Application of Genetic Algorithm to Solving Better SRAF Rules

Sub-resolution assist feature (SRAF) is critical to the process window for lithography in advanced integrated circuit manufacturing. SRAF provides a large enhancement to the process window (PW) of isolated (ISO) pattern. Currently, the SRAF rule is determined through the design of experiments (DOE) method in the industry. This method can only optimize one or two parameters at a time, but the SRAF problem needs to solve at least three parameters. In this paper, the genetic algorithm (GA) with improved selection method, which is highly efficient in multi-parameter optimization, is combined with lithography model and Calibre OPCVerify tool to determine the SRAF rule. A much shorter rule setup cycle is achieved by using GA, with a better performance in process window enhancement.