Source-linked AI summary
Quantum memories: emerging applications and recent advances
Khabat Heshami, Duncan G. England, Peter C. Humphreys, Philip J. Bustard, Victor M. Acosta, Joshua Nunn, Benjamin J. Sussman
TL;DR
Quantum memories are needed to synchronize probabilistic photonic events and extend photonic quantum technologies, but practical devices face noise, bandwidth-matching, and implementation challenges. This review surveys memory protocols, physical platforms, recent experiments, and emerging applications. It reports progress including long-coherence solid-state memories and photon-storage demonstrations while identifying remaining barriers to practical use.
Problem
Photonic quantum technologies need quantum memories to synchronize probabilistic events, but practical devices face noise, bandwidth-matching, and implementation challenges.
Method
The review surveys optically controlled, engineered-absorption, and hybrid protocols alongside physical platforms, recent developments, and emerging applications.
Results
370±60 minutes of ground-state hyperfine coherence at 2 K was demonstrated in 151Eu3+:Y2SiO5, while photon-storage experiments addressed narrow linewidth and wavelength compatibility.
Takeaways & Limitations
Quantum memories are progressing toward advanced photonic technologies involving long-distance entanglement distribution, optical signal processing, and optical-to-microwave conversion.
Takeaways & Limitations
Technical challenges remain in storage lifetime, signal-to-noise ratio, duty cycle, efficiency, and implementation complexity.
Abstract
from arXiv · showhide
Quantum light-matter interfaces are at the heart of photonic quantum technologies. Quantum memories for photons, where non-classical states of photons are mapped onto stationary matter states and preserved for subsequent retrieval, are technical realizations enabled by exquisite control over interactions between light and matter. The ability of quantum memories to synchronize probabilistic events makes them a key component in quantum repeaters and quantum computation based on linear optics. This critical feature has motivated many groups to dedicate theoretical and experimental research to develop quantum memory devices. In recent years, exciting new applications, and more advanced developments of quantum memories, have proliferated. In this review, we outline some of the emerging applications of quantum memories in optical signal processing, quantum computation, and nonlinear optics. We review recent experimental and theoretical developments, and their impacts on more advanced photonic quantum technologies based on quantum memories.
1. Introduction
Quantum memories are central photonic quantum components because they store and recreate photonic quantum states while coherently manipulating, buffering, and re-timing signals. This review emphasizes recent advances and applications beyond quantum-repeater cryptography, including optical signal processing and other photonic technologies.
- Quantum memories store single-photon quantum states and recreate them as part of photonic quantum technologies.
- Quantum memories can extend quantum cryptography through repeaters and support larger photonic processing systems.They provide coherent manipulation, buffering, and re-timing of photonic signals.
- Their applications extend beyond entanglement-swapping synchronization to coherent light-matter interfacing, non-classical-state preparation, and optical readout.
- The review complements earlier reviews by examining recent advances and spectral/temporal signal-processing techniques that enable new applications.
- The paper surveys optical-control and engineered-absorption protocols, experimental platforms, and implementation challenges for practical photonic quantum memories.
2. Implementations
Quantum memory implementation requires matching a storage medium with a protocol that controls photon absorption and re-emission. The review organizes implementations across five physical platforms and stresses that application requirements determine the preferred system and protocol.
- A quantum memory controls the absorption and re-emission of a photonic qubit using a selected medium and protocol.
- The review groups quantum memories into rare-earth-ion doped solids, diamond color centers, crystalline solids, alkali vapors, and molecules.
- No protocol-platform combination yet provides all desirable properties in one package.
- Application requirements guide system choice: storage time matters for long-distance communication, while time-bandwidth product or multimode capacity may matter for local processing.
- The implementation overview identifies outstanding research challenges and ongoing efforts to address them.
2.1. Protocols
Quantum memory protocols follow optically controlled, engineered-absorption, or hybrid strategies. They trade on-demand retrieval and mode selectivity against control-induced noise, while engineered absorption offers multimode storage but constrained readout timing.
- Optically-controlled memories: Optically controlled memories use strong control pulses in a Λ-level system to mediate signal absorption and retrieval.
- Optically-controlled memories: Optically controlled memories provide on-demand retrieval up to the memory’s coherence time.
- Optically-controlled memories: Their intense control fields can generate noise photons and restrict storage to a single optical mode.
- Engineered absorption: Engineered-absorption memories require no control field, avoiding the same noise problems and supporting multiple temporal and spectral modes.
- Engineered absorption: Engineered-absorption readout is not truly on-demand because CRIB and AFC rephasing impose timing constraints.
- Hybrid Schemes: Hybrid schemes combine engineered absorption with optical control, including shelving-state approaches intended to enable on-demand readout.
- Alternative control schemes: Other proposals control transition dipoles, transition frequencies, phase matching, or spatial and temporal control to reproduce memory behavior.
