{"id":"92bc9f00-c2bb-4fe3-aa7a-a9475cd9b1ec","arxiv_id":"1909.00675","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"ALMA imaging of NGC 1068 reveals a stratified molecular torus of 10-30 parsec size whose inner dense gas is being swept by an AGN wind while an equatorial reservoir remains available to fuel the black hole.","lead":"Using ALMA observations of three molecular gas tracers, the team mapped a ring of gas around the heart of the galaxy NGC 1068 and found a compact, messy molecular torus that spans 10-30 parsecs. The images show that the bright active nucleus is blowing a wide wind into this torus, pushing out roughly half of its gas while the rest can still feed the black hole for a few million years.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 0.4-0.6 entrained torus fraction is set by the assumed uniform-density ring geometry, while the OUT-vs-CR discrimination is only a 1.2-1.3x chi2 preference with absolute chi2>=4; a clumpy or warped alternative could change the reservoir mass.","rationale":"The reader's conditional verdict already captures the core risk: the direct ALMA imaging and the radial stratification (HCO+ size vs CO size) are solid and independently supported, and the kinemetry/3DBarolo detection of large-scale non-circular motions is consistent with prior CO(3-2) work. My stress-test pass looked for a place where the central quantitative claim could fail even if those morphological results stand. The weakest point is the 0.4-0.6 entrained torus fraction in Sect. 7.2. It depends on a uniform-density, fixed-geometry toy torus whose fit quality is admitted to be poor and whose preference over a counter-rotating alternative is marginal (factor 1.2-1.3 in chi2). The model's key additional discriminator, the minor-axis velocity gradient, is only 'tentatively detected' in CO(6-5). Consequently the 'large gas reservoir unaffected' and the 1-4 Myr timescale are not direct measurements; they inherit the assumed geometry. This is not an internal inconsistency or a claim outside consensus; it is a robustness risk in the quantitative step. Since the reader already assigned CONDITIONAL for exactly this model dependence, my pass does not move the verdict; it sharpens the test that would validate or retire the claim.","tokens_in":44444,"tokens_out":4523,"duration_ms":52255,"concrete_test":"Re-run the Section 7.2 model grid replacing the constant-density toroidal ring with a clumpy density prior derived from the observed HCO+(4-3)/CO(2-1) ratio map, with theta and vout sampled on a grid or by MCMC. If theta=80 deg is not preferred over theta=50 deg by Delta-chi2 >= 4 (or if the CR-model is within the 1-sigma contour), then the 0.4-0.6 entrained fraction and the 1.2-1.8x10^5 M_sun unaffected reservoir are not constrained, and the 1-4 Myr fueling conclusion should be treated as a model-dependent upper limit rather than a robust claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The quantitative headline rests on the OUT-model of Sect. 7.2. That model fixes a uniform-density toroidal ringed disk (size 20-22 pc, PA=113 deg, i=-80 deg, maser rotation curve) and assigns the entrained gas fraction purely by the geometric intersection of a wide bicone with that ring, with the half-opening angle theta as the main free parameter. The paper itself states that the goodness-of-fit parameters are 'admittedly large (>=4)' and that the OUT-model beats the counter-rotating CR-model by only a factor ~1.2-1.3 in chi2. With absolute chi2 values this poor, the best-fit theta~80 deg (implying 0.4-0.6 M_torus entrained) is not a robust measurement; a clumpy or vertically stratified density distribution, or a slightly warped ring, could yield a substantially different intersection volume and thus a different residual equatorial reservoir. The 1-4 Myr fueling time then scales directly with that residual mass. The morphological detection of the stratified torus and the large-scale kinematic distortions are independently supported and are not the issue; the load-bearing fragility is specifically the conversion of a marginally preferred toy model into a quantitative entrained fraction and reservoir mass.