{"id":"1ed921e4-7fcc-46c6-b7be-96ee6a8fcb8b","arxiv_id":"2507.12622","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"New JWST data robustly detect CH4 and CO2 in K2-18 b's atmosphere, indicating a water-rich interior; DMS, CH3SH, and N2O remain marginal, and abiotic organosulfur chemistry is a viable explanation.","lead":"JWST transmission spectra of the temperate sub-Neptune K2-18 b now robustly show methane and carbon dioxide in its atmosphere, pointing to a water-rich interior with at least 10 percent water by volume in a thick envelope or a thin atmosphere above a liquid ocean. The result sharpens the case for K2-18 b as a key target in the search for habitable ocean worlds, while explicitly leaving the existence of a liquid-water ocean unconfirmed.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The water-rich inference rests on the Yang & Hu (2024a) mapping, yet the same EPACRIS models overproduce CO, NH3, HCN, and CH3SH while sulfur-carbon chemistry is flagged as poorly constrained, so a dry high-metallicity envelope is not excluded.","rationale":"The observational core is strong: two independent reductions agree, three retrieval codes give consistent CH4/CO2 constraints, and repeated visits tighten the limits. I do not contest the CH4/CO2 detections. The load-bearing step is the interpretation of those abundances as bulk water. The reader's weakest_assumption already identifies the Yang & Hu mapping; I agree and sharpen it: the mapping's parent model fails on other species in the same dataset, so it is not independently validated for the species pair it is used to interpret. The paper's own limitation statements in Sections 5.3 and 6 flag the relevant uncertainty. The constrained retrievals in Appendix F further show that alternative compositions with H2O, NH3, or CO fit nearly as well, so the exclusivity language in the central claim is stronger than the evidence. A dedicated model-grid test with an independent code and varied sulfur/OCS kinetics would settle whether a dry envelope is truly excluded. If it is not, the paper should be revised to conditional or pending language; the current CONDITIONAL verdict remains appropriate.","tokens_in":43621,"tokens_out":9082,"duration_ms":108520,"concrete_test":"Take the retrieved 1σ ellipse in Table 5 (log CH4 = −1.06 +0.24/−0.37, log CO2 = −3.35 +0.66/−0.80) and recompute the Yang & Hu (2024a) grid for a 100×solar C/N/S envelope with H2O/H2 = 5:95 using an independent photochemical code, varying the OCS formation and H-abstraction rates (Table 9) across their published uncertainty. If any low-H2O model falls inside the retrieved ellipse, a dry envelope is not excluded and the 10–25% H2O claim fails; if all low-H2O models fall outside, the concern is retired.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative step is in Section 4: the retrieved CO2/CH4 ratio is converted to a bulk H2O/H2 of 10-25% by volume using the Yang & Hu (2024a) grid for a 100x solar C/H envelope. That grid is a product of EPACRIS, the same model family that, in Section 3.2.1 and Figure 7, substantially overproduces CO relative to the retrieval upper limit, overproduces NH3 by at least two orders of magnitude, overproduces HCN, and overproduces CH3SH. The authors attribute the NH3 problem to unconstrained interior nitrogen depletion and call the CO overproduction 'unexplained' (Section 7). Section 5.3 states that 'the interplay between sulfur and carbon chemistry in H2-dominated atmospheres remains poorly understood, with many reaction rates still unconstrained,' and Section 6 advises combining CO and OCS because OCS production kinetics are uncertain. If the same model's CO2/CH4 ratio responds to these unconstrained sulfur/OCS pathways, a massive envelope with H2O/H2 below the claimed 10% could still reproduce the observed CH4 and CO2. Because the small-atmosphere branch is only one of two allowed scenarios and is itself inferred, not directly observed, the 'conclusively water-rich' statement is not yet supported. This is a correctness risk, not an internal inconsistency.