Short answer: A 2015 Nature paper reported a nuclear spin coherence time of 370 minutes, about six hours, in europium ions doped into an yttrium orthosilicate crystal at 2 kelvin. The transition can be addressed with light, which is why the authors consider it relevant to quantum memories for communication. The reported result is a coherence measurement on a spin ensemble, and no quantum repeater was built.
The problem the paper addresses
The Nature abstract explains that sending entanglement beyond a few hundred kilometres by light alone is prevented by losses from scattering, diffraction and absorption. A proposed fix is a quantum repeater protocol, which distributes entangled pairs of optical modes among many quantum memories along the channel. To be effective, the memories must hold quantum information for times long compared with the direct transmission time of the channel.
What was measured
The authors measured a decoherence rate of 8 times ten to the minus five per second over 100 milliseconds, which they note is the time required for light transmission on a global scale. The measurements used a ground-state hyperfine transition of europium-151 ion dopants in yttrium orthosilicate, with optically detected nuclear magnetic resonance. They say the observed decoherence rate is at least an order of magnitude lower than that of any other system suitable for an optical quantum memory.
With dynamic decoupling, they report a coherence time of 370 minutes with an uncertainty of 60 minutes, at 2 kelvin. The article title calls this a six-hour coherence time.
Context from the journal summary
A summary on the INSPIRE literature database, reproduced from the publisher, states that the previous record coherence time was three hours, in phosphorus donors in silicon-28, and that in the europium-doped crystal coherence had previously been limited to tens of milliseconds. It notes the crucial advantage that the transition is optically addressable, which it calls promising for long-lived quantum memory applications. A third listing on the RePEC site carries the article's bibliographic information and author list only.
A speculative remark
The abstract also notes that a coherence time this long means nuclear spins carried at 9 kilometres per hour in a crystal would have lower decoherence with distance than light in an optical fibre. The authors say this enables some early entanglement distribution approaches, in which spins are transported rather than light, to be revisited. This is an argument about physical limits and not a plan or an experiment.
Why long coherence helps
A memory that holds a state far longer than light takes to cross the globe gives a repeater design more slack, because the nodes need not be synchronized as tightly. The abstract frames the result in these terms, and it does not describe a network or a timeline.
Limits
The coherence time is a property of the spin transition under specific conditions, namely 2 kelvin and dynamic decoupling. The abstract does not report storing or retrieving a photon with six-hour retention. A quantum memory also needs efficient light storage and readout, which the abstract does not cover.
FAQ
What did the 2015 paper measure?
A nuclear spin coherence time of 370 minutes at 2 kelvin, using dynamic decoupling.
Which material was used?
Europium-151 ions in yttrium orthosilicate.
Why does it matter for communication?
Quantum repeaters need memories that outlast the transmission time, per the authors.
Does it work as a memory today?
The abstract reports coherence of the spin, not a full memory demonstration.
What was the earlier record?
Three hours, in phosphorus donors in silicon-28, per the journal summary.
All field notes