Intel pushes out EUV lithography

SAN JOSE, Calif. — Unable to obtain the tools and materials in time, Intel Corp. has apparently pushed out its efforts to bring extreme ultraviolet (EUV) lithography into “high-volume production” for the 32-nm node by 2009. Instead, the company said that it plans to extend 193-nm lithography for mass production at 32-nm.

For years, Intel (Santa Clara, Calif.) has publicly stated that it would like to deploy EUV lithography for chip production at the 32-nm node. Now, the company hopes to insert EUV in a “pilot line” setting instead of mass production for the 32-nm node — if the tools and materials are available at that time, said Yan Borodovsky, Intel senior fellow and director of advanced lithography in its Technology and Manufacturing Group.

At Intel, EUV, along with various 193-nm techniques, are possible options for chip production at the 22-nm node in 2011, Borodovsky said. But for 32-nm “high-volume production,” the company is now exploring various means to extend 193-nm wavelength lithography, including the use of dry, immersion and double-exposure techniques, he said.

“We will look at dry and wet solutions [for the 32-nm node],” he said in a presentation at the SPIE Microlithography conference on Monday (Feb. 20).

EUV lithography had been seen as a likely next-generation chip processing technology, based on 13-nm wavelength illumination, allowing chip makers to print features sizes of 32-nm and below on integrated circuits.

According to recent statements by Intel, the company plans to deploy 193-nm “dry” scanners for the 45-nm node in the 2007 time frame. Then, at 32-nm, the chip giant said that it was exploring several options, such as EUV and 193-nm wavelength lithography, including immersion.

And for years, ASML Holding NV, Canon Inc. and Nikon Corp. have been separately developing EUV tools for mainstream chip production. ASML of the Netherlands is expected to ship two “alpha demonstration” EUV tools in 2006, while Nikon of Japan is supposed to deliver one machine in 2007, according to analysts.

Intel, the major proponent of EUV, had hoped to obtain production-worthy EUV tools in 2007 or so in order to insert the systems in time for the 32-nm node. “We hoped to have the tools in 2007,” Borodovsky said. “The tools and materials are not available.”

Indeed, EUV still faces a plethora of technical challenges despite billions of dollars of R&D being poured into the technology. At present, there are no adequate power sources for EUV. The shelve life of a condenser for an EUV tool is about a month now. Many had hoped to see shelf lives of four years.

There are still photoresist and line-edge roughness problems with EUV. And there is no way to inspect the masks, which are supposed to be defect free.

“I’m not disappointed with the progress,” according to the Intel technologist. “[The vendors] are doing better than I thought. It’s a complex technology.”