A new ballgame for lithography beyond higher NA
06/01/2006
While the leading lithography equipment suppliers race to achieve systems with higher numerical apertures, they are employing radically different means to achieve their ends. Understanding these various approaches can be critical when evaluating which technology best fits your company’s needs
For the first time in decades, the big three lithography system suppliers are pursuing really different strategies, and it shows. Of course, numerical aperture, or NA, remains important. Nikon talked about its 1.30 NA tool that will be ready this year, just in time for 45nm. Canon is pursuing a 1.35 NA and a hyper-NA capability of ≥1.5, in the belief that the theoretical limits of immersion can be pushed with new "cool-NAid" liquids (my term, not theirs) being developed. Also, ASML will have its 1.20 NA tool ready this year and should follow this with a 1.35 NA tool in 2007 or 2008. But underneath this race to higher NAs are some fascinating differences that may have a huge impact on your decision about which system to buy. Here’s my take on the key differentiators in the hardware you’ll be evaluating.
New fluids versus EUV
The hyper-NA fluids will be proprietary. So it is quite possible that one tool supplier could gain an unassailable advantage and dominate. It’s clear that ASML is pursuing EUV as its primary horse beyond 32nm and will actually ship two tools this year. Canon is focused on Cool-NAids. Nikon has been straddling the fence, with efforts in both areas. Success or failure in one or the other will change the competitive landscape of the semiconductor industry. The last time this happened was in the late 1970s, when the choice was between DUV projection, e-beam, and steppers. Poor choices by some chipmakers caused them to lose their market positions. Others emerged with the right choice. This is not a simple question of how to decorate your fab. Lithography choices at these junctures can be a life or death choice.
Immersion nozzles
For the first time, lithography systems are doing something more than shedding light on the wafer. Managing the liquid interface between the wafer and the lens is critical to yield. It also adds complexity to the system, which could be a reliability differentiator between the systems. I really like Nikon’s immersion nozzle because it has been kept relatively simple, and tests show no bubbles exist at high speed scan rates. Nikon has done something really special with the nozzle because the company has always claimed no bubbles from the beginning and every new test customers devise reproduces the same results. I’d like to see independent data from a customer or ISMI, but it’s certainly something to watch based on the results I’ve seen. Of course, ASML and its customers have the most field experience, with 13 immersion systems in the field as of this writing.
Catadioptric lens design
For NAs above 1.10, everyone has agreed to abandon refractive lenses and go catadioptric. (SVGL was right, just 20 years too early.) Catadioptric lenses have both reflective and refractive (through the glass) elements. Reflective mirrors are more difficult. They tend to have more flare; the thermal aberrations are harder to compensate for; and the field aberrations are harder to control. Worse, lensmakers cannot add elements with abandon as they have done in the past because aberrations are additive, and about 30% of the light is absorbed at each mirror. Although ASML has loads of technology derived from their SVGL acquisition, lensmakers in general are not known for their skills here. This is a whole new playing field. Expect some upsets.
Moreover, the shapes of these lenses will be radically different, so shoe-horning them into existing stages will be difficult. Nikon’s new tandem stage was designed with future catadioptric lenses in mind. Canon is designing an entirely new dual stage to go with its new 1.35 NA lens. ASML is extending its industry-leading twin-stage design. What I like about the newer stages is that they allow three elements with lower degrees of off-axis positioning. This means it is unlikely that they will need the 80 to 90W lasers being developed for four-element lenses. Existing 60W lasers will be fine, so this could be a huge savings. Also, element life will be longer and thermal management issues in the lens will be less critical because you’re pumping less energy into the elements. If three-element suppliers decide to use higher wattage lasers, they will have higher throughput.
Design innovation
The issues are no longer about simply raising NAs. System design and innovation will have a huge impact on the on-wafer results from new-generation systems. ASML’s TwinScan stage and Nikon’s Polano polarization source are good examples of how innovations in system design are already adding real differentiation points today. The unique features of ASML’s TwinScan enabled it to pull from behind and take the lead in immersion. Nikon’s Polano has reinvigorated the company; the source adds much higher depth-of-focus (DOF) than can be achieved with nonpolarized light. At 45nm, DOF is about 2×, which adds huge amounts of process latitude in manufacturing. This results directly in higher yields, so it is a real moneymaker. Nikon’s new Tandem stage is another example of innovation. And, of course, I can’t wait to see what Canon develops. All in all, it’s an exciting time in lithography with great innovations and cool new systems coming along, making it a whole new ball game.
Rules of the game
First and foremost, never bet on a technology until you must do a change-lock-down for manufacturing. There are simply too many good people out there working independently, and often a single idea from one of them can make or break a technology. For example, just imagine where we would be today if the pellicle hadn’t been developed. This simple idea transformed chipmaking. Imagine where you would be if you had bet on proximity x-ray. Here the breakthrough ideas needed to make it manufacturing-ready never came to be. Innovation or lack of it can make or break careers. Be wary of committing to a technology in which somebody still has to think of something for it to be a manufacturing reality.
Second, be cognizant of how the supplier’s roadmap aligns with your company’s. Even though the technology is the hottest thing, if your company is not on the bleeding edge, you may be able to wait until it has been fully debugged and even through several generations of debugging. If you are on the bleeding edge, it is almost always advisable to bet on lithography technologies that are manufacturing-worthy when you are ready to move to manufacturing.
There is no single correct decision for all companies. You must strike a delicate balance between the technology you need and its affordability, and then align it to your manufacturing and market strategies.
 |
G. Dan Hutcheson is the CEO of VLSI Research Inc. and a well-known semiconductor industry visionary whose career experience spans more than 20 years. He has authored numerous publications, developed many industry models, and researched most aspects of the semiconductor industry. VLSI Research Inc., 2880 Lakeside Dr., Suite 350, Santa Clara, CA 95054; ph 408/453-8844.