Hong-Ou-Mandel effect with two frequency-entangled photons of vastly different color

TL;DR

The paper addresses whether Hong-Ou-Mandel interference can occur between photons of vastly different color. It demonstrates this by using a quantum frequency converter as an active beam splitter to couple photons at 637 nm and 1587 nm with a $282$ THz energy difference, preserving the frequency-entangled biphoton state. The work reports a color HOM demonstration with raw visibility above 90% and provides a detailed characterization of the converter as a unitary, frequency-mixing element, highlighting its potential to interface heterogeneous photonic qubits. This approach paves the way for integrating frequency conversion with quantum interference in heterogeneous quantum networks, enabling spectral-domain interconnects across diverse photonic platforms.

Abstract

In the original formulation of the Hong-Ou-Mandel (HOM) experiment, when two otherwise indistinguishable photons are incident upon the two input ports of a balanced beam splitter, they coalesce, always leaving via the same output port. It is often interpreted that this interference arises due to the indistinguishability of the single photons at the beam splitter; the situation, however, is often more nuanced. Here, we demonstrate an analog of HOM interference between two photons of completely different color. To do so, we utilize a quantum frequency converter based on sum- and difference-frequency generation as an `active' beam splitter -- coupling frequency-entangled red and telecom single photons with an octave-spanning energy difference of 282 THz. We achieve an uncorrected two-photon interference visibility beyond 90%. This work presents the first demonstration of HOM interference between two single photons of distinctly different color, deepening our understanding of what underlies quantum interference. It also suggests a novel approach to interfacing photonic qubits in heterogeneous quantum systems where frequency conversion and quantum interference are unified.

Figures (4)

• Figure 1: The principle of the Hong-Ou-Mandel (HOM) effect with a) photons of the same color and with b) photons of different color. In both scenarios, the probability amplitudes for four possible paths need to be considered when calculating the state at the output. Paths 2 and 3 are indistinguishable and interfere destructively and consequently -- in the case of an ideal implementation and a balanced beam splitter -- the measurement outcomes associated with the output photons having different color disappear.
• Figure 2: Experimental setup for the Hong-Ou Mandel effect in color. A blue pump photon decays via spontaneous parametric down-conversion into a photon pair at red and telecommunication wavelengths, with the photon frequencies matched to the transition frequencies of a quantum frequency converter. A tunable relative time delay $\Delta\tau$ is introduced for the photon pair in the telecom arm. Subsequently, both photons are aligned into a quantum frequency converter operating at 50% conversion efficiency. To resolve the coincidence events within both colors, the red and telecom output modes are further split via ordinary balanced beam splitters before detection. Coincidence counts are recorded with four single photon detectors as a function of the relative time delay $\Delta\tau$.
• Figure 3: Transition probability of the active beam splitter. The experimentally inferred transition probability was fitted with $\sin^2(\pi/2\sqrt{P_p/P_{max}})$albota, predicting 100% transition probability at $P_{max}=149 \hbox{W}$ and a 50% transition probability at $\sim 37$W.
• Figure 4: Hong-Ou-Mandel (HOM) dip in color: (a) measured telecom-red coincidences as a function of the optical delay in the telecom arm. The visibility of the fitted Gaussian is (91.4 $\pm$ 0.3)% without the subtraction of accidental coincidences. The width of the fitted Gaussian is (7.78 $\pm$ 0.06) ps (FWHM). Measured (b) telecom-telecom and (c) red-red coincidences as a function of the optical delay in the telecom arm. The corresponding widths of the fitted Gaussian anti-dips are for telecom (8.99 $\pm$ 0.11) ps and for red (5.99 $\pm$ 0.34) ps.