Short answer: The nitrogen nucleus inside every nitrogen-vacancy (NV) centre in diamond can act as a quantum memory for the centre's electron spin. A Nature Physics demonstration reported room-temperature storage with transfer in as little as 120 ns and total storage fidelity of 88 +/- 6%. Other work with carbon-13 nuclear spins reports much longer storage. All of it is laboratory research.

An intrinsic memory

A quantum memory pairs a long-lived qubit with each processing qubit, and the two must be linked by a fast, high-fidelity operation. The Nature Physics paper demonstrates a room-temperature memory based on the spin of the nitrogen nucleus that is part of each NV centre. The authors point out that this nucleus is deterministically present in every NV defect, which suits scaling. They store the electron spin state in the nuclear spin using Landau-Zener transitions across a hyperfine-mediated avoided level crossing. By working outside the asymptotic regime they reach coherent state transfer in as little as 120 ns with total storage fidelity of 88 +/- 6%.

Longer-lived registers

A 2019 Physical Review X paper realised a ten-qubit register made of an NV electron spin and nine nuclear spins. It reports entanglement between all 45 qubit pairs and genuine multipartite entangled states with up to seven qubits. Used as a memory, it protected an arbitrary single-qubit state for over 75 s, which the authors describe as the longest reported for a single solid-state qubit. The gates combine dynamical decoupling of the electron spin with selective, phase-controlled driving of nuclear spins.

Memory in a network setting

A 2022 npj Quantum Information paper addresses a network problem: dephasing during optical entanglement distribution limits scaling. The authors show a carbon-13 nuclear spin memory in isotopically engineered diamond that is robust to the optical link operation of an NV centre. They report memory lifetime improved by two orders of magnitude over the state of the art, surpassing reported times for entanglement distribution, and the nuclear-spin state can survive ionisation and recapture of the NV electron. Simulations in the paper suggest that combining this memory with demonstrated entanglement links and gates could enable network primitives such as deterministic non-local two-qubit gates.

Limits

These are results from individual experiments, in different setups, so the numbers above are not directly comparable. The simulated network capabilities are projections from the paper, not demonstrations.

Why a nuclear memory at all

The electron spin of an NV centre is the part that is read out optically and driven quickly, so it works as the processing qubit. Nuclear spins are the long-lived partner. The papers above address different parts of the same idea: how to get a state into a nuclear spin quickly, how many nuclear spins a register can hold while staying connected, and whether the stored state survives the optical operations that a network node needs. Each paper reports its own conditions, so a fidelity from one should not be read as a property of the others.

FAQ

What makes the nitrogen nuclear spin a convenient memory?

It is deterministically present in each NV centre.

How good was the storage?

Total storage fidelity of 88 +/- 6% with transfer in as little as 120 ns, at room temperature.

How long can nuclear spins hold a state?

The 2019 register paper reports over 75 s for an arbitrary single-qubit state.

Does the memory survive optical operations?

The 2022 paper reports that a carbon-13 memory in isotopically engineered diamond is robust to the NV optical link and to ionisation and recapture of the electron.

What is still simulated?

Network primitives such as non-local two-qubit gates are shown in simulation in that paper.

All field notes