{"id":"0af3a2c0-e542-4218-b42c-2091e1d7350f","arxiv_id":"2506.08077","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"JWST aperture-masking images of Circinus show a 2x5 pc dust component along the torus funnel, a <1% outflow arc, and NIR holes where the equatorial disk blocks light, supporting a disk-fed AGN.","lead":"Using a new interferometric mode on the JWST, astronomers mapped the hot dust within the central 10 parsecs of the Circinus galaxy and found that most of the near-infrared emission comes from an elongated structure around the black hole, with only a small arc of dust caught in an outflow. The result supports the picture in which a disk of dust and gas feeds the supermassive black hole, and shows that the technique can map the nuclei of other nearby active galaxies.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Main text's '<10% line contribution' claim is inconsistent with the Methods' 71% continuum fraction in F430M, so the AMI photometry supporting the disk-vs-wind model choice is not secure.","rationale":"I agree with the reader's weakest-assumption analysis: the <10% line-contamination claim conflicts with the Methods' own 71% continuum fraction in F430M, and this matters because the new AMI points are what tip the balance (or nearly tip it) between the disk-like clumpy model and the wind model. The concern is load-bearing, not because line contamination is certain, but because the paper's central, journal-level claim is presented as a physical finding while resting on photometry whose spectral purity is asserted but not derived. The check is straightforward: synthetic photometry on an actual AGN spectrum (preferably Circinus itself) will settle whether F430M is 90%+ continuum as claimed or only ~70% as the Methods templates suggest. It is also worth noting that Table 2 without AMI already prefers clumpy over wind (χ² 1.24 vs 2.44), so the AMI data are not the only support; still, the new data are the paper's headline contribution and they are currently adding only marginal discrimination. The morphology (extended component, North arc, holes) is likely real given the bootstrapping and independent reconstructions, so I am not moving to REJECT. A conditional acceptance requiring the corrected photometry and a more explicit statement that the mass distribution is model-inferred is the right call.","tokens_in":22609,"tokens_out":9586,"duration_ms":120945,"concrete_test":"Recompute the synthetic photometry in F380M, F430M, and F480M using an observed Circinus NIR spectrum (or the type-2 AGN templates) with the spectral features included and with a feature-free continuum, obtaining the true continuum contribution per band. Then rerun the 2D Gaussian photometry and the Table 2 torus-model fits on the corrected 3.8, 4.3, and 4.8 µm fluxes. If the F430M flux changes by more than ~10%, or if the χ² gap between the clumpy and CAT3D-WIND models (currently 3.64 vs 3.69) narrows or reverses, the abstract's disk-mass conclusion needs to be qualified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that the dust mass reservoir is a disk feeding the SMBH—depends on the 1–1000 µm SED model selection, where the clumpy torus is preferred over the CAT3D-WIND model. Including the AMI/JWST points, the preference is marginal (clumpy χ²ALL=3.64 vs CAT3D-WIND=3.69 for the PSF aperture; Table 2), so the AMI photometry can decide the outcome. That photometry is assumed to be continuum-dominated: the main text states 'emission lines have small contributions, <10%, within the AMI filters.' However, Methods 'Emission line contribution' gives continuum fractions of 94%, 71%, and 84% in F380M, F430M, and F480M for type-2 AGN templates, implying up to ~29% non-continuum flux in F430M from CO2 ice and 12CO features. The 71% is called a lower limit because the templates (NGC 3256) have stronger lines than typical AGN, but the paper does not show how <10% follows from the quoted values. If F430M is not continuum-dominated, the derived dust temperatures (450 K, 730 K, 420 K), the aperture photometry, and the SED fit that prefers the clumpy torus are all biased, directly weakening the 'most dust mass in disk' conclusion.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents JWST/NIRISS aperture-masking interferometric (AMI) images of the Circinus galaxy at 3.8, 4.3, and 4.8 μm. Image reconstruction with SQUEEZE, supported by bootstrapping and synthetic-observable checks, reveals an extended 2×5 pc component at PA ≈ −64°, a 'North arc' contributing <1% of the flux and cospatial with outflow tracers, and 'Holes' coincident with the 700 μm disk. From a 2D Gaussian decomposition, 87% of the 3.8–4.8 μm emission is assigned to the central extended component, with dust temperatures of about 450 K, 730 K, and 420 K for the central component, North arc, and extended NLR/jet component, respectively. Fitting the 1–1000 μm SED with four torus models, the authors report a statistical preference for the clumpy torus over the CAT3D-WIND model, and