Nano research challenges storage limit


ARLINGTON, Va., Aug. 26 (UPI) -- The computer hard-drive industry might get an unexpected research boost from a study about how densely magnetic bits can be packed, which was debated Monday at a nanotechnology conference.

Most of the research at the Institute of Electrical and Electronic Engineers event, which focuses on developments in the science of manipulating matter at the atomic or molecular levels, look several years into the future.


The nanomagnetics work of Larry Bennett, a research faculty member at George Washington University's Ashburn, Va., campus, and Ed Della Torre of the National Institute of Standards and Technology's Metallurgy Division in Gaithersburg, Md., could affect today's hard-drive designers, however.

Current thinking among storage experts suggests that as individual magnetic storage sites, or domains, grow smaller, they become more susceptible to randomly changing at room temperature, which would destroy any information the domains hold.


This "superparamagnetic" effect limits how small domains can get, threatening to slow the current rapid growth in storage densities.

The equations for determining the superparamagnetic limit, however, fail to take into account several factors, including barriers to domain change in the storage material's chemical potential, Bennett said.

"There is a whole folklore here, running over the past 30 years, that we feel is wrong," Bennett told the conference's nanomagnetics session.

Theoretical models of how domains change disagree with some experimental evidence, lending credence to the idea that superparamagnetic equations could be too conservative in some cases, he said.

If true, the finding would give hard-drive designers a longer timeframe to pack more and more information into every square inch of their magnetic media materials before trying complicated technological approaches. For example, IBM has pioneered the idea of strengthening its disk-drive media's domains by inserting a metal layer only a few atoms thick, a challenging proposition for even the most advanced production methods.

More importantly for storage researchers, Bennett said, is the idea that correcting flaws in the equations might revive the possibilities of less-complex materials previously thought to be too temperature-sensitive for long-term information retention.

Bennett urged researchers in the nanomagnetics arena to send him other magnetic storage material samples for further experimentation.


Such confirming work is essential if such a key factor in hard-drive design is to be challenged, said Vitali Metlushko, an associate electrical engineering professor of at the University of Illinois at Chicago, who co-chaired the session.

"This is the beauty of (nanomagnetics)," Metlushko told United Press International after the session. "It's mutltidisciplinary, everybody brings knowledge from totally different fields; sometimes you get a new perspective that's been overlooked by the specialists."

One complication of Bennett and Della Torre's study is they focused on a cobalt/platinum structure with some layers far thinner than a nanometer, Metlushko told UPI. A nanometer is to an inch what an inch is to 400 miles. Not only is it a challenge to accurately control the thickness of such infinitesimal layers, Metlushko said, but the material must be defect-free and nearly perfectly smooth to preserve the structure's storage ability.

Other, longer-range research at the nanomagnetics session could pave the way for replacing today's electronic transistors with arrays of domains. Such an approach would remove the need for complicated interconnections between individual computing features in a processor, said Gyorgy Csaba, a researcher at Notre Dame University in Indiana.

"Nanopillars" of magnetic materials would serve as transistors, Csaba said, since domain changes could propagate through an array of pillars according to the rules of logic. The sorts of sensors currently used to read data from hard drives would record the results of the array's calculations, he said.


This approach could pack logic elements more tightly that today's processors, but can't match their speed; the current design would run at only about 100 megahertz, he said. That is about 10 times slower than average Pentium chips available in home computers. Speeding up the system would rely on shrinking the nanopillars, he said, but increasing the array size is also proving difficult, Csaba told the conference.

Latest Headlines


Follow Us