Abstract: Metrology prediction based on equipment sensor data, i.e. virtual meteorology, is generally regarded as a methodology to improve efficiency of metrology instrument usage. In addition, a robust model with solid physical understanding can be used as a benchmark to monitor the health-of-equipment and capture anomalies at an early stage. This ability helps preventing the catastrophic malfunction of process equipment, which malfunction can result in long downtimes and even massive yield loss. In this paper, we demonstrate an online deployment of a robust prediction model for silicon carbide (SiC), capping layer thickness in a dual damascene copper process in a 65 nm mass production line.
Keywords: FDC, Yield Modeling, Virtual Metrology