Postselected Entanglement between Two Atomic Ensembles Separated by 12.5 km

Quantum internet gives the promise of getting all quantum resources connected, and it will enable applications far beyond a localized scenario. A prototype is a network of quantum memories that are entangled and well separated. In this Letter, we report the establishment of postselected entanglement between two atomic quantum memories physically separated by 12.5 km directly. We create atom-photon entanglement in one node and send the photon to a second node for storage via electromagnetically induced transparency. We harness low-loss transmission through a field-deployed fiber of 20.5 km by making use of frequency down-conversion and up-conversion. The final memory-memory entanglement is verified to have a fidelity of 90% via retrieving to photons. Our experiment makes a significant step forward toward the realization of a practical metropolitan-scale quantum network.

Figure 1

Experimental layout. Bird’s eye view (top) of the memory nodes (A and B) that are 12.5 km apart and connected with fibers of 20.5 km (17 km deployed together with a 3.5 km long fiber loop). Each node includes a setup with ${}^{87}\mathrm{Rb}$ atoms and a quantum frequency converter. Schematic (bottom) and corresponding level schemes (insets). In node A, a cavity-enhanced DLCZ-type setup is used to generate atom-photon entanglement. The write-out photon propagates along the clockwise mode of the cavity and emits from the partially reflective mirror (PR). The emitted photon is then converted to 1342 nm via DFG in a PPLN waveguide and transmitted to node B via the fiber channel. In node B, the photon is first converted back to 795 nm via SFG in another PPLN waveguide chip. A series of filters, including dichroic mirrors (DM), long-pass filters (LP), and bandpass filters (BP), are used to suppress the noise in each conversion module. (The types and quantities shown in the figure are not in one to one correspondence with the actual situation). The photon’s two time-bin modes are then converted to two spatial modes and stored in the atomic ensemble by applying a control pulse via the EIT mechanism. The time-bin to spatial conversion is achieved by a combination of a Pockels cell, an asymmetric Mach-Zehnder interferometer (AMZI), and a polarizing beam splitter (PBS). An etalon in the readout path of the EIT quantum memory is for blocking the control pulse leakage. QWP and HWP are quarter- and half-wave plates, respectively. Markers I, II, and III indicate three checkpoints for the write-out photon during the characterization of the photon transfer and the EIT storage. Map data is from Google and Maxar Technologies.

Figure 2

Benchmarking the DLCZ quantum memory. (a) Scheme of the spin-wave freezing. After the spin wave was created, a Raman $\pi$ pulse that couples the $|s⟩$ to $|s^{\prime }⟩$ transition via a two-photon Raman process is applied. The Raman process introduces a new momentum $\hslash {\overrightarrow{k}}_R$ to the $|s⟩/|s^{\prime }⟩$ state atom, resulting in an altered spin wave with the wave vector $\mathrm{\Delta }{\overrightarrow{k}}^{\prime }=\mathrm{\Delta }\overrightarrow{k}+{\overrightarrow{k}}_R\approx 0$. The Raman $\pi$ pulse is applied again before the readout to recover the wave vector. (b) Relative retrieval efficiency ($\eta /{\eta }_0$) as a function of the readout delay. The data for two spin-wave states are dark and light blue for the spin-wave freezing case and dark and light red for the spin-wave freezing free case. The purple curve is the theoretical expectation of the spin-wave freezing (see Supplemental Material [43]). The orange bar indicates the entanglement distribution time in the experiment. (c) Measured atom-photon correlations at $\chi =5.4\%$ as a function of the readout delay. The oscillation of $⟨XX⟩$ originates from the time evolution of the atomic phase induced by the magnetic field. Error bars indicate one standard deviation of the photon-counting statistics.

Figure 3

Characterization of the photon transfer and the EIT quantum memory. (a) Red dots shows the measured SNR of the write-out photon before transfer (I), after transfer (II), and after EIT storage (III). The normalized cross-correlation function $g_{w,r}^{(2)}$ between the write-out and the read-out photon at the corresponding three conditions are shown in blue squares. (b) Measured $⟨ZZ⟩$ and $⟨XX⟩$ correlations of $|{\mathrm{\Psi }}_{\mathrm{ap}}⟩$ as a function of readout delay, i.e., the atomic phase evolution. Blue circles, red triangle, and green diamond refer to measuring the write-out photon at I, II, and III. The purple curve in each graph shows the fitting for data in case I. Error bars indicate one standard deviation of the photon-counting statistics.

Figure 4

Two-node entanglement. Results of the Bell test and the correlation measurement with the remote atom-atom entanglement $|{\mathrm{\Psi }}_{\mathrm{aa}}^{+}⟩$. Error bars indicate one standard deviation of the photon-counting statistics.