{"id":"457b71f5-3387-4308-813d-5e3cb63ccf53","arxiv_id":"2501.00104","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"MIRI coronagraphic photometry of GJ 504 b detects ammonia absorption at 10.6 µm; the reported 12.5σ significance falls to 3σ when PSF normalization errors are included.","lead":"JWST's MIRI coronagraph produced the first mid-infrared images of the cold companion GJ 504 b, and the team reports ammonia absorption in its atmosphere. The detection matters because ammonia traces gravity and could help tell whether this object is a young planet or an old brown dwarf.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Headline NH3 significance relies on dropping σ_PSFnorm, but the paper's own 4.1% vs 12.6% filter-dependent PSF-normalization errors contradict the full-correlation assumption.","rationale":"The reader's weakest_assumption correctly identifies the PSF-normalization correlation and model-relative nature of the NH3 detection as the load-bearing premises. My concern sharpens the first premise: the paper's own measured σ_PSFnorm values are strongly filter-dependent (4.1% in F1065C, 12.6% in F1140C), which directly contradicts the claim that this error cancels in the F1065C/F1140C ratio. A common-mode error would have the same fractional value in both filters; the mismatch is quantitative evidence that the normalization is wavelength-dependent. The paper's conservative 3σ calculation is therefore not just a 'lower limit' but the appropriate significance unless the correlation is demonstrated, not asserted. This does not invalidate the valuable new MIRI photometry or the atmospheric parameter constraints, but it means the central claim of an unambiguous 12.5σ ammonia detection is overstated. The reader's CONDITIONAL verdict remains the right call: the paper should be accepted only after either removing the 12.5σ headline in favor of the conservative significance or providing a measured, filter-by-filter justification for the error correlation. My agreement is 'partial' because I add a concrete internal contradiction (the σ_PSFnorm values) that the reader mentioned only generally, and I also note the model-dependence as a secondary but real limitation that the paper acknowledges but does not quantitatively address.","tokens_in":19593,"tokens_out":8230,"duration_ms":71203,"concrete_test":"Compute the F1065C/F1140C flux ratio for each of the three PSF-normalization methods (target-acquisition contrast, WebbPSF simulation, JWST pipeline calibration) shown in Appendix C. Report the dispersion of this ratio across methods. If that dispersion is of order sqrt(0.041² + 0.126²) ≈ 13%, then Eq. (3) must include the PSF-normalization term and the NH3 S/N is ~3σ, requiring the headline claim to be downgraded; if the dispersion is <2%, the 12.5σ claim may be defensible.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.2 (Eqs. 2-3) defines the NH3 S/N from the F1065C deficit relative to the no-NH3 model, with σtot = sqrt(σ_F1065C² + σ_F1140C²). Using the full uncertainties from Table 2 gives S/N = 3σ. The paper then asserts σ_PSFnorm is common-mode and zeroes it, obtaining 12.5σ. But Section 3.1 reports σ_PSFnorm = 4.1% (F1065C), 12.6% (F1140C), and 10.7% (F1550C). A truly common-mode multiplicative error would produce identical fractional errors in every filter; the observed threefold mismatch shows the normalization uncertainty is wavelength-dependent and does not cancel in the F1065C/F1140C ratio. This error therefore propagates directly into the inferred NH3 absorption, and the detection significance reverts to roughly 3σ, below the usual bar for an 'unambiguous' detection. Additionally, the null hypothesis is the no-NH3 Exo-REM model, and the paper does not show that no-NH3 models within the posterior, or with other cloud or opacity prescriptions, cannot reproduce the measured two-filter color. The NH3 attribution is thus model-relative even if the photometry were perfectly calibrated.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents JWST/MIRI coronagraphic observations of the planetary-mass companion GJ 504 b in the F1065C, F1140C, and F1550C filters. It builds a large reference-star library and applies a quadrant-based PCA subtraction to detect the companion and measure its mid-infrared photometry. Using Exo-REM atmospheric grids, the authors fit the companion's SED, derive updated parameters (Teff = 512 ± 10 K, R = 1.08 RJup), and claim a detection of NH3 at 12.5σ, with a volume mixing ratio of 10^-5.3±0.07. The paper also reports the conservative estimate of 3σ when the full photometric uncertainties are used, and discusses consequences for the age, mass, and formation