Development of an Eccentricity Computation System using Image Processing

Thispaperproposesadesignmethodforaprecisionwaferpositioningsystemusinganeccentricity computation system. That wafer positioning system is useful for improving the semiconductor manufacturing process. The CD-SEM is a tool that can be used to improve semiconductor manufacturing processes. The CD- SEM measures the dimensions of fine circuit patterns with electron beams. In recent years, the design rule of less than 2 nm is required, and its density has also increased. There is a wafer positioning system in CD-SEM. In order to identify the dimensions of the circuit pattern to be measured, the Si wafer is placed in a reference position by the wafer poisoning system. The eccentricity of the wafer center position for the center of rotation is computed to place the wafer at the reference position. Conventional eccentricity computation methods have been computed based on the wafer circumference data. However, the conventional methods have two issues. The first issue is the long data sampling time. The second issue is that the computation accuracy is degraded by disturbances. To reduce the data sampling time, this paper proposes an eccentricity computation system that computes the eccentricity by image processing using an image of an entire wafer. At first, this paper mentions a wafer imaging system for capturing wafer images. The wafer is placed on a stage and the entire wafer is imaged from overhead using a digital camera. The brightness of the surface of the wafer is adjusted by a light box so that the edge of the wafer can be clearly recognized. Then, this paper mentions a method to compute the eccentricity of the wafer. The contour information of the captured wafer image is obtained by the Monte Carlo method. Based on the Monte Carlo method, multiple circular images are generated by randomly determining the center coordinates and radius. The degree of conformity of the contour information between the captured images and the generated images is used as an evaluation value. The eccentricity is the difference between the coordinates, which points to the maximum value and the origin coordinates. This algorithm computes the goodness-of-fit of the contour information from a combination of uniform random numbers. Therefore, it can compute the eccentricity with an accuracy less than the resolution of the image by searching for the center coordinates that maximize the evaluation value. As a result, we propose a precision wafer positioning system based on a new eccentricity computation algorithm.