Industry Takes a Hard Look at Immersion Lithography

As the keynote speaker at last year's Lithography Breakfast Forum at SEMICON West, Tony Yen pointed to immersion lithography as a way to extend optical lithography's resolution capabilities while maintaining a k₁ factor that lithographers are used to dealing with. "Implementation will not be easy," said Yen, a TSMC assignee to International SEMATECH (ISMT, Austin, Texas), and co-director of ISMT's lithography division. "We would like the suppliers to think about this problem." That was in July. Suppliers and other developers have been thinking about it, and are building a critical mass. But there are still several issues to consider.

More than 100 people gathered for ISMT's Immersion Lithography Workshop in San Francisco in December, getting down to the nitty-gritty of how to make the alternative technique work in the face of continuing battles with next-generation lithography (NGL) techniques. There are a few different immersion techniques being considered. Although none of them is really practical yet, neither is extreme ultraviolet (EUV) lithography, which is considered to be the main NGL solution. The question at hand is whether making immersion lithography work in a production environment is more practical than trying to get EUV or another NGL technique to work.

The practical limits of optical lithography assume that the medium through which imaging is occurring is air. By introducing a liquid between the last lens element and the wafer being imaged, the refractive index changes, thereby enabling enhanced resolution with a lower effective wavelength of the lightsource. Immersion lithography would effectively lower a 157 nm lightsource to a 115 nm wavelength, enabling the printing of critical layers at the 45 nm node with the same k₁ factor that the industry is accustomed to today (~0.43), Yen said.

Lithographers are considering that immersion lithography could actually push 193 nm lithography down into the 45 nm node, with 157 nm lithography going as far as 32 nm. Immersion techniques effectively shorten the 193 nm wavelength to ~132 nm. "And yet, in theory, you're still dealing with 193 nm tools," noted Phil Ware, senior fellow, lithography at Canon U.S.A. Inc. (Irving, Texas). All the headaches of 157 nm lithography — large amounts of CaF₂, hard pellicles, a nitrogen purge, etc. — are not necessary, he noted.

Of course, immersion lithography is by no means a slam dunk. "Probably resists will have to be reformulated, there's an issue of microbubbles, and there's the terrible messiness of it all," said Ware, who was an attendee at the recent workshop. The question, though, is whether making immersion lithography work is a bigger problem than making 157 nm lithography work, or bringing EUV lithography into the fab, he said, adding that there were certainly several optimists at the workshop. The goal of the workshop was essentially to answer this question within a six-month timeframe, deciding whether one or more of the possible immersion techniques is likely to offer a cost-effective extension of optical lithography.

After hearing presentations from key researchers, workshop attendees divided into four brainstorming sessions to focus on key areas of concern: resist, fluid-tool, fluid-optical, and high numerical aperture (NA). Critical issues from each of these areas was compiled, and the assembly decided on the top 10 critical issues that immersion lithography faces overall (Table).

"The results from this fact-finding workshop have already evolved, literally overnight, into several action-oriented working groups that will tackle some of the more difficult questions surrounding immersion technology," said Walt Trybula, ISMT senior fellow and workshop chairman, in a statement immediately following the workshop. "If 193 nm immersion lithography can be implemented in time and with competitive cost-of-ownership figures, it could be a very attractive technology to semiconductor manufacturers."

There are three different projections on what immersion lithography will look like, according to Trybula. For 193 nm imaging, the most effective liquid is water; for 157 nm lithography, fundamental research is still required on which liquid would be best. Researchers at MIT Lincoln Laboratory (Lincoln, Mass.) have put considerable effort into immersion studies, but note that further experiments are required in several areas. With the involvement of more players in the industry, immersion lithography certainly has a chance at mainstream acceptance. ISMT plans to hold another immersion technology workshop in June.