Earth’s internal heat budget
How much of the heat flowing out of Earth’s interior comes from radioactive decay, and how much is left over from its formation?
Summary
Why it is open
By the main estimates, 44 to 47 terawatts (TW, 10¹² watts) of heat leave Earth's surface [1, 5]. Part of it is made now, by the decay of uranium, thorium and potassium. The rest is primordial heat, from the collisions of the material that built Earth and, to a lesser extent, the sinking of metal to form the core [10], lost as the interior cools.
The surface heat flow does not show the split. In 2010 J. H. Davies and D. R. Davies set out the budget behind their 47 TW estimate: 6–8 TW is generated in the crust and 5–13 TW comes from the core, a range they call highly uncertain, leaving 24–38 TW to radioactivity in the mantle and the cooling of the planet [5]. Models of the bulk silicate Earth (crust and mantle together) give about 20 TW of radiogenic heat; after the crust's share, that leaves 7–29 TW to be supplied by cooling, a rate that recent studies had judged highly unlikely, and Davies and Davies wrote that their higher total makes this "energy paradox" more difficult to understand [5].
The composition is itself disputed. Published models of the bulk silicate Earth fall into three groups that predict 10–15, 17–22 and more than 25 TW of radiogenic heat: one assumes a mantle like enstatite chondrites, one combines carbonaceous chondrites with terrestrial samples, and one is motivated by the heat needed for realistic mantle convection [10]. The share of heat loss supplied by radioactivity, the Urey ratio, is therefore uncertain too [4].
No body sets a bar for a resolution. Geoneutrinos, the antineutrinos emitted by radioactive decay, measure uranium and thorium inside Earth directly [6, 11]; in 2022 Laura Sammon and William McDonough expected the next detectors, SNO+ and JUNO, to constrain the budget of heat-producing elements better [11].
What is known
Heat at the surface (1993–2019). Henry Pollack, Suzanne Hurter and Jeffrey Johnson compiled 24,774 heat-flow observations at 20,201 sites and found a global heat loss of 44.2 TW, more than half of it from Cenozoic ocean floor; their total includes corrections for heat carried by water circulating through the ocean crust, which conductive measurements miss [1]. Using 38,347 measurements and a half-space cooling model for young ocean floor, Davies and Davies obtained 47 ± 2 TW, overlapping the 46 ± 3 TW of Jaupart and colleagues [5]. A. M. Hofmeister and R. E. Criss argued in 2005 that the half-space cooling model, which gives infinite heat flux at ridge centres, inflates such totals, that the gap between modelled and measured fluxes is not due to circulating water, and that the data give 31 ± 1 TW [2]. In 2019 F. Lucazeau compiled about 70,000 measurements: on young seafloor, recent measurements average 92 milliwatts per square metre against 79 for older ones, which he relates to better quality and sampling where water circulates, and they give 40–42 TW in total, 3–5 TW less than a conductive cooling model [7].
Core and mantle (2008). Thorne Lay, John Hernlund and Bruce Buffett noted that estimates of the heat flow across the core–mantle boundary had "generally increased by a factor of 2 to 3" in a decade, implying a younger, faster-growing inner core and/or "significant radiogenic material in the outer core or lowermost mantle" [3]. What the outer core contains and whether the lowermost mantle is chemically distinct are open problems of their own: Light elements in Earth’s outer core and Chemical heterogeneity at the core–mantle boundary. Jun Korenaga estimated the bulk Earth Urey ratio at about 0.35 and the convective Urey ratio, for the mantle alone, at about 0.2, so that most of the mantle's heat loss is supported by cooling [4].
