{"id":"07628c16-6833-44bd-a0c3-7b7840239583","arxiv_id":"2502.05283","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Cold molecular outflows are detected in 8 of 11 IR-luminous z>6 quasars, and their high speeds imply AGN feedback.","lead":"Using ALMA observations of 11 very bright quasars from when the universe was less than a billion years old, astronomers found fast cold gas outflows in 8 of them. This is the first survey showing that these early quasars blow away large amounts of molecular gas, likely powered by the black hole rather than by stars.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. 2's additive intercept b≈180–190 M☉/yr dominates the derived z>6 outflow rates; if this low-z calibration constant is not universal, the AGN-energy requirement in Sec. 4.5 is not secured.","rationale":"The paper has two nested claims. The first—blueshifted OH absorption is common (8/11) and faster than in DSFGs—is well-supported by the spectra and by the self-consistent reanalysis of comparison samples. The second—AGN energy is required because SN coupling would need to be ~30%—depends on kinetic powers that are not directly observed. Eout = 0.5 Mdot v^2 uses v from the spectra, but Mdot comes from Eq. 2, a fitted mapping from EW√L calibrated on low-z objects. My stress-test focuses on the least-secure component of that mapping: the large additive intercept b≈180–190 M☉/yr. For the relevant parameter range, this constant is as large as or larger than the multiplicative term, and it is an extrapolated, physically unmotivated offset. A zero-intercept sensitivity test directly quantifies how much of the AGN-requirement conclusion depends on this fitted constant. Even if the test shows some sensitivity, the paper's more basic claims stand, so the reader's CONDITIONAL verdict is appropriate; the caveat attached to Eq. 2 should be made explicit in the abstract and title, which currently overstate certainty with 'Direct Evidence.' No code is shipped, but the test can be performed from the published tables and Eq. 2.","tokens_in":36117,"tokens_out":7158,"duration_ms":84533,"concrete_test":"Recompute Table 3 and the right panels of Fig. 8 using Eq. 2 with b=0 for both the LIR and LFIR variants (or refit the low-redshift training sample forcing a zero intercept). Compare the median required supernova coupling fraction (Eout/ESNe) to the 10% ceiling. If the median remains ≳20% and no individual source needs less than ~10%, the AGN-energetics conclusion is robust; if the median falls below ~10%, the conclusion is an artifact of the intercept and should be downgraded to emphasize that only faster velocities are directly established. In addition, apply Eq. 2 to the two non-outflow quasars: if it predicts substantial outflow rates, the intercept is demonstrably unphysical outside the low-z training domain.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central energy argument compares Eout = 0.5 Mdot v^2 to supernova and AGN power (§4.5, Fig. 8, right). Eout inherits all assumptions in the empirical outflow-rate calibration of Eq. 2. That calibration contains a large additive constant: b ≈ 180–190 M☉/yr (S20b and refit with LFIR). For the quasars, the multiplicative term m × EW_v<−200 × sqrt(L/1e12 L☉) is typically 60–750 M☉/yr, so b contributes roughly 20–75% of the quoted Mdot values in Table 3. The constant was fitted to low-redshift galaxies and is not physically motivated; indeed, for EW→0 Eq. 2 predicts a 190 M☉/yr outflow, which is inconsistent with the authors' own classification of P009-10 and J0305-3150 as non-outflows (§3.1, Table 2). Because Eout scales linearly with Mdot, setting b=0 would lower several derived kinetic powers by factors of 2–4. The required supernova coupling would drop from ~30% toward values near the 10% ceiling, so the conclusion that AGN energy is 'required' becomes a calibration-dependent statement rather than a robust direct inference. The velocity and detection-rate results are not affected.