{"id":"de21b95d-dba2-4c5e-9e1d-7918c69d7b7b","arxiv_id":"2505.02641","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A synthesis of geochemical and geoneutrino data concludes that Earth's present radiogenic power is about 20 TW and the bulk silicate Earth is enriched in refractory lithophile elements by 2.5 to 2.7 times CI chondrites.","lead":"This paper updates the compositional model of the bulk silicate Earth and uses geoneutrino measurements to argue that Earth's radiogenic heat production is about 20 terawatts. It claims that the refractory element enrichment of the bulk silicate Earth is 2.5 to 2.7 times that of CI chondrites, and that Earth is more volatile-depleted than Mars.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 20 TW affirmation rests on a geoneutrino uncertainty budget that omits the dominant geological-model uncertainty and is inconsistent with the paper's own 1σ rejection claim.","rationale":"The reader's weakest assumption identifies the sensitivity of geoneutrino-derived power to local crustal models, and I agree this is the central soft spot. My stress test sharpens it: the paper's own quoted uncertainty (20±8 TW) is inconsistent with its subsequent claim of 1σ rejection of >2.5× CI models, since Table 4 includes >2.5× CI models (Palme & O'Neill, Jagoutz) whose total power lies inside 12–28 TW. The abstract's '≤20 TW' also conflicts with the body's '20±8 TW'. The Tuscan example in Section 8 demonstrates a ~10 TW shift from a geological model change, larger than the quoted ±8 TW mantle uncertainty, indicating that the dominant uncertainty is not propagated. This makes the central claim's strength ('affirms', 'rejects') exceed what the data and analysis support. The paper remains a useful compositional review, and the BSE table is a valuable update, so a conditional acceptance with revision of the abstract and the 1σ statement is appropriate. No basis for rejection or unverdictability emerges; the core synthesis is credible but the headline certainty is overclaimed.","tokens_in":43529,"tokens_out":5124,"duration_ms":55573,"concrete_test":"Recompute the combined KamLAND+Borexino geoneutrino inversion using the published spectra (Abe et al. 2022; Agostini et al. 2020) and at least two published crustal models (e.g., Coltorti et al. 2011 without Tuscan rocks; Sammon and McDonough 2022 with them, plus an independent third model if available). Propagate crustal U/Th abundance uncertainties by Monte Carlo and report the resulting 1σ interval for global radiogenic power. If the interval is wider than 20±8 TW (e.g., roughly 10–30 TW), the paper's affirmation of 20 TW and its 1σ rejection of >2.5× CI models are not supported; if the interval remains 20±8 TW under all models, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 8 reports mantle 13±8 TW and crust 7±1 TW, i.e. 20±8 TW, and then states that at 1σ the geoneutrino data reject BSE models with >2.5× CI enrichment. With 20±8 TW the 1σ interval is 12–28 TW. Table 4 lists Palme & O'Neill at 21.6 TW (U=22.9 ppbw, >2.5× CI) and Jagoutz at 23.8 TW (U=26 ppbw), both inside this interval; only Turcotte & Schubert (29.3 TW) falls outside. So the rejection claim is not supported by the quoted uncertainty. Moreover, the abstract says '≤20 TW' while the text says '20±8 TW'; these are different statements. The Tuscan example in Section 8 (mantle estimate shifting from 30 TW to 20 TW when local magmatic rocks are included) shows geological-model uncertainty of order 10 TW, larger than the quoted ±8 TW mantle uncertainty, so the quoted uncertainty omits the dominant term. The 20 TW affirmation therefore rests on a self-cited re-analysis (Sammon and McDonough, 2022) whose model uncertainty is not fully propagated.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper is a broad review and reinterpretation of Earth's bulk composition, origin, and energy budget. It updates the McDonough and Sun (1995) bulk silicate Earth (BSE) model, argues that the terrestrial planets are neither solar-photosphere nor CI-chondrite analogs, and uses the NC/CC meteorite dichotomy to frame early solar system accretion. The central quantitative claims are that KamLAND and Borexino geoneutrino data affirm a present-day radiogenic power of about 20 TW (abstract: \"≤20 TW\"; text: \"20±8 TW\"), that the BSE has refractory lithophile elements at 2.5–2.7 times CI chondrite, and that bulk Earth and