{"id":"f9ff8eb5-a967-4934-94aa-ebb2e09ddd03","arxiv_id":"2505.03884","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Luminous unobscured quasars at z~2 have molecular gas fractions of 2-32 percent, generally lower than those of inactive galaxies and obscured quasars, supporting an obscured-to-unobscured evolutionary sequence.","lead":"ALMA observations of 41 bright quasars at redshift about 2 show that their host galaxies contain less molecular gas relative to their stars than inactive or dust-hidden galaxies do. This supports the idea that quasars move from gas-rich, obscured systems to gas-poor, unobscured ones as they blow out their surroundings.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The gas-depletion claim hinges on an assumed 5 kpc CO radius for every source; a 2 kpc radius raises the median gas fraction by roughly a factor of three, so the 'universal' conclusion is conditional on source sizes that only three resolved objects support.","rationale":"The reader's weakest assumption correctly identifies the single most load-bearing point: Eq. 3's fixed 5 kpc radius and 30-degree inclination feed directly into Eq. 7, so every gas fraction inherits a geometric prior that is measured for only three of 41 sources. The paper's own r = 2 kpc sensitivity check quantifies the impact: the median fgas rises by about a factor of three, reducing but not eliminating the contrast with star-forming galaxies and obscured quasars. This is therefore a conditional-support situation rather than a refutation. The homogeneous ALMA sample, the transparent reduction checks, and the attempt to homogenize line ratios and alpha_CO in Figure 8 are real strengths, but they do not remove the dependence of all fgas values on the assumed CO size. Because the reader already reached a CONDITIONAL verdict on this basis, no verdict change is needed; the concern is real but does not overturn the paper's central result, which remains plausible for typical r = 5 kpc sizes.","tokens_in":25328,"tokens_out":13377,"duration_ms":145236,"concrete_test":"Recompute all gas fractions in Table 3 while sweeping the CO radius from 1 to 8 kpc and the inclination from 15 to 45 degrees; for each combination, compare the resulting median log10(fgas) with the Tacconi et al. (2020) main-sequence expectation at matched stellar mass and redshift, and check whether the combinations that erase the depletion are consistent with the three measured sizes and with SED-based stellar masses for the sample. If SED masses are unavailable, this check at least quantifies the geometric prior needed for the central claim to fail.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.3 computes every dynamical mass from Eq. 3 using a single assumed CO-emitting radius r = 5 kpc and inclination i = 30 degrees, even though only three sources are spatially resolved, with measured sizes 0.53–0.65 arcsec (about 5 kpc). This assumption is load-bearing because stellar masses are not independently measured but are defined through Eq. 7 as M_dyn - M_BH - M_gas, so the adopted radius propagates directly into every gas fraction. The paper's own sensitivity test in Section 4.1 shows that adopting r = 2 kpc changes the median log10(fgas) from -1.22 to -0.76, i.e. a median fgas of about 0.06 rising to about 0.17. The conclusion of depletion relative to inactive galaxies and obscured quasars survives at r = 2 kpc, but with reduced contrast; at still smaller radii or more face-on inclinations the gap can close further. Since only a discrete sensitivity check is reported rather than a marginalization over physically plausible radii, the central claim is conditional on an unverified geometric prior.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents ALMA CO observations of 41 luminous unobscured quasars at z∼2 from SDSS, comprising a main sample of 32 sources observed in CO(3-2) plus 9 archival sources observed in CO(3-2), CO(4-3), or CO(7-6). The authors measure CO luminosities, derive gas masses using fixed conversion factors (α_CO = 0.8, r31 = 0.97, r41 = 0.87, r71 = 0.2), and compute dynamical masses from CO line widths under the assumption of ordered rotation with a fixed radius r = 5 kpc and inclination i = 30° for all sources (Eq. 3). Stellar masses are then obtained as M★ = M_dyn − M_BH − M_gas (Eq. 7), so gas fractions (Eq. 8) depend directly on the