In Line Real Time Yield Impact Prediction by Regression Method

摘要

In process of semiconductor, some defect inspectors are used to scan defect at some process stages. SPC control limit of defect can be defined to control process quality. However it is difficult for defect reduction engineer to estimate to what extend yield will be impacted after defect inspection, and to determine whether contaminated wafer should be scraped. Many defective wafers are released for further processing and then scraped after yield testing because of low yield. Process cost is thus wasted on scraped wafers. The cost can be reduced if yield impact can be predicted. At "In line real time yield impact prediction model" is derived from regression method. Yield impact can be predicted by the regression line of defective dice (DD) and defective bad dice (DBD) after obtaining in line defect inspection data.