If the Oxford Physics Department sounds grey and sobersided to you, you might want to watch the scientists themselves acting in the movie The DiVincenzo Code about a mad science scheme to reduce the value of the h-bar Planck constant to zero. But life is not all fun and games at Oxford's Ultrafast Group.
White PapersMore >>
Led by Prof. Ian Walmsley they labor in the dawn of quantum computing over the quantum light-matter interface, performing theoretical leaps and developing engineering solutions for devices which will eventually become quantum communication channel repeaters and quantum memory.
I spoke by phone with Prof. Walmsley recently ...
JW: Before we discuss your work on quantum memory, I'd like to ask you a few questions about your earlier paper Ionization effects on spectral signatures of quantum-path interference in high-harmonic generation.
IW: Quite a different subject! What caught your interest there?
JW: It was that everything you said about the harmonics generated from the path interferences which you were studying seem to apply to the acoustics of music as well.
IW: There are some very similar analogues there, having to do with the fact that the underlying entities in all of these physical theories is a field, which is a weird sort of concept. Once you have that, interferences and oscillations play a big role in quantum mechanics as they do in music.
IW: As in an organ pipe.
JW: Or a clarinet .. the 1st, 3rd, 5th, 7th, etc. harmonic generation being especially strong because the second normal mode is 4/3 the length of the pipe.
IW: It all depends on where your boundary conditions are. That's an important thing.
JW: Further I was intrigued by the electromagnetic phenomenon of plateau harmonics, where the rapid decrease in intensity of the harmonics ceases and nearly levels out when you reach the plateau around the 12th harmonic.
IW: When you get into these high harmonics, the amplitude does not decrease with harmonic order. It's really a fundamentally different mechanism, a different way to look at this perturbative n-photon absorption.
JW: Something like this happens in brass instrument playing. As you're playing up the harmonic series, you have to put more effort into achieving the higher and higher notes, but when you get up around the twelfth harmonic, it takes about the same effort to ascend therefrom, which is why classic valveless trumpet and horn playing worked. Anyway, that's what intrigued me in reading your earlier paper!
IW: There are certainly some very strong analogues.
JW: With regard to your current work on quantum memory, I have talked to many people working on quantum processors. You are striving towards, by contrast, an indispensible peripheral, since among the many questions in quantum computing is, "What does one do with an intermediate result if you want to continue a quantum computation?"
IW: There are some immediate applications to our present work [apart from the longterm use as memory for a quantum computer]. The obvious one is the next stage of a quantum communications link, which is called a quantum repeater.
If you want to communicate using photons over a long link, the loss goes exponentially with the length of the link. That turns out not to give you any enhancement, quantum mechanically, over that which you could get with a regular communication link.
What you have to be able to do is break that link up into sections so that you have efficient transmission over each section. But if you are going to do that, you have to have a way to store entanglement of these intermediate sections.
That's the utility of quantum memory. I think that will turn out to be one of the early applications before we get to the level of computers.
JW: The problem of storing entanglement is indeed, is it not, one of the fundamental problems of practical quantum computing?
IW: It is. Solving that problem will certainly be of value to a quantum computer.
JW: And your current focus is to be able to make secure quantum channels which repeat the entanglement in stages so you get the same quality of entanglement at the end as at the beginning of the pipe?
IW: Over a long distance, that's right. We certainly have an eye on quantum computers, but there are many other things that will have to come together before one can envisage building a quantum computer out of light, or indeed, out of anything.
So one is looking for intermediate applications that might come to pass in the next few years, rather than the next ten years or so.
JW: And which might incidentally finance further research!
IW: Yes, that's true. One of the interesting things that quantum information processing has brought is an interesting way to look at the contrast between the classical and quantum worlds, that is, looking essentially at how much it costs to do something. How do resources scale when you're looking at a quantum problem compared to a classical problem? At the boundary between the classical and quantum worlds you have quantum computers, which are a macroscopic scale machine that operates on a preserved quantum phenomenon. It's a really interesting physics task.
JW: I have seen ion-trap experiments in quantum computing. I've talked to people who are working with Josephson junctions. I have not managed yet to chat with the electron-spin crowd nor have I gotten briefed on photon-based quantum computing, which I believe is the approach your team would favor.
IW: Yes, certainly. These memories were developed in part with that sort of quantum computing in mind.
JW: So you are colliding photons into an atomic gas, which are later re-emitted.