scenario of GJ 504 b.","tokens_in":19876,"tokens_out":5369,"duration_ms":57010,"significance":"If the central claim holds, this would be the first robust NH3 detection in a directly imaged planetary-mass companion and the first mid-infrared photometry of GJ 504 b, providing a valuable anchor for cold-atmosphere chemistry and for NH3 as a gravity diagnostic. The paper has clear strengths: the companion is detected consistently across many stellar-subtraction methods, the reference-library approach is carefully described, and the authors are transparent about the two significance estimates. However, the headline 'unambiguous' claim rests on a 12.5σ value that is obtained by dropping the PSF-normalization uncertainty, and the attribution to NH3 is tested against only a single atmospheric model. These issues are load-bearing for the title-level claim, even though the underlying data and analysis are of lasting value.","major_comments":[{"comment":"The headline significance of 12.5σ is obtained by setting σ_PSFnorm to zero, justified only by the assertion that the NH3 measurement is relative. This assertion is not supported by the paper's own error budget: Section 3.1 reports σ_PSFnorm = 4.1%, 12.6%, and 10.7% in F1065C, F1140C, and F1550C, respectively. If the PSF-normalization errors were fully correlated across filters, their fractional contributions would be equal; the observed factor-of-three spread shows that the normalization uncertainty has a wavelength-dependent component that does not automatically cancel in the F1065C/F1140C ratio. The paper itself states that using the full uncertainties yields S/N = 3σ, so the abstract and title should either quote 3σ as the primary detection significance or provide a proper covariance analysis demonstrating which fraction of σ_PSFnorm cancels. As written, the 'unambiguous' claim is not supported.","section":"Section 3.2, Eqs. (2)-(3)"},{"comment":"The NH3 test is performed against a null hypothesis consisting of the no-NH3 Exo-REM model, and no evidence is shown that the two-filter F1065C/F1140C color cannot be reproduced by variations in cloud properties, metallicity, C/O, or other opacity sources, either within the posterior distribution or with an independent atmospheric model. The discussion in Section 4.2 acknowledges that clouds may affect the derived NH3 abundance but does not quantify the effect. For a claim of an 'unambiguous' first detection, the paper should demonstrate, for example, that a retrieval with free cloud/chemistry parameters or a second atmospheric model cannot fit the measured color without NH3. Without such a test, the attribution of the flux deficit to NH3 remains model-relative.","section":"Section 3.2, Eq. (2)"},{"comment":"The significance calculation and the quoted NH3 volume-mixing-ratio range are based on a second grid built from the best-fit atmospheric parameters, without propagating the posterior uncertainties in Teff, log g, metallicity, C/O, and radius into the predicted no-NH3 continuum flux. Because those parameters are uncertain (Table 4), the model-predicted F1065C flux used in Eq. (2) has an uncertainty that is not included in σ_tot. The resulting S/N is therefore optimistic even before the PSF-normalization issue, and the 1σ VMR range should be interpreted as conditional on the best-fit continuum model.","section":"Section 3.2 and Fig. F.1"}],"minor_comments":[{"comment":"The phrase 'proving the first mid-infrared (MIR) images' appears to be a typo; it should read 'providing the first mid-infrared images.'","section":"Section 2.1"},{"comment":"The legend labels are difficult to read because of repeated filter names and inconsistent spacing in program names (e.g., 'GTO1413' vs. 'GTO 1413'); please use a single, consistent notation and a concise legend.","section":"Figures 3 and B.4"},{"comment":"Equations referenced in the text (e.g., 'Eq. 1') are not all explicitly numbered in the provided manuscript, and the symbols σ_PSFnorm and σ_stellarsub are not typeset consistently; please ensure all equations and symbols are defined once and used consistently.","section":"Section 3.1"},{"comment":"The comparison of photometric normalization methods would be clearer with a table listing the individual flux values from each method, since the text describes variations of 0.8, 1.3, and 6.9% for stellar subtraction but the corresponding per-method fluxes are not given in the main text.","section":"Appendix C"}],"recommendation":"major_revision","confidential_remarks":"The paper has the makings of an important contribution, and the companion detection and first MIR photometry are solid. My main concern is that the headline