Geoneutrinos: looking inside the Earth
Radioactive elements deep inside the Earth emit antineutrinos that cross the planet unhindered. Catching a few of them each year tells geologists how much of the Earth's internal heat comes from radioactivity.
- 1Boulby Underground LaboratoryUnderground laboratory
- 2Laboratorio Subterráneo de Canfranc (LSC)Underground laboratory
- 3Laboratoire Souterrain de Modane (LSM)Underground laboratory
- 4Laboratori Nazionali del Gran Sasso (LNGS)Underground laboratory
- 5KM3NeTNeutrino telescope
- 6Neutrino Energy Group, BerlinNeutrinovoltaic research
- 7Baksan Neutrino ObservatoryUnderground laboratory
- 8Baikal-GVDNeutrino telescope
- 9China Jinping Underground Laboratory (CJPL)Underground laboratory
- 10JUNOReactor experiment
- 11Kamioka ObservatoryUnderground laboratory
- 12SNOLABUnderground laboratory
- 13Sanford Underground Research Facility (SURF)Underground laboratory
- 14IceCube Neutrino ObservatoryNeutrino telescope
Beginner
A planet that is still warm
The Earth is slowly losing heat. Through its surface, from volcanoes, hot springs and, above all, the steady warmth that seeps out of rocks and the ocean floor, it releases about 47 terawatts6. That is more than twice the total energy use of humanity. The heat comes from two sources: warmth left over from the planet's violent formation 4.5 billion years ago, and heat produced today by the decay of radioactive elements, mainly uranium, thorium and potassium. This second part is called radiogenic heat. How the total splits between the two is one of the big open questions of geology, because it controls how fast the mantle churns, how plates move and how long the Earth will stay geologically active.
Radioactive rocks glow in antineutrinos
Uranium-238 and thorium-232 do not decay in one step. They pass through long chains of short-lived daughters before ending as stable lead. Several of these steps are beta decays, and each beta decay sends out an antineutrino. Potassium-40 does the same. These geoneutrinos fly straight out of the Earth at almost the speed of light. In a sense the planet glows faintly in antineutrinos, and the brightness of that glow is tied directly to how much uranium and thorium it contains. The idea of using them was proposed in the 1960s by George Marx and Gernot Eder.
Catching them
Geoneutrinos are measured in large tanks of liquid scintillator, an oily liquid that gives off light when charged particles pass through it, placed deep underground to block cosmic rays. KamLAND in Japan holds about 1,000 tonnes, Borexino in Italy about 280 tonnes. Occasionally an antineutrino hits a proton and turns it into a neutron, releasing a positron. The positron makes a flash at once, and the neutron makes a second flash a fraction of a millisecond later. This double signal, called inverse beta decay, is very hard to fake. It works only for antineutrinos above 1.8 MeV, so part of the uranium and thorium signal can be seen, but none from potassium.
KamLAND reported the first evidence in 2005: about 25 events in two years1. Borexino confirmed the signal in 20104. Today the two experiments together have recorded a few hundred geoneutrinos.
What they tell us
The results show that radioactivity provides a large share of the Earth's heat, but not all of it. KamLAND's latest analysis points to roughly 10 to 15 terawatts from uranium and thorium3; Borexino's points somewhat higher5. Both have large uncertainties, but together they rule out the extremes: the Earth is neither almost free of radioactive heat nor heated by radioactivity alone. The remainder is primordial heat slowly leaking out.
Why location matters
Most geoneutrinos reaching a detector come from the crust within a few hundred kilometres, because the continental crust is rich in uranium and thorium and lies close by. Geologists know the crust fairly well from rock samples, so they can subtract its share. What remains comes from the mantle, the hot rock layer below. Borexino showed that this mantle signal is not zero5. Detectors on thin oceanic crust would see the mantle more clearly, while detectors under thick mountains see more crust.
What comes next
The giant JUNO detector in China, which began taking data in 2025, should record about 400 geoneutrinos per year8. SNO+ in Canada is making the first measurements in the western hemisphere, and a planned detector at Jinping in China would sit beneath very thick crust far from nuclear reactors7. Comparing sites on different kinds of crust is the way to map the radioactive elements inside the Earth.
Quiz
Test your knowledge
4 questions on this level · Print as worksheet
Intermediate
Decay chains and their antineutrinos
The three main heat-producing elements release both energy and antineutrinos in fixed proportions. Summed over the full chains:
Because the ratio of heat to antineutrinos is fixed by nuclear physics, counting antineutrinos measures heat production directly. Only a few steps produce antineutrinos above the inverse beta decay threshold of 1.806 MeV: ²¹⁴Bi (endpoint 3.27 MeV) and ²³⁴Paᵐ (2.27 MeV) in the uranium chain, ²¹²Bi (2.25 MeV) and ²²⁸Ac (2.08 MeV) in the thorium chain. Thorium therefore contributes only below about 2.25 MeV, and the shape of the measured spectrum separates the two elements. Potassium-40, whose antineutrinos stop at 1.31 MeV, stays invisible.
