Multi-phase 반도체 제조공정에서 functional 입력 데이터를 위한 모니터링 시스템

Abstract

In general, monitoring output variables has difficulty in detecting the changepoint and take action. Even earlier detection of the process variation causes cannot completely prevent yield loss in reality because a number of wafers in processing after detecting them must be re-processed or cast away. Semiconductor process engineers have their attention to monitoring process to monitoring process by early detecting change using input parameters. To solve this process control problem, in the thesis, we propose the method to monitor input variables efficiently in multi-phase semiconductor manufacturing process. Measured input parameters in the multi-phase process tend to have functional structure. Data pre-processing is needed to monitor process using these functional input data. Next, change-point is estimated for pre-processed data set. If process variation occurs, key variables affecting process variation will be selected using contribution plot. To evaluate the propriety of proposed monitoring method, we used real data in semiconductor manufacturing process. The experiment shows that the proposed method has better performance in prediction and process monitoring than previous output monitoring method.