I cannot take credit for the stuff below. I took the time to scan the numerous discussion forums (SI, Raging Bull, Yahoo, and Motley Fool) and came across the following useful information. I will take credit for collating, excerpting, summarizing, and adding my short opinion since the information should really be credited to the message board posters such as the people here, that are kind enough to post information for all to digest. The short summary for what you are about to read is that DUV will be around for quite a few years, and as we have been telling readers for the longest time, we should not be at all concerned about the so-called advanced process technologies that could obsolete the need for CYMI products.
BTW - the first article is really slanted towards the SVGL Micrascan but we will not get into our beliefs as to how much (or little) SVGL will penetrate the market. The more overriding situation is the "controversy" over when and which laser will be associated with each advanced technology along with the cost effective use of OPC or PSM retcles to extend the life of the existing lithography systems. In other words, like I have said for years and experienced for years, e-beam and x-ray lithography will not go mainstream for at least the next 5-10 years.<GGG>
Andrew Vance
RadarView
March 22 - Just How Fast Will 193-nm Litho Take Over?
Semiconductor Business News - Despite all the hype pushing argon fluoride 193-nm lithography, the new leading-edge exposure tool is having trouble dislodging the current state-of-the-art systems. Industry experts say the main reason for this delay is that recent advances in optics and photoresist have extended the life of existing krypton fluoride, 248-nm tools. This means that the older tools are taking over the latest-generation of 0.13-micron processing, which had long been expected to require the new 193-nm tool. 193-nm wavelength lithography systems will eventually be installed in fabs and continue optical processing's long reign. But just how many and how fast they will be installed is still open to debate. These next-generation production systems are slated to start modest shipments this year, but no one expects them to hit major production rates until 2002 or 2003 at the earliest.
No one should be too surprised to learn that memory-chip makers are expected to use whatever technical assists they can find like phase shift masks and optical enhancements, for example, to stay with their present 248-nm tools through the 0.13-micron generation node. The transition will go slow even at the logic chipmakers that were the early adopters. They will need only modest numbers of the 193-nm units to turn out the critical layers in the 0.13-micron process. They will still use 248-nm and even i-line systems for the bulk of non-critical layers. Market forecasts of lithography tool shipments by Dataquest bear out this slow transition. Even sales of i-line tools will continue to outsell 193-nm scanners until 2004, the market researcher predicts, and shipments of 248-nm systems will continue to make up half of the lithography market through 2004.
Only 24 of the 193-nm tools will ship this year, Dataquest predicts. This modest number will grow to 54 units in 2001, 84 in 2002, 167 in 2003, and 307 in 2004, it says. This rate will run far below the production rates of 248-nm tools. The market researcher expects 632 of these systems will be shipped this year, 859 in 2001, 913 in 2002, 792 in 2003 and 720 in 2004.
The slow takeoff of 193-nm litho should not be a surprise. This year, chipmakers will start installing 193-nm tools in 200-mm wafer development lines to gain experience with the new technology. Then as they move to 0.13-micron processing, the chipmakers will start using 193-nm litho on 200-mm wafer production lines. Then, he says, when "many firms open new 300-mm wafer fabs in 2002, we can expect to see a big jump in 193-nm production tools.
The switch to the larger-diameter wafers could be tricky. And the three major lithography system vendors differ in their approaches to moving 193-nm processing from 8-inch to 12-inch wafer lines.
SVGL designed its Micrascan-193 as a bridge tool that can be used on either size of wafer. Nikon's 193-nm system was designed so that a 200-mm wafer tool can be quickly modified in the field so that it can pattern 300-mm wafers. ASM Lithography, on the other hand, developed separate 193-nm platforms for 200-mm wafers and for 300-mm wafer. ASML management thinks it makes sense to have the most advanced platform for a tool that might be extended down to the 70-nm (0.07-micron) node. The lenses and laser illuminator are common to both 200-mm and 300-mm systems, so the transition can be made without great difficulty.
