The Quest for the Spin Transistor

Daktaklakpak writes: "Found this interesting article on the IEEE Spectrum. It details the different attempts to make transistors based on electronic spin. Apparently, this technology is related to the MRAM that we've been hearing so much about."

Re:Benefits

[Kamran]

If you read the article a few paragraphs in it says:

'Fast, rugged, and nonvolatile, MRAMs are expected to carve out a niche from the US $10.6-billion-a-year flash memory market.'

I think those are the concrete benefits such a technology will provide. Especially the speed.

Re:Benefits

[Angry White Guy]

From what I understand, faster switching..

Props and much bold text to your on-topic FP

Re:Zionist trickery

[yangsta]

what the fuck are you smoking? wtf does the green party have to do with jews? how dare you slander ralph nader and the green party. what we should do is bomb your tight little asshole.

Re:Zionist trickery

[geekoid]

That right, don't slander the man whose ego allowed Bush in office!

Multi-State Processors

[Angry White Guy]

If a whole computer platform could be developed using spintronics, we would be no longer be bound by a binary system. Even a trinary system would give itself a seven fold speed increase.
But that would require radical thinking and a complete redesign of the computer industry, which could take decades (plus a week for someone to port linux to it).

AWG

Re:Multi-State Processors

[Anonymous Coward]

Spintronics is not very promising whereas using photonic bandgap materials to build optical transistors is. See the latest SciAm for a nice article on these materials of the future computing.

Re:Multi-State Processors

[diddalick]

We are not bound to a binary system; the idea of a trinary computing has been around for a long time.
Think about it, a thing can essentially have (one of) three different types of charge: positive, negative and none, so that would give you a trinary system.

There was a slashdot story [slashdot.org] on this a while ago, in fact.

Spin as I know it...

[Joe 'Nova']

Article mentions ferromagnetics cross polarised as a switch, a memory cell.

If you cool Xenon to around 0k, it becomes an Einstein condensate-the atoms align and act as one.(Don't have link handy, search will prolly churn plenty) The idea is coherence, just as a laser aligns all it's photons polarizations. It sounds like they have learned how to do this on an atomic level. From what I also understand, domains(think quantized magnet 'particle') tends to degrade unless they are cooled/remain undisturbed.
It does sound like a neat idea, flip an electron without having to take it anywhere, then you don't need a conductor, only a 'resonating structure' to channel the effect somewhere.

Also mentioned is the fact it has no gain, too bad, everything we interface with needs amplification in order to operate. Even your retinas send cascades of electrons with only a single photon. If they can solve the gain problem, this would seem like one of those Moore's Law things, but I wonder how they hold up against stray magnetic/electric fields?

Also mentioned is the energy stored,(n*2+1)/2, which suspiciously sounds like the energy levels of the electrons, ignoring the spins. Even that could be used to store information, but it would certainly be a bugger to keep the electron from transferring the energy.

If they can come up with something equivalent to hi-temp superconductors for spins, I see alot of good coming, just not this week ;)

Re:Spin as I know it...

[esonik]

but with spin, you have to destroy its state to read it. you cannot just look at an electron and know it's spin, once you look and know what its spin was, it then changes to a random state. kinda wierd, but that's what nature

You are thinking about a single-electron-spin transistor, however the article is not about single-eletron effects. Second, spin doesn't necessarily change randomly: say you want to distinguish up/down (z) polarization (as opposed to -x/+x or -y/+y polarization). If your electron is in either up or down and you read its z-component (up/down) it get's projected on up/down, i.e. it is not changed. Of course you don't know its x and y components but you're not interested in those anyway. If you read x or y components (i.e. project to x or y eigenstates) then of course you lose the z-polarization, but you don't have/want to do that.

Re:Spin as I know it...

[Anonymous Coward]

i work in the building where the bose-einstein condensates were first made five years ago (and got the nobel prize in physics this year).
You or the building?

Damn...

[danwarne]

... from the subject line I thought this story was going to be about the quest for a product which would take a company's press release and translate the PR spin into something meaningful... ... and here I was thinking we might be onto something truly useful ;-)

Re:Damn...

[Alien54]

I thought this story was going to be about the quest for a product which would take a company's press release and translate the PR spin into something meaningful

Just remember that the entire universe is based on spin. Down to the smallest bit.

It would be horrible if the two varieties were somehow related. ;-)

Another Article

[Weedstock]

Here is another article [unisci.com] about electronic spin based computing. It clearly explains, toward the end, what will be the practical applications of those experiments.

Back to the Future

[firewort]

I can't have been the only one to think of Flux Capacitor, can I?

Possible Problems

[Renraku]

The only problems I can see with MRAM other than the usual early-generation stuff that happens with most electronics is that we'd have to redefine the way we use our computers. Sure, we'd have a lot more power, but all of those nasty memory resident virses would tend to have a bit more bite. Before, you could just boot with a disk, clean the HDD, and you'd be free. But now, it might be a bit harder. Also, the electronics needed to run the MRAM might be a bit wasteful in terms of space and power.

Re:Possible Problems

[steve_l]

well, it would give you notebooks with instant wakeup and zero static power consumption; that would be cool. Or You could rip out all your main memory server side and have a persistent in memory database.

It all depends on cost and capacity. Never underestimate the ability for existing technology to catch up by the time some advanced technology comes to market, or the willingness of DRAM vendors to lose billions of $ selling their product at a loss.

Magnetic fields in my computer?!?

[dizzydogg]

I can see it now, a month after they release the first computer running with this technology, somebody is going to write a virus that sets all the values in the pc to 1, creating a magnet and wipeing out my disks and monitor. I don't know about you guys, but considering that most of this technology requires a magnetic field to set up the electron spins makes me a little nervous about installing anything using this technology with conventional electronics, where magnetism causes some very strange problems. And i doubt we will be seing a chip using lasers to change the spins due to the fact that it would be rather tough to install millions of lasers on a chip, not to mention the power draw/heat problems it would cause.

Bad Pun

[FrankDrebin]

<pun> I guess all those PHD's mentioned in the article would be spin doctors... </pun>

A problem here...

[TWX_the_Linux_Zealot]

If something works off of spin, wouldn't we have to build a mirror image of one for use in Australia and New Zealand? I mean, toilets and sinks and drinking fountains spin the other direction down there, so wouldn't electrons too?

*smile*

Spindizzy

[snake_dad]

Has anyone read James Bliss' "Cities in Flight"? Great sf, if you remember what year it was written. The story is based on the discovery of some weird things you supposedly could do to electron spin, enabling complete cities to enter interstellar space, using a machine called a spindizzy. Oh well, off topic probably. It's just that the article reminded me of the book.