Feature Extraction for Predictive Maintenance for Semiconductor Plasma Etching Equipment

Abstract

The semiconductor market continues to expand and the importance of improving productivity in semiconductor manufacturing is increasing. In addition, with the increasing difficulty of process shrink, the importance of the plasma etching process has increased especially in recent semiconductor manufacturing, and it is important to improve the productivity of plasma etching equipment. To improve the productivity of plasma etching equipment, predictive maintenance (PdM) using sensor data of the system condition has recently been proposed in contrast to conventional periodic maintenance. In this paper, a feature extraction method based on sensor waveform component separation and a predictive maintenance approach using this method are proposed for the characteristic sensor data of plasma etching equipment, where the effects of disturbances caused by the process and multiple degradation factors can overlap, in order to realize PdM of plasma etching equipment. Simulation verification based on a case study of actual collected experimental data showed the effectiveness of the proposed method in terms of prediction accuracy.