introduced for reducing the required memory size and the calculation time. We show the usefulness of the present thermal model and verify the performance of the present controller through computer simulation as well as the control of an actual system.
