Measuring Photolithographic Overlay Accuracy and Critical Dimensions by Correlating Binarized Laplacian of Gaussian Convolutions
Issue No. 01 - January (1988 vol. 10)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/34.3864
<p>A technique is described for measuring overlay accuracy and critical dimensions in IC manufacture and similar fields, based on a theory originally developed for matching binocular stereo images. The method uses targets composed of small elements that can be at the minimum feature size of the photolithographic process. Alignment is measured using clusters of those elements rather than the small elements individually. This makes the method insensitive to many of the imaging effects that have plagued other approaches, such as interference fringes and edge topology differences between process steps. The method is tolerant of high noise levels, which allows operation on process layers that produce low-contrast images or high-noise backgrounds as is the case when aligning resist over metal. Adding an appropriate bar grating to the alignment target causes element size changes to induce a proportional shift in alignment, allowing critical dimensions to be measured by the alignment technique.</p>
computer vision; positioning accuracy measurement; convolution correlation; picture element clusters; noise tolerance; resist alignment; feature matching; pattern recognition; photolithographic overlay accuracy; critical dimensions; binarized Laplacian of Gaussian convolutions; IC manufacture; interference fringes; edge topology differences; low-contrast images; bar grating; computer vision; computerised pattern recognition; integrated circuit technology; Laplace transforms; photolithography
H. Nishihara and P. Crossley, "Measuring Photolithographic Overlay Accuracy and Critical Dimensions by Correlating Binarized Laplacian of Gaussian Convolutions," in IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 10, no. , pp. 17-30, 1988.