Readable Nuclear Spins Advance Quantum Computing

eldavojohn writes, "A University of Utah researcher and his team of German colleagues have shown that it is possible, using electronics, to read data stored as nuclear 'spins'. The lead researcher in the experiment was Dr. Christoph Boehme and his team's letter is available via Nature Physics (at a cost of $18 unless you are a subscriber). This is looking to be a large advance in quantum computing because prior to this, measuring the number of spins of a single phosphorus nucleus was very difficult." From the article: "The researchers used a piece of silicon crystal about 300 microns thick — about three times the width of a human hair — less than 3 inches long and about one-tenth of an inch wide. The silicon crystal was doped with phosphorus atoms. Phosphorus atoms were embedded in silicon because too many phosphorus atoms too close together would interact with each other so much that they couldn't store information. The concept is that the nuclear spin from one atom of phosphorus would store one qubit of information. The scientists used lithography to print two gold electrical contacts onto the doped silicon. Then they placed an extremely thin layer of silicon dioxide — about two billionths of a meter thick — onto the silicon between the gold contacts. As a result, the device's surface had tiny spots where the spins of phosphorus atoms could be detected."

Comments

Readable Nuclear Spins Advance Quantum Computing

[Anonymous Coward]

Easy for you to say.

frist post!

[Anonymous Coward]

I wonder if this means we will have nuclear spin DRAM available soon?

Charmed, I'm sure.

[Tackhead]

Cue the strange jokes. I'm charmed. You're Bohred. And the cat is both.

Re:Charmed, I'm sure.

[Anonymous Coward]

*reaches for his pistol*

Re:Charmed, I'm sure.

[mikefitz]

Yo brah, Im sorry some people like important topics such as this one. The relation that this topic bares on our country is unanimous. Please post your personal views on an emo message board. Jim Boring approves of this message!

Re:Charmed, I'm sure.

[Anonymous Coward]

> *reaches for his pistol*

Unfortunately, it was in the box with the cat. The cat's state is dead, its momentum is zero, and its position is indeterminate along the inside walls of the box.

Re:Charmed, I'm sure.

[Iron Condor]

at a cost of $18 unless you are a subscriber

Or get it for free at arXiv...

Nuclear spins - that explains it

[commodoresloat]

That explains why the black hole they found was spinning so fast.

Re:Nuclear spins - that explains it

[Anonymous Coward]

Definitively, it spins so fast because it does so in very little time!

quantum this quantum that

[Anonymous Coward]

when we finally get one built, we'll realize that we spent vast amounts of time and resources into doing something that doesnt matter, and we will wish that we could go back and correct all our past mistakes.

and thats how quantum leap really started.

Re:quantum this quantum that

[Dunbal]

we spent vast amounts of time and resources into doing something that doesnt matter, and we will wish that we could go back and correct all our past mistakes.

      What, like fusion you mean? :)

They read the spin of 10,000, but can't read one

[Mobile Mineral]

Once they can, that will be news, especially if they get beyond a few qubits...

3x as wide as a human hair?

[r_jensen11]

The researchers used a piece of silicon crystal about 300 microns thick -- about three times the width of a human hair -- less than 3 inches long and about one-tenth of an inch wide.

3x the width of a human hair? Maybe, for you. But me, I use *insert name brand* Volumizing Shampoo! Now, my hair is 3x stronger, smoother, and thicker than before!

Er...

[Anonymous Coward]

The silicon crystal was doped with phosphorus atoms. ... The scientists used lithography to print two gold electrical contacts onto the doped silicon. Then they placed an extremely thin layer of silicon dioxide -- about two billionths of a meter thick -- onto the silicon between the gold contacts. As a result, the device's surface had tiny spots where the spins of phosphorus atoms could be detected.

The seek time for that device sounds horrible.

Ah!

[dkf]

So this is what spin doctors do all day!

Re:Ah!

[DittoBox]

The good folks on Capitol Hill aren't that smart. Sorry.

Spin Cycler?

[Doc Ruby]

What's the cheapest (and maybe smallest, lowest powered) device that can flip the spin of electrons? Even if lots of electrons (coulombs) at once. Flip them up and down, singly or en masse (pun intended). I know they're different machines; I want to know abuot the cheapest machine I could get. If civilians can even get them.

Re:Spin Cycler?

[Legendre]

Stern-Gerlach apparatus, circa 1920.

Re:Spin Cycler?

[Doc Ruby]

That's a pretty simple device. Does the magnetic field need to be very strong, and the duration of operation very long (eg. slow photons through small field in short space not OK) to flip the states?

Does it take more energy to flip one way or another? Is either flip direction endothermic (to the electron, not net to the device with its power consumption)?

I'd investigate a microscopic device etched in silicon for flipping throughput, but it looks like these devices are MRIs, which are still very large and power hungry. Is that the detector making it big? If I didn't need imaging, but just a flipper and detector, could I make a chip with fairly high current, like a deciampere or more?

Re:Spin Cycler?

