Fast curvilinear thick-mask model for deep ultraviolet lithography incorporating boundary fluctuations.

To address the thick mask effect involving significant edge diffraction perturbations in curvilinear masks, this paper proposes a fast thick mask model that adopts the slope chain code with direction (SCCD) to characterize the influence of curvilinear edge diffraction. For a given curvilinear mask, it is decomposed into several distinct feature regions according to the different characteristics of the curvilinear boundary. When the local shift-invariant assumption of the mask model is violated, the mask transfer matrix (MTM) is introduced to compute the diffraction near field (DNF) of the curvilinear mask for each feature region, with its values obtained by fitting from a pre-established training set. Simulation results show that, compared with the rigorous EMF method, the proposed model can accurately reproduce the pronounced fluctuations in the transparent regions of the DNF, while the computational efficiency is greatly improved. The proposed model also exhibits satisfactory applicability to the various fluctuations of the DNF at different observation planes.