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents ALMA observations of CO(2-1), CO(3-2), HCO+(4-3), and continuum at 229.7 and 344.5 GHz toward the circumnuclear disk (CND) and torus of NGC 1068 at 2-6 pc resolution. The authors resolve the CND as an asymmetric ~400 pc ringed disk with a central gas deficit, find edge-brightened NIR polarization arcs at its inner edge, and detect a molecular torus of ~3×10^5 M_sun extending over 10-30 pc, with the HCO+ torus (11 pc full size) a factor 2-3 smaller than the CO tori (26-28 pc), indicating radial density stratification. The kinematics show large non-circular motions and line splitting, modeled with kinemetry, 3DBarolo, and a custom morpho-kinematic torus model (OUT-model) that attributes the apparent counter-rotation to an AGN wind entraining 0.4-0.6 of the torus gas, leaving a residual equatorial reservoir of 1.2-1.8×10^5 M_sun that can fuel the AGN for 1-4 Myr.","tokens_in":44784,"tokens_out":8000,"duration_ms":69234,"significance":"If the quantitative claims hold, this is a landmark result: it directly images the long-postulated torus and demonstrates that it is not a passive rotating doughnut but a stratified structure actively shaped by AGN feedback, while still containing a reservoir for continued fueling. The strengths are the high angular resolution (2-6 pc), the multi-transition approach spanning n(H2)~1e3-1e7 cm^-3, the direct Gaussian-fitted sizes and orientations, the careful flux-recovery sanity checks against earlier IRAM data, and the consistency of the morphological results with independent NIR polarimetric and MIR interferometric evidence. The paper's observational framework is reproducible and the qualitative scenario of an outflowing torus with an equatorial reservoir is well supported. The quantitative values in the abstract, however, go beyond what the model discriminates, as discussed in the major comments.","major_comments":[{"comment":"The central quantitative claim, that the AGN wind entrains ~0.4-0.6 of the torus mass and leaves a 1.2-1.8×10^5 M_sun reservoir, is derived entirely from the geometric intersection of a fixed wide bicone (free half-opening angle θ) with an assumed uniform-density toroidal ringed disk of fixed size (20-22 pc), PA=113°, i=-80°, and the maser rotation curve. The paper itself states (Sect. 7.2) that the goodness-of-fit parameters are 'admittedly large (≥4)' and that the OUT-model beats the counter-rotating CR-model by only a factor ~1.2-1.3 in χ². With absolute χ² values this poor, the best-fit θ≈80° is not a robust measurement; a clumpy or vertically stratified density distribution, a warped ring, or a different inclination (the observed aspect ratio only gives i≥60-70°) could change the intersection volume substantially. Because the 1-4 Myr fueling time scales directly with the residual reservoir mass, the quantitative headline is not supported by the evidence as presented. I request either a sensitivity analysis of the entrained fraction and reservoir mass to the assumed density distribution and geometry, or a revised abstract/conclusions that present 0.4-0.6 and 1-4 Myr as order-of-magnitude model-dependent estimates.","section":"Section 7.2"},{"comment":"The absolute torus mass M_torus~3×10^5 M_sun is derived from CO(2-1) using a galactic XCO and line ratios R21=2.5, R31=2.9 taken from Viti et al. (2014). The paper acknowledges that XCO in AGN environments carries up to an order-of-magnitude uncertainty (Sect. 5.2). Since the reservoir mass (1.2-1.8×10^5 M_sun) and the fueling time (1-4 Myr) are fractions of M_torus, this systematic uncertainty propagates directly into the abstract's quantitative conclusions. The abstract should carry the same caveat that the paper applies to the column densities, or the authors should quote the mass-dependent quantities with an explicit XCO-dependence.","section":"Section 5.2"},{"comment":"The 3DBarolo fit assumes axisymmetric tilted rings, and the vertical velocity component vvert is not fitted iteratively but explored over discrete values (0, 25, 50, 100, 150 km/s; Appendix A), yielding vvert~100±50 km/s. This velocity enters the OUT-model's outflow velocity vout=(vrad^2+vvert^2)^{1/2}, which is used to generate the model p-v diagrams that discriminate OUT from CR. The paper itself notes (Sect. 7.1) that the model cannot reproduce 'any substantial deviation from the axisymmetric dependence of the kinematic parameters.' The degeneracy between vrad, vvert, and the assumed inclination in the torus region (r<20 pc) is not quantified; this weakens the discrimination between the two models to a greater degree than the reported χ² ratio alone suggests.","section":"Section 7.1 and Appendix A"},{"comment":"The model's rotation curve is taken from the H2O megamaser kinematics, which are confined to r<1 pc (Greenhill et al. 1996), and extrapolated to the 20-22 pc torus using a power-law vrot∝r^{-0.31}. Extrapolating the maser curve by a factor ~20 in radius is a strong assumption; a steeper or flatter rotation curve, or a substantial torus self-gravity, would change the model p-v diagrams and hence the best-fit θ. This assumption should be tested (e.g., by using the CO-derived rotation curve from the 3DBarolo fit where available) or its effect on the inferred entrained fraction quantified.","section":"Section 7.2 and Figs. 