Entanglement between a Telecom Photon and an On-Demand Multimode Solid-State Quantum Memory

Entanglement between photons at telecommunication wavelengths and long-lived quantum memories is one of the fundamental requirements of long-distance quantum communication. Quantum memories featuring on-demand readout and multimode operation are additional precious assets that will benefit the communication rate. In this Letter, we report the first demonstration of entanglement between a telecom photon and a collective spin excitation in a multimode solid-state quantum memory. Photon pairs are generated through widely nondegenerate parametric down-conversion, featuring energy-time entanglement between the telecom-wavelength idler and a visible signal photon. The latter is stored in a ${\mathrm{Pr}}^{3+}:{\mathrm{Y}}_2{\mathrm{SiO}}_5$ crystal as a spin wave using the full atomic frequency comb scheme. We then recall the stored signal photon and analyze the entanglement using the Franson scheme. We measure conditional fidelities of 92(2)% for excited-state storage, enough to violate a Clauser-Horne-Shimony-Holt inequality, and 77(2)% for spin-wave storage. Taking advantage of the on-demand readout from the spin state, we extend the entanglement storage in the quantum memory for up to $47.7\mu \mathrm{s}$, which could allow for the distribution of entanglement between quantum nodes separated by distances of up to 10 km.

Figure 1

(a) Experimental setup. Entangled photon pairs generated at the source are separated with a dichroic mirror (DM). Telecom photons are sent directly to an interferometer for entanglement analysis. Signal photons are stored in the quantum memory crystal (QM), and are later filtered and analyzed in the filter crystal (FC). PPLN: periodically poled lithium niobate; FCav: Fabry-Perot filter cavity; BPF: bandpass filter; PZ: piezo actuator; PC: polarization controller; SPD: single-photon detector. (b) Energy level scheme of ${\mathrm{Pr}}^{3+}:{\mathrm{Y}}_2{\mathrm{SiO}}_5$, where the relevant transitions are indicated. (c) Coincidence histograms showing the time difference between the detection of idler and signal photons. The darker regions indicate the portion considered for analysis. We plot the three cases that we analyze in the text: input photons, AFC storage for $10\mu \mathrm{s}$, and spin-wave (SW) storage with a $T_s$ of $6.9\mu \mathrm{s}$.

Figure 2

Visibility fringes and relevant coincidence histograms for maximum and minimum values (darker regions represent the coincidences window used). (a) AFC storage. We varied the phase between the short and long paths of the AFC-based interferometers for two settings of the idler interferometer differing by $\pi /2$. See [51] for the normalization of the AFC coincidences. (b) Semiconditional SW storage. We scanned the phase between the short and long paths of the idler interferometer for two settings of the AFC-based interferometer differing by $\pi /2$. The shaded area represents the maximum value that we expect for the ideal case, with a span of 1 sigma, while the dashed gray line is the minimum one.

Figure 3

(a),(b) Variation of the storage efficiency and second-order cross-correlation with $T_s$. They follow a Gaussian trend, which we use for the fit, obtaining an inhomogeneous broadening of the spin state of 16.1(7) and 14.8(9) kHz, respectively. The shaded area corresponds to 1 standard deviation of the error of the fit. (c) Variation of the entanglement visibility with $T_s$. The shaded area represents our model accounting for imperfect analyzers, while the dashed lines correspond to the upper and lower bounds of the ideal case, only limited by photon statistics. The dotted gray line represents the bound for separable states.