High-k poses threat to Moore's Law - News - lithography

Electronic News, August 5, 2002 by Jeff Chappell

Lithography has always been viewed as the potential roadblock to Moore's Law, but it may be dielectric materials that force it into the breakdown lane.

As R&D for the 90nm node gets into full swing, the industry is looking for the successor to silicon dioxide (Si[O.sub.2]), a material it has used for decades, and its sister material, silicon oxynitride. While integrating a low-k film with copper in interconnect structures has proved problematic, finding a suitable high-k dielectric for use in gate stacks, DRAM capacitors and other applications may prove twice the headache.

Like low-k dielectric films, they have been pushed out on industry roadmaps, even as everything else has been accelerated in recent years. And chipmakers will push the limits of existing technologies, trying to find ways to work around materials problems to keep on track with feature shrinks.

"It is potentially a showstopper," said Dean Freeman, a senior analyst with Gartner Dataquest. "But every time this happens, we find a work-around," he added, citing the adoption of copper with fluorinated silicon glass, as opposed to a low-k film.

DRAM makers will seek to make trenches longer and deeper, for example, said Scott Becker, VP of product management for FSI International Inc. Becker recently participated in a panel discussion on dielectric materials at Semicon West along with ASMI, Genus, ASML and Dataquest.

"I believe they are going to extend their current approaches as far as they can," Becker said, speaking of DRAM makers.

"Competitiveness being what it is, people are willing to extend what they have," agreed Tom Seidel, CTO of Genus Inc.

Seidel also suggested that metal electrodes might be introduced at the 70nm node as well. Genus is seeing a shift away from tantalum oxide toward substances such as aluminum and halfnium oxides.

But while one substance may have good mobility, its leakage may not prove any lower than silicon oxynitride, or vice versa. So far, no one substance seems to have the right combination of properties to make it a leading candidate, Chris Werkhoven, VP of strategic marketing for ASMI subsidiary ASM America, said: "Is there one answer? No there is not," he added.

And the very nature of the industry itself is proving problematic to solving these issues. The competitive nature of semiconductor manufacturing has made some customers loath to share their advanced R&D data with their suppliers about what may work and what might not. "If you could get a straight answer from our customers, I'd be very happy," Werkhoven said.

And yet to overcome the high-k dielectric problems, it is going to require collaboration on the part of chipmakers and their suppliers, the panelists concluded.