Skip to Main Content
The goal of this paper is to develop a multi-step optimization framework for a semiconductor manufacturing process. A simple physics model of a flash memory cell is presented to demonstrate the idea, and a multi-step supervisory control strategy is simulated on a flash memory cell manufacturing process based on this model. The multistep supervisory control strategy performs multiple optimizations of target inputs for a lot with newly available metrology after each step in the course of its processing. The proposed strategy is shown to confer a significant improvement in the control of product quality over a supervisory control strategy which uses a one-time optimization and an open loop implementation. This improvement is primarily due to the ability of the multi-step optimizer to compensate for the effect of error introduced at earlier steps using end-of-step, metrology.