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents ALMA OH 119 µm observations of 11 IR-luminous quasars at z>6 (8 new targets plus 3 reanalyzed from the literature) and reports a 73%±13% detection rate of blueshifted molecular outflows, outflow velocities typically ~300 km/s faster than the z~5 non-quasar DSFGs and low-redshift ULIRGs, modest mass-loading factors (~0.5), and kinetic powers of order 0.1% of the AGN luminosity. The authors conclude that cold molecular outflows are ubiquitous in the most IR-luminous reionization-era quasars and that AGN energy injection is required to explain their kinetic power, based primarily on the high outflow velocities and the comparison of outflow kinetic power to available supernova and AGN energy budgets.","tokens_in":36298,"tokens_out":5634,"duration_ms":60574,"significance":"If the conclusions are robust, this is the first statistical sample of cold molecular outflows in z>6 quasars and provides the strongest current evidence that AGN feedback operates on the cold ISM during reionization. The paper has clear strengths: the outflow classification is conservative (blueshifted OH absorption beyond the [CII] systemic velocity), the spectral detections appear visually unambiguous, the literature data are reanalyzed in a self-consistent way, and the authors explicitly cross-check their SFR and LAGN estimates with independent tracers. They also honestly flag the circularity in their outflow-rate calibration and the model dependence of the derived energetics. However, the central AGN-energy argument rests on an empirical outflow-rate calibration that contains a large additive constant fitted to low-redshift galaxies; the paper does not demonstrate that the conclusion survives plausible variations of that calibration.","major_comments":[{"comment":"The additive constant b≈180–190 M☉/yr in the empirical outflow-rate calibration contributes substantially to the derived outflow rates for several quasars. For example, for J1319+0950 the multiplicative term m×EW_{v<-200}×√(L/10^12 L☉) is roughly 60 M☉/yr, so b contributes about 75% of the 260 M☉/yr quoted in Table 3. Because both momentum flux and kinetic power scale linearly with Mdot, the claim in §4.5 that supernovae cannot power the outflows without an unrealistically high coupling fraction is directly contingent on this low-redshift constant. The internal inconsistency is also apparent: for EW→0, Eq. (2) predicts a constant ~190 M☉/yr outflow, which contradicts the authors' own classification of P009-10 and J0305-3150 as non-outflows in Table 2. The paper should demonstrate that the AGN-energy conclusion is robust to the calibration: for instance, repeat the energy budget analysis with b set to zero, with the multiplicative term only, or with an independent estimate of Mdot from column density and radius. As written, the central claim is calibration-dependent rather than a direct inference from the data.","section":"§3.3, Eq. (2); Table 3; §4.5"},{"comment":"The paper correctly acknowledges a partial circularity between Mdot and the IR-based SFR, since Mdot uses sqrt(L) through Eq. (2) while LSF is derived from the same far-IR photometry. However, the impact on the derived mass-loading factors and on the Edot/SFR comparison in the right panels of Fig. 8 is not quantified. The statement that the relation is not totally circular because Mdot is not proportional to SFR^1/2 is insufficient, because the normalization (m and b) is still set by the same photometry. The authors should provide a quantitative sensitivity test, e.g., recompute η and Edot using only the [CII]-based SFRs, or using the IR-based LAGN values (which they note are ~2× lower), and state how the conclusions in §4.4 and §4.5 shift. Without such a test, the conclusion that the mass-loading factors are 'modest' and that supernovae are insufficient remains coupled to the assumed SED decomposition.","section":"§3.3, §4.4, Fig. 8"}],"minor_comments":[{"comment":"The statement that the flux-scale discrepancy with Butler et al. (2023b) 'appears to be in the data and not in any analysis steps' is vague; please clarify what tests were performed (e.g., comparing with the ALMA sensitivity calculator and other observations) and whether any of the literature values used in the analysis rely on the discrepant absolute flux scale.","section":"§2.2"},{"comment":"The spectral fits are described as 'certainly not unique'; please describe how the systematic uncertainties on v50, v84, vmax, and EW_{v<-200} were estimated, for example by repeating the fits with different initial conditions or different numbers of Gaussian components, so that the 10–20% uncertainties quoted in Table 2 are reproducible.","section":"§3.1, Fig. 1"},{"comment":"The KS tests comparing the vout distributions are reported