Long-Distance Distribution of Atom-Photon Entanglement at Telecom Wavelength

Entanglement between stationary quantum memories and photonic channels is the essential resource for future quantum networks. Together with entanglement distillation, it will enable efficient distribution of quantum states. We report on the generation and observation of entanglement between a ${}^{87}\mathrm{Rb}$ atom and a photon at telecom wavelength transmitted through up to 20 km of optical fiber. For this purpose, we use polarization-preserving quantum frequency conversion to transform the wavelength of a photon entangled with the atomic spin state from 780 nm to the telecom $S$ band at 1522 nm. We achieve an unprecedented external device conversion efficiency of 57% and observe an entanglement fidelity between the atom and telecom photon of $\ge 78.5\pm 0.9\%$ after transmission through 20 km of optical fiber, mainly limited by decoherence of the atomic state. This result is an important milestone on the road to distribute quantum information on a large scale.

Figure 1

Experimental setup and entanglement generation scheme. A single ${}^{87}\mathrm{Rb}$ atom (upper left side), serving as quantum memory, is stored in the focus of a dipole trap (850 nm wavelength, $2.05\mu \mathrm{m}$ waist, and 42 mW power), where a high-NA objective collects the atomic fluorescence. The atom-photon entanglement is generated in the spontaneous decay following the excitation to the state $5{{}^{2}P}_{3/2}$ $|F^{\prime }=0,m_{F^{\prime }}=0⟩$. The MEMS switch guides the emitted photons toward the frequency converter where the 780 nm single photons are overlapped with 1600 nm pump light within a PPLN waveguide in a Sagnac-type interferometer to transfer the entanglement to 1522 nm photons. In the polarization analyzer, the single photons are first spectrally filtered by a Fabry-Perot filter cavity (FC), volume Bragg grating, bandpass filter (BPF), and short-pass filter (SPF). Next, after setting the analysis basis with a half-wave plate (HWP) and a quarter-wave plate (QWP) and splitting the polarization components by a Wollaston prism, the single photons are detected with two SNSPDs. Classical reference light, inserted along the readout path, can be coupled out with a flip mirror to analyze and compensate for polarization drifts. Further abbreviations used are mirror (M) and dichroic mirror (DM).

Figure 2

External device efficiency of the frequency converter. The external device efficiency ($\eta$) of the two polarization components depends on the pump power ($P$) in the respective arm. The data are fitted with $\eta (P)={\eta }_{\max }{\sin }^2(\sqrt{{\eta }_{\mathrm{nor}}P}L)$ [38]. The power in each arm is set to the operating point such that both conversion efficiencies are equal and one efficiency is maximized, 175 and 189 mW for the $p$- and $s$-polarization arms, respectively. At this point, the external device efficiency equals 57%.

Figure 3

Observation of atom-photon entanglement over 20 km of optical fiber ($A$). (a) Detection time histogram of the frequency converted photons. Within a hardwired acceptance window of 50 ns, indicated with dashed lines, approximately $2/3$ of the converted single photons were accepted. Note that the QFC does not influence the photon shape [25]; see Ref. [20] for an unconverted photon shape. (b) The corresponding atom-photon state correlations in two bases ($H/V$ and $D/A$) for varying atomic analysis angles (i.e., readout polarization, where 0° corresponds to vertical polarization). The sinusoidal fits give estimated visibilities of $73.4\pm 2.0\%$, $89.6\pm 1.1\%$, $72.5\pm 1.1\%$, and $68.6\pm 4.1\%$ for horizontal $|H⟩$, vertical $|V⟩$, diagonal $|D⟩$, and antidiagonal $|A⟩$ photonic linear polarization states, respectively. This results in an estimated state fidelity of $\ge 78.5\pm 0.9\%$.

Figure 4

Expected atom-photon and atom-atom entanglement fidelities. Points marked $A$, $B$, and $C$ correspond to the measurements presented in the text and summarized in Table 1. The expected fidelity for short distances ($<1\mathrm{km}$) is mainly limited by the imperfect atomic state preparation and readout. For distances $<100\mathrm{km}$, the fidelity is also reduced by atomic state decoherence due to position-dependent dephasing, which could be strongly suppressed in a future setup. For distances $>100\mathrm{km}$, detector dark counts will eventually limit the fidelity.