claim is currently stronger than the statistical evidence: the 12.5σ NH3 detection depends on an unvalidated correlation assumption, and the conservative estimate is 3σ. I would be comfortable with acceptance after the authors either provide a rigorous covariance treatment that supports the higher significance or revise the title, abstract, and conclusions to present the NH3 detection at the conservative significance and explicitly frame the model-dependence of the attribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is the first MIR photometry of GJ 504 b, the companion detection is robust, and the reference-library subtraction work is useful — but the 12.5σ NH3 claim in the title and abstract does not survive contact with their own error budget. The defensible number is the 3σ they relegate to a 'conservative lower limit.'\n\nWhat's genuinely good: the companion is detected across every stellar-subtraction method they tried (PCA, LOCI-style linear optimization, different reference subsets), the PSF normalization — the dominant systematic — is quantified with three independent methods, and they print the 3σ number in the text rather than hiding it. The updated parameters (Teff = 512 K, R = 1.08 RJup) and the factor-of-3 shrinkage in the radius uncertainty when the MIR points are added look reasonable, and the mass/age discussion is properly hedged. The citation pattern is clean; the GJ 504 b literature and the Whiteford 51 Eri b NH3 retrieval are both cited.\n\nThe soft spot is load-bearing. The 12.5σ comes from zeroing σ_PSFnorm, justified as common-mode between filters. Their own Table 2 reports fractional normalization dispersions of 4.1% in F1065C and 12.6% in F1140C. A common-mode multiplicative error would produce the same fractional scatter in both filters; the threefold mismatch means most of that error is filter-specific and does not cancel in the ratio that carries the NH3 signal. The stress-test note is right, and the abstract-level 'unambiguous detection' claim should be scaled back to the conservative significance.\n\nThe second issue is model relativity. The null hypothesis is no-NH3 Exo-REM, and the paper does not show whether no-NH3 models with other clouds, C/O, or metallicity can reproduce the F1065C/F1140C color. Around 10 µm that matters. The NH3 grid is also built on best-fit parameters that were themselves fit including the MIR points, so the NH3 significance does not carry atmospheric-parameter covariance — a mild circularity, not a fatal one.\n\nBottom line: send it to a serious referee. The photometry is new, the companion detection is solid, and the field needs this benchmark MIR datapoint. But the NH3 detection should be presented at the conservative significance, and the correlation assumption either defended with evidence or dropped. The observed color anomaly is plausibly NH3; at 3σ with a model-relative interpretation, 'unambiguous' is the wrong word.","headline":"The first MIR photometry of GJ 504 b is genuinely new and the companion detection is robust, but the 12.5σ NH3 claim rests on a correlation assumption that their own error budget contradicts; the defensible significance is 3σ.","tokens_in":20612,"tokens_out":6181,"would_cite":true,"duration_ms":69808,"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":"This paper reports the first unambiguous detection of ammonia in the atmosphere of the planetary-mass companion GJ 504 b, at 12.5 sigma.","keywords":["NH3","ammonia","GJ 504 b","directly imaged exoplanets","JWST MIRI coronagraphs","mid-infrared photometry","Exo-REM atmospheric models","planetary-mass companions"],"falsifier":"A medium-resolution MIRI/MRS spectrum of GJ 504 b resolving the 10.5 micron ammonia band, or any independent re-reduction that lets the F1065C and F1140C PSF-normalization terms float freely, would settle the claim: a band absent or much weaker than the photometric deficit implies, or a significance dropping to about 3 sigma when the normalization term is included, would overturn the 12.5 sigma ammonia detection.","tokens_in":19362,"feed_emoji":"🪐","tokens_out":8636,"duration_ms":76855,"temperature":0.7,"pith_summary":"This paper claims the first unambiguous detection of ammonia (NH$_3$) in the atmosphere of a directly imaged planetary-mass companion, the roughly 500 K object GJ 504 b. The detection rests on a flux deficit measured in the JWST/MIRI F1065C coronagraphic filter, centered on NH$_3$ absorption, relative to the F1140C continuum filter, and is reported at 12.5 $\\sigma$ with a conservative lower limit of 3 $\\sigma$. Combining the new mid-infrared photometry with archival near-infrared data in the Exo-REM