arXiv:2609.26889
Deterministic photon waveform adaptation for quantum connectivity
Jeffrey Mohan, Jérémy Berroir, Filippo Borselli, David Libault, Ferhat Loubar, Kilian Müller, Jed Rowland, Félix Hoffet, Eleni Diamanti, Tom Darras, Tommaso Mazzoni, Julien Laurat
Abstract
The scalability of quantum technologies will depend on the ability to interconnect independent quantum systems through photonic channels. However, heterogeneous quantum platforms emit and absorb photons with widely differing properties, severely limiting inter-node interference and modular connectivity. Here we demonstrate a cold-atom optical quantum memory that simultaneously achieves near-unity storage-and-retrieval efficiency and deterministic temporal adaptation of single photons between arbitrary and programmable input and output pulse waveforms. Operating at high optical depth and within a fully integrated architecture, the system can store photons with durations spanning over three orders of magnitude and reshape them arbitrarily without compromising efficiency, achieving compatibility with many current platforms. By augmenting the role of a quantum memory from a passive storage element to an active programmable photonic interface, our results establish a key building block for scalable entanglement-based quantum networks and modular quantum computing architectures.
AI Summary
The paper addresses temporal-mode mismatch between photons from heterogeneous quantum platforms: can a quantum memory both store light efficiently and convert known input waveforms into programmable output waveforms? The implementation uses electromagnetically induced transparency in a cold $^{87}\mathrm{Rb}$ ensemble with optical depth approximately 750, housed in a transportable, fibre-coupled rack system. Modulating the control-beam power changes the retrieved pulse’s duration and shape. For arbitrary waveform conversion, an optimal-control algorithm jointly optimises writing and retrieval by solving Maxwell–Bloch equations and maximising the efficiency-weighted target-mode overlap. Measurements with weak coherent pulses containing approximately one photon on average give mean background-corrected storage-and-retrieval efficiency $\eta=90.4(1.6)\%$ and conditional waveform overlap $\mathcal{I}_c=97.9(2)\%$ across more than three orders of magnitude in bandwidth. Demonstrations include storing pulses as short as 20 ns, generating pulses as long as 30 μs, stretching and compressing Gaussian pulses, converting Gaussian and double-exponential inputs into an ion-emission-like waveform, splitting one input into up to ten output time bins, and storing two successive pulses. The operating memory lifetime is approximately 150 μs. Minimum output bandwidth is limited by this lifetime; maximum bandwidth is limited by modulator response. Other stated limitations include laser-lock noise, control leakage and amplified spontaneous emission, waveform-generator resolution, and calibration nonlinearity. The optimisation model neglects spin-wave dephasing. Phase control between time bins is proposed rather than demonstrated, and the experiments use weak coherent inputs rather than single-photon sources.
Central claim may not hold: The single-photon adaptation claim is supported by weak coherent-pulse measurements, without a demonstration that nonclassical statistics or entanglement survive the noisy memory channel. The experiments establish intensity-profile adaptation, while the quoted complex temporal-mode overlaps require an unverified temporal-phase assumption and programmable phase control is left to future work. The claimed 60-fold entanglement-rate enhancement does not follow from the stated 9% squared mode overlap and rate scaling, which permit at most an approximately 11-fold enhancement before memory losses. The printed adjoint polarization equation has the wrong decay sign relative to the displayed constrained functional, making the stated gradient calculation mathematically inconsistent.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.