{"id":"7812ef85-4f2e-46b4-9acf-797c8be0947b","arxiv_id":"2411.15117","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"PDS 70b has a carbon-to-oxygen ratio of 0.28 (+0.20/-0.12), similar to its host star and lower than the surrounding disk gas.","lead":"Using Keck/KPIC high-resolution spectra, researchers detected carbon monoxide and water in the atmosphere of the young planet PDS 70b and measured its carbon-to-oxygen ratio for the first time. The value is consistent with the host star, suggesting the planet's building materials were likely dust and ice rather than disk gas.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Retrieval fixes a cloudless Sonora P-T profile that the paper's own BT-Settl forward model disfavors (log10 BF=0.7); the C/O upper limit may shift if this assumption is relaxed.","rationale":"The reader's conditional verdict is justified. The paper is transparent about its assumptions and the marginal SNR, but the Section 4.2 comparison means there is direct evidence within the paper that the retrieval's fixed P-T/cloudless model is disfavored. This is more specific than a generic concern about clouds. The proposed test is computationally cheap: it reuses the existing retrieval framework and priors, and it directly targets the quantity that the headline claim rests on (the <0.63 upper limit). If the upper limit survives both P-T changes, the stellar-like interpretation is considerably strengthened; if not, the claim should be downgraded to 'consistent with a range that includes stellar-like values' or an upper limit with an explicit model-dependence caveat. I agree with the reader's weakest assumption and recommend no change to the conditional verdict.","tokens_in":101,"tokens_out":7798,"duration_ms":192603,"concrete_test":"Re-run the retrieval from Section 5 on the same data with the P-T profile changed to the best-fit BT-Settl atmosphere (Teff~1000 K, log g~4.7) and, separately, with a simple cloud parameterization (e.g., a gray cloud deck) while keeping all other priors identical. Compare the 95% upper limit on C/O. If the upper limit moves above 0.63 or the posterior becomes unconstrained (e.g., 95% upper >1), the central stellar-like C/O claim is not robust and the paper should explicitly present the result as model-dependent rather than as a measurement.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (C/O=0.28+0.20-0.12, <0.63 at 95% CL) is produced by the Section 5 petitRADTRANS retrieval with the Sonora cloudless P-T profile fixed at Teff=1100 K, log g=4.5. However, Section 4.2's forward modeling of the same spectra with BT-Settl grids gives Teff=1003+134-75 K, log g=4.7+0.5-0.6 and explicitly reports that the Sonora cloudless model is a worse fit (log10 Bayes factor=0.7, 'substantial'). Thus the retrieval's assumed thermal structure is not the one favored by the data. C/O in K-band is inferred from the relative strengths of CO and H2O lines, which depend on the temperature-pressure profile and on cloud opacity; a different P-T or the presence of clouds can change the column where lines form and bias C/O. The 3.5 sigma H2O detection makes the constraint especially fragile. If the C/O upper limit is not robust to replacing the Sonora profile with the BT-Settl best-fit or an explicitly cloudy P-T, the 'stellar-like C/O' conclusion is a modeling artifact rather than a measurement.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This Letter reports Keck/KPIC high-resolution spectroscopy of the protoplanet PDS 70b, with cross-correlation detections of CO (3.8 sigma) and H2O (3.5 sigma), a combined detection significance of 4.2 sigma, and an upper limit on the projected rotational velocity of v sin i < 29 km/s. The authors use a petitRADTRANS atmospheric retrieval to derive C/O = 0.28+0.20-0.12 (<0.63 at 95% confidence) and [C/H] = -0.2+0.8-0.5 dex, which they compare with the host star's C/O of 0.44 +/- 0.19 and the outer disk's gas-phase C/O of greater than or approximately 1. They conclude that PDS 70b's C/O is stellar-like and not gas-like, and discuss two formation scenarios: solid-dominated accretion or late carbon enrichment of the disk gas.","tokens_in":18742,"tokens_out":3567,"duration_ms":34755,"significance":"If correct, this is the first abundance measurement of a protoplanet embedded in its natal disk, a unique and important result for planet formation theory. The detection is supported by a very strong model comparison between a star-only and a star-plus-planet model (log10 Bayes factor 344.9, delta chi-square 772), which addresses