An Atomic Interface for High-Dimensional Temporal Mode Quantum Networks

Abstract

Temporal modes of photons are a promising encoding scheme for high-dimensional quantum networks due to their high channel capacity and fiber compatibility. However, realizing their full potential requires devices capable of synchronizing, processing and interfacing these modes across photonic and atomic bandwidths. In this work, we demonstrate a programmable high-dimensional temporal mode processor using a Raman quantum memory in warm cesium vapor. We exploit the single-mode nature of the Raman interaction kernel, dynamically shaping the control field to synthesize a tunable coherent filter that selectively addresses specific temporal waveforms. This mechanism enables on-demand storage, filtering, and conversion, providing a coherent interface between MHz- and GHz-bandwidth modes. We validate the platform's selectivity across a basis of 30 orthogonal Hermite-Gaussian modes and certify high-fidelity quantum operation via 5-dimensional process tomography. By combining deterministic mode conversion with bidirectional bandwidth interfacing, we establish the Raman memory as a critical active node for scalable quantum information processing.

Figures (12)

• Figure 1: Conceptual overview of the Raman quantum memory as a versatile high-dimensional temporal mode processor and network interface. The Raman quantum memory is capable of processing high-dimensional temporal modes, such as orthogonal Hermite-Gaussian wavepackets, by mapping them into and out of an atomic spin-wave via a programmable two-photon transition. By tailoring the classical control field, the memory performs on-demand retrieval, coherent filtering, as well as deterministic mode and bandwidth conversion. This versatility enables direct interfacing between disparate quantum network nodes, bridging the bandwidth gap between broadband emitters (e.g., GHz-bandwidth quantum dots and parametric down-conversion sources) and narrowband matter-based systems (MHz-bandwidth atomic nodes).
• Figure 2: Temporal mode storage and mode crosstalk. (a) Crosstalk matrix for the first 30 Hermite-Gaussian temporal modes, represented by the storage efficiencies of different signal-control combinations. Rows correspond to input signal modes, and columns represent control write modes. (b) Examples of storage for different mode combinations. When the control (dashed black) is in the same mode as the input signal (gray), the difference between the input and the leaked signal (red) indicates successful storage. Conversely, when the control and signal modes are orthogonal, minimal storage occurs.
• Figure 3: Storage process matrix. Real and imaginary components of the reconstructed, efficiency-normalized storage process matrix $\boldsymbol{Q}$. The gray bars represent the ideal process matrix corresponding to a perfect single-mode temporal filter, which is purely real.
• Figure 4: Mode conversion and optimized retrieval. (a) Total efficiency for different combinations of mode conversions within the first 8 Hermite-Gaussian modes. (b) Example of retrieval mode optimization converting an HG$_1$ input to an HG$_3$ output. The non-optimized retrieval is shown in blue, and the optimized retrieval is shown in orange. A standard HG$_3$ intensity profile is included for comparison. The input signal (gray) and corresponding leaked signal (red) are scaled by a factor of 1/5.
• Figure 5: Bandwidth conversion. (a) Conversion efficiency of the Raman memory as a function of the bandwidth-compression factor (larger factors correspond to longer output durations). The input signal duration is 10 ns. Blue and red points denote read control pulses with constant peak amplitude and constant energy, respectively. The solid black line shows the response of an ideal passive top-hat frequency filter. Error bars are smaller than the markers. (b) Temporal traces demonstrating conversion using constant-energy control pulses. The 10 ns input yields output durations of 8, 20, 40, 60, and 100 ns (light to dark red). The input signal (gray) and corresponding leaked signal (dark red) are scaled by a factor of 1/20.