Is EUV Ready for Prime Time?
Thursday, 22 February 2006
Delegates exchange cards at the ASML booth during the SPIE Microlithography 2006 Exhibition in San Jose. Netherlands-based ASML announced that the first of two EUV alpha demo systems is fully assembled. ASML plans to ship the world's first 0.25 NA EUV systems in Q2 2006 to Albany NanoTech at the State University of New York at Albany, N.Y., and the Interuniversity Micro Electronics Center (IMEC) in Leuven, Belgium.
The SPIE Microlithography 2006 Exhibition closed yesterday, and judging from the buzz on the show floor, and the raft of company press releases, product announcements and news reports, the advanced lithography industry is enjoying healthy growth. Tools and processes emphasized at Microlithography 2006 typically fall into two camps: those that facilitate the extension of 193-nm lithography to the 32-nm node, such as immersion lithography, and those tools researchers hope will help them render the 193nm process obsolete altogether, like those using extreme ultraviolet lithography (EUVL). Talk of EUVL seems to be as much on people's lips this year as immersion lithography has been in years past, a sign indeed that the practice of advanced lithography could be headed for a shift as toolmakers chase higher yields.
Of the companies chasing EUVL, Netherlands-based ASML is thought to have their tools in the most advanced stage of readiness, thought the engineers and researchers at Nikon would argue that their EUVL tools, while not quite ready for prime time, have more of the engineering in place for volume production.
So how does EUVL work? Kim Eun-Jin, Hanyang Univ. (Korea), gave an interesting overview of EUVL as part of the "Emerging lithographic Technologies" Conference chaired by Michael J. Lercel, SEMATECH, Inc. and IBM. EUVL uses a very short exposure wavelength of 13.5 nm. So it has many characteristic in common with optical lithography, but EUVL are different from the conventional mask applied to the projection optical lithograph. Industry experts generally agree, Kim pointed out, that the biggest challenges and risks for the next generation of lithography systems involve the mask.
In EUVL, a mask is produced by applying multilayers of molybdenum and silicon to a flat substrate. The circuit pattern is produced by applying a final EUV-absorbing metal layer and then etching away the metal to form the image of the circuit. Also, the light shining with 6 degrees oblique to mask can not get target CD easily because of the shadow effect on pattern resolution.
Kim and his colleagues have been trying to change the structure of the mask to decrease this effect while keeping enough process latitude for the 32 nm node. EUV mask is affected by the thickness and type of absorber and buffer material. First, they changed the absorber material--typically Cr, TaN and Ge etc.--without changing the buffer material. Second, they changed the thickness of the absorber materials, while trying to minimize the shadow effect by adjusting the side wall angle of the absorber layer parallel to the oblique incidence.
SPIE Microlithography 2006 Plenary Session Plays to Packed House
Monday, 20 February 2006
SPIE Microlithography Symposium chair Anthony Yen, Cymer, Inc. introduces speakers to a packed house in attendance for the Monday Evening Plenary session. Now in its 31st year, Microlithography, held at the San Jose Convention Center and Marriott Hotel, has become the "go-to" event for the advanced lithography industry.
SPIE's 31st International Symposium on Microlithography officially began today at the San Jose Convention Center. This meeting has become the premier event for advanced lithography, and as pressure mounts to squeeze out higher yields while controlling costs, the role of the SPIE's Microlithography as a venue for lithography professionals to explore cost-effective ways to attain these yields will only increase.
In an attempt to frame the challenges and obstacles that confront the practice of advanced lithography, Yan Borodovsky of Intel began Micolithography's plenary presentations with a discussion of three broad classes of lithography solutions for keeping up with the demanding march of Moore's law: evolutionary solutions, revolutionary solutions, and disruptive solutions. He defined these in terms of their impact to existing technology and their cost and risk factors to schedules and yields.
He defined "evolutionary" as solutions extending current technology but at low increase of cost and schedule risk. "Revolutionary" he defined as discarding parts of current technology at similar cost but higher schedule risk, and "disruptive" he defined as a solution with high promises but unpredictable risk of schedule and yield.
