Stanford, IBM Team To Explore Spintronics

saxylife writes "NYTimes and various other media are carrying a story on the latest venture between IBM and Stanford," which will concentrate on spintronics, in other words, controlling "the magnetic orientation of atoms to store data. It's supposed to ease the pressure of hitting the barrier of Moore's law."

Comments

Magnetics

[(54)T-Dub]

For those of us who have never heard of spintronics here is a quick summation from the article:

Designing electronics based on how electrons spin instead of how they transmit electrical charges could, in theory, lead to far smaller devices with much lower power requirements and fewer problems with heat buildup. Unlike standard electronics, which represent the ones and zeros of digital information by manipulating voltage and current, spintronics uses magnetic fields to manipulate electronic spin into one of two states called up and down.

This sounds like a great idea to me. It also seems to me that there has been a lot of talk about using magnetics in data transmision (not storage) for a long time without any real results. It seems very promising considering that a magnetic field moves at the speed of light once it's been created.

One final interesing quote from the artice:

One area of concentration will be exploration of Dr. Zhang's research on spin currents. He has reported theoretical support for the concept that spin states can flow from electron to electron just as a charge does, but without generating the resistance that causes energy to be lost every time a charge moves from one transistor to another over a short copper interconnect.

Re:Magnetics

[fbform]

I remember there was similar research at Purdue University some months back. Here's the link and here's a pic.

Re:Magnetics

[wass]

It seems very promising considering that a magnetic field moves at the speed of light once it's been created.

Well, it's really the 'electromagnetic field' that can propogate at the speed of light in a vacuum (in the form of photons, which are of course the fundamental quanta of electromagnetic radiation.

Magnetic and electric fields are quite related, and only independent phenomena in time-independent processes (ie, electrostatics and magnetostatics). Namely, if you write Maxwell's Equations out and put all time derivatives to zero you really get separation of electric and magnetic fields. But for real systems, changes of one of these induces spatial variations of the other. So they're truly interconnected, and in fact they're most conveniently written in 4-dimensional form that describes special relativity perfectly.

Re:Magnetics

[wass]

What you are talking about is really just Quantum Field Theory explained from a different perspective, which actually has been done since at least Landau and Lifshitz (see their course on Theoretical Physics, volumes 2 and 4). Classical field theory is essentially electrodynamics (relativistic classical EM fields). Quantum Field Theory basically quantizes the field operator, but is difficult for a number of reasons. (Ie, in the 3+1 four-vector, the momentum is a standard operator but time is more of a parameter than operator, so one cannot merely generalize non-relativistic QM that easily. It involves going through alot of clever manipulation.)

And magnetism does exist, the magnetic and electric fields are really one and the same (in the proper 4-vector formalism). Magnetism can come from electron spin (explained very well in QFT) as well as moving charges (moving electron, for example). Spin has alot of quantum weirdness due to being angular momentum that's always 'just there' and discretized. But it's explained well enough w/ quantum field theory and group theory.

obligatory wikipedia...

[qrash]

Spintronics

http://en.wikipedia.org/wiki/Spintronics

Re:Honestly, folks.

[Anonymous Coward]

No, Moore's Law applies to transistor density. Transistor density depends on the smallest line we can draw on a microchip. Storage media sizes depend on the smallest line we can draw on a platter.

Platter density and transistor density are more closely related than you might think.

General information on spintronics

[leeum]

For those (like myself) who have little idea about spintronics, Wikipedia has a general article that seemed to explain it to me quite well. Of course, I'm not a physicist so I have no idea whether or not it's accurate although I'm tempted to find out more from the referenced article. PhysicsWeb has more of the same. Apparently this will have far-reaching implications on RAM and cable bandwidth.

Not quite ...

[DarkMan]

Although this may sound similar at the level of description given in the articles, don't let the journalists deep and impressive knowledge of this technology blind you.

The devices that are being talked about work in profoundly different ways to the old ST506 disks. Plus that fact that spintronics has been expanded to cover anyhting with magnets doesn't help clarification much.

For example, despite zdnets claims that IBM use GMR heads in their hard disks - that's not true, they are spin valves. These show a change in elecrical resistance in the prescence of a magnetic field - but no where near the magnitude of effect of a GMR device. That's fundementally different from the older method used in the read heads, which was to have a coil of wire, and detect the current induced in that coil.

If you can align the spin of electrons (do-able), then you can orient the spin, and thus have two independant channels within a single wire (horizontal and vertical, or whatever you want to call them). That's pretty novel.

Re:Will spin tunnel as well?

[Compuser]

Electrons carry both charge and spin. They can tunnel.
Spin is a property not a particle, hence your
question makes no sense (even RVB diehards who argue
for spin-charge separation in some materials will
assign spin to some quasiparticle, a "spinon", and
even in those cases tunneling is reserved for
electrons).
Your question is a bit like: "what does blue taste
like?"

This is Cool Stuff!

[ewhac]

I saw a presentation on spintronics given at WorldCon by Kevin Roche, who is one of the IBM researchers developing this stuff. He will be giving another presentation on it at -- of all places -- BayCon 2004.

I found his talk absolutely fascinating. He's basically created a "transistor" that allows through only electrons of a particular spin. Once you have an electric current composed of electrons spinning all the same way, you can do lots of unexpected things. One example: Light-emitting diodes emit polarized light! Even if you have only a cursory exposure to physics or chemistry, you'd probably enjoy his talk.

Schwab

Re:This is Cool Stuff!

[MustardMan]

He's basically created a "transistor" that allows through only electrons of a particular spin.
What, you mean, like, say... a magnetic field? *grin* Seriously though, for those who aren't really familiar with spin, theres a decent quick-and-dirty spin primer here, which includes a bit of details on the stern-gerlach experiment, which shows one way one might select electrons of only a certain spin.

Re:Will spin tunnel as well?

[wass]

Electrons can tunnel across a gate: can variables like spin do the same thing? If so, that's another barrier.

Yes, it's still the electrons tunneling across. And it's quite appropriate that you use the word 'barrier'.

There are spin tunnel junctions, where the electron tunnels through an insulator, and people are measuring how long the spin can be preserved if the electron tunnels into a standard metal. Ie, after enough scattering points the spin will be effectively randomized.

But yes, electrons are tunneling, and in some cases the spin of the electron (whether up or down) determines how well it will tunnel through the barrier. So spin is really another parameter that can be controlled to make spin-transistors or spin valves more dynamic than traditional transistors.

Re:Not News

[brarrr]

Not to be rude or intend to flame or anything, but spintronics has nothing to do with bubble memory. I'm doing a phd in spintronics under an advisor who focuses on magnetism, so i feel qualified in saying this. Bubble memory i don't know much about, other than it uses novel orientations and sizes of domains for magnetic recording.

spintronics, on the other hand, uses the charge and spin of electrons and holes in a similar method as electrons and holes are used in standard electronics. for example, the energy required to depopulate a channel in a transistor (turning it on or off) is far greater than the energy required to flip the spins of the charge carriers... so using that, you could have a smaller and lower energy transistor.

the limitation at the moment is in the materials, which is what we do... making them work at and above roomtemp for example.

if you be wanting to see a little more, check out our research page: http://depts.washington.edu/kkgroup/research/spint ron/index.html

MRAM

[anethema]

MRAM uses spintronics to store data. Its supposed to be very fast (dram speeds), dense, and not too expensive.

Oh did I mention non-volatile ?

This isnt some fancy technology thats going to maybe apear in ten years.

There are preliminary datasheets out now right here.

I cant wait to change my hdd over to this stuff (welll, that may be years away ;)