Major Breakthrough In Spintronics Research

Invisible Pink Unicorn writes "Spintronics is the field of research into developing devices that rely on electron spin rather than electron charge to carry information. A major advance has been made by the Naval Research Laboratory (NRL), where they have for the first time generated, modulated, and electrically detected a pure spin current in silicon. Progress in this field is expected to lead to devices which provide higher performance with lower power consumption and heat dissipation. Basic research efforts at NRL and elsewhere have shown that spin angular momentum, another fundamental property of the electron, can be used to store and process information in metal and semiconductor based devices. The article abstract is available from Applied Physics Letters."

Comments

I don't get it

[Quadraginta]

Now, the press release says the exciting thing about "spintronics" (ugh) is that " it frees one from the constraints of capacitive time constants and resistive voltage drops and heat buildup which accompany charge motion."

Well, fair enough; I can readily imagine that if you could get information to flow through a magical material without having to actually make electrons move, that would be great. No more of that pesky knocking into the lattice that they do which converts their motion into heat.

But...um...how exactly do you get a spin current without the electrons actually moving? I mean, given that the spins in question are nailed to the electron? Seems tricky. Like driving down the highway without having your car move...

Furthermore, if we read further down the abstract, we find this:

"NRL scientists first inject a spin polarized electrical current. . . .which generates a pure spin current flowing in the opposite direction. . ."

Sounds to me like the existence of their spin current depends on the existence of an old-fashioned charge current. So how's this help? How is this a "key enabling advance" (as the press release calls it), still less a "major breakthrough" as the /. article excitedly and credulously calls it?

Re:I don't get it

[silverpig]

Spins are transferred from electron to electron as the spins flip. Imagine a series of bar magnets. You can flip one magnet and it will affect the energy of the next one, and then the next one etc. The exact solution is difficult to calculate quite often, but in general, if you have a high population of spin up electrons localized in one area, the spins will tend to diffuse away from that via a few mechanisms:

1. The spin ups will turn to spin downs and cause nearby spin downs to turn spin up.
2. The spin up electrons will move to the right (just picking a random direction), and this will be compensated for by having spin down move left. The result is a net spin current with no net charge current.

To generate this, a spin polarized charge current is generally used. In this paper they used a ferromagnet contact as a source. The setup is basically a 3-way intersection.

Lead 1 is just a floating lead not connected to any ground.
Lead 2 is the ferromagnetic lead
Lead 3 is a ground connection

A voltage is applied between Lead 2 and Lead 3 causing an electrical current to flow. The electrons come out of the ferromagnet partially polarized. This current then goes into the ground Lead 3. All charge current flows from Lead 2 to Lead 3. However, the excess spin up electrons in the junction cause spins to diffuse down the floating Lead 1. No charge current flows down Lead 1 because it has nowhere to go. The result is a net spin current with no net charge current.

Re:I don't get it

[silverpig]

Spin-orbit can still be quite strong yes, but it is very dependant on the material. An interesting way to think about it is that when you have a standard piece of conducting material, it's not that there is no current flowing in it while it sits there not hooked up to a source; actually the electrons go all over the place inside the material. Current flows right to left, left to right, but it all balances out and there is no net current. Resistive heating only occurs when you have net charge current. In an ideal spintronics device you would have charged currents flowing just like in any other material, but there would be no NET charge current. The spin current can diffuse along your channel. Why is this better in terms of heat? I'd have to check, but I think you're on the right track with the magnetic dipole being much weaker than an electric monopole. 1/r^4 vs 1/r^2 IIRC. One of the major benefits occurs if you can pass a coherent spin current along a channel. This leads to the possibility of quantum computations involving spins.

Re:I don't get it

[counterfriction]

All particles have an associated spin, just as all particles have an associated net charge.
Spintronic devices make use of the spin of electrons. Whence, Spintronics.

Re:I don't get it

[brarrr]

well, it's 1am and i'm writing up my phd thesis draft in... spintronics... so i'll jump on this as best i can, having skimmed the article (but not the press release because really, what science comes from press releases)

the idea for spintronic devices is to use different device physics utilizing the spin of charge carriers vs just their charge. a common device is a GMR read head on hard drives - developed in '88, widespread now. the next step is to make transistors that use spin - this requires a new class of materials (GMR is a metal/macro structure effect), essentially making non-magnetic materials ferromagnetic is the goal. (personally i use ZnO, not Si, but the idea is similar). if you use a ferromagnetic semiconductor of some kind, then there is better charge transfer to other semiconductors vs a ferro metal to semiconductor... and then what you're looking for is a material that has a long spin polarization lifetime (time before the knocking around flips the spin and all of a sudden you have no polarization). so i *think* that they mean a spin current to be something that is 100% spin polarized (ie all spin up) - which means that if aligned with an applied magnetic field there will be minimal scattering therefor lower resistivity and lower heat/phonon interaction. vs. the case of a partially polarized or random spin up/down distribution where the available states in a material subjected to a field are only open to half of the free carriers (ie only spin up states are available because of the field, so only spin up electrons are efficient carriers). all this is very much so like GMR heads, obviously (well i suppose to me).

i've met the authors at conferences, and i'm sure they're less than thrilled with this being labeled a "major breakthrough" though i'm sure they like a bit of the attention. this is pretty cool and interesting stuff that they've done, but it isn't a breakthrough - just another piece of an extremely large and complex puzzle.