[Legendre]

Yeah, it's a pretty simple setup. Tiny magnetic fields will do the job, and it's almost instantaneous. Now, you asked "Does it take more energy to flip one way or another? Is either flip direction endothermic?", well that is a QM phenomena. Basically the thing doesn't have a direction per-se until you do the experiment, and when you do it, it comes out 50-50. You will always need to put in some energy to the measurement. So, you wanted to flip the spin, that's how you'd do it cheap & easy. To build a quantum computer out of it is another story.

Re:Spin Cycler?

[JohnsonJohnson]

Strictly speaking the Stern-Gerlach experiment: passing a beam of electrons through a magnetic field and detecting the two resulting beams, doesn't flip the spin it sorts electrons based on their spin.

Controlling the spin of electrons is hard, if you work with an electron beam, perhaps one you filtered to contain only one spin orientation, you have to insulate the beam from the environment to make sure the spins don't interact and change orientation later on. Furthermore the electrons within the beam themselves will interact and a beam of purely one spin state will eventually contain both spin states unless you put energy in the system to keep your spin state energetically favorable, usually by passing the beam through a constant magnetic field, but that will deflect the beam since electrons are charged and that puts limits on how far the beam can travel before it hits something or intersects itself.

Working with electrons bound to atoms is a little easier, you don't have to worry about maintaining a magnetic field along the path of a beam since the electrons aren't going anywhere. On the other hand in a bulk material to have electrons which are free to change spin state they have to be unpaired and atoms or molecules with unpaired spins tend to be highly reactive. Thus they will tend to combine with other atoms or molecules to form combinations whose spin cannot be measured.

This leads to the most common technique of spin manipulation which controls the spins of atomic nuclei in bulk material. Because the electrons shield the nuclei they tend to remain in one spin state for a little while, in fact because the local environment of each nucleus in a bulk material is determined by the combination of a known external magnetic field and the local electron environment you can get information about molecular structure from NMR. To be precise though magnetic resonance techniques both electron and nuclear depend on the fact that in an external magnetic field there will be a slight population difference in spin states for a bulk material. The individual spins will still transition between states due to interactions with the environment but you can hold a large enough number of them in a particular state for long enough to be able to manipulate spins in the desired state for a little while before they decohere.

So to answer your question, the cheapest practical spin manipulation device is an NMR spectrometer. I'm having locating one for sale for cheap, there used to be a couple of companies selling 60MHz and tabletop permanent magnet NMRs for educational use but I can't find any of them now. You can build one yourself for on the order of $2,000, all you need is time, some soldering skills, a permanent magnet in a solenoid configuration, an oscilloscope which is probably the most expensive part, and a circuit diagram for 60MHz oscillator. Or you can use software to simulate NMR experiments.

The holy grail of spin manipulation of course is to trap and manipulate a single atom or molecule, or a small ensemble of such in an entangled state, which is of course what the research article reference above is about.

Re:Spin Cycler?

[Doc Ruby]

That is a fascinating explanation of the engineering made possible by the physics of spintronics. I used NMR spectrographs in undergrad organic chem lab 20 years ago - there should be plenty of them, especially after they all got rebranded as "MRI" tech to avoid the marketing poison of calling themselves "nuclear" :).

I wonder whether photons are more manageable, because they're organized along a line separated from each other, and don't directly interact (except transiently interfering at a single momentary locus), though all traveling at c. Or whether the sizes and energies are too small for current cheap equipment to manipulate.

What I'm looking for is a way to "charge" spins, of any particle, in large amounts of particles. Perhaps by randomly setting spins to either state, then separating them into populations of mostly one state (higher energy state) in the large majority of that filtered population. Then later "discharging" them, setting their state back to the lower energy state, collecting the discharged energy. The re/flipping (and separation) processes would have to be low energy consumption, at least not many times larger than the energy recoverable in the dis/charge cycle. Photons are even more interesting, if they're not much more difficult/inefficient, because the charged photons are already traveling fast for delivery to remote discharge.

Is this at all achievable (or immediately foreseeable) with current engineering? Convertible to chip scales that can process milli/deci/centiamperes of charged electrons, or some comparable spin-energy bearing amount of photons per second.

Re:Spin Cycler?

[JohnsonJohnson]

What I'm looking for is a way to "charge" spins...Then later "discharging them...

What you are describing is precisely and NMR experiment. You use radiofrequency pulses to set the spin state, you then turn off the pulses and listen, the spin system will relax to the ground state emitting a radiofrequency pulse. There's no particular reason to prefer one particle over another, any quantum system can emulate another of equivalent size, that's one of the reasons Feynman got excited about developing a quantum computer. So you should chose the particle that's easiest to work with, currently for spin physics that's atomic nuclei. There is active research on microscopic NMR. The bigger problem with quantum computing with NMR is that it is difficult to create entangled states, which are necessary for quantum computers, the largest I know of is using the carbons in an alanine molecule to make a 5 qubit computer.

Photons are a reasonable candidate for spin manipulation, although usually what is manipulated is the polarization of the photon not its spin. The problem is to change the spin or polarization of a photon you have to make it interact with some sort of matter and that makes life more difficult.

Getting Over the Hump

[Doc Ruby]

What's the net potential energy difference between the difference between the different spin states, if any? And what does the curve look like - is there a big hump between them, or a small hump relative to any energy difference? If it's a hump, is it a trough to flip the states back?

