A study on process control to improve yield in semiconductor manufacturing

Abstract

We present the process analysis system that can analyze causes, like an expert, after processes. Also, the plasma etching process that affects yield is controlled using an artificial neural network to predict output before the process. In modeling, a method that uses history for input data is considered, it offers advantages in both learning and prediction capability. This research regards the critical dimension that is considerable in highly integrated circuits as the output variable of the model. Based on a model using this method, we propose an algorithm to analyze and control the effect of input variables for predicted defects. Both the weight of input variables and their historical trend are examined for this algorithm.