only as p<0.05; please report the sample sizes, the test statistic, and whether the result survives exclusion of the most extreme object (e.g., J2054-0005), given the small number of quasars in the sample.","section":"§4.2, Fig. 5"},{"comment":"The adopted mid-IR power-law index α=1.0 is a fixed choice for all quasars, and the paper states that varying α between 0.5 and 2.0 changes LFIR by ≲15%; consider showing this range explicitly for a representative source in the main text, since the appendix only compares fits with limited versus full photometry.","section":"§3.2, Appendix A"},{"comment":"The abstract's phrase 'Direct Evidence for AGN Feedback' is stronger than the paper's own concluding caveat that the result is 'far from smoking gun' evidence. Consider tempering the title or abstract (e.g., 'strong evidence' or 'consistent with') so that the presentation matches the calibrated, assumption-dependent nature of the energetics argument.","section":"Abstract, §5"}],"recommendation":"major_revision","confidential_remarks":"The paper is a valuable observational contribution and the authors are transparent about many of their assumptions. The main risk is that the headline conclusion ('AGN feedback required') is driven by the additive constant in the empirical outflow-rate calibration of Eq. (2), which contributes 20–75% of the derived Mdot values. If the authors add a robustness test (e.g., setting b=0 or using an independent Mdot estimator) and the energy argument still requires AGN participation, the paper would be acceptable after a minor revision; if not, the conclusions need to be scaled back to 'consistent with AGN feedback' rather than 'direct evidence.' I would encourage the editor to request this sensitivity analysis as a condition of acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is the first real sample of OH 119um outflows in z>6 quasars, and the primary observational results—8/11 detection rate and outflow velocities roughly 300 km/s faster than z~5 non-quasar DSFGs—are solid and important. The spectra look convincing, the reanalysis of literature targets is careful, and the authors are unusually candid about the limits of their outflow-rate estimates. They even flag the circularity between outflow rates and IR-based SFRs in Sec 4.4, and the conclusions walk back the title's 'direct evidence' language to 'far from smoking gun.'\n\nThe soft spot is exactly where you put a finger: Eq. 2. The empirical calibration has an additive constant b ~ 180-190 Msun/yr, which is physically odd—it predicts an outflow even at zero equivalent width—and it contributes 20-75% of the quoted Mdot values for this sample. The kinetic powers scale linearly with Mdot, so setting b=0 (or recalibrating to high-z) lowers Eout by factors of 2-4. That shifts the supernova coupling requirement from ~30% down to near the ~10% ceiling, so the claim that AGN energy is required becomes a calibration-dependent statement. The velocity and detection-rate results are not affected, so the central observational contribution stands.\n\nAlso, the SED separation of AGN vs star-forming dust is necessarily crude with limited photometry for some targets. The paper checks this with [CII] and with fits restricted to 120/160um, which helps, but the assigned LAGN values carry real uncertainty.\n\nWho should read it: anyone working on high-redshift AGN feedback or cold outflows. It's a genuine step beyond the handful of single-object studies.\n\nRecommendation: send it to review. It deserves referee time. Ask the authors to show how the energetics depend on the intercept in Eq. 2, and to soften the 'direct evidence' framing unless they can make the argument robust to that calibration choice.","headline":"First systematic OH 119um survey in z>6 quasars: the detection rate and fast outflows are robust, but the AGN-required energetics lean on an extrapolated low-z calibration with a large additive constant.","tokens_in":36964,"tokens_out":2674,"would_cite":true,"duration_ms":27177,"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":"Blueshifted OH absorption in 8 of 11 infrared-luminous z>6 quasars reveals cold molecular outflows ~300 km s−1 faster than in star-forming galaxies, with kinetic power beyond supernovae alone, implicating AGN energy injection.","keywords":["AGN feedback","molecular outflows","OH 119 micron","z>6 quasars","ALMA","reionization