Semiconductor Manufacturing

With the development and popularity of sensors installed in manufacturing systems, complex data are collected during manufacturing …

This paper explores the application of reinforcement learning (RL) methods for process control in semiconductor manufacturing, presenting a new RL-based controller that shows significant advantages over traditional run-to-run controllers in dealing with both linear and nonlinear processes.

We propose a novel tensor-based process control and monitoring method for semiconductor manufacturing. This method addresses high-dimensional image-based overlay errors using limited control recipes and monitors control residuals to manage high-dimensional disturbances. The effectiveness of the proposed method is validated through extensive simulations and real case studies.