16-17"}],"minor_comments":[{"comment":"The χ² formula reads χ2 = ∑[D(i, j)− M(i, j)/σ]^2, which is not the standard expression; it should be ∑[(D(i, j)-M(i, j))/σ]^2. Please correct the typo since the model comparison is based on this quantity.","section":"Section 7.2"},{"comment":"The phrase 'of≃ up to an oder of magnitude' contains a typo; it should read 'an order of magnitude.'","section":"Section 5.2"},{"comment":"The notation 'v−vHEL sys ⊂ [−350, 350] km s−1' is ambiguous because the subscript and superscript are scrambled; please write the systemic velocity frame explicitly.","section":"Section 2.1"},{"comment":"The labels 'Knot N' and 'Knot S' are introduced in the caption but not defined in the figure or the main text; please define them or remove the labels.","section":"Figure 22 caption"},{"comment":"The [SiVI] data are credited to 'Ric Davies, private communication'; for reproducibility, a reference to the published dataset should be provided if one exists.","section":"Section 6.2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is well within the scope of A&A and the observational data are of high quality. My main concern is that the abstract and conclusions present the entrained fraction and fueling time as robust measurements, whereas the underlying model has poor absolute chi2 and a marginal preference over the counter-rotating alternative. I believe this is fixable with a cautious rewrite and possibly a robustness test, so I recommend major revision rather than rejection. The editor may also want to check that the reference list includes the data credited as 'private communication' in Sect. 6.2."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: this is the ALMA paper that finally shows the torus in three tracers at once, and the stratification is real: HCO+(4–3) torus is ~11 pc full size, while CO(2–1) and CO(3–2) tori are ~28 and ~26 pc. That alone is worth the read. The morphological work — Gaussian fits, deconvolved sizes, PA, offsets — is careful and the quoted errors are believable. The CND maps are also a step up from the 2014/2016 work, and the kinemetry/3DBarolo comparison is honest.\n\nThe genuinely new piece is the claim that a wide-angle AGN wind is entraining 0.4–0.6 of the torus mass, leaving ~1.2–1.8e5 Msun to fuel the AGN for 1–4 Myr. That is the headline, and it is the softest part. The entrained fraction is not measured from the data; it comes from a purpose-built morpho-kinematic model (Sect. 7.2) that assumes a uniform-density toroidal ring with fixed size, PA, inclination, and maser rotation curve. The free parameter is the bicone half-opening angle, fitted by chi2. The paper is transparent that the chi2 values are 'admittedly large (>=4)' and that the OUT-model beats the counter-rotating alternative by only a factor ~1.2–1.3. With absolute chi2 that poor, theta ~80 deg and the resulting 0.4–0.6 fraction are not robust; a clumpy or warped torus geometry could shift them substantially. The 1–4 Myr timescale scales directly with that residual mass.\n\nAlso, the mass numbers sit on assumed XCO and line ratios; the authors flag the order-of-magnitude XCO uncertainty in AGN environments. So the quantitative conclusions are conditional, and the paper mostly says so. I don't think the stress-test is wrong here; it correctly identifies that the load-bearing quantitative claim is the OUT-model, not the data.\n\nCredit where due: the density stratification, the connection between the torus and the CND streamers, and the vertical outflow component inferred from line splitting are all well-supported and important. The paper does not oversell its model comparison; it shows the p-v diagrams and lets you see the disagreements.\n\nWho is this for? Anyone working on AGN torus physics or AGN feedback. It deserves a serious referee — and in fact the version I have thanks Masatoshi Imanishi for a detailed referee report, so it has already been through peer review. My own verdict would be: accept with the understanding that the entrained fraction is a model-dependent estimate, not a measurement. For citing, I'd cite the stratification and the torus sizes, not the 0.4–0.6 number without a caveat.","headline":"A genuinely new, well-resolved view of the NGC1068 torus; the density stratification is the solid result, while the entrained fraction is a model-dependent extra.","tokens_in":45481,"tokens_out":2105,"would_cite":true,"duration_ms":20977,"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":"A wide-angle wind from the accreting black hole in NGC 1068 is currently blowing 40–60% of its obscuring torus outward.","keywords":["NGC 1068","Seyfert 2 galaxies","active galactic nuclei","molecular torus","AGN feedback","molecular outflows","circumnuclear disk","galaxy