Neutrino Atlas
EN
  1. Learn
  2. Watching nuclear reactors with neutrinos

Chapter 5: Applications · Lesson 5.1

Watching nuclear reactors with neutrinos

Open

Every reactor emits a stream of antineutrinos that no wall can block. Counting them reveals how hard a reactor runs and how much plutonium its fuel contains, but distance and detector size limit what can be seen.

Inverse beta decay and the delayed coincidenceAn electron antineutrino converts a proton into a positron and a neutron. The positron gives a prompt flash, the captured neutron a delayed one.ν̄e from a reactorpe⁺Prompt: positron annihilates (2 × 511 keV)nDelayed: neutron captureν̄e + p → e⁺ + nflash 1flash 2Δt ≈ 30 µs (Gd) to 200 µs (H)time
Inverse beta decay, ν̄e + p → e⁺ + n, the reaction behind Cowan and Reines’ discovery and today’s reactor experiments (threshold 1.8 MeV). The positron annihilates at once (prompt signal); the neutron wanders and is captured microseconds later, emitting gamma rays (delayed signal) – about 8 MeV on gadolinium, 2.2 MeV on hydrogen. Demanding both signals in a short time window suppresses background by orders of magnitude. Schematic.

Beginner

Reactors glow in antineutrinos

A nuclear reactor produces heat by splitting uranium and plutonium nuclei. The fragments are radioactive and undergo beta decay, and on average each fission leads to about six electron antineutrinos. A large power reactor therefore emits several hundred billion billion antineutrinos every second. They pass through concrete, steel and rock as if these were not there. No wall can stop them and no shielding can hide them. This makes reactor antineutrinos an unusually honest witness of what is happening inside a reactor core.

An idea from 1978

In 1978 Lev Mikaelyan and Anatoly Borovoi of the Kurchatov Institute in Moscow proposed using this stream of particles to watch reactors from the outside1. In the 1980s their group placed a detector of about one cubic metre underneath a reactor at the Rovno power plant in Ukraine. It counted several hundred antineutrinos a day and measured the reactor's power to within about two percent2.

Counting and colour

A detector can learn two things. The number of antineutrinos tells how hard the reactor is working: double the power and the count doubles. The details reveal the fuel. Think of judging a campfire both by how many sparks fly and by their colour. When plutonium-239 splits, it gives fewer detectable antineutrinos, and less energetic ones, than uranium-235. As fresh uranium fuel burns, plutonium builds up in it, and at constant power the antineutrino count slowly drops by roughly ten percent over a fuel cycle. Because plutonium is the material of concern for nuclear weapons, this burn-up effect is what makes antineutrinos interesting for reactor monitoring.

From experiment to monitoring device

From 2003 a team led by Adam Bernstein ran a 640-kilogram detector called SONGS1 about 25 metres from the core of the San Onofre reactor in California. It worked almost unattended, noticed reactor shutdowns within a few hours and saw the burn-up effect3. Later detectors went further. NUCIFER in France worked close to a research reactor with very little shielding from cosmic rays4. DANSS, under a reactor in Russia, has tracked the reactor's power to about one percent per week for seven years6. PROSPECT in the United States showed that such detectors can work at ground level.

Who would use it?

The International Atomic Energy Agency (IAEA) checks that nuclear material in civilian reactors is not diverted to weapons. Antineutrino detectors are attractive because they do not touch the reactor and cannot be fooled by shielding. A study for the US government that interviewed safeguards experts found, however, that for ordinary reactors existing tools are considered adequate8. The most promising uses are new types of reactors that are hard to inspect and future international agreements.

The problem of distance

The signal weakens with the square of the distance. Moving a detector from 25 metres to 25 kilometres away means a million times fewer events. A British–American project called WATCHMAN planned a tank holding thousands of tonnes of water in a mine about 26 kilometres from the Hartlepool power station. It was cancelled in 2022. Finding secret reactors from far away would need even larger detectors and is not practical today.

Smaller detectors on the horizon

A different reaction, coherent elastic scattering, lets antineutrinos give a tiny push to a whole nucleus. In 2025 the CONUS+ experiment observed it at a Swiss power plant with a few kilograms of germanium7. Such detectors could one day be much smaller, but the pushes they must measure are minute.

