SANTA CLARA, Calif.--In a new and intriguing twist in the lithography industry, major scanner suppliers are reportedly investigating the move to develop workhorse 248-nm tools equipped with immersion capabilities.
ASML Holding NV, Canon Inc., and Nikon Corp. are each reportedly looking into developing a 248-nm immersion scanner--a tool that in theory could extend 248-nm technology and push chip designs to the 70-nm node, according to analysts at the SPIE Microlithography conference here this week. However, some believe that 248-nm immersion would not be a mainstream technology, but rather a niche tool for "depth of focus" applications in chip making, sources said.
ASML, Canon, and Nikon reportedly have not committed on developing 248-nm immersion scanners for the commercial market--yet, according to sources. But "a few" undisclosed chip makers are said to be interested in 248-nm immersion and have prodded the scanner vendors to investigate the technology, according to sources.
The commercial development of 248-nm immersion could put a new twist in the already confusing market. At present, current "dry" 248-nm scanners are the mainstream lithography tools running in production fabs today, but the technology is expected to hit the wall and run out of gas at the 130-nm node. However, some chip makers are looking to push 248-nm scanners to the 100-nm node, with the help of resolution enhancement techniques (RETs).
And until recently, ASML, Canon and Nikon have been gearing up their immersion programs for 193-nm wavelength platforms only. Expected to hit production fabs in 2006 or so, the 193-nm immersion scanners are targeted for 65- and 45-nm chip production. Some believe that 193-nm immersion will extend to the 32- or perhaps the 22-nm node, thereby pushing out the need for 157-nm and extreme ultraviolet (EUV) technology for advanced chip making.
A possible scenario for both 248- and 193-nm immersion scanners is mix-and-match applications for advanced chip fabrication, although some believe that 248-nm "wet" tools will become nothing more than a novelty item. Some do not see the technology entering the commercial fray.
Is 248-nm wet for real?
"When I first heard about 248-nm immersion, I dismissed it," said lithography expert Griffith Resor. "I dismiss it, because I don't see the industry going backwards," said Resor, who is also the editor of Lithography Review, a newsletter published by the Semiconductor Equipment and Materials International (SEMI) trade group in San Jose.
At present, the industry is moving from 248- to 193-nm lithography. Resor believes it's unlikely that the industry will move backwards from 193- to 248-nm lithography--even with immersion on the horizon.
"The lithography vendors also don't need another project on their plates," he said. Lithography vendors already have their hands full with a number of projects, including 193-nm, 193-nm wet, and NGL. And on top of massive R&D budgets, vendors are losing money despite the upturn.
But if 248-nm wet lithography does in fact make it to the market, it will be a niche product for use in "depth of focus" applications, according to Resor.
On the other hand, there is a business issue with immersion. Some analysts pointed out that chip makers like Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC) could potentially be a customer for both 248- and 193-nm immersion tools. TSMC is already a major proponent of immersion and ASML is developing 193-nm tools for the silicon foundry giant.
One engineer at a major scanner maker also believes that 248-nm immersion is much easier to develop than 193-nm "wet" technology. In theory, a 248-nm immersion tool, which is geared for 70-nm chip production, could have a numerical aperture of 1.1, according to analysts.
The specifications and chip-production targets for a 248-nm immersion tool are somewhat similar to that of a 193-nm dry scanner, analysts said. But needless to say, 248-nm immersion would be less expensive than 193-nm wet technology.
The 193-nm technology also suffers from what some call compaction, which is damage caused by light to the fused silica material of the lithography lens. In general, 193-nm scanners use a combination of fused silica and calcium fluoride materials within the lithography lens. In contrast, 248-nm technology does not suffer from compaction; it uses only fused silica material on the lens, it was noted.
In any case, all scanner vendors are reportedly looking into 248-nm immersion. "We are investigating it," according an engineer at Nikon.
Nikon has not made a decision to commercialize 248-nm immersion. The Japanese lithography giant believes that 193-nm wet technology will be the mainstream platform and 248-nm immersion will become a niche--if that tool is even offered by vendors in the marketplace.
Rivals Canon and ASML are reportedly looking into the technology as well.
In fact, some believe that 193-nm will be the last technology wavelength in lithography. Optical scanners based on 193-nm wavelength technology are expected to last for 10 years or more in production fabs--a development that will eradicate the need for 157-nm and EUV lithography, according to an executive from KLA-Tencor Inc.
In a keynote address at the SPIE Microlithography conference on Tuesday (February 24, 2004), Chris Mack, vice president of lithography at chip-equipment provider KLA-Tencor of San Jose, said that 193-nm lithography is expected to last for several chip generations, especially with the advent of immersion techniques.
The combination of 193-nm and immersion may extend 193-nm to the 22-nm node, he said. "It's more than possible," he said. "It's quiet likely."
With 193-nm immersion on the horizon, Mach sees no hope for 157-nm dry or immersion lithography in the marketplace. "157-nm seems almost too challenging," he said during a keynote address today. "We almost would like that not to happen."
He also sees no room for EUV in spite of a major push and rhetoric by the champion of the technology--Intel Corp. "I don't read the rhetoric," he said.
"I am not an NGL guy, as you might suspect," he said. "I'm not a friend of NGL," he said during the keynote address (see February 23 story).
Karen Brown, an expert in the lithography industry and consultant for Idea Industries, believes that immersion is feasible, based on her cost of ownership model. But based on Brown's model, NGL technologies such as EUV and maskless are expected to be too expensive for production fabs--if or when these technologies hit the market.
In this cost of ownership model, the throughput of a lithography tool must be equal to or greater than the selling cost of the machine times six. For example, a 193-nm tool sells for between $14-to-20 million per unit. Therefore, the tool must have a throughput of 84-to-120 wafers per hour before it is economically feasible. Today's 193-nm scanners are said to have an overall throughput of 90-to-110 wafers an hour.
If or when a production-worthy EUV tool is introduced in the marketplace by 2010 or so, the price tag is projected to be $35-to-$70 million for a single unit, Brown said. In other words, the tool must have an overall throughput of 210-to-420 wafers an hours--an unlikely scenario based on the current power source issues with EUV, she implied.
Based on this model, maskless lithography also "is unaffordable for production," she said. A maskless tool is expected to run $25 million each, but it can produce only 5 wafers an hour or 22,000 wafers a year, she said.
This is not to say maskless should be abandoned. On the contrary, the industry is in need of new maskless lithography breakthroughs--and for good reason: Rising reticle costs are becoming major roadblocks in order to develop next-generation designs.
"Mask costs at $1 million are becoming unaffordable," Brown said.
Soaring mask costs could also hamper the new technology of choice. "Immersion lithography is affordable, but not if mask cost is $4 million," she said during a presentation at SPIE.