Mars are enriched in refractory elements by about 1.9 times CI. The paper also discusses mantle convection, volatile and noble gas budgets, core composition, and future neutrino geoscience experiments.","tokens_in":43877,"tokens_out":6448,"duration_ms":76900,"significance":"If the 20 TW geoneutrino result were robust, it would be a landmark global-scale constraint on Earth's radioactive heat budget and would help discriminate among competing BSE compositional models. The paper's most useful contributions are the updated BSE table (Table 3), the side-by-side comparison of published BSE models and their heat production (Table 4), the clear separation of the geochemical BSE derivation from the geoneutrino measurement, and the careful discussion of biases in the meteorite record. The paper is transparent about the subjective nature of many abundance uncertainties and explicitly states that a full error propagation is not possible, which is an honest and important caveat. However, several of the headline statements, especially the abstract's \"≤20 TW\" and the 1σ rejection of >2.5× CI models, go beyond what the paper's own uncertainty budget supports. The future SNO+ and JUNO measurements are correctly identified as tests that could sharpen or overturn the present inference.","major_comments":[{"comment":"The claim that \"at the 1σ level, the geoneutrino data reject bulk Earth compositional models that propose high concentrations of refractory lithophile elements (>2.5× CI)\" is not supported by the paper's own quoted uncertainty. The text gives a total radiogenic power of 20±8 TW, i.e., a 1σ interval of 12–28 TW. Table 4 lists Palme and O'Neill (2014) at 21.6 TW with U = 22.9 ppbw (about 2.9× CI) and Jagoutz et al. (1979) at 23.8 TW with U = 26 ppbw (about 3.3× CI); both models exceed 2.5× CI and both fall inside the 12–28 TW interval. Only the Turcotte and Schubert (2002) model at 29.3 TW lies outside. The data therefore favor, but do not reject at 1σ, the >2.5× CI class of models. This statement should be softened or the uncertainty must be reduced.","section":"Section 8, abstract"},{"comment":"The abstract states that geoneutrino data \"affirm that the Earth has ≤20 TW of radiogenic power,\" while Section 8 reports \"a total of 20±8 TW.\" These are materially different statements: a symmetric 1σ interval of 12–28 TW cannot be summarized as one-sided upper bound of 20 TW without an explicit prior or a one-sided statistical treatment. The abstract either must be revised to \"20±8 TW\" or the paper must justify the change to a one-sided claim. As written, the two statements are internally inconsistent and the stronger abstract claim is not derived from the quoted uncertainty.","section":"Abstract / Section 8"},{"comment":"The quoted uncertainty of 20±8 TW appears to omit the dominant geological-model uncertainty. The text correctly states that the local lithosphere contributes about 50% of the geoneutrino signal and then describes how including the Tuscan magmatic rocks in the Borexino near-field model changed the mantle estimate from 30 TW to 20 TW, a 10 TW shift. That model-driven shift is comparable to or larger than the quoted mantle uncertainty of ±8 TW. Unless the quoted uncertainties already propagate the full covariance of the crustal models around KamLAND and Borexino, the total uncertainty is understated and the \"20 TW affirmation\" is not a robust experimental result but a model-dependent estimate. The paper should either include this term in the uncertainty budget or explicitly state that the 20 TW value is conditional on the adopted crustal models.","section":"Section 8"},{"comment":"Section 6.4 states that \"one cannot conduct a full and rigorous error propagation of all of the uncertainties as is carried out in physics experiments\" and describes the Table 3 uncertainties as \"a subjective judgment.\" Given that admitted limitation, the abstract's wording that the data \"set the proportions of refractory lithophile elements in the bulk silicate Earth at 2.5 to 2.7 times that in CI chondrites\" overstates the certainty of the BSE model. The EFRLE value of 2.65 and its range are not derived through a formal statistical procedure, and the same subjective uncertainties feed into the heat-production comparison in Table 4. The paper should either supply a more rigorous uncertainty for EFRLE or qualify the abstract's claim as an estimate rather than a determination.","section":"Section 6.4 / Table 3"}],"minor_comments":[{"comment":"The word \"terrawatts\" should be \"terawatts.