is there a LIMIT to data STORAGE? - magnetic recording

Emedia Professional, May, 1999 by Hal Glatzer

the 100-year history of magnetic recording (a "Cliff Notes" version, anyway) was the subject of my previous column. After a century, I said in closing, "there is unsettling evidence that we are now reaching its technical limits. Or are we?"

Data is recorded magnetically by polarizing tiny spots--accretions of crystalline grains--in the media's recording layer. The smaller the spots, and/or the more densely they are packed into a given area (typically measured per square inch), the greater the storage capacity. For simplicity's sake, assume that one spot represents one data bit--a 1 or 0 depending on whether it's oriented toward a magnetic north or south pole. Those spots have to be large enough for a drive's head to recognize them as spots before reading their polarity.

But magnetism's a funny thing. If you bind a handful of bar magnets together for a while, their magnetic fields will so interact that the north and south polarities of at least some of them will be changed. So too, data spots packed too closely together will destabilize or reverse one another's polarity, causing the head to experience errors in distinguishing 1s from 0s.

MO to the rescue

Current state-of-the-art magnetic areal densities measure at roughly 10 to 12Gb/[in.sup.2], at room temperature. Conventional wisdom holds that the "superparamagnetic effect"--that tendency of adjacent magnets to interfere with one another--will kick in somewhere in the range of 20 to 40Gb/[in.sup.2]. Consequently, a lot of R&D money is going into overcoming this limitation. Current approaches include adapting magneto-optical (MO) recording techniques, rethinking the purely magnetic recording process itself, and moving toward some non-magnetic recording technology like phase-change.

The polarity of MO spots doesn't change at room temperature. Because the recording layer is a continuous magnetic material with no apparent crystalline grain structure, it takes high heat from a laser to "soften" a spot for the drive's electromagnetic head. MO materials, therefore, exhibit higher areal density than purely magnetic materials, so intensive efforts are underway to somehow merge the two technologies.

Quinta Corporation and TeraStor Corporation are leading the charge, though neither has taken products to market yet. Greatly simplified, Quinta's "optically-assisted Winchester" approach builds on hard disk know-how, miniaturizing the MO process with fiber optics and tiny mirrors. TeraStor's approach, called "near-field recording," refracts the laser's light through a second lens, which shrinks the diameter of the illuminated spot, and, likewise, the size of the recorded spot.

your spots are showing

But is pure magnetic recording really in trouble? David Lambeth, professor of electrical and computer engineering at Carnegie Mellon University and associate director of the university's National Science Foundation's Data Storage Systems Center, doesn't think so. "With today's media and perhaps closer head-to-disk spacing," he contends, "we should be able to achieve as much as 25Gb/[in.sup.2]. We could do this by getting the head even closer to the media, and making the grains in the data spots smaller, while cranking up their anisotropy energy." (That's tech-talk for making the spots' material more resistant to having their polarity reversed.)

Lambeth's research suggests that adjusting spots' "orientation" also may help overcome the superparamagnetic effect. If all the tiny magnetic grains can be aligned so they have the same orientation, and if the shape of the resulting spot is then kept as round as possible (i.e., not too irregular), then the noise in the readout signal will be lower. "This implies," he says, "that a smaller number of grains are needed, per spot, to have the same signal-to-noise ratio, and that a spot can be made smaller to achieve higher areal densities."

and then there's phase-change

Tom D. Milster, a professor of optical sciences at the University of Arizona, advocates phase-change recording as an alternative to both magnetic and MO. In the recording layer, the "phase" of a spot can be either crystalline (a good reflector of laser light) or amorphous (a poor reflector); the difference is read as a 1 or 0. If you have a CD-RW or DVD-RAM drive, you're employing phase-change already.

Adjacent phase-change spots do not affect one another; their phases can be set or changed only by laser heat, so they can be packed very tightly together. But for any given disk size, phase-change storage capacity is not significantly higher than the other technologies. To compete--certainly to replace magnetic drives--phase-change developers will have to achieve at least a four-fold improvement in areal density.

Some believe the best way to do this is to shorten the wavelength of the laser, from today's infrared (780nm) or red (650nm) wavelengths down to the blue light's 440nm range. "PC media sensitivity does not dramatically change with wavelength," Milster says, "so it should be a relatively easy transition to short wavelengths, and new phase-change systems should be backward-compatible with older ones." But until someone produces a commercially viable blue laser diode, phase-change won't be a viable option.