Photoresist Line Edge Roughness Research Articles

Reaction-diffusion power spectral density

Characterization of a stochastic process in lithography, giving rise to photoresist line-edge roughness (LER), requires elucidation of the power spectral density (PSD) for that process. Thus, any analytical model for LER requires an analytical model for the PSD. Using a previously derived formulation for the reaction-diffusion autocorrelation function, the PSD can be derived from its Fourier transform. The resulting analytical expression for the reaction-diffusion PSD provides an interesting and useful form that will aid modeling work in LER prediction. Numerically calculating the PSD for the stochastic development rate shows that this same analytical expression approximately matches the simulation but with a correlation length that decreases as the amount of development noise increases.

Line-edge-roughness transfer during plasma etching: modeling approaches and comparison with experimental results

Both modeling and experimental results for the effects of plasma etching on photoresist line edge roughness and linewidth roughness (LER/LWR) and their transfer to underlayer films are presented and compared. In particular, we investigate the roughness formation on both photoresist and underlayer sidewalls during (1) isotropic trimming of photoresist, and (2) anisotropic plasma etching and LER transfer to substrate. The trimming process is modeled with an isotropic movement of the resist sidewall. In the anisotropic plasma etching process, the resist sidewall is used as a mask to anisotropically transfer the pattern to the underlying film. Experiments include trimming of a photoresist patterned with 193-nm photolithography in O2 plasma with no bias voltage and anisotropic etching of BARC and Si underlayers in CF4 and HBr/Cl2/O2 with bias. Both model and experimental results show that resist trimming causes reduction of resist LER and increase of the correlation length and roughness exponent with trimming time. This means that surface nano-protrusions versus trimming time become shorter and wider. In the case of anisotropic etching, the model predicts noticeable reduction of LWR, whereas correlation length and roughness exponent remain almost unaffected. The first experimental results seem to confirm these predictions.

Improving lithography pattern fidelity and line-edge roughness by reducing laser speckle

In this article the authors discuss the impact of coherence, or laser speckle, of current generation 193nm argon fluoride (ArF) excimer sources on lithographic patterning. They report a new metrology capability to characterize single-pulse speckle patterns at the exit of the laser aperture and quantify the speckle dependence on time integral square (TIS) pulse duration for different system configurations. The authors performed lithographic exposures on high-numerical-aperture immersion and dry ArF systems as a function of source pulse duration and have quantified the speckle impacts on measured photoresist line-edge roughness (LER) and linewidth roughness (LWR) using immersion and dry lithography processes. Measurements were obtained for multiple feature sizes, pitches, and illumination modes using both static and scanning exposures. They have compared the measured LWR due to laser speckle to results of a line-roughness image model, which accounts for the LER, LWR, and critical dimension uniformity due to the effective dose variation from laser speckle. Finally, the authors present measurements that demonstrate the proportionality between laser speckle contrast and the inverse root of the TIS pulse duration and find that the lithographic LWR exhibits a similar relationship with pulse duration.

Simulation of material and processing effects on photoresist line-edge roughness

Sub-100 nm device fabrication rules require extremely tight control of Line-Edge Roughness (LER) of patterned structures. During lithographic processes the resist film introduces an initial LER due to its chemical structure and processing. This initial LER evolves during the subsequent etching step. A stochastic simulator is presented which takes into account material and process effects on photoresist LER. Its application in model cases reveals that LER decreases with lower degree of polymerisation, but is sensitive on the acid diffusion process during post exposure bake in Chemically Amplified Resists (CARs). Further simulations confirm that LER can be reduced during etch patterning, at the expense of critical dimension control.

Material and process effects on line-edge-roughness of photoresists probed with a fast stochastic lithography simulator

Stochastic simulation of photoresist line-edge roughness (LER) is attempted using a three dimensional (3D) lithography simulator incorporating a fast dissolution algorithm based on a modified critical ionization model. The fast 3D simulation permits detailed evaluation of the material and process effects on LER. In this article the effects of deprotection fraction, critical ionization fraction, photoacid generator concentration, acid diffusion range, and polymerization length on LER are investigated through simulation. It is found that the relation of LER to photoresist polymerization length is greatly affected by the photoacid concentration and diffusion range.