{"id":"04aad92c-2596-4868-b15b-edbd2050e8f7","arxiv_id":"2