Fault detection and classification of plasma CVD tool

A fault detection and classification (FDC) study by means of principle component analysis (PCA) is conducted on a high density plasma (HDP) chemical vapor deposition (CVD) tool. The design of experiment (DOE) of gas flow and RF power effects is introduced to study the feasibility of PCA for FDC as well as to investigate the correlation of tool parameters. In this study, the first wafer effect of HDP CVD tool is identified and attributed to thermal system variation. It is shown that the PCA model is sensitive to detect (+/-)5% variation of tool parameters. It is also found that the fault space vector method might provide an approach for fault classification.