Trapped-Ion Quantum Network Link

Two atoms in separate vacuum chambers, linked by nothing but a fiber and a coincidence of photons — and suddenly sharing the most intimate quantum connection possible.

Generates entangled Bell pairs between ion qubits residing in physically separate ion traps, potentially metres to kilometres apart. Each link attempt works by exciting an ion in each trap to emit a single photon whose polarisation or time-bin encoding is entangled with the ion's internal state. The photons travel through single-mode optical fibres to a central beam splitter; a coincident click on two detectors ('heralded entanglement') announces that the two ions have been projected into a Bell state, without revealing or disturbing the individual qubit values. The link is probabilistic per attempt, but the heralding signal makes it deterministic upon success — and the entangled state can then be stored in the ion for hundreds of milliseconds to seconds while classical communication confirms the outcome.

An ion (commonly ⁸⁸Sr⁺ at Oxford/Innsbruck, ¹³⁸Ba⁺ at Duke, or ¹⁷¹Yb⁺ at IonQ) is first optically pumped into a specific ground state, then excited by a short laser pulse. It decays back to two possible ground states via spontaneous emission, emitting a photon whose polarisation is correlated with the final ionic state: the ion-photon system is in an entangled Bell state. High-numerical-aperture optics (typically 0.6–0.8 NA) or an optical cavity collect the photon and couple it into a single-mode fibre. At a central 'heralding station', photons from two separate modules interfere on a 50:50 beam splitter (Hong-Ou-Mandel interference). Certain two-photon coincidence patterns are only consistent with the ions being in a Bell state, heralding successful remote entanglement. Because the photon collection efficiency per ion is typically 1–10%, most attempts fail and the cycle must be repeated, but the entanglement rate (Bell pairs per second) has improved from ~once per nine minutes in 2007 to ~250 per second by 2024.

In plain terms

Imagine two people in different rooms, each holding one end of a magical string. Neither person can see the string, but if one person tugs their end in a particular way, the other instantly feels a matching tug — even though no information passed through the wall. A quantum network link makes two atoms in different boxes share that kind of invisible connection, using particles of light as the go-between.

Also called: photonic ion-ion interconnect, remote ion entanglement link, quantum matter-link.

Why it matters

Photonic links are the only demonstrated method for entangling ions in separate vacuum systems over macroscopic distances, a prerequisite for modular quantum computers that scale beyond the size limit of a single trap.
Trapped ions already have the highest two-qubit gate fidelities of any platform (~99.9%); the network link extends that advantage across modules without the crosstalk and connectivity limits of a single large trap.
Remote entanglement with >10 s storage at Oxford (2025) means the link can serve as both a compute interconnect and a quantum memory within a network, opening the door to distributed quantum algorithms and quantum repeaters.
The heralded nature of the link means entanglement is certified before it is used — a property that is mathematically required for device-independent quantum cryptography.

Where it lives

Oxford University (Stephenson, Nadlinger et al., PRL 2020 / Main et al. 2025): Sr⁺/Ca⁺ mixed-species modules with 182 s⁻¹ entanglement rate, 94% Bell fidelity, remote entanglement storage >10 s
Duke Quantum Center (O'Reilly, Monroe et al., PRL 2024): Ba⁺ modules with 250 s⁻¹ continuous rate, >94% Bell fidelity, using Yb⁺ sympathetic cooling
IonQ (2024–2026): demonstrated remote ion-ion entanglement across separate commercial trap wells; first networked commercial quantum computers announced April 2026
University of Innsbruck (Lanyon group, 2025): 10-ion register interfaced with optical cavity, 92% average ion-photon Bell fidelity, scalable to larger registers

