No new record, a better scoreboard
Week of November 2, 2026. Some weeks move a number; this one organized the numbers. The most useful publication of the week is not an experiment but a review from the group that holds most of the records, and it arrives with a taxonomy that makes every future claim easier to check. A modeling preprint for erbium rounds out the week. And because a headline about an “hour-long breakthrough” made the rounds, a hype check is in order.
The state of the art, from the group that set much of it (Journal)
Yu-Yu Sa, Pei-Xi Liu, Zong-Quan Zhou and Chuan-Feng Li of the University of Science and Technology of China have published *Long-lived quantum memories with rare-earth ions doped in crystals* in Quantum Frontiers (5, 14, 2026, open access). This is the group behind the europium records, so the review doubles as an authoritative account of where the field actually stands.
Its most useful contribution is a three-level taxonomy that any future announcement can be measured against. The **spin coherence time** is how long the crystal’s spins stay phase-coherent, and it caps everything else. The **light storage time** is how long an optical signal can be parked in the spins and retrieved coherently. The **quantum storage time** is how long a signal at the single-photon level survives with a signal-to-noise ratio high enough to beat any classical strategy. Each lifetime bounds the next, and confusing them is the most common source of inflated headlines.
Against that taxonomy, the records the review consolidates are: hour-scale spin coherence, with six hours in Eu:Y<sub>2</sub>SiO<sub>5</sub> (Zhong et al., 2015, still the longest known spin coherence time in any material); hour-scale light storage, with a 1/e lifetime of 52.9 minutes and coherence verified after one hour in Eu:Y<sub>2</sub>SiO<sub>5</sub> (Ma et al., 2021, a 75-fold improvement at the time); and minute-scale single-photon-level quantum storage, with a 1/e storage time of 27.6 seconds and qubit fidelity of 88.0% at 5.6 seconds against a classical limit of 82.1% (Lv et al., more than a 60-fold improvement of the quantum storage record). Europium-doped Y<sub>2</sub>SiO<sub>5</sub> remains the only material system that has demonstrated hour-scale coherent light storage.
The review is equally direct about what blocks the next step. There is a structural trade-off between lifetime and efficiency: long coherence wants low doping, strong absorption wants high doping. Reaching the hour scale takes thousands of dynamical-decoupling pulses, and small pulse errors and heating accumulate into noise that single-photon signals cannot tolerate. Integration into photonic chips and the engineering of transportable memories remain open problems.
Hype check: the “hour-long breakthrough” headline
Get Alfio Musumeci’s stories in your inbox
Join Medium for free to get updates from this writer.
At least one popular outlet ran a piece this week presenting “hour-long quantum memory” as a fresh breakthrough. Traced to its source, the article is a writeup of the USTC review above: the hour-scale results it describes are the 2015 spin-coherence and 2021 light-storage experiments, not a new experiment this week. The records are real and remain unbeaten; what happened this week is that they were summarized, not extended.
Erbium: computing where the quiet spots are (Preprint)
Erbium’s appeal is its telecom-wavelength optical transition; its weakness is magnetic noise from its own unquenched electronic spin. The standard mitigation is to work at ZEFOZ points, magnetic-field configurations where the transition’s first-order sensitivity to field fluctuations vanishes. A preprint posted November 5 by Shashank Srivastava maps the ZEFOZ landscape of site-2 <sup>167</sup>Er:Y<sub>2</sub>SiO<sub>5</sub> with a full-Hessian spin-Hamiltonian treatment, folding in the anisotropic magnetic noise of the crystal’s own <sup>89</sup>Y nuclei, and asks a practical question: which operating points best balance modeled coherence against microwave control authority?
The paper reports a Pareto-optimal frontier of four transition pairs, with one transition’s modeled yttrium-bath pure-dephasing time exceeding 2.2 x 10<sup>5</sup> seconds and another offering strong microwave addressability. It also shows that if the real Hamiltonian deviates from the assumed one, the shifted ZEFOZ points can be recovered by numerical recalibration, with field relocations of roughly 11 to 28 mT in the sampled cases. Two caveats are stated in the abstract itself and are worth keeping: all timescales are modeled and conditional on the adopted noise-correlation model, and the analysis assumes operation near 10 mK. This is groundwork for future experiments, not a measurement.
Follow-up: last week’s praseodymium result gets noticed
The ICFO praseodymium heralded-memory preprint covered here last week was picked up by popular science coverage on November 3. The coverage contains no new data; it is a sign that the result is circulating beyond the arXiv, which matters for how the field’s incremental work gets perceived.
What this means for quantum networks
The review states the requirement side of the equation cleanly: quantum repeaters need memory lifetimes from milliseconds to minutes depending on architecture and scale, while transportable quantum memories, physically shipped between nodes, need hours. Nothing published this week changes what is deployable today. What changed is that the field’s record-holding group published its own map of the gap between the two regimes, and the map is open access. For anyone evaluating quantum-memory claims, the three-lifetime taxonomy is the tool: ask which lifetime a headline refers to, and most of the confusion disappears.
*Sources are linked per paragraph. Journal labels mark peer-reviewed publications; Preprint labels mark arXiv postings that have not completed peer review.
