A Case Study of Deploying Run-to-Run Strategy with Deep Learning Model in High-Mix Semiconductor Manufacturing

In the past ten years, primarily driven by a tremendous expansion in the availability of data and computing power, artificial intelligence (AI) and machine learning (ML) technologies have found their way into many different areas and disruptively changed our life and way of solving problems. For example, they refine results in online search and shopping, customize advertising, tailor news feeds, and even drive cars. Recently, the ability of AI/ML to learn from data autonomously and quickly find patterns and correlations has found its applications in metrology and inspection in the semiconductor manufacturing industry. In the past few years, there has been some great work on applying AI/ML technologies in run-to-run (R2R) control strategy design and implementation., e.g., Jim Redman gave a tutorial on AI for semiconductor manufacturing, and Ivan Chen evaluated different neural networks in R2R control modeling and compared their performance with traditional exponentially weighted moving average (EWMA) models, etc. Motivated by their work, this paper presents a case study of applying a recurrent neural network (RNN) deep learning model in a deposition control strategy that we built with Discover Run-to-Run software.