Re:Getting Over the Hump

[PhysicsPhil]

What's the net potential energy difference between the difference between the different spin states, if any? And what does the curve look like - is there a big hump between them, or a small hump relative to any energy difference? If it's a hump, is it a trough to flip the states back?

I had to pull out my quantum mechanics book for this one. As a rough estimate for the energy difference between "up" and "down" spins, you can use the energy of the Zeeman effect (energy level splitting in an atom when in a magnetic field). The magnitude of that effect is (B/2.4e9 gauss) * 13.6 eV, where B is the size of the applied magnetic field. A supermagnet would produce fields on the order of 1e5 gauss, so we're not talking very much energy here. As another very crude estimate, consider that random thermal effects have enough energy to flip spins randomly, which is one of the big problems facing spintronics.

As to the humping issue, this is quantum mechanics; there is no curve. Only discrete states are allowed, with nothing in between.

Re:Getting Over the Hump

[Doc Ruby]

Thanks for the very specific answer.

By "hump" I meant the amount of energy required to push the spin from one to another state - I understand that "quantum" mechanics don't change in continuous functions.

So I suppose the Zeeman effect is that energy consumed by keeping the magnetic field at strength while the spin is changed. Does your book (or other source) indicate whether the spin flips back when the magfield is removed? Does it emit the energy difference? Or does an external stimulus have to force the flip back?

Is there an online discussion of the actual mechanics of these quantum changes? Not necessarily the quantized amounts in the grad-level math in which I'm mostly illiterate. Something accessible to a 1980s college physics minor like me, but not all stoned like _The Tao of Physics_.

What I really wonder is whether a chip like the one mentioned in the story that we're discussing could be used to store a significant amount of energy in the spins of amperes of electrons. Which could be released for use later by another chip. Which would consume less energy in the total spintronic dis/charge cycle than could be stored. And whether this works any better with electrons, and how frequency relates to the manageability of these phenomena.

And if not a chip, then in a larger machine, though I suspect better efficiencies are to be had at the microscale. Or maybe at the nanoscale, though these electrons are on the 10E15 femtoscale, just a couple notches lower.

Re:Getting Over the Hump

[slughead]

What's the net potential energy difference between the difference between the different spin states, if any? And what does the curve look like - is there a big hump between them, or a small hump relative to any energy difference? If it's a hump, is it a trough to flip the states back?

I had to pull out my quantum mechanics book for this one. As a rough estimate for the energy difference between "up" and "down" spins, you can use the energy of the Zeeman effect (energy level splitting in an atom when in a magnetic field). The magnitude of that effect is (B/2.4e9 gauss) * 13.6 eV, where B is the size of the applied magnetic field. A supermagnet would produce fields on the order of 1e5 gauss, so we're not talking very much energy here. As another very crude estimate, consider that random thermal effects have enough energy to flip spins randomly, which is one of the big problems facing spintronics.

As to the humping issue, this is quantum mechanics; there is no curve. Only discrete states are allowed, with nothing in between

Uheheheh.... you said 'humping.'

Re:Getting Over the Hump

[wass]

I'm too lazy to look up values or calculate it, but when using the magnitude of Zeeman splitting, you'd need to use nuclear magneton, not the Bohr magneton (as the textbook for Zeeman splitting almost certainly did), since this effect is looking at nuclear spins. Since the proton is about 2000 times as massive as the electron, the nuclear magneton is about 2000 times smaller.

Re:Getting Over the Hump

[wass]

So I suppose the Zeeman effect is that energy consumed by keeping the magnetic field at strength while the spin is changed. Does your book (or other source) indicate whether the spin flips back when the magfield is removed? Does it emit the energy difference? Or does an external stimulus have to force the flip back?

Zeeman effect is the splitting of degenerate 'atomic' states in the presence of a magnetic field. Atomic really means here hydrogen atom, because that's the only exactly solvable model.

What this means is that there would be a number of discrete states that an electron in a hydrogen atom can be in, none can be the same due to Pauli exclusion. However, many of these have the same energy (called degenerate states). The degeneracy is split (ie, the energy levels are slightly changed) due to an applied magnetic field. When you consider other effects like the spin-orbit interaction (the electron's motion around the nucleus makes it seem like the charged nucleus is moving around it, creating a magnetic field which interacts with the electron's spin), and also to a lesser extent spin-spin interactions (magnetic dipole interaction between nucleus and electron) they also slightly split those degenerate energy levels.

Anyway, the electron can be in a number of different discrete states. Applying a field changes those discrete states that the electron can be in. If you knew for certain the electron was in one state (eg, as per a measurement), and then applied or removed a field such that the state the electron was in is now not a valid state, the electron will be in a linear combination of the other now-valid states.

Finally, I'm too lazy to punch numbers, but I suspect the grandparent's post is incorrect in that when calculating the magnitude of Zeeman splitting, it assumed an electron in a magnetic field when in this case it's really a proton in the magnetic field (different magnetic moment, by a factor of about 2000).

Re:Getting Over the Hump

[PhysicsPhil]

So I suppose the Zeeman effect is that energy consumed by keeping the magnetic field at strength while the spin is changed. Does your book (or other source) indicate whether the spin flips back when the magfield is removed? Does it emit the energy difference? Or does an external stimulus have to force the flip back?

