“…When the line/feature size is decreased to several tens nanometers, the bias due to the edge effect would become more obvious. A lot of works including both experimental study (Bunday et al, 2007;Choi et al, 2006;Jones et al, 2003;Kawada et al, 2003;Maeda et al, 2008;Matsumoto et al, 2006;Morokuma et al, 2004;Novikov et al, 2007;Rice et al, 2006;Shishido et al, 2002;Tanaka et al, 2003;Tanaka et al, 2004;Tanaka et al, 2005;Tanaka et al, 2007;Tanaka et al, 2008a;Tanaka et al, 2008b;Wang et al, 2007;Yamane & Hirano, 2005) and theoretical investigation (Abe et al, 2007;Babin et al, 2008a;Babin et al, 2008b;Bunday & Allgair, 2006;Dersch et al, 2005;Frase et al, 2007a;Frase et al, 2007b;Gorelikov et al, 2005;Villarrubia et al, 2004;Villarrubia et al, 2005a;Villarrubia et al, 2005b) with MC simulation methods have been done, aiming at accurate estimation of the CD values with CD-SEM. However, when the dimension decreases to tens nanometers most of the experiential methods, such as, the maximum derivative method, the regression to baseline method and the sigmoidal fit method, face different difficulties (ITRS, 2007;Villarrubia et al, 2005a).…”