Masks for flash memory gates for the 45nm node: binary or attenuated?
暂无分享,去创建一个
The introduction of the XT:1700Fi has brought the 45nm node well within reach for memory makers. Simulations have shown that at these small feature sizes attenuated phase shift masks may not be optimal anymore: binary intensity masks show a reduced MEEF for dense lines ans spaces using dipole illumination in combination with TE-polarization. The increased angle of incidence of the light falling on the mask causes all kinds of 3D effects that could be largely neglected up till now. Both binary and attenuated phase shift masks start to behave differently than can be expected based on scalar diffraction theory (thin-mask approach). These differences lead to some interesting observations, which make it imperative to include mask absorber material explicitly when optimizing imaging performance for the 45nm node. Flash Memory Gate structures contain both 1:1 dense lines and large semi-isolated lines. In this paper, experimental results will be shown comparing binary intensity masks with 6% attenuated phase shift masks for 1-D L/S gratings and NAND-type flash memory structures. The various imaging relevant parameters (process window, MEEF, EL, DOF and pattern placement) will be addressed and compared to simulations
[1] Andreas Erdmann,et al. Mask modeling in the low k1 and ultrahigh NA regime: phase and polarization effects (Invited Paper) , 2005, Other Conferences.
[2] Alfred Kwok-Kit Wong,et al. Resolution enhancement techniques in optical lithography , 2001 .
[3] Jo Finders,et al. Mask error factor: causes and implications for process latitude , 1999, Advanced Lithography.
[4] Mark van de Kerkhof,et al. Pushing the boundary: low-k1 extension by polarized illumination , 2007, SPIE Advanced Lithography.