With the dramatically increase of design complexity and the advance of semiconductor technology nodes, huge difficulties appear during design for manufacturability with existing lithography solutions. Sub-resolution assist feature (SRAF) insertion and optical proximity correction (OPC) are both inevitable resolution enhancement techniques (RET) to maximize process window and ensure feature printability. Conventional model-based SRAF insertion and OPC methods are widely applied in industrial application but suffer from the extremely long runtime due to iterative optimization process. In this paper, we propose the first work developing a deep learning framework to simultaneously perform SRAF insertion and edge-based OPC. In addition, to make the optimized masks more reliable and convincing for industrial application, we employ a commercial lithography simulation tool to consider the quality of wafer image with various lithographic metrics. The effectiveness and efficiency of the proposed framework are demonstrated in experimental results, which also show the success of machine learning-based lithography optimization techniques for the current complex and large-scale circuit layouts.

With the rapid advance of semiconductor process technologies, layout features in integrated circuits (ICs) become highly prone to process variations. Lithography hotspots are a set of problematic layout patterns with poor printability even if they pass design rule checking (DRC). These hotspots need to be detected and fixed as early as possible in the design flow to improve manufacturability and yield. While most of existing studies focus on hotspot detection, hotspot pattern classification is rarely addressed but plays an important role in determining the efficiency of hotspot detection. In hotspot pattern classification, similar hotspots are classified into a cluster with tolerance constraints, and a representative hotspot is chosen for each cluster for future application. To minimize the problem size of subsequent hotspot detection, the number of clusters, and thus the number of representative hotspots, should be minimized. In this paper, a clip shifting method is adopted to further reduce the cluster number during hotspot classification. We propose a two-stage pattern matching algorithm flow and derive optimal solutions by solving a set cover problem. Experiment results show that our flow can reduce the cluster number by about 30% compared to reference results in 2016 CAD Contest at ICCAD.