the concern that the signal could be residual starlight. The paper is also honest about its limitations, explicitly acknowledging the noisy spectra and the need for future verification. The central abundance claim, however, rests on a fixed, cloudless thermal profile that the authors' own forward modeling disfavors, so the reported C/O should be treated as model-dependent unless its robustness is demonstrated.","major_comments":[{"comment":"The retrieval fixes the cloudless Sonora pressure-temperature profile at Teff = 1100 K and log g = 4.5, yet the same data analyzed with BT-Settl model grids in Section 4.2 yield Teff = 1003+134-75 K and log g = 4.7+0.5-0.6 and explicitly find the cloudless Sonora model to be a worse fit (log10 Bayes factor = 0.7, 'substantial'). Because the C/O ratio is inferred from the relative strengths of CO and H2O lines, which form at pressure levels controlled by the thermal structure and cloud opacity, this choice could bias the derived C/O and its confidence interval. The manuscript should test the sensitivity of C/O and the 95% upper limit to (a) replacing the Sonora profile with the BT-Settl best-fit P-T profile, (b) including clouds, and (c) marginalizing over Teff and log g within the forward-model posterior.","section":"Section 5, Table 2"},{"comment":"The C/O and [C/H] constraints come from spectra with CO and H2O detections at only 3.8 sigma and 3.5 sigma, respectively. Given that the retrieval priors on C/O span 0.1-1.6, the reported median of 0.28 with an upper limit of 0.63 may be significantly prior-dependent. The authors should provide a quantitative comparison of the priors and posteriors for C/O and [C/H], for example by reporting the Hellinger distance or the information gain, so the reader can assess how much information the noisy KPIC spectra actually add beyond the assumed priors.","section":"Section 3, Section 5"}],"minor_comments":[{"comment":"There is an inconsistency between the text and the table for the best-fit effective temperature: Section 4.2 reports Teff = 1003+134-75 K, while Table 2 lists Teff = 1103+134-75 K. Please correct the typo.","section":"Section 4.2, Table 2"},{"comment":"In Section 7, the text writes 'CO ∼ 0.44' when referring to the host star's carbon-to-oxygen ratio; this should be 'C/O ∼ 0.44' for clarity.","section":"Section 7, Abstract"},{"comment":"The description of adding a 2.5 km/s systematic uncertainty to the radial velocity of PDS 70b is brief; please clarify whether this value was derived purely from Ruffio et al. (2023) or also from the observed RV scatter in this data, as it directly affects the claimed 2.5-sigma RV offset from the host star.","section":"Section 4.2"}],"recommendation":"major_revision","confidential_remarks":"The paper presents an exciting and potentially first-of-its-kind measurement for a protoplanet still embedded in its disk, and the detection validation via Bayes factor is convincing. However, the C/O abundance is the central scientific result, and the retrieval's fixed thermal profile is the opposite of what the same data favor in the forward-modeling section. I would like to see the authors either demonstrate the C/O robustness to alternative P-T/cloud assumptions or substantially soften the claim about stellar-like C/O. The two major comments above are intended to be addressable within a revision rather than requiring new observations."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First: this is the first molecular detection (CO and H2O) and first C/O measurement for a protoplanet still embedded in its disk. That is a genuine step forward, and the spin upper limit is a useful extra. The paper is honest about the limits: it reports 3.8 and 3.5 sigma detections, combined 4.2 sigma, and the C/O posterior is broad (0.28+0.20-0.12, <0.63 at 95%). The star+planet model is strongly favored over star-only (log10 BF=344.9), which gives me confidence the signal is real even at low SNR.\n\nThe soft spot is the retrieval setup. Section 4.2 forward-modeling with BT-Settl grids finds Teff=1003+134-75 K, log g=4.7+0.5-0.6, and explicitly shows Sonora cloudless is a worse fit (log10 BF=0.7, described as substantial). Yet the Section 5 petitRADTRANS retrieval fixes the Sonora cloudless P-T at Teff=1100 K, log g=4.5, v sin i=0. That is an internal inconsistency. C/O in the K-band comes from the relative CO and H2O line strengths, and those depend on where in the P-T profile the lines form. If the true thermal structure is the BT-Settl one, or if clouds matter, the C/O upper limit can shift. The paper does not test this. Given the central claim is \"stellar-like C/O,\" that omission matters.