Abstract
As the global competition in semiconductor industry becomes fiercer, yield enhancement, the basis of quality, is the key factor to maintain competitive advantages for semiconductor companies. With the shrinking feature size in advanced semiconductor technology generations, the process window is tightened to meet high quality requirement, and thus increase the difficulty of yield enhancement. In addition to the diagnosis of low yield for trouble shooting, the online process control plays a critical role to enhance the yield in advanced technologies. Advanced equipment control/advanced process control (AEC/APC) including run-to-run (R2R) control and fault detection and classification (FDC) is the core methodology in semiconductor process control. In particular, R2R control functions aims to reduce the variability of process output through manipulating the process/equipment parameters. Indeed, developing a R2R control model not only considers process modeling and controller design, but also involve the preference of the decision-maker and the uncertainty in production environments. The decision-maker determines the control hierarchy and the scope of influence factors, and thus has inter-relationship with process modeling and controller design. The uncertain events and production limitations in fab affect the performance of R2R control model, so it is needed to structure the relationship, sense the influence outcomes, and overall evaluation to select the best fitted model in real problems. Focusing on the research gap in literatures and realistic needs, this study proposes a quality engineering framework for advanced process control in semiconductor manufacturing. The decision context is embedded in the proposed framework to support implementation in semiconductor industries. Two empirical studies were conducted to explain the framework process and validate the derived advanced process control model including dynamic adjusted proportional-integral (DAPI) feedback controller for lithography processes and tool similarity feed-forward controller for etch processes.