Quiz

Test your knowledge

4 questions on this level · Print as worksheet

  1. 1On average, how many electron antineutrinos follow from each fission?
  2. 2Why does the antineutrino count drop over a fuel cycle at constant power?
  3. 3Who proposed in 1978 to watch reactors from outside with antineutrinos?
  4. 4How does the event count change if a detector moves from 25 metres to 25 kilometres?

Intermediate

The source

Each fission releases about 200 MeV, and the subsequent beta decays of the neutron-rich fragments emit on average about six electron antineutrinos with energies up to roughly 10 MeV. This fixes the emission rate per unit thermal power:

Ṅ_ν̄ ≈ 6 Pth / ⟨Ef⟩ ≈ 1.9 × 10²⁰ s⁻¹ × (Pth / 1 GW)Antineutrino emission rate of a reactor with thermal power Pth and mean energy release per fission ⟨Ef⟩ ≈ 200 MeV.

Monitoring detectors use inverse beta decay, ν̄e + p → n + e⁺, with a threshold of 1.8 MeV, so only roughly a quarter of the emitted antineutrinos can be seen. The positron gives a prompt flash and the neutron is captured some tens of microseconds later, on gadolinium or lithium-6 in modern designs. This delayed coincidence suppresses backgrounds by orders of magnitude.

Fuel composition from rate and spectrum

Four isotopes, ²³⁵U, ²³⁸U, ²³⁹Pu and ²⁴¹Pu, produce almost all fissions, and each has its own antineutrino spectrum. Daya Bay measured the inverse-beta-decay yield per fission as the plutonium fraction changed during fuel cycles5:

  • ²³⁵U: (6.17 ± 0.17) × 10⁻⁴³ cm² per fission
  • ²³⁹Pu: (4.27 ± 0.26) × 10⁻⁴³ cm² per fission, about 30 percent less, with a softer spectrum
  • Change with the ²³⁹Pu fission fraction F₂₃₉: dσf/dF₂₃₉ = (−1.86 ± 0.18) × 10⁻⁴³ cm² per fission

Since ²³⁹Pu also releases slightly more energy per fission, the count per unit of thermal power falls further. At constant power the rate of a typical pressurised water reactor drops by roughly 10 percent over a cycle. A rate alone cannot separate a change in power from a change in fuel, but the spectral shape can, provided enough events are collected.

Demonstrations

  • Rovno (Ukraine, 1980s): about 1,050 litres of gadolinium-loaded scintillator 18 m below a VVER-440 reactor; about 760 antineutrinos per day, thermal power to 2 percent2.
  • SONGS1 (California, 2003–2006): 640 kg, 24.5 m from a 3.4 GWth core in a tendon gallery; about 460 net events per day, shutdowns detected within 5 hours, weekly power to 3 percent3.
  • NUCIFER (Saclay, France): 0.85 m³ at 7.2 m from the 70 MW Osiris research reactor under only about 12 m water equivalent; 281 ± 7 (stat) ± 18 (syst) antineutrinos per day despite accidental backgrounds about 12 times the signal4.
  • PROSPECT (Oak Ridge, USA): about 4 t of lithium-6-doped segmented scintillator at 7–9 m from the 85 MW HFIR, at the surface.
  • DANSS (Kalinin, Russia): about 1 m³ of plastic scintillator strips on a movable platform 10.9–12.9 m below a 3.1 GWth reactor; about 5,000 events per day, power to 1.0 percent per week over seven years6.

These results drew the attention of the International Atomic Energy Agency, which discussed antineutrino detection at an expert meeting on new safeguards techniques in Vienna in 2003. The appeal is a measurement that is continuous, non-intrusive and independent of operator records. The drawback is that a detector of about a tonne has to sit within a few tens of metres of the core.

Compact detectors with coherent scattering

In CEvNS the antineutrino scatters off a whole nucleus. The cross-section grows with the square of the neutron number and has no reaction threshold, so a few kilograms of target can replace a tonne of scintillator. The cost is a tiny signal: a 3 MeV antineutrino gives a germanium nucleus at most about 0.27 keV, and only a fraction of that appears as ionisation. In 2025 CONUS+ observed reactor CEvNS 20.7 m from the 3.6 GWth Leibstadt reactor with 395 ± 106 events, a 3.7σ signal7. For monitoring, such detectors are still research instruments.

