Defect diagnosis using in line product control data in semiconductor industry

Defect diagnosis in semiconductor manufacturing is crucial to improve the product quality and to reduce the production cost. When defect is recognized, the objective is to identify which equipment generates it. This paper defines the problem of different types of equipment failures and the impact on the defect diagnosis using the in line control data of the process. A defect diagnosis based on combination of Tool commonality Analysis and Suspected Equipment Confirmation techniques is proposed. Analysis begins by identifying two data horizons: Equipment horizon that specifies the set of suspected equipment and Lot horizon which specifies the inspected samples that are useful for the analysis. A signature table is used to make a binary decision in order to identify the set of suspected equipment and the computing algorithm is described at the end of the paper with an illustration of a numerical example.