Quantum Computing with Microwaves

It's been a while since I did any ResearchBlogging, first because I was trying to get some papers of my own written, and then because I was frantically preparing for my classes this term (which start Wednesday). I've piled up a number of articles worth writing up in that time, including two papers from an early-August issue of Nature, on advances in experimental quantum computation (the first is available as a free pdf because it was done at NIST, and thus is not copyrightable). These were also written up in Physics World, but they're worth digging into in more detail, in the usual Q&A format.

So, have they built a quantum computer to factor big numbers and hack credit card encryption yet? No, your credit cards are still safe. These papers are reporting on some technical advances in ion trap quantum computing. Specifically, they're using techniques that allow you to control the state of trapped ions with microwaves, rather than lasers or magnetic fields.

Whoa. No wonder it got written up in Nature. Talk about repurposing everyday technology... "Let's see, do I want to pop some popcorn, or entangle the states of two trapped ions?" We're not talking about a microwave oven, we're talking about light in the microwave region of the electronic spectrum. The two groups-- one at NIST, the other in Germany-- have demonstrated the ability to manipulate the states of trapped ions using only microwave radiation.

Yeah, but haven't they already done lots of experiments manipulating trapped ions? Why is this interesting? What's interesting about this is that it greatly simplifies some of the important processes. Previous experiments have used optical frequencies to manipulate the states of the ions, using light from very complicated laser systems. They've been able to do some pretty amazing things this way-- to lift a phrase from Winter's Tale, "Light under flawless tutelage knows no limits," but the level of flawlessness required takes an awful lot of work.

Microwaves, on the other hand, are an extremely well-understood technology, and there's a vast industry devoted to integrating them with computer chips and the like, in the form of cell phones. If quantum computing operations can be done with microwaves alone , that makes life a lot easier for the people who would need to build and operate quantum computers in the future. You can even build the whole thing into a chip, which is what the NIST group has done:

The ion trap control electrodes are labeled C1-C6, and the RF frequencies needed to make the trap are brought in along the orange wires. There are also three yellow microwave transmission lines, that provide the fields used to do the state manipulation.

If this is such a great idea, why hasn't anyone done it before? Well, because it's tricky. One of the key things needed for a practical quantum computer is addressability-- that is, the ability to pick out a particular ion and change its state at will. You also need to be able to affect a group of ions collectively, so that you can entangle their states, which in the ion trapping system is done by setting all of them moving.

These things are relatively easy to do with lasers, because as a general rule, you can focus light down to a spot size roughly comparable to the wavelength. the ions in a typical trap system are separated by distances on the order of microns, and optical frequency light has wavelengths of hundreds of nanometers, so this is relatively easy to do-- you just get a big lens, and focus your laser on the particular atom you want.

Microwaves, on the other hand, have wavelengths on the scale of centimeters or meters (the NIST experiment uses a frequency of 1.69 GHz, which corresponds to a wavelength of 18cm), which is bigger than the whole ion trap. It's not easy to arrange for microwaves to "talk" to only one ion, or to set the group in motion.

So how did they manage it? They used the fact that if you're working with an ion trap on the scale of a chip, like they are here, you're in what physicists call the "near field," which means the behavior of waves is more complicated than the usual rule-of-thumb situations, but if you're clever enough, you can find ways to engineer lots of interesting features into the radiation field. Some of those configurations are useful for doing the things you need to do to make an ion trap quantum computer, and the NIST paper is basically a demonstration that they can, in fact, make this work.

You say "demonstration," which often seems to mean "only sorta-kinda works." Is this another of those cases? Well, it definitely works. They can manipulate the state of a single ion to make what they want with about 98% probability, and they can prepare an entangled state with about 76% fidelity. The data are pretty clean, too, so it's not like they're backing this out of some complicated and noisy signal.

Now, those are nowhere near good enough to make a practical quantum computer-- you'd like both of those numbers to be over 99%-- but as a proof-of-principle experiment demonstrating a new technology for quantum information it's, well, good enough to get into Nature.

I don't know, dude, that's kind of weak. Well, as previously noted, there's a multi-billion dollar global industry in microwave processing, so there's every reason to believe that the technical issues limiting the performance in this experiment can be beaten down. And there are some other features that are very nice-- the single-ion gate times are really fast, for example, faster than can be done with CW lasers (though Chris Monroe at Maryland has done better with intense pulsed lasers). But it's very much a proof-of-principle.

