Semiconductor industry ready to take 193-nanometer plunge


By Hank Hogan

AUSTIN, Texas—Less than a month into 2004, semiconductor industry representatives have decided that the end of the decade should be all wet.

A group of technical movers and shakers from hundreds of companies attended a two-day International Sematech-sponsored meeting in Los Angeles at the end of January. The participants settled on 193-nanometer (nm) immersion lithography as the most likely candidate for semiconductor manufacturing in 2007 and 2009.

Immersion lithography: "ripples still spreading out."
Ralph Dammel, Clariant Corp.
Click here to enlarge image

The plunge into 193-nm technology has implications for contamination control because the liquid will be in contact with the wafer. What's more, the liquid's optical properties are critically important and can be altered by contaminants. Other aspects of manufacturing will also be impacted in ways that are just now being discovered.

As Ralph Dammel, director of technology for 193 nm and 157 nm at photoresist maker Clariant Corp. (Charlotte, N.J.), quipped at the Sematech meeting, "The ripples are still spreading out."

But the vote wasn't unanimous. In particular, Intel Corp. (Santa Clara, Calif.), the world's largest semiconductor manufacturer, appears to have reservations about the expense of immersion lithography.

"No one really knows what immersion tools will cost," says Peter Silverman, an Intel fellow and director of the company's lithography capital equipment. "That's an issue for immersion lithography."

For its part, Intel appears to be targeting extreme ultraviolet (EUV) lithography for end-of-decade manufacturing. EUV uses light sources with wavelengths about 13 nm and requires technological advances so that manufacturing can take place. Just before the Sematech meeting, Intel announced an agreement to provide light source maker Cymer Inc. (San Diego, Calif.) with $20 million in funding over three years to accelerate development of production-worthy EUV lithography light sources.

Intel, like other companies, is bypassing 157-nm lithography. Although at one time 157 nm was a leading candidate, Silverman and others say it requires extensive investment in calcium fluoride technology. That isn't taking place, and so the technology appears dead; hence, the need for alternatives for 2007 and beyond.

"The only problem at the 45-nm half-pitch node is extensibility."
Burn Lin, Taiwan Semiconductor Manufacturing Company Ltd.
Click here to enlarge image

Not everyone has Intel's reservations about immersion lithography, where high-purity water or another suitable liquid acts as the final optical element. The liquid is injected between the lens and wafer, forming a skin and allowing narrower lines to be imaged. This is seen as a way to extend the usefulness of current 193-nm technology. One advantage that this approach offers is the use of lenses with higher numerical apertures, which can produce narrower features but can also be more expensive to manufacture.

"The only problem that I can see for 193-nm immersion at the 45-nm half-pitch node in 2009 is extensibility," says Burn Lin, senior director of the micropatterning division at Taiwan Semiconductor Manufacturing Company Ltd (TSMC; Hsinchu, Taiwan)—the world's largest semiconductor foundry. "The numerical aperture required for 193-nm immersion at the 45-nm half-pitch node can be kept at 1.3, which is the threshold of cost escalation."