SEOUL, South Korea—Lithography tool vendor ASML Holding Tuesday reported third quarter sales and profit roughly in line with analysts' estimates and said that productivity improvements demonstrated by its extreme ultraviolet (EUV) source providers have ASML on track to deliver EUV scanners suitable for commercial semiconductor production in 2013 and 2014.
In a statement, Eric Meurice, ASML's president, chairman and CEO, acknowledged that EUV source power development is progressing more slowly than planned. But Meurice said, two of ASML's power source suppliers have now demonstrated technologies for power levels that should enable ASML to improve the throughput of EUV scanners from the current level--less than 10 wafers per hour--to about 15 wafers per hour. The new source technologies will be implemented in the fourth quarter. Meurice said the new development puts ASML on a roadmap to improve throughput to commercially viable levels by the summer of 2012.
"We are therefore proceeding with ramping our capacity to address a number of DRAM, NAND and logic critical layers for customer wafer production in 2013 and 2014," Meurice said.
Meurice's pronouncement that EUV will be ready for some commercial production by 2013 is likely to be greeted with skepticism by some. Several attendees and presenters at last month's Bacus Photomask Conference in Monterey, Calif., seemed resigned to the idea that a robust, reliable EUV power source will not be available in time to meet current schedules.
Samsung Electronics Co. Ltd. wants to insert EUV into volume memory production sometime in the 2013 to 2015 timeframe, while Intel Corp. hopes to use EUV as its mainstream lithography for critical layers at the 10-nm node in the second half of 2015. Both companies are also developing contingency plans to extend optical lithography further in the event that EUV is not ready for commercial production in time to meet their schedules. But further extension of optical lithography down to the 10-nm node would be complicated and costly, likely requiring triple-, quadruple- or even quintuple-patterning.
Low throughput due to lack of an adequate source has dogged EUV development for years. The technology at one time was to be implemented at the 22-nm node, which Intel is initiating this year, but has been pushed out several times. In addition to the source, there are a number of other hurdles still facing EUV, including photoresists, metrology, and photomask inspection, repair and cleaning.
Chip makers would like EUV scanners with throughput of 100 to 150 wafers per hour. But to date pre-production tools have offered throughput of less than 10 wafers per hour, a rate that would make commercial production economically unfeasible. Some have suggested that throughput of 80 wafers per hour might be good enough, but others are skeptical that even that rate would make EUV attractive to broad adoption.
ASML also announced Tuesday that its latest optical immersion lithography scanner, Twinscan NXT:1950i, has achieved throughput of over 200 wafers per hour and that one of those tools has surpassed the productivity milestone of more than 4,000 wafers in a single day at a customer manufacturing site.
Meurice did not say which of ASML's three EUV source providers demonstrated the improved technologies. But at Bacus, Bill Arnold, ASML's chief scientist, said the company had sources from Cymer Inc. and Ushio Inc. and said the company expects to get a third source from GigaPhoton Inc. next year. Arnold said at the time that ASML recognizes the urgency of developing an adequate EUV source and had deployed some of its best engineers to be stationed on-site at suppliers' facilities to oversee and help with the development.
I just don't see it happening anytime soon for EUV litho. There are so many issues that still need to be addressed. I see only more delays ahead. A 2015 insertion point for "real manufacturing" seems optimistic to me.
years ago I worked for PE which became svgl now it is ASML...I remember microscan twinscan etc etc....but most of all our little company BBE was ahead of ASML and it is now 20 years....Charlie K of PE, Frank Reidel of PE both dead now...and I am the only one left...we were and are ahead of our times.....BBE, xray stepper just google....
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.