Wafer Failure Graphic Automatic Identification System Applied Research, Yield Management In Ulsi

A complete Yield Management System is crucial to semiconductor manufacturing as yield is a determining factor in the profitability of a fab.This system manages and analyzes large volume of complicated IC manufacturing data and results in faster yield ramp and efficiently maintain guaranteed yield.This system also creates significant cost savings in manufacturing,manpower,and reducing defective wafers to ensure the company receives the most benefit.Yield management starts with the identification of different wafer failure patterns and this thesis focuses on the research and popularization of "Wafer Pattern Classification System".The Wafer Pattern Classification System is a vast improvement over the traditional method where the engineers manually and subjectively classify different fail patterns.This wafer classification system objectively classifies wafers according to pre-set parameter settings in an automated and efficient manner.This "Wafer Pattern Classification System" is successfully applied in three areas: Automatic Tested Wafer Disposition System,Automatic Downgrade System,and subsequent automatic data analysis.The Wafer Pattern Classification System completes the traditional yield management system and helps engineers to find low yield root causes quickly and significantly reduces the yield ramp cycle time.In the competitive semiconductor industry,a complete and automated data analysis system is crucial in ramping up and maintaining high yield.As the most important portion of Yield Management System,Wafer Pattern Classification System is already successfully used by one of semiconductor manufacturing company to solve a bottleneck of traditional yield management system.