Research On Reconstruction Of SEM Image Of Integrated Circuit Wafer

In the last half a century, the development of IC industry has been keeping its pace with Moore’s Law. As a critical method of pattern transferring for Very Large Scale Integrated Circuits(VLSI) manufacturing, photo-lithography has become more significant. In recent years, the mainstream semiconductor manufacturing technology advances from 90nm,65nm to 40nm,28nm,20nm and below. Smaller Critical Dimension(CD) requires more advanced manufacturing technology, stricter manufacturing environment and fewer manufacturing defects. The photo-lithography has been dramatically affected by the introduction of new process step and the manufacturing environment, and the introduction and environment lead to new defects in lithography pattern. Research on lithography pattern defects on wafer has become a very important topic in semiconductor industry.The lithography pattern defect is defined as the difference of graph between design and wafer, which is beyond allowing. When CD is reduced to nanometer level, the effect of the lithography pattern defects on integrated circuit yield is very serious. After the completion of lithography step, the inspection of lithography pattern defects on wafer is an important means to improve the yield of integrated circuit. Scanning Electron Microscope(SEM) image is an important basis to observe the topology of the wafer and find the lithography pattern defects. The next step lithography pattern size measurement and defect inspection are based on the reconstruction of SEM image and the binary analysis of pattern image.This thesis studies photo-lithography technology, pattern defect, SEM image and the preprocessing of image, and particularly aiming at the reconstruction of gray image of chip surface from SEM, it studies the method of reconstruction based on edge information and puts forward a method about reconstruction based on the gradient of SEM image and methods of statistics. The presented algorithm uses histogram to analyze the noise of SEM image, and filters the noise, and gets the gradient information of image with Kirsch operator, then according to the characteristic that the gradient of outer edge is larger than inner edge, the algorithm uses statistic information to fill the image. Experimental results show that compared with the edge detection algorithm, with the same scanning area, the algorithm has higher degree of automation under the premise of the same stability in high resolution images and in low resolution images, it avoids the effect of edge extracting failing and improves the success ratio of image reconstruction. The algorithm is also useful for different types of SEM images.