Franklin Kalk, CTO, Toppan Photomasks

Franklin Kalk (Source: Toppan Photomasks)

Franklin Kalk is CTO at Toppan Photomasks (Round Rock, Texas), the same position he held at DuPont Photomasks before it was acquired by Toppan Printing. He leads Toppan Photomasks' global R&D programs, supporting customers' technology roadmaps. He joined DuPont Photomasks in 1992, and served as manager of the R&D group at the DPI Reticle Technology Center (RTC). For years, Kalk has focused on photomask resolution enhancement technology issues and led company-wide programs aimed at improving the quality and yield of photomask products. Prior to joining DuPont Photomasks, he held a number of engineering and research positions within the electronics group of E.I. duPont de Nemours & Co. Kalk holds 15 U.S. patents and has published numerous technical papers. He has a Ph.D. in optics from the University of Rochester and a B.S. in physics from Emory University.

SI: Last April, DuPont Photomasks was acquired by Toppan Printing and renamed accordingly. Can you tell us something about that transition?

Kalk: There were a few hurdles to jump over. From a technology standpoint, looking at any acquisition or merger of this kind (where the two companies have been competitors in the past), there are many opportunities one can take advantage of, although we didn't directly compete in many markets and had fairly complementary market and manufacturing footprints.

SI: Still, some time must pass before you stop thinking in terms of "them" and "us."

Kalk: Certainly. And this holds throughout all company functions, whether sales or technology. The first thing to do is to create a bridge. This is a fairly short-term activity. Then there is a longer-term one that can take a year or longer, depending on the particular function, which is merging the roadmaps. The integration process was intense, but we were helped by the fact that engineers like to communicate with each other. We had one set of technical meetings where one group would talk, then the other, and when either group spoke the other nodded its communal head affirmatively. We came out of this realizing that we all faced the same fundamental issues, and that technology is becoming increasingly important in the mask industry as, surely, it is just about everywhere else.

SI: So your original technology map experienced changes because of the acquisition?

Kalk: It was a very curious thing. We put the companies' technology roadmaps side by side, and studied them in great detail. We found that what drove our technologies was the lithography options that our customers were planning to use over the next three to five years. We also looked at our respective specification and toolkit roadmaps, and discovered that the toolkit roadmaps were virtually identical. That was encouraging, but not startling, since the mask industry is not very big and has a limited range of suppliers, all of who face the same problems. For example, there aren't 10 different companies you can go to for inspection tools. The spec roadmaps also matched up well. ...Although there wasn't a complete year-by-year alignment of all values (and the spec roadmap is very detailed), there was still a very good alignment regarding what both of us thought we could and couldn't do. To follow on with that, on the areas we thought we could not do, there was good alignment about what was necessary to get there — whether it was a new tool, additional engineering, new materials, etc. Plus, we were almost perfectly aligned on the customer litho maps.

SI: I suppose that's a testimony to the fact that there aren't many ways of doing high-end lithography.

Kalk: Indeed!

SI: So now that the cuffs and collars match, what is your R&D effort's focus?

Kalk: Today, 90 nm is in early-stage production, and we're doing it with 193 nm ArF lithography using weak phase shifters. The push is toward finishing off qualifications, bringing in customers who are on the tail end, and driving costs down through better yields and cycle times. Presently, 65 nm is under very heavy development, and full 65 nm mask sets are being shipped to leading-edge customers. Again, this is 193 nm litho for criticals with some use of strong phase shifting, which is a more complicated technology, but one that DuPont Photomasks started in the early 1990s working with Sematech; so we have a rich history with strong phase shifters, and view this node as a straightforward extension of existing technologies. Now, 45 nm is generally considered in the ITRS as the last non-EUV (or next-generation lithography [NGL]) node. It will be 193 immersion — or at least that is the conventional wisdom — for criticals, and there will be wider use of strong phase shifting.

SI: How do you fit your roadmap to that of the end user?

Kalk: It requires close collaboration on both parts. You don't want to have to develop five technologies and masks to end up with one or two at a given design rule — it's too costly. At 130 nm, for example, we had to develop phase-shifting masks for both 248 and 193 nm wavelengths, but when all was said and done, only 248 was used. Thus, we did twice the necessary amount of development to put the technology in the ground. This was a lesson learned about narrowing options early on, and not trying to build everything. ...Toppan recently signed an agreement with IBM to do development work for 45 nm masks. This should generate some truly solid technology we can fan out to our manufacturing base.

