Characterizing and resolving unobservability in run-to-run control of high mix semiconductor manufacturing

Run to run control is a major tool used in semiconductor manufacturing to keep the unit processes within the required manufacturing constraints. Typically, the difference, or bias, between the desired and actual result of processing a particular wafer is affected by not only the particular product being produced, but the prior processing path, which can complicate the control. Previously, several authors have discussed a method of describing the bias for a particular wafer as a linear combination of fundamental bias causes and then using a Kalman Filter to estimate these biases. One known complication of this method is a system structure that can cause the bias states to be unobservable. In this paper we address a second form of unobservability that can be caused by the production process, (i.e. which products are sent to which tools.) We then analyze two methods for adjusting the estimation process to mitigate this unobservability, one of which is new. The performance of this novel method then will be shown using the simulations.