Axions, if they exist, would interact with ordinary matter for microseconds at most. A dark matter signal passing through a detector would perturb nuclear spins for a duration far shorter than any practical measurement. The standard response is to build more sensitive detectors. If the signal is faint, amplify it. If it's brief, sample faster.
A collaboration across five quantum sensor nodes spanning 320 kilometers between Hefei and Hangzhou (Physical Review Letters, 2026) did something different. Instead of measuring the signal while it happens, they stored it. When an axion-like perturbation nudges nuclear spins, the nudge is captured in a long-lived nuclear-spin coherent state — a quantum memory that holds the perturbation stable long after the event has passed. The microsecond-scale signal becomes accessible as a minute-scale readout. The team didn't build a faster detector. They built a slower one.
The sensitivity gain is about four orders of magnitude beyond previous techniques, reaching spin rotation sensitivity of approximately one microradian. The constraints on axion-nucleon coupling across the mass range from 10 picoelectronvolts to 0.2 microelectronvolts surpass astrophysical limits — the first laboratory experiment to do so for axion topological-defect dark matter.
The 320-kilometer baseline serves a different purpose: it distinguishes real signals from local noise. A true dark matter signal would perturb all five nodes in a pattern consistent with a coherent wavefront. Local electromagnetic interference would not. The spatial distribution is not about coverage — it's about discrimination.
The general principle: when a signal is too brief to measure directly, the problem is framed as one of sensitivity — build a better detector. But sometimes the problem is temporal, not energetic. The signal carries enough information; it just doesn't persist long enough for the measurement apparatus to complete its work. Storing the signal in a quantum memory and reading it out slowly is a reframing that bypasses the sensitivity constraint entirely. Not every hard measurement needs a more powerful instrument. Some need a more patient one.