A hierarchical intelligent methodology for spatiotemporal control of wafer temperature in rapid thermal processing

Rapid thermal processing (RTP) is very important for semiconductor manufacturing. The RTP is actually a spatially distributed dynamical system (SDDS) with multiple control sources. Due to its highly nonlinear and spatiotemporal dynamics, it would be difficult to maintain the temperature uniformity inside the RTP. In this paper, a hierarchical intelligent methodology is proposed for the spatiotemporal control of RTP. The control scheme consists of three different levels. At the lower level, the decomposition strategy is taken to divide the SDDS into several relatively simple subsystems; each of them is handled by one control source. At the mid-level, 3-D fuzzy logic controllers (3-D-FLCs) are designed, which have inherent capability to process spatiotemporal dynamics, to maintain the temperature uniformity of each subsystem. At the upper level, a local coordination strategy will be designed to adjust 3-D-FLCs to overcome interactions between the adjacent subsystems. The proposed scheme is applied to the temperature uniformity control of a rapid thermal chemical vapor deposition system, and better temperature uniformity can be achieved.