Efficient integrated quantum memory for light

TL;DR

This work demonstrates highly efficient integrated quantum memories in two architectures—a waveguide-cavity (WGC) and a fiber-based cavity (FBC)—using rare-earth Eu$^{3+}$:Y$_2$SiO$_5$ coupled to impedance-matched cavities. By applying absorption-enhanced AFC preparation and, in the WGC, Stark-shift-based on-demand retrieval, the devices achieve record efficiencies up to $80.3(7)\%$ for weak coherent pulses and $69.8(1.6)\%$ for telecom-heralded single photons, while storing up to 20 temporal modes with an average of $51.3(2)\%$. The thin-membrane Eu$^{3+}$:Y$_2$SiO$_5$ architecture additionally offers spectral tunability via variable strain, enabling flexible interfacing for quantum networks. Collectively, these results combine high efficiency, large multimode capacity, and spectral tunability in compact on-chip memories, providing a versatile hardware foundation for scalable quantum repeaters and chip-scale photonic processors, with performance surpassing the no-cloning limit of $50 ext{%}$ in a micro-volume device.

Abstract

Scalable implementation of quantum networks and photonic processors require integrated photonic memories with high efficiency, yet current integrated systems have been limited to storage efficiencies below 27.8%. Here, we demonstrate highly efficient integrated quantum memories based on rare-earth-iondoped crystals coupled with impedance-matched microcavities, realized in two novel architectures: 200-micrometer-thin membranes of Eu3+:Y2SiO5 integrated with fiber-based microcavities, and waveguide-based cavities fabricated using femtosecond lasers. Our approach achieves reliable integrated quantum storage with record efficiencies of 80.3(7)% for weak coherent pulses and 69.8(1.6)% for telecom-heralded single photons, alongside the storage of 20 temporal modes with an average efficiency of 51.3(2)%. Moreover, the thin-membrane Eu3+:Y2SiO5 architecture enables spectrally tunable efficient quantum storage via variable strain, providing a flexible interface for quantum networks. By combining high efficiency, large multimode capacity, and tunability, our devices establish a versatile hardware foundation for scalable quantum repeaters and chip-scale photonic processors.

Figures (15)

