Abstract: Analysis of data gathered during a experiment to demonstrate control of the polysilicon gate etch revealed the possible presence of a first wafer effect, i.e., a different response of the etch metrics for the first wafer run after a delay or pump down. In this paper, we investigate this first wafer effect on spatial etch rate and in situ ellipsometer metrics more thoroughly. The spatial metrics were standard deviation and Median Absolute Deviation from the Median, as well as contrasts (such as average etch rate of fast sites - average of slow sites). Ellipsometer metrics were Mean Etch Rate, initial etch rate, rate of change of the etch rate during etch, initial thickness - estimated initial thickness, and polysilicon loss during the deglaze step. Multivariate Statistical Quality Control statistics of the ellipsometer metrics were also examined. In addition, a comparison of the ellipsometer data with data measured ex situ on a site near the ellipsometer measurement die were made. This document demonstrates that data obtained in situ with an ellipsometer can indicate when the etch rate spatial pattern is different from that expected.!9
展开▼