use HyperCAT synthetic images to argue that the observed morphology is reproduced by a clumpy torus whose dust mass lies in an equatorial disk feeding the SMBH.","tokens_in":22886,"tokens_out":8267,"duration_ms":100378,"significance":"The observational material is novel and the image reconstruction appears careful: the use of two epochs with ~83° uv rotation, bootstrapping to assess feature significance, and comparison of synthetic or reconstructed visibilities with the data are all appropriate and well documented. The multi-filter consistency of the North arc at >16σ and its spatial correlation with [CI], H36α, and CO(6-5) outflows are valuable and likely robust. If the disk-vs-wind conclusion is correct, this is an important step in resolving the long-standing question of the origin of the 3–5 μm excess in AGN, and the demonstration of NIRISS AMI on an extragalactic source is a technical milestone. The main limitations are in the photometric calibration of line contamination and the marginal statistical preference between the clumpy torus and the wind model, both of which bear directly on the central claim.","major_comments":[{"comment":"The main text states that 'emission lines have small contributions, <10%, within the AMI filters,' but the Methods section reports continuum fractions of 94%, 71%, and 84% in F380M, F430M, and F480M for type 2 AGN templates, implying up to ~29% non-continuum flux in F430M from the 12CO band and CO2 ice features. The 71% value is described as a lower limit because the NGC 3256 template has stronger features than typical AGN, but the derivation of <10% from these numbers is not shown. Because the F430M photometry (Table 1) is an input to the blackbody temperature estimates, the aperture photometry, and the SED fits in Table 2, unaccounted line/ice contamination could bias the derived dust temperatures and the model comparison that prefers the clumpy torus. Please provide a direct measurement or an explicit calculation of the line contribution in the Circinus AMI filters, or treat F430M as non-continuum and propagate the resulting corrections through the SED analysis.","section":"Methods: 'Emission line contribution'; main text 'NEW JWST INTERFEROMETRIC OBSERVATIONS'"},{"comment":"The reported discrimination between the clumpy torus and CAT3D-WIND is marginal: for the PSF aperture, χ2_ALL=3.64 for clumpy vs 3.69 for CAT3D-WIND, a difference of 0.05 on what appears to be an absolute χ2 (not explicitly stated as reduced or normalized by the number of data points). No confidence interval, Δχ2 significance, or model-comparison statistic (e.g., AIC/BIC) is given. Given that the AMI photometry may carry systematic errors from line/ice contamination, this difference cannot support the strong claim that most of the dust mass is located in an equatorial disk. The authors should quantify the significance of the model preference and show that it is robust to line-contamination corrections, aperture choices, and the assumptions in the SED construction.","section":"Table 2 and 'Origin of the central continuum emission'"},{"comment":"The synthetic HyperCAT images shown in Fig. 3 are generated from the best-fit clumpy model that was itself fitted to the same 1–1000 μm SED that includes the AMI/JWST photometry, so the agreement in Fig. 3b–d is not an independent validation of the model. In addition, the tilt angle of 50° is chosen to be 'cospatial with the orientation along the lack of emission' in the AMI images, which means the 'Holes' are matched partly by construction. Please provide a quantitative comparison between the synthetic and observed surface brightness distributions (e.g., image-plane residuals or χ2), or fit the tilt angle as a free parameter and report its posterior distribution.","section":"Fig. 3 and Methods: 'Torus models'"},{"comment":"The model preference is aperture-dependent: for the 8 pc aperture, the smooth torus is preferred (χ2_ALL=3.02 versus 6.54 for clumpy), whereas the clumpy model is preferred at the smaller apertures that define the central 2×5 pc component. The paper acknowledges this and attributes it to diffuse extended emission at >4 pc, but the central claim about the dust mass distribution relies specifically on the small-aperture fit. Please demonstrate that the small-aperture preference is not an artifact of the chosen aperture or the exclusion of the North arc and NLR emission, and state explicitly how the large-aperture smooth-torus preference affects the 'disk feeding' conclusion.","section":"Appendix Table 2 and Fig. 9 (aperture dependence)"}],"minor_comments":[{"comment":"The text defines F^T_obj(u=0,v=0,λ) as the total flux of the zero-baseline in units of counts and then gives '(i.e., ADU: analog diginal unit)'; correct the typo 'diginal' and ensure the units in Eq. A1 are consistent (Jy, ADU, and the normalization of the reconstructed image).","section":"Methods: 'Flux calibration' and Eq. A1"},{"comment":"The beam size for H36α is given as '29×24 mas ◦', which