The Fine Print

As I rushed off to the Denver airport yesterday to catch my flight back home, the panel discussion on double exposure was still going on at the Immersion Symposium in Keystone, Colo. The discussion was running late, but a half hour just wasn’t enough time for such a compelling and complex topic, and the conversation was just getting interesting. I was sorry that I had booked my flight so tightly.

As a basic outline for the discussion, Intel’s Robert Bristol, symposium program chair, put forth a few questions for the panel, the first of which was whether finding/creating new double exposure materials was even worth the bother. At this point, the industry seems to be pretty firmly planted in the “yes” category. Nonetheless, it is a task that will be more than just a little difficult.

I’ll back up for a quick explanation. Double patterning, much talked about of late, is the next step in keeping optical lithography alive (in tandem with immersion lithography). Although not always so simple, the basic premise is to split a design into two looser designs and print them on two separate masks, thereby alleviating the pressure on k₁ and the process window. There are several different schemes, but the most basic of them involve exposing twice, developing twice, etching twice, etc. In contrast, double exposure offers a higher throughput and lower cost by only exposing twice and getting rid of the extra processing steps. Although this would certainly be preferred by chipmakers, most explanations of the technology involve a mention of a magical, elusive material that can be exposed twice for two different results — a material that has yet to be found.

Yesterday’s panel included perspectives from a nice cross section of the industry. Besides Intel’s Bristol as moderator, panelists included Geert Vandenberghe of European research center IMEC, Ralph Dammel of AZ Electronic Materials (hard at work on the materials), Hyun-Woo Kim of memory maker Samsung, Jos Benschop of litho toolmaker ASML, and Mordy Rothschild of the Massachusetts Institute of Technology (MIT, also hard at work on materials).

Two common themes at the symposium throughout the week were cost and timing. Although the timing on these solutions leaves much to be desired, the cost issue is just too big to ignore. “The key thing is cost,” Vandenberghe said in response to the “Why bother?” question. “If we really want to have cost of ownership improvement, we need to eliminate the intermediate etch process.” Progress is needed, he added, in alternative processes such as double development, image freezing and different solvent systems, among other ideas in the works.

Progress is needed quickly. Time is running out. As MIT’s Rothschild noted, if anybody has a great idea for a magic material right now, it’s already too late for 38 nm, and probably too late for 32 nm as well. “The technology really needs to come within the next two years.”

Samsung’s Kim agreed, saying that double exposure needs to be ready for mass production by the 2009-2010 timeframe. But research groups are lacking the resources to get the job done quickly enough — another common symposium theme. AZ’s Dammel gave a quick overview of various efforts that were going on in the industry, including a look again at a silicon-containing material that AZ is working on, and which was detailed earlier in the day by Munirathna Padmanaban. Although the material is looking promising, it’s not perfect, Dammel said, and may never get to a usable point. “This is going to be very, very difficult,” he noted. “The timing just might not be there.”

Double patterning is a realistic solution, and it’s just a matter of time before it goes into volume production, Kim said. Benschop of ASML said that, in fact, it’s the only solution at the industry’s disposal to keep the roadmap going. But the real question is how to make it cost-effective, Kim said. Double exposure is the most cost-effective solution, he said, so Samsung can’t help but pursue it.

With or without double exposure, the industry will no doubt use double patterning. And perhaps double exposure will find its way into manufacturing further down the road, helping to alleviate the pain of extra processing steps and reduced throughput. “If there’s no alternative, then people will use the complex technologies,” Rothschild said. “But simple always trumps complex.”

Many in the industry can’t help but be optimistic about ultimately finding that solution. Really, who can blame them? The brilliant minds of this industry have always found ways to keep the ticker going on optical lithography. But while I heard Dammel say, for example, how difficult it was going to be to develop the right materials in time, what Bristol took away from the comment was that it was “not necessarily impossible.”

As I mentioned, I had to rush off just as the Q&A portion was getting started. If anybody can fill us in on the concerns and ideas from the audience, please let us know here. Even if you weren’t at the symposium, I would love to hear your thoughts on double exposure.