Automatic Classification Using Decision Tree and Support Vector Machine

Abstract

The EDS wafer test yield is the most important criteria to evaluate FAB’s productivity, so the manufacturing operation’s main purpose is to secure new product yield early and maintaining the yield of mass-produced products high. Defining a failed characteristic that’s compatible to the device and classifying wafers depending on failure type helps tasks searching for error from FAB become automated. This would be more efficient then existing failed analysis operations and strive to become the basis for improvement in yield and quality. For this method, this research is trying to use a high speed recognition algorithm called SVM (support vector machine) that will define wafer’s failed type and automatically classify each one.