Most Earth models assume that thorium and uranium occur in the mass ratio Th/U ≈ 3.9 found in primitive meteorites (chondrites). With this ratio fixed, a detector needs to measure only one number, the total rate. Leaving the ratio free and fitting the spectrum is a direct test of that assumption, but it requires many more events, because the thorium component is small and sits at the lowest energies, where reactor and other backgrounds are largest.
The heat budget
Compilations of borehole and ocean-floor measurements give a surface heat flow of 47 ± 2 TW6; other analyses quote values between about 44 and 47 TW. Models of the bulk silicate Earth, the crust and mantle together, predict very different radiogenic powers depending on how they derive the abundances: models based on enstatite chondrites give around 10–15 TW, geochemical models based on carbonaceous chondrites and mantle rocks about 20 TW, and models built to make mantle convection work smoothly 30 TW or more. Potassium contributes about 4 TW in most of these models2. The continental crust alone produces roughly 7–8 TW, so the mantle's radiogenic share is the most uncertain piece.
Counting in TNU
Predicted signals range from about 10 TNU on oceanic crust to over 40 TNU on thick continental crust. A kilotonne detector thus sees only a few tens of geoneutrinos per year, and must identify them against reactor antineutrinos, which have the same signature and overlap in energy between 1.8 and 3.3 MeV. The delayed coincidence of positron and neutron capture removes almost all other backgrounds.
Results so far
- KamLAND 2005: first evidence, about 25 events in 749 days, and an upper limit of 60 TW on uranium and thorium heat1.
- Borexino 2010: observation at 4.2σ, with a much smaller reactor background at Gran Sasso4.
- KamLAND 2011: uranium and thorium together contribute 20.0 (+8.8/−8.6) TW, about half the total heat flow2.
- Borexino 2020: 52.6 (+9.4/−8.6) events in 3,263 days, a signal of 47.0 (+8.4/−7.7) TNU; zero mantle signal excluded at 99% confidence5.
- KamLAND 2022: 5,227 days of data; 174 (+31/−29) events, a uranium plus thorium flux of (3.4 ± 0.5)×10⁶ cm⁻² s⁻¹ at the surface and the absence of geoneutrinos excluded at 8.5σ3.
KamLAND benefited from the long shutdown of Japanese reactors after 2011, which lowered its main background. In 2025 SNO+ presented a preliminary geoneutrino rate from the Canadian Shield, still with large uncertainties.
Quiz
Test your knowledge
4 questions on this level · Print as worksheet
Expert
From abundances to signal
The expected spectrum at a detector at position R is a volume integral over the Earth of the local abundances, weighted by the inverse-square distance and the oscillation probability:
Since source distances greatly exceed the oscillation length of about 100 km at a few MeV, the survival probability averages to ⟨Pee⟩ ≈ 1 − ½ sin²2θ₁₂ cos⁴θ₁₃ − ½ sin²2θ₁₃ ≈ 0.55; matter effects change it by less than 1%. Oscillations therefore scale the flux without distorting the spectrum. With ρ taken from seismology (PREM), the unknowns are the abundance distributions aU(r) and aTh(r).
Crust, mantle and the local problem
At continental sites, 70–85% of the signal comes from the lithosphere, and roughly half from crust within about 500 km. Separating the mantle therefore requires a local geological model with uncertainties well below the experimental ones. Borexino subtracted a lithospheric prediction and found a mantle signal of 21.2 (+9.6/−9.0) TNU, corresponding to 24.6 (+11.1/−10.4) TW of mantle radiogenic power; combined with the crust and with potassium, this implies a total radiogenic heat of 38.2 (+13.6/−12.7) TW, in 2.4σ tension with models that have the lowest abundances of heat-producing elements5. KamLAND, whose data are more precise but whose local crust (Japan arc) is more complex, prefers lower values: 15.4 (+8.3/−7.9) TW for U + Th with the Th/U ratio free, 10.6 (+5.2/−4.2) TW with Th/U fixed at the chondritic 3.9, and exclusion of a fully radiogenic Earth at 5.2σ under the assumption of a homogeneous mantle3. The two results lean in different directions but are compatible within their uncertainties, and both disfavour the highest-power geodynamical models.
Spectroscopy and exotic sources
KamLAND's spectral fit with Th/U free gives 117 (+41/−39) uranium and 58 (+25/−24) thorium events, consistent with the chondritic ratio; the no-uranium hypothesis is disfavoured at 3.3σ3. Geoneutrino data also constrain a hypothetical natural fission reactor at the Earth's centre, the 'georeactor': its power is below 1.26 TW (KamLAND, 90% CL) and 2.4 TW (Borexino, 95% CL)35.
Directionality
Isotropic counting cannot tell whether antineutrinos come from nearby crust or the deep mantle. In inverse beta decay the neutron retains a memory of the antineutrino direction and is displaced forward by a few centimetres before capture, but in kilotonne scintillators this is washed out by vertex resolution and neutron diffusion. Proposed remedies include capture agents with short diffusion length (⁶Li, ¹⁰B), fine-grained or segmented detectors, and elastic scattering on electrons in detectors able to separate Cherenkov from scintillation light, which would also give access to ⁴⁰K geoneutrinos below the IBD threshold. A directional measurement could test lateral mantle structure, such as the two large low-shear-velocity provinces beneath Africa and the Pacific.
