The incorporation of the pattern matching approach into a post-OPC repair flow

Abstract

The model based optical proximity correction (OPC) systematically computes the mask compensation that will be applied to the main features of circuits with sub-wavelength sizes. Even a sophisticated OPC recipe could render thousands of weak points, below the specs. An automatic repair flow may correct most of these post-OPC weak points. The remaining errors will have to demand engineers' visual inspections and subsequent manual fixings; and it might cost a considerable amount of human efforts and hence compromise the turnaround time (TAT). After performing several tape-outs, we have also noticed some weak points that need to be fixed afterward share certain commonalities. This inspires us to incorporate the pattern matching (PM) approach into our post-OPC repair flow. For the previous tape-outs, the remaining weak points will be fixed manually or be fixed by a special OPC recipe. Thus our old knowledge can directly provide proper OPC solutions for these known weak points. For a new tape-out, the design patterns associated with these weak points scan the post-OPC layer and find the match. Then, the proper OPC solutions will be pasted to these matched locations to complete repair process, allowing us to avoid repeatedly performing the manual fixings for the same types of weak points. This approach will also help identify certain OPC weak points that are proven to be fine by the wafer data. This type of weak points can be automatically waived by the OPCV verification. The incorporation of the PM approach into our repair flow can significantly reduce the TAT for a new tape-out.