The representative figure method for the proximity effect correction [III]

Abstract

The representative figure method for the proximity effect correction was proposed in the previous papers in order to eliminate the correction time dependence on the large-scale integrated (LSI) pattern density. In this report, a new calculation method for the proximity effect correction is proposed. The method uses both the representative figure method and the dose formula method. When the formula of Pavkovich is adopted as a dose formula, the total correction error for the proposed method is at most 4%. The intrinsic error of the representative figure method itself is at most 0.5%. The proposed method reduces the total calculation time to (1/28) compared with the conventional method, when the minimum feature size is 0.15 μm and the acceleration voltage is 50 kV. Here, the total calculation time includes (1) preparation time for representative rectangles and (2) shorter calculation time for the correction. Furthermore, the total time is found to become less than the data format conversion time, when the minimum feature size is <0.5 μm. The calculation time for the correction is found to be <1 h for any pattern density, when the newly proposed method and the 50 MIPS engineering workstation (EWS) are used. The problem of the proximity effect correction is reduced to an easier problem; i.e., how to reduce the calculation time for obtaining the areas and the center of gravity of the LSI patterns.