Machine learning versus Deep Learning in Low Yield Wafer Map Classification

Wafer maps provide important information in investigating the root causes of low yields during semiconductor manufacturing processes. This paper discusses and compares two methods of wafer map classification, machine learning and deep learning. Wafer map classification is important, because different wafer map defect signatures can be attributed to different root causes for low yield investigation. The automated classification of wafer maps will allow for early defect detection which leads to improved yield enhancement.