The energy is a one-shot deal, only required to flip the spin. In principle, the magnetic field doesn't need to stay on after the spins are flipped. In practice, once the field is removed, thermal effects take over and the spin population relaxes towards a random distribution again. Energy would be reemitted as photons for transitions to a lower energy level.

I must correct my previous posting. As a responder pointed out, we are talking about nuclear spins rather than electronic. The energy required to flip a spin will be lower, but the nuclear spins do not couple as strongly to the outside world. The lifetime of a spin polarized system could be substantial because thermal interactions with other stuff is more attenuated.

As for useful textbooks, I'm afraid I'm out of ideas. Dealing with spin is a higher-level concept; I didn't touch it until my third term of QM. As such, it's tough to find a book that won't have a lot of math along the way.

Re:Getting Over the Hump

[Doc Ruby]

Well, maybe if we kept up this thread, someone could edit it into _Spintronics for Dummies_ ;).

So I think you're saying that a calibrated device could require energy transduced first into a magnetic field of strength to flip the spin of an electron, operating just long enough to probably flip the spin. A much lower strength than the nuclear formula you cited - do you have the formula for the energy to flip an electron's spin?

The electron spin, much more decoupled from the kinetics that transfer thermal energy (random kinetic energy in particles) among nuceli, can remain persistent after being flipped. To flip them back requires the same device. If the device acts on many electrons, not requiring any specific electron's state to be read to flip between the lower/higher energy states, does the energy for flipping have to be much larger than the energy difference? Is there a way to flip the spins "deterministically" of a large electron population to one selected state of the two (ie. almost certainly flip a large majority of the spins), then later to the other state (again, probabilistically)? Maybe by flipping them randomly, but in a roughly (probabilistically) expected quantity of each, then somehow separating them into two populations mostly in on or another state.

If the magnetic field is a permanent magnet, then the prior spin state will resist the magnet flipping it with a small force equal to the force generated by the energy input to the magnet by an augmenting electrical field (or rotation, etc). Or maybe an apparatus like the phosphorus doped device in the story we're discussing (tangentially ;). Could one of these techniques back up a device that could take "wild" electrons by the ampere (or deci, or centi, or even milli), and output coulombs of "charged" electrons in the higher spin (and maybe about the same amount of "discharged" spin electrons separated into a separate circuit)? Could a complementary device accept the charged spin electrons, discharge them, collect the energy from their spins into another medium (like an opposing magnetic field against the device's magfield, driving a current)?

And would all these devices be more efficient in large quantity, and maybe easier to make, if we were dis/charging low-mass photons rather than heavier electrons?

Schottky Diode?

[dduardo]

It sounds to me like this is some type of Schottky diode. The phosphorus is the N-type semiconductor that is doped in the silicon and you've got metal (gold) contacts.

My question is, since the principle behind Schottky diode is the use of quantum tunneling, does this technique rely on how spin affects tunneling? If so, in what way?

round round baby?

[Edial]

What can we do with this new devolopment? Is there anything we can finally do? Will this change our views on certain things? enlighten me please!

Re:round round baby?

[meregistered]

Hmmmmm

To the best of my knowledge storing data as spin, therefore creating transistors the size of atoms* will, at the very least, bypass the limitations of the current transistors measured in nanometers. A Nanometer is 10 to the -9th power of a meter**. An atom is approximately 10 to the -11th power of a meter***. Therefore this technology, when fully functional would theoretically allow two orders of magnitude greater number of transistors per area of measurement.
So if a Pentium IV has approximately 42million transistors**** it could (in theory) contain 42,000,000 to the 2nd power more transistors.
Accept the increase is far greater than this because the P IV die process is 0.18 microns which is 180 nanometers (if I'm correct). So the actual increase in available transistors per area of measurement would be more on the order of 42,000,000 to the 5th power: 5,489,031,744,000,000 transistors (well atoms).

Now add to that the current problems with heat. I would expect (although I most definitely do not remember/know the laws of thermodynamics well enough to do more than vague speculation) that the amount of heat created by such a quantum system would be impressively small compared to the current system... although I would conjecture there are limitations to speed when measuring and changing spin... this would hugely increase the ability to clock the processor higher (an over abundance of heat is the primary limiting factor in clocking the processor system higher).

Wow, so now I am looking forward to having my conjecture ripped to pieces by those who actually know :D.

I hope that's at least a little helpful

*(although I think of spin being associated with quarks, a much smaller, sub-atomic particle... obviously a hole in my knowledge)

**Nanometer: http://whatis.techtarget.com/definition/0,,sid9_gc i514407,00.html

***Atom, Size of: http://trshare.triumf.ca/~safety/EHS/rpt/rpt_1/nod e7.html

**** http://72.14.203.104/search?q=cache:foWPHOKFqoMJ:w ww.soc.staffs.ac.uk/mss1/hsn/hsn-lect9.ppt+transis tors+in+a+Pentium+IV&hl=en&gl=us&ct=clnk&cd=2&clie nt=firefox-a

Re:round round baby?