Geoneutrinos (2011–2026). Geoneutrinos pass through Earth virtually unaffected; those from uranium-238 and thorium-232 can be detected, but those from potassium-40 are below detection limits, so its roughly 4 TW comes from geochemistry [6, 10]. Combining KamLAND in Japan with Borexino in Italy, the KamLAND Collaboration found in 2011 that uranium and thorium supply 20.0 (+8.8/−8.6) TW, about half of the heat flow together with potassium, and concluded that "Earth's primordial heat supply has not yet been exhausted" [6]. In 2020 Borexino rejected the absence of a mantle signal at 99.0% confidence and estimated Earth's total radiogenic heat at 38.2 (+13.6/−12.7) TW, a convective Urey ratio of 0.78 (+0.41/−0.28), in tension at 2.4 standard deviations (2.4σ) with the models with the fewest heat-producing elements [8]. In 2022, with more than 18 years of data, KamLAND found its result consistent with the chondrite-based models, disfavoured the high group at 99.76% confidence and a fully radiogenic Earth at 5.2σ, or 4.0σ if all mantle uranium and thorium sit at the base of the mantle, and obtained a convective Urey ratio of 0.13 (+0.15/−0.06) [10]. It put total radiogenic heat at about 14.6 TW and noted that such estimates "can strongly depend on the accuracy of crust representation", especially within 500 km of the detector [10]. Sammon and McDonough found that this nearby crust supplies about half of the signal, mapped potassic volcanic rocks within 50 km of Borexino that its crustal model had left out, and estimated three times more uranium in the local upper crust, which lowers Borexino's estimate; with their revised local models, combined KamLAND and Borexino data "strongly favor" an Earth with about 20 TW of radiogenic power [11].
Recent results. In September 2026 the SNO+ Collaboration, at SNOLAB in Ontario, Canada, reported a geoneutrino signal of 48 (+14/−12) terrestrial neutrino units (TNU, one event per year per 10³² target protons); its model predicts 36.3 ± 8.7 TNU from uranium and 9.7 ± 2.3 TNU from thorium for 20 TW of total radiogenic heat, and a geological interpretation is under way [13]. The JUNO Collaboration in China expects to collect as many geoneutrinos as the whole existing world dataset in less than a year, and about 8% precision on the total flux over ten years [12].
Main approaches
Counting heat at the surface. Compile measurements and correct for heat carried by water in young ocean crust [1, 5]. Strongest evidence: compilations from 20,201 sites to about 70,000 measurements give 44–47 TW with model-based corrections [1, 5, 7]. Main difficulty: the correction rests on a cooling model that Hofmeister and Criss reject [2], totals from recent measurements still fall 3–5 TW short of the model [7], and Antarctica, Greenland, Africa and other regions are poorly sampled [5].
Counting geoneutrinos. Measure uranium and thorium at several sites and subtract the crust's share [6, 11]. Strongest evidence: KamLAND's 5.2σ result against a fully radiogenic Earth [10] and the combined result of about 20 TW [11]. Main difficulty: the nearby crust supplies about half of the signal [11], potassium is invisible [6], and for the highest-abundance models a 3σ detection of the mantle at JUNO over six years needs the lithospheric flux known to within 15% [12].
A chondritic Earth that is cooling fast. Radiogenic power is about 20 TW and most of the mantle's heat loss is cooling [4, 9]. Strongest evidence: a composition-based calculation giving 19.9 ± 3.0 TW [9], KamLAND's agreement with the chondrite-based models [10] and the combined geoneutrino result [11]. Main difficulty: KamLAND notes that these models may require changes to conventional thermal-evolution models [10]; Korenaga suggests that today's imbalance may result from the start of plate tectonics [4], while Davies and Davies report that such rapid cooling had been judged highly unlikely [5].
More radiogenic heat. Earth holds more heat-producing elements, as the high group assumes for realistic mantle convection [10], or is near a steady state in which radioactivity supplies most of a 31 TW total, as Hofmeister and Criss argue from an enstatite-chondrite composition [2]. Strongest evidence: Borexino's central estimate of 38.2 TW and convective Urey ratio of 0.78 [8]. Main difficulty: KamLAND disfavours the high group at 99.76% confidence [10], the revised local crust lowers Borexino's estimate [11], and the 31 TW total is below other compilations [5, 7].
Open sub-questions
- The SNO+ mantle signal. Once the crust around SNOLAB is modelled from local geology, does the 48 TNU signal leave a mantle contribution consistent with about 20 TW of total radiogenic heat [11, 13]?
- Borexino's near field. With the potassic volcanic rocks near the detector included, do Borexino and KamLAND agree on a single mantle composition [8, 10, 11]?
- JUNO's lithosphere. Can the lithospheric geoneutrino flux around JUNO be modelled to within 15%, the precision a 3σ mantle detection needs for the highest-abundance models [12]?
- Young ocean floor. How much of the 3–5 TW gap between totals from recent measurements and a conductive cooling model remains when only well-sampled young seafloor with circulating water is used [2, 7]?
