High-rate subgigahertz-linewidth bichromatic entanglement source for quantum networking

The generation of entangled photon pairs, which are compatible with quantum devices and standard telecommunication channels are critical for the development of long-range fiber quantum networks. Aside from wavelength, bandwidth matching, and high fidelity of produced pairs are necessary for high interfacing efficiency. High-rate, robust entanglement sources that satisfy all these conditions remain an outstanding experimental challenge. In this work, we study an entanglement source based on four-wave mixing in a diamond configuration in a warm rubidium vapor. We theoretically and experimentally investigate an alternative operating regime and demonstrate an entanglement source, which produces highly nondegenerate 795- and 1324-nm photon pairs. With this source we are able to achieve in-fiber pair-generation rates greater than ${10}^7/\mathrm{s}$, orders of magnitude higher than previously reported atomic entanglement sources. Additionally, given our source’s native compatibility with telecom infrastructure and atomic systems, it is a step towards scalable quantum networks.

Figure 1

Experimental setup for generation and analysis of entangled photon pairs. Inset: relevant rubidium level diagram for the four-wave mixing process. PBS, polarizing beam splitter; DM, dichroic mirror; BPF, bandpass filter; LCR, liquid crystal retarder; QWP, quarter-wave plate; HWP, half-wave plate; L, lens.

Figure 2

Heralding efficiency of the source as a function of the spectroscopically measured vapor-cell temperature, and the two-photon detuning, $\delta$. The single-photon detuning is fixed, $\mathrm{\Delta }/2\pi \approx 1.1$ GHz. The pump power is fixed at approximately equal to $250\text{μ}\mathrm{W}$. For each detuning and temperature we vary the coupling power to set the signal rate to approximately equal to $100\times {10}^3\mathrm{counts}/\mathrm{s}$.

Figure 3

Numerical simulations of the simplified three-level system in the near (blue) and far-off (orange) two-photon resonance regimes. Plot (a),(b) shows the scattering probability, unweighted (weighted) by the Maxwell-Boltzmann distribution, as a function of atomic velocity. (b) has been renormalized so that the peak weighted scattering probability is unity for ease of comparison. Dashed green line in (b) shows the Maxwell-Boltzmann distribution for vapor temperature used in the simulation. In both cases the pump detuning is fixed far from resonance ($\mathrm{\Delta }/2\pi =1150$ MHz), the pump and coupling Rabi frequencies are held at reasonable experimental values (${\mathrm{\Omega }}_p/2\pi ={\mathrm{\Omega }}_c/2\pi =350$ MHz), and temperature is fixed at $T=80{}^{\circ }\mathrm{C}$.

Figure 4

Scaling behavior of the source with pump and coupling power. (a) displays measured signal-idler coincidence rate as a function of coupling power, for various pump powers. We use a $\propto P_{\text{coupling}}{P}_{\text{pump}}$ fit that accounts for the finite detector dead times. We calculate a measured scaling constant of the $|VV⟩$ mode coincidence rate approximately equal to $3\times {10}^5/\mathrm{s}/{\text{mW}}^2$ (approximately equal to $6\times {10}^5/\mathrm{s}/{\text{mW}}^2$ when accounting for detection efficiencies). Inset table displays the coincidence values displayed in the plot. (b) shows the peak value of the signal-idler cross-correlation, $g_{\mathrm{si}}$, as a function of the $|VV⟩$ mode coincidence rate. Orange line is a $\propto 1/\text{coincidences}$ fit, taking into account the finite-detector dead time in the system. Inset displays a typical $g_{\mathrm{si}}$ curve, using 100-ps bins. In both plots error bars, arising from statistical uncertainties, are smaller than the data points.

Figure 5

Real and imaginary parts of the maximum-likelihood density-matrix reconstruction of the signal-idler state. Lower bound on the fidelity to the expected $|{\mathrm{\Phi }}_{+}⟩$ Bell state is 95%.