Aaron Hand

EUV Lithography Workshop: A Guest Perspective
June 17, 2008

In my June 11 Lithography Report, I asked readers for feedback on the International Workshop on EUV Lithography, a new conference organized by Vivek Bakshi and EUV Litho Inc., which took place last week in Maui. I got such a detailed and informative response from one reader that I wanted to share it with you here.

So this latest installment is a guest post from Kenneth Goldberg, a physicist at Lawrence Berkeley National Laboratory (LBNL) who works on EUV lithography issues, primarily optics testing (EUV interferometry) and EUV actinic mask inspection (AIMS-type and scanning). He attended the conference, where he led a mask inspection panel discussion on Thursday, co-chaired a session on metrology, and chaired the poster session.

This is what he had to say:

I’ve been a regular attendee at almost all of the EUV meetings for the past 10+ years. This meeting certainly had a different flavor than SPIE Advanced Lithography, EIPBN (International Conference on Electron, Ion, and Photon Beam Technology and Fabrication), or the annual International EUVL Symposium. A large number of presentations came from universities and national labs, which have been a primary source of innovation and original ideas for EUVL. Instead of focusing on commercial products and roadmaps from major suppliers, presentations had an R&D focus, as per the intention of the workshop. Thus the usual emphasis on “latest and greatest” was replaced with a “How are we going to make this work?” approach to the field. According to Vivek, the conference attendees were split approximately 37% United States, 37% Asia, and 26% Europe.

There were a large number of 20-minute invited and review talks, giving speakers an opportunity to share their experiences with a longer view of the field, past and future. I also noticed that projects and topics that often don’t get much attention were featured here, particularly on sources.

The mask inspection panel was originally put together by Kurt Kimmel (Advanced Mask Technology Center, Dresden, Germany), but when Kurt couldn't make it, I was asked to step in, rather late in the game. Mask inspection and defects are consistently near the top of the list of critical issues for EUVL. One of the main questions Dr. Kimmel raised asked how commercial mask inspection tools can be ready in time for the 22 nm node, given the current state of the field and the lack of commercial suppliers engaged in making tools. Professor Kinoshita (University of Hyogo, Japan), widely regarded as “the father of EUVL,” presented his group’s EUV mask inspection efforts, including two EUV microscopes based on very different concepts. Professor Pui (University of Minnesota) discussed the production of PSL spheres with high size uniformity, for use in the calibration of inspection tools. My own presentation echoed Dr. Kimmel’s (which I delivered for him). We both showed a number of converging facts from recent conference presentations (experiments and simulations) with evidence that defects below 29 nm SEVD are critical at EUV wavelengths, and that such defects are below the detection sensitivity of today’s UV laser-based inspection tools. I then showed that the total power requirements for an EUV-based brightfield mask-blank inspection tool, with sensitivity to defects on the order of 10-20 nm SEVD, are well within the reach of even today’s EUV sources — so on paper, it’s entirely feasible to create such a tool.

Goldberg summarized the current state of EUV mask inspection using a graph presented by Samsung’s Dr. Wonil Cho at last year’s SPIE Photomask Technology conference. Recent experiments with EUV (actinic) imaging have demonstrated that small defects may in fact be less printable that simulations predict. That’s the good news. On the other hand, even with the recent advances in commercial tool inspection sensitivity, the defects relevant for 40 nm HP lines are still below detection limits using UV light.
Note: “AIT” refers to data from the Actinic Inspection Tool, a SEMATECH-funded EUV-light microscope at Lawrence Berkeley National Laboratory.

The panel discussion took some interesting turns. Here are a few points: Everyone agrees that defects are a huge concern for EUVL, with no clear commercial inspection solution now or in the pipeline. Presentations at recent conferences reveal that worldwide, there is so much work taking place on innovative sources and exotic source concepts. There is also a lot of attention being paid to resists and line edge roughness (LER) with new formulations and theories being created and tested routinely. But in the area of mask inspection, by comparison, there is almost no new work taking place outside of a few national labs and university projects — really just a few people are working on it, and that makes no sense strategically. In my personal opinion, with the low level of research investment being applied in this area, relative to the importance and urgency of the work, one would have to conclude that the semiconductor industry is relying on magic to somehow solve the inspection problem.

To be fair, I did mention that a lot of money was spent on EUV mask inspection, but the bankruptcy of Exitech took critical momentum away from this area — and that was in 2006. However you slice it, attention to this field was dropped for some reason, and it hasn't been replaced. Today’s prototype inspection tools, including my group’s EUV microscope at LBNL, cannot keep up with the demand (in terms of time on the tool) for mask development research data (including aerial images at different NA values), whether for defect sensitivity testing, defect repair work, or simply mask-blank inspection.

One questioner raised the issue that companies will not get involved in the creation of EUV mask inspection tools until we show for sure that actinic inspection (i.e., using EUV light) is required; so the experts need to tell us whether or not it is required. I suggested that since we the panelists are scientists, evidence-based thinkers, we must be able to conduct experiments in order to answer one way or the other, conclusively. The missing piece is not just the lack of engagement of commercial suppliers; it is in funding research that is specifically geared toward the 22 nm node, or beyond.

My suggestion was that, before the industry or commercial companies invest tens of millions of dollars on integrated, standalone mask inspection tools based on a single inspection concept, we (the EUVL community) need flexible research test beds where various ideas can be tested quickly, and concepts can be proven.

Separate from the panel, other mask-related authors at the conference discussed particle-removal methods, and there was an excellent presentation comparing the defect-resistant quality of new mask boxes that had been shipped all around Japan and dropped from varying heights.

All in all, it was a very worthwhile meeting, with a collegial atmosphere, and (of course) a fantastic setting as well. I expect the participants will bring their enthusiasm and renewed focus back to their home institutions. Many of us will reconvene at the upcoming SEMATECH-sponsored EUVL Symposium in Lake Tahoe, Nev.

I have certainly felt over the past several years that the content of conferences within the lithography community are being controlled more and more by the big corporations and what they dictate is, as Goldberg puts it, the “latest and greatest.” There have been complaints that fundamental research just is not being funded lately, and the picture that Goldberg paints about EUV mask inspection is indeed a frightening one.

I’d love to hear more thoughts on the situation, and more feedback from last week’s workshop. Please post your comments here, and keep the thread going.

Posted by Aaron Hand on June 17, 2008 | Comments (2)