A static-random access memory (SRAM) for a generic 22-nm node with a cell size of on a contact layer (see in 8) was used for the evaluation. Since, a minimum pitch of the SRAM was 81.2 nm, exposure pattern was divided into two parts for double patterning to relax the pitch to 110 nm as shown in Fig. 2. An NEGA-process,9 which used a negative-tone resist of 100-nm thick on bottom-ARC with 92-nm thick expanding PW in this kind of contact pattern with tighter pitch, was applied. A first iteration of SMO was carried out with a thin-mask model (Kirchhoff), an aerial image base, properties of Intelligent Illuminator and scanner fingerprint of Jones pupil of NSR-621D as Nikon scanner signature file,10 and the target pattern. During this aerial image based SMO run, an edge placement error (EPE) based optimization metric was determined by the intensity contours at fixed -position inside the resist bulk, here 16.5 nm above the resist bottom. Figures 3(a) and 3(b) show SMO result of contour on drawn and corrected patterns with SRAF, MEEF. An MEEF value less than a target value of 4 was obtained. PW-tree, which consisted of an acceptable area of DOF and exposure latitude (EL), was calculated from a common PW as shown in Fig. 3(c). To ensure the effectiveness of the co-optimization, a PW-tree with a quadrupole source and an optimum optical proximity correction (OPC) as the same algorithm of SRAF insertion on SMO was compared as shown in Fig. 4. Over 100 nm of DOF at 6% EL, whose values were PW target value, was obtained and an area of common PW was improved up to 48%.