appears to be missing a position-angle value or unit; check this and similar beam-size entries for consistency.","section":"Methods: 'Archival observations'"},{"comment":"The sentence 'Both observations ensure a ~90° rotation of the uv-plane' should be more precise: the two epochs provide a ~83° rotation, as stated in the Methods, with the 90° being the requested target.","section":"Main text: 'NEW JWST INTERFEROMETRIC OBSERVATIONS'"},{"comment":"Column (e) 'Contribution ext. emission' gives 13%, 13%, and 11% for F380M, F430M, and F480M, while the text quotes 12+4/−6% for the >5 pc emission; clarify how the Gaussian-removal residual and the separately measured 'North arc' combine to give these percentages.","section":"Table 1 and 'Origin of the central continuum emission'"},{"comment":"The χ2 values are presented without the number of data points or degrees of freedom; please state whether these are reduced χ2 values and add the number of photometric points used in each fit.","section":"Table 2"},{"comment":"The WCS registration assumes the peak pixel of the reconstructed image is the AGN position and then aligns to the 1200 μm ALMA peak; this should be stated as an assumption with an estimated uncertainty on the registration, especially since the AMI images are resolved and the peak may be offset from the true nucleus by up to a fraction of the beam.","section":"Methods: 'WCS correction'"}],"recommendation":"major_revision","confidential_remarks":"The paper is a strong technical advance, but the headline conclusion currently rests on a marginal model-selection result and an unquantified line-contamination correction in F430M. These issues are fixable in revision, so I recommend major revision rather than rejection. The fit between the manuscript and a broad journal is acceptable if the central claim is made robust; otherwise the paper would still be a solid specialized-journal contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First: the AMI/JWST 3.8–4.8 µm images are genuinely new, and they show a 2×5 pc extended component at PA ~ −70°, a North arc, and NIR holes aligned with the 700 µm disk. That fills a real observational gap between MATISSE/VLTI and single-dish maps. Second: the claim that most of the dust mass lies in an equatorial disk is a model inference, not a direct detection, and the SED fit that drives it is less decisive than the text suggests.\n\nThe imaging analysis is solid: bootstrapping, synthetic observable checks, and consistent morphology across three filters. The North arc is a nice result—likely real, cospatial with the molecular and ionized outflows. The paper also makes good use of archival ALMA/VLT data to place the components in context.\n\nThe soft spot is the line-contamination issue. The main text says emission lines contribute <10% in the AMI filters, but the Methods section reports continuum fractions of 94%, 71%, and 84% for F380M, F430M, and F480M for type 2 templates. The 71% in F430M implies up to ~29% non-continuum flux, and the paper never shows how the <10% figure follows. The authors call the continuum fractions lower limits because the NGC 3256 template has stronger lines, so the true contamination may be smaller—but the gap is real and it matters. If F430M photometry carries a significant line contribution, the derived temperatures, aperture photometry, and SED-based model preferences in that band are all biased.\n\nThis is not academic, because the model discrimination is already marginal. In Table 2, with the AMI/JWST PSF aperture, the clumpy torus gives χ²ALL = 3.64 and CAT3D-WIND gives 3.69—statistically indistinguishable. The morphology does real work here: the holes aligned with the 700 µm disk and the extended component perpendicular to it support a funnel geometry. But 'most dust mass is in a disk' is a result of fitting a clumpy torus to the SED and then using that same model to generate synthetic images that match the morphology. That is a consistency check, not an independent validation.\n\nWho is this for? AGN observers, anyone following the torus-versus-wind debate, and anyone interested in pushing JWST AMI on extragalactic nuclei. The paper deserves a serious referee. The authors need to reconcile the line-contamination numbers, clarify the model-selection language, and ideally release the reconstructed image cubes.