Photospeed considerations for extreme ultraviolet lithography resists

Abstract

Photospeed is a prime consideration for wafer throughput of extreme ultraviolet ͑EUV͒ lithography. Faster photoresists additionally provide system advantages such as less thermal management of the mirrors and mask, and potentially increased component lifetimes. However, there are some predicted detrimental considerations when using fast photoresists such as shot noise. In this article, we report details of the formulation of photoresists exposed at 248 nm and identical formulations exposed at 13.4 nm. Compositions typically contained co-or terpolymers of poly-4-hydroxystyrene, t-butyl acrylate and as an option, styrene, a photoacid generator of bis-t-butylphenyl iodonium camphorsulfonate or perfluoroxbenzensulfonate and tetrabutyalummonium or triphenylsulphonium hydroxide base. With these formulations, the EUV photospeed was varied from 34 to 2.7 mJ/cm 2 . Scanning electron microscope analysis was done for all wafers at Sandia using GORA software to determine the line-edge roughness ͑LER͒. Identical formulations were exposed at photon dense deep UV ͑DUV͒ wavelengths and comparatively photon-sparse EUV wavelengths. Therefore, ratioing the LER of identical formulations exposed at DUV minimizes the confounding effects of formulation changes that affect dissolution, resolution, etc. A plot of the line-edge roughness ratio of LER EUV /LER DUV as a function of the photospeed was used to gain insight into the effect of shot noise on the LER of dense lines. If shot noise effects were dominating the LER for EUV, then the ratio would increase above the nominal value as the dose to size was decreased. It is clear that the current generation of DUV resists, with the invention of no new material required, can be reformulated to satisfy the photospeed specification of 5 mJ/cm 2 for EUV lithography and no shot noise effects have been observed.

Understanding Extreme Stochastic Events in EUV Resists

“…If we simply assume the dose noise to go as the square root of the mean dose, a 25% reduction in noise would require a 56% increase in dose. This arguably represents the "shot noise limit" [8][9][10][11][12][13][14][15][16][17][18][19]], yet we find that by biasing the CD by 2 nm we significantly outperform this "limit". Figure 11 shows the corresponding PDFs for the original case in Fig.…”

High performance resist for EUV lithography

“…However, the advent of several micro exposure tool over the next year has ignited an interest in developing an EUV resist meeting strict tool and performance requirements. For example, the developed resist must possess reasonable photospeed while maintaining a low level of outgassing components [2]. Another factor that has to be considered is the absorption coefficient of resist materials in the EUV.…”

Resist Materials and Processes for Extreme Ultraviolet Lithography

“…Experimental studies on the exposure dose dependence of LER demonstrated that LER decreases with increasing exposure dose. [44][45][46][47] However, it cannot be reduced below a certain value (3-4 nm) when exposure dose is increased. 50,51) A common explanation of this lower limit of LER upon high-dose exposure is that it is of material origin, namely, it is associated with the molecular size and/ or high-order interaction of polymers.…”

“…[44][45][46][47][48][49] Equations (2) and (3) indicate that LER is inversely proportional to the square root of exposure dose. Experimental studies on the exposure dose dependence of LER demonstrated that LER decreases with increasing exposure dose.…”