Key numbers

Entanglement rate (Oxford Sr⁺, 2020): 182 s⁻¹ (Stephenson et al., PRL 2020; 94% Bell fidelity)
Entanglement rate (Duke Ba⁺, 2024): 250 s⁻¹ (O'Reilly et al., PRL 2024; continuous rate with sympathetic cooling; >94% fidelity)
Remote entanglement fidelity (Oxford, 2025): 96.94% (Main et al., arXiv 2025; Sr⁺-Sr⁺ after full state tomography)
Entanglement storage (Oxford Ca⁺, 2025): >10 s (Memory qubit lifetime; >100× entanglement creation time)
First remote ion entanglement (Monroe group, 2007): ~1 per 9 min (~3×10⁻⁹ success probability per attempt; Moehring et al., Nature 2007)

Surprising facts

In 2007 the first remote ion-ion entanglement was heralded at a rate of about once every nine minutes — with a success probability of ~3 in a billion attempts. By 2024 the rate had reached 250 Bell pairs per second, a gain of more than seven orders of magnitude in just 17 years.
The photons used as the link 'messenger' are often at visible wavelengths (422 nm for Sr⁺, 493 nm for Ba⁺); telecom conversion to 1,550 nm is required for fibre transmission over more than a few kilometres.
Oxford's 2025 experiment stored remotely entangled ions in a Ca⁺ 'memory' qubit for more than 10 seconds — over 100 times longer than the entanglement creation time — which is sufficient in principle to run complex distributed quantum algorithms.
The heralding mechanism means the two ions 'know' they are entangled before any quantum gate is applied — something that is impossible classically, where you cannot certify a shared secret without sending it.

Open questions

Photon indistinguishability across separate traps: The Hong-Ou-Mandel interference that heralds remote entanglement requires photons from two independent ions — in separate vacuum chambers, driven by separate lasers — to be nearly perfectly indistinguishable in frequency, timing, and polarisation. Any residual distinguishability limits entanglement fidelity. How precisely two spatially separated, thermally fluctuating atoms can produce identical photons, and how well the fundamental quantum indistinguishability limit can be approached in practice, remains an active metrological challenge with no fully satisfying answer.

A short history

The idea of using photons as 'flying qubits' to link stationary matter qubits was formalised by Cirac, Zoller, Kimble and Mabuchi in 1997, and made practical for atomic ensembles by the Duan-Lukin-Cirac-Zoller (DLCZ) protocol of 2001. The crucial step — entangling single ions via photon interference — was first demonstrated by Monroe's group at the University of Michigan in 2007 (Moehring et al., Nature 449, 68), using Yb⁺ ions separated by about one metre. High-fidelity, high-rate operation came with Stephenson, Nadlinger and colleagues at Oxford (2020, PRL 124, 110501), achieving 182 Bell pairs per second at 94% fidelity with Sr⁺ ions. Continuous-rate operation at 250 s⁻¹ was reported by O'Reilly, Monroe et al. at Duke in 2024. Distributed quantum computations across two modules — including blind quantum computing — were demonstrated at Oxford in 2023–2025, and IonQ commercially demonstrated networked trapped-ion systems in 2026.

Sources

Moehring et al., Nature 449, 68 (2007) — first remote ion-ion entanglement (https://www.nature.com/articles/nature06118)
Stephenson, Nadlinger et al., Phys. Rev. Lett. 124, 110501 (2020) — high-rate, high-fidelity Sr⁺ link (https://link.aps.org/doi/10.1103/PhysRevLett.124.110501)
O'Reilly, Monroe et al., Phys. Rev. Lett. 133, 090802 (2024) — 250 s⁻¹ Ba⁺ entanglement with sympathetic cooling (https://link.aps.org/doi/10.1103/PhysRevLett.133.090802)
Main, Drmota, Nadlinger et al., arXiv:2506.14334 (2025) — multipartite mixed-species entanglement, 10 s storage (https://arxiv.org/abs/2506.14334)
NIST quantum networking with trapped ions in optical cavities — programme overview (https://www.nist.gov/programs-projects/quantum-networking-trapped-ions-optical-cavities)