Optimal Z-axis Find Algorithm in Ellipsometry Semiconductor Process based on Local Search using Machine Vision

Abstract

By the latest method, wafers of semiconductors have been sliced very thin for manufacturing efficiency, and the manufacturing process of stacking various thin films has been used. In order to measure such a thin film during the semiconductor manufacturing process, an Elipsometer, a non-destructive optical device, is used. Ellipsometer analyzes the thin film by checking the change in the polarization state of the incident light after the light irradiated to the wafer surface is reflected from the incident surface. However, thinly sliced wafers are often bent during the manufacturing process, so in industrial sites Therefore, it was difficult to efficiently measure the thin film by maintaining an accurate optical state. Accordingly, this study analyzed data based on the image of Machine Vision and compared algorithms that efficiently enable precise measurement on vented wafers by using it and changing the Z axis. Thus, we propose a focusing optimizing algorithm based on machine vision image processing and evaluate the data and features to support it, and we open data sets and algorithm codes that can prove this process in GitHub repository1. In addition, the efficiency of these algorithms was interpreted through simulation figures, and through this, an optical system capable of precise measurement applying a method of efficiently moving the Z-axis is proposed.