SI: When will photomask technology run out of tricks — at 45 nm?

Kalk: I have seen all the modeling, calculations, as well as the experiments, and I'm not so sure that 45 is the last node for long-wavelength photons, "long wavelength" meaning "not EUV." As we've gone down in design rules, we have increasingly restricted what the lithographer can do to print something. You can imagine printing just lines in the vertical or the horizontal dimensions as we go further and further. So it isn't obvious to me that 45 nm will be the last node to use 193 for critical layers — we may be able to do it at 32. In 1997, the roadmap predicted that 193 would come in around 1999, and that we'd have 180 nm design rules in early production in 2001; in fact, the opposite occurred; 180 was in early production in 1999, and 193 began being used in volume during the last year or so. We successfully push out wavelength technology changes, while pulling in design-rule reduction.

SI: Doing photomask defect inspection is increasingly difficult and expensive. Is there anything that the inspection OEMs ought to be doing?

Kalk: If we go NGL — EUV, for example, since it has experienced the greatest development work — very significant changes will be required to do inspection for EUV masks. They are reflective, not transmissive, and feature sizes will be incredibly small. Inspecting masks will be slightly simplified merely because, with EUV masks (assuming that they come in at 32 nm), the amount of OPC will be significantly reduced compared to what it has been, because the k₁ is comparatively higher for EUV. A relaxed k₁ makes for a relaxed OPC, which simplifies inspection. The flip side is that feature sizes will be incredibly small and it is not clear that long-wave — DUV — inspection will be easy to do.

SI: Much will be required, though.

Kalk: Yes. Right now, inspection tools run at about 257 nm, around lithography's DUV wavelength. Do we need a shorter wavelength below 200 nm or must we go all the way to a very short wavelength, like 13 nm? If we must go to 13, it won't be ready when needed. However, I don't believe that this will be necessary. I think that it's going to be less than 200 nm — 198 or 193 wavelength. Sophisticated algorithms will be needed to handle the mask feature sizes; however, at the same time, some relaxation of the requirements will take place, because the OPC won't be as intense. Another possibility is to push 193 immersion one more node. If we use it for critical layers at 32 nm, the inspection tool has its work cut out for it. It'll be difficult, because even today the very top tools have some major limitations when attempting to inspect things like strong shifters; they aren't as capable when inspecting these very complex masks.

SI: Inspection OEMs must measure the unmeasurable, so to speak, which means an increasing reliance on models. How do you view this?

Kalk: For current 248 and 193 nm lithographies, there are mask inspection tools that, instead of doing a very high-NA image acquisition and looking for defects, use an imaging approach that is closer to a scanner's, and just look for linewidth variations instead of actual physical defects. If we go to EUV masks, that'll be hard to do because, at the beginning, there won't be any EUV wavelength inspection tools; we'll be back to the traditional direct image acquisition approach.

SI: What would you consider the device designer and device maker's greatest challenge over the next few years?

Kalk: If you look at design over the last six years, the total number of designs has decreased. This is because design now takes far more time than it used to and is costlier. Then you have increasing NRE costs and longer cycle times. You must be able to predict that you will have a winner before you even start. And who wants to drop $10M or $20M in nine or 10 months and not be able at least to assess the probability that you will have a successful product? This is a key issue.

People talk about killer apps. What is today's killer app? I don't know. Is it convergence? If it is, the designer must ensure it will fit the killer app a year from now, when it sees production. So, in a nutshell, the device manufacturer's major challenge is a combination of high costs, long cycle times, and fairly low predictability of success.

SI: What do you see as the biggest problem facing the industry today?

Kalk: Without a doubt, integrating the supply and engineering chains. We use terms like "DFM" a lot, and that's exactly what partnering and collaboration are all about: integrating the supply chain, the engineering chain. This is difficult, because it doesn't require money; it requires resolve and a good dose of courage. We must start doing business differently. Now, from the technical side, from my perspective, one of the major problems the industry faces today in the lithography area is haze. This is where growth occurs on the mask — and it can occur just about anywhere on it: the pellicle, the glass, the pattern side, anywhere. Everyone owns this problem to a certain degree. It is not simply a mask, exposure tool, or fab environment issue; everyone owns this issue. Haze mitigation provides us a textbook opportunity to integrate the engineering chain.