\n\nI don't think this is fatal. The detections stand on their own, and the C/O is presented with wide error bars and clear caveats. But the headline claim should be read as tentative. A referee should ask for a robustness check: redo the retrieval with the BT-Settl best-fit P-T and one cloudy model, and see if the C/O limit moves outside the quoted range. If it does, the \"stellar-like\" conclusion needs rephrasing.\n\nWho benefits: this is for the exoplanet atmosphere and planet formation communities. The first in-situ abundance of a forming planet is a milestone even if the precision is low. I'd send it to a serious referee. My own verdict would be \"publish with revisions\" conditional on the robustness test.\n\nIn short: worth engaging with, but the C/O number is a modeling-dependent estimate, not a settled measurement.","headline":"First molecular detections and C/O measurement for PDS 70b, but the C/O limit rests on a cloudless P-T profile the paper's own forward model disfavors.","tokens_in":19351,"tokens_out":2745,"would_cite":true,"duration_ms":25888,"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":"First abundance measurement of a planet still forming in its disk finds a stellar-like carbon-to-oxygen ratio.","keywords":["protoplanet","PDS 70b","carbon-to-oxygen ratio","high-resolution spectroscopy","atmospheric retrieval","planet formation","KPIC"],"falsifier":"A higher-signal-to-noise spectrum of PDS 70b analyzed with a retrieval that lets clouds, the pressure-temperature profile, and v sin i vary freely would either confirm or overturn the stellar-like C/O; alternatively, a detection of refractory elements like Fe or silicates in the planet's atmosphere would discriminate between the solid-dominated and late-enrichment scenarios.","tokens_in":18272,"feed_emoji":"🪐","tokens_out":3457,"duration_ms":33446,"temperature":0.7,"pith_summary":"This paper reports the first measurement of the atmospheric carbon-to-oxygen ratio of a protoplanet still embedded in its natal disk, using Keck/KPIC high-resolution spectroscopy. The authors detect carbon monoxide and water in PDS 70b and retrieve a C/O ratio of 0.28+0.20-0.12, consistent with the host star and lower than the carbon-rich outer disk gas. Because PDS 70b is the only confirmed planet still forming inside its birth disk, this measurement offers a direct test of how planets acquire their volatile elements. If the result holds, it suggests that the planet's bulk carbon and oxygen came largely from dust and ice solids rather than from disk gas, or that the disk gas became carbon-enriched only after the planet formed.","feed_headline":"A forming planet's atmosphere matches its star's chemistry","feed_subtitle":"Keck spectra find CO and water in PDS 70b, with a stellar-like carbon-to-oxygen ratio that challenges gas-dominated accretion models.","key_machinery":"Atmospheric retrieval with the petitRADTRANS package applied to the KPIC high-resolution spectrum, with the pressure-temperature profile fixed to a cloudless Sonora model at Teff = 1100 K and log g = 4.5 and with v sin i fixed to zero. The C/O and [C/H] values are constrained by the relative line depths of CO and H2O in spectral orders 31–33 of NIRSPEC, using a nested-sampling fit over radial velocity, quench pressure, and planet flux.","core_discovery":"The paper establishes that PDS 70b's atmosphere contains CO and H2O, detected via cross-correlation at combined significance of about 4.2σ, and derives a carbon-to-oxygen ratio of 0.28+0.20-0.12 (upper limit 0.63 at 95% confidence) and a metallicity [C/H] of -0.2+0.8-0.5 dex. These values are consistent with the host star PDS 70 A, whose C/O is about 0.44, and clearly lower than the gas-phase C/O of the outer disk inferred from ALMA to be ≳1. The authors also report a spin upper limit of <29 km/s for the planet. They interpret the stellar-like C/O as evidence that the planet's volatile budget is dominated by solids rather than gas, or that the disk underwent late carbon enrichment after the planet formed.","pith_inferences":["If this measurement holds, it implies that a planet's C/O ratio can stay stellar-like even when its surrounding disk is carbon-rich, so disk gas composition alone may not predict the composition of the forming planet.","The late carbon-enrichment interpretation predicts that planets forming early in a disk's lifetime should have stellar-like C/O, while those forming later, after