posted on 2024-10-31, 16:27authored byHuiyuan Cheng, Melanie Ooi, Ye Chow Kuang, Eric Sim, Bryan Cheah, Serge Demidenko
Recurring defect cluster patterns on semiconductor wafers can be linked to imperfectness/faults in specific manufacturing processes or alternatively - to failure or malfunctioning of production equipment (in our research we assume that defects associated with deficiencies/errors in the circuit design are not present). By identifying these patterns as they occur, a fast and effective process monitoring and control mechanism can be achieved, shortening the time-to-yield period and reducing the loss in revenue due to avoidable yield drop. Identifying these patterns manually could be a too complex and time consuming task. This research presents an automatic yield management system to extract and identify defect clusters as well as perform yield analysis in a high-volume semiconductor devise manufacturing.