Abstract: FDC data from inline sensors can be used to better control PCM variation using multivariate modeling. However, identifying which FDC sensor contributes most to PCM variability has been challenging since individual process equipment has hundreds of sensors, and the total number of sensors over the production line can be tens of thousands or more. In this paper, we demonstrate PCM variability control with multivariate modeling in a 65 nm mass production line using centralized FDC data management system.
Keywords: Predictive Learning, FDC, Yield Modeling, Virtual Metrology