At the heart of quantum mechanics is the wave-particle duality in which light, for instance, can be made to display wave-like or particle-like properties depending on the experimental setup used. Now, researchers headed by the French scientist Jean-François Roch provide compelling evidence to show that photons, the smallest units of light, do not "know" whether to behave as a wave or a particle, and it is the observer who makes this choice by deciding on what properties of light to measure. Their research appeared in the Feb. 16 edition of Science.
Most scientists today believe that the behavior of a quantum system (like those involving photons) depends on the type of measurements performed, which is in contrast to the view of some scientists in the 20th century (including Einstein) who advocated that the choice of a photon to behave as a wave or a particle is made long in advance of performing an experiment. To resolve this conflict, the theoretical physicist John Wheeler proposed a "delayed choice" experiment in 1978 which could provide evidence for the former view. The research conducted by Roch and his colleagues is an experimental realization of the delayed choice experiment.
In quantum mechanics, the photon can be treated as a wave that has a certain probability of being found at a particular place and time. If we wish to observe the wave-like properties of the photon, we can imagine having a single photon incident on a beam-splitter A that can split the wave into Path I and Path II, and then recombine these waves using another beam-splitter B. The two waves will interfere with each other, like two ripples occurring on the surface of water that might reinforce each other at some points to cause larger ripples, and weaken each other at other points to cause smaller dampened ripples.
In the experimental setup, due to interference, the probability of the photon being perceived at one detector (Detector I) may be different from that at the other detector (Detector II) depending on the difference in length of Path I and Path II. For certain path differences, the two waves may interfere to give no probability of detection at Detector I (and consequently all probability of detection at Detector II) whereas for other path differences, the reverse pattern may be observed. Therefore, the probability of detecting the photon at one detector or another is a function of the path difference though any individual photon will only be detected at one of the detectors-but not both.
If interference occurs, it implies that each photon has simultaneously traveled through both Path I and Path II, which is in contrast with our common-sense understanding of particles that only adopt one path or another but not both.
On the other hand, to observe particle-like properties, we can decide to repeat the experiment without beam-splitter B so that the photon waves, after being split by beam-splitter A, are not recombined; as a result, the photon is perceived by Detector I or Detector II with equal probability independent of path length. Since the photon couldn't have split into Path I and Path II because there is no way for these two paths to recombine together at the same detector in the absence of B, the photon must have taken only one of the two paths, akin to a particle. Half the time the photon takes Path I and shows up at Detector I whereas the other half of the time it takes Path II and appears at Detector II.
John Wheeler's "delayed choice" experiment uses the same experimental setup as above with one principal modification-the decision to include beam-splitter B in the experiment is made randomly after the photon has passed through beam-splitter A. After A, if the photon were a particle it would only take Path I or Path II but not both and consequently no interference should be seen, whereas if the photon took both Path I and Path II and not either, wave-like behavior should ensue.
The experimental data obtained by Roch and others indicate that, when the beam-splitter B is randomly included in the setup after the photon has passed A, then interference is observed, whereas if the splitter B is randomly excluded, particle-like behavior results. Thus the decision of the photon to behave as a particle or a wave after it has left A is delayed until it reaches B.
Though the experiment conducted by Roch and his colleagues wasn't the first delayed choice experiment, it was closest to that envisioned by Wheeler. In fact, the first delayed choice experiment, reported in 1987, was carried out by an experimental group including Arthur Zajonc, professor of physics at Amherst College.
Professor Zajonc, commenting on the findings, said that the delayed choice experiment conducted "is an elegant experiment." "This is the experiment you wanted to do, but it was too hard," he said. "It's going to be seen as a kind of a landmark." Professor Zajonc also added that the setup used had some essential improvements over previous experiments like "the use of true single-photon states for the input, relativistic space-like separation of the 'choice' and the input, and use of a quantum random number generator to make the choice [a truly random choice ensures that no deterministic process exists that again might provide information about apparatus configuration]." This eliminates any trace of "classical" electromagnetic wave nature to the incoming light, and ensures that the choice can have no influence on the input state.
Professor of Astronomy George Greenstein summed it up: "This experiment makes vivid one of the most amazing things about the quantum world: that it simply does not exist in our ordinary sense of the word. Quantum objects have no separate, independent reality. It is a little like learning that you drive only one vehicle, and that it is an SUV and that it is a bicycle."