era","cold gas","quasar winds"],"falsifier":"Take one or two of these quasars and measure the outflow rate independently — for example with spatially resolved CO emission or additional OH transitions — and compare with the calibration of Eq. 2. If the independent rates come out several times lower, or if a $z\\sim5$ dusty galaxy with a confirmed AGN shows the same fast velocities once matched in AGN luminosity, the kinetic-power argument that AGN drive the winds would lose its footing.","tokens_in":35828,"feed_emoji":"💨","tokens_out":11916,"duration_ms":102553,"temperature":0.7,"pith_summary":"This paper tries to show that accreting supermassive black holes in the most luminous quasars of the reionization era are actively pushing cold molecular gas out of their host galaxies. Using new and archival ALMA observations of the OH 119 µm doublet, the authors detect blueshifted absorption in 8 of 11 infrared-luminous $z>6$ quasars, implying outflowing gas. The outflow velocities are on average about $300~\\mathrm{km\\,s^{-1}}$ faster than in roughly co-eval non-quasar dusty galaxies, while the outflow rates are modest, with mass-loading factors around 0.5. From the kinetic power, the paper argues that supernova feedback alone cannot drive these winds, so AGN energy injection is required. If correct, this is the first statistical evidence that quasar feedback operates on the coldest gas within the first billion years of cosmic history.","feed_headline":"Quasar winds outrun starburst winds by 300 km/s","feed_subtitle":"The winds move ~300 km/s faster than in star-forming galaxies, pointing to black-hole energy as the driver.","key_machinery":"The workhorse tracer is the ground-state OH $2\\Pi_{3/2}\\,J=3/2\\rightarrow5/2$ doublet at rest-frame 119 µm, observed in absorption against the quasar host's far-infrared continuum. Because OH 119 µm is optically thick even in the line wings, blueshifted absorption directly signals outflowing cold gas, and the absorption depth yields the covering fraction of the outflow. To convert observed equivalent widths into outflow properties, the analysis uses an empirical calibration (Eq. 2), $\\dot{M}_{\\rm out} = m\\,\\mathrm{EW}_{v<-200}\\,\\sqrt{L/10^{12}\\,L_\\odot} + b$, trained on low-redshift galaxies with CO- or OH-detected outflows, together with a time-averaged thin-shell geometry relating outflow rate, momentum, and kinetic power.","core_discovery":"The central claim is that cold molecular outflows are very common and unusually fast in the most infrared-luminous quasars at $z>6$, and that the fastest outflows carry enough kinetic energy to require AGN driving. The evidence is blueshifted OH 119 µm absorption seen in 8/11 quasars ($73\\%\\pm13\\%$), with median outflow velocity $v_{84}$ near $900~\\mathrm{km\\,s^{-1}}$, roughly $300~\\mathrm{km\\,s^{-1}}$ faster than the $z\\sim5$ non-quasar dusty-galaxy comparison sample. The implied outflow rates, roughly $300$–$1500~\\mathrm{M}_{\\odot}\\,\\mathrm{yr}^{-1}$, give a median mass-loading factor near 0.5, so the outflows are not extreme in rate but are fast. The outflow momentum is consistent with momentum-driven winds with boosts of order a few times $L/c$, but the kinetic power would require about 30% of available supernova energy to couple to cold gas, considered unrealistic, whereas about 0.1% of the AGN luminosity suffices. The paper therefore concludes that the AGN must at least partially drive the winds.","pith_inferences":["Inference: The absolute outflow rates rest on a low-redshift calibration, so an independent measurement of outflow rate in one or two $z>6$ quasars (e.g., via resolved CO or multiple OH transitions) would be the sharpest test; if the calibration fails, the magnitude of the required AGN energy changes accordingly.","Inference: The ~300 km/s velocity gap between quasars and non-quasars predicts a systematic trend: fainter $z>6$ quasars, with lower AGN luminosity, should show slower outflows; deeper observations of less luminous quasars could confirm this.","Inference: The smaller covering fractions measured for the quasar outflows suggest they may be more collimated than starburst-driven winds, which means part of the velocity difference could be geometric rather than energetic; inclination-selected samples would disentangle these.","Inference: The absence of broad [C II] wings in the fastest OH outflows calls into question