Machine Learning Approaches for Process Optimization

Abstract

We have optimized semiconductor manufacturing processes by machine learning (ML) approaches. The optimization of the nonuniformity of plasma enhanced atomic layer deposition (PEALD) film thickness, the PEALD film stress, the carbon etching profile and the PEALD film thickness profile have been successfully achieved the targets. In some cases (stress and thickness nonuniformity optimization), the ML approach is found to be more powerful than the knowledgeable engineers. The authors showed the effectiveness of ML approach for the semiconductor manufacturing processes.