This Week in Rare-Earth Quantum Memories: Erbium Reaches a Millisecond at Telecom

Second entry in a weekly series tracking what rare-earth ions have actually demonstrated as quantum memories and network interfaces. Every claim names its source, and each item is labeled Journal (peer reviewed or accepted) or Preprint (not yet peer reviewed). Nothing here is a product claim, and where the researchers themselves urge caution, that caution is included.

Erbium: on-demand millisecond storage at telecom wavelength (Journal, accepted)

The strongest result of the week comes from Daniel Oblak's group at the University of Calgary. Physical Review Letters lists as accepted (30 September 2026) "On-demand millisecond storage of spectrotemporal modes at telecom wavelength" by Anuj Sethia, Nasser Gohari Kamel and Daniel Oblak. The team stores weak coherent pulses in an erbium-doped yttrium orthosilicate crystal (Er:YSiO) and recalls them on demand for up to 1 millisecond, which the authors describe as an order of magnitude beyond previously reported erbium-based demonstrations. Source: journals.aps.org/prl/accepted/10.1103/6rns-8wj3.

The reported numbers give a honest picture of where the field stands. Memory efficiency is 10.36% at a storage time of 300 microseconds, with a signal-to-noise ratio of 10.9, limited mainly by spontaneous emission noise. The team stores two spectral modes as a frequency-bin encoding and retrieves the state with an overlap of 81.8%, and demonstrates multimode capacity by simultaneously storing 20 temporal and 3 spectral modes with on-demand and selective recall. Source: same PRL accepted abstract as above. A public preprint has existed since December 2025 (arXiv:2512.17181); what is new this week is peer-review acceptance.

Why this could matter: the memory operates at telecom wavelength, so it is in principle compatible with existing fiber infrastructure, and on-demand recall plus multimode capacity are two properties quantum repeater architectures require. The caveats are just as important: the stored pulses are weak coherent states, not single photons, and a 10% efficiency is far from anything deployable. The authors frame it as "a promising step toward a scalable quantum repeater architecture", which is the right scale of claim. Source: PRL abstract cited above.

Erbium again, in another crystal: a spectral imprint that persists for over a week (Journal, 2025, highlighted this week)

On 26 October 2026 the CEA highlighted a result its Quantronique group published with several collaborators in Nature Communications: "Week-long-lifetime microwave spectral holes in an erbium-doped scheelite crystal at millikelvin temperature" (Wang, Lin, Le Dantec, Rancic, Goldner, Bertaina, Chaneliere, Liu, Esteve, Vion, Flurin, Bertet). The paper itself appeared on 10 October 2025 (Nat Commun 16, 9032 (2025)), so this is a year-old journal result that a major lab chose to explain to the public this week; the date distinction matters and is kept deliberately. Sources: CEA news, 26 October 2026 and the Nature Communications paper.

What the team showed: in a calcium tungstate crystal (CaWO4) containing roughly 3 erbium ions per billion atoms, cooled to 10 millikelvin and driven through a superconducting resonator, a microwave-written modification of the erbium spectrum (a spectral hole, plus a periodic train of holes producing "accumulated echoes") remains measurable after six days, and the decay analysis shows a slow component with a time constant of about one month. The majority of erbium ions drift in resonance frequency by less than 10 kHz over a month. The proposed explanation is a "frozen core" effect: some tungsten nuclear spins near the erbium ions interact so weakly with the rest of the crystal that the local magnetic environment barely evolves. Sources: CEA news and Nat Commun paper cited above.

The caution comes from the researchers themselves, and it is worth quoting in substance: this is the microwave domain, not the optical domain, and according to the CEA text, progress is still needed, in particular to strengthen the spectral modification and the coupling between the erbium ions and the resonator, before anything like a real quantum memory can be considered. Read it as a demonstration of extreme spin-environment stability at millikelvin temperatures, not as a storage device. Source: CEA news cited above.

Europium: driving the crystal's own spins to extend coherence (Journal, accepted)

Physical Review A lists as accepted (27 October 2026) "Enhanced coherence of rare-earth nuclear spins in a crystal" by Bassam Nima, Yuiki Takahashi and Amar Vutha (University of Toronto). The abstract reports that the nuclear spin coherence time of 153Eu3+ ions doped into yttrium orthosilicate (the same host family used in several leading memory experiments) can be significantly enhanced by driving the environmental nuclear spins of the host crystal. The authors point to improved precision measurements using rare-earth-doped crystals as the direct application. Source: journals.aps.org/pra/accepted/10.1103/s3dn-1wf7. A preprint is available as arXiv:2608.03939.

Why it is on this list: europium-doped Y2SiO5 holds some of the longest optical and spin coherence records in the field, and coherence time is the raw resource that memory protocols spend. A technique that quiets the host's own nuclear spins attacks one of the known decoherence channels. How large the enhancement is, and whether it translates into longer memory storage, is not stated in the accepted-paper abstract, so no stronger claim is made here. Source: PRA accepted abstract cited above.