assumed geometry. The central finding is that these quasars have low gas fractions, median ∼0.06, 'generally lower than both inactive galaxies and obscured quasars at similar redshifts', which the authors attribute to an evolutionary trend from gas-rich obscured quasars to gas-poor unobscured quasars. They also report a tentative correlation with C IV blueshift / He II EW and identify a few CO-detected companions. A sensitivity analysis in Section 4.1 shows that changing the radius to 2 kpc raises the median log10 f_gas from −1.22 to −0.76 (median f_gas ≈ 0.17) and that implausible values (r31 ≈ 0.1 or α_CO ≈ 11) would be required to erase the deficit entirely.","tokens_in":25577,"tokens_out":5558,"duration_ms":52222,"significance":"If the result holds, it is an important piece of evidence for AGN-driven gas depletion in the most luminous quasars at cosmic noon, a regime where simulations currently do not predict depleted reservoirs. The paper's strengths are its relatively large and uniformly selected sample, the homogeneous ALMA processing, the explicit sensitivity analysis in Section 4.1 (which correctly identifies the magnitude of the geometric prior), and the connection drawn between BLR/NLR outflow indicators and the molecular ISM. The paper also includes upper limits for non-detections and makes the reduced data available on request. These features make the work potentially valuable to the AGN feedback community, provided the central geometric assumption is either better justified or the conclusions are appropriately caveated.","major_comments":[{"comment":"The central claim of universal gas depletion rests on a single assumed CO-emitting radius r = 5 kpc and inclination i = 30° for every source, although only three objects are spatially resolved (sizes 0.53–0.65 arcsec ≈ 5 kpc). Because the stellar mass is not independently measured but is defined by M★ = M_dyn − M_BH − M_gas (Eq. 7), the assumed radius propagates directly into every gas fraction. The paper's own test in Section 4.1 shows that adopting r = 2 kpc changes the median log10 f_gas from −1.22 to −0.76, i.e., the median f_gas rises from ≈0.06 to ≈0.17, reducing the contrast with star-forming galaxies by roughly a factor of three. At smaller radii or larger inclinations (the sin^2(i) dependence in Eq. 3 is significant) the gap could close further. Since only a single discrete sensitivity check is reported rather than a marginalisation over physically plausible (r, i), the 'universal' conclusion is conditional on an unverified geometric prior. I ask the authors to (a) adopt resolved sizes where available and marginalise over plausible radius and inclination distributions, or (b) substantially soften the wording in the abstract and conclusions, quantifying how the significance of the deficit changes with r and i.","section":"3.3 (Eq. 3) and 4.1"},{"comment":"The comparison that justifies the statement that gas fractions are 'generally lower than both inactive galaxies and obscured quasars' mixes literature samples with widely different α_CO and line-ratio conventions (Table 2: α_CO ranges from 0.8 to 3.6 for Perna+18, is 3.6 for Circosta+21, and 0.66–15.22 for Sanders+23 SFGs). Since f_gas scales linearly with the assumed α_CO (Eq. 6), the apparent deficit relative to star-forming galaxies could partly reflect these convention differences rather than an intrinsic physical depletion. The sensitivity test in Section 4.1 (r31 ≈ 0.1 or α_CO ≈ 11 would erase the deficit) is reassuring, but it applies only to the authors' own measurements, not to the comparison samples. I recommend that the authors recompute all literature gas fractions with a common set of conversion factors (as already done for the obscured quasars in Fig. 8) or explicitly quantify how the deficit depends on adopting plausible alternative α_CO and line ratios for the comparison samples.","section":"4.1, Fig. 7, Table 2"}],"minor_comments":[{"comment":"The abstract uses 'CIV' but the main text uses 'Civ'; please standardise the notation for the C IV line.","section":"Abstract"},{"comment":"Eq. (3) as printed appears dimensionally inconsistent (FWHM appears without a square, and sin^2(i) appears in the numerator); please verify the expression against Wang et al. (2013) and Bischetti et al. (2021), and specify the units of each term.","section":"3.3 (Eq. 3)"},{"comment":"The