A Quality Metric for Defect Inspection Recipes

This paper describes a metric that measures the quality of defect inspection recipes. It takes into account several factors including nonvisual rate, defect of interest rate, defect count per wafer, and inspection time. The calculation runs automatically, and only minimal user input is necessary. Different weighting models allow giving each factor more or less importance, thus making the metric flexible for different applications in development and manufacturing. The result is a final score for each recipe. A list of inspection recipes can be sorted by scores to show that recipe needs optimization. Several examples show the use of the quality metric in IBM´s East Fishkill development facility.