Short answer: Erbium ions absorb light at telecom wavelengths, so erbium-doped materials are studied as quantum memories that can store photons suited to fibre networks. A 2014 Nature Photonics paper reports storing a 1,532 nm photon entangled with a 795 nm photon in an erbium-doped fibre, a 2018 arXiv paper reports storing heralded 1532 nm photons in an erbium-doped lithium niobate waveguide, and a December 2024 arXiv preprint reports storing telecom photonic qubits with a fidelity above 92%. All three are cryogenic laboratory experiments.
The 2014 fibre experiment
The authors say that, to date, neither an atomic quantum memory for non-classical states of light at a wavelength compatible with standard telecom fibre, nor a fibre-based quantum memory, had been reported. They report storage and faithful recall of the state of a 1,532 nm photon entangled with a 795 nm photon, in an ensemble of cryogenically cooled erbium ions doped into a 20-metre silica fibre, using a photon-echo quantum memory protocol. The abstract excerpt notes that the memory's performance was limited at the time, and the authors frame the result as relevant to a future quantum Internet.
The 2018 waveguide experiment
The arXiv authors report storing and retrieving heralded 1532 nm photons in a solid-state waveguide memory. The photons come from a photon-pair source based on parametric down-conversion. The memory is a 6 GHz-bandwidth atomic frequency comb prepared in the inhomogeneously broadened absorption line of a cryogenically cooled erbium-doped lithium niobate waveguide. Using persistent spectral hole burning under varying magnetic fields, they determine that the memory is enabled by population transfer into niobium and lithium nuclear spin levels. They note limited storage time and efficiency.
The 2024 preprint
The preprint says efficient storage of telecom-band quantum optical information is a crucial milestone for distributed quantum optical networks. It says erbium ions in crystalline hosts are a promising platform, but that practical applications have been hindered by demanding operational conditions such as ultra-high magnetic fields and ultra-low temperatures. The authors report storing telecom photonic qubits encoded in polarization, frequency and time-bin bases using the atomic frequency comb protocol in an Er3+-doped crystal. They report a memory initialization scheme that improves storage efficiency by over an order of magnitude under practical experimental conditions, and a fidelity exceeding 92% from quantum process tomography. It is a preprint, so it may change after peer review.
Limits
All three experiments need cryogenic cooling. Two of the three papers describe limited efficiency or storage time, and the third reports an efficiency improvement without giving a figure in the abstract. None describes a network deployment.
FAQ
Why erbium?
The papers use erbium because the 2014 work stores a 1,532 nm photon, in the telecom range, and the 2018 paper describes memories compatible with telecommunication-wavelength photons for quantum networks.
What is an atomic frequency comb?
A memory protocol that prepares a comb-like structure in an absorption line. The 2018 and 2024 papers use it. The 2014 fibre work used a photon-echo protocol.
What fidelity was reported?
The 2024 preprint reports a fidelity exceeding 92% from quantum process tomography.
Were the memories efficient?
The 2014 and 2018 papers report limited efficiency or storage time. The 2024 preprint reports an improvement of over an order of magnitude from its initialization scheme.
Is it deployed?
No. These are laboratory demonstrations.
All field notes