Wavelet-based virtual metrology technique

In this paper, a systematic methodology for virtual metrology is proposed. This virtual metrology system which is mainly designed for the process subject to process drift consists of a fuzzy neural network for calculating the process outcome and wavelet transform for estimating the trend of process drift. Because many semiconductor processes exhibit inevitable steady drifts in nature, virtual metrology is a novel technology to predict the process results based on the previous metrology measurements, instead of measuring practically. The system was implemented to the sputtering deposition process in TFT-LCD fabrication for experimental verification. The results show that it has good generalization capability and performance. Thus, it provides an effective and economical solution for metrology prediction.