Group-velocity dispersion (GVD) places a fundamental limitation on the resolution possible when measuring the propagation time of short optical pulses through dielectric media. We show that due to a nonlocal, quantum-mechanical effect, a surprising cancellation can occur in fourth-order interference, allowing the time interval between photons emitted in spontaneous parametric down-conversion to be measured without significant degradation of the resolution due to the spreading of the individual photon wave packets. This may also prove useful for measuring the higher-order contributions to GVD.
Group-velocity dispersion (GVD) places a fundamental limitation on the resolution possible when measuring the propagation time of short optical pulses through dielectric media. We show that due to a nonlocal, quantum-mechanical effect, a surprising cancellation can occur in fourth-order interference, allowing the time interval between photons emitted in spontaneous parametric down-conversion to be measured without significant degradation of the resolution due to the spreading of the individual photon wave packets. This may also prove useful for measuring the higher-order contributions to GVD.
In the process of spontaneous parametric down-conversion, conjugate pairs of broadband photons are emitted simultaneously from a nonlinear crystal. We use coincidence detection in a fourth-order interferometer to measure the propagation delay of one photon with respect to that of its conjugate. One photon of every pair enters each arm of the interferometer, and when the path-length difference is adjusted such that the two photons meet essentially simultaneously at the output beam splitter, a dip is observed in the coincidence rate between photon counters placed at the two output ports. In our experiment, the rms width of this dip is 15 fs. By placing various dielectric media in the path of one of the photons and then adjusting the free-space path lengths until the coincidence dip reappears, we have measured the propagation time of single photons and found it to agree with predictions based on the group velocity.1 Because of a surprising quantum-mechanical effect, group-velocity dispersion cancels out, such that the expected wave packet broadening has essentially no effect on the high time resolution of this technique.2
It is a well-known and remarkable fact that in certain coincidence photon-counting experiments with cw-pumped parametric down-converters, the effects of group-velocity dispersion arising from media interposed between source and detectors are completely canceled, even if the media physically affect only one of the photons of the pair. Recently Perina et al. [Phys. Rev. A 59, 2359 (1999)] showed that this phenomenon does not occur when certain classical timing information is available about the arrival of individual photons at the detectors, as is the case when the photon pairs are produced via spontaneous parametric down-conversion using an ultrashort pump pulse. In this paper we show that the nonlocal cancellation of dispersion for such a source of entangled photons can be restored in principle by proper engineering of the source properties. In particular, we describe techniques for recovering interference in coincidence-counting experiments by suppressing distinguishing information without the post selection of photons. Moreover, a precise classical timing signal coincident with the photon pair is still available.
Nonlinear SU(1,1) quantum interferometers based on non-degenerate optical parametric down-conversion exhibit strong unbalanced group velocity dispersion (GVD). This feature is intrinsic to this type of interferometer as correlated photons of vastly different frequencies propagate through a dispersive nonlinear crystal; consequently, the dispersion arises from the source itself. The resulting GVD degrades the axial point-spread function (PSF) in optical coherence tomography (OCT) with undetected photons; and physical compensation is less straightforward, in particular for non-degenerate broadband regimes due to the limited number of suitable materials. In this contribution, we analyze dispersion in bulk nonlinear interferometry and describe its implications for OCT imaging. Aspects of hardware compensation are addressed, and a novel empirical numerical method of compensation is proposed. The approach is based on the extraction of the phase component directly from the time-domain modality (high precision linearized quantum Fourier transform infrared spectrometer) and its injection into the mid-IR spectral-domain OCT signals (central wavelength of around 3770 nm) before the Fourier transform. The proposed method is compared with an alternative numerical technique. The results demonstrate a 2.2-fold improvement in axial resolution and outperform the alternative correction method in overall imaging performance.
We have demonstrated for the first time that single photons in glass travel at the group velocity, and have observed a novel, nonlocal dispersion-cancellation effect. We used a two-photon interferometer in which a conjugate pair of photons produced in parametric fluorescence travel separate paths and are detected in coincidence after being recombined at a beam splitter. A piece of glass was placed in the path of one of the photons, and a variable delay was adjusted to precisely compensate for it. The single-photon propagation time was thus measured to within approximately 4 fsec.
The concept of dispersion cancellation has been established in light optics to improve the resolution of interferometric measurements on dispersive media. Odd order dispersion cancellation allows to measure phase shifts without defocusing the interferometer due to wave packet displacements, while even order dispersion cancellation allows to measure time lags without losing resolution due to wave packet spreading. We report that either type of dispersion cancellation can be realized very easily in a triple Laue interferometer. Such interferometers are Mach–Zehnder interferometers based on Bragg diffraction, and are commonly used for neutrons and x-rays. Although the first x-ray interferometer was built nearly five decades ago, the feature of dispersion cancellation hasnʼt been recognized so far because the concept was hardly known in the neutron and x-ray community. However, it explains right away the surprising decoupling of phase shift and spatial displacement that we have discovered recently in neutron interferometry (Lemmel and Wagh 2010 Phys. Rev. A 82 033626). Furthermore, this article might inspire the light optics community to consider whether a triple Laue interferometer for laser light would be useful and feasible. We explain how dispersion cancellation works in neutron interferometry, and we describe the setup rigorously by solving the Schrodinger equation and by calculating the path integral. We point out, that the latter has to be evaluated with special care since in our setup the beam trajectory moves with respect to the crystal lattice of the interferometer.
In the experiment presented in the paper, HBT (Hanbury-Brown and Twiss) interferometer is used to measure the second-order correlation function (G2) for two entangled photons, denoted as signal and idle ones, generated by the spontaneous parametric down-conversion process. When an extra optical fiber is inserted in the signal's path, the arrival times of these two photons are different, so that the length of the fiber can be estimated by measuring the time offset in the G2 function.#br#As the laser light we used is not monochromatic, the experimental result is affected by the group velocity dispersion (GVD). Light of different wavelength has different velocity and leads to different travelling time in the fiber. Through calculation, we find that the G2 function of each wavelength has the same shape but different center time offset. Thus, the final G2 function, as a superimposition of the functions with different wavelengths, is broadened in a long-distance measurement, which seriously reduces the estimation accuracy. We analyze this phenomenon in both theory and experiment, and the experimental results fit our theoretical simulations quite well.#br#In the theoretical analysis, we find that the group velocity dispersion introduces an additional time offset to the final G2 function which is proportional to the total time offset. In our experiment, this additional time offset is about 1% of the total time offset. In particular, this effect results in a 63600 ps extra time offset, which can be compensated in post process, with a time accuracy of about 1 ps.#br#This paper focuses on the G2 function affected by the group velocity dispersion, and also calculates the additional time offset produced by group velocity dispersion, and thus gives a method to compensate the additional time offset through calculation.