Researchers claim 248-nm lithography could have longer life than believed
MIT-led group produces experimental 9-nm device to show the end isn't near
By Mark LaPedus
Semiconductor Business News
(03/07/01 14:37 p.m. PST)
SANTA CLARA, Calif. -- In what could have a profound impact on semiconductor manufacturing, a group of researchers said they have demonstrated that 248-nm lithography tools could still have a productive life in 0.10-micron and below process technologies.
The group, led by Lincoln Laboratories at the Massachusetts Institute of Technology (MIT), also claims it has used 248-nm lithography to produce a 0.009-micron device (9 nanometers) to show just how far today's exposure tools could go in shrinking critical dimensions in ICs using subwavelength photomask technologies.
The breakthrough, reported during last week's SPIE Microlithography Conference, theoretically means that chip makers could extend the life of their existing 248-nm exposure tools to perhaps 2005, thereby pushing out the need for advanced and more expensive scanners now in development. This includes the new 193- and 157-nm tools, and quite possibly, the next-generation lithography (NGL) technologies, such as extreme ultraviolet (EUV) and electron projection lithography (EPL), according to some experts.
In a paper presented at the SPIE conference here, MIT researchers described the development of the 9-nm poly gate silicon-on-insulator (SOI) transistor by using a 248-nm tool from Japan's Canon Inc. The 9-nm device was produced with so-called reduced-enhanced technologies.
MIT also said it set another world's record with the help of Canon's EX-6 line of 248-nm lithography tools. With this system and a chromeless phase-shift mask from reticle maker Photronics Inc., researchers at MIT said they demonstrated a device with a 210-nm pitch grating (105-nm lines and spaces) design rules.
In an interview with SBN, officials from MIT acknowledged that the 9-nm device produced with 248-nm lithography is not a working part. It is merely an experimental device devised in a controlled laboratory setting.
But what the research organization attempted to prove is that 248-nm tools continue to push the limits in semiconductor manufacturing far beyond what experts had previously thought possible. More specifically, MIT researchers hoped to prove that 248-nm tools are suitable for minimum feature sizes at the 100-nm technology node (0.10 micron)--and perhaps beyond.
Last year, in fact, MIT demonstrated a working SOI transistor, based on 25-nm design rules using a 248-nm tool from Canon and phase-shift, double-exposure technology from Numerical Technologies Inc. of San Jose.
"My main point is that there is plenty of life left in 248-nm lithography tools, with strong resolution enhancement technologies (RETs)," declared Michael Fritze, a staff member at Lincoln Labs, who presented the paper at SPIE last week in Santa Clara. "I believe that 100-nm node is possible using high-NA (numerical apeture) 248-nm sources with resolution enhancements such as strong phase-shifting [photomasks]," Fritze claimed, in an interview with SBN.
In other words, the workhorse 248-nm tools can be extended to limits that were not predicted by the industry experts.
"This possibility is not even mentioned in the 2000 ITRS Roadmap," said Fritze, referring to the International Technology Roadmap for Semiconductors. "The 130-nm node is being done around the world using 248-nm tools with RETs. But I believe these methods can be extended to make high-NA 248-nm tools viable for the 100-nm node, which would carry this technology out to the 2004 to 2005 time frame," he added.
This could be a major benefit for chip makers, but not necessarily a good thing for lithography suppliers, which are now eager to start shipping higher-priced 193-nm scanners. "This [extension of existing tools] would be a great advantage for the industry since 248-nm technology has a large installed base," suggested the MIT researcher. "This message may not be popular with the folks who have a lot invested in the quick introduction of 193-nm technology."
But just how far will 248-nm scanners will go, in terms of practical design rules? When the 248-nm tools were introduced in the late 1990s, they were mainly targeted for processing devices down to 0.25-micron, recalled analyst Klaus-Dieter Rinnen, who tracks the industry segment at Dataquest Inc. in San Jose.
"Now, the 248-nm tools are being used in 0.18- and 0.13-micron fabs," Rinnen noted. "Maybe they will also be used in the 100-nm [0.10-micron] node."
Not surprisingly, tool vendors disagreed with MIT's assessment of the potential for greater extension of 248-nm lithography. Leading scanner suppliers believe the 248-nm systems are quickly running out of steam, and some say it is getting too expensive to extend the existing tool sets much future in volume production.
"With a conventional [248-nm tool], you can achieve 100-nm [devices]," observed Phillip Ware, director and general manager of marketing for the Semiconductor Equipment Division at Canon U.S.A. Inc. "But that's with an extremely low K1 factor," Ware said. "That's going to be painful."
How far 248-nm will go in device shrinks depends greatly upon the type of products being made by a wafer fab, said other suppliers. "If you talk to the foundry guys right now, they are going to try to extend their 248-nm tools," said Dave Chavoustie, executive vice presidentof sales for ASM Lithography of the Netherlands. "It's a different story with the DRAM makers. But from 0.12-micron on down, we're talking about 193-nm technology," he said.
Other suppliers agree. "The trigger point appears for 248-nm [tools] appears to be 130-nm [process technology]," said John Wiesner, senior vice president of engineering for Nikon's U.S. subsidiary, Nikon Precision Inc., in Belmont, Calif. "But the goal [for chip makers] is to push these tools as far as possible," he added. |
