When light passes through an interferometer, the visibility of the resulting fringe pattern depends on the amount of "which-path" information available to the experimenter. If the experimenter is able to determine, by any means, which path the light takes through the interferometer, then there is no interference. If the experimenter has no information about the path (and cannot even in principle determine which path the light took) then high visibility interference can be observed.
In some experiments it is possible to switch between having full which-path information and having no which-path information by a subtle modification of the experimental apparatus. In cases where the which-path information is not available, the which-path information is said to be "erased", and this erasure can allow the observation of high visibility interference fringes. We refer to an interferometer with these properties as a "quantum eraser". For example, a waveplate inserted into one arm of an interferometer can be used to modify the polarization of one beam in the interferometer. With suitable polarization analysis after the interferometer, rotations of this waveplate can either yield or erase which-path information, hence changing the visibility of the observed interference pattern.
We have demonstrated quantum erasure in a polarization interferometer using two different sources of polarization-correlated photon pairs: an entangled-state source which exhibits quantum mechanical correlations, and a mixed-state source which exhibits classical correlations. One of the source beams (the signal beam) traverses the polarization interferometer before being measured. The other (idler) beam passes through a polarization analyzer before being measured. We look for interference in the measured coincidence counts between these beams, hence interference depends not only on the properties of each individual beam, but on the correlations between them. We can thus perform erasure not only by modification of the interferometer apparatus, but also by modifying the polarization analysis made on the idler beam.
We find that the visibility of the measured interference pattern does indeed depend on how the polarization of the idler beam is analyzed; this is true for both the entangled-state and the mixed-state sources. However, we find that the results obtained with these two sources are not identical in all respects. For the entangled-state source we find that interference is lost when which-path information for the signal beam can be obtained by measuring the polarization of the idler beam. For the mixed-state source which-path information is never available, and the lack of an interference pattern is due to the inability to separate two overlapping interference patterns which are out of phase with each other.
· More details: A Gogo, W.D. Snyder, and M. Beck, "Comparing quantum and classical correlations in a quantum eraser,".Phys. Rev. A, 71, 052103 (2005).
Top shows measured coincidence counts between entangled photon pairs when the polarizer in the gate beam is set to yield full which-path information. No interference is observed.
Bottom shows measured coincidence counts between entangled photon pairs when the polarizer in the gate beam is set to erase the which-path information. High visibility interference is observed.
webpage updated 2/17/05
beckmk at whitman.edu (replace "at" with @)