Run-to-run fault detection based on ARX model and PCA for semiconductor manufacturing processes

This paper proposes a run-to-run(RtR) fault detection approach for general semiconductor manufacturing processes. In this paper, a data-based model, auto-regressive with exogenous inputs (ARX) model, will be introduced as an alternative of a mechanical model for a semiconductor manufacturing process. In this model a recursive least-squares (RLS) algorithm is proposed to identify the on-line parameter. Once the process abnormalities occurred, the fault will be detected with a statistical principal component analysis (PCA) method applied to ARX parameters. And the results will be illustrated by several simulations.