\n\nRecommendation: send it to peer review. The imaging result is important and the weaknesses are fixable.","headline":"Genuinely new AMI/JWST images of Circinus reveal a 2x5 pc NIR structure and North arc, but an internal gap on F430M line contamination and a marginal model discrimination keep the 'disk feeding the AGN' conclusion from being fully secure.","tokens_in":23606,"tokens_out":2700,"would_cite":true,"duration_ms":34836,"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":"Using JWST's aperture-masking interferometry, this paper shows that 87% of the 3–5 micron emission from the Circinus galaxy's hidden black hole comes from a 2×5 parsec dusty structure in the funnel of an equatorial disk, while the dusty…","keywords":["aperture masking interferometry","AGN torus","Circinus galaxy","dusty torus","Seyfert 2 galaxy","near-infrared excess","supermassive black hole accretion","AGN outflows"],"falsifier":"Measure the fraction of non-continuum emission inside each AMI filter using spatially resolved spectroscopy of the inner 10 pc of Circinus, for example with JWST NIRSpec IFU or MIRI MRS spectra, by comparing narrow-band photometry at 3.8, 4.3, and 4.8 microns on the observed spectrum with a continuum-only baseline; if the F430M non-continuum fraction is close to the template-based value near 29% rather than the claimed <10%, the derived 450 K temperature and the 87% disk-dominated flux split must be re-evaluated.","tokens_in":22333,"feed_emoji":"🌌","tokens_out":13251,"duration_ms":137719,"temperature":0.7,"pith_summary":"The paper uses JWST's aperture-masking interferometry to image the inner 10 parsecs of the Circinus galaxy—the nearest hidden (type 2) active galactic nucleus—at 3.8, 4.3, and 4.8 microns, at roughly twice the resolution of ordinary JWST imaging. It finds that 87% of that hot-dust emission comes from a 2×5 parsec structure at a position angle of about −70 degrees, less than 1% comes from a 'north arc' coinciding with molecular and ionized outflows, and the remaining 12% comes from dust beyond 5 parsecs along the radio-jet direction. The apparent 'holes' in the near-infrared image coincide with the optically thick equatorial disk seen at 700 microns, and model fits to the full 1–1000 micron spectral energy distribution prefer a clumpy torus over wind or smooth geometries. The paper concludes that the long-puzzling near-infrared excess of this AGN is produced by dust in the funnel of a disk-like torus, and that the bulk of the dusty mass feeding the black hole sits in that equatorial disk rather than in a wind.","feed_headline":"87% of hot dust near Circinus's black hole lies in a feeding disk","feed_subtitle":"JWST interferometry resolves the nearest hidden supermassive black hole: a dusty disk, not a wind, dominates.","key_machinery":"Aperture-masking interferometry on JWST's NIRISS instrument is the central mechanism: a seven-hole non-redundant mask across the 6.5 m primary turns the telescope into an interferometer, and from measured squared visibilities and closure phases the images are reconstructed at about 0.1 arcsecond resolution, twice that of direct imaging, while suppressing the telescope's blur pattern and large-scale starlight. This mechanism brings out the faint structures—the 2×5 pc funnel emission, the north arc, and the near-infrared holes—that are then compared with archival sub-millimeter continuum and gas-tracer maps and with radiative-transfer torus models to separate disk from wind.","core_discovery":"On its own terms, the paper establishes that the 3–5 micron excess of Circinus is dominated not by an outflow but by directly heated dust in the funnel of an edge-on, clumpy torus. The AMI images at 3.8, 4.3, and 4.8 microns show an extended 2×5 pc component at a position angle of about −70°, contributing 87+5−7% of the total flux; a 'north arc' contributing <1%, cospatial with the [CI], CO(6-5), and H36α outflows; and low-surface-brightness dust beyond 5 pc along the pc-scale jet contributing 12+4−6%. The regions where near-infrared emission is missing ('Holes') coincide with the optically thick equatorial disk traced by 700 micron dust, HCN(3-2), CO(6-5), and the water maser disk. Fitting 1–1000 micron SEDs with four torus geometries, the paper finds that clumpy torus models best describe the central 2×5 pc SED, and radiative-transfer images from the best-fit model reproduce the observed 3.8-to-700 micron morphology with a 5×3 pc disk. It therefore concludes that most of the dust mass in the central 10 pc lies in an equatorial disk feeding the supermassive black hole, with the dusty outflow and jet-heated narrow-line-region dust as minor contributors.","pith_inferences":["A direct test of the disk-versus-wind picture would be to repeat this interferometric imaging on other nearby type 2 AGNs; if the pattern is general, one expects compact elongated near-infrared structures aligned with the torus funnel and outflow components contributing only a few percent.","Because the 4.3 micron filter may carry a larger spectral-feature contribution than the other two filters, spatially resolved spectroscopy between 3.8 and 4.8 microns could check whether the 450 K temperature and the 87% flux split survive; the morphological gap between the near-infrared and 700 micron emission would likely remain.","The paper's joint mapping of 3–10 micron thermal dust, 700–1200 micron cold dust, and molecular and ionized gas tracers provides a general recipe for separating accretion from outflow in AGN that could be applied to larger samples with the same or future instruments."],"forward_implications":["The 3–5 micron excess that has been debated for