Realization of a programmable multi-purpose photonic quantum memory with over-thousand qubit manipulations

TL;DR

This work tackles the need for a programmable, high-capacity photonic quantum memory suitable for quantum networks. It implements a random-access quantum memory (RAQM) using a 2D $^{87}$Rb atomic cloud with AOD addressing and EIT storage, achieving $72$ qubit cells across $144$ ensembles, coherence time $>0.5\ \mathrm{ms}$, and $1000$ on-demand operations. The authors demonstrate functional quantum data structures (queue, stack, buffer) and show storage/reshuffle of four entangled pairs with fidelities in the $83$–$92\%$ range, highlighting the memory’s versatility for entanglement distribution and routing. This programmable memory constitutes a key platform for future quantum repeaters and large-scale networks, with clear paths toward longer coherence, higher efficiency, and integrated quantum processing capabilities via lattice loading, wavelength conversion, and in-memory gate implementations.

Abstract

Quantum networks can enable various applications such as distributed quantum computing, long-distance quantum communication, and network-based quantum sensing with unprecedented performances. One of the most important building blocks for a quantum network is a photonic quantum memory which serves as the interface between the communication channel and the local functional unit. A programmable quantum memory which can process a large stream of flying qubits and fulfill the requirements of multiple core functions in a quantum network is still to-be-realized. Here we report a high-performance quantum memory which can simultaneously store 72 optical qubits carried by 144 spatially separated atomic ensembles and support up to a thousand consecutive write or read operations in a random access way, two orders of magnitude larger than the previous record. Due to the built-in programmability, this quantum memory can be adapted on-demand for several functions. As example applications, we realize quantum queue, stack, and buffer which closely resemble the counterpart devices for classical information processing. We further demonstrate the synchronization and reshuffle of 4 entangled pairs of photonic pulses with probabilistic arrival time and arbitrary release order via the memory, which is an essential requirement for the realization of quantum repeaters and efficient routing in quantum networks. Realization of this multi-purpose programmable quantum memory thus constitutes a key enabling building block for future large-scale fully-functional quantum networks.

Figures (19)

• Figure 1: Multi-purpose photonic quantum memory and experimental setup. (a) Quantum network architecture with a communication backbone based on a quantum repeater, and many local quantum nodes. The red dashed lines demonstrate the connection scheme for distributed quantum computing based on this architecture. (b) Different types of quantum data containers and their utilizations. (c)-(d) Schematic of the multipurpose quantum memory. The system consists of two encoding converters, two addressing units, and a two-dimensional memory array based on a $^{87}$Rb atomic cloud. Both the input and output qubits are encoded in polarization.
• Figure 2: Performance of individual memory units. (a),(b), Coherence time and storage efficiency of the 144 micro-ensembles. (c), The storage and retrieval of an arbitrary polarization qubit. We first convert a polarization qubit $a|H\rangle+b|V\rangle=cos(\theta)|H\rangle+e^{i\phi}sin(\theta)|V\rangle$ to a time-bin qubit with an encoding converter. Then the time-bin qubit is further converted into a path qubit to be stored in two memory cells and finally read out and converted back into a polarization qubit $a_4|H\rangle+b_4|V\rangle=cos(\theta_4)|H\rangle+e^{i\phi_4}sin(\theta_4)|V\rangle$, which has a high overlap fidelity with the input state if $\theta_4=\theta$ and $\phi_4=\phi$. This can be achieved by calibrating the amplitude and phase in the read-out AODs. (d), The index of the $72$ qubit cells paired from 144 micro-ensembles. (e)-(f), State fidelity of the retrieved qubit in six typical qubit cells $1$, $4$, $20$, $22$, $34$ and $53$. Fidelities for different input polarizations with $15\mu$s storage are in (e), and average fidelities over $4$ polarizations at different storage time are in (f). The crosstalk error induced by accessing nearby micro-ensembles of the stored qubit. The green micro-ensemble pair represents the qubit for storage, and we iteratively access the $6$ neighboring regions marked grey for different rounds to characterize the fidelity loss. Note that the storage time is fixed to $55\,\mu$s with variable round number. A round of operations on the 6 neighbors leads to about $1\%$ infidelity. The error bars represent one standard deviation in (e), (f), and (g).
• Figure 3: Random access quantum memory. (a) Snapshot of the RAQM with $72$ qubit cells. (b) A random sequence of $500\,\mu$s. All the domains in each $2\,\mu$s operation are randomly generated including write-or-read (W/R), qubit cell index, and input polarization. (c) A similar random sequence of $1000$ write or read operations with a length of $2\,$ms. We update the storage time of each stored qubit within a scrolling window of $500\,\mu$s, and execute forced read-out to the stored qubits about to expire in the next clock cycle. (d)-(h) The filling number of the memory, the total accesses in each qubit cell, the distribution of storage time of all output qubits, and the distribution of the state fidelities of all output qubits, for both $500\,\mu$s (red) and $2\,$ms (blue) cases. The inset in (g) shows the forced read-out happens at a chance of $0.4\%$. The inset in (h) demonstrates the corresponding fidelity for each input polarization. The error bars represent one standard deviation.
• Figure 4: Quantum queue, stack, and buffer. (a) Quantum queue (FIFO). We demonstrate a special sequence of $72$ contiguous enqueues followed by $72$ contiguous dequeues under the rule of a queue. (b)-(e) The polarization distribution of the input states and average fidelity for each input polarization, the real-time filling number, the distribution of storage time for individual qubits, and the fidelities for individual qubits and corresponding threshold for quantum storage (the orange lines) jiangnan, respectively. (f)-(j) Schematic and results for a quantum stack. Again we use a special sequence in which we first push $72$ qubits and then pop them out in a reversed order. (k)-(o) Schematic and results for a quantum buffer. The distribution of time intervals between successive input qubits are shown in (l). The error bars represent one standard deviation in (b), (e), (g), (j), and (o).
• Figure 5: Storage and reshuffle of multiple entangled photon pairs. (a) Four Einstein-Podolsky-Rosen (EPR) pairs are sequentially generated through the DLCZ protocol and are heralded by the detection of signal photons. The $4$ idler photons with random arrival times are stored into the quantum memory and are read out together in an arbitrarily chosen order ($2$-$4$-$1$-$3$). (b) The sequence of the experiment. (c) The distribution of the random arrival times of the four idler photons $I_1$, $I_2$, $I_3$, and $I_4$. Note that the four EPR pairs are generated sequentially so that the distributions are not independent. The error bars represent one standard deviation. (d) The entanglement fidelity of the four EPR pairs $S_1$-$I_1'$, $S_2$-$I_2'$, $S_3$-$I_3'$, and $S_4$-$I_4'$ after storage and reshuffle. The fit follows an exponential decay with the gray shaded area representing the $68$% confidence interval. The error bars also represent $68$% confidence interval. (e) Two potential applications in a quantum network, i.e. entanglement connection of two quantum repeater segments and entanglement routing between different network nodes.