Novel fine-tuned model-based SRAF generation method using coherence map

Abstract

We have developed the comprehensive sub-resolution assist features (SRAFs) generation method based upon the modulation of the coherence map. The method has broken through the trade-off relation between processing time and accuracy of the SRAF generation. We have applied this method to a real device layout and the average of Process Variation band width (PV band width) has improved to 40% without any processing time loss. １．INTRODUCTIONAs Integrated Circuit(IC) pattern shrinks, the mask layout used in optical lithography differs from the corresponding IC design layout owing to the optical proximity effect correction (OPC). In order to satisfy the robustness to the process variations such as exposure dose variation and focus variation, recent mask layout includes not only the modified IC design patterns (OPCed patterns) but also the sub-resolution assist features (SRAFs). Establishment of the technology for the "holistic" optimization of this SRAF layout has been a hot issue for lithography engineers for a long time. However, as is well known, such an optimization is generally a difficult problem because the goal should be a "holistic" one. This means that it should be verified that the optimized mask layouts are robust against not only a single process variation but against a number of them in order to gain the maximum process yield. In addition, Issues specific to mask layout pattern and extremely low k1 lithography conditions should be considered in order to fix all hot-spot and warm-spot patterns.Recently, several novel techniques have been developed for the generation of the model-based SRAF at the final stage of the lithography. The local search heuristics to generate the model-based SRAF such as genetic algorithm (GA) [1], the optimal gradient method[2], inverse lithography technology (ILT)[3][4] can greatly improve the generation of the model-based SRAF, but these approaches have not been widely applied due to the intractable computer run-times. The coherence map technology that includes 2D-TCC method [5][6], interference map technology (IMT) [7], and SRAF guidance map technology(SGM) [8] is a novel promising method of optimizing SRAFs with drastically shorter turnaround time (TAT). They described a mask optimization algorithm using a "coherence map" that shows the optimum degree to be placed in a mask layout as to robustness against a specific process variation such as image contrast, or exposure latitude (EL). This intelligent method can actually increase the process robustness to some extent, but there are still some challenging problems to be solved in order to increase the robustness to the ultimate level for huge numbers of IC random layouts with an easy conscience. For example, our extensive studies of various mask designs have clarified that the acquired mask layouts optimized in order to increase a single process variation such as EL "frequently" result in poor robustness against the unconsidered process variations such as focus variations. For instance, in some contact hole patterns, the positions...