Aaron Hand

Lithography Answers Blowin’ in our Wind?
February 28, 2008

As is often the case during the nighttime panels at SPIE, there was a fair bit of silliness in Tuesday night’s panel discussion, “Future Projection Lithography: Optical or EUV?” It’s a serious enough question, but after a very long day of listening to detailed technical presentations, attending customer/supplier meetings, and perusing the exhibit floor, these leaders of the teams trying to find answers to extremely difficult lithographic challenges were ready to loosen up a bit.

The tone was set very aptly by IBM’s Kit Ausschnitt, who entertained the audience with a reading of his latest musical adaptation — this time, with apologies to Bob Dylan, with Blowin’ in Our Wind:

How many nodes must a man walk down

Before you call him a man?

Yes, 'n' how many chips must he fabricate

Before using all the sand?

Yes, 'n' how many times must the molten tin fly

Before EUV is banned?

The answer, my friend, is blowin' in our wind,

The answer is blowin' in our wind.

The topic for the night took a closer look at just what was likely or necessary to keep walking down those nodes. There are of course no easy answers, and the panelists came with often starkly differing views.

One of the more telling moments of the evening was watching Freescale’s Will Conley laugh continuously as his friend Kevin Cummings so doggedly presented ASML’s company line on EUV. As Ausschnitt alludes to in his lyrics, there has been pretty much talk this week at the Advanced Lithography conference with regard to the “molten tin” flying through the EUV systems (some are referring to the exposure systems as deposition systems because of those sputtering tin targets).

But one of the key arguments that evening was not related to the technology and all the challenges that it faces, but the questionable economic models surrounding it. Ausschnitt stood up during the Q&A period to ask how ASML could have a cost-of-ownership (CoO) model showing EUV as the more cost-effective solution when they have no high-power source, no defect-free mask and no workable resist on which to base their model. It is understood that any cost model at this point is going to have to be based on a certain amount of assumptions. But what was disconcerting was Cummings’ answer to the question: Since they have to assume something, they assume the numbers that are needed to make EUV successful. That wasn’t a very comforting explanation in itself, but when pressed on that point, Cummings followed it up with, “We have a program in place that we believe in, and you’re the judge. Either we convince you or we don’t.”

How many screens must a man project

Before he can fill the sky?

Yes, 'n' how many jeers must one man get

Before he can hear hecklers cry?

Yes, 'n' how many snores will it take till he knows

That too many brain cells have died?

The answer, my friend, is blowin' in our wind,

The answer is blowin' in our wind.

But I don’t mean to just partake in EUV bashing here. Indeed, a lot of slides were projected Tuesday evening, and it’s not as if optical lithography faces a particularly rosy outlook itself. Double patterning, for example, is widely considered the route for extending the lifetime of immersion lithography, but it doesn’t present a very friendly cost model, particularly for chipmakers producing fewer wafers per mask, which Conley pointed out, being with one such chipmaker. In separate conversations, Conley has told me how much Freescale is taking its designs back to the drawing board to make them more litho-friendly in an effort to increase the process window without having to resort to double patterning. During the panel discussion, he said, “You can call it DFM, but it’s just survival.”

Tony Yen of TSMC labeled double patterning as a one-generation extension of immersion lithography. Although that is debatable, the EUV contingent has presented some compelling information on how they plan to take that technology all the way down to the 11 nm node.

Looking at high-index immersion lithography as a means of extending optical lithography also does not keep the hecklers at bay. Ben Eynon, Sematech’s associate director of lithography, presented a relatively balanced view of the challenges presented by both EUV and 193 nm immersion lithography. One important point he made about the high-index lens material (LuAG) needed for 1.70 NA imaging is that the lens material absorption is at about 0.07/cm, and needs to reach 0.01/cm. Those numbers are a little outdated, since Schott is talking this week about having achieved 0.05/cm, but especially given that the absorption target is now <0.005/cm, the concerns are still valid.

How many years can a speaker present

Before the same SPIE?

Yes, 'n' how many molds can he nanoimprint

Before one is defect free?

Yes, 'n' how many talks can a man suffer through,

Pretending he just doesn't see?

The answer, my friend, is blowin' in our wind,

The answer is blowin' in our wind.

There were several other excellent presentations given during the panel by Tim Brunner of IBM, Tatsuhiko Higashiki of Toshiba, Winfried Kaiser of Carl Zeiss, and Kurt Ronse of IMEC. And the main recurring theme was certainly that nothing is going to be easy. Although the panel was focused on projection lithography, Higashiki did a nice job of pulling in discussion of more of the options, including maskless lithography and nanoimprint. But, as he pointed out, they all have significant development challenges. For a company like Toshiba, competing to push NAND flash as quickly and as far as they can, they will grab hold of any technology that can promise a boost in resolution and continued scaling. But those solutions need to come sooner rather than later.

As Conley noted among his multiple quotations of Yogi Berra, “It ain’t over till it’s over.”

Meanwhile, there’s more to be said about nanoimprint. Molecular Imprints had a couple announcements this week — one on the release of the Imprio 300, with improvements in throughput and more, and another on Sematech’s purchase of an Imprio 300 (yes, the journalists around here were a bit annoyed that we weren’t told about the Sematech purchase during our briefings with Molecular Imprints and/or Sematech just one or two days before, but such is the nature of press release approval processes). Anyway, there’s more to be said about all of that, and the buzz around here at the show, so stay tuned…

Posted by Aaron Hand on February 28, 2008 | Comments (0)