The Research Of Defect Segmentation And Short Critical Area Optimization Based On Graph Theory

In IC (Integrated Circuit) manufacturing, the defect which can cause circuit failurein some critical areas is a major factor in affecting the IC functional yield loss. Defectsegmentation and further analysis are the premise of yield optimization based onrandom defects. And then we carry out the layout optimization research based on thecritical area and the characteristics of defect segmented.Firstly, the paper analyzes the types of the defects which can cause various circuitfailure and the concept of the short critical area based on the random defects. Then weapply an emerging image segmentation technology-IFT(Image Forest Transform) forthe defects segmentation of IC image by studying most existing segmentation method. Anew improved threshold-constrained IFT is proposed on the base of the originalalgorithm, which shortens the execution time and speeds up the execution efficiency.For traditional interactive IFT selecting seeds manually, its time complexity is higher.This paper also proposes a new IFT algorithm that can select the initial seeds automaticbased on edge detection and morphological processing operations. The new algorithmcan produce better segmentation effect at a rapid speed and reduce the time complexityeffectively finally.Finally, according to the characteristics analysis of the defects segmented, thepaper systematically discusses the application of critical path and shortest path in therange selection of yeild optimization in the COE(Critical area On Edge network) whichis based on the graph theory. Meanwhile, optimized threshold can be obtainedsuccessfully by the organic combination of them and the maximum value reachable canalso be found. By moving the nets to be optimized step by step, the IC yield achievesthe best optimization. Therefore, it can contribute to the development of DFM (Designfor Manufacturing) research.