OK, so that's the NIST paper, what's the German one? The same thing with all the verbs pushed to the end of the sentences? No, Nature publishes in English.

The German group also used microwaves to manipulate the states of trapped ions (ytterbium, if you care, where the NIST group used magnesium), but they attacked a slightly different issue, namely the need to have stable states to store information for a long-ish time during a calculation. This is a problem because if you want to do a real calculation, you're necessarily going to have some of your qubits hanging around in an entangled state for a while as you do operations on other qubits in your quantum computer.

There are some clever ways to get around this by using superpositions of multiple states that are prepared in just the right way to avoid interactions with the environment that would cause them to fall apart-- what people in the trade call "decoherence-free subspaces." Such states generally require particular combinations of laser fields to set them up, and fairly specific magnetic fields to be applied. They're vulnerable to tiny fluctuations in the magnetic fields, particularly of a type that produces magnetic field gradients across the sample, and those are damnably difficult to get rid of.

And microwaves get around this problem? Yes. Very loosely speaking, it's because they're oscillating fields rather than dc fields, and it's easier to control oscillating fields in a lot of respects (again, multi-billion-dollar industry, blah, blah).

Rather than creating and maintaining their superposition states with dc magnetic fields, they use microwave fields to mix different levels together.

So, it's a microwave blender? Not exactly. The idea is, roughly speaking, that if you take a microwave field that is nearly resonant with the energy difference between two states, an ion exposed to that field will flip back and forth between them so rapidly that it's essentially in a superposition of the two different states. With any pair of states that you connect with microwaves, you can get two different sorts of superpositions, and these "dressed states" are the appropriate states to talk about when analyzing what happens to an ion exposed to the field.

The system they use is considerably more complicated-- it involves four different levels in ytterbium, plus a few different microwave frequencies, but when you put the whole thing together, you can make some dressed states that are very robust with respect to the sorts of field fluctuations that usually kill these things.

"Very robust" meaning...? They measure the lifetime of a superposition state created through this method, and find that it's almost two seconds. They can also use microwaves to drive ions back and forth between states, and demonstrate high-contrast oscillations for the better part of a second. That's a pretty long time as such things go, better than 100 times longer than you get with the bare states, without the microwave field.

This is just another piece of the puzzle, though, right? I mean, they haven't made a real computer out of this. That's right. In this case, they have a scheme for using microwave transitions between these dressed states to do entangling operations, but as far as I can tell, they don't have an experimental demonstration of it yet.

So, both groups have a lot of work to do to turn these papers into really practical devices, but they're interesting steps in the direction of improved quantum information processing.

I dunno, dude. I mean, I expect papers in Nature to be really exciting, but this all seems suspiciously like engineering. In a sense, I suppose it is. These are technical refinements, providing a new way to do things that people have already done with other systems.

But if you think about it, that's kind of exciting in and of itself. Back when I started in grad school, there were plenty of people who doubted we'd ever get to the point of needing to do engineering type activities with quantum information systems. But now, here we are, with the messy technical details getting brought more and more under control, laying the groundwork for working with quantum information on larger and larger scales.

I guess if you look at it that way, it is kind of impressive. Still, it's not like they're factoring huge numbers, here. True, but as an old joke in the field has it, you won't see that work published in Nature. If Dave Wineland at NIST ever makes a working quantum computer capable of doing serious code-breaking type calculations, the way the world will find out is that he'll drop out of sight, because the NSA will classify his whole lab...

We're a long, long way from that, but at least this stage gives us peer-reviewed literature to talk about on blogs...

Ospelkaus, C., Warring, U., Colombe, Y., Brown, K., Amini, J., Leibfried, D., & Wineland, D. (2011). Microwave quantum logic gates for trapped ions Nature, 476 (7359), 181-184 DOI: 10.1038/nature10290

Timoney, N., Baumgart, I., Johanning, M., Varón, A., Plenio, M., Retzker, A., & Wunderlich, C. (2011). Quantum gates and memory using microwave-dressed states Nature, 476 (7359), 185-188 DOI: 10.1038/nature10319