Etch rate prediction in plasma etching using feed forward Error-Back Propagation neural network model

Abstract

In this paper, a Virtual Metrology (VM) model is proposed to predict etch rate which is one of the most important etching profile in etch process. Error Back Propagation (EBP) neural network is used to make the VM for etch rate prediction. Etching process recipe data obtained through the Design of Experiments (DOE) are used to train the VM. The etch rate data are gained through the experiments, and the EBP neural VM model is trained to satisfy the allowable error between predicted etch rate and experimental etch rate. With this trained EBP neural network VM model, it can be possible to predict the etch rate without real experiments.