kinematics"],"falsifier":"Observe the torus minor axis in a dense-gas line at sub-parsec resolution and look for the south-redshifted, north-blueshifted velocity gradient predicted when an outflowing bicone is projected onto the disk plane. The paper reports that this gradient is absent in CO(2–1) yet tentatively present in CO(6–5); a clean detection or a firm non-detection after accounting for beam smearing would decide whether 40–60% of the torus is genuinely entrained.","tokens_in":44213,"feed_emoji":"🌪️","tokens_out":9688,"duration_ms":79969,"temperature":0.7,"pith_summary":"The paper tries to establish that the molecular torus of NGC 1068 is not a quiet, static doughnut but a radially stratified structure that is being partly blown apart by the AGN's own wind. Using high-resolution millimetre observations of three molecular transitions, it shows that low-density CO gas forms a disk 26–28 pc across while the dense HCO$^+$ gas is confined to an 11 pc core. The kinematics require a wide-angle wind that currently entrains 40–60% of the torus mass, yet a reservoir of $1.2$–$1.8\\times10^5\\,M_\\odot$ near the equatorial plane survives to keep feeding the black hole for at least 1–4 Myr. If correct, the result means AGN feedback shapes the very obscuring structure that defines Type 2 Seyfert galaxies, and that the torus has a finite, feedback-limited lifetime.","feed_headline":"Black hole wind is pushing 40-60% of NGC 1068's torus outward","feed_subtitle":"Dense gas near the torus equator survives and can keep feeding the black hole for 1-4 million years.","key_machinery":"The load-bearing mechanism is the OUT-model: a morpho-kinematic model of the torus as a toroidal ringed disk (4–5 pc tube radius orbiting at 5–6 pc, full size 20–22 pc, position angle 113°, inclination about 80°) whose gas follows the maser rotation curve, with a fraction of the gas—determined by where a wide-angle AGN-wind bicone with half-opening angle $\\theta\\simeq80^\\circ$ intersects the disk—given an extra radial outflow of about 100 km/s. Comparing synthetic and observed position-velocity diagrams along the torus major and minor axes, the model reproduces the apparent counter-rotation, the three velocity components seen on the minor axis, and a shallow minor-axis velocity gradient, and it fits better than a counter-rotating disk model. The model converts morphology and kinematics into the paper's quantitative statements about entrained mass and surviving reservoir.","core_discovery":"The central claim is that the torus of NGC 1068 is simultaneously the fuel supply and the target of AGN feedback: a wide-angle ionized wind launched from the accretion disk is sweeping through the torus along a hollow bicone, entraining roughly half of its molecular mass while leaving an equatorial reservoir intact. The evidence is morphological and kinematic: CO(2–1) and CO(3–2) trace an elongated disk with full sizes of 26–28 pc, HCO$^+$(4–3) traces a denser, smaller disk of 11 pc, and the velocity field contains gas at forbidden velocities that looks like counter-rotation but is better reproduced by outflow projected along the line of sight. The paper's favored model sets the entrained fraction at 0.4–0.6 of $M_{\\rm torus}\\simeq3\\times10^5\\,M_\\odot$ and identifies the untouched equatorial gas as a reservoir of $1.2$–$1.8\\times10^5\\,M_\\odot$, enough to sustain accretion for roughly 1–4 Myr even though supply through the 15–50 pc streamers is currently throttled.","pith_inferences":["A testable extension is that Seyfert type is partly a phase: a torus partially blown open by its own wind could expose the broad-line region for a few million years, then re-cover as material from the circumnuclear disk rebuilds the equatorial reservoir.","The measured density stratification predicts a multi-transition size sequence: in other nearby Seyferts, low-J CO tori should appear systematically larger than HCN/HCO$^+$ tori, which a modest survey could check without full kinematic modeling.","If the 1–4 Myr reservoir is consumed while the 15–50 pc streamers are outflowing, the AGN duty cycle in NGC 1068 may be set by a competition between wind erosion and streamer resupply, implying self-regulating, flickering accretion."],"forward_implications":["Torus size becomes tracer-dependent: low-density CO(2–1) and CO(3–2) disks span 26–28 pc, while the dense HCO$^+$(4–3) core spans only 11 pc.","Obscuration and feedback coexist: a wide-angle wind from the accretion disk is actively removing roughly half the torus mass even as the torus continues to hide and feed the nucleus.","The surviving equatorial reservoir of $1.2$–$1.8\\times10^5$ solar masses can sustain accretion for 1–4 Myr at the inferred rate of 0.05–0.1 solar masses per year.","Fresh gas