Abstract

This paper explores the application of principal component analysis (PCA) to the monitoring of within-lot and between-lot plasma variations that occur in a plasma etching chamber used in semiconductor manufacturing, as observed through Optical Emission Spectroscopy (OES) analysis of the chamber exhaust. Using PCA, patterns that are difficult to identify in the 2048-dimension OES data are condensed into a small number of principle components (PCs). It is shown, with the aid of experimental data, that by simply tracking changes in the directions of these PCs both inter-lot and intra-lot patterns can be identified.