STORAGE LIMITS: Postponing The Inevitable.

It's going to happen: there is a finite limit to the areal density of magnetic media. Magnetic particles on the surface of hard disks can only be "so" small or packed "so" close together. Any smaller or closer and they'll interfere with one another, reversing adjacent magnets' polarities. Then, the patterns of ones and zeros they represent will be distorted or randomized, noise will overcome signal, and users will lose their data to it.

"It" is the superparamagnetic effect and postponing it was topic-A, once again, at DISKCON. Being a phenomenon of nature, it probably cannot be defeated; but it has been held off, and there is some evidence that its onset can be pushed farther away, according to the R&D professionals who presented their various approaches at this annual technical conference.

The issue is important because new applications for HDDs are about to emerge, many of which involve video rather than computer data--and video is notorious for consuming disk capacity. A home electronics company called TiVO is planning to bring out a product this year that's intended to replace the VCR: the difference being that it uses an internal HDD instead of removable tape media. This could give manufacturers a boost, although it will initially force them to concentrate production on inexpensive (and relatively quiet) drives. Eventually, however, consumer demand for greater storage will pull higher capacity HDDs into that market and manufacturers will have to be ready with them.

11.56bit/ [in.sup.2] Today

Today, the state of the art in areal density is in the 1012Gbit/[in.sup.2] range, although most HDDs offer--and need--far less. Seagate's huge 50GB model Barracuda 50 tops the 3.5-inch field in terms of capacity, but it does so with a mere 3.225Gbit/[in.sup.2] areal density over the 11 disks inside. IBM's model Travelstor 25GS squeezes 25.3GB into its five 2.5-inch disks with an areal density of only 8.774Gbit/[in.sup.2]. The highest areal density so far announced for a commercial drive is Toshiba's model MK64l4MAP. Targeted at the laptop/notebook market and due for release by year's end, its single 2.5-inch disk holds 6.4GB at a whopping ll.577Gbit/[in.sup.2]. But is that all that can be achieved?

It's "Physics"--Not "Engineering"

To date, it's Read-Rite's R&D lab that has demonstrated the highest areal density: 26.5Gbit/[in.sup.2]. Subrata Dey said his team did it by tinkering with the Giant Magneto-Resistive (GMR) head. What he called a "dual spin-valve design" helps to reduce the bit-error rate and a new combination of metallic "layers" increases the head's sensitivity. "The sensor design can be extended to 40Gbit/[in.sup.2]," he predicted.

At Seagate, the R&D lab has demonstrated 23.8Gbit/[in.sup.2], but, as vice president for advanced concepts, Nigel Macleod said of the achievement--and by extension, the entire areal density effort--"This is more of a physics problem than an engineering problem." Seagate's approach involves making the tiny magnetic grains as nearly uniform in size and shape as possible. But they're also working to boost the coercivity of the media (in effect, its "willingness" to have the grains' polarities re-oriented).

Richard Fontana of IBM's Storage Systems Division is basically optimistic about pushing back the onset of the superparamagnetic effect. "Magnetic recording areal densities are increasing at annual rates of 60100 percent," he said, "which implies a 300Gbit/[in.sup.2] recording density within five to ten years." (Earlier this year, IBM reported achieving 20.3Gbit/[in.sup.2]; and Fontana reported that they, too, are working on the "head" side of the equation--specifically with thin-film heads.)

He concluded, however, that "areal density increases result from scaling the entire recording system--not simply scaling the bit-cell [grain] dimensions--and a 30X increase in areal density implies a 5X increase in sensor [head] sensitivity and a 5X decrease in fly-height or head-disk spacing."

Fujitsu, which has demonstrated 20.4Gbit/[in.sup.2] areal density, is taking a similar approach. Akira Kakehi reported that their achievement was largely due to what he called a "stiction-free slider": a "padded" head assembly that can make contact with a more highly smoothed disk surface than what was previously employed.

Perhaps Chris Bajorek of Komag put the problem in the truest perspective. He declared that "current designs are extendible to areal densities well beyond 5OGbit/[in.sup.2]." He added: "a 10X extendibility beyond current areal densities looks just as difficult [today] as it did in 1980."