is currently blocked at intermediate radii: the streamers connecting the torus to the circumnuclear disk are outflowing at about 0.6 solar masses per year, so fueling is thwarted between 15 and 50 pc.","The circumnuclear disk is globally affected: about half of its $1.4\\times10^8$ solar masses participates in an outflow with average radial speeds of about 85 km/s from 50 to 200 pc."],"supporting_citations":[{"why":"Supplies the H2O megamaser rotation curve and black-hole mass used to set the torus model's rotational field.","marker":"Greenhill et al. (1996)"},{"why":"Earlier ALMA CO(6–5) imaging of the compact torus whose minor-axis velocity gradient and outflow interpretation this work extends to larger scales.","marker":"García-Burillo et al. (2016)"},{"why":"Combined CO(6–5) torus data independently show the minor-axis velocity gradient that the OUT-model predicts.","marker":"Gallimore et al. (2016)"},{"why":"Provides the AGN-wind bicone geometry (PA 30°, inner and outer FWHM of 40° and 80°) used to define where the wind strikes the torus.","marker":"Das et al. (2006)"},{"why":"NIR polarimetry revealed an extended, roughly 50–60 pc dusty disk at PA about 120°, supporting the extended torus traced by CO.","marker":"Gratadour et al. (2015)"},{"why":"HCN/HCO$^+$ imaging that proposed a counter-rotating torus, the alternative scenario the OUT-model is designed to beat.","marker":"Imanishi et al. (2018)"},{"why":"The tilted-ring fitting method used to derive the global 3D outflow geometry and velocities that feed the torus model.","marker":"Di Teodoro & Fraternali (2015)"},{"why":"Supplies the CO-to-H2 conversion formula used to derive the torus, reservoir, streamer, and CND masses.","marker":"Bolatto et al. (2013)"},{"why":"Provides the CO line brightness-temperature ratios used to convert CO(2–1) and CO(3–2) fluxes into H2 masses.","marker":"Viti et al. (2014)"}],"fun_headline_variants":["AGN wind blasts 40-60% of NGC1068 torus, equatorial gas survives","Black hole wind sweeps half of NGC1068 torus, leaving fuel for 1-4 Myr","NGC1068's AGN wind: half of torus entrained, dense core persists","Black hole wind evicts half of NGC1068's torus, reservoir remains","Torus under fire: NGC1068's wind strips 40-60%, core fuels on"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper's quantitative claims—40–60% of the torus mass outflowing and $1.2$–$1.8\\times10^5$ solar masses still available—rest on a model that assumes the torus is a smooth, axisymmetric ringed disk of fixed size and orientation with uniform gas density; if the true torus is warped, clumpy, or non-axisymmetric, both numbers would change.","fun_headline_variants_meta":{"raw":{"variants":["AGN wind blasts 40-60% of NGC1068 torus, equatorial gas survives","Black hole wind sweeps half of NGC1068 torus, leaving fuel for 1-4 Myr","NGC1068's AGN wind: half of torus entrained, dense core persists","Black hole wind evicts half of NGC1068's torus, reservoir remains","Torus under fire: NGC1068's wind strips 40-60%, core fuels on"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000996,"raw_usage":{"total_tokens":4374,"prompt_tokens":1255,"completion_tokens":3119,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":871,"completion_tokens_details":{"reasoning_tokens":2997}},"tokens_in":871,"tokens_out":3119,"duration_ms":274366,"temperature":1.0,"reasoning_tokens":2997,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T05:39:32.407051+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Observe the torus minor axis in a dense-gas line at sub-parsec resolution and look for the south-redshifted, north-blueshifted velocity gradient predicted when an outflowing bicone is projected onto the disk plane. The paper reports that this gradient is absent in CO(2–1) yet tentatively present in CO(6–5); a clean detection or a firm non-detection after accounting for beam smearing would decide whether 40–60% of the torus is genuinely entrained.","supporting_citations":[{"cited_title":"J., Gwinn , C","cited_arxiv_id":null,"evidence_quote":"Supplies the H2O megamaser rotation curve and black-hole mass used to set the torus model's rotational field."},{"cited_title":"2015, , 581, L8","cited_arxiv_id":null,"evidence_quote":"NIR polarimetry revealed an extended, roughly 50–60 pc dusty disk at PA about 120°, supporting the extended torus traced by CO."},{"cited_title":"2018, , 853, L25","cited_arxiv_id":null,"evidence_quote":"HCN/HCO$^+$ imaging that proposed a counter-rotating torus, the alternative scenario the OUT-model is designed to beat."},{"cited_title":"2014, , 570, A28","cited_arxiv_id":null,"evidence_quote":"Provides the CO line brightness-temperature ratios used to convert CO(2–1) and CO(3–2) fluxes into H2 masses."}],"review_version":1}