Multivariate Statistical Process Control in Etching Process

The purposes of multivariate statistical process control (MSPC) are to enhance process ability by quickly detecting process abnormalities and pinpointing the specific cause of the fault. This paper presents the implementing of MSPC system in etching process including principal component analysis (PCA), and fault contributions from square prediction error (SPE) statistic and Hotelling's T2 statistic. In addition, a combined index called PHI is introduced, which can incorporate information from the SPE and Hotelling's T2 to make fault monitoring more conveniently. The effectiveness of the system to process ability and the uniformity of product quality is demonstrated using etching process tool data.