past claims of molecular outflows inferred from [C II] stacking at $z>6$, suggesting OH should be treated as the primary cold-gas outflow tracer."],"forward_implications":["Cold molecular outflows are common (at least 73%) in the most infrared-luminous $z>6$ quasars, so AGN feedback on cold gas appears to be in place within the first billion years of cosmic time.","Outflow velocity correlates with AGN luminosity across four orders of magnitude and shows no similar trend with star-formation luminosity, tying the wind speed to black-hole accretion.","The outflow kinetic power, typically about 0.1% of $L_\\mathrm{AGN}$, is within theoretical AGN coupling limits but far above what supernovae can supply at realistic efficiency, so the AGN must contribute energy even if star formation helps launch the wind.","Momentum boosts of order $L/c$ place these outflows in the momentum-driven regime, so energy-conserving wind phases are not required to explain them.","OH 119 µm emission appears in roughly 82% of the quasars but in none of the $z\\sim5$ dusty galaxies, offering a kiloparsec-scale tracer of warm dense gas tied to AGN activity."],"supporting_citations":[{"why":"Supplies the z~5 non-quasar dusty-galaxy comparison sample and the absorption-based outflow classification scheme.","marker":"S20a"},{"why":"Provides the empirical outflow-rate calibration (Eq. 2) and the time-averaged thin-shell geometry used to derive outflow properties.","marker":"S20b"},{"why":"Contributes the OH 119 µm template for z>6 quasar outflows and the form of the empirical relation adopted in Eq. 2.","marker":"HC20"},{"why":"Provides OH observations of three quasars, one of which (P036+03) is reanalyzed here, and prior data for two more.","marker":"Butler et al. 2023b"},{"why":"Supplies the resolved OH study of J2054-0005 and the outflow properties re-estimated with this paper's methodology.","marker":"Salak et al. 2024"},{"why":"Extends the low-redshift training sample used to recalibrate the outflow-rate relation.","marker":"Lamperti et al. 2022"},{"why":"Sets the roughly 10% ceiling on supernova energy coupling to cold gas against which the required ~30% is judged unrealistic.","marker":"Kim & Ostriker 2015"},{"why":"Sets the 2-5% limit on AGN luminosity coupling to outflow kinetic power, used to argue the AGN has enough energy.","marker":"King & Pounds 2015"}],"fun_headline_variants":["ALMA sees AGN-driven outflows in 8 of 11 reionization-era quasars","Quasar winds 300 km/s faster than starbursts at cosmic dawn","Cold winds from earliest quasars point to black hole feedback","Reionization-era quasars blow 900 km/s molecular winds","AGN drive fast outflows in 73% of z>6 quasars"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The outflow rates come from an empirical calibration fit to low-redshift galaxies, and the paper assumes this relation keeps working when extrapolated to the far higher luminosities and different physical conditions of $z>6$ quasars; if that calibration fails, the derived outflow rates, kinetic powers, and the case for AGN driving would all change.","fun_headline_variants_meta":{"raw":{"variants":["ALMA sees AGN-driven outflows in 8 of 11 reionization-era quasars","Quasar winds 300 km/s faster than starbursts at cosmic dawn","Cold winds from earliest quasars point to black hole feedback","Reionization-era quasars blow 900 km/s molecular winds","AGN drive fast outflows in 73% of z>6 quasars"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00021,"raw_usage":{"total_tokens":1501,"prompt_tokens":1123,"completion_tokens":378,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":739,"completion_tokens_details":{"reasoning_tokens":276}},"tokens_in":739,"tokens_out":378,"duration_ms":4015,"temperature":1.0,"reasoning_tokens":276,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T19:53:10.380069+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take one or two of these quasars and measure the outflow rate independently — for example with spatially resolved CO emission or additional OH transitions — and compare with the calibration of Eq. 2. If the independent rates come out several times lower, or if a $z\\sim5$ dusty galaxy with a confirmed AGN shows the same fast velocities once matched in AGN luminosity, the kinetic-power argument that AGN drive the winds would lose its footing.","supporting_citations":[],"review_version":1}