Etalon-based integrated microchip inspection system

This paper presents an automated etalon-based integrated microchip inspection system, aimed to minimize manual quality control tasks, by analyzing images of integrated microchips taken by electronic microscope. To perform inspection the system requires three to five images of the analyzed region to average them into etalon image; no other a priori information is needed. The system uses rule-based approach, that assigns each defect to one of three groups: 1) shorting of two or more adjacent conductors; 2) islet or pin hole, representing defect contained within conductor's boundaries or in metallization layer; 3) group type of defect including conductor's breakups, mousebites and spurs. The paper proposes two algorithms to tackle two major problems of reference-based systems: slope and alignment correction. To detect topological elements affected by the defect, the algorithm of random polygon intersection is introduced. The decision whether a particular defect is significant is based on the comparison of the defect and affected conductor dimensions.