[zevans]

Interesting, but i think you missed the point. This is quantum tech, not just nanotech.

A byte's worth of qubits can be in all 256 states *at once*, so as well as the 2d density of info you describe, you have to factor in that the thing is in many universes as a multiplier for the bit density it has merely in this one...

*Feynman fans read 'history' for 'universe'

Re:round round baby?

[wass]

Storing data as spin has a far more interesting effect, and that is it's a specifically quantum mechanical value, meaning that it can be in a superposition of the two states, unlike a classical bit, and more importantly can be operated on and entangled with other qubits to allow for quantum computation. That's the whole quest for this research, to make workable quantum computers, which means at this point having workable qubits capable of storing a value for a 'reasonable' length of time.

Re:round round baby?

[AnwerB]

A Nanometer is 10 to the -9th power of a meter**. An atom is approximately 10 to the -11th power of a meter***. Therefore this technology, when fully functional would theoretically allow two orders of magnitude greater number of transistors per area of measurement.
So if a Pentium IV has approximately 42million transistors**** it could (in theory) contain 42,000,000 to the 2nd power more transistors.

A nm is 1e-9, while an atom is nominally 1e-11 (different atoms have different sized nuclei and numbers of shells). This is a linear (1-D) measurement, so you can fit 100 atoms into a nm if you are going in a line, and 100*100=10,000 if you are talking about 1 nm^2. Also, remember that this resulting number is quite approximate and arbitrary, since it doesn't take atomic structure (how the atoms are arranged), among other issues, into account.

So, I think the number you are looking for is not NumberOfTransistors^2, but rather NumberOfTransistors*(100*100).

Stick to a standard

[ndogg]

Please, stick to a standard when writing your articles--preferably metric. I want none of this crap of switching between the metric and English.

Re:Stick to a standard

[Anonymous Coward]

We gave up using imperial measures for scientific work many years ago. Even my sixty year old engineering lecturer scolded me when I used them, and I only did that because most of my experience so far has come from working on classic cars.

I suspect we only cling to them in day to day life because of their convenient size and divisors: a pint is a good measure of beer, and a quarter pound of ham will make a generous round of sandwiches. One has to get one's priorities in order.

Re:Stick to a standard

[Anonymous Coward]

We gave up using imperial measures for scientific work many years ago.

Not in all fields. We're still waiting for the old stuck in the mud hydrologists to die, so the next generation can switch to metric.

Re:Stick to a standard

[Anonymous Coward]

america needs to ditch english in favor of metric like the rest of the world and fuck all those too dumb or lazy to adjust.

and we should all syncronize our watch to universal time and just deal with it...same time all over the world makes things more efficient in the long run.

current time: 0153

Re:Stick to a standard

[radtea]

I suspect we only cling to them in day to day life because of their convenient size and divisors: a pint is a good measure of beer, and a quarter pound of ham will make a generous round of sandwiches.

A litre is an even more convenient size for a beer! And a kilos can be treated as pounds times two, so 150 grams of hame will do not too badly, 200 is lots and 250 is really rather much.

What is convenient is what you know, what you're used to. Metric measures are just as convenient as Imperial ones if you know how to use them, and FAR more convenient if you've never been able to remember how many ounces are in a pound, or a pint, or whatever.

Re:Stick to a standard

[Archibald Buttle]

The measures you're talking about are usually referred to as "Imperial", rather than English.

In England there's only three things I can think of where we stick with Imperial measures: beer, milk, and driving distances. There is also a tendency for individuals here to measure their own weight in stones (14 lb to a stone), however the medical profession uses kg. Besides those things we're completely metric. Indeed, it's illegal to sell groceries in England in pounds and ounces, and petrol (gasoline) is legally required to be sold in litres.

The USA is dominated *far* more by imperial measures than England. It should also be noted that when it comes to volume measures, US pints, quarts, and gallons are all smaller measures to their English equivalents.

Oh - and Universal Time - that's essentially English - it's otherwise known as Greenwich Mean Time (GMT). :-)

Re:Stick to a standard

[Paua Fritter]

We gave up using imperial measures for scientific work many megaseconds ago. Even my 1.89 gigasecond old engineering lecturer scolded me when I used them, and I only did that because most of my experience so far has come from working on classic cars.

There, fixed that for you

Re:Stick to a standard

[GospelHead821]

It should also be noted that when it comes to volume measures, US pints, quarts, and gallons are all smaller measures to their English equivalents.

So very true. This is why my heart leaps for joy whenever I find a bar that provides beer by the Imperial pint. That is really the perfect size for a glass of beer. Unfortunately, it doesn't convert so nicely to metric. Who wants to order 5.7 deciliters of lager?

Spins

[inKubus]

Actually, TFS is incorrect; they only measured the "net spin" of millions of phosphorous atoms. According to TFA, they took a measurement of the hair thing at room temperature (where the spins are pretty evenly 50-50), then they measured it at liquid helium cold (where spins are "down") and when heated by microwaves (where spins are "up"). It's important to note that "spin" really refers to the electrons. I'm guessing that in a nice silicon matrix the "spin" affects the surrounding silicon either making it more or less conductive around the phosphorous. They don't really get into what "spin" is, so you think they are actually talking about a spinning ball or something which couldn't be further from the facts. Since electrons are like photons and they are waves at small scales, it's more about these little probability eddies or whirlpools where the electrons hang out more. There's a wikipedia article that explains the concept, they say "spin angular momentum cannot be associated with rotation but instead refers only to the presence of angular momentum." So like I said, it appears the particles are affected by angular momentum (statistically), but they are not actually "spinning" because there's no such thing that that scale.

