Precise Visual Inspection Algorithm for LSI Wafer Patterns Using Grayscale Image Comparison

A sub-micron defect detection algorithm for LSI wafer patterns has been developed. This algorithm is based on a comparison of corresponding images of 2 chips or cells. Two grayscale images are aligned by their detected edge patterns and compared by the new algorithm called Local Perturbation Pattern Hatching, by shifting one image in 8 plane-directions and in grayscale, and finding the best match in a local window between the shifted images and the other image. The resulting unmatched regions are recognized as defects. This method is performed so as to detect 0.3 p m defects of photoresist patterns without being influenced by small irregularities of the grayscale images.