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents four new JWST/NIRSpec transit observations (two G235H, two G395H) of the temperate sub-Neptune K2-18 b and combines them with re-reduced NIRISS/SOSS and G395H data from earlier programs. Two independent NIRSpec reductions (Eureka! and ExoTEDRF) and one NIRISS reduction (NAMELESS) are shown to agree. Three retrieval frameworks (ExoTR, AURA, SCARLET) consistently detect CH4 at log(VMR) ≈ -1.0 to -1.1 and CO2 at approximately 10^-3.5 to 10^-2.5, with 2σ upper limits on H2O, NH3, CO, and HCN. The paper then uses self-consistent EPACRIS photochemical-climate models to argue that the observed CO2-to-CH4 ratio implies a massive high-metallicity envelope with 10–25% H2O by volume, or alternatively a thin H2 atmosphere above a liquid-water ocean, and concludes that K2-18 b has a water-rich interior. It also reports marginal, sub-3σ evidence for DMS, CH3SH, and N2O, and presents abiotic photochemical pathways for DMS and CH3SH in massive high-metallicity envelopes, updating the 2021 characterization roadmap for temperate sub-Neptunes.","tokens_in":43876,"tokens_out":9376,"duration_ms":95225,"significance":"If the central claim is accepted, this is an important step in sub-Neptune characterization: it would be the first robust demonstration that a temperate sub-Neptune's bulk volatile inventory is water-dominated, with direct implications for interior structure and habitability. The observational core is genuinely strong: the paper ships multiple independent reduction pipelines, three independent retrieval frameworks with consistent CH4 and CO2 constraints, quantitative Bayes-factor detection tests, public data products, and transparent constrained-retrieval experiments that probe alternative compositions. The repeated-visit analysis is also a valuable methodological contribution, showing that tentative DMS/CH3SH/N2O signals weaken as more data are added. However, the headline water-rich conclusion is not a direct measurement: it depends on the Yang & Hu (2024a) model grid, whose reliability for the specific CO2/CH4 diagnostic is undercut by the same models' large overpredictions of CO, NH3, HCN, and CH3SH, and on a small-atmosphere branch that is inferred from theory rather than directly observed.","major_comments":[{"comment":"The central quantitative inference of a bulk H2O/H2 ratio of 10–25% by volume is obtained by comparing the retrieved CO2-to-CH4 ratio with the Yang & Hu (2024a) grid for a 100×solar C/H envelope. The same EPACRIS model family, however, overproduces CO relative to the retrieved upper limit, overproduces NH3 by at least two orders of magnitude, and produces CH3SH above the data-allowed level; Section 5.2.2 calls the CO overproduction \"unexplained,\" and Section 5.3 states that \"the interplay between sulfur and carbon chemistry in H2-dominated atmospheres remains poorly understood, with many reaction rates still unconstrained.\" Because the model CO2/CH4 ratio could respond to exactly these unconstrained OCS/sulfur pathways, the mapping from observed CO2/CH4 to a water-rich bulk composition is not yet robust, and a dry, high-metallicity H2-dominated envelope is not excluded. This is a load-bearing correctness risk in the central claim; the paper should either provide a quantitative propagation of the model-chemistry uncertainty into the inferred H2O/H2 range or explicitly reframe the 10–25% inference as conditional on the EPACRIS/Yang & Hu chemistry being accurate.","section":"Section 4; Sections 3.2.1 and 5.3"},{"comment":"The statement that the results \"conclusively demonstrate that K2-18 b has a water-rich interior\" is stronger than the evidence supports. The massive-envelope branch depends on the model grid criticized above. The small-atmosphere branch assumes, rather than demonstrates, that a thin H2 atmosphere must overlie a liquid-water ocean (Section 5.2, citing Gupta et al. 2025), and the small-atmosphere model in Section 3.2.2 adopts the retrieved CH4 and CO2 abundances as lower-boundary conditions, so it is a consistency test rather than an independent prediction. In addition, the nondetections of H2O, NH3, and CO are not decisive: Sections 5.1, 5.2.1, and 5.2.2 show that constrained retrievals with elevated H2O, NH3, or CO produce visually acceptable fits with likelihood odds of only about 7:1, 3:1, and 2:1 against them, respectively. The conclusion should be softened to state that the data are consistent with a water-rich interior under the adopted photochemical/structural models, or the paper should provide a genuinely independent test of the water-rich hypothesis.","section":"Abstract and Section 7; Sections 