\"","section":"Section 8"},{"comment":"The sentence \"a PPD is a rotating gas and dust cloud with a composition comparable to that of the its star\" contains a duplicated article; it should read \"that of its star.\"","section":"Section 3"},{"comment":"The text refers to a \"T50 accretion age\" for Mars, but the same section uses \"τaccretion\"; the notation should be made consistent and the typo corrected.","section":"Section 9"},{"comment":"The specific heat capacity is given as \"1,000 J kg−1 s−1\"; the correct unit for specific heat capacity is J kg−1 K−1.","section":"Section 9"},{"comment":"The abstract contains \"the Suns O/Fe,\" which should be \"the Sun's O/Fe\" with an apostrophe.","section":"Abstract"},{"comment":"The phrase \"the closet 250 km\" appears to be a typo for \"the closest 250 km\" or \"the nearest 250 km\"; please clarify.","section":"Section 8"}],"recommendation":"major_revision","confidential_remarks":"This manuscript is largely a synthesis and reinterpretation of the author's own prior work, including the geoneutrino re-analysis (Sammon and McDonough 2022) on which the 20 TW claim rests. That is not a defect in itself, but it means the paper does not provide an independent confirmation of the central geoneutrino result, and the novelty relative to the author's earlier publications (e.g., Yoshizaki and McDonough 2021; Sammon and McDonough 2022) is limited. The paper would be a stronger fit for a review-oriented venue or as a chapter-length synthesis; for a research journal the headline claims need to be reconciled with the paper's own uncertainty budget before acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read it for the updated BSE table and the review of meteoritic biases; don't take the abstract's celebration of 20 TW at face value. The paper does some real work: the Nb revision from 0.75 to 0.605 ppmw is defended, the Th/U constraint from Wipperfurth et al. is used sensibly, and the comparison of BSE models in Table 4 is handy. The synthesis of NC/CC dichotomy and PPD evolution is competent.\n\nThe soft spot is the central claim. Section 8 quotes 20±8 TW (mantle 13±8 + crust 7±1), then says at 1σ the geoneutrino data reject BSE models with >2.5× CI enrichment. With ±8, the 1σ interval is 12–28 TW, which includes Palme & O'Neill (21.6 TW) and Jagoutz (23.8 TW); only Turcotte & Schubert (29.3 TW) is outside. So the rejection claim doesn't follow from the quoted uncertainties. The abstract's '≤20 TW' also doesn't match the text's 20±8 TW; those are different statements. And the Tuscan example right there in Section 8 shows the crustal model shifts the mantle estimate by about 10 TW (30 to 20), which is larger than the quoted ±8. So the uncertainty budget omits the largest term. These aren't nitpicks; they're about the main argument.\n\nThe reliance on Sammon and McDonough (2022) for the re-analysis is a self-citation but not a flaw by itself; the problem is that the model uncertainty from that analysis isn't fully propagated. The paper honestly admits rigorous error propagation is impossible for the BSE, which is fine, but then the 1σ rejection language is too strong.\n\nBottom line: it's a credible review and the updated BSE table will be cited. The geoneutrino synthesis needs revision: fix the abstract, propagate the crustal model uncertainty, and soften the rejection claim to something like 'the data are most consistent with ~20 TW' rather than 'reject >2.5× CI'. A serious referee should see this; it deserves review, not desk rejection.","headline":"Useful review with an updated BSE table, but the headline geoneutrino claim overstates what the data show and the paper's own uncertainty budget undercuts its rejection claim.","tokens_in":44363,"tokens_out":2300,"would_cite":true,"duration_ms":23356,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Geoneutrino measurements place Earth's radiogenic power at about 20 terawatts and fix the bulk silicate Earth's refractory element enrichment at 2.5 to 2.7 times CI chondrite.","keywords":["bulk silicate Earth","geoneutrino","radiogenic heat production","pyrolite","refractory lithophile elements","CI chondrite","planetary accretion","mantle convection"],"falsifier":"Run the same geoneutrino inversion for a new detector with an