Re:Spins

[inKubus]

And a link to the wikipedia article, since I forgot to insert it before submitting ;)

Re:Spins

[halftrack]

Actually, according to the fine article, they measured the net spin of ~10,000 phosphorous atoms. Previously it was only possible to measure the net spin of billions of phosphorous atoms. So it's a big improvement (but still not single atoms.) Furthermore, spin isn't a property of electrons only. The article mentions that they are trying to measure the atom nuclei's spin indirectly by measuring the electron spin, since these are connected.

Re:Spins

[klaun]

It's important to note that "spin" really refers to the electrons.

I'm not sure why you say this. It is clear from the article that they are talking about nuclear spin, not electron spin. In fact the basis of the experiment seems to be the nuclear magnetic moment... it is definitely not an electron artifact... There is a quote from the researcher (an assistant professor of physics) refering to nuclear spin in the second paragraph of the article.

Your explanation of spin is a little off the mark as well... You seem to be confusing a couple of different concepts. I'd say spin is really a quanity associated with a particle more than anything else... much like charge. And it is not on off. One of the main divisions of particles is into fermions and bosons based on whether there spin is an non-integer or integer, and thus whether they obey the exclusion principle or not. The wave/particle nature of electrons is only tenously connected to the concept of spin, in as much as they are both quantum mechanical concepts.

Re:Spins

[dido]

Actually, if their goal really is quantum computing, the spin of atomic nuclei would have been a better bet than electron spin. The atomic nucleus is one of the best-isolated quantum mechanical systems known, and it usually takes a very long time for decoherence and similar behavior to destroy any superposition of spin states that you create (an absolute necessity for any type of useful quantum computation). The most promising methods to date (such as this and this) for quantum computation are largely based on nuclear spin. Electron spin superpositions are not only hard to create but they succumb to decoherence very quickly, making them difficult to use.

Since electrons are like photons and they are waves at small scales, it's more about these little probability eddies or whirlpools where the electrons hang out more.

Now, while even people who are experts in the field of quantum mechanics are at a loss to explain what the spin of a quantum particle actually is, this explanation you give is about as far from the phenomenology of spin as it gets. It has nothing to do whatever with "probability eddies" where "electrons hang out more", but is an intrinsic property of the electron or whatever other quantum particle (note that it isn't only electrons that have this property, as you seem to think: almost everything appears to have it). If you put a charged particle such as an electron inside a nonuniform magnetic field (e.g. inside a Stern-Gerlach apparatus), for example, it will have only two possible values for its magnetic moment thanks to this intrinsic spin, and hence the electron would be deflected by the field in only two possible directions.

Re:Spins

[wass]

Umm, the latter part of your post is a bunch of incoherent rambling that may sound profound but is not really saying anything useful (like the alien as the professor in the episode of American Dad last night).

Since electrons are like photons

Okay, I'll just stop you there, electrons are NOTHING like photons, those two couldn't be further apart. Phonons are massless chargeless relativistic spin-1 bosons, electrons are massive spin-1/2 fermions with charge -e. Big HUGE difference, in every quantum effect. They are of course related in that the electrostatic force between two electrons is mediated by exchange of photons, but they are so entirely differnet. Massive electrons versus massless photons (big implications of chemical potential when considering thermodynamic quantities). Boson vs fermion (HUGE implications, photons aren't subject to Pauli exclusion and can be in a collective ground state, not true with electrons, electrons have a fermi surface at reasonably-cold temperatures leading to concepts like metals, semiconductors, insulators). Also photons are relativistic, with kinetic energy proportional to momentum, non-relativistic electrons have kinetic energy proportional to momentum squared. All of these give these two particles major MAJOR differences in the way they act, and how their collective properties occur.

Spin is what's called 'intrinsic angular momentum', and it's present in many subatomic particles like electrons and photons. There's an angular momentum that is 'just there', you cannot get rid of it. Electrons are spin-1/2, which means they can have a value of spin either +hbar/2 or -hbar/2, no middle value. It's unlike classical mechanics because a spinning top can go clockwise or counterwise, with a magnitude of angular momentum from 0 to infinity. Not true in quantum mechanics where angular momentum is quantized in units of hbar. Electrons only have two allowable spin values (protons too, as they are also spin-1/2 fermions). Those two values are why electrons, or nuclei, make good candidates for quantum computers. Photons are spin-1, so they can have spin +hbar, 0, or -hbar. (And actually for complicated relativistic reasons they cannot have value 0.) The two states +hbar and -hbar would correspond to clockwise and counterclockwise circular polarization of the photon, which you can write as a linear combination of horizontal and vertical linear polarization.

