Endpoint Detection for Single Layer Plasma Etch Using Hybrid Fuzzy Inference Systems

Abstract

The need for the accurate endpoint detection (EPD) technique in plasma etching process is increasing in the semiconductor industry for yield improvement and higher device performance. To monitor and analyze plasma process, the one of the most popular sensor is optical emission spectroscopy (OES). The OES can collect optical data emitted from plasma in the processing chamber and these data can be used in diverse methods for monitoring and analyzing plasma status. In this paper, a fuzzy inference system is applied for the endpoint detection using OES data. The fuzzy inference system is an interpretation method of uncertain data set. The normal etching stage and over-etching stage is classified by the fuzzy inference system using OES data. The proposed method successfully detected endpoint of silicon oxide film etching. The performance comparison between fuzzy inference system with linear regression method and with polynomial regression method is also investigated.