Wafer Defect Root Cause Analysis with Partial Trajectory Regression

Identifying upstream processes potentially responsible for wafer defects is challenging due to the inherent variability in processing routes, which arises from factors such as reworks, the randomness of process waiting times, etc. This paper presents a novel framework for root cause analysis, called Partial Trajectory Regression (PTR), which leverages recent advances in representation learning and explainable AI. PTR is designed to handle variable-length process trajectories and timestamp sequences. We demonstrate its effectiveness on real wafer history data from the NY CREATES fab in Albany.