2.2. Physical systems and their recent developments
Recent developments span solid-state and atomic platforms, with progress in multimode, long-coherence, microwave, and integrated quantum memories. Rare-earth crystals are especially notable for combining optical access, tailored broadening, and extended coherence, although efficiency and practical integration remain challenging.
- Rare-earth-ion doped solids: Rare-earth-ion doped crystals offer narrow homogeneous linewidths, optical access to electronic and nuclear spins, tailored inhomogeneous broadening, and integrable implementations.
- Rare-earth-ion doped solids: Rare-earth crystals support time-bin and polarization-qubit storage, including temporally multiplexed polarization qubits.
- Long coherence times: Long-distance quantum communication requires storage times exceeding milliseconds, motivating the strong coherence properties of cryogenic rare-earth crystals.
- Long coherence times: Spin-echo and dynamical-decoupling techniques enabled low-noise single-photon and spin-wave storage, including retrieval after about 1 ms.The cited experiments also demonstrated high signal-to-noise ratio and time-bin-qubit storage.
- Long coherence times: 370±60 minutes of ground-state hyperfine coherence at 2 K, combined with photonic-entanglement storage, supports reconsideration of long-distance entanglement distribution.
- Microwave storage and conversion: Rare-earth spin states enable microwave storage and optical-to-microwave conversion, including a 9 ms nuclear-spin coherence time in Nd3+:YSO.
- Efficiency and integration: High efficiency remains difficult to combine with multimode functionality and long storage time, motivating cavity-enhanced light-matter interfaces.
- Diamond color centers: NV centers support microwave photon storage and hybrid optical-to-microwave interconversion through coupling to superconducting qubits.
2.2.3. Raman scattering in solids
Raman-based memories in solids, vapours, and molecules extend quantum storage across bandwidths, lifetimes, spatial modes, and processing regimes, but each platform presents distinct trade-offs.
- Solids: Diamond’s 40 THz Raman shift and inhibited four-wave mixing enable broadband, room-temperature memory with quantum-level noise properties.Its 3.5 ps optical-phonon decay nevertheless limits communication and many processing applications.
- Solids: A modified Raman interaction produced optically controlled, continuously tunable slow light for femtosecond pulses in a potassium titanyl phosphate waveguide.A narrowband control pulse near two-photon Raman resonance created the dispersion responsible for slowing the signal.
- Alkali vapours: Alkali vapours provide high optical depth, long coherence times, and accessible near-infrared transitions for EIT, Λ-GEM, and Raman storage.Warm and ultracold operation trade atomic motion and collisional effects against longer coherence times.
- Alkali vapours: Vapour memories can store structured photons by mapping orbital angular momentum into the phase and amplitude of distributed ensemble excitations.Cesium-MOT storage demonstrated OAM qubits with raw average fidelity 92.5 ± 2% and efficiency η = 15 ± 2%.
- Alkali vapours: Cooling and optical trapping extend alkali-vapour coherence from microseconds to milliseconds, while dynamic decoupling reached 16 s for a rubidium clock transition.These techniques reduce atomic motion and compensate differential Stark shifts.
- Molecules: Hydrogen memories stored 100-fs pulses for nearly 1 ns and produced nonclassical correlations between approximately 150-fs Stokes and anti-Stokes photons.The results support molecules for ultrafast local processing when long storage times are unnecessary.
2.3.1. Single photon storage
True single-photon storage is limited by control-induced noise and mismatches between photon-source bandwidths and memory resonances. The review describes noise mechanisms and source-engineering strategies that address these constraints.
- Challenges: True single-photon storage is necessary for quantum communication and processing but is more vulnerable to spurious noise than bright-pulse storage.Noise photons contaminate the output and degrade storage fidelity.
- Noise: Intense read and write pulses generate noise through resonant fluorescence, thermal storage-state population, and spontaneous four-wave mixing.These photons can remain indistinguishable from the signal even when strong control fields are filtered.
- Noise: In warm memories, Doppler-broadened fluorescence requires large detunings, whereas ultracold or cryogenic substrates permit near-resonant operation with reduced fluorescence.For pulses ≲1 ns, time-gating can also mitigate fluorescence because it decays exponentially.
- Noise: Far-off-resonant Raman protocols can suffer significant four-wave-mixing noise, which can be reduced using polarization selection, dispersive phase matching, or non-collinear geometry.Large ground-state separation can make Raman coupling stronger than the four-wave-mixing pathway.
- Bandwidth matching: Single-emitter photons are narrowband but frequency-fixed, while heralded sources are tunable but typically have linewidths around THz, exceeding most memories’ ≲5 GHz bandwidth.This source-memory mismatch motivates filtering, cavity engineering, or memory-based photon generation.
- Bandwidth matching: Cavity-enhanced sources reduce linewidth and increase tunability, while atomic heralded sources are intrinsically matched to absorptive memories.These approaches have enabled interfaces with AFC and EIT memories.
- Bandwidth matching: Diamond Raman memory stored single photons with a 1.7 THz bandwidth, directly addressing the bandwidth of SPDC sources.Its broad optical response distinguishes it from memories limited to narrow dipole-transition bandwidths.