carbon enrichment, would show superstellar C/O; a survey of protoplanets at different disk evolutionary stages would test this.","Measuring refractory species such as Fe or silicates in PDS 70b would break the degeneracy between the two scenarios without needing extreme signal-to-noise, because a dust-and-ice dominated accretion history should show enhanced refractory abundances.","A similar C/O measurement for the outer companion PDS 70c would test whether accretion chemistry varies with orbital radius inside the same disk."],"forward_implications":["If the stellar-like C/O is real, the bulk carbon and oxygen of PDS 70b must have been delivered mostly by dust and ice aggregates rather than by the carbon-rich gas of the outer disk.","Alternatively, the disk gas became carbon-enriched after PDS 70b finished accreting, as predicted by models of disk chemical evolution and observed in some old disks.","The spin non-detection (<29 km/s) supports the picture that the planet is still contracting and has not yet undergone the spin-up phase expected after accretion ends.","Future higher-SNR observations of PDS 70b and PDS 70c, and measurements of isotope ratios such as 12C/13C and 16O/18O, could distinguish the dust-and-ice scenario from the late carbon-enrichment scenario.","A star-only model for the observed spectra is highly disfavored, supporting that the detected CO and H2O lines truly come from the planet atmosphere."],"supporting_citations":[{"why":"Supplies the KPIC data-reduction and forward-modeling cross-correlation methodology used to detect the planet's molecular lines.","marker":"Wang et al. 2021c"},{"why":"Provides the joint star-planet forward-modeling and fitting procedures applied to the KPIC spectra.","marker":"Hsu et al. 2024a"},{"why":"Presents the petitRADTRANS retrieval framework with quench-pressure parameterization used to derive C/O and [C/H].","marker":"Mollière et al. 2020"},{"why":"Supplies the Sonora cloudless pressure-temperature profiles that are fixed in the retrieval at Teff = 1100 K and log g = 4.5.","marker":"Marley et al. 2021"},{"why":"Provides the BT-Settl CIFIST model grids used in the forward-model fits that fix effective temperature and surface gravity.","marker":"Baraffe et al. 2015"},{"why":"Gives the host star C/O measurement and the disk chemistry model that explains a stellar-like planet C/O through late carbon enrichment.","marker":"Cridland et al. 2023"},{"why":"Provides the ALMA-based measurement of the outer disk gas-phase C/O ≳ 1 that contrasts with the planet's stellar-like value.","marker":"Facchini et al. 2021"}],"fun_headline_variants":["PDS 70b's carbon-oxygen ratio matches its star, not its disk","Keck detects CO and water in young planet PDS 70b, C/O stellar-like","Forming exoplanet PDS 70b shows stellar-like C/O, hints at solid accretion","PDS 70b's atmosphere: CO and H2O, C/O similar to host star","Young planet's C/O ratio challenges gas-dominated formation models"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The retrieval assumes a cloudless, non-rotating atmosphere with a pressure-temperature profile fixed to a model at 1100 K; if the real atmosphere has clouds, a different temperature structure, or significant rotation, the inferred carbon-to-oxygen ratio would shift.","fun_headline_variants_meta":{"raw":{"variants":["PDS 70b's carbon-oxygen ratio matches its star, not its disk","Keck detects CO and water in young planet PDS 70b, C/O stellar-like","Forming exoplanet PDS 70b shows stellar-like C/O, hints at solid accretion","PDS 70b's atmosphere: CO and H2O, C/O similar to host star","Young planet's C/O ratio challenges gas-dominated formation models"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00026,"raw_usage":{"total_tokens":1682,"prompt_tokens":1131,"completion_tokens":551,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":747,"completion_tokens_details":{"reasoning_tokens":438}},"tokens_in":747,"tokens_out":551,"duration_ms":5242,"temperature":1.0,"reasoning_tokens":438,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:27:51.783838+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A higher-signal-to-noise spectrum of PDS 70b analyzed with a retrieval that lets clouds, the pressure-temperature profile, and v sin i vary freely would either confirm or overturn the stellar-like C/O; alternatively, a detection of refractory elements like Fe or silicates in the planet's atmosphere would discriminate between the solid-dominated and late-enrichment scenarios.","supporting_citations":[],"review_version":1}