Re:Spins

[inKubus]

Anyway, you have an electron rotating a nuclei, not a free electron. So the spin has a different effect because the electron is locked in the orbital (and maybe even the orbital of the surrounding silicon). It doesn't just fly off in one of two directions. It's going around in a circle. So, the electron has it's spin (which is not really spin, but some sort of momentum without rotation) and it's also orbiting in a cloud of probability. At quantum level and timespace, it's pretty much a solid sheet of varying thickness because the spin (angular momentum, not really spin) causes the path it takes to sort of "spiral". So I visualize this little electron by the path it makes say one orbit and I see that it is not just a big circle but really it's constantly trying to dive one way or another but the orbital force attracting it fights back. Now, they are talking about using electricity in the article to measure the spin. That implies a flow of electrons so you know electrons are involved somewhere. I'm saying the electrons are flowing through the silicon and when they encounter a sort of thinner area of the silicon where it stretches around the phosphorus (like a tennis ball in a sock, or something), they are regulated somehow by the underlying spin of the phosphorus, which they think they can change by cooling or heating. I'm not sure why that is because the spin (which is not really spin) is just, as you say, "there". But anyway, the electrons are somehow regulated at this site by the spin. I THINK that the spin of the phosphorus somehow affects physically the silicon that's stretched around it and that's what allows it to be measurable. If that's the case, it's a simple matter to isolate a single phosphorus and make a quantum transistor. But of course, the real goal is getting a bunch of them working together to somehow just magically "have" the answer because we're asking for it, which is what quantum computing is. It just is. Anyway, I'll let you get back to your textbook now, I'm just someone trying to make a picture in my mind of what's going on. I'm not really interested in proving how smart I am or regurgitating the wikipedia article I already linked above..

Science could be much better if people were more casual from time to time. That's where ideas come from. Not memorizing long words and terminology. Where is Feynman when you need him? Oh wait, he invented all this shit. Haven't you ever heard of wobble?

Rimshot

[I Like Pudding]

Annoyingly, they don't know how fast the memory is

Re:Rimshot

[zevans]

They built the first one and measured the speed exactly... but afterwards they could no longer find it...

Free link to the full-text publication

[Anonymous Coward]

http://arxiv.org/format/quant-ph/0607178

Re:Free link to the full-text publication

[Anonymous Coward]

I'll second that. There are small differences in wording between the arxiv preprint and the actual article, but they don't seem to be significant. Here is the abstract from the Nature article to let slashdotters get a feel for the scale of the differences.

In recent years, a variety of solid-state qubits has been realized, including quantum dots1, 2, superconducting tunnel junctions3, 4 and point defects5, 6. Owing to its potential compatibility with existing microelectronics, the proposal by Kane7, 8--on the basis of phosphorus donors in silicon--has been pursued intensively9, 10, 11. A key issue of this concept is the readout of the 31P quantum state. Electrical measurements of magnetic resonance have been carried out on single spins12, 13, but the statistical nature of these experiments based on random-telegraph-noise measurements has impeded the readout of single spin states. Here, we demonstrate the measurement of the spin state of 31P donor electrons in silicon and the observation of Rabi flops by purely electric means, that is by coherent manipulation of spin-dependent charge-carrier recombination between the 31P donor and paramagnetic localized states at the Si/SiO2 interface. The electron spin information is shown to be coupled through the hyperfine interaction to the 31P nucleus, suggesting that recombination-based readout of nuclear spins is feasible.

Re:They read the spin of 10,000, but can't read on

[badboy_tw2002]

Puh, you call that jaded? You may not care about several order of magnitude improvement over previous attempts, but _I_ don't want to hear anything until I have a quantum computer wrist watch. Wake me in 2050!

haven't (can't, currently) read tfa, but...

[StandardDeviant]

how is this really news? Every approach to qc that I'm aware of uses spin setting/reading (via NMR in every case that's coming to mind). Bringing this back to the g33k/slashdot crew, check out the work done around 2001 to implement Shor's Algorithm at IBM (by Vandersypen et. al.) The wikipedia summary is a bit dense, but the original paper (cryptome appears to have a mirror) is a bit better.

(NB: I'm far from being an expert in this field, it's just something I was interested in a while back when I was wrapping up my chemistry bachelors. There could also certainly be something newsworthy in the present article that I can't presently see.)

"Extremely thin" -- wow!

[LotsOfPhil]

Then they placed an extremely thin layer of silicon dioxide -- about two billionths of a meter thick -- onto the silicon between the gold contacts.

Holy cow! 20 angstroms? You can't get much thinner than that. The Si-O bond length is probably 3-4 angstroms. That is stunning.

Re:"Extremely thin" -- wow!

[grrrl]

20 angstroms (2 nanometers) is not difficult, especially given that SiO is generally not deposited on Si but rather grown out of it by oxidation (at least, for standard MOS that's how it's done). But depositions of films of this thickness is pretty regular in semiconductor work.

In fact, all the dimensions outlined in the article are pretty standard, if not large, for this type of research.

How many units?

[Anonymous Coward]

The summary talks about these different measurements in microns, inches, and meters. Is there a good reason why we can't use one unit for all of them? It kind of makes it hard to follow. And please, don't choose meters. It's very hard to understand what 1/2,000,000,000 meters is. At least use milimeters or micrometers or something.

