Abstract
A method has been developed for correcting line width variations due to midrange flare with a scattering range of over a few tens of micrometers (which we call local flare). It is shown that the conventional single Gaussian point spread function (PSF) is not sufficient and that a double Gaussian point spread function is needed to explain the line width variation caused by local flare. The remaining errors after correction are discussed under the assumptions that the mask correction is linear with respect to local flare intensity and is independent of pattern layout considering the order of the local flare correction (LFC) and optical proximity correction (OPC). This simple sizing method can reduce the critical dimension (CD) variation regardless of whether LFC is done before or after OPC. The LFC performance was evaluated using actual 90-nm-node LSI data. A much faster correction time than that of OPC was achieved by introducing the area density map method. The CD variation due to local flare was reduced from 22 to 5 nm.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have