Yield prediction using critical area analysis with inline defect data

Abstract

This paper presents methodologies for using critical area analysis with inline defect data to predict random defect limited yield and for partitioning yield losses by process step. The procedure involves (1) calculating critical areas, (2) modeling defect size distributions, and (3) combining critical area information and defect size distributions to estimate yield loss. We introduce a method to model defect size distribution from inline defect data. We develop two yield prediction methods that can overcome the difficulties caused by the inaccuracies in determining defect size when using laser scatterometry detection. We compare the predicted yield with the actual yield and show that the two are in good agreement.