Cymer Focuses Work for High-Volume EUV Source

Although the battle is far from over, at least one source developer for extreme ultraviolet (EUV) lithography has been nailing down some critical decisions for continued high-volume production development. Cymer Inc. (San Diego) announced recently that it has settled on a laser-produced plasma (LPP) source as its approach for commercial EUV lithography, citing the increasing challenges of discharge-produced plasma (DPP) for production tools. Cymer made the announcement last month at the EUVL Symposium in Miyazaki, Japan.

Despite some formidable challenges, EUV lithography is considered the most likely candidate for production at the 32 nm node and beyond. One of the key struggles has been the ability to increase EUV source output power high enough to meet high-volume production demands, while also remaining cost-effective. Several groups have been working on a variety of source platforms, primarily centered around DPP and LPP mechanisms (see "EUV Sources Continue to Fight the Fight ," Semiconductor International, June 2004, p. 34). LPP, Cymer has concluded, is the source solution capable of achieving the power scaling and cost effectiveness required. The high-volume output requirement that has been traditionally quoted is ~115 W at intermediate focus (IF). Cymer's LPP source has so far achieved an output of 4.52 W IF, but is expected to reach 15.1 W next year, primarily by increasing the drive laser power from 800 W to 2.28 kW. Cymer expects output power to reach 102.23 W by 2007.

Cymer’s proposed EUV source for high-volume manufacturing is a laser-produced plasma source that uses excimer as the drive laser technology and lithium as the target material.

Even after all of that, history indicates that those power requirements are only expected to rise over time, noted Dave Myers, vice president of EUV development at Cymer. DUV sources, for example, began high-volume manufacturing in 1996 at ~10 W, and have reached 60 W today. "You start with lower powers, and lithographers find ways to need higher powers," he said. "For Moore's Law to happen, you have to have higher throughputs.

All of this makes DPP sources less viable for volume production, primarily because LPP sources are more scalable. Other disadvantages of a DPP source include its relatively large size and the challenge of electrode erosion. Electrode lifetime has reached 10 million to 35 million shots today, but requirements call for 10 billion shots. "You can't have a solution where you're changing the electrodes often," Myers said. "Tool downtime would increase, which is a big problem in a fab."

Consumables issues have also had an influence on Cymer's decision to use lithium as the source's target material — not xenon, which has been used extensively in research; and not tin, which most source developers have been exploring for its conversion efficiency. While xenon's CE is <1%, lithium has a CE of >2.5%, which is as good or better than that of tin. The primary attraction to lithium as the target material, however, is its positive influence on collector lifetime. Myers estimates current collector lifetimes at billions of shots. "At 10 kHz, you get a billion shots in a few days of operation," he said. "If you're changing the optical system every few days or every month or even every few months, that's not acceptable."

Collector lifetime is influenced by the source material and contaminants that deposit and build up on the collector optics. Tin and xenon ions sputter more readily onto the collector because of their significantly higher energies. In addition, Cymer is using a capping layer and heating system that can evaporate any lithium sputtering from the collector before significant diffusion occurs. The concept would not work well for tin, Myers noted, because it has a much higher melting evaporation point.

Cymer has also announced its decision about the drive laser technology, which will be an excimer laser. This may come as no surprise, given that Cymer dominates the excimer laser field, but Myers said that it had to do with more than just the fact that the excimer is something the company understands. The drive laser is not a strong variable with lithium, he said. The main problem with diode-pumped solid-state (DPSS) or CO₂ laser is simply a cost issue.

For additional information on lithography, go to www.semiconductor.net/lithography