Experts Detail Immersion's Successes, Challenges

Although immersion lithography has made great strides even just in the past year, there are still several challenges to overcome. At Semiconductor International's latest webcast on immersion lithography, part of a two-day online conference (see www.emsummit.com), key industry players detailed their latest successes and also talked about the challenges that lie ahead. Panelists were Dan Corliss, program manager of immersion lithography development at IBM Microelectronics (Hopewell Junction, N.Y.); Andrew Grenville, an Intel assignee to Sematech (Austin, Texas), where he is program manager for immersion lithography strategy; and Will Conley, a member of Freescale Semiconductor's Advanced Optical Lithography Group (Austin, Texas), and an assignee to Sematech.

Corliss described IBM's immersion work with ASML and Albany NanoTech in a multi-year program called IMPLSE. The first phase of the program was to look at immersion lithography's capabilities with ASML's prototype tool that had previously been in use at ASML's facilities in the Netherlands. As IBM announced in December (see "Chipmakers Immerse Themselves in 193 Wet ," Semiconductor International, February 2005), the chipmaker used an immersion scanner to produce a via level of a commercial chip. They ran most of the production through the standard product line, then split the via level between a normal manufacturing line and the Albany NanoTech research facility. The net result was that the microprocessors yielded, he said. "Not only did they yield, they yielded comparable to the dry processes."

This SEM shows a microprocessor imaged with immersion lithography for one via level. (Source: IBM)

Corliss showed a SEM of a Power microprocessor that was imaged with immersion lithography (Figure ). Of the four via levels and metal levels shown, one was produced with immersion lithography. "And the challenge here is really to find which one it is," he said. "And I think most lithographers would conclude that there's no difference between any of these, which is the greatest statement that one could have."

Despite the latest successes, there are still several challenges to overcome for immersion lithography. Grenville and Conley discussed ongoing work being done through Sematech, taking a closer look at some of the issues still looming. Grenville and his group tested optical coatings on CaF₂ lenses. Although the material is likely to be used as the final optic element because of its durability, it degrades in water, so it needs to be coated for protection. With samples provided by the three major tool suppliers, researchers were able to test a variety of coating compositions and thicknesses, about a quarter of which showed subtle or no degradation over the course of a one-year equivalent lifetime. Some 50% showed nominal, non-catastrophic changes, such as surface roughness, thinning of the coating layers, darkening of the coating, and discoloration of the irradiated area.

Showing some examples of degradation, Grenville also showed a very high-performing coating — with <1% change in the coating over approximately a two-year equivalent lifetime. Ultimately, the industry is looking for a 10-year lifetime result for the optical coatings, but Grenville contends that the prognosis is good. "We feel that, at this point, up to about a quarter of the coatings we've tested so far from the tool suppliers have shown what really is close to acceptable performance," he said. "Looking at the 10-year requirement, to get up to sort of the 20 MJ/cm² types of requirements, the trends are very, very encouraging."

Grenville's team also studied the possible sources and effects of contaminants on the lens, and found what were also encouraging results. Even with concentrated levels of contaminants introduced through accelerated experiments, they found very low rates of contamination on the final optic element. "The overall conclusion here is that we're rather less likely — perhaps by a couple orders of magnitude — rather less likely to have issues with photocontamination of the final element," he said.

Conley's team did work with a 1.05 NA Exitech system at the Rochester Institute of Technology (Rochester, N.Y.), experimenting with various levels and types of photoacid generators (PAGs) and how the PAGs interacted with water and the coated wafer surface in immersion lithography. The researchers experimented with several PAGs, including TPS-Tf, TPS-Nf and TPS-Of (PFOS) types, demonstrating that PAG leaching increased with anion chain length. So PFOS (with an anion chain length of 8) showed PAG levels of ~80 ppb, compared with TPS-Tf (anion chain length of 1), which was closer to ~5 ppb.

One study also examined the effects of the amount of PAG in a resist. Although 1% PAG produced surface measurements of ~5 ng/mL that stayed consistent over time, 5% PAG rose sharply within the first minute of contact with water, then somewhat leveled off. Given this tendency, the researchers decided to try a pre-rinse of the resist. "When we looked at the amount of PAG coming off vs. time, we saw that it comes off rather quickly, and then has a tendency to not really level off, but gradually increase after that," Conley noted. "So if we can rinse the wafer with process water before, perhaps we can remove an excess amount of PAG or other contaminants at the surface."

Although the experiments showed little or no variation in the process window for rinses of 0, 1, 10 or 15 seconds, increasing the rinse to 30 and 60 seconds caused a loss of exposure latitude and a significant loss in depth of focus. They also looked at contrast through the various rinse times. Again, there was little change through 15 seconds, but the contrast went down by as much as 30% for the 30 second rinse, and by almost 50% for the 60 second rinse.