IW: Our first stage was to say, "How do we get what you might call practical bandwidth in a practical system?" And we were successful and increased the bandwidth dramatically. The second feature, again practical more than fundamental, is that we are able to do this in warm gas. You don't have to have a big cooling apparatus. What we are working on now and have some preliminary results on is showing that we can really operate this device at the quantum limit.
JW: So how does this work?
IW: The key feature is that you take an excitation of the electromagnetic field, the photon, and you map it to an oscillating excitation of that atomic system. Two things then happen.
First of all, the atomic system is fixed in space, stationary, so you can hold it one place. And second, it being an atomic system, with very little, relatively rare gas, has a fairly long lifetime, on the order of hundred of microseconds up to milliseconds, and longer if you do other things.
The point is that light is not fixed in space. You're mapping a temporal excitation into a spatial excitation. It's not an original idea, it's been around for some time. There have been various schemes proposed over the past decade for doing this. Our scheme is one that is particularly well-suited to using large bandwidths and warm atoms.
JW: How warm are they?
IW: These are about 50 degrees centigrade.
JW: A little warmer than room temperature in the Sahel.
IW: Yes, you don't have to have them at microkelvin temperatures.
JW: So you've made a significant engineering advance based on theoretical ideas which have been kicking around for some time.
IW: I would like to claim there is a physics advance, too, because the particular process we are using had been looked at once briefly but dismissed as being impractical, and we found a way to make that physical process practicable.
The fact that one can do it at practical bandwidths in warm vapors, you're right, that's an engineering step.
JW: What is the physics insight that made this work?
IW: The physics insight was to recognize that, when you go a long way from electronic resonance, the bandwidth of the memory you have, the bandwidth of the storage, is determined almost entirely by the shape and bandwidth of the control field, as opposed to many of the other systems where it is determined by the atomic absorption resonance. That gives you a lot of advantage in bandwidth and tunability.
People has sort of looked at this one paper that had decided that it wasn't practicable because you couldn't get efficiency. So we decided to study the physics of the interaction and optimize it.
Our control pulse energies are about a factor of ten (10) more than the people who do it on resonance. A factor of ten is not outrageous in nanojoule pulse energies, nothing really dramatic.
A standard way to do memory is that you take an atom and absorb the light directly into the atomic electronic transition, you excite an atomic electron. Then you apply afterwards you apply a control field to take that excited electron and store it in a more quiet state.
In that scheme, your absorption is determined by the properties of the atom you are using.
What we do is this: we never excite a real electron. We only excite virtual electrons and that gives us a flexibility the other schemes are missing. The tradeoff is that we have to use more energy in our control pulses than the other scheme.
JW: What's a virtual electron?
IW: The electron is real. You shake it around with the optical field, but you never give it enough energy that it can stay stably in an excited state. So it's never put into a real excited state, it's put into what we term a virtual excited state.
Imagine water waves. If you put a float on a water wave, it bobs up and down. It follows the amplitude of the wave. Imagine if that wave got bigger and bigger, it could lift the float onto a rock. Then as the wave went down, the float would stay on the rock, it would have reached a "real excited state", if you like.
So we never get into that condition. The electron is always following the control wave as we wiggle it up and down.
What this means is that there is more flexibility in terms of the bandwidth of the memory and wavelength you can store.
JW: If the electron gets stuck "on the shelf" in the excited state as you described before, have you lost your quantum information?
IW: You haven't lost your quantum information, you're more or less restricted to where that shelf lies, where that state is. Certainly people have demonstrated quantum memories that make use of the fact that you generate a real excitation. But there you are constrained to the atomic absorption resonance, whatever frequencies the atomic absorption allows. Those are usually quite narrow bands.
JW: What can you do today with what you have built? And is it something which can be mass-manufactured as a relay on a fibre-optic system?
IW: We have not yet demonstrated something that can be integrated into an optical system. We have some ideas about how to pursue that.
What we are working on at the moment is proving that this room-temperature warm gas really can operate to store light at the quantum limit and can store entanglement efficiently.
Our overall efficiency was reasonable with respect to other systems. It still needs to be increased, so we are working to improve that.
I think we will not try to engineer something into a technology until we have nailed down all of these benchmarks that will be crucial for a practical system.
To read more about the Oxford Ultrafast Group's work on quantum memory, please see the paper Towards high-speed optical quantum memories.