The Detection of Plasma Endpoint Process Using OES-Wavelet-Modified Univariate Charts

Abstract

As the exposed area to the film being etched is very thin (< 0.5%), showing the incapacity of the traditional endpoint detection method to determine the endpoint in Plasma etch process. A new proposed Algorithm designed in order to pick the optimal level of decomposition (OLD) for wavelet analysis and then a modified wavelet univariate control charts X bar is presented to detect endpoint based on the mean, SD and CV respectively. When the mean and the variance are not stable and the coefficient of variation is steady, its application to details coefficients enhances the ability of detecting endpoint in plasma etch process. The control limits are also Constructed from the main etch interval for mean and from Over etch interval to ameliorate the EPD.