5.1, 5.2.1, 5.2.2"}],"minor_comments":[{"comment":"The text contains the stray phrase \"as shown hello thank youFigures 6 and A5,\" which appears to be an editing artifact and should be removed.","section":"Section 3.1.2"},{"comment":"The list of six primary molecules reads \"H2O, CH4, CO2, NH3, CO2, HCN,\" with CO2 repeated; presumably CO was intended in one of the two positions.","section":"Section 3.1.2"},{"comment":"The quoted 10–25% H2O range in Section 4 is based on the ExoTR abundances, but AURA returns a CO2 abundance higher by about 0.8 dex (log CO2 = -3.35^{+0.66}_{-0.80} versus -2.52^{+0.48}_{-0.57}); the text should state explicitly that the compositional range is conditional on the ExoTR retrieval and the Yang & Hu (2024a) grid, or provide a range that spans both retrieval frameworks.","section":"Tables 5 and 7"},{"comment":"The caption spells \"Madhusudan et al. (2023)\" with a missing 'h'; the spelling should be corrected to \"Madhusudhan\" for consistency with the reference list.","section":"Figure A2 caption"},{"comment":"The skin-temperature formula is rendered ambiguously as \"T skin = 1 21/4\" before the bracket; please typeset the (1/2)^{1/4} factor clearly.","section":"Section 5.1, Eq. (3)"}],"recommendation":"major_revision","confidential_remarks":"This is a strong observational paper with an overreaching headline. The CH4/CO2 detections and the multi-framework consistency are publishable and important. In revision, the authors should either add a quantitative treatment of the model-chemistry uncertainty that propagates into the H2O/H2 inference, or change the abstract and conclusion to present the water-rich interior as a model-dependent interpretation. I do not see grounds for rejection, but the current wording would mislead readers about the certainty of the central claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The useful part of this paper is the data: four new NIRSpec transits, two independent reductions, three retrieval codes, and a CO2 detection that settles the earlier dispute with Schmidt et al. The CH4 and CO2 constraints are consistent across frameworks, and the paper is honest about its own caveats, including the constrained-retrieval tests that fit almost as well. That is real progress and worth taking seriously.\n\nThe soft spot is the jump from retrieved CO2/CH4 to a bulk H2O of 10–25%. That step goes through the Yang & Hu (2024a) grid, and the same EPACRIS models overproduce CO (with the paper calling it 'unexplained'), overproduce NH3 by orders of magnitude, and rely on sulfur chemistry the authors describe as poorly constrained. If that mapping is biased, a dry high-metallicity envelope is not excluded. The two-branch argument — massive water-rich envelope or thin atmosphere over an ocean — is only conclusive if both branches are secure, and the small-atmosphere branch is itself inferred, not observed. So 'conclusively demonstrate a water-rich interior' overstates what the evidence supports. The nondetections of H2O, NH3, and CO are also weaker than the abstract implies: forced-abundance retrievals fit within odds of 7:1, 3:1, and 2:1, respectively, which is not decisive.\n\nThe shifted-average spectrum is constructed with manually applied offsets. The authors checked against the direct average and got similar results, so this is a minor concern, not a fatal one. The DMS/CH3SH abiotic pathway is novel and tied to lab data, but the rate coefficients carry large uncertainties; the paper appropriately stops short of claiming a detection.\n\nWho gets value from this: anyone working on sub-Neptune atmospheres, JWST transmission spectroscopy, or biosignature interpretation. The cross-framework consistency and public data products make it a useful reference even if the interior claim does not fully hold.