Process proximity correction by using neural networks

Making an accurate and quick critical dimension (CD) prediction is required for higher integrated device. Because simulation tools are consisted of many process parameters and models, it is hard that process parameters are calibrated to match with the CD results for various patterns. This paper presents a method of improving accuracy of predicting CD results by applying /spl Delta/ (the difference between simulation and experimental data) value to neural network algorithm (NNA) to reduce CD the difference caused by optical proximity effect.