sentence 'Molecular gas at 5kpc has been identified but as outflows and not part of the mainbulkrotation' is incomplete and unclear; please rephrase to state whether such 5 kpc gas is associated with rotation or with outflows, and cite the relevant literature.","section":"4.1"},{"comment":"For non-detections the gas fraction column lists values such as '<0.05', but the caption does not explicitly state that these are 3σ upper limits computed with the assumed median line width; please clarify the caption.","section":"Table 3"},{"comment":"The grey line for main-sequence star-forming galaxies from Tacconi et al. (2020) should be accompanied by a statement of the stellar mass and the α_CO / line-ratio assumptions used to place it on the plot.","section":"Fig. 7"},{"comment":"There is a typo: 'J1606+16735' should be 'J1606+1735' as listed in Table 3; also check 'J0052+0104' versus 'J0052+0140' for consistency.","section":"3.7"}],"recommendation":"major_revision","confidential_remarks":"The paper's main claim is sensitive to the assumed CO radius and inclination in a way that the authors themselves quantify, but they do not fully propagate this into the conclusions. The 'universal' wording in the abstract and conclusions is stronger than the current sensitivity analysis supports. A revision that either obtains better size constraints, marginalises over geometry, or carefully rephrases the claims would address this. The literature-comparison issue is secondary but worth fixing for a robust central claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Best just to know the central claim before reading: the paper argues that luminous unobscured quasars at z~2 have depleted molecular gas reservoirs, but the strength of that claim depends on one geometric assumption that only three resolved sources support. The data are real and worth having; the headline is a bit stronger than the evidence.\n\nWhat's new: a uniform ALMA CO(3-2) sample of 32 quasars plus 9 archival targets, all with rest-UV and rest-optical spectra from the same group's prior work. That lets them connect BLR winds (CIV blueshift), NLR outflows ([OIII] W80), and the molecular ISM in a way that hasn't been done systematically at this luminosity. The detection rate is about 47%, gas fractions 0.02-0.32 with median about 0.06. The sensitivity analysis in Section 4.1 is the best part: they show that erasing the depletion claim would require r31 ~0.1 or alpha_CO ~11, both implausible for luminous quasars. Fair and transparent.\n\nThe soft spots are real but not fatal. Every dynamical mass uses r=5 kpc and i=30 deg, even though only three sources are resolved; at r=2 kpc the median fgas goes from about 0.06 to about 0.17, which lowers but doesn't erase the contrast with star-forming galaxies. The paper acknowledges this, but it shows how load-bearing the radius assumption is. The literature comparison is not fully homogenized: different samples use different alpha_CO and line ratios, and while they re-compute gas fractions in Fig 8 with common values, the underlying dynamical mass methods still differ. 'Universal' is a stretch for a sample that is deliberately luminous and unobscured, with 20 detections out of 41. The CIV correlation is tentative and they say so, so no problem there.\n\nOverall: a solid, honest observational paper with a valuable new dataset. The depletion claim is probably right in direction, but the 'universal' framing and the exact amplitude hinge on an assumed radius that only three sources measure. A serious referee should send this back for a softened interpretation and more explicit treatment of radii/inclination. I'd accept it for peer review; the data deserve to be in the literature.","headline":"Useful new CO sample, honest sensitivity analysis, but 'universal depletion' leans on a single geometric assumption only three sources justify.","tokens_in":26209,"tokens_out":4139,"would_cite":true,"duration_ms":39010,"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":"Luminous unobscured quasars at z~2 are gas-poor: gas fractions of 0.02-0.32 lie below inactive galaxies and obscured quasars, marking an evolutionary shift from gas-rich obscured to gas-poor unobscured hosts.","keywords":["quasars","molecular gas","gas fractions","CO emission","AGN feedback","galaxy evolution","cosmic noon","Type 1 quasars"],"falsifier":"Resolve