Supervisory Control for Semiconductor Processing

A control strategy employing Artificial Intelligence concepts has been developed for supervisory control of batch processing operations in the semiconductor industry. The control strategy integrates statistical quality control and conventional feedback control with a rule-based supervisory control system that can be customized to any desired degree of automation. A set of process-independent criteria has been incorporated into the rule base to determine the appropriate type of action required to return the process to the target value. Incorporation of process-specific data is accomplished through the framework of a simple rule structure. The new control strategy is implemented in a prototype controller usig object-oriented programming. The prototype controller features a graphical user interface and is designed to readily accept control laws derived from off-line empirical process characterization studies, a common form of analysis within the semiconductor industry. Results of a simulation study for a photolithography line are presented and discussed.