independently characterized local crust, for example JUNO; if the total radiogenic power required to fit its observed signal falls outside the 12 to 28 TW range, the paper's 20 TW claim is wrong.","tokens_in":43339,"feed_emoji":"🌍","tokens_out":10365,"duration_ms":121190,"temperature":0.7,"pith_summary":"The paper sets out to show that the Earth's radiogenic power supply is now a measured quantity, not a free parameter: the combined geoneutrino signal recorded by detectors in Japan and Italy puts the total at about 20 terawatts, with roughly 13 from the mantle and 7 from the continental crust. This matters because Earth's surface heat loss is 46 terawatts, so the radioactive fraction determines how much primordial heat remains to drive mantle convection, plate tectonics, and the geodynamo. The same data fix the bulk silicate Earth's refractory lithophile elements at 2.5 to 2.7 times CI chondrite, which narrows the range of acceptable compositional models of the planet. Along the way, the paper argues that the rocky planets are not direct compositional copies of the Sun or of CI chondrites, and that Earth assembled rapidly from differentiated inner-solar-system planetesimals.","feed_headline":"Geoneutrinos put Earth's radiogenic power near 20 terawatts","feed_subtitle":"The same flux data fix the mantle's refractory elements at 2.5–2.7 times CI chondrite, narrowing the composition debate.","key_machinery":"The load-bearing object is the geoneutrino: an electron antineutrino produced by beta decay of uranium and thorium in the Earth's interior. Counting geoneutrinos with liquid-scintillator detectors gives the only direct, whole-planet measure of the U and Th budget, which is then converted into terawatts of radiogenic power. The conversion rests on a model of each detector's local lithosphere, which supplies about half the signal, plus a mantle contribution of about a quarter and a distant-lithosphere contribution of about a quarter; the paper shows that changing the local model (adding the U- and Th-rich Tuscan magmatic rocks near the Italian detector) moves the mantle power estimate from 30 TW down to about 20 TW. A second supporting device is the pyrolite compositional model of the bulk silicate Earth, built from peridotite residues and basalt melts, which fixes the refractory lithophile elements at 2.65 times CI and supplies the Th and U abundances (20.6 ppbw U, Th/U about 3.8) that the geoneutrino results are compared against. The paper also uses a newly precise Pb-isotope ratio (KPb = 3.90) to tie the BSE's Th/U to the chondritic value independently of the neutrino data.","core_discovery":"The paper's central discovery claim is that Earth has about 20 TW of present-day radiogenic power, derived from the U and Th content of the whole planet as seen in geoneutrino flux. It reports that the mantle produces about 13±8 TW and the continental crust about 7±1 TW, leaving about 26 TW of surface heat flux to be primordial in origin. A combined analysis of the two geoneutrino experiments also places the bulk silicate Earth's refractory lithophile element enrichment at 2.5 to 2.7 times CI carbonaceous chondrite, consistent with a pyrolite mantle composition. The paper further concludes that the bulk Earth and Mars share a refractory enrichment of about 1.9 times CI, while Earth is more volatile-depleted and less oxidized than Mars. These results are presented as an update and confirmation of earlier compositional models, now tied to a direct measurement of the planet's radioactive content.","pith_inferences":["The sensitivity of the result to local geology suggests that every future geoneutrino site should be paired with a pre-registered, high-resolution crustal model; otherwise the mantle component remains hostage to the same 10 TW swings seen at Borexino.","If the 20 TW budget is correct, then any geochemical model invoking a large hidden reservoir of heat-producing elements in the deep mantle or core would need to store those elements without contributing to the geoneutrino flux, which is physically difficult; the paper's Th/U constraint already hints at this tension.","Extending the method to a mobile ocean-bottom detector, which the paper mentions as a future goal, would test whether the mantle's U and Th are as homogeneous as the pyrolite model assumes; strong lateral variation in geoneutrino flux would challenge the single-value BSE