摘要
This paper reports a proposal for an advanced and efficient method to evaluate the pattern transfer completeness (PTC) in terms of line edge roughness (LER) by quantifying the deviations of printed patterns statistically in regards to their original designed patterns. Three substantial errors in the existing method are corrected by the proposed method with evidence from iterative examinations. With the use of identical images of complex patterns expressible in parametric forms such as Archimedean, logarithmic, and hyperbolic spirals, error corrections and efficiency improvements compared to the existing method are proven. Comprehensive studies for image operation, reference point definition, deviation acquisition, contour point creation, and LER calculation were performed. In addition, this work involves analyses of the errors in the existing method, the efficiency improvement of the proposed method, the impact of variations on point density, and the validity of the LER calculations. The results show that the proposed method not only correctly evaluates the PTC of printed patterns with on average 97.6% efficiency enhancement, with at most 37.7% correctness improvement, but also displayed operation flexibility with the controllable point density in comparison to the existing method.