Bound bits could bring bigger disks

By Kimberly Patch, Technology Research News

Despite the constant speed and capacity improvements to magnetic computer disks, the basic technology behind them has stayed largely the same for the past 40 years.

In order for the capacity improvements to continue, the bits that store digital information must continue to shrink. The practice of shrinking bits, however, is expected to run up against the laws of physics within the next few years.

Although it is still too early to tell how these problems will be solved, researchers from IBM's Almaden Research Center have moved one approach a step forward by reading and writing to patterned media, which uses a single magnetic domain to represent a bit.

This is in contrast to the hundred or so magnetic grains, or domains that make up each bit on today's thin-film media. The average magnetic pole direction of its grains determines whether the bit represents a one or a zero. Below that many grains per bit, the signal to noise ratio is too low to be sure of the average magnetic pole direction of the bit. This is because the grains do not point in exactly the same direction, or even always hold their positions. The grains that change randomly or point less directly contribute noise to the signal from the majority of the bits pointing in the correct direction.

The problem with cramming more bits onto conventional media is once the grains of material shrink below certain size, larger numbers of grains become unstable and will flip their magnetic poles randomly due to changes in temperature.

This is one reason it's a bad idea to leave a floppy disk in a hot car -- over time enough heat will randomly flip enough grains that the bits will become unreadable.

Conventional wisdom says the most bits that can be crammed onto a square inch of disk space using today's thin film technology will be around 100 million, or 100 gigabits. This is five times the state-of-the-art 20 gigabits per inch, but disk space traditionally doubles every year, meaning grain size could come into play in as soon as three years. Today's grain sizes are about 15 nanometers, or the size of 150 carbon atoms in a row.

Patterned media isolates bits physically by, for instance, carving a ditch around a small square, which leaves a magnetic island. The grains in this small, isolated group are coupled magnetically and so change poles in lockstep, are very stable, and can function as a bit.

Because grain size is not relevant in patterned media, this type of media could store much more information before running into the grain size limitations facing today's technology.

Researchers from IBM's Almaden Research Center have demonstrated a patterned media with single domain islands 80 nanometers across, which translates to a density of 100 gigabits per square inch. "We've taken a magnetic film and used a focused ion beam to carve it up into small islands," said Bruce Terris, manager of exploratory magnetic media and phenomena at Almaden. "The islands are small enough that each island is a single domain. And then we used a magnetic recording head to read and write to those islands," he said.

Patterned media has its own challenges, however, and most of them involve changing the disk design that has worked well for 40 years.

For instance, the thin films on conventional disks are like blank slates that can be written on starting at any point. With patterned media, however, information must be written precisely on the islands that make up the bits. It's a more complicated engineering problem to write within the lines. "The writing will need to be more precise because you have to have the write... head synchronized to those islands," said Terris.

Another challenge is tracking. Bits line up around a disk, and the read head must stay on the correct track to read bits in the correct order. It will be more difficult to write and read on the narrower tracks of patterned media, said Terris. "You need a mechanical servo system that will keep the head on track in those dimensions," he said.

Although the researchers' 80-nanometer islands are only five times smaller than today's conventional bits, they are much more than five times narrower. This is because patterned bits are generally square, while conventional bits are 10 to 15 times wider than they are long, with the wide side perpendicular to the edge of the disk, making for a relatively wide track for the read and write heads to follow, said Terris.

The researchers read and wrote information to the islands using sensors that were in direct contact with the bits. In order to make this type of technology viable, they would have to use read and write heads that fly above the disk without touching it. To work with patterned media, the disk heads would have to fly closer to the disk than today's state-of-the-art media, said Terris. "The heights would probably be on the order of 10 nanometers or lower," in contrast to be 20 to 30 nanometer range of today's disks, he said.

Perhaps the biggest challenge, however, is making the islands small enough to make it worth it to invent a whole new technology around them.

In principle, it is possible to use lithography to make 12.5-nanometer islands, which would make a disk density of 1,000 gigabits, or one terrabit per square inch, said Terris. "But that is extremely challenging, and to do that in a manufacturing way is still a problem to be solved," he said.

The research is a step toward making patterned media practical, but the field "is in its infancy [and] there are a couple of really difficult hurdles," said Caroline Ross, an associate professor of material science at the Massachusetts Institute of Technology. "I think probably the major [IBM] contribution is they've shown they can read and write bits" onto the islands, she said.

The big challenge is making the patterned bits small enough; the second problem is you have to be able to do this cheaply over a very large area like the area of hard disk, said Ross. The lithography methods used in semiconductor manufacturing, for instance, won't make small enough structures, she said.

The ion beam method the IBM researchers used is only good for research purposes. "It's very slow and costly. So you need a way to pattern full disks with 50-nanometer features in a cost-effective manner, said Terris. According to both Terris and Ross, it's not yet apparent what could work effectively.

It is also a possibility that today's disk technology could be pushed further than conventional wisdom says is possible. "There will be a limit to conventional media but it's a question of where the limit really is," said Ross. "People originally thought maybe it was 10 gigabits per square inch and then they said 50 and now they say 100. It's always very hard to put numbers on those predictions because people seem to find a way around them," she said.

It will be at least three years before patterned media technology could be technically viable, said Terris.

Terris' research colleagues were Jens Lohau, Andreas Moser, Charles T. Rettner, and Margaret E. Best of IBM's Almaden Research Center. They published the research in the February 12, 2001 issue of Advanced Physics Letters. The research was funded by IBM, the Department of Energy (DOE) and the Defense Advanced Research Projects Agency (DARPA).

Timeline:   > 3 years
Funding:   Corporate, Government
TRN Categories:  Data Storage Technology
Story Type:   News
Related Elements:  Technical paper, "Writing and Reading Perpendicular Magnetic Recording Media Patterned by a Focused Ion Beam," Applied Physics Letters, February 12, 2001.


February 28, 2001

Page One

Robots learn soft touch

Evolution breeds cooperation

Chip promises brighter wearable displays

Light source brightens prospects for security

Bound bits could bring bigger disks


Research News Roundup
Research Watch blog

View from the High Ground Q&A
How It Works

RSS Feeds:
News  | Blog  | Books 

Ad links:
Buy an ad link


Ad links: Clear History

Buy an ad link

Home     Archive     Resources    Feeds     Offline Publications     Glossary
TRN Finder     Research Dir.    Events Dir.      Researchers     Bookshelf
   Contribute      Under Development     T-shirts etc.     Classifieds
Forum    Comments    Feedback     About TRN

© Copyright Technology Research News, LLC 2000-2006. All rights reserved.