- Radiogenic heat at depth. How much radiogenic material in the outer core or lowermost mantle would allow a two- to threefold higher core–mantle boundary heat flow without a young inner core [3]?
Work on this in Papex
If you have a testable answer to one of the sub-questions above, you can draft it as a private hypothesis in Papex, or prepare the draft from Claude or ChatGPT with the Papex connector. You review the exact version, sign it and decide whether to publish it; a public record that names this problem is listed on this page. A Papex record shows who signed which version and when. It is not peer review and does not mean the hypothesis is correct.
Key references
- Pollack, H. N., Hurter, S. J. & Johnson, J. R. (1993). Heat flow from the Earth's interior: analysis of the global data set. Reviews of Geophysics 31(3), 267–280. 10.1029/93RG01249 · Checked 15 Oct 2026
- Hofmeister, A. M. & Criss, R. E. (2005). Earth's heat flux revised and linked to chemistry. Tectonophysics 395(3–4), 159–177 (published online 2004). 10.1016/j.tecto.2004.09.006 · Checked 15 Oct 2026
- Lay, T., Hernlund, J. & Buffett, B. A. (2008). Core–mantle boundary heat flow. Nature Geoscience 1(1), 25–32. 10.1038/ngeo.2007.44 · Checked 15 Oct 2026
- Korenaga, J. (2008). Urey ratio and the structure and evolution of Earth's mantle. Reviews of Geophysics 46(2), RG2007. 10.1029/2007RG000241 · Checked 15 Oct 2026
- Davies, J. H. & Davies, D. R. (2010). Earth's surface heat flux. Solid Earth 1(1), 5–24. 10.5194/se-1-5-2010 · Checked 15 Oct 2026
- The KamLAND Collaboration (2011). Partial radiogenic heat model for Earth revealed by geoneutrino measurements. Nature Geoscience 4(9), 647–651. 10.1038/ngeo1205 · Checked 15 Oct 2026
- Lucazeau, F. (2019). Analysis and mapping of an updated terrestrial heat flow data set. Geochemistry, Geophysics, Geosystems 20(8), 4001–4024. 10.1029/2019GC008389 · Checked 15 Oct 2026
- Agostini, M., Altenmüller, K., Appel, S., Atroshchenko, V., Bagdasarian, Z., Basilico, D. et al. (Borexino Collaboration) (2020). Comprehensive geoneutrino analysis with Borexino. Physical Review D 101(1), 012009. 10.1103/PhysRevD.101.012009 · Checked 15 Oct 2026
- McDonough, W. F., Šrámek, O. & Wipperfurth, S. A. (2020). Radiogenic power and geoneutrino luminosity of the Earth and other terrestrial bodies through time. Geochemistry, Geophysics, Geosystems 21(7), e2019GC008865. 10.1029/2019GC008865 · Checked 15 Oct 2026
- Abe, S., Asami, S., Eizuka, M., Futagi, S., Gando, A., Gando, Y. et al. (KamLAND Collaboration) (2022). Abundances of uranium and thorium elements in Earth estimated by geoneutrino spectroscopy. Geophysical Research Letters 49(16), e2022GL099566. 10.1029/2022GL099566 · Checked 15 Oct 2026
- Sammon, L. G. & McDonough, W. F. (2022). Quantifying Earth's radiogenic heat budget. Earth and Planetary Science Letters 593, 117684. 10.1016/j.epsl.2022.117684 · Checked 15 Oct 2026
- Adam, T., Ahmad, S., Ahmed, R. et al. (JUNO Collaboration) (2026). Prospects for geoneutrino detection with JUNO. Chinese Physics C, accepted manuscript published online 13 February 2026 (Crossref lists only the corresponding author, Y.-F. Li). 10.1088/1674-1137/ae457d · Checked 15 Oct 2026
- Abreu, M., Allega, A., Anderson, M. R., Andringa, S., Asner, D. M., Auty, D. J. et al. (SNO+ Collaboration) (2026). Measurement of reactor antineutrino oscillations with 1.46 kilotonne-years of data at SNO+. Physical Review Letters 137(10), 101807. 10.1103/mv5z-pfb7 · Checked 15 Oct 2026
Source
Open, as listed by Wikipedia · Unsolved problems in geoscience. Checked 08 Oct 2026.
The status is the source’s. Papex does not decide whether a problem is solved.
Records on this problem
No public record addresses this problem yet.