Some handy Google conversions:

300 microns = 0.0118110236 inches
300 inches = 7 620 000 microns
(1/10) inches = 2 540 microns
(1/2 000 000 000) meters = 0.0005 microns
(1/2 000 000 000) meters = 1.96850394 × 10-8 inches

Re:How many units?

[r_jensen11]

But how many human hairs wide are all of those? AhhhHA!

This sounds like a pretty big deal

[Anonymous Coward]

One of the fundamental barriers to quantum computing has been finding an ideal method of representing qubits. No matter how you impliment qubits they have a limited time before they decohere due to interference from the environment (though this lifespan may be able to be extended through error correcting codes). If I recall correctly, magnetic spins have the longest dechoerence times of any method of representing a qubit. The problem with magnetic spins is that they're hard to measure. In fact, you could more or less say, they last a long time because they are hard to measure (i.e. they're relatively isolated from outside influence).

But if you can sidestep this tradeoff and come up with a method of storing qubits with long decoherence times that's easy to read, you've got a good candidate for at least a component of a quantum computing implimentation.

Heat?

[morgan_greywolf]

Then the device was chilled with liquid helium to 452 degrees below zero Fahrenheit

And here I thought AMDs had cooling problems...

Oh come on!

[Assassin bug]

Two billionths of a meter thick!? Come on! Just say, 2 nanometers please.

Quantum virtual machine

[edschurr]

This made me wonder, are there any quantum virtual machines? Surely nobody is waiting for a physical implementation. Well, apparently there are a few: Linear Al, libquantum, and a Java quantum circuit simulator. Now I wonder how difficult it's going to be to program something...

Re:Quantum virtual machine

[dido]

Sure there are, the only trouble is simulating more than a handful of qubits will require a really powerful machine. If you had as few as 256 qubits your simulator would need to keep track of 2^256 states, about as many states as there are sub-atomic particles in the visible universe! It's one of the reasons why computer modeling of quantum mechanical phenomena on classical computers is so hard, and probably that this application, far more than Shor's or Grover's algorithms, will be the primary use for the quantum computers that do get built. Gauge quantum chromodynamics simulations, which use the Standard Model to calculate the masses and other properties of sub-atomic particles, have been known to tie up supercomputers for months on end. A quantum computer would be able to perform such computations far more easily.

Re:Quantum virtual machine

[master_p]

...the universe?

I mock theeeee

[Impy the Impiuos Imp]

> Then they placed an extremely thin layer of silicon
> dioxide -- about two billionths of a meter thick

Holy crap! That's about 2 trillionths of a kilometer thick!

Re:They read the spin of 10,000, but can't read on

[Anonymous Coward]

so what my memory stick can store 128 billion times more information, and rewrite it over and over and over again in a fraction of the time, sure its just usb, but come on, we dont need no stinkin golden silicon oxide interface. besides who can read braile these days anyway, not this joker.

Flash Showing Animation

[fedrive]

This flash albeit published years ago by another inventor shows how the concept will work using ferroelectrics, silicon,etc. using phosphor as a visual data agent.

http://colossalstorage.net/display/atomic_switch_d isplay.htm

So far

[Cracked Pottery]

They have developed nonprogrammable read only memory.

Re:So far

[someone1234]

Well, you just have to pick the wafers which contain shakespeare and reprocess the waste.

Standard Units

[istartedi]

For those who aren't sure, "the width of a human hair" is based on the average width of a clump of random hairs kept under a glass Bell jar at constant tempterature and humidity in Paris, France. I've never seen it, but allegedly the clump is about the size of a hailstone.

I really hope this is true

[Randy Jian]

Advancement in computer hardware engineering has/will really slow down. Successful development in quantum computing and storage will hopefully kick of a actual information revolution.

Units man

[sauron_of_mordor]

"[snip] about 300 microns thick [snip] less than 3 inches [snip] one-tenth of an inch wide [snip] about two billionths of a meter [snip]" Some consistency would be good! -S

http://www.bash.org/?2999

[dunkelfalke]

There was a 23% drop in temperature.
  That's almost 25%! ... That was one of the most worthless comments I've ever heard.

Re:They read the spin of 10,000, but can't read on

[Rei]

To me, this is bad news. Not because of the technological implications -- the implications of this are great. The bad news, for me, is that I'm going to need to go back and revise parts of my last novel before it goes out. Kane quantum computers ("Kane chips") played a significant role, and I don't want it to have any technically inaccurate portions. ;)

FYI, what the article doesn't mention -- the reasoning why this is important is twofold.

1) Kane quantum computers have very little problem with "decoherence". That is, interaction with the outside world can ruin your state, and the error correction to compensate for this is so expensive for most quantum computer designs (excepting Kane's, and a few others) that you might as well be using a normal computer.

2) Kane's design is highly scalable, and builds on existing chip fabrication infrastructure.

The biggest problem with the design was qubit readout, and it looks like they're close to getting this one nailed down!

Re:They read the spin of 10,000, but can't read on

[Anonymous Coward]

Yes, but with the computer that uses your memory stick, it matters whether P=NP.