\n\nRecommendation: send to peer review. The observations deserve referee time. But the authors should be pushed to either soften the 'conclusively water-rich' language or provide independent validation of the CO2/CH4-to-H2O mapping. As it stands, the central claim is conditional, not conclusive.","headline":"Strong new transit data and a robust CO2 detection, but the 'conclusively water-rich' conclusion rests on a model mapping the paper's own models do not fully validate.","tokens_in":44527,"tokens_out":2362,"would_cite":true,"duration_ms":28620,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Repeated JWST spectra of K2-18 b show a water-rich interior, whether through a 10–25% water envelope or a thin atmosphere over a liquid ocean.","keywords":["exoplanet atmospheres","transmission spectroscopy","sub-Neptunes","K2-18 b","water-rich interior","methane","carbon dioxide","biosignatures"],"falsifier":"Measure the CO2-to-CO ratio with deeper NIRSpec/G395H transits: a confirmed CO2/CO below 1 would rule out the thin-atmosphere-over-ocean scenario, while CO2/CO above 3 would rule out standard massive-envelope models. Laboratory or quantum-chemical rate coefficients for the reaction CH3S + CO → CH3 + OCS would directly test the mapping used to infer the planet's water content.","tokens_in":43375,"feed_emoji":"🌊","tokens_out":5989,"duration_ms":61309,"temperature":0.7,"pith_summary":"This paper reports four new JWST transit observations of the temperate sub-Neptune K2-18 b and combines them with earlier data into a high-precision transmission spectrum. The authors claim the spectrum robustly detects methane (CH4) and carbon dioxide (CO2) while showing no clear water vapor, ammonia, or carbon monoxide. From the measured CO2-to-CH4 ratio they infer that the planet's interior is water-rich: either a massive hydrogen envelope with 10–25% water by volume, or a thin hydrogen atmosphere above a liquid-water ocean. The marginal signals of dimethyl sulfide, methyl mercaptan, and nitrous oxide are all below 3σ, and the paper's photochemical models show the first two can form abiotically in massive high-metallicity envelopes.","feed_headline":"K2-18 b is water-rich, JWST spectra show","feed_subtitle":"Methane and carbon dioxide detections point to a thick water-bearing envelope or a liquid ocean.","key_machinery":"The argument turns on using the observed CO2-to-CH4 ratio as a proxy for the bulk H2O-to-H2 ratio in a massive hydrogen envelope. In water-rich envelopes, higher water content shifts carbon chemistry toward CO2 relative to CH4; the paper compares the retrieved abundance ratio to a precomputed grid of self-consistent pressure–temperature and chemistry models to read off an interior H2O fraction of 10–25% by volume. The alternative branch of the argument uses the absence of NH3 and CO together with the high CO2-to-CO ratio, as predicted for a thin H2 atmosphere resting on a liquid-water layer, to support the ocean scenario.","core_discovery":"The central claim is that K2-18 b has a water-dominated interior. The evidence is the simultaneous presence of CH4 and CO2 at abundances that, according to the paper's self-consistent atmospheric models, can only be produced by either a massive H2 envelope with roughly 100 times solar metallicity and a bulk H2O content of 10–25% by volume, or a small H2 atmosphere over a liquid-water ocean. The spectrum contains no detectable H2O, NH3, or CO; the H2O absence is interpreted as a cold trap with a high Bond albedo, and the NH3 and CO nondetections favor the thin-atmosphere-over-ocean scenario, though massive-envelope alternatives with interior nitrogen sequestration are not ruled out. The paper concludes that whether or not the ocean exists, the planet's bulk volatile inventory is water-rich.","pith_inferences":["If the interior is water-rich, similar temperate sub-Neptunes such as TOI-270 d become high-priority targets for the same repeated-transit strategy, since their CO2-to-CH4 ratios could be measured the same way.","The paper's demonstration that repeated visits weaken tentative DMS signals suggests that single-epoch biosignature claims on sub-Neptunes should be treated with caution until reproducibility is established.","A testable corollary is that ground-based high-resolution spectroscopy targeting CO and OCS could break the degeneracy between the massive-envelope and ocean scenarios without waiting for more JWST transits.","The cold-trap interpretation implies that transmission spectra of cool sub-Neptunes systematically under-report bulk H2O, so interior water content must be inferred from carbon chemistry—a lesson that transfers to other planets in this class."],"forward_implications":["If the water-rich interior claim holds, K2-18 b joins a class of temperate sub-Neptunes whose volatile inventory is dominated by water rather than H2/He, informing formation models that accrete icy solids.","The nondetection of water