this AGN is identified as torus-funnel dust: the dominant structure is the same elongated component seen at 8–13 microns, not starlight or a dusty wind.","Because the 700 micron dust, HCN(3-2), CO(6-5), and the maser disk all lie along the equatorial axis while the near-infrared images show holes there, the obscuring torus is an optically thick equatorial disk whose cold dust mass feeds the black hole.","The dusty outflow contributes less than 1% of the near-infrared continuum and is spatially tied to the molecular and ionized outflows, so the wind scenario can explain only a minor fraction of the hot-dust emission.","The remaining 12% of emission from beyond 5 pc along the pc-scale jet direction requires dust heated by the jet or AGN in the narrow-line region, separate from the torus component.","Clumpy torus models with a compact 5×3 pc disk reproduce the observed morphology from 3.8 to 700 microns, whereas wind-dominated and smooth monolithic models fail the sub-millimeter data."],"supporting_citations":[{"why":"This MATISSE/VLTI N-band map provides the 8–13 micron extended dust morphology and temperatures that the AMI emission is shown to be cospatial with.","marker":"(J. W. Isbell et al. 2022)"},{"why":"These L- and M-band VLTI observations give the previous higher-resolution view of the compact disk and the standard-star fluxes used for calibrating the AMI images.","marker":"(J. W. Isbell et al. 2023)"},{"why":"The ALMA 700, 890, and 1200 micron continuum maps and [CI], H36α, and HCN(3-2) gas-tracer maps supply the disk, jet, and outflow positions against which the AMI features are registered.","marker":"(T. Izumi et al. 2023)"},{"why":"The VISIR N-band imaging and the disk-plus-hyperboloid torus model provide the previous morphological and model baseline for the extended dusty emission.","marker":"(M. Stalevski et al. 2017)"},{"why":"The clumpy torus model grid is the model family the SED fits statistically prefer for the central 2×5 pc region.","marker":"(M. Nenkova et al. 2008a,b)"},{"why":"The CAT3D-WIND torus-plus-wind model provides the competing geometry that underpredicts the sub-millimeter dust emission.","marker":"(S. F. Hönig & M. Kishimoto 2017)"},{"why":"The smooth torus model is the monolithic alternative that overpredicts the sub-millimeter dust emission.","marker":"(J. Fritz et al. 2006)"},{"why":"The two-phase clumpy torus model, with an interclump medium, is the third geometry that the paper tests against the SED.","marker":"(O. González-Martín et al. 2023)"},{"why":"The water maser disk establishes the edge-on inclination and the equatorial axis used to fix the torus model geometry and to interpret the near-infrared holes.","marker":"(L. J. Greenhill et al. 2003)"},{"why":"The CO(6-5) map shows the roughly 6 pc molecular disk cospatial with the 700 micron dust and offset by about 66 degrees from the AMI extended emission.","marker":"(K. R. W. Tristram et al. 2022)"}],"fun_headline_variants":["JWST spots dusty disk feeding Circinus's black hole","Most of Circinus's hot dust is a disk, not a wind","JWST resolves black hole's feeding disk: 87% of hot dust","Circinus's black hole: dusty disk dominates, outflow minor","JWST shows black hole's dust disk, not outflow, feeds it"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central claim assumes that the 3.8, 4.3, and 4.8 micron images are dominated by continuum dust emission; if the 4.3 micron filter is substantially contaminated by carbon-monoxide or carbon-dioxide-ice spectral features, the derived dust temperatures, the 87% flux share for the disk, and the preference for the clumpy torus model would all be biased.","fun_headline_variants_meta":{"raw":{"variants":["JWST spots dusty disk feeding Circinus's black hole","Most of Circinus's hot dust is a disk, not a wind","JWST resolves black hole's feeding disk: 87% of hot dust","Circinus's black hole: dusty disk dominates, outflow minor","JWST shows black hole's dust disk, not outflow, feeds it"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000726,"raw_usage":{"total_tokens":3329,"prompt_tokens":1095,"completion_tokens":2234,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":711,"completion_tokens_details":{"reasoning_tokens":2142}},"tokens_in":711,"tokens_out":2234,"duration_ms":17096,"temperature":1.0,"reasoning_tokens":2142,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T05:21:36.898672+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the fraction of non-continuum emission inside each AMI filter using spatially resolved spectroscopy of the inner 10 pc of Circinus, for example with JWST NIRSpec IFU or MIRI MRS spectra, by comparing narrow-band photometry at 3.8, 4.3, and 4.8 microns on the observed spectrum with a continuum-only baseline; if the F430M non-continuum fraction is close to the template-based value near 29% rather than the claimed <10%, the derived 450 K temperature and the 87% disk-dominated flux split must be re-evaluated.","supporting_citations":[],"review_version":1}