Smart Sampling Methodology for Yield Defect Inspection in a 200mm Foundry Wafer Fab

Abstract

In Semiconductor Manufacturing, inline Inspection monitoring by Brightfield and Darkfield scan tool platform is a 2nd line of defence to detect abnormal events to prevent defectivity escape to Wafer Sort or Assembly Test. Early detection and good process tool coverage are essential to the quality of wafers. These have to be balanced with competitive cycle time and cost. There are many newer scan tool, software and manufacturing systems available today that can optimize this sampling monitoring. As a legacy 200mm foundry wafer fab, system upgrade may not be feasible or can be expensive. As a result, an innovative solution had to be developed in order to stay relevant in the semiconductor industry. This paper introduces an inline inspection sampling Artificial Intelligent (AI) methodology to ensure maximum process tool scan coverage without additional CAPEX cost - getting more out from the same capacity; getting more coverage with same scan capacity and improve line of defence.