Following the Moore’s law, 157nm photolithography would be the logic next step after the 193nm dry photolithography. But it turned out that the gain from the 157nm photolithography was too less compared to enormous technical challenges to implement the technology. So it was not a surprise when the 157-nm lithography was abandoned as not worthwhile by Intel and other major chipmakers in early 2000’s. The decision was also based on the great success from a new 193 nm (ArF) immersion lithography, which was implemented by engineers from two great companies TSMC and ASML. New IC process innovation such as double/multi-patterning and computational photolithography helped to further seal the fate of 157nm photolithography.
The immersion ArF photolithography has experienced great success for the process nodes scaling from 65nm to the current 10nm. Due to the exposure media change from air to water (with a larger refractive index), depth of focus increases and the resolution is thus improved. However, a disadvantage from the new immersion process is that the new exposure media of water is in direct contact with both photoresist and masks. This has caused some serious issues, one of which is photoresist component diffusion into water. Leaching of PAG and other photoresist components not only contaminates the mask (very costly to fabs), but also degrades photoresist performance. To overcome this new challenge, top-coating layer has been applied to prevent resist components from leaching.
To some extent, investigation on top-coating materials becomes the main theme on material development for the new ArF immersion photolithography. The top-coating layer can be solvent-developable and alkali-developable. Alkali-developable top coatings are preferred because they are compatible with photoresist development process. With a decent top-coating layer material, most dry ArF photoresists could be used for the new immersion process. However, material scientists would never be satisfied with the two-layer solution involving top coating, as significantly increased process complexity. One layer process is still much preferred and has been actively sought. One innovation was embedded barrier layer (EBL). These EBL material can be phase-separated from bulk polymer resin during spin coat, forming a protect layer on top of ArF photoresist. So two layers (top coating + photoresist) are technically realized by one spin coating step. Surface energy match and process control are keys to form such uniform EBL. Meanwhile, designing new photophobic resist structures, which could greatly prevent PAG from leaching, has also been investigated. Many design ideas from the earlier 157nm photoresist development, such as fluorinated polymers, finally found their destiny in these new investigations on top-coatings and one-layer ArF immersion photoresists. The fluorinated polymers have high transparency and are less prone to cause swelling in addition to their terrific hydrophobicity. This is probably still one of the most studied areas for the ArF immersion photoresists and new results will remain to be trade secret-protected for a while.
As discussed earlier, the new 193nm immersion photoresist development became more complex as it is needed to address issues such as component leaching, PAG diffusion during PEB (size/charge effect on diffusion), new water/photoresist interface, etc. In addition, for the most recent and advanced process nodes including 14nm and 10nm, which are being implemented at TSMC, Intel, and Samsung, process innovation becomes more and more important. This is new for the photolithography industry. It would certainly take off some pressure from material scientists’ shoulder but new material innovation is still a big piece for the industry. One particular challenge, Line Edge Roughness (LER), became vital as the industry is marching toward printing smaller features ≤10nm. Uncontrolled PAG diffusion is believed to be one main reason that lead to large LER. PAG diffusion during exposure/PEB is an intrinsic property for all photoresists with chemically amplified mechanism so it can’t be completely avoided. However, there are ways that diffusion could to be limited to the minimum extent. High Tg resins and PAGs with proper size/charge could be explored further to limit PAG diffusion. Besides that, a quencher is usually needed to limit PAG diffusion for most ArF immersion photoresists. One thing that the folks at Dow Chemicals have been working on is to use quenchers of photo base generators, which lead to localized and synchronized quenchers with the acid generation and therefore greatly limits acid diffusion during bake. The other thing that people from Dow worked on was to incorporate PAG component into bulk resin for their EUV photoresists to limit PAG diffusion. If succeed, the same principle may be applied to the ArF immersion technology.
Other patterning mechanisms such as negative-tone had also been actively explored for the ArF photoresists and commercial products are available on current market. Non-chemically amplified resists had been reconsidered to circumvent the PAG diffusion issue. However, it remains to be challenging as these new photoresists are difficult to meet the sensitivity requirement of ArF lithography.
To pursue the next generations of nodes below 10nm, double/multi-patterning has been demonstrated to be capable with the ArF immersion method. However, process complexity explodes for these new nodes from the multi-patterning. For example, the upcoming 7nm node would require over 30 lithography steps and around 60 critical alignment overlay steps assuming only ArF immersion process is used. This is unacceptable for fabs unless it is the last resort. New generation of EUV photolithography technology is in much greater demand than ever and the technology is already on the horizon for the next chapter of semiconductor industry.