To get a handle on the lithography and laser source markets, SBN caught up with Robert Akins, co-founder, chairman and CEO of Cymer Inc. The San Diego-based company not only is a leading supplier of excimer lasers for litho tools, but it is also making a run at the NGL market with a line of light sources.

SBN: Where does Cymer fit in the overall chip-equipment market?

Akins: Cymer is the leading supplier of excimer illumination sources for deep ultraviolet photolithography. Our primary products are high power and extremely narrow-spectrum DUV lasers for leading-edge chip manufacturing, and our three direct customers are today's leading lithographic exposure tool manufacturers--Canon Inc., ASM Lithography and Nikon Inc.

Currently, we are focused on developing a high-volume production, 193-nm laser using Cymer's new Master Oscillator Power Amplifier (MOPA), dual-gas-discharge-chamber technology. This new architecture is a paradigm shift from current lithography light source technology that will enable next-generation light sources to deliver higher power, tighter bandwidth and lower cost of operation for future optical lithography applications across all three DUV wavelengths of 248-nm, 193-nm and 157-nm.

On July 22, Cymer announced the introduction of the XLA 100 argon fluoride (ArF) light source, which incorporates our MOPA technology--delivering significant throughput advantages for 193-nm lithography applications.

SBN: What kind of demand are you seeing in terms of product types right now? Is the market shifting from 248-nm to 193-nm lithography tools and laser sources?

Akins:The current generation of high-throughput scanners for 193-nm chip production are expected to hit fabs in mid-2003 for patterning at below 0.13-micron. Cymer, however, expects 248-nm lithography tools to comprise the majority of lithography tools until 2006 when 193-nm tools will take the lead.

SBN: How are you faring in comparison to your competitors, Lambda Physik and Gigaphoton?

Akins: This has been a long and deep downturn, and we are lucky to be in a strong financial position, with plenty of cash and working capital. Because Gigaphoton is a private company, it's difficult to say how they're faring financially, since they're not required to report their numbers. We know that Lambda Physik has experienced a significant decline in their semiconductor business. Both of our competitors receive government financial support-- something we expect to continue indefinitely. Perhaps the most telling comparison is that of the success at chipmakers over the past four quarters: 91% of DUV light sources installed at chipmakers were Cymer's.

SBN: What is your overall outlook in terms of sales and/or growth for Cymer for the remainder of this year? 2003?

Akins: We experienced sequential revenue and earnings growth through the first three quarters of 2002, but because we experienced a slowdown in bookings in the third quarter, we are projecting that fourth quarter revenue will be between 20 percent and 25 percent below third quarter revenue of $84.5 million. So we don't expect to see sales growth in the fourth quarter.

Visibility into 2003 is quite limited at this time, and the market is very volatile. We are not willing to try to project revenue for 2003, because of the lack of visibility. We don't expect a significant change from the late part of 2002 in the early part of 2003, however. Of course, we believe the upturn will come, but it is difficult to understand why it would happen any sooner than the second half of 2003.

SBNWhat is your outlook for the overall IC and/or semiconductor equipment industries for 2002 and 2003?

Akins: Again, our visibility is limited enough that we can't see a pickup in the near-term future. From past experience, we know that for the industry to see a pickup by the end of Q1 of '03, or early in Q2, we should be talking with our customers about our ability to ramp production and deliver systems in high quantities at this time.

We are not having conversations of that type presently. This tells us the pick up is, at the earliest, a second half of 2003 event, and that it is probably later than that. But because visibility is so limited, this should be considered speculation.

SBN: I assume that the laser source market reflects overall lithography trends. At present, chipmakers are scrambling to extend their 248-nm tools to the 130-nm node and beyond to cut costs. How far will chipmakers be able to extend their 248-nm tools before they must obtain 193-nm scanners? At what nodes?

Akins: The transition point from krypton fluoride (KrF) to ArF will be different for various chipmakers, depending primarily on the type of products they manufacture. The memory chip manufacturers will push KrF to feature sizes of 100-90-nm by employing numerous optical techniques--especially exotic and expensive mask technologies. They can amortize the mask costs across larger production runs of identical chips.

However, for the foundry model, where frequent mask changes are the norm, or if you are constantly revising the design of a microprocessor, it can be more economically advantageous to adopt ArF earlier--say just below 0.13-micron--and use less expensive masks. The complexity of the decision implies that such technologies would lead to a mixed installation of 248-nm and 193-nm at sub-100-nm exposure nodes.

