Phase-Stable Optical Fiber Links for Quantum Network Protocols

TL;DR

This work demonstrates phase-stable distribution of path-entangled single-photon pulses over deployed optical fiber links by adopting precision time/frequency metrology methods from optical clocks. A co-propagating, temporally multiplexed classical stabilization channel achieves high phase stability while maintaining strong quantum/classical isolation, enabling Mach-Zehnder interference with fidelities exceeding 0.99 over 2.1 km links. Key results include optical timing jitter below 100 as over 10 minutes, path-indistinguishability above 99.6%, and quantum/classical isolation surpassing $8\times 10^{10}$, underscoring the viability of scalable, high-rate quantum networks. The study also analyzes phase-noise dynamics, fundamental limits of stabilization in fiber, and practical paths for extending to multi-node networks with improved passive stabilization and alternative fiber technologies.

Abstract

We demonstrate the distribution of single-photon-level pulses from a mode-locked laser source over a phase-stable fiber link, achieving an optical timing jitter of less than 100 as over 10 minutes of data accumulation. This stability enables a fidelity greater than 0.998 between two stabilized 2.1 km long deployed fiber links. Building on time and frequency metrology techniques traditionally used for high-stability optical atomic clock signal distribution, we use time and frequency multiplexing to achieve an isolation of quantum and classical channels by better than $10^{10}$. Our results mark a necessary step towards scalable, high-rate quantum networks with a provable quantum advantage.

Figures (7)

• Figure 1: A two-node path-entangled quantum network with a classical channel (purple lines) and quantum channel (red lines). A classical laser is used to stabilize network fiber links and to synchronize single-photon events at each node. A quantum repeater helps overcome network link losses and heralds entanglement. Entanglement heralding and local state projective measurements are recorded locally and transmitted over a different classical communication line (not shown).
• Figure 2: Experimental setup for the co-propagation of a quantum single-photon channel and a classical single-frequency phase-stabilization channel. The channels are isolated from one another by time-interleaving the classical and quantum channels using a mechanical chopper. The quantum signal, generated by attenuating a 500 MHz repetition rate, 1550 nm mode-locked laser to an average of $<1$ photon per pulse, is combined with the classical channel via a 50/50 optical beamsplitter. Length compensation is achieved using a fiber stretcher and motorized delay line. Two independently stabilized fiber links are employed to assess the performance of stabilization on both channels by measuring residual phase noise power spectral density on the classical channel and optical fringe contrast on the quantum channel. SNSPD - superconducting nanowire single-photon detector, AOM - acousto-optic modulator, PID - proportional/integral/derivative loop filter
• Figure 3: (a) Phase noise power spectral density (PSD) of the 1550 nm optical beatnote between two stabilized optical fiber links (2.1 km and 2 km), showing the frequency-dependent phase fluctuations as a function of chopping rate. (b) Corresponding integrated root-mean-square (RMS) phase deviation, illustrating the cumulative RMS phase noise as a function of integration time.
• Figure 4: Phase noise power spectral density of the unlocked 2 km fiber, $S_\text{unlocked} (f)$ (black), the theoretical best locked fiber $S_\text{locked} (f)$ (gray), and the experimental locked fiber without optical chopping (purple).
• Figure 5: Demonstration of quantum-compatible indistinguishability between two 2.1 km stabilized fibers arranged in a Mach-Zehnder interferometer configuration. A free space optical path length is swept via piezo stage (see Fig. \ref{['fig:setup']}) while an SNSPD detects single photons out of one of the interferometer ports for four different numbers of average photons. Sinusoidal fits show visibilities of 99.61 $\pm$ 0.035 % (average of 0.9 photons/pulse), 99.72 $\pm$ 0.076 % (0.08 photons/pulse), 97.33 $\pm$ 0.116 % (0.006 photons/pulse) and 87.27 $\pm$ 0.190 % (0.001 photons/pulse), from top to bottom. The stated uncertainties represent the 95% confidence intervals.