2.3.2. Telecom wavelength storage
Telecom-band memories are motivated by fiber-based long-distance communication, but direct storage remains difficult because suitable transitions and storage performance are limited. Frequency conversion, hybrid wavelengths, and narrowband sources provide alternatives.
- Motivation: Silica-fiber attenuation is lowest from approximately 1.2 to 1.6 µm, motivating quantum memories that operate in telecom bands for quantum repeaters.The S- and C-bands are especially widely used because of erbium-doped fiber amplifiers.
- Direct and indirect interfaces: Far-off-resonant Raman memories are broadly tunable and could operate at telecom wavelengths, but achieved storage times were not yet suitable for relevant applications.Indirect interfaces therefore use frequency conversion or teleportation.
- Direct storage: An erbium-doped crystal memory stored approximately 200 ns pulses with mean photon number n̄ = 0.6 for up to 600 ns, at peak efficiency 0.25%.The demonstration used CRIB at 2.6 K.
- Frequency conversion: Rubidium frequency conversion shifted single photons from 780 nm to 1367 nm using a resonantly enhanced four-wave-mixing transition.Atomic transitions constrain the conversion wavelength and bandwidth.
- Hybrid wavelengths: A telecom-visible entangled source can send the telecom photon through fiber while storing the visible photon locally, avoiding telecom-wavelength memory storage.This increases the number of repeater stations required per unit length.
- Narrowband sources: A cavity-enhanced SPDC source produced 1436 nm and 606 nm photons with approximately 2 MHz linewidth, enabling storage of the 606 nm photon for on-demand readout up to 4.5 µs.The visible wavelength was compatible with a Pr3+:Y2SiO5 AFC memory.
2.3.3. Integrability of implementations (Nano/micro fabricated systems)
Integrated quantum memories can improve efficiency and reduce control-power requirements, but fiber-scale implementations face coherence-lifetime and optical-depth constraints.
- Waveguides and optical fibers provide tight modal confinement over long interaction lengths, potentially increasing memory efficiency and reducing read/write-field power.
- HCPCFs support nonlinear-optics experiments with low-intensity light and offer a route toward integrated memories containing atoms or molecules.
- Optical depths above 3000 were achieved in a Cs-filled Kagome HCPCF using light-induced atomic desorption, which transiently increases fiber optical density.
- Wall collisions can flip atomic spins and limit coherence, restricting cesium HCPCF memory lifetime to approximately 100 ns at room temperature versus over 1 µs in bulk.
3. Emerging applications for quantum memories
Quantum memories extend photonic quantum technologies beyond synchronization by enabling temporal and spectral signal processing, scalable cluster-state generation, and storage-mediated optical nonlinearities. Reviewed developments include arbitrary signal reordering, wavelength and bandwidth manipulation, memory-based quantum computation schemes, and photon-level phase shifts.
- 3.1. Optical signal processing: Quantum memories provide buffering for temporal-mode control and can combine frequency and temporal manipulation on a single platform.
- 3.1. Optical signal processing: Combining FIFO and FILO retrieval enables arbitrary reordering of optical signals, with the technique applicable to single-photon pulses.
- 3.1. Optical signal processing: A Ti:LiNbO3 waveguide with multiple AFC channels can process optical signals using distinct comb spacings and phase modulation.
- 3.1. Optical signal processing: Raman-coupling temporal engineering enables broadband quantum memories and multipulse addressing for optical state engineering.
- 3.1. Optical signal processing: Heralded 720 nm photons were recalled with wavelength shifts of approximately ±9 nm, while bandwidth was manipulated over a 3.75 THz range.
- 3.2. Linear optical quantum computation: For arbitrary N-mode unitary operations, N quantum memories are sufficient, providing favorable scaling for time-encoded linear optical quantum computing.
- 3.2.2. Photonic cluster states for one-way quantum computing: Quantum memories can address two-dimensional time-frequency modes for optical continuous-variable cluster states, combining temporal and spectral selectivity.
- 3.2.2. Photonic cluster states for one-way quantum computing: Beam-splitter and two-mode-squeezing Raman interactions generate cluster states across d frequency modes and arbitrary time-bin modes using 7d −3 quantum memories.
4. Summary and outlook
Quantum memories have progressed from long-distance communication applications toward versatile components for photonic quantum information processing. Practical adoption still requires overcoming application-dependent technical challenges and adding functionalities such as tunability and multiplexing.
- Quantum memories now serve as controllable light-matter interfaces for optical quantum information processing beyond their initial communication role.
- Memory requirements vary by application, so practical designs must balance performance characteristics rather than optimize a universally ideal platform.
- Technical challenges remain in storage lifetime, signal-to-noise ratio, duty cycle, efficiency, and implementation complexity across many experimental platforms.
- Wavelength tunability, spectral and spatial multiplexing, temporal and spectral control, and wavelength conversion would make memories more versatile processing elements.