SBN: Do you believe 157-nm lithography will have a place in chip production in spite of the delays with the technology? Will 193-nm tools, or even NGL, displace 157-nm at the 65-nm node?

Akins: In general, the most desirable approach to high-resolution lithography for leading chipmakers is to shrink wavelength and history teaches us that, so far, no viable wavelength has been skipped over. So, if 157-nm lithography and its process infrastructure were ready for introduction at the 65-nm node, it would be the preferred choice. It is uncertain, however, whether 157-nm technology will be ready for the 65-nm node. Of course, chipmakers have back-up plans, such as the use of high numerical aperture (NA) 193-nm tools for this node.

At this time, it seems that ArF is the most likely approach for the 65-nm node, although it may not be possible to meet all the design rules with this technology. With regard to next-generation lithography (NGL) technologies, extreme ultraviolet (EUV) lithography will probably not be ready in time for the 65-nm node. Proponents of electron-beam projection lithography (EPL), however, believe that this technology may be ready in time to play a limited role for low-volume applications.

SBN: What's your outlook for NGL? Some say it will happen by 2007, but some say NGL will never happen.

Akins: The leading industry chipmakers are working to put all the infrastructure for EUV lithography in place by 2007, and this is the earliest that it would be introduced. There are many challenges, however, not just in light sources, but in optics, defect-free masks and resists. Although the leading chipmakers would prefer to use the low-wavelength solution, they will have back-up approaches in case of delays and if EUV lithography is not mature enough for full-scale production. We expect that conventional optical lithography, however, will eventually need to be replaced and that EUV will be the preferred choice for NGL.

SBN: What type of EUV laser technologies are you developing and how does it compare against competitive technologies?

Akins: Here at Cymer, we are not working to develop an EUV laser. We are working to develop a Discharge Produced Plasma (DPP) EUV light source. There are two viable types of EUV sources, Laser Produced Plasma (LPP) and DPP. EUV light, in both cases, is produced by an emission from hot plasma. The LPP uses a high-power laser to heat the plasma, but the laser itself is not an EUV laser.

We are developing a DPP source in which we use an electrical discharge to heat a pinched plasma. The electrical pulse power system is based on proven technology from our excimer lasers. We believe that the industry would prefer a DPP source because it offers better overall efficiency, lower cost of operation and a much smaller footprint, compared to the LPP sources. Competitors are developing both LPP and DPP sources.

The performance of the different DPP sources is quite similar right now, but the key for success is being able to scale output power to very high levels and to engineer a reliable, production-worthy source-something Cymer has the best track record for doing.

SBN: TRW claims it will shortly ship the world's first production-worthy EUV source. Is Cymer ahead or behind TRW in this arena?

Akins: Production-worthy sources are not expected to be available for several more years. The sources being developed, and possibly shipped, over the next year or so will probably generate much less power than will eventually be needed for high-volume production, and will be employed in R&D and process development application.

SBN: Some say the standalone laser-source market could disappear all together. For example, in the optical era, the laser is sold as a standalone system with the lithography tool. But for NGL, the laser will be integrated with a EUV tool. Can you comment on this change or trend?

Akins: Today's laser may stand separately from the exposure tool but it is tightly integrated into the complete system. This level of integration will be greater with EUV sources, due to additional constraints on matching the collection and illumination optics and on vacuum compatibility. We believe the source will continue to be a specialized sub-system provided by a separate supplier, however, and then integrated into the exposure tool.

SBN: At present, there are three major competitors in the laser-source market, including Cymer. But a dozen or so companies are developing laser sources for NGL, namely EUV. Is there room enough for dozen or so laser-source providers for EUV. If not, how many are there room for?

Akins: The current situation parallels that of DUV light source development in the 1980s, when there were about a dozen companies who had thrown their hats into the ring. Over time, that dozen companies became three.

We don't believe that all the EUV source research projects share the goal of developing a production-worthy EUV source. There is a lot more involved in building a reliable source than simply developing a promising concept for EUV emission. We don't expect the market to be able to support a large number of suppliers, and we expect the large number of companies currently working on EUV sources to consolidate in a similar way to that in which we experienced in the DUV light source arena.