early days of experimental quantum cryptography, The

IBM Journal of Research and Development, Jan 2004 by Smolin, J A

This paper describes the first quantum cryptography experiment, performed at the IBM Thomas J. Watson Research Center in the summer of 1989 by Charles H. Bennett and the author. The apparatus and some of the lesser-known details of the experiment are illustrated and discussed, and quantum cryptography is discussed in the light of some of the more recent research. Also included as an appendix is a short essay about Bennett written by Rolf Landauer.

Introduction

I would like to share with the reader some of my experiences in the early days of quantum cryptography [1, 2], beginning with the summer of 1989 when I had just graduated from MIT. I would have been happy to do nothing all summer, but, since I already knew Charlie Bennett, mostly through his stepson George Dyer, I knew that Charlie wanted help building his quantum cryptography experiment, so I came to IBM for the summer to help him out.

I assume that readers, for the most part, know how the Bennett-Brassard scheme for quantum key distribution works [3]. For those who don't, a good review can be found in [4]. Briefly, Alice (the sender in the protocol) sends a single photon in one of four possible polarizations, either a vertical or horizontal photon, representing a 0 or 1 in the rectilinear basis, or a photon polarized along one of the diagonal directions, representing a bit in the diagonal basis. Actually, in the experiment we used a third choice of basis: right and left circular polarizations, instead of the diagonal basis. When Bob (the receiver) receives the photon, he measures in one of the bases, and only then does Alice announce which basis she used. Half the time Bob uses the wrong basis, and they have to throw away that bit and do it again. When they are right, however, they have a bit that ought to agree. Anyone eavesdropping in the middle also would not know which basis to use and thus would be likely to disturb the photon. If Alice and Bob check their agreement on some of their photons, they can detect attempted eavesdropping.

The apparatus

Our apparatus (Figure 1) consisted of a light-emitting diode (LED) that emits dim pulses, a lens and pinhole to collimate the beam, a 550-nm filter, a polarizing filter, and two Pockels cells (crystals that rotate the polarization of light as a function of applied voltage) for Alice, which allowed her to choose a basis and a bit. By the time the light pulses left Alice's section of the apparatus, they were attenuated to an expectation value of less than one photon per pulse. Next was the quantum channel (30 cm long) and then Bob, who got a Pockels cell and a calcite crystal. The crystal split orthogonal polarization states into two different beams, and the Pockels cell let him choose his basis. Finally, there were two photomultipliers able to detect single photons with a quantum efficiency of a few percent.

Neither Charlie nor I knew much about building anything, but we knew enough to be dangerous. As an example of Charlie's experimental agility, I remember a time I was visiting George in their apartment in Cambridge, Massachusetts. Charlie was excited about some fancy new tea he had gotten somewhere. He had set up a little double boiler using a pot and a teapot, explaining how this was the right way to cook the very delicate tea. George and I left the house for some time, returning hours later; when we came into the kitchen, we noticed the teapot. If you know about black-body radiation, you've probably seen how a black body turns invisible in a furnace, radiating the same spectrum as fills the cavity. The situation we found was not quite that, but there was a red teapot sitting in an empty pot on the stove. This would not have been disturbing, except for the fact that at room temperature the teapot had been green. I had forgotten this and wouldn't have known except that George pointed it out, and proved it by turning off the stove. The delicate tea had left nothing but a faint burnt aroma.

I can't resist a further aside about the time George, Charlie, and I attended an engine fair near Wendell, Massachusetts. There were many hobbyists with these antique gasoline engines with big flywheels. At a certain point in the flywheel's revolution, a bit of gas would get squirted into a piston, there would be a kind of cough, and the wheel would pick up enough speed to rotate another revolution. There were maybe a hundred such engines belching away, and all kinds of stalls selling parts and old tools. We bought a gas-or maybe kerosene-blowtorch to play with, an idea as bad as the teapot. We took it home, filled it with gas, and lit it. This resulted in a lovely flame for a time, but as the thing heated up, the solder bits mending the many holes in it started to melt, and we ended up with tongues of flames bursting out in all directions. There was no way to go near the thing and nothing much to do but let it run out of gas.

Anyway, that's the kind of experimental background Charlie and I were starting with. We also had no budget. The stockroom at the time was very good, and we could get all kinds of small parts more or less for free. Charlie had discovered the difference between capital and expense budgets. He said he could order nothing worth more than $300, but as much as he wanted under $300. I think there may have been a slight exception made for the photomultiplier tubes, but mainly we had to improvise. Charlie had come up with a poor-man's version of a laser table. He had some pieces of angle-iron (actually aluminum) with holes drilled for small set screws. This allowed for all six degrees of freedom that were possible with professional laser-table mounts, but they cost almost nothing. (They didn't work all that well, either.) Figure 2 shows a Pockels cell in one of the mounts. To adjust it, the screws were turned, which moved the cell in a complicated function of the degrees of freedom you actually wanted to adjust. You also had the additional excitement of the apparatus moving substantially when you stopped touching the screw, because of the significant amount of lash in the threads. (For readers who know more about physics than screws, that could also be called mechanical hysteresis.)