EPIC: Efficient Prediction of IC Manufacturing Hotspots With A Unified Meta-Classification Formulation

In this paper we present EPIC, a new generic and unified formulation to seamlessly combine the advantages of various types of lithographic hotspot detection techniques. With such formulation, we develop an efficient CAD flow and optimize it with quadratic programming techniques under industry-strength data at an advanced lithography node. After integrating various machine learning and pattern matching detection methods, we evaluate EPIC with a number of industry benchmarks under advanced manufacturing conditions. EPIC demonstrates so far the best capability in selectively combining the desirable features of various hotspot detection methods (3.5-12% accuracy improvement) as well as significant suppression of the detection noise (upto above 80% falsealarm reduction). These characteristics make EPIC very suitable for conducting high performance physical verification and guiding efficient manufacturability-friendly physical design.