atmospheric model, the authors derive $T_{\\mathrm{eff}} = 512 \\pm 10$ K, a radius of $1.08\\,R_{\\rm Jup}$, and an NH$_3$ volume mixing ratio of $10^{-5.3 \\pm 0.07}$. If correct, the result opens a direct observational path to using ammonia as a gravity-sensitive tracer in the coldest directly imaged companions, helping to decide whether such objects are young planets or old brown dwarfs.","feed_headline":"JWST detects ammonia in a planet-mass companion","feed_subtitle":"The 12.5-sigma sighting in GJ 504 b's atmosphere may pin down its gravity and true nature.","key_machinery":"The measurement is carried by the two-filter flux ratio between the F1065C filter, centered at 10.575 microns inside the 10.5 micron ammonia band, and the F1140C filter, a continuum window at 11.30 microns, combined with a large library of previously observed reference stars whose diffraction patterns are subtracted quadrant-by-quadrant with principal component analysis. Photometry is extracted by fitting a coronagraphic point-spread-function model to each reduced image, correcting for mask transmission (0.84, 0.93, and 0.77 in the three filters), and the ammonia signal is the F1065C deficit relative to an Exo-REM model with negligible NH$_3$. The significance quoted as 12.5 sigma treats the two filters' PSF-normalization uncertainties as fully correlated, and including them independently lowers the detection to 3 sigma, which the paper calls a conservative lower limit.","core_discovery":"Using the MIRI four-quadrant phase-mask coronagraphs on JWST, the paper obtains the first mid-infrared photometry of GJ 504 b in three filters, F1065C, F1140C, and F1550C, and reports the detection of ammonia at 12.5 $\\sigma$. The signal is the lower flux observed at F1065C, where NH$_3$ absorbs, compared with the continuum at F1140C; the authors express it as $S/N = (F_{\\rm no\\,NH_3} - F_{\\rm obs})/\\sigma_{\\rm tot}$ relative to the Exo-REM model with negligible NH$_3$ abundance. The best-fit Exo-REM atmosphere has $T_{\\mathrm{eff}} = 512 \\pm 10$ K, radius $R = 1.08^{+0.04}_{-0.03}\\,R_{\\rm Jup}$, $\\log g = 3.45^{+0.35}_{-0.25}$, metallicity $0.54^{+0.09}_{-0.11}$, and $C/O = 0.70 \\pm 0.06$; the measured NH$_3$ mixing ratio of $10^{-5.3 \\pm 0.07}$ is at least an order of magnitude below Jupiter's. The paper also notes this is the first unambiguous NH$_3$ detection in a directly imaged planetary-mass companion, with the only previous direct-imaging detection, for 51 Eri b, reported at 2.7 $\\sigma$. Because the NH$_3$/N$_2$ ratio is insensitive to atmospheric mixing, the authors argue the NH$_3$ abundance favors a low surface gravity, and thus a planetary-mass rather than brown-dwarf interpretation, while cautioning that the mass remains degenerate.","pith_inferences":["If the 12.5 sigma detection survives re-analysis, the same two-filter F1065C/F1140C strategy could be applied to other cold companions as a relatively cheap NH$_3$-based gravity screen before expensive spectroscopy.","A natural test is the already-planned MIRI/MRS spectrum: a resolved 10.5 micron NH$_3$ band at the reported strength would confirm the photometric claim, while a null or much weaker band would refute it.","The radius of $1.08\\,R_{\\rm Jup}$ sits slightly below evolutionary-model predictions, and if this persists it may point to gaps in cooling-track models or cloud physics rather than to the ammonia detection itself.","Since the NH$_3$ abundance is compared with Jupiter's, similar two-band photometry across a range of temperatures could sharpen empirical constraints on nitrogen chemistry across the T/Y boundary."],"forward_implications":["The MIRI coronagraphic filters, designed around the 10.5 micron NH$_3$ band, can deliver unambiguous ammonia detections in the coldest directly imaged companions.","Adding mid-infrared photometry reduces the radius uncertainty of GJ 504 b by at least a factor of three and improves the luminosity estimate by a similar factor.","The measured NH$_3$ abundance is consistent with an object of $\\log g < 4$, favoring the low-mass, planetary interpretation over a brown-dwarf one, while the authors stop short of confirming the mass.","NH$_3$ emerges as a practical gravity diagnostic for cold companions, because the NH$_3$/N$_2$ ratio is insensitive to mixing.","The first mid-infrared photometry of GJ 504 b anchors forthcoming JWST spectroscopy of the same object with MIRI/MRS and NIRSpec/IFU."],"supporting_citations":[{"why":"Discovery of GJ 504 b and first near-infrared photometry, plus the young-age estimate for the host star.","marker":"Kuzuhara et al. 2013"},{"why":"SPHERE near-infrared photometry and the age/mass