Optimizing return on inspection trough defectivity smart sampling

In this paper we describe the use of an index to continuously monitor the risk of producing defective parts in a semiconductor production line. This index is used by product inspection policy to select the best lot to be controlled in order to decrease uncertainty on production tool's health. The approach was tested on data straight from STMicroelectronics Crolles300 production line. We propose to evaluate the efficiency of the inspection policy based on the percentage of wafers inspected unnecessarily from the total number of inspected wafers.