Quiz

Test your knowledge

4 questions on this level · Print as worksheet

  1. 1About how many antineutrinos per second does a reactor emit per gigawatt of thermal power?
  2. 2Why can inverse beta decay detect only about a quarter of reactor antineutrinos?
  3. 3How does the IBD yield per fission of ²³⁹Pu compare with ²³⁵U, according to Daya Bay?
  4. 4What can the spectral shape reveal that the rate alone cannot?

Expert

Observables and their degeneracy

For a detector with fixed geometry the inverse-beta-decay rate is proportional to the fission rate times the fuel-averaged yield. Daya Bay's linear model captures the burn-up dependence5:

R(t) ∝ (Pth / ⟨Ef⟩) · [σ̄f + (dσf/dF₂₃₉)(F₂₃₉(t) − F̄₂₃₉)]σ̄f = (5.90 ± 0.13) × 10⁻⁴³ cm² per fission at F̄₂₃₉ ≈ 0.30 and dσf/dF₂₃₉ = (−1.86 ± 0.18) × 10⁻⁴³ cm² per fission. ⟨Ef⟩ rises from about 202 MeV for ²³⁵U to about 211 MeV for ²³⁹Pu.

Rate alone constrains only the product of thermal power and a fuel-dependent factor; a few-percent drop can be either lower power or more plutonium. The spectral shape breaks the degeneracy because ²³⁹Pu antineutrinos are softer above 4 MeV. Statistically, a 1 percent rate measurement needs about 10⁴ background-subtracted events, which SONGS1 (about 460 net per day) collected in a few weeks3 and DANSS collects in days6. Absolute predictions remain limited by flux models: Daya Bay found a 7.8 percent deficit for ²³⁵U relative to the Huber–Mueller prediction, linked to the reactor anomaly and the 5 MeV bump. Monitoring therefore relies mostly on relative changes calibrated against declared operation.

Backgrounds near the surface

Reactor buildings offer little overburden. Cosmic-ray muons produce fast neutrons, whose proton recoil and capture mimic the inverse-beta-decay pair, and the β–n emitters ⁹Li and ⁸He. Gamma rays from the reactor and its surroundings raise the accidental rate; at NUCIFER accidentals were about 12 times and cosmogenic correlated events about 4 times the antineutrino rate4. Solutions include fine segmentation, pulse-shape discrimination (as in PROSPECT with ⁶Li capture), muon vetoes and reactor-off subtraction. DANSS exploits the overburden of the reactor itself and a movable platform for systematic checks.

CEvNS as a monitoring channel

With Tmax = 2Eν²/(M + 2Eν), reactor antineutrinos give germanium recoils below about 1 keV, and quenching leaves ionisation signals of order 100 eV or less. CONUS+ reached this regime with germanium point-contact detectors at Leibstadt7; the νGeN experiment at Kalinin has set limits with 1.41 kg of germanium (νGeN). Advantages for monitoring are kilogram-scale targets and sensitivity to antineutrinos below the 1.8 MeV IBD threshold. Disadvantages are the lack of a coincidence tag, a recoil spectrum that carries less fuel information, quenching uncertainties and sub-keV thresholds that are hard to maintain outside a laboratory.

Far field and the WATCHMAN story

Rates scale as 1/L². For an idealised IBD detector, a 10 MWth reactor 800 km away gives about five events per year in a megatonne of target. The Advanced Instrumentation Testbed, announced by the US NNSA with UK partners in 2018, planned WATCHMAN, a water Cherenkov detector of several kilotonnes with gadolinium loading, 1.1 km underground at Boulby, about 26 km from Hartlepool's two 1.5 GWth reactors (WATCHMAN/AIT). Its goals were to detect a reactor at that range, distinguish one core from two and confirm declared cycles. After the planned early closure of the Hartlepool reactors, the project at Boulby was cancelled in 2022. Work continues with BUTTON, a 30-tonne testbed for water-based liquid scintillator and fast photodetectors.

