Optical modeling and resist metrology for deep-UV photolithography

Abstract

This thesis first presents a novel and highly accurate methodology for investigating the kinetics of photoacid diffusion and catalyzed-deprotection of positive-tone chemically amplified resists during post exposure bake (PEB) by in-situ monitoring the change of resist and capacitance (RC) of resist film during PEB. Deprotection converts the protecting group to volatile group, which changes the dielectric constant of resist. So the deprotection rate can be extracted from the change of capacitance. The photoacid diffusivity is extracted from the resistance change because diffusivity determines the rate of change of the acid distribution. Furthermore, by comparing the R and C curves, the dependence of acid diffusivity on reaction state can be extracted. The kinetics of non-Fickean acid transportation, deprotection, free volume generation and absorption/escaping, and resist shrinkage is analyzed and a comprehensive model is proposed that includes these chemical/physical mechanisms. Then in this thesis a novel lithographic technique, liquid immersion contact lithography (LICL) is proposed and the simulations are performed to illustrate its main features and advantages. Significant depth-of-field (DOF) enhancement can be achieved for large pitch gratings with deep-UV light (λ=248nm) illumination with both TM and TE polarizations by liquid immersion. Better than 100nm DOF can be achieved by when printing 70nm apertures. The simulation results show that it is very promising to apply this technique in scanning near field optical microscopy. Finally, a rigorous, full vector imaging model of non-ideal mask is developed and the simulation of the imaging of such a mask with 2D roughness is performed. Line edge roughness (LER) has been a major issue limiting the performance of sub-100nm photolithography. A lot of factors contribute to LER, including mask roughness, lens imperfection, resist chemistry, process variation, etc. To evaluate the effect of mask roughness on LER, a rigorous full vector model has been developed by the author. We calculate the electromagnetic (EM) field immediately after a rough mask by using TEMPEST and simulate the projected wafer image with SPLAT. The EM field and wafer image deviate from those from an ideal mask. LER is finally calculated based on the projected image.