Automatic detection of semiconductor mask defects

Abstract This paper presents the method and device for the detection of mask defects, referring to the masks used for semiconductor production. A brief outline of the method is as follows: the examined mask image is approximated to the set of M × N points by means of the set of converters. Each point acquires the value 0 or 1, according to the degree of lightness of the corresponding sector of the image. Then, the formal criteria of belonging to the category of interferences and defects or to the group of correct images are set for each point of the image. Detection of the defects is performed independently of the examined mask positions, rotations or shifts in the field of the device lens. The results of the developed algorithm action on the mask graphic images are also presented. An idea of the specialist system construction for the automatic detection of mask defects is also presented.