ASML’s EUV Roadmap Points to New Wavelength
By Mark LaPedus, SemiMD senior editor
ASML Holding NV has just begun to ship its first pre-production, extreme ultraviolet (EUV) lithography tools based on 13.5nm wavelength technology.
Even before 13.5nm EUV is in full production — and amid constant delays for the technology — ASML for some time has been talking about extending EUV with smaller wavelengths. More recently, the company has been making a strong case in the industry for EUV based on wavelengths ranging from 6.6nm to 6.8nm, according to a recent presentation made by ASML, which was obtained by SemiMD. The presentation, made at the 2011 International Workshop on EUV and Soft X-Ray Sources, was held in Dublin, Ireland earlier this month.
ASML will continue to push current 13.5nm EUV technology for at least four more tool generations, according to the presentation that outlines the company’s roadmap.
Following 13.5nm EUV, ASML is exploring — or “under study” — for 6.6nm to 6.8nm wavelength EUV technology for chip production beyond the 8nm node, according to the roadmap. The proposed EUV scanner is slated for the 2018 time frame, according to the roadmap.
In other words, ASML believes it can extend EUV beyond 13.5nm to propel Moore’s Law. ASML did not acknowledge the contents of the presentation slides and roadmap. A company spokesman issued the following statement: “ASML expects 13.5nm wavelength-driven EUV to go down to 10nm imaging. We plan to accomplish this by larger NA (numerical aperture), improved resist sensitivity and off-axis illumination techniques. Of course our primary focus at the moment is getting EUV to commercially viable production levels. Beyond that, ASML is investigating the possibilities of lithography to further expand on 13.5nm to sub 10nm nodes, but these are pure research projects.”
ASML has recently shipped its pre-production EUV tool — the NXE:3100 — to Intel, Samsung, Toshiba, Hynix and TSMC. One customer, TSMC, is pleased with the 13.5nm wavelength EUV machine, but the throughput is lacking.
At a recent event, Obert Wood, a lithography research manager at GlobalFoundries and the Albany Research Alliance, also said source power remains a “worrisome” issue for EUV lithography. The EUV source power must increase to 100 Watts to support a throughput of 60 wafers per hour, the minimum threshold for commercial production, he said. EUV tools are currently processing wafers at less than 5 wafers an hour.
ASML outlines EUV roadmap at 13.5nm and 6xnm wavelengths (Source: Company)
In 2012, ASML is expected to ship the NXE: 3300B, a full-blown, 13.5nm production tool. Following the NXE:3300B, ASML is expected to deliver the NXE:3300C, the NXE:3300D and the NXE:3500.
According to the roadmap, ASML’s last 13.5nm EUV tool, the NXE:3500, will deliver 11nm resolution in the 2016 time frame. Beyond that, ASML is exploring the ways to extend EUV down to sub-5nm resolutions by increasing apertures up to 0.7 with a wavelength reduction down to 6.6nm to 6.8nm, according to the presentation.
C.J. Muse, an analyst with Barclays Capital, said that it is viable to develop EUV with smaller wavelengths, but a more pressing concern today is to make current 13.5nm, EUV work in production fabs. As reported, ASML has shipped pre-production tools, but the throughput is troubling due to the power source. “I think they are just hoping to get to 60 wafers per hour on 13.5nm EUV,” Muse said.
“I guess theoretically you could get wavelength down from current 13.5nm to sub-10nm,” he added. “But you would have to change the light source to get below 13.5nm. Perhaps Cymer, Extreme/Ushio or Gigaphoton are working on this.”
As before, there are several challenges for 6xnm EUV: imaging optics, source power, resists, among others. According to the presentation, ASML has demonstrated the first pilot multilayer mirror coating for the range of 6.6nm to 7.0nm.
Two potential source fuels are being investigated: Terbium (Tb) and Gadolinium Gd. A CE of 1.8 percent has been demonstrated, according to the presentation.
At the event in Ireland, Japan’s Gigaphoton presented its findings on a 6xnm source. “6xnm EUV light has a possibility to be used as light source for lithography process in semiconductor device manufacturing. Thus investigation has been started to get information to generate 6.x nm EUV,’’ according to an abstract given by the firm.
“The basic research of 6xnm EUV generation by Laser Produced Plasma (LPP) have been made with experiments in a LPP test rig build for 13.5 nm EUV light source development. The test rig has a target system in a vacuum vessel, drive laser system and metrology system for 6xnm EUV light. The target system can set planer target of Gadolinium. 6xnm EUV light is detected through La/B4C mirror as a spectrum band pass filter,’’ according to the company.
Three Russian entities, RnD-ISAN, Institute of Spectroscopy RAS and Keldysh Institute of Applied Mathematics, presented the following data: “Experimental studies of spectra of plasma of Gd and Tb in 6.X spectral region are reported. Spectra were excited in LPP and DPP types of radiation sources. Conversion efficiency of laser radiation (LPP) and stored electrical energy in 0.6 percent ‘in band’ spectral interval have been measured. CE as high as 1.8 percent was demonstrated for special type of LPP target geometry.”
A group led by Utsunomiya University said presented the following data: “We demonstrate an efficient extreme ultraviolet (EUV) source for operation at 6.7 nm by optimizing the optical thickness of Gadolinium (Gd) plasmas. Using low initial density Gd targets and dual laser pulse irradiation, we observed a maximum EUV conversion efficiency (CE) of 1.8 percent, which is 1.6 times larger than the 1.1 percent CE produced from a solid density target. Enhancement of the EUV CE by use of a low-density plasma is attributed to the reduction of self-absorption effects. In addition, strong resonant line emission at 6.76 nm from a discharge-produced Gd plasma was also obtained.”
–David Lammers contributed to this story
December 1st, 2011 at 2:11 am
This Grimm’s fairy tale, propagated by ASML and other tool suppliers who benefit dramatically from double and triple patterning at 193nm (read double or triple existing tool sales), continues endlessly. It started roughly 2 decades ago with the name change from SXPL (soft x-ray projection lithography) to EUV, successfully duping more people than it should have, that the technology was closer to DUV and x-ray. But you cannot fool Mother Nature, so this particular Grimm fairy tale will end more like a Grim fairy tale. We were promised an x-ray solution at 1 micron, 20 years ago. We were promised this particular x-ray solution (SXPL) at 65nm. Now we are told maybe 13nm and we are gratuitously told throughput needs to improve from “less than 5WPH (exactly what is “less than 5WPH: 3, 1, or is the appropriate metric really hours per wafer?). ASML has created more dramatic fairy tales about future generations, running at >200WPH, shorter wave lengths,etc., and they are nothing more than head fakes, designed to distract us from focusing on their egregiously missed promises, and how far removed they are today from a functional, not to mention economical, x-ray solution. Most fairy tales have happy endings. This Grim one does not.