Funding for E-Beam Lithography

Good question, Aaron.  I don't think it's chicken-and-egg.  E-beam direct write (EBDW) was funded back in the "NGL sweepstakes" days, but that was focused on technologies intended to replace optical lithography for high volume wafer manufacturing.  For low volume applications, alternatives weren't funded because they weren't really necessary.  Conventional optical lithography met the cost requirements of IC manufacturing for all designs.  The fundamental issue we'd like to address with EBDW now is the high cost of NRE (non-recurring engineering) for low-wafer-volume designs.  That's a relatively recent phenomenon, say since the 90nm technology cycle.

One other item: has anyone ever done a good market study to quantify the impact of mask cost on design activity, by node, by potential wafer volume?  I've only heard anecdotes and seen "models," but real data seem to be in short supply.  And I think that is at least partly responsible for the long induction time we've seen for direct write funding since EUV "won" the high-wafer-volume NGL sweepstakes.

That's a very good point, Franklin. Thank you for the reminder about those "NGL sweepstakes" days. I tend to still think of the lithography world back like it was when I was first introduced to the industry. I still have the t-shirt ASML handed out at Semicon West '98 — "2010: Clash of the Technology Titans," with four comic-book heroes representing X-ray, SCALPEL (e-beam), ion-beam and EUV. ASML had attendees vote at their booth to choose which technology would take over from optical lithography by 2010. If I remember correctly, Extreme UV Man was the chosen hero of the day.

But yes, times have changed and the technology needs have changed, with mask development costs becoming much more of a burden on low-volume designs.

I would definitely be interested in seeing a study like you mention, quantifying the impact of mask cost on design activity. Has anyone seen such a study?

Well, let me add my pessimistic view. I've spent 5 years in one of the most "advanced" recent EBDW litho projects (do not want to mention its name in vain). Not there anymore. I believe it is an illusion that there is something new the EBDW is capable to bring to the lithography plate. Regardless the funding volume. Maybe 1 or 2 niche applications with very low patterning density. Even  the currently most promising   approach requires specificalyy designed resists with very unique set of properties.

The illusion that funding can help comes from the comparison of the images of a monstrous EUV machine and a simple looking single e-beam litho tool. However, experts who spent some reasonable time in attempt to bring up the productivity know that the complexity of the EBDW machine and supporting SW explodes drastically once some reasonable productivity is targeted. At this point I think my pessimistic point is pretty realistic.

Wishing the best,

YG.

P.S. I am providing consulting for interested parties (Litho, CMP, wafer clean and prep, ALD films with targeted ellectrical properties, e-beam induced thin film and surface modifications and more [email protected])

Yehiel, thanks for your insightful comments.  Without divulging anything proprietary, could you clarify some of the technical targets for the EBDW effort in which you worked?  I'm interested in target throughput and accuracy specifications (e.g., was it 10 wph at 32nm design rules, etc.)?  Thanks again.

Actually it is in the public domain, just Google with REBL.

Best,

Yehiel

There are dark spots even on the Sun's surface. Well, a skilled observer can clearly see them next to some items of the Litho bright spots agenda. Not to say more...

Franklin, I just paid attention at the targets in your message (10@32@ and you did not specify patterning density, which appears to be a very very critical parameter). As you can figure out from the DARPA REBL specification, the targets are much more modest, fewer wph@45@<6%. If you will follow the DARPA specification historical adjustments, you will easily find out that the trend is more or less the same as the ITRS LER specification historical corrections,  say, starting from year 2005 (since I do not work for this project anymore, but I am still under legal obligations, I would prefer not to go into more specifics.

The targeted LER trend was corrected in reverse direction, unlike in the previous hystory, reflecting the enormous complications associated with improvement of this performance parameter. I have a couple of papers on these matters presented at several litho forums. The most recent one is atteched. I also attempetd to attach the associated slides (SPIE-09) but for some reson it did not work. If there somebody interested in this talk, I can present it where it will be available.

est,

Yehiel

Attachments: