{"id":"d04014d9-d119-4303-82ed-e4898fc26999","arxiv_id":"2608.12482","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":9,"one_line_summary":"Outflows in 17 CLASSY galaxies are phase-dependent, with loading factors that fall with stellar mass in line with FIRE-2 and maximum cool-gas velocities near 620 km/s in line with CGOLS.","lead":"This paper measures the gas outflows of 17 nearby star-forming galaxies using ultraviolet light from three forms of silicon, and finds that cool gas moves faster and stays closer in than warm gas. It provides a direct observational check of two widely used outflow simulations, FIRE-2 and CGOLS.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Ionization-completeness assumption in Eq. 2 (all Si as SiII-III-IV) scales every Mdot_H and loading factor; if the missing Si fraction varies with mass or age, the claimed stellar-mass scaling and FIRE-2/CGOLS agreement are not robust.","rationale":"The paper's main kinematic and compositional results (SiII versus SiIII/SiIV differences in alpha, psi, fc, RW, and v_rho) come directly from the SALT fits of the three silicon transitions and do not depend on the hydrogen conversion; those appear well-supported by the model validation in Carr et al. (2023, 2025a). The load-bearing weakness is the conversion from silicon to hydrogen in Eq. 2. The reader's weakest_assumption identifies this precisely: the paper assumes all silicon is in the three observed ionization stages and scales by a single galactic metallicity and solar silicon abundance. This assumption multiplies every Mdot_H and therefore every loading factor in Table 1 and Figure 4. The scaling with stellar mass and the comparison with FIRE-2/CGOLS would be invalidated if the missing silicon fraction (higher ionization stages, dust) is large and varies systematically across the sample. Such variation is physically plausible because outflow ionization depends on the same young stellar populations that drive the age and mass trends. The paper acknowledges related limitations (hot phase not traced, footnote on SFR degeneracy), which supports CONDITIONAL rather than REJECT. A photoionization-model test with CLASSY SEDs would settle whether the correction is constant or mass-dependent. We therefore leave the reader's verdict unchanged.","tokens_in":20632,"tokens_out":6365,"duration_ms":62904,"concrete_test":"Run Cloudy photoionization calculations for each CLASSY galaxy using the SED from Parker et al. (2026) and the SALT-derived density, radius, and ionizing conditions to predict the full silicon ionization balance, including SiV/SiVI and dust depletion. Use the predicted total Si/H correction to recompute Mdot_H in Eq. 2, then re-derive eta_H, eta_p, and eta_E and their slopes against stellar mass, plus the age-split N_SiII/N_Si trend. If the mass-loading slope shifts by more than its quoted uncertainty, or loses significance (Spearman p > 0.05), the central scaling and simulation-agreement claims depend on the unmeasured ionization-completeness assumption.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Equation 2 of Section 2 converts SALT-fitted silicon outflow rates to hydrogen rates via Mdot_H = Mdot_Si / (Z_Galaxy * Z_Si_sun), with Mdot_Si summed over only SiII, SiIII, and SiIV. The paper states that it is 'assuming that silicon is only in the three observed ionization states,' justified by simulations (C. A. Carr et al. 2025c), but this is not measured in the CLASSY data. This conversion is the multiplicative normalization for every Mdot_H, eta_H, eta_p, and eta_E in Table 1 and Figure 4. If a non-negligible silicon fraction resides in higher ionization stages (e.g., SiV, SiVI) or in dust, and if that fraction varies with stellar mass, metallicity, or stellar age, then the reported inverse scaling of loading factors with stellar mass, and the agreement with FIRE-2 and CGOLS, could be partly or wholly artifacts of the adopted correction rather than intrinsic outflow physics. The age-dependent Mdot_H trend is similarly affected because it uses the same conversion. The assumption is adopted, not verified; the paper's own caveat that the hottest gas is not traced by the UV silicon lines (Section 4) underlines how large the missing phase could be.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript analyzes 17 CLASSY galaxies previously modeled with the SALT radiative transfer code by Huberty et al. (2024). It compares SALT-derived properties of SiII, SiIII, and SiIV outflows, converts silicon outflow rates to hydrogen rates via Eq. (2), and uses these to derive mass, momentum, and energy loading factors. The central claims are that the cool SiII-traced gas has distinct kinematics and spatial distribution (faster, less porous, smaller terminal radius) relative to the warmer SiIII/SiIV gas; that loading factors decrease with stellar mass and match FIRE-2 for mass and momentum; that the average terminal velocity of about 620 km/s matches CGOLS; and that outflows from young, single-burst stellar populations have cooler-gas-dominated columns and radially decreasing mass outflow rates.","tokens_in":20833,"tokens_out":5038,"duration_ms":46042,"significance":"If correct, the paper would strengthen the empirical case that star-formation-driven outflow properties depend on gas phase and stellar age, and it would provide observational support for FIRE-2 and CGOLS feedback prescriptions. The work builds on a well-defined CLASSY sample and uses a radiative transfer model (SALT) that has been validated against RASCAS and simulation outputs in earlier papers. The paper also reports 16th/84th percentile uncertainties and places the results alongside LzLCS measurements. The main caveats are that the hydrogen normalization rests on an ionization-completeness assumption that is not directly verified in the CLASSY data, and that the apparent agreement with simulations compares fitted quantities rather than independent predictions; these caveats should be addressed before the stronger conclusions can be accepted.","major_comments":[{"comment":"The conversion Mdot_H = Mdot_Si/(Z_Galaxy * Z_Si_sun) assumes that all silicon in the outflow is in SiII, SiIII, or SiIV. This is a multiplicative normalization for every mass, momentum, and energy loading factor in Table 1 and Figure 4, so any silicon that resides in other ionization stages or in dust will shift all derived quantities. If the missing fraction varies with stellar mass, metallicity, or age, the reported inverse scaling of eta_H, eta_p, and eta_E with stellar mass, and the agreement with FIRE-2 and CGOLS, could be partly artifacts of the assumption. The paper's own caveat in Section 4 that the hot phase is not traced by the UV silicon lines makes this concern concrete. Please add a robustness test, for example using photoionization calculations to estimate the unobserved SiV/SiVI fraction across the sampled stellar mass and metallicity range, or demonstrate what mass-dependent missing fraction would be required to erase the trends. At minimum, the abstract and conclusions should state that the loading-factor scaling is conditional on this ionization-completeness assumption.","section":"Section 2, Eq. (2)"},{"comment":"The comparison between CLASSY and CGOLS is framed as agreement in maximum velocity, but v_w is a free parameter of the SALT model fitted to the spectra (Section 2), not a measured quantity independent of the model. The same applies to the loading factors in Figure 4, which are functions of the fitted SALT parameters (tau_0, alpha, psi, f_c, and the power-law indices). Thus the statements that the measurements agree with FIRE-2 and CGOLS are consistency checks between the adopted outflow model and simulation outputs, not validation of the simulations against independent observables. Please revise the abstract and Section 4 to use 'consistent with' language and, where possible, identify which combinations of SALT parameters drive the apparent agreement.","section":"Section 4.2 and Figure 5"},{"comment":"The claim that single-burst versus multi-burst galaxies differ in N_SiII/N_Si rests on a two-sample KS test with p = 0.047 for a sample of 17 galaxies, after dividing the sample into two groups using the Parker et al. (2026) classification. Given the small sample and the multiple ways the population could be split, this p-value alone is weak evidence. Please report the KS statistic (already given as 0.639), the effect size, the number of galaxies in each group, and a bootstrap or permutation p-value. If the evidence does not survive these tests, the age-composition claim should be downgraded.","section":"Section 3.2 and Figure 2"},{"comment":"The radial trend in Mdot_H(r), decreasing for young sources and increasing or constant for older sources, is presented largely by visual inspection of the right panel of Figure 5. With large uncertainties and SALT's known tendency to overestimate Mdot(R_W) as cited from Carr et al. (2023), this age-dependent radial claim needs a quantitative test, such as fitting the slope of Mdot(r) per galaxy and comparing the posterior distributions of slopes between young and old groups. As written, the conclusion in Section 5 that young outflows have Mdot decreasing with radius is not supported by a stated statistical measure.","section":"Section 4.2 and Figure 5 (right panel)"}],"minor_comments":[{"comment":"The ionization-state labels are typeset inconsistently (for example, 'Siii', 'Siiii', and 'Siiv' appear in the abstract and text), which makes the phase labels difficult to read. Please use unambiguous SiII, SiIII, and SiIV notation throughout.","section":"Abstract and throughout"},{"comment":"There are typographical errors such as 'realtive' and 'supperbubbles' that should be corrected during copyediting.","section":"Section 2"},{"comment":"The statement that the cooler gas traces faster outflows is based on the density-weighted velocity v_rho; please clarify that this does not imply all cool gas is faster than all warm gas, as the text already notes later in the same paragraph.","section":"Section 3.1"},{"comment":"The self-correlation between eta_H and SFR is acknowledged in a footnote, but the text should more prominently state that part of the anti-correlation in the upper-right panel of Figure 4 is expected because eta_H is defined as Mdot/SFR and Mdot correlates with SFR.","section":"Section 4 and footnote 19"},{"comment":"The comparison with FIRE-2 is between observed kinetic-energy loading and simulated total-energy loading; the text mentions this, but the figure caption should explicitly repeat it so the two quantities are not mistaken for the same definition.","section":"Figure 4, lower right panel"}],"recommendation":"major_revision","confidential_remarks":"The ion-completeness assumption in Eq. (2) is the single most load-bearing point; the editor may wish to ask the authors to provide the SALT posterior distributions or a companion reproducibility note when the revision is submitted. The paper is otherwise a solid CLASSY series contribution that would be a good fit after the robustness tests and statistical rephrasing are completed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things about this paper. First, it's a credible, well-organized follow-up to H24 that teases out genuinely new physical trends from the existing SALT fits: the cool SiII phase moves faster, is denser, and terminates closer in than the warm SiIII/SiIV phases, and the mass-loading factors fall with stellar mass and line up with FIRE-2. Second, the absolute numbers – the loading factors and the 620 km/s terminal velocity – are outputs of the SALT model, not independent measurements, and the mass scale carries a normalization assumption that all silicon lives in SiII–SiIV. That last point is the one to probe in review.\n\nWhat the paper does well: it is explicit about its methods, reports 16th/84th percentiles, flags its own caveats (including the SFR self-correlation in footnote 19), and compares against two sets of hydrodynamic simulations and the LzLCS sample. The phase-dependent kinematic differences (Figure 1) and the age trends (Figures 2–3) are new relative to H24 and give the field concrete patterns to explain. The paper does not oversell the KS test result; it calls it 'evidence' at p=0.047, which is appropriate for 17 galaxies.\n\nThe soft spots, in proportion. The ionization-completeness assumption in Eq. 2 is load-bearing: if a non-negligible silicon fraction sits in higher stages or dust, every Mdot_H and loading factor shifts by a constant (or worse, a mass/age-dependent constant). The paper cites Carr+25c simulations as support, but that is not a measurement in these CLASSY data. This needs to be either quantitatively justified with ionization modeling for this sample or explicitly framed as a systematic floor. The simulation agreement is also a consistency check rather than a validation, because the velocity and loading-factor points are fitted, not predicted. The paper is honest about this implicitly, but a referee should ask for a sentence that says it directly. Finally, the Mdot(r) radial profiles in Figure 5 have large uncertainties, and the authors note SALT tends to overestimate Mdot at R_W; the 'decreases with radius' claim for young galaxies is plausible but rests on a handful of objects.\n\nBottom line: this is a serious, useful paper for anyone working on galactic outflows, loading factors, or the CLASSY survey. It earns a serious referee. I'd send it to review with a request to address the silicon ionization completeness and to moderate the language on simulation agreement. It is not a desk reject.","headline":"Careful analysis of 17 CLASSY outflows with useful phase-resolved results, but the absolute loading factors and simulation agreement rest on an unmeasured ionization-completeness assumption.","tokens_in":21567,"tokens_out":2527,"would_cite":true,"duration_ms":22205,"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":"In 17 nearby star-forming galaxies, cooler silicon-traced wind gas moves faster, is less porous, and terminates closer to the galaxy than warmer gas, and mass, momentum, and energy loading factors fall with stellar mass.","keywords":["galactic outflows","star-forming galaxies","ultraviolet absorption spectroscopy","radiative transfer modeling","mass loading factor","stellar populations","circumgalactic medium","galaxy feedback"],"falsifier":"In one of the 17 galaxies, obtain spectra covering additional silicon ionization stages, such as Si V and Si VI, and estimate silicon depletion onto dust: if the summed Si II + Si III + Si IV column accounts for substantially less than the total silicon expected from the galaxy metallicity, then the hydrogen outflow rates and all loading factors are systematically low by that missing fraction. Likewise, spatially resolved integral-field observations that resolve the outflow into multiple distinct cones, or a clumpy medium with strongly varying porosity, would invalidate the single-bicone geometry assumption.","tokens_in":20329,"feed_emoji":"💨","tokens_out":9235,"duration_ms":78458,"temperature":0.7,"pith_summary":"This paper tries to establish that galactic outflows are genuinely multiphase in their spatial and kinematic structure, and that their power depends on galaxy mass and stellar age. Using radiative transfer fits to ultraviolet silicon absorption lines in 17 nearby, highly star-forming galaxies, it finds that the cool gas traced by Si II flows faster, is less porous, and terminates at smaller radius than the warmer gas traced by Si III and Si IV. It then converts those line fits into hydrogen mass, momentum, and kinetic-energy outflow rates and reports that the corresponding loading factors decline as stellar mass increases, in agreement with state-of-the-art hydrodynamic simulations for mass and momentum, while kinetic-energy loading sits far below simulated total energy budgets. Finally, it links outflow phase to stellar age: young (<5 Myr) stellar populations drive cool-gas-dominated, high-mass outflows whose rates decrease with radius, whereas older systems are more ionized and show flatter or rising outflow profiles. If right, these results validate the feedback prescriptions used in galaxy simulations and imply that low-mass, young starbursts are the most efficient at ejecting gas.","feed_headline":"Cool gas runs faster and stops sooner in galaxy winds","feed_subtitle":"UV spectra of 17 starbursts show outflow power falls as stellar mass and age rise.","key_machinery":"The central object is the Semi-Analytical Line Transfer (SALT) model, a radiative transfer code that fits ultraviolet resonance-line profiles under the Sobolev approximation, including resonant absorption and re-emission infilling. From the Si II, Si III, and Si IV profiles SALT returns the outflow opening angle $\\alpha$, orientation angle $\\psi$, porosity $f_c$, optical depth $\\tau_0$, power-law velocity and density indices $\\gamma$ and $\\delta$, launch radius $R_{\\rm SF}$ (set to the NUV half-light radius), and terminal radius $R_W$. The paper converts the summed silicon outflow rates to hydrogen rates with Eq. (2), dividing by galaxy metallicity and the solar silicon fraction under the assumption that all silicon is in the three observed ions. These ingredients give radial profiles of $\\dot M_H$, momentum flux, and kinetic-energy flux, from which the mass, momentum, and energy loading factors are computed at $R_{\\rm SF}$.","core_discovery":"The paper argues that the outflows of 17 local, actively star-forming galaxies are not single-phase winds: the cool gas traced by Si II flows faster (average maximum cool-gas velocity about 620 km/s), is less porous, terminates at a smaller radius, and carries a larger share of the mass outflow than the warmer gas traced by Si III and Si IV, whose angular geometries correlate closely with each other but not with Si II. From the SALT fits, it derives hydrogen mass, momentum, and kinetic-energy outflow rates at the half-light radius and finds that the mass, momentum, and energy loading factors (outflow rate normalized by star-formation rate, supernova momentum deposition, and supernova energy deposition) all decrease with stellar mass, with mass and momentum loading in agreement with high-resolution galaxy-formation simulations. The paper also connects outflow phase to stellar age: galaxies with young light-weighted populations (<5 Myr) tend to be dominated by cool Si II, have higher mass outflow rates, and show mass outflow rates that decrease with radius, whereas older systems are more ionized and show flat or rising radial mass outflow profiles.","pith_inferences":["If the assumed ionization completeness is wrong, the absolute loading factors shift by one common factor, so the reported scaling with stellar mass could survive even though the normalization would not; a direct measurement of the silicon ionization balance would separate these two cases.","The age sequence suggests a two-stage wind lifecycle that is testable at fixed stellar mass: young bursts should show cool, compact, decelerating outflows, while older bursts should show more ionized, extended, flat or rising mass profiles; comparing galaxies across burst ages with the same mass and metallicity would test this.","Because the ultraviolet lines miss the hot phase where simulations put most of the energy, the kinetic-energy loading deficit may be partly an observational aperture effect; adding X-ray absorption or emission measurements of the same outflows would show whether the missing energy is in a hot component or radiated away.","Extending the same radiative transfer analysis to additional ions, such as C II, C IV, and O VI, and to spatially resolved or lensed high-redshift analogues, would test whether the mass-loading decline with stellar mass is universal."],"forward_implications":["More massive galaxies have lower mass, momentum, and energy loading factors, so low-mass starbursts return a larger fraction of their star-formation budget to the circumgalactic medium per unit star formation.","Agreement between observed mass and momentum loading factors and hydrodynamic simulations validates the feedback prescriptions in those simulations for cool and warm outflows, while the much smaller observed kinetic-energy loading factors than simulated total energy loading indicates that the hot phase is not captured by ultraviolet silicon lines.","Because cool Si II gas is faster, less porous, and shorter-lived in radius than warmer gas, single-ion estimates of outflow rates are systematically biased, and multi-phase measurements are needed for total mass and energy budgets.","Outflows from young (<5 Myr) stellar populations are cooler, more massive, and decelerate with radius, whereas older (>10 Myr) populations drive more ionized outflows whose mass outflow rates stay flat or rise, tying outflow phase to the onset of supernovae.","The average maximum cool-gas wind velocity of about 620 km/s matches hydrodynamic simulations of starburst-driven winds, supporting the simulated relation between wind speed and galaxy potential."],"supporting_citations":[{"why":"Supplies the SALT radiative transfer fits and parameter distributions for the 17 outflows analyzed here.","marker":"M. Huberty et al. (2024)"},{"why":"Introduces the SALT model and its semi-analytical formalism for fitting outflow line profiles.","marker":"C. Carr et al. (2023)"},{"why":"Validates SALT parameter recovery against the RASCAS numerical radiation transfer code and simulation mock spectra.","marker":"C. A. Carr et al. (2025a)"},{"why":"Provides the simulation predictions for mass, momentum, and energy loading factors that the measurements are compared against.","marker":"V. Pandya et al. (2021)"},{"why":"Supplies the CGOLS IV hydrodynamic simulations of cool outflow velocity profiles used as the comparison standard.","marker":"E. E. Schneider et al. (2020)"},{"why":"Supplies the CGOLS V simulations with a cluster mass function, the second velocity-profile comparison standard.","marker":"E. E. Schneider & S. A. Mao (2024)"},{"why":"Provides the light-weighted stellar ages and single-burst versus multi-burst classifications that drive the age-dependent results.","marker":"K. S. Parker et al. (2026)"},{"why":"Provides SALT-based loading factors for more massive LzLCS galaxies, extending the stellar-mass trend.","marker":"C. A. Carr et al. (2025c)"},{"why":"Supplies the supernova momentum deposition rate $p_*$ used to normalize momentum loading.","marker":"N. Murray et al. (2005)"},{"why":"Supplies the supernova kinetic-energy deposition rate $E_*$ used to define the energy loading factor.","marker":"C. Leitherer et al. (1999)"}],"fun_headline_variants":["Cool winds outrun warm gas in starbursts","Fast Si II winds cap at 620 km/s in galaxies","Young starbursts favor cool, fast outflows","Outflow power drops as galaxy mass rises","Starburst winds: cool gas leads, then fades"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The mass, momentum, and energy outflow rates all assume that every silicon atom in the wind is in one of the three observed ionization states (Si II, Si III, or Si IV) and that the wind is a single power-law bicone with one porosity; if much silicon is in other ions or in dust, or if the geometry is more complicated, every loading factor shifts by a large factor.","fun_headline_variants_meta":{"raw":{"variants":["Cool winds outrun warm gas in starbursts","Fast Si II winds cap at 620 km/s in galaxies","Young starbursts favor cool, fast outflows","Outflow power drops as galaxy mass rises","Starburst winds: cool gas leads, then fades"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000165,"raw_usage":{"total_tokens":1288,"prompt_tokens":1023,"completion_tokens":265,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":639,"completion_tokens_details":{"reasoning_tokens":185}},"tokens_in":639,"tokens_out":265,"duration_ms":3155,"temperature":1.0,"reasoning_tokens":185,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T00:07:52.280992+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"In one of the 17 galaxies, obtain spectra covering additional silicon ionization stages, such as Si V and Si VI, and estimate silicon depletion onto dust: if the summed Si II + Si III + Si IV column accounts for substantially less than the total silicon expected from the galaxy metallicity, then the hydrogen outflow rates and all loading factors are systematically low by that missing fraction. Likewise, spatially resolved integral-field observations that resolve the outflow into multiple distinct cones, or a clumpy medium with strongly varying porosity, would invalidate the single-bicone geometry assumption.","supporting_citations":[],"review_version":1}