degeneracy for the system.","marker":"Bonnefoy et al. 2018"},{"why":"LBTI near-infrared photometry and the supersolar metallicity estimate for GJ 504 b.","marker":"Skemer et al. 2016"},{"why":"Provides the method for quantifying the NH3 detection level by comparing filter photometry against a model with negligible NH3.","marker":"Danielski et al. 2018"},{"why":"Defines the Exo-REM atmospheric model used for the grid fitting.","marker":"Baudino et al. 2015"},{"why":"Updates Exo-REM for the L-T transition, the regime occupied by the T-type companion.","marker":"Charnay et al. 2018"},{"why":"MIRI coronagraph commissioning that identified the background glowstick effect and the reduction steps reused here.","marker":"Boccaletti et al. 2022"},{"why":"Data reduction and photometry extraction procedure applied to MIRI coronagraphic images.","marker":"Boccaletti et al. 2024"},{"why":"Previous NH3 inference in 51 Eri b at 2.7 sigma, which defines the context for a first unambiguous detection.","marker":"Whiteford et al. 2023"},{"why":"Theoretical basis for NH3/N2 insensitivity to mixing, making NH3 a surface-gravity probe.","marker":"Zahnle & Marley 2014"}],"fun_headline_variants":["First unambiguous ammonia detection in a planetary-mass companion","JWST detects ammonia in a planet-mass companion for the first time","Ammonia spotted on a directly imaged planetary-mass companion","JWST confirms ammonia in the coldest directly imaged companion","Ammonia discovered in a young giant planet's atmosphere"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim stands or falls on whether the F1065C flux deficit comes from NH3 opacity rather than a cloud, another molecule, or a wavelength-dependent calibration error, plus the assumption that the PSF-normalization errors of the two filters are fully correlated, which is what lifts the conservative 3 sigma to the headline 12.5 sigma.","fun_headline_variants_meta":{"raw":{"variants":["First unambiguous ammonia detection in a planetary-mass companion","JWST detects ammonia in a planet-mass companion for the first time","Ammonia spotted on a directly imaged planetary-mass companion","JWST confirms ammonia in the coldest directly imaged companion","Ammonia discovered in a young giant planet's atmosphere"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000523,"raw_usage":{"total_tokens":2693,"prompt_tokens":1276,"completion_tokens":1417,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":892,"completion_tokens_details":{"reasoning_tokens":1335}},"tokens_in":892,"tokens_out":1417,"duration_ms":13206,"temperature":1.0,"reasoning_tokens":1335,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T22:59:22.739660+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A medium-resolution MIRI/MRS spectrum of GJ 504 b resolving the 10.5 micron ammonia band, or any independent re-reduction that lets the F1065C and F1140C PSF-normalization terms float freely, would settle the claim: a band absent or much weaker than the photometric deficit implies, or a significance dropping to about 3 sigma when the normalization term is included, would overturn the 12.5 sigma ammonia detection.","supporting_citations":[{"cited_title":"Water depletion and 15NH3 in the atmosphere of the coldest brown dwarf observed with JWST/MIRI","cited_arxiv_id":"2410.10933","evidence_quote":"Discovery of GJ 504 b and first near-infrared photometry, plus the young-age estimate for the host star."},{"cited_title":"2018, A&A, 618, A63","cited_arxiv_id":null,"evidence_quote":"SPHERE near-infrared photometry and the age/mass degeneracy for the system."},{"cited_title":"J., Morley, C","cited_arxiv_id":null,"evidence_quote":"LBTI near-infrared photometry and the supersolar metallicity estimate for GJ 504 b."},{"cited_title":"2018, The Astronomical Jour- nal, 156, 276","cited_arxiv_id":null,"evidence_quote":"Provides the method for quantifying the NH3 detection level by comparing filter photometry against a model with negligible NH3."},{"cited_title":"2018, ApJ, 854, 172","cited_arxiv_id":null,"evidence_quote":"Updates Exo-REM for the L-T transition, the regime occupied by the T-type companion."},{"cited_title":"2022, A&A, 667, A165","cited_arxiv_id":null,"evidence_quote":"MIRI coronagraph commissioning that identified the background glowstick effect and the reduction steps reused here."},{"cited_title":"2024, A&A, 686, A33","cited_arxiv_id":null,"evidence_quote":"Data reduction and photometry extraction procedure applied to MIRI coronagraphic images."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Theoretical basis for NH3/N2 insensitivity to mixing, making NH3 a surface-gravity probe."}],"review_version":1}