|Analyst: EUV Misses 14nm Node and Now Aims for 10nm|
By Mark LaPedus, SemiMD senior editor
Extreme ultraviolet (EUV) lithography has missed the 20nm — and the 14nm — process nodes, according to an analyst, who added that the tool technology is down but not out of the running.
At last count, EUV was supposed to be inserted for IC production at the 22/20nm node, but the technology has apparently missed that window. Although chip makers won’t publicly admit it, the word at this week’s SPIE Advanced Lithography conference is that EUV has also missed the 14nm node.
The optimists still believe that EUV has a possible chance to hit the 14nm node. But one analyst, who has been (and still is) on the EUV bandwagon, is not so optimistic. “It’s missed 14nm,” said G. Dan Hutcheson, president of VLSI Research Inc. “To do 14nm, you need the tools in the fab today.”
The EUV tool supplier — ASML Holding NV — is not expected to deliver its first production unit until the second half of 2012 and through 2013. And the throughput is lagging due to the lack of an adequate power source.
Now, the industry is shooting for the insertion of EUV at the 10nm node, Hutcheson said. “ Samsung might do some layers at 14nm, but real production is 10nm,” he said.
Another chip maker, Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC), “is focusing on EUV for high-volume manufacturing production at 14nm,” but that won’t occur at least until “2015,” said C.J. Muse, an analyst with Barclays Capital. Publicly, TSMC said it is still evaluating EUV, maskless and multi-patterning for the 14nm node. Intel Corp. won’t insert EUV until 10nm.
The pulse on EUV
At one time, there was a sense that EUV would never succeed, because it was too costly and complicated. More recently, as the technology progressed, the sentiment for EUV was not “if but when.” This week, there was a general feeling of disappointment — and pessimism — with the lithographic odds makers betting against EUV again.
VLSI’s Hutcheson believes that EUV is still viable and could get inserted into production fabs as early as 2015. Hutcheson said he always thought EUV would not get inserted until sometime between 2015 to 2020. Because “people over promised” the technology and delivery schedules, the perception is that EUV is wildly late and in trouble, he said.
The reality, according to the analyst, is that EUV is “the last man standing in NGL.” Nano-imprint and maskless are not ready for prime time now or in the distant future. “Imprint has too many issues like defectivity,” he said. “E-beam has huge cost and infrastructure issues.”
EUV has its own inherit problems, namely the power source. “Gaps between device requirements and the current capabilities of EUV litho are still present in: scanner throughput and availability; EUV to 193 overlay and in-use mask defectivity; mask blank defectivity and mask substrate flatness; and resist line-edge roughness,” said Obert Wood, principal member of the technical staff for strategic lithography technology at GlobalFoundries, during a presentation at SPIE.
EUV lags on several technical fronts (Source: GlobalFoundries)
Because of the lack of adequate power sources, EUV throughputs are running about 4 wafers an hour right now. The industry requires throughputs of around 100 wafers an hour to make EUV economically viable in production fabs. For some time, Cymer Inc. has been shipping an EUV power source operating at 8 Watts, which is designed for ASML’s NXE:3100, a pre-production EUV machine. The industry is looking for at least 100 Watts, and eventually, 250 Watts, which support production-worthly throughputs of 60 and 125 wafers an hour, respectively.
ASML is expected to ship the NXE:3300B, a full-blown, 13.5nm EUV production tool, to undisclosed customers in the second half of 2012. For “early NXE:3300 adopters,” ASML vows it will deliver an EUV tool with a “stable performance” and a throughput at 69 wafers an hour.
On Thursday (Feb. 16), ASML and Cymer will separately claim progress on the power front during various presentations at SPIE, according to an analyst. “Here (at SPIE) we heard few surprises, with ASML telling customers that the company was currently showing 50 Watt at 80 percent duty cycle internally with plans to reach this level at customer’s facilities by the end of the year,” Muse said.
“The source power has increased significantly,” said Hans Meiling, product manager for ASML, during a presentation at SPIE, but “we have steps to go.”
Meiling said EUV has generally made solid progress. In the first quarter of last year, the company’s NXE:3100 pre-production tool demonstrated the ability to print 22nm and 18nm images, with an overlay of about 4nm. The unit had a 2 Watt source, which could support a throughput of 1 wafer an hour.
In January of 2012, the company’s EUV tool demonstrated the ability to print 16nm images, with an overlay of about 2nm. The unit had a 9 Watt source, which could support a throughput of 5 wafers an hour, he said.
One source, developed by Cymer, has demonstrated the ability to achieve 90 Watts burst mode with a duty cycle of 20 percent, he said. It has also demonstrated the ability to achieve 50 Watts with a duty cycle of 80 percent, he said. A 50 Watt EUV source is capable of throughputs of 40 wafers an hour, he added.
During a separate presentation, David Brandt, senior director of marketing for Cymer, said the company will shortly “deploy” or ship its previously-announced 20 Watt source. By the end of this year, the company is expected to ship a 100 Watt source. That source is based on what it calls a pre-pulse technology.
Cymer pushes LPP for EUV source (Source: Company)
Another EUV source vendor, Xtreme, “has shipped two tools to ASML for (the) 3300 and is driving to achieve 50 Watt at 100 percent duty cycle at module level,” Muse said.
At present, Xtreme’s source has achieved 7 Watts of power. The company said it has shipped a 20 Watt system to ASML in the first quarter of 2012. The company will deliver a 50 Watt unit in the second quarter and a 70-to-90 Watt unit by the third quarter.
Xtreme and the third EUV source vendor – Gigaphoton Inc. – are still behind Cymer. At SPIE, Phil Alibrandi, director of the sales and account management division at Gigaphoton USA Inc., acknowledged the company is “late” with its EUV source, but it’s far too early to declare a winner in the arena.
Gigaphoton took one step forward, by announcing that its mass-production laser-produced plasma (LPP) light source, scheduled to be shipped in 2012, generated 7 watts of EUV power. This was achieved with a tin droplet irradiated by a combination of solid-state laser pre-pulse and 2.7 kW, 90kHz carbon dioxide (CO2) laser main pulse.
A maximum of 0.3 mJ, 27 Watt EUV light was obtained at the plasma point running the LPP light source at 30 percent duty cycle for one hour. “The achievement of first light with a mass-production technology LPP light source confirms that Gigaphoton’s unique LPP light source solutions can be implemented to ensure stable performance and low-cost operation,” said Yuji Watanabe, president of Gigaphoton.
Share and Enjoy: