Roadblocks seen in immersion lithography
Mark LaPedus
EE Times
(05/19/2006 4:20 PM EDT)
SAN JOSE, Calif. — There is a growing sentiment that high refractive index materials for 193-nm immersion lithography are not panning out, thereby implying that immersion will run out of gas at the 32-nm node.
Advanced 193-nm immersion tools, which use de-ionized water with a refractive index of 1.44, are projected to process wafers down to the 32-nm “half-pitch” node, according to experts. ASML, Canon and Nikon are developing 193-nm immersion scanners and expect its respective customers will insert the tools at the 45-nm — and possibly 32-nm — "half-pitch" nodes.
Several companies, including DuPont, JSR and others, are working on non-water solutions — or so-called high refractive index materials — for 193-nm immersion lithography. Many of those solutions claim to have a refractive index of 1.64, enabling 193-nm immersion scanners to extend beyond the 32-nm node.
Many believe, however, that these materials are not viable — yet. This implies that 193-nm immersion will not scale beyond 32-nm.
“There are people working on fluids other than water for immersion lithography,” said Shang-Yi Chiang, senior vice president of research and development for silicon foundry giant Taiwan Semiconductor Manufacturing Co. Ltd. “So far, they haven’t worked out.”
As a result of this and other dynamics in the marketplace, the semiconductor industry is going back to the drawing board and looking at a number of next-generation lithography (NGL) candidates for 32-nm and beyond: direct-write e-beam, extreme ultraviolet (EUV), nano-imprint, among others. Other possibilities include 193-nm immersion with complex double-patterning or double-exposure techniques.
Still, the industry is worried about the lack of viable and future lithography candidates in the market. “We don’t see a simple solution,” Chiang said.
TSMC itself proposes to insert 193-nm immersion lithography tools for production at the 45-nm node. Then, the company plans to use 193-nm immersion at 32-nm, but it’s still unclear what the foundry provider will use beyond that node, he said.
On its roadmap, TSMC hopes to use fast direct-write electron-beam tools for 22-nm. “EUV is a possibility,” he said.
The problems with direct-write e-beams are clear: the tools are expensive and the throughputs are slow. Several companies are developing high-speed e-beam tools, but most — if not all — of those efforts are paper tigers and still in the R&D stage, according to analysts.
ASML Holding NV of the Netherlands has a slightly different viewpoint about the market. High refractive index solutions for 193-nm immersion “are very attractive,” said Bill Arnold, chief scientist for the lithography-tool giant.
“You will need new glass materials [for these solutions], but that’s where everyone is stumped right now,” Arnold said. “ASML does not feel that any of these materials can be made. We think that (high refractive index materials with immersion] is the least likely candidate for 32-nm.”
Like TSMC, ASML believes that 193-nm immersion will be inserted at the 45-nm “half-pitch” node. Beyond that, ASML contends that EUV has resurfaced as a viable candidate for 32-nm and below. “Long term, EUV is the technology that will take over,” he said.
ASML and Nikon are developing EUV systems. But EUV also has its drawbacks, and, simply put, the infrastructure is questionable. There are still no viable EUV sources and resists in the market. Plus, EUV is expensive: a tool is projected to run $50 million or more if or when the technology is available.
Another possibility is to extend 193-nm immersion scanners by using double-exposure or double-patterning techniques. “There are two problems [with those technologies],” said Hamid Zarringhalam, vice president of technical sales and marketing at Nikon Precision Inc. “One is cost. The other is overlay.”
Overlay and throughput are among the key problems for another NGL called nano-imprint lithography, according to analysts. Many vendors are selling nano-imprint tools, but those systems are being used for niche applications. Some experts believe that nano-imprint is not a viable technology for semiconductor production.
|