Dr. Borodovsky then compared assorted lithographic processes on these terms to help predict key events that will define where emerging technologies will play a part in future large-scale manufacturing.
Eli Yablonovitch, University of California, speaks at the SPIE Microlithography 2006 plenary session about research currently underway by Yablonovich and colleagues to use the properties inherent in photonic crystals to extend current lithography practices far into the sub-wavelength domain.
Eli Yablonovitch of the University of California then gave a compelling talk about the wavelength-scale optics of photonic crystals, and sub-wavelength emerging technologies such as plasmonics for near field, and new techniques in biological microsopy for far field.
His thorough discussion about the basics of photonic bandgaps in crystals began with - surprisingly - peacocks, because a peacock's colorful plume is just one of many natural examples of the interference effect in photonic crystals. Yablonovitch went on to discuss how photolithography has made nanophotonic circuits (such as ring modulators, CMOS optical modulators, detectors and waveguides) possible, bringing on the field of "silicon nanophotonics."
He then began a fascinating discussion about plasmonics, describing how plasmons -being at optical frequencies, but x-ray wavelengths - give incredibly high refractive indexes (n could equal 100!). He then demonstrated how a plasmonic lens could be effectively used for the slider of a hard disk, and how people currently working with such applications are using working distances of only 5-6 nm.
The question then posed was: How far can we go? What are the true limits of optical lithography? Prof. Yablonovitch proposed that the resolution limit relevant to the future will actually be related to the phase angle accuracy.
Biological microscopy afforded yet another look at sub-wavelength advances. In particular, stimulated-emission depletion microscopy (STED) becomes sub-wavelength through a lithographic-type technique very similar to photolithography, and will likely share the same limits as optical lithography.
The final discussion in Prof. Yablonovitch's talk re-examined the fundamentals of resolution enhancement techniques (RET). When one considers the intersection of computational power advances and inverse algorithm requirements, he showed that inverse lithography technology (ILT) will provide non-unique and counterintuitive answers that will lead to superior photomask patterns. These will be mathematically rigorous, he asserted, but essential as we move into the nano realm.
Kurt Ronse of IMEC (Belgium) then gave an excellent overview of trends in lithography that will continue the scaling in integrated circuits. He began with the simplicity of the Rayleigh equation and methodically attacked each variable to show how different technologies could contribute to resolution improvement.
He first approached numerical aperture (NA) with immersion lithography. Although immersion has evolved from research to the development stage in record time, the control of defects such as air bubbles, watermarks, drying stains, and resist/TC-water interaction are determining its readiness for volume manufacturing. However, defects per wafer are decreasing and the outlook is that hyper-NA immersion lithography is coming. Polarization of the exposure light will play a new, important role, and this will depend on feature orientation, which will give room for design optimization. Unfortunately, NA is limited to 1.35 with water immersion and 2nd and 3rd generation high-index fluids have been investigated as a replacement, but need further study.
Dr. Ronse then attacked lambda with EUV lithography. With the shorter wavelengths, he showed how EUVL could open up the wall beyond 32 nm, although many challenges such as source power and optics lifetime act as barriers to this realization. Dr. Ronse gave an overview of the worldwide coordination to tackle those very difficulties.
Finally, he went after k1 with ArF immersion with double patterning. This, he showed, would push us toward a half-pitch of 32 nm, but this method is the worst in terms of cost-of-ownership (CoO), so anything that can improve CoO is very welcome.
Having assaulted Rayleigh on all fronts, Dr. Ronse also showed how new materials such as Ni-FUSI, Poly/TiN, and TaN are being investigated to reverse the flattening of interconnect scaling.
In summary, he stated that ArF immersion is well underway for the 45nm mode, and there are three options for obtaining 32-nm half-pitch mode: 1.65-NA ArF immersion, EUV lithography, and 1.3-NA ArF immersion lithography with double patterning. Furthermore, although new materials are needed, he stated, interconnect scaling will certainly continue.