All-photonic quantum repeaters

Send individual photons—the particles of light—through a long optical fibre, and many never reach the detector. Can stations along the route deliver quantum connections fast enough for useful network tasks despite this loss? They cannot simply copy an unknown quantum state as an ordinary signal booster copies information. All-photonic repeaters instead prepare groups of entangled photons, whose joint state links them, and measure selected photons to connect the remaining end points.[1] [3]

A photon sent through fibre can be lost before any detector registers it, and an unknown quantum state cannot be perfectly copied to replenish the signal. Can an all-photonic repeater chain overcome this loss while producing end-to-end entanglement at useful rates? Entanglement gives a shared state quantum correlations that cannot be explained solely by classical random choices of separate local states.

A repeater graph state represents photons as vertices and specified entangling operations as edges. The graph describes quantum correlations. Neighbouring resources send photons to joint measurement stations. Successful outcomes select branches to connect; other measurements remove unused branches. Redundant tree encodings can infer some required measurement results despite lost photons. Ordinary classical messages tell the end nodes whether the whole attempt succeeded and how to interpret the surviving state.

The 2019 12-photon interferometer demonstrated a simplified small parallel repeater without the full loss-tolerant encoding. Separately, the 2026 theoretical “emit-then-add” method joins a photon to a growing resource only after collection is heralded, so a failed collection need not erase the resource already built. It requires auxiliary spins and added gates; general stored graph states also require suitable nondestructive heralding and photon storage.[1] [2] [3]

A small laboratory demonstration used a 12-photon interferometer. A separate theoretical proposal makes large entangled groups less costly to generate when photon collection is poor. Neither establishes a practical chain of repeaters: sources, loss-tolerant measurements and success messages must work together at useful rates.[1] [2] [3]

State-generation scaling is not the rate of a working multi-hop service. Local spin coherence, gate fidelity, source speed, fibre loss and detector performance must meet the same protocol’s requirements. Classical outcome processing and end-node waiting remain necessary even when repeater stations avoid long waits in matter memories. The small experiment and theoretical construction do not certify this integrated performance.[1] [2] [3]

Shared entanglement is a resource for quantum communication and distributed quantum processing. A rare successful connection is useful evidence, but a service must supply new connections as they are consumed. Repeater performance therefore determines which tasks can actually run across separated sites.[1] [2]

Quantum network tasks consume entanglement rather than merely observing it once. Loss-tolerant redundancy can improve success probability while increasing the photons, gates and processing required for each attempt. Understanding that trade-off would show when the architecture supplies more usable pairs per unit time, connecting a source’s physical capabilities to a communication or computation workload.[1] [2]