Where does a reactor’s antineutrino spectrum come from?

The fragments keep decaying after a fission

When a heavy nucleus splits, the fragments usually carry more neutrons than stable nuclei of similar mass. Many move toward stability by beta-minus decay: a neutron becomes a proton, and an electron and an electron antineutrino are emitted. Each daughter can decay again, so the source keeps going.

A single decay branch shares its energy among several products, so its antineutrino energy is spread out, not fixed. A reactor contains many branches with different endpoints, all overlapping. The smooth-looking spectrum is a sum over a complicated population of nuclei, not the signature of one universal decay.

Power gives the rate of energy release but does not list every fragment or branch. Two cores at similar thermal power can have different fission mixtures and different antineutrino spectra. Linking power to a prediction also needs the energy released per fission and the changing contribution of each isotope.

Sources: [1] · [2] · [3]

The source changes as the fuel changes

In common power-reactor calculations, most of the signal comes from fissions of uranium-235, uranium-238, plutonium-239 and plutonium-241. These names identify the nuclei that split; the antineutrinos come mostly from their radioactive products. Each fissioning isotope yields its own distribution of fragments.

As the fuel burns, the weights in that mixture shift. Plutonium is produced and consumed during operation, and the uranium contributions change too. A spectrum averaged over a fuel cycle therefore reflects a different mixture from one recorded early in the cycle, and any average must use the actual operating history.

Longer-lived fragments make the irradiation history matter too. Residual activity and spent fuel can still contribute after the reactor changes its operation. Treating the signal as an instantaneous thermometer of power would miss this memory. Detailed predictions track these effects, with their sizes and uncertainties, instead of assuming all decays stop at once.

Sources: [1] · [4]

Two ways to predict the spectrum from nuclear data

The summation method adds up the antineutrino spectra of individual decay branches, weighted by fragment yields and branching probabilities. It is physically explicit, but incomplete decay schemes and uncertain nuclear transitions can distort the total. A large database helps only as far as its contents describe the actual source.

The conversion method starts from measured total electron spectra for selected fissioning isotopes. It describes those spectra with a set of effective beta branches and converts the electron information into antineutrino information. That takes more than flipping the axis of a histogram: nuclear charge, branch endpoints and nuclear corrections all matter.

The 2011 calculations by Mueller and Huber are an important historical reference for these methods. Their predictions helped define the later rate discrepancy, but a prediction is not a direct measurement of what a reactor emits. Nuclear-structure calculations, such as those of Dwyer and Langford, offer another way to examine features in the spectrum.

Sources: [1] · [2] · [3] · [5]

The detector sees a different spectrum from the one the reactor emits

Scintillator experiments usually select inverse beta decay on free protons. This reaction has an energy threshold, and its probability rises across much of the reactor energy range. The sample therefore does not weight every emitted antineutrino equally, and many low-energy antineutrinos never enter this channel at all.

The prompt signal is mainly the positron’s kinetic energy plus its annihilation light. If small recoil effects are ignored and that light is fully contained, the incoming antineutrino energy is roughly the prompt energy plus the offset below. Real reconstruction also has to handle light yield, nonlinearity and energy resolution.

A fair comparison carries the emission prediction through distance, survival probability, interaction cross section and detector response. Comparing an unprocessed emitted spectrum directly with prompt counts mixes different quantities. How the detector defines its energy axis is part of the scientific argument.

Sources: [6] · [7] · [1]

Changing fuel gives the prediction another test

Daya Bay’s 2017 fuel-evolution analysis compared measured yields as the fission mixture changed. That adds a second question: beyond whether the overall rate agrees, does the rate change with the fuel the way the isotope-by-isotope prediction says? An error shared by all isotopes and an error in one isotope’s contribution need not look alike.

When testing oscillations, relative measurements at different distances cancel much of a shared source uncertainty. Absolute measurements and fuel comparisons instead test the source model itself. Neither replaces the other: an experiment can measure a mixing parameter convincingly and still find that its adopted source spectrum needs revision.

When reading a reactor graph, first identify the energy variable, the fuel mixture and the normalization. Then ask whether the line is an emitted prediction, a prediction folded through the detector response or a measured reference. These distinctions explain how a well-controlled reactor supports precision particle physics even though not every nuclear prediction is exact.

Sources: [4] · [8] · [7]

Try it in the Lab

Primary sources & revision

1. Mueller et al. · Improved Predictions of Reactor Antineutrino Spectra (2011)
2. Patrick Huber · On the determination of anti-neutrino spectra from nuclear reactors (2011)
3. Dwyer & Langford · Spectral Structure of Electron Antineutrinos from Nuclear Reactors (2014/2015)
4. Daya Bay Collaboration · Evolution of the Reactor Antineutrino Flux and Spectrum (2017)
5. Hayes et al. · Systematic Uncertainties in the Analysis of the Reactor Neutrino Anomaly (2013/2014)
6. Strumia & Vissani · Precise quasielastic neutrino/nucleon cross section (2003)
7. Daya Bay Collaboration · Measurement of the Reactor Antineutrino Flux and Spectrum (2015/2016)
8. Daya Bay Collaboration · Observation of electron-antineutrino disappearance (2012)

First published 2026-10-10; last revised 2026-10-10. Original explanatory text and diagrams by Sandbox Physics. Illustrations are schematic; no experimental event records are reproduced here. This is an independent educational publication, not an official Nobel or experiment collaboration publication.