The next generation
- JUNO, 20 kt near Jiangmen: about 400 geoneutrino events per year and a predicted 39.7 (+6.5/−5.2) TNU; with Th/U fixed, the signal should reach a precision of about 8% after five years, limited by the large reactor flux and by the local crust, for which a dedicated model (JULOC) was built8.
- SNO+ at SNOLAB: on the old Canadian Shield, with about 25 events per year expected and a known local geology.
- Jinping at CJPL: beneath the thickened crust at the edge of the Tibetan plateau, about 1,400 km from the nearest reactor; a predicted 45.9 ± 6.4 TNU, 86% from the lithosphere, and about 100 events per year for a multi-kilotonne detector7.
- Ocean-bottom detectors: proposed for sites far from continents, where the mantle would provide most of the signal.
A global fit of several sites with different crust-to-mantle ratios, rather than any single detector, is expected to determine the mantle's radiogenic power and to test whether its heat-producing elements are uniformly distributed or concentrated in deep reservoirs.
Quiz
Test your knowledge
4 questions on this level · Print as worksheet
Open questions
- How much of the Earth's roughly 47 TW heat flow is radiogenic, and which class of Earth composition models is correct?
- Is the mantle's uranium and thorium uniformly mixed, or concentrated in deep reservoirs such as the large low-shear-velocity provinces?
- Can potassium-40 geoneutrinos be detected, and is there potassium in the Earth's core?
- Can a practical directional antineutrino detector separate crust and mantle signals?
Terms to explore
- Geoneutrinos
- Radiogenic heat
- Terrestrial neutrino unit (TNU)
- Inverse beta decay
- Delayed coincidence
- Liquid scintillator detector
- Reactor antineutrinos
- Beta decay
- KamLAND
- Borexino
- JUNO
- SNO+
- Jinping Neutrino Experiment (JNE)
- Earth density profile (PREM)
- Neutrino tomography
People
- George Marx (Marx György)
- Gernot Eder
- Kunio Inoue
- Atsuto Suzuki
- Gianpaolo Bellini
- Mark C. Chen
- Wang Yifang
- Giorgio Gratta
On the timeline
- 1966Antineutrinos from Earth's interior proposed
- 2002KamLAND sees reactor antineutrinos disappear
- 2005KamLAND detects geoneutrinos
- 2010Borexino confirms geoneutrinos
- 2025JUNO begins data taking
References
- KamLAND Collaboration, T. Araki et al., 'Experimental investigation of geologically produced antineutrinos with KamLAND', Nature 436, 499 (2005), doi:10.1038/nature03980
- KamLAND Collaboration, A. Gando et al., 'Partial radiogenic heat model for Earth revealed by geoneutrino measurements', Nature Geosci. 4, 647 (2011), doi:10.1038/ngeo1205
- KamLAND Collaboration, S. Abe et al., 'Abundances of uranium and thorium elements in Earth estimated by geoneutrino spectroscopy', Geophys. Res. Lett. 49, e2022GL099566 (2022), arXiv:2205.14934, doi:10.1029/2022GL099566
- Borexino Collaboration, G. Bellini et al., 'Observation of geo-neutrinos', Phys. Lett. B 687, 299 (2010), doi:10.1016/j.physletb.2010.03.051
- Borexino Collaboration, M. Agostini et al., 'Comprehensive geoneutrino analysis with Borexino', Phys. Rev. D 101, 012009 (2020), arXiv:1909.02257, doi:10.1103/PhysRevD.101.012009
- J. H. Davies, D. R. Davies, 'Earth's surface heat flux', Solid Earth 1, 5 (2010), doi:10.5194/se-1-5-2010
- O. Šrámek, B. Roskovec, S. A. Wipperfurth, Y. Xi, W. F. McDonough, 'Revealing the Earth's mantle from the tallest mountains using the Jinping Neutrino Experiment', Sci. Rep. 6, 33034 (2016), doi:10.1038/srep33034
- JUNO Collaboration, A. Abusleme et al., 'JUNO physics and detector', Prog. Part. Nucl. Phys. 123, 103927 (2022), arXiv:2104.02565, doi:10.1016/j.ppnp.2021.103927
Last updated: · Editorial standard
Cite this page
Text
Neutrino Atlas (2026). “Geoneutrinos: looking inside the Earth”. Retrieved 22 November 2026, from https://neutrino-atlas.com/learn/geoneutrinos/BibTeX
@misc{neutrinoatlas2026learngeoneutrinos,
title = {Geoneutrinos: looking inside the Earth},
author = {{Neutrino Atlas}},
year = {2026},
url = {https://neutrino-atlas.com/learn/geoneutrinos/},
note = {Accessed 2026-11-22},
language = {english}
}