Where antineutrinos could help safeguards

The NuTools study for the NNSA interviewed 41 experts on end-user needs8. Its findings are sober:

  • Existing IAEA safeguards at conventional reactors: no specific role seen for neutrinos.
  • Advanced reactors (for example with online refuelling or liquid fuel) and future nuclear agreements, including verifying that a reactor is not operating: the most promising use cases.
  • Non-intrusive assay of spent fuel in dry casks: a plausible need, but technically very hard.
  • Non-cooperative discovery of reactors at long range: ruled out by detector size, dwell time, distance and backgrounds.
  • Reactor control: too slow compared with second-to-minute instrumentation timescales.

Adoption also requires full-scale demonstrations, acceptable cost for the IAEA and siting close enough for performance yet not so close as to be intrusive.

Quiz

Test your knowledge

4 questions on this level · Print as worksheet

  1. 1In Daya Bay's linear burn-up model, what is the slope dσf/dF₂₃₉?
  2. 2About how many background-subtracted events does a 1 percent rate measurement need?
  3. 3Which is an advantage of CEvNS as a monitoring channel?
  4. 4According to the NuTools study, where are antineutrino detectors most promising?

Open questions

  • Can a compact, robust antineutrino monitor reach the technology readiness and cost that the IAEA would require for routine use?
  • How well can rate and spectrum together determine the plutonium inventory of advanced reactors with online refuelling or liquid fuel?
  • Will CEvNS detectors with sub-keV thresholds become practical monitoring tools outside research settings?
  • Is there a realistic role for kilotonne detectors in verifying future agreements at mid-field distances?

Terms to explore

People

On the timeline

  1. 1953Project Poltergeist at the Hanford reactor
  2. 1956Cowan and Reines detect the neutrino
  3. 1978Neutrinos proposed for reactor monitoring
  4. 2003SONGS1 demonstrates practical reactor monitoring
  5. 2012Daya Bay discovers a non-zero θ₁₃
  6. 2025CONUS+ observes CEvNS at a nuclear reactor

References

  1. A. A. Borovoi, L. A. Mikaelyan, 'Possibilities of the practical use of neutrinos', Sov. At. Energy 44, 589 (1978), doi:10.1007/BF01117861
  2. Yu. V. Klimov, V. I. Kopeikin, L. A. Mikaelyan, K. V. Ozerov, V. V. Sinev, 'Neutrino method remote measurement of reactor power and power output', At. Energy 76, 123 (1994), doi:10.1007/BF02414355
  3. N. S. Bowden et al., 'Experimental results from an antineutrino detector for cooperative monitoring of nuclear reactors', Nucl. Instrum. Meth. A 572, 985 (2007), doi:10.1016/j.nima.2006.12.015
  4. Nucifer Collaboration, G. Boireau et al., 'Online monitoring of the Osiris reactor with the Nucifer neutrino detector', Phys. Rev. D 93, 112006 (2016), arXiv:1509.05610, doi:10.1103/PhysRevD.93.112006
  5. Daya Bay Collaboration, F. P. An et al., 'Evolution of the reactor antineutrino flux and spectrum at Daya Bay', Phys. Rev. Lett. 118, 251801 (2017), doi:10.1103/PhysRevLett.118.251801
  6. DANSS Collaboration, I. Alekseev et al., 'Long term remote reactor power and fuel composition monitoring using antineutrinos', Phys. Lett. B (2025) 139575, arXiv:2410.18914, doi:10.1016/j.physletb.2025.139575
  7. CONUS+ Collaboration, N. Ackermann et al., 'Direct observation of coherent elastic antineutrino–nucleus scattering', Nature 643, 1229 (2025), doi:10.1038/s41586-025-09322-2
  8. O. A. Akindele et al., 'Nu Tools: exploring practical roles for neutrinos in nuclear energy and security', PNNL-31870 (2021), arXiv:2112.12593, doi:10.2172/1826602

Last updated: · Editorial standard

Cite this page

Text

Neutrino Atlas (2026). “Watching nuclear reactors with neutrinos”. Retrieved 22 November 2026, from https://neutrino-atlas.com/learn/watching-reactors-with-neutrinos/

BibTeX

@misc{neutrinoatlas2026ingreactorswithneutrinos,
  title        = {Watching nuclear reactors with neutrinos},
  author       = {{Neutrino Atlas}},
  year         = {2026},
  url          = {https://neutrino-atlas.com/learn/watching-reactors-with-neutrinos/},
  note         = {Accessed 2026-11-22},
  language     = {english}
}