The three big litho vendors also are going their different ways with argon-fluoride laser light sources. SVGL uses one from Lambda Physik while Nikon gets its source from Cymer Inc. Proprietary differences in the lasers from the two major suppliers will mean that most tool makers will end up using either one or the other, says Pascal Didier, Cymer senior vice president of customer operations. ASML will supply 193-nm tools with either the Lambda Physik or the Cymer laser, depending on customer preference. The two lasers can't be used interchangeably, so ASML will have two versions of its 193-nm tool, one for each illuminator.
Another laser supplier also hopes to gain a foothold in this deep-UV lithography market. Japan's Komatsu Ltd. already has developed its 193-nm argon fluoride laser. But none of the new high-powered lasers will be easy to design into lithography imaging systems. For one thing, they can easily damage lenses in the 193-nm tools. One way to eliminate much of the thermal stress that could damage the sensitive optics is to fabricate lenses with calcium fluoride. There was some initial worry that the calcium fluoride material could turn into a major supply problem. But most toolmakers now agree that this is no longer an issue.
SVGL's CEO Shamaly says the Micrscan-193 gets around this problem by using proprietary catadioptic lenses that require less calcium fluoride material. The lens design also is inherently more resistant to thermal stress from the high-power 193-nm laser.
Cymer's argon fluoride laser has doubled the operational pulse rate to 4-kiloHertz and cut the power in half to 5 millijoules per pulse as compared with its existing 248-nm laser, says Didier. "This allows a calcium fluoride lens to handle the high power laser without the threat of thermal damage," he claims. Japan's Seiko Epson turns out 248-nm krypton fluoride lasers for Nikon and Canon tools under license from Cymer. But Cymer will produce all the new 193-nm lasers initially at its San Diego plant, Didier notes. "We will transfer technology to Sieko Epson after the manufacturing process is mature," he says. Cymer will do it the same way with its upcoming argon fluoride laser, adds Didier.
Moving lithography down to sub-0.18-micron processing also is causing new resist and particle contamination problems. As a result, the chemically augmented resists and antireflective-coating resists that are used in 193-nm lithography must be very tightly controlled, he says. The depth of focus of exposure tools also is getting so minuscule, as low as 200 nm, that it is difficult to get a flat wafer over the entire field of view. This problem can cause tiny defects in the wafer pattern where particles may have distorted or blocked exposure. The wafer can also pick up particles from the chuck that places it on the exposure tool. At the feature line sizes used in 193-mm lithography, even the smallest particle can cause a problem.
Chipmakers will need a range of techniques to cope with the increased particle contamination threat at 193-mm, from traditional inspection tools and simulation to fine-tuning processes to prevent contamination in the first place. Real-time feedback of inspection data also can be analyzed in a network database to catch immediately any processing problems before they lower the yields of an entire production run.
Once that 193-nm lithography is running well in the fabs, chipmakers expect that it will be extended to finer and finer design rules, just as 248-nm systems have. As a result, most toolmakers believe that 193-nm systems will be able to handle 100-nm (0.10-micron) processing with little problem. Indeed, some lithography vendors predict that better lenses, optical enhancements, and phase-shift masks could even push 193-nm down to the 70-nm (0.07-micron) level.
If that happens, some industry observers are now beginning to question whether the next lithography breakthrough, the 157-nm F2 laser tool called for in the semiconductor technology roadmap for the 70-nm process will even be needed.