the CO emission in a representative subset of about a dozen of these quasars at sub-kiloparsec resolution and measure the actual radii and orientations; if the median radius is close to $2\\,\\mathrm{kpc}$ rather than $5\\,\\mathrm{kpc}$, the dynamical-mass equation raises the median $\\log_{10} f_{\\rm gas}$ from $-1.22$ to $-0.76$, largely erasing the claimed depletion relative to the star-forming main sequence. Alternatively, measure stellar masses independently from rest-frame optical or near-infrared SED fitting and recompute the gas fractions directly.","tokens_in":25125,"feed_emoji":"🌌","tokens_out":11787,"duration_ms":102056,"temperature":0.7,"pith_summary":"This paper tries to establish that the brightest unobscured quasars at redshift $\\sim 2$ — the epoch of peak star formation, 'cosmic noon' — are systematically gas-depleted: they hold far less molecular gas per unit stellar mass than typical star-forming galaxies or than obscured quasars of similar luminosity. Using 41 quasars with carbon-monoxide (CO) observations from a millimetre interferometer, it derives gas masses and dynamical masses and finds gas fractions in the range $0.02$–$0.32$, with a median near $0.06$. If this is right, it matters because it places luminous AGN activity at the end of a gas-clearing sequence: a galaxy fuelling a bright quasar first appears dusty, gas-rich and obscured, then sheds its molecular reservoir and emerges as a blue, unobscured, gas-poor quasar. The paper also reports a tentative connection between higher gas fractions and stronger broad-line-region winds traced by C IV blueshifts.","feed_headline":"Bright quasars at z≈2 run low on star-forming gas","feed_subtitle":"Median gas fractions near 6 percent in 41 quasars support an obscured, gas-rich to unobscured, gas-poor sequence.","key_machinery":"The load-bearing machinery is the chain from CO line measurements to gas fractions. A CO(3–2) luminosity is converted to CO(1–0) using fixed line ratios ($r_{31}=0.97$, $r_{41}=0.87$, $r_{71}=0.20$) and then to molecular gas mass with $\\alpha_{\\rm CO}=0.8\\ M_\\odot\\,({\\rm K\\,km\\,s^{-1}\\,pc^2})^{-1}$; dynamical masses come from the CO line width through a rotating-disk model, $M_{\\rm dyn} = 1.16\\times 10^5 \\times 0.75 \\times \\mathrm{FWHM_{CO}} \\times \\sin(i)^2 \\times r_{\\rm kpc}$, with inclination $i=30^\\circ$ and radius $r=5\\ {\\rm kpc}$ assumed for every target; stellar masses are the remainder $M_\\star = M_{\\rm dyn} - M_{\\rm BH} - M_{\\rm gas}$, so the gas fraction is $M_{\\rm gas}/M_\\star$. Because only three targets are spatially resolved, the single assumed radius is what carries the depletion contrast.","core_discovery":"The central claim is that luminous, unobscured (Type 1) quasars at $z\\sim 2$ have depleted molecular gas reservoirs. In the 20 of 41 targets with CO detections the median gas mass is $8.0 \\pm 1.5 \\times 10^9\\,M_\\odot$, the median dynamical mass is $1.4 \\pm 0.9 \\times 10^{11}\\,M_\\odot$, and the gas fractions $M_{\\rm gas}/M_\\star$ range from $0.02$ to $0.32$ with median $0.06 \\pm 0.09$. These values sit below both inactive main-sequence galaxies at the same redshift and obscured quasar populations (red quasars and hot dust-obscured galaxies) at comparable luminosity. The authors interpret the ordering obscured/gas-rich $\\rightarrow$ unobscured/gas-poor as an evolutionary trend: the same luminous phase that evacuates nuclear dust and gas leaves a gas-poor host. They also present tentative evidence that quasars with stronger disk winds (C IV blueshifts above $1000\\ \\mathrm{km\\,s^{-1}}$ and weak He II) have somewhat higher gas fractions, and they find no strong sign that these quasars live in extremely dense environments.","pith_inferences":["A direct extension the authors leave implicit: if the gas-poor unobscured phase is the end state, the same targets should show recently removed gas — for example broad molecular outflows or excess ionised gas — at a rate that scales with how far below the main sequence each object sits; stacking the non-detections for broad wings, which the paper begins to do, is a cheap test.","The tentative C IV/gas-fraction link could be sharpened by testing whether gas fraction correlates more tightly with Eddington ratio or with wind velocity, using the same sample's measured black-hole masses and $L/L_{\\rm Edd}$ values.","Because stellar masses are here derived rather than measured, the cleanest independent check is rest-frame optical or near-infrared SED fitting for a subsample; higher stellar masses would make the depletion stronger, while lower stellar masses would weaken it."],"forward_implications":["If the depletion is real, the most luminous unobscured quasars at $z\\sim2$ are not sites of ongoing vigorous star formation but hosts whose molecular fuel has been largely consumed or expelled.","Gas fractions rise with obscuration across comparable luminous samples, so the molecular reservoir shrinks as the quasar sheds its dusty cocoon; the ordering is consistent with a gas-rich obscured phase preceding the unobscured phase.","Quasars with the strongest broad-line winds (C IV blueshift above $1000\\ \\mathrm{km\\,s^{-1}}$) tend toward higher gas fractions, suggesting that disk winds are more easily launched when some cold gas remains rather than in fully depleted hosts.","Only about 15 percent of the sample shows CO-bright companions at the achieved sensitivity, so luminous quasars at this epoch do not appear to require extreme overdensities; deeper observations could still uncover fainter companions."],"supporting_citations":[{"why":"Supplies the adopted line ratio $r_{31}=0.97$ and the standard framework for converting CO luminosities to gas masses.","marker":"Carilli & Walter 2013"},{"why":"Provides the $\\alpha_{\\rm CO} = 0.8$ conversion factor used to turn CO(1-0) luminosity into molecular gas mass.","marker":"Downes & Solomon 1998"},{"why":"Gives the formula relating integrated CO intensity to line luminosity $L'_{\\rm CO}$ used in the analysis.","marker":"Solomon & Vanden Bout 2005"},{"why":"Supplies the dynamical-mass prescription of equation (3) and the main comparison sample of similarly luminous quasars.","marker":"Bischetti et al. 2021"},{"why":"Provides the lower-luminosity AGN sample at the same redshift whose higher gas fractions set the contrast.","marker":"Circosta et al. 2021"},{"why":"Hot dust-obscured galaxy comparison with markedly higher gas fractions, anchoring the obscured-to-unobscured trend.","marker":"Sun et al. 2024"},{"why":"Red quasar comparison used to place unobscured quasars at the gas-poor end of the evolutionary sequence.","marker":"Banerji et al. 2021"},{"why":"Defines the star-forming main-sequence scaling used as the non-AGN benchmark for gas fractions.","marker":"Tacconi et al. 2020"}],"fun_headline_variants":["Quasar gas fractions sink in unobscured phase","Luminous quasars at z≈2 starve for gas","Gas-poor hosts: 41 quasars at z≈2","Type 1 quasars show universal gas depletion"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire depletion signal rests on assuming every quasar's molecular gas sits in a rotating disk of radius $5\\,\\mathrm{kpc}$ inclined at 30 degrees; if the true sizes are closer to $2\\,\\mathrm{kpc}$, the paper's own calculation shifts the median gas fraction upward by about half a decade and the gap to star-forming galaxies narrows considerably.","fun_headline_variants_meta":{"raw":{"variants":["Quasar gas fractions sink in unobscured phase","Luminous quasars at z≈2 starve for gas","Gas-poor hosts: 41 quasars at z≈2","Type 1 quasars show universal gas depletion"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000351,"raw_usage":{"total_tokens":2030,"prompt_tokens":1174,"completion_tokens":856,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":790,"completion_tokens_details":{"reasoning_tokens":789}},"tokens_in":790,"tokens_out":856,"duration_ms":8476,"temperature":1.0,"reasoning_tokens":789,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:43:20.472417+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Resolve the CO emission in a representative subset of about a dozen of these quasars at sub-kiloparsec resolution and measure the actual radii and orientations; if the median radius is close to $2\\,\\mathrm{kpc}$ rather than $5\\,\\mathrm{kpc}$, the dynamical-mass equation raises the median $\\log_{10} f_{\\rm gas}$ from $-1.22$ to $-0.76$, largely erasing the claimed depletion relative to the star-forming main sequence. Alternatively, measure stellar masses independently from rest-frame optical or near-infrared SED fitting and recompute the gas fractions directly.","supporting_citations":[],"review_version":1}