Temporal Multimode Storage of Entangled Photon Pairs

Multiplexed quantum memories capable of storing and processing entangled photons are essential for the development of quantum networks. In this context, we demonstrate and certify the simultaneous storage and retrieval of two entangled photons inside a solid-state quantum memory and measure a temporal multimode capacity of ten modes. This is achieved by producing two polarization-entangled pairs from parametric down-conversion and mapping one photon of each pair onto a rare-earth-ion-doped (REID) crystal using the atomic frequency comb (AFC) protocol. We develop a concept of indirect entanglement witnesses, which can be used as Schmidt number witnesses, and we use it to experimentally certify the presence of more than one entangled pair retrieved from the quantum memory. Our work puts forward REID-AFC as a platform compatible with temporal multiplexing of several entangled photon pairs along with a new entanglement certification method, useful for the characterization of multiplexed quantum memories.

Figure 1

Conceptual scheme of the experiment. Temporally multiplexed photon pairs generated from spontaneous parametric down-conversion (SPDC) are stored in a multimode quantum memory (QM) and released after a predetermined time ${\tau }_M$. The pump is pulsed with a 10 MHz repetition rate and with duration ${\tau }_p$. Two polarization-entangled pairs are generated probabilistically by the same pulse and separated by a time $\delta t\le {\tau }_M$. The duration of the pulse equals the storage time such that the stored photons are both in the QM for a time (${\tau }_M-\delta t$). To certify entanglement after absorption and remission by the QM, we analyze the correlations in polarization of the four-photon state.

Figure 2

Detailed experimental setup. Our source of polarization-entangled photon pairs is based on parametric down-conversion inside two periodically poled nonlinear waveguides (ppnw) phase matched for Type-0 down-conversion. Both crystals are coherently pumped by a 532 nm continuous wave (cw) laser. The pump laser is intensity modulated by an acousto-optic modulator (AOM) at a 10 MHz repetition rate, producing a train of 50 ns square pulses. Signal (883 nm) and idler (1338 nm) photons are spatially separated by a dichroic mirror (DM) and spectrally filtered using a cavity and a volume Bragg grating (VBG) in each output mode. The signal mode is coupled to a polarization-preserving solid-state quantum memory (QM), based on two ${\mathrm{Nd}}^{3+}:{\mathrm{Y}}_2{\mathrm{SiO}}_5$ crystals separated by a half-wave plate. To increase the storage efficiency, we use a double-pass configuration through the QM. Finally, we analyze the polarization states of the photon pairs with a set of half-wave (HWP) and quater-wave (QWP) plates, polarizing beam splitter (PBS) and single photon detectors $D_{\pm }^{(s,i)}$, fiber-coupled superconducting nanowires on the idler side and free-space avalanche photodiodes on the signal side.

Figure 3

Temporal multimode storage. (a) The twofold coincidences between detections of the signal and idler photons as a function of the delay between two detection events. The first peak at 0 ns stems from the signal photons not absorbed by the QM, while the second peak at 50 ns corresponds to the signal photons absorbed by the QM and released after the storage time. Such an histogram is accumulated for each pair of detectors between signal and idler photons. A coincidence window of 4 ns (depicted by the dotted lines) is used to calculate the rates. (b) The total four-fold coincidences collected during the experiment is plotted as a function of the delay $\delta t$ between photon pairs. The events corresponding to the stored state (2) inside the storage time of the QM are delimited by dotted vertical lines from 5 to 50 ns. There are 9 distinguishable time divisions, demonstrating storage of 10 modes containing single-photon excitations. For longer delay ($>50\mathrm{ns}$), two photons do not overlap at any time in the QM. Error bars represent one standard deviation assuming Poisson noise for the counts.

Figure 4

Entanglement witness. From the measured four-fold correlations, we can compute our central figure of merit $\mathcal{T}$ [Eq. (3)]. A value of $\mathcal{T}$ implies that the operator ${\mathcal{W}}_k(\mathcal{T})$ has an expectation value equal to exactly zero. For a sufficiently large value of $\mathcal{T}$, this implies that the expectation value of $W_1(\mathcal{T})$ [or $W_2(\mathcal{T})$] is negative. Since we can prove that ${\mathcal{W}}_1$ is an entanglement witness and ${\mathcal{W}}_2$ a Schmidt number witness, we can thus conclude on one or two entangled pairs.