A team led by Thierry Lasserre and Anthony Onillon at the Max-Planck-Institut für Kernphysik (MPIK) in Heidelberg has measured electron antineutrinos from spent nuclear fuel for the first time, logging 106 ± 18 events in the 1–3 MeV window — a 5.9σ excess above background — during a 24.4-day simultaneous shutdown of the two Chooz B reactors in the French Ardennes.

What changed about the measurement window?

Independent safeguards agencies have used reactor antineutrinos since the 1970s to fingerprint fissile content and flag illicit plutonium production, but only while reactors operated. After shutdown, long-lived fission products in burnt assemblies keep decaying; the residual flux drops to roughly 1% of the running-reactor signal and sits in a low-energy band where natural backgrounds dominate. The 2017 outage at Chooz — both 4.25 GWth pressurized-water cores down at once for refuelling and maintenance — gave the Double Chooz collaboration the rare, clean window it needed. After muon-veto dead time, the team accumulated 17.2 days of near-detector live time and 22.2 days at the far site.

Why is the near detector doing the heavy lifting?

The experiment runs a "near" detector at about 400 m from the cores and a "far" detector at 1.05 km. The near unit sees a much larger solid-angle fraction of the spent-fuel pools' emissions, so the team built its spectrum there. Both detectors hold over 30 m³ of liquid scintillator — a medium that emits a characteristic prompt-plus-delayed light doublet when an antineutrino inverse-beta-decays on a proton. Outer shielding uses 15 cm of demagnetized steel (far) or 1 m of water (near) to suppress rock gamma backgrounds, mineral oil inside the target volume, and a segmented muon veto above the structure to cut cosmogenic false positives.

How well does the spectrum agree with fuel inventories?

The observed 106 events sit 1.0σ above the 88 ± 7 events the team predicted from a detailed simulation of the irradiation and cooling history of the remaining fuel assemblies. Each Chooz core carries 205 assemblies of roughly 600 kg of enriched uranium dioxide, predominantly ²³⁸U with a few percent ²³⁵U; neutron capture on ²³⁸U also breeds ²³⁹Pu and ²⁴¹Pu during a typical >1-year operating cycle. That the measured residual flux matches the inventory model — rather than diverging — is the result's main operational value: it validates a non-intrusive way to track what is in the cooling pool without opening it.

What does this mean for safeguards R&D budgets?

The work traces directly to a 2003 question raised at an IAEA safeguards meeting with the neutrino community: can antineutrinos extend monitoring past reactor shutdown? "The main difficulty has been detecting the faint residual antineutrino signal," the authors told Physics World, noting that success required a simultaneous dual-reactor outage, low and well-controlled backgrounds, and a fuel-assembly-by-assembly decay simulation. They expect the dataset to become a benchmark for agencies and detector designers building purpose-built spent-fuel monitors, framing the capability as an independent complement to existing inventory tools while cautioning that field deployment will face cost and footprint constraints.

The MPIK team is blunt about the engineering trade-off: "Such monitoring is feasible and could provide an independent, non-intrusive complement to reactor-status and spent fuel inventory," but it "will not necessarily be easy, compact or inexpensive to implement." For R&D managers weighing detector architectures, the Chooz dataset supplies the first quantitative residual spectrum against which competing designs — gadolinium- or lithium-loaded scintillators, segmented liquidO targets, or scaled-down near-only stations — can now be benchmarked. The full results appear in Physical Review Letters.