EETimes.com – Direct-write litho still facing uphill climb
SAN JOSE, Calif.—Multiple development efforts focused on e-beam direct-write lithography have reported progress this week at the SPIE Advanced Lithography conference. But, at least according to one prominent lithography researcher, production tools are still a minimum of five years away.
Based on the length of time it has historically taken for each new lithography technology to move from proof-of-concept to production, e-beam direct-write lithography tools will be available no sooner than 2015, according to Kurt Ronse, lithography department director at nanoelectronics research center IMEC.
Ronse offered some advice to the many companies and consortiums developing e-beam direct write technology: target the 16-nm node, because the technology won’t be commercially viable by the 22-nm node—the target of many of the development efforts.
Ronse also recommended that these groups initially apply their technology to mask-writing tools—where throughput requirement would not be so arduous—as a shorter term, intermediate step.
Chip makers continue to look longingly at direct-write lithography, which could potentially reduce or remove the need for photomasks, which are getting more expensive—according to data presented by Ronse, the cost of a mask set doubles at each new technology node. Analysts and industry executives label the rising cost of masks as the chief culprit behind an ominous trend: declining ASIC starts.
But technical issues—including unacceptably slow wafer writing times—have to date kept e-beam direct write lithography from moving closer to commercial production.
Providing an overview of the latest direct-write development work being done by three European-based companies, Ronse said the resolution of tools is getting closer to acceptable range for the 22- and 16-nm nodes, but that overlay control and throughput remain well short of what is needed.
“It’s a very interesting technology, and I have a lot of respect for the people developing it,” Ronse said. “But in terms of overlay and throughput I think there is a long way to go.”
A number of firms and consortiums that are developing e-beam direct write technologies gave presentations at SPIE, including Mapper Lithography BV, IMS Nanofabrication AG and the eBeam Initiative, a consortium of more than 25 companies headed by Direct2Silicon Inc. Others are also developing e-beam direct-write technologies, including KLA-Tencor Corp., Micronic Laser Systems AB, Vistec Electron Beam Lithography Group and Tokyo Electronic Ltd., in addition to government-backed research and universities.
That there remain so many distinct direct-write development efforts is testament to the technology’s potential market opportunity. Anyone who can bring to production an e-beam direct-write lithography technology stands to cash in, particularly in light of next-generation extreme ultraviolet (EUV) lithography being pushed out further to target production at the 16-nm node. This means lithographers will push 193-nm immersion lithography down to at least 22-nm, but there is widespread consensus that that technology is not extendible to the 11-nm node.
“I don’t want to say that we are desperate, but when the industry is on the cusp of a major change, you are going to see all of these new technologies emerge,” said Franklin Kalk, executive vice president and chief technology officer Toppan Photomasks Inc.
Toppan (Austin, Texas), a subsidiary of Japan’s Toppan Printing Co. Ltd., is a member of the E-Beam Initiative. Kalk said Toppan recognizes that for some chips that do not sell huge volumes, a full mask set can be cost prohibitive. Toppan supports e-beam direct-write in the hopes that it will eventually enable companies doing lower-volume designs to do critical layers using direct-write while continuing to purchase photomasks for the non-critical layers, which are less expensive but still the majority of a design, Kalk said.
“We see direct-write as keeping open an application space that is going to close,” Kalk said.
At SPIE, the E-Beam Initiative announced the additional of six companies, bringing the total number of members in the consortium to 26. D2S, the managing sponsor of the group, appears to be one of the companies following Ronse’s advice to apply direct-write technology to mask writing. The company announced at SPIE a new design for e-beam mask technology for the production of advanced optical photomasks with circular and curvilinear shapes, which the company said could help extend 193-nm immersion to the 22-nm node.
Another company that is doing as Ronse suggests is IMS (Vienna, Austria). In a keynote presentation at SPIE, Elmar Platzgummer, chief operating officer at IMS, said the company built two proof-of-concept tools last year and is continuing to develop its Projection Mask-Less Lithography (PLM2) technology, with a goal of building systems that offer 256,000 programmable electron multi-beams of 50 keV energy. But IMS is also developing the electron-optical column of a mask exposure tool for writing leading-edge, complex masks. The company’s eMET technology targets the 16-nm node and below.
During a paper presentation Tuesay (Feb. 23), Christof Klein, a project manager at IMS, said the company’s maskless technology should benefit from the knowledge being gained on the mask exposure tool. Klein said IMS is working on the mask exposure tool “very intently.”
Platzgummer said IMS has achieved good results in resolution, but acknowledged that throughput remains too slow. Writing a single 300-mm wafer at 60-nm half pitch still takes 20 hours, he said.
Mapper Lithography (Delft, The Netherlands), meanwhile, is not applying its multiple e-beam technology to mask writing. The company said last week that one of its tools, located at Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC)’s Fab 12 in Hsinchu, Taiwan, is printing features so far unachievable with current immersion lithography technology.
In another paper presentation Tuesday, Bert Jan Kampherbeek a co-founder of Mapper and currently the company’s vice president of sales and marketing, said Mapper shipped two “pre-alpha tools” last summer—one to TSMC and one to French technology research organization CEA-Leti.
Kampherbeek said Mapper has shown progress on resolution, including exposure of 20-nm SRAM cells. The company is currently focusing on increasing the throughput of its tools, including adding a moveable stage, Kampherbeek said. Mapper hopes to establish a throughput for its tools of 10 wafers per hour and then cluster 10 tools together in a system that can write 100 wafers per hour.
Mapper’s tools feature 110 electron beams that can be individually switched on and off by means of an optical blanker array.