Will magnetic switching by light keep storage vendors spinning?
Circularly polarised lasers - the new magnetism
Hard drive storage is based on switching magnetic states by magnetism. Switching by laser light could be much faster and could be used on non-spinning media.
Laser light switching of a magnetic medium is described in a PhD thesis, Laser-Induced Femtosecond Magnetic Recording, written by Dr Daniel Stanciu, currently a researcher for Océ Technologies in The Netherlands. He and Dr Fredrik Hansteen discovered how to use light to reverse magnetic polarity in 2006. Stanciu developed the idea more as an intern at Seagate's Pittsburgh research centre in 2007.
The PhD thesis was published last year and can be downloaded here (pdf).
Front cover of Stanciu's thesis.
The background is that magnetic-induced switching of a magnetic field becomes progressively harder as the size of the magnetic field shrinks. We want to reduce the size of the magnetic area so as to increase the storage capacity of the medium, such as a hard disk drive. Current shipping hard disk drives (HDD) have a storage areal density at or in excess of 325Gbit/sq inch.
Post-perpendicular recording technologies such as discrete track recording, bit-patterned media and HAMR (Heat-Assisted Magnetic Recording) are set to take this to and perhaps beyond 1Tbit/sq inch.
Unless the write and read time per bit increases in speed as the areal density increases then the I/O rate of the disk will appear slower and slower relative to its capacity. Stanciu reckons that, in 2007, disk drives were operating at an internal data transfer rate of approximately 200 MB/sec, corresponding to a channel data rate of about 1.6 Gbit/s. He says that the writing time for a single bit is a little less than one nanosecond (billionth of a second) at that speed.
(For reference here a picosecond is one trillionth of a second. A femtosecond is one billionth of one millionth, or one quadrillionth of a second.)
Stanciu asserts that HDDs with terabit areal densities will need sub-nanosecond switching times; their data bits will need their magnetic polarities set at such speeds. How could it be done?
One way is to increase the strength of the applied magnetic field and this, he says, can get us into speeds measured in picoseconds. But the smallness of the bits requires magnetic material with a high anisotropy (directional differences in magnetic field strength).
The strength of the writing magnetic field must be high to overcome this and it must also be very narrow in focus or else contained some how so that neighbouring bit values are not affected. He cites a potential limit on magnetic switching that indicates it can never be faster than two picoseconds.
Lasers are already involved in data storage, with CDs, DVDs and Blu-ray drives using laser light as well as the developing holographic drives. The HAMR development HDD technology involves the use of a laser to heat a tiny area of a hard disk drive's recording surface which then has its magnetic polarity changed via a magnetic write head.
It had been thought that it was impossible for a magnetic field to be switched between north and south poles by light. What Stanciu discovered was that it was possible to switch magnetic polarity by using circularly polarised light.
The direction of magnetism is controlled by the light's helicity, left or right. The angular momentum of the photons sets the magnetic direction. What happens is that there is localised and very fast heating of the material's magnetic systems, which renders it susceptible to the magnetic field produced by the circularly polarised pulse of light.
It is possible to produce pulses of this type of light from so-called ultra-fast or femtosecond lasers, and Stanciu demonstrated sub-picosecond magnetic switching of ferrimagnetic rare-earth-transition-metal amorphous alloys - the type of material already used, he says, in magnetic media-based data storage devices.
Two problems were identified by Stanciu in 2007 preventing the use of this method in real-life data storage. Femotosecond lasers are large and focussing a circularly polarised beam of light at the sub-micron levels required caused the polarity to suffer and the magnetic switching to be prevented.
He says that during his Seagate internship he demonstrated the effect with picosecond lasers which are less expensive and smaller than femotosecond lasers. Also, the plasmon mirrors used to focus the circularly polarised light have been improved and can now focus the light at the sub-micron level required.
Stanciu reckons that a data storage device, a laser hard drive, using this technology could have an internal channel speed of 1Tbit/s.
For commercialisation to happen a data storage manufacturer must take this technology up and use it. Isn't flash memory a better bet? Stanciu reckons a flash drive would not be as fast, as they operate at 2-3Gbit/s and, the implication is, a laser hard drive could go faster than flash memory.
Stanciu said: "Initially, I think this technology can be integrated within an HDD. (It) will be easy and cheaper to do this since the new (coming) HDD based on Heat Assisted Magnetic Recording will have a laser integrated in the writing/reading head, and used for heating the magnetic disk (for higher storage density)."
"Otherwise I think that now with the all-optical switching available we can not only transfer information by light but also store. The fact that you can use light to switch the magnet might indeed give access to new storage techniques other than spinning platters of hard disk drives."
If the hard drive was replaced by static media then there is a great big problem.
As you don't have a read/write head hovering above a moving media surface then you need a read/write head moving over a static media surface. Moving such a head in two dimensions to cover the recording medium's surface area at the speed and accuracy required would be horribly difficult.
The alternative is to use a battery of read heads - perhaps enough of them so that you only move them in one dimension. Naturally the expense of the system increases rapidly as you do this.
Stancius thinks: "You might think of storage techniques such as PROBE (a lot of heads - plasmon antenna), a holographic storage or even a storage using instead of a moving head, a moving mirror!?"
The ultimate idea would be to have one read/write head per recording bit, which is pretty much the DataSlide idea (PowerPoint deck download). That would mean a recording surface oscillating under a reading/writing surface using a pizzo-electric effect.
Is there any interest inside Seagate in Stanciu's discovery? "There is a strong interest from the Seagate side. In fact I am currently writing a paper on this subject with Dr. J. Hohlfeld and Dr. A. Rebei from Seagate Research."
We are, Stanciu reckons, at least five years from commercialisation. It's more likely to be well in excess of five years and is not an at all dependable outcome. What we have here is terrifically interesting potential technology generating a lot of interest.
Stanciu said: "What is clear is that the interest in this field is growing fast; e.g. storage institutes are now working on all-optical switching." Who knows what might eventually emerge? ®