Praseodymium: telecom-heralded single-photon storage with spin rephasing (Preprint)

A preprint from Hugues de Riedmatten's group (ICFO and collaborators), posted 2 September 2026, reports "Long-lived telecom-heralded single-photon storage in an absorptive spin-rephased quantum memory" (Rodriguez-Moldes, Appas, Hanni, Rakonjac, Grandi, de Riedmatten; arXiv:2609.02579). It is included this week because it completes the verification of an item flagged in the previous review. According to the abstract, this is the first demonstration of non-classical storage of telecom-heralded single photons in an absorptive quantum memory that includes a spin rephasing mechanism: an atomic frequency comb spin-wave protocol in Pr3+:Y2SiO5 combined with an XY4 microwave rephasing sequence. Source: the arXiv abstract.

The reported numbers: long-lived AFC photon echoes in the classical regime up to about 3 milliseconds, and non-classical correlations between heralding photons and stored signal photons (from a cavity-enhanced parametric pair source) for storage times up to 180 microseconds, with cross-correlation values as high as 4.6(4). Two honest caveats: this is a preprint, not yet peer reviewed; and the non-classical regime reaches 180 microseconds, not the 3 milliseconds of the classical echoes. The authors present it as a significant step toward scalable long-distance quantum repeater links, which is their framing, quoted here as such. Source: arXiv:2609.02579 abstract.

Erbium, thulium and europium in molecules: one ligand, three different behaviors (Preprint)

A second preprint worth noting compares Er3+, Tm3+ and Eu3+ in the same molecular ligand environment at cryogenic temperature: "Comparative Optical Spectroscopy of Er3+, Tm3+, and Eu3+ Molecular Complexes at Cryogenic Temperature" (Rasekh, Sahil, Mansourzadeh, Galico, Hunger, Murugesu, Oblak; arXiv:2610.27223, posted in October 2026). Spectral hole burning, the basic mechanism behind several rare-earth memory protocols, is observed in the Eu3+ complex but not measurably in the Er3+ and Tm3+ complexes. The authors also document strong laser-induced local heating in the Er and Tm samples, with recovery over seconds and in some erbium conditions tens of seconds, and they argue that heating alone is unlikely to explain the full contrast between the three ions; the differences point to ion-dependent relaxation and decoherence tied to electronic level structure and coupling to the molecular environment. Source: arXiv:2610.27223 abstract.

Why it matters, with conditionals: rare-earth molecular complexes attract interest because chemistry can tune them, unlike doped crystals. This comparison suggests the ions do not transfer their crystal behavior into molecules automatically, and it doubles as a methodological warning about heating artifacts in cryogenic spectroscopy of powders. It is a preprint and a negative result for two of the three ions, which is exactly the kind of evidence that keeps this series honest. Source: arXiv:2610.27223 abstract.

Neodymium

Nothing new on neodymium-based quantum memories surfaced this week. As in the first review, it stays on the watch list rather than being forced into the story.

What the week adds up to

Read together, and only with the conditionals each source carries, the week's items share one theme: the fight is no longer only about optical coherence, it is about controlling the spin environment around the ion. Calgary's erbium result pushes on-demand telecom storage into the millisecond range with real multiplexing. The CEA-highlighted CaWO4 work and Toronto's europium result both attack decoherence by quieting nearby nuclear spins, one by cooling into a frozen-core regime, the other by actively driving the host spins. The ICFO praseodymium preprint tries to bring heralding and rephasing together in one memory. And the molecular comparison warns that a new platform has to re-earn each ion's reputation. None of this is a working repeater, and no source cited here claims one. It is incremental, checkable physics, which is what this series exists to track.

Sources

  • Sethia, Gohari Kamel, Oblak, "On-demand millisecond storage of spectrotemporal modes at telecom wavelength", Phys. Rev. Lett., accepted 30 September 2026. link; preprint arXiv:2512.17181.
  • Wang et al., "Week-long-lifetime microwave spectral holes in an erbium-doped scheelite crystal at millikelvin temperature", Nat Commun 16, 9032 (2025). link; CEA highlight, 26 October 2026: link.
  • Nima, Takahashi, Vutha, "Enhanced coherence of rare-earth nuclear spins in a crystal", Phys. Rev. A, accepted 27 October 2026. link; preprint arXiv:2608.03939.
  • Rodriguez-Moldes et al., "Long-lived telecom-heralded single-photon storage in an absorptive spin-rephased quantum memory", arXiv:2609.02579 (preprint, 2 September 2026).
  • Rasekh et al., "Comparative Optical Spectroscopy of Er3+, Tm3+, and Eu3+ Molecular Complexes at Cryogenic Temperature", arXiv:2610.27223 (preprint, October 2026).

Comments