Observability and state estimation for multiple product control in semiconductor manufacturing

Run-to-run control in semiconductor manufacturing is complicated by the use of multiple processing tools and many different products being made during a given time period. This paper investigates a Kalman filter-based state estimation scheme that views a manufacturing area with all the tools, products, and processes it contains as a single interrelated system. This formulation maximizes the amount of information that is shared across different batches by capturing their common characteristics in shared parameters. The estimation scheme performs state updates correctly even when measurement data is missing or delayed. A set of simulations are used to demonstrate the performance of the algorithm under different operating conditions.