Single-photon interference seen.

Duality is commonplace in modern physics. We are taught that things have a double nature, particlelike and wavelike. Things that people tend to think of as particles, such as electrons, also exhibit wavelike behavior; things that people tend to think of as waves, such as light, also exhibit particulate behavior.

The wave-particle duality was first enunciated by Louis de Broglie in 1923. Now Alain Aspect, a physicist from the University of Paris at Orsay, reports what he says is the first experiment that demonstrates the dual behavior of light, particularly the wavelike behavior of single photons, particles of light. His presentation at last week's Symposium Commemorating the Centennial of Niels Bohr, held at the American Academy of Arts and Sciences in Cambridge, Mass., left an appreciative audience silent.

Over the decades many experiments have shown or claimed to show either the wavelike or particlelike behavior of light, electrons, neutrons, etc., and we are now to the point where technological artifacts, such as electron microscopes, make use of one or another aspect of the duality. However, it has always been a commonplace that an experiment (or a technological application) designed to see one side of the duality saw that side but not the other and vice versa. This elusive quality of the duality is one of the thngs that has fueled the longstanding and extremely complicated philosopical debate over the reality of the duality and the observer's influence on the outcome of experiments in quantum mechanics, the domain of atoms and smaller things: How much does what the observer sees depend on what he or she sets up to see? Of particular experimental interest has been the attempt to drive the paradox to its most elemental manifestation: whether single particles exhibit wavelike behavior. (Where astronomically large numbers of particles are in play, it is easy to invent statistical reasons why wavelike behavior could show up.) Experiments have repeatedly shown that no matter how faint the light, it exhibits wavelike behavior. Some experiments have also claimed to be observing single photons. Aspect criticizes these because they used calculated probabilities to determine whether they were dealing with single photons and recorded data with the photoelectric effect. The photoelectric effect has a well-known quantal or particlelike quality, but, Aspect says, the light that triggers it need not be particulate or quantal. The appearance of the quantal photoelectric effect in itself "tells nothing about one or two photons."

Aspect's experiment proposed to show particlelike behavior of light directly by modern quantum optical means that do not require the intermediation of the photoelectric effect. A beam of light is split by a halfsilvered mirror: The two halves of the beam go off at 90 [deg.] angles to each other. Detectors are put at the ends of equal paths in those directions.

If light is a wave, the wave will split into two. The two halves will take equal time to reach the detectors, and the recordings of the detectors will show a certain coincidence. If light is a stream of particles, says Aspect, one photon should go one way from the halfsilvered mirror, the next perhaps the other way, and so on. Over time the detectors' records should show much less than the coincidence proper to waves.

The experimenters tried various fine and delicate sources of light and could get no results better than the edge between the two. Then they turned to what Aspect calls "our secret source." This is cadmium energetically excited in such a way that it gives up its energy by emitting two photons in rapid succession, what atomic physicists call a cascade process. Each of the two photons has its own characteristic wavelength. The first photon of the cascade is used as a trigger. Equipment detecting its wavelength opens an optical gate long enough to admit the second, immediately following, photon. With this arrangement the experimenters got a particlelike coincidence reading. "We have proved that light behaves like a single photon," Aspect says.

Next they looked for wavelike behavior of the single photons. If what the halfsilvered mirror splits is a wave, and the two half waves are recombined after they have gone over unequal distances, they will be out of phase with each other. As a result, when added together, they will reinforce each other in other places. This produces a pattern of bright and dark "fringes" known as an interference pattern, and it is the standard test for wave behavior in everything from water in ponds to ionized hydrogen in the magnetosphere. To look for single-photon interferences, the experimenters inserted into the experiment an arrangement of transparent cyrstals that makes the experiment into waht is called a Mach-Zehnder interferometer. When they did this, interference effects appeared. They appeared in just the way that physicists would expect in a case like this: At first the data are a random mush; gradually, as the seconds of observation pass, a distinct interference pattern builds up.

"I suspect," says Aspect, "that we have shown wave-particle duality, wavelike behavior, particlelike behavior. Here, you can think of the photon going one way or the other; here, you can think of the photon split into two. Sometimes it gives me a headache."