| Some non-classical and non-local effects in two-photon interference processes are investigated theoretically and experimentally. When the sources of the two photons are randomly phased, ordinary second-order interference effects disappear, but higher-order effects may still be present. Fourth-order interference effects, that can be established by observing intensity correlations rather than the intensity itself, are predicted even classically for two independent sources. But quantum mechanical theory predicts much stronger correlations than those possible from classical considerations. We have shown that if the simultaneous photon pairs produced in the process of parametric down-conversion in a nonlinear medium are allowed to interfere, the joint probability of detecting two photons at two points x and x{dollar}spprime{dollar} in the interference plane with two photoelectric detectors exhibits a cosine modulation with the separation {dollar}vert{lcub}rm x - xspprime{rcub}vert{dollar} of the two detectors, with a relative depth of modulation or "visibility" of 100%, whereas classically the maximum realizable visibility is only 50%. The photons in this case have a wide bandwidth, but the interference effects persist even when the measurement time interval is much larger than the coherence length (given by the reciprocal bandwidth) of the interfering photons. A visibility of 100% of the interference pattern implies the vanishing of the joint probability of two-photon detection for certain pairs of points in the detector plane, namely those separated by an odd integral multiple of the half interference spacing, even though the single-photon detection probability has no spatial modulation, i.e., one photon can be detected anywhere in the interference plane. This phenomenon is reminiscent of violation of local realism in the E-P-R sense, since the outcome of a measurement at one position x is strongly influenced by where the other detector is located. The results of an interference experiment on down-conversion are presented where the laws of classical probability are shown to be violated by a significant degree. When the sources for the interference experiment are just two fully excited atoms, emitting spontaneously and hence independently of each other, there are no second-order interference effects, but fourth-order interference effects are predicted theoretically. By selecting two atoms at an intermediate point of their Rabi oscillation cycle, i.e. when the two atoms are subject to the same degree of partial excitation, one should be able to observe second-order interference effects. |