vapor implies an efficient cold trap and a Bond albedo of at least 0.3 at 2σ, with values near 0.5–0.7, close to what climate models need to avoid runaway greenhouse and sustain an ocean.","Dimethyl sulfide and methyl mercaptan should not be treated as standalone biosignatures on massive high-metallicity sub-Neptunes, because the paper's models produce them abiotically; co-production of C2H6 or C2H2 could instead indicate biogenic surface fluxes.","The CO2-to-CO ratio emerges as a practical diagnostic for future observations: values above roughly 3 favor a thin atmosphere over an ocean, while values below roughly 1 favor a massive H2 envelope."],"supporting_citations":[{"why":"Supplies the earlier NIRISS/G395H spectrum and the tentative DMS signal that this work reanalyzes and extends with repeated visits.","marker":"Madhusudhan et al. 2023"},{"why":"Raises the reduction-dependent ambiguity of the CO2 detection that the new repeated observations are designed to resolve.","marker":"Schmidt et al. 2025"},{"why":"Provides the theoretical grid mapping the CO2-to-CH4 ratio to the bulk H2O-to-H2 ratio in massive envelopes, the load-bearing bridge to the 10–25% water inference.","marker":"Yang & Hu 2024a"},{"why":"Predicts a CO2-rich, NH3-poor atmosphere with CO2/CO greater than about 3 for a thin atmosphere over a liquid-water ocean, the basis of the ocean-scenario diagnostic.","marker":"Hu et al. 2021"},{"why":"Argues that a small H2 atmosphere necessarily rests on a liquid-water layer, otherwise the atmosphere and interior would mix, supporting the ocean interpretation.","marker":"Gupta et al. 2025"},{"why":"Supports the idea that high CH4 can be a primordial remnant and constrains NH3 photochemistry in small atmospheres.","marker":"Yu et al. 2021"},{"why":"Challenges photochemical CH4 production in small H2 atmospheres, motivating the alternative of primordial or biogenic CH4.","marker":"Wogan et al. 2024"},{"why":"Provides DMS photochemistry and the threshold behavior of organosulfur buildup used for comparison with the paper's ocean-surface flux models.","marker":"Tsai et al. 2024"},{"why":"Gives the climate-model requirement of high planetary albedo for a liquid-water ocean, used to check consistency with the inferred Bond albedo.","marker":"Leconte et al. 2024"}],"fun_headline_variants":["JWST reveals water-rich interior for K2-18 b","K2-18 b: water-rich interior, maybe a hidden ocean","Water-rich sub-Neptune K2-18 b confirmed by JWST","Methane and CO2 point to water-rich K2-18 b interior","K2-18 b's interior is water-rich, JWST spectra say"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The inference of 10–25% interior water depends on the theoretical mapping from the observed CO2-to-CH4 ratio to the H2O-to-H2 ratio in massive hydrogen envelopes; if that mapping is biased by poorly constrained sulfur chemistry such as OCS formation, a massive envelope with much less water could still reproduce the spectrum.","fun_headline_variants_meta":{"raw":{"variants":["JWST reveals water-rich interior for K2-18 b","K2-18 b: water-rich interior, maybe a hidden ocean","Water-rich sub-Neptune K2-18 b confirmed by JWST","Methane and CO2 point to water-rich K2-18 b interior","K2-18 b's interior is water-rich, JWST spectra say"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001108,"raw_usage":{"total_tokens":4686,"prompt_tokens":1083,"completion_tokens":3603,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":699,"completion_tokens_details":{"reasoning_tokens":3506}},"tokens_in":699,"tokens_out":3603,"duration_ms":35703,"temperature":1.0,"reasoning_tokens":3506,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T16:43:38.792270+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the CO2-to-CO ratio with deeper NIRSpec/G395H transits: a confirmed CO2/CO below 1 would rule out the thin-atmosphere-over-ocean scenario, while CO2/CO above 3 would rule out standard massive-envelope models. Laboratory or quantum-chemical rate coefficients for the reaction CH3S + CO → CH3 + OCS would directly test the mapping used to infer the planet's water content.","supporting_citations":[{"cited_title":"F., & Schwieterman, E","cited_arxiv_id":null,"evidence_quote":"Provides DMS photochemistry and the threshold behavior of organosulfur buildup used for comparison with the paper's ocean-surface flux models."}],"review_version":1}