• Figure 1: The devices for efficient integrated quantum memories based on rare-earth-ion doped crystals. a, Efficient quantum memories based on the waveguide cavity (WGC) fabricated on $^{151}$Eu$^{3+}$:$\mathrm {Y_2SiO_5}$ crystals. Laser-written optical waveguides are coated with reflective layers at both ends to form optical cavities. On-chip coplanar electrodes deliver electrical pulses, enabling active control over the readout times of stored photons. Input and readout photons are separated by a polarizing beam splitter (PBS) depending on their different polarization states, with multiplexing achieved in the time domain. b, Efficient quantum memories based on the fiber microcavity (FBC). This cavity consists of a fiber concave mirror and a 200-µ m-thin membrane $^{151}$Eu$^{3+}$:$\mathrm {Y_2SiO_5}$ crystals with reflective coatings on its outer surface. The membrane is bonded to a sapphire substrate with a hole. The concave fiber mirror fixed on the stacked shearing piezoelectric transducer (SS-PZT) maintains a distance of approximately 100 µ m from the membrane surface. The input and output signal photons are also separated using a PBS.
• Figure 2: Efficient and multiplexed single-photon-level storage in the WGC. a, Normalized photon counting histograms for the 1st AFC echo. The red solid line represents the 1st AFC echo, compared to the reference input signal (black solid line with hollow squares) at an average photon number per pulse $\mu=0.35$. Noise measurements with zero input are shown as a blue dot-dashed line. The storage efficiency of the 1st AFC echo is $80.3(7)\%$, while the simulation (green dashed line) predicts a slightly higher efficiency of $80.9\%$. b, Photon counting histograms of higher-order AFC echoes with $\mu=0.35$. Two electric pulses actively control the readout times through Stark-shift induced interference. The fitted efficiency decay (pink solid line) yields an AFC finesse of $F_{\mathrm{AFC}}$ of 11.9(1), consistent with the expected value for the impedance-matched condition. c, On-demand storage of time-bin qubits. Three input states $\left\vert e\right\rangle$, $\left\vert l\right\rangle$, and $\left\vert e\right\rangle+\left\vert l\right\rangle$ are presented in the histogram. The average photon number per qubit $\mu_q = 0.35$, with an average storage efficiency of 62.0(6)%. The inset shows interference measurements on the superposition state $\left\vert e\right\rangle+\left\vert l\right\rangle$, with constructive (brown solid line with hollow squares) and destructive interference (cyan solid line with hollow triangles). d, Storage of 8 temporal modes with an average efficiency of 52.9(2)%. The FWHM of the temporal modes is set as $0.23$ µ s with each mode containing $\mu=0.17$. The inset provides the interference between the retrieved multimode echoes and reference pulses with two initial phases differing by $\pi$, yielding an average visibility of $0.94(4)$.
• Figure 3: Efficient and spectrally-tunable quantum storage using $^{151}\mathrm {{Eu}^{3+}}$:$\mathrm {Y_2SiO_5}$ membranes coupled into the FBC. a, Photon counting histograms of storage for weak coherent pulses with $\mu=0.35$, with an efficiency of 75.2(5)% at 2 µ s, while the simulation (green dashed line) predicts a slightly higher efficiency of $78.4\%$. The inset shows efficiency decay fitted with an exponential decay function $\mathrm{exp}(-4 t/T_{\mathrm{AFC}}^{\mathrm{eff}})$, yielding an effective AFC lifetime $T_{\mathrm{AFC}}^{\mathrm{eff}}$ = 87.6(2.4) µ s. b, Storage of time-bin qubits with $\mu_q=0.7$. The inset shows the interference measurements for the superposition state $\left\vert e\right\rangle+\left\vert l\right\rangle$. c, Quantum storage of 20 temporal modes with average efficiency of 51.3(2)% at $\mu=0.7$. The inset shows the interference between retrieved multimode echoes and reference pulses, yielding an average visibility of 0.94(1). d, Photoluminescence excitation spectra of $^{151}\mathrm {{Eu}^{3+}}$:$\mathrm {Y_2SiO_5}$ membranes under variable strain. Lorentz fits reveal fluorescence centers at 516.8479 THz (black), 516.8479 THz (purple), 516.8521 THz (blue), 516.8532 THz (red), and 516.8593 THz (green). Quantum storage efficiencies achieved at the latter three frequencies are 62.1(6)%, 75.2(5)%, and 61.6(4)%, respectively.
• Figure 4: Efficient storage of telecom-heralded single photons in the FBC. a, A 421-nm laser pumps a nonlinear crystal, generating photon pairs at 1537 nm and 580 nm via cavity-enhanced spontaneous parametric down-conversion (cSPDC). Photon pairs are separated by a dichroic mirror (DM). The 1537-nm telecom-band photons are detected by a superconducting nanowire single-photon detector (SNSPD), providing heralding signals. The 580-nm signal photons are spectrally filtered using an optically-pumped filter crystal, resulting in a bandwidth of 4 MHz. After storage in the fiber microcavity (FBC) quantum memory, the readout photons are extracted by an optical circulator and detected by a silicon-based single-photon detector (Si-SPD). Coincidence measurements of the photon pairs are performed using a time-correlated single-photon counting (TCSPC) module. b, Photon counting histograms for storage of heralded single photons. The measured storage efficiency is 69.8(1.6)% at 2-µ s with a cross-correlation of 16.4(2) within a 123-ns window, confirming non-classical correlations between retrieved photons and telecom heralding photons.
• Figure 5: Performance overview of quantum memories for light. The horizontal axis represents the inverse of effective volume which corresponds to the number of potential memory units per volume of mm$^3$. The surveyed physical systems include cold atomic gases Cao:20Wang2019Cho:16, warm atomic vapors Guo2019Ma2022Hosseini2011, rare-earth ion (REI)-doped solids Hedges2010Duranti2024EfficientZhong2017NanophotonicCraiciu2019NanophotonicRakonjac2022StorageLiu2020On-demandSaglamyurek2011Seri2019Quantum, optical phonons in diamond England2013FromEngland2015Storage, and single atoms in cavities Korber2018Decoherence, with a focus on those that have demonstrated efficient or small-volume quantum storage. For each experimental work plotted, the reported efficiency corresponds to the maximum measured storage efficiency. The effective volume $V$ of a quantum memory is determined by multiplying the transverse mode area of the optical field with the effective device length (the medium length or the cavity length). The effective number of stored modes is calculated as the product of the number of independent modes and the corresponding storage efficiency, with the color depth of each data point representing this value.