budget."],"forward_implications":["About 26 TW of Earth's 46 TW surface heat flux would be non-radiogenic, implying substantial primordial power from core cooling and mantle secular cooling rather than radioactive decay.","BSE models with roughly 30 ppb U and 30 TW of radiogenic power, the high end of the published range, would be rejected at the 1-sigma level by geoneutrino data.","The continental crust would hold 30 to 40 percent of the planet's potassium, thorium, and uranium, which requires mantle processing beyond the upper 660 km and supports whole-mantle convection.","The bulk Earth's refractory element enrichment would be about 1.9±0.2 times CI, a tighter anchor for planetary accretion models than the earlier factor-of-three spread among BSE models.","New detectors with longer exposure or larger target mass will provide a direct, independent check of the 20 TW budget within a few years."],"supporting_citations":[{"why":"Supplies the KamLAND geoneutrino flux measurement that anchors the U and Th budget and thus the radiogenic power estimate.","marker":"Abe et al. (2022)"},{"why":"Supplies the Borexino final geoneutrino analysis, the second independent flux measurement used in the combined result.","marker":"Agostini et al. (2020)"},{"why":"Provides the initial geological model of the region around Borexino whose omission of Tuscan magmatic rocks is the key local-model uncertainty.","marker":"Coltorti et al. (2011)"},{"why":"Shows that including the Tuscan magmatic rocks lowers the inferred mantle power from about 30 TW to about 20 TW.","marker":"Sammon and McDonough (2022)"},{"why":"Establishes the pyrolite BSE composition and the 2.65 times CI refractory lithophile enrichment that the geoneutrino results are compared against.","marker":"McDonough and Sun (1995)"},{"why":"Provides the time-integrated Pb-isotope ratio KPb = 3.90 that fixes the BSE Th/U ratio independently of geoneutrino data.","marker":"Wipperfurth et al. (2018)"},{"why":"Reports the first geoneutrino flux measurement and notes that about half the signal comes from the local lithosphere around the detector.","marker":"Araki et al. (2005)"},{"why":"Supplies the bulk Earth and Martian refractory element enrichment of about 1.9 times CI used for planet-scale comparisons.","marker":"Yoshizaki and McDonough (2021)"}],"fun_headline_variants":["Geoneutrinos fix Earth's radiogenic power at 20 TW","Earth's radioactive heart: 20 TW from geoneutrino data","Geoneutrino flux pins Earth's radiogenic heat to 20 TW","Earth's composition and power budget tied to geoneutrinos"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The 20 terawatt number assumes that the uranium and thorium content of the rocks within a few hundred kilometers of each neutrino detector is known well enough to subtract; roughly half of the measured signal comes from this local crust, and the paper shows that changing one regional rock package shifts the mantle estimate by 10 terawatts.","fun_headline_variants_meta":{"raw":{"variants":["Geoneutrinos fix Earth's radiogenic power at 20 TW","Earth's radioactive heart: 20 TW from geoneutrino data","Geoneutrino flux pins Earth's radiogenic heat to 20 TW","Earth's composition and power budget tied to geoneutrinos"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000717,"raw_usage":{"total_tokens":3296,"prompt_tokens":1097,"completion_tokens":2199,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":713,"completion_tokens_details":{"reasoning_tokens":2121}},"tokens_in":713,"tokens_out":2199,"duration_ms":17613,"temperature":1.0,"reasoning_tokens":2121,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T00:45:55.068207+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same geoneutrino inversion for a new detector with an independently characterized local crust, for example JUNO; if the total radiogenic power required to fit its observed signal falls outside the 12 to 28 TW range, the paper's 20 TW claim is wrong.","supporting_citations":[{"cited_title":", author Boraso, R","cited_arxiv_id":null,"evidence_quote":"Provides the initial geological model of the region around Borexino whose omission of Tuscan magmatic rocks is the key local-model uncertainty."},{"cited_title":", author Guo, M","cited_arxiv_id":null,"evidence_quote":"Provides the time-integrated Pb-isotope ratio KPb = 3.90 that fixes the BSE Th/U ratio independently of geoneutrino data."}],"review_version":1}