BOTTOM LINE: The three big litho vendors also are going their different ways with argon-fluoride laser light sources. Proprietary differences in the lasers from the two major suppliers will mean that most toolmakers will end up using either one or the other. ASML will supply 193-nm tools with either the Lambda Physik or the Cymer laser. The two lasers can't be used interchangeably, so ASML will have two versions of its 193-nm tool, one for each illuminator. Japan's Komatsu Ltd., another laser supplier, hopes to be a player in the deep-UV lithography market. None of the new high-powered lasers will be easy to design into existing lithography imaging systems. Fabricating lenses with calcium fluoride is a possible solution. Cymer's argon fluoride laser has doubled the operational pulse rate to 4-kiloHertz and cut the power in half to 5 millijoules per pulse as compared with its existing 248-nm laser. This allows a calcium fluoride lens to handle the high power laser without the threat of thermal damage. Japan's Seiko Epson turns out 248-nm krypton fluoride lasers for Nikon and Canon tools under license from Cymer. The roadmap for DUV lithography may be unclear as to which laser will be used at certain feature sizes but it is evident that DUV lasers from suppliers like CYMER will dominate the market for at least the next 4 advanced technologies, going out more than 5 years and probably 10 years, putting a damper on the implementation of x-ray or e-beam systems as mainstream litho tools. Also, as we have stated before, reticle technology (OPC and PSM) will be an integral part of the lithography process, helping to push out the requirement for advanced lasers for at least one or two generations. All in all, this means that the business visibility for CYMI is measured in years and not quarters, as there are still a great deal of i-line processes that will eventually migrate to DUV as feature sizes shrink. CYMI controls more than 80% of the laser market. They are very well positioned within the chip industry and almost all of the chipmakers will rely on CYMER's laser technology to manufacture advanced devices for faster and more complex circuits.
March 21, Transcript of CNBC Interview With Cymer Chairman And CEO Robert Akins
SUMMARY: Akins explains the company's excimer laser illumination sources for deep ultraviolet (DUV) photolithography systems. The company's stock has risen with the rest of the sector in the past year, more than doubling its share price.
AKINS: Business is doing well and is ahead of our own expectations and being driven primarily by the demand for a 0.18 micron leading devices and beyond. This is where our technology really starts to make a difference in the production of advanced integrated circuits.
MARK: If I am making .25 and want to make 0.18u, I need a new laser too, right?
AKINS: That's correct. We produce a family of lasers, each one with a shorter wavelength than the preceding one. And the way mother nature has arranged it, the shorter the wavelength of light, the higher resolution you can attain at the wafer. So these very fine features are being produced by light, whose wavelength is short, and far beyond the range of human vision.
MARK: If I want to work with copper instead of aluminum, do I need a new laser?
AKINS: In general, copper is being used for the high conductivity where you don't have much room to put the conductor. So in that respect, the DUV lithography is patterning the copper lines to make the most of them. Think of copper and DUV as complimentary technologies to help accomplish the big fast chips necessary for the Internet and communications and so on and so forth.
MARK: So do I need a new laser to use copper instead of anything else?
AKINS: For every generation, you need a new laser light sort, first for a quarter micron, then for 0.18 and for 0.15 and 0.13.
MARK: How about going form 200mm to 300mm wafers?
AKINS: We don't make a laser that is wafer size specific, but if you are a chipmaker and you want to go 300mm, which has twice the area of a 200mm, you would like a laser with twice the power. And we have introduced a laser with twice the power just for that reason.
MARK: The point of the question is obvious. I'm trying to figure out what is going to drive growth in the future and it sounds like you have the same drivers that applied has or anyone else.
AKINS: That is absolutely correct. I think you probably have discussed the fact that, historically, PCs have been the major driver in the sector. Now, the Internet and other communications and telecommunications devices are driving it just as strongly, if not more strongly for all the leading edge devices. You need the small dimensions to give you small chips that are energy efficient, so that batteries last longer. So any kind of portable access device is driving the business.
Charles Kadlec: How much further can it go? Now you are saying .15. Can we have another 20-years of doubling?
AKINS: I believe so. I am a big fan of optical lithography. And I believe in Moore's Law. The future is clear. You'll see optical lithography using light sources, such as the ones we make, become the technology of choice for the next 10-years. |
