Experimental and simulation comparison of electron-beam proximity correction

Abstract

In electron-beam lithography, electron scattering in the resist, and underlying substrate limits the control over feature sizes [M. Parikh, J. Vac. Sci. Technol. 19, 1275 (1981)]. To correct for this effect, one applies the so-called “proximity correction” that adjusts the beam dose of each individual feature. The algorithms that are commonly used to calculate this correction model the electron scattering as a double-Gaussian function. However, a point-spread function (PSF) describes the scattering process more accurately. Therefore, we have investigated whether the use of the PSF in the proximity correction algorithm leads to an increase in feature size control; as compared to the double-Gaussian PSF. In order to compare the two PSFs directly, we have performed a proximity correction on a pattern of contact holes, using each of the approximations. Our measurements show that the algorithm using the PSF performs a substantially better proximity correction, leading to a more accurate control over the contact hole sizes. Moreover, we have developed a method to assess the proximity correction’s success using the SELID [M. Bohn et al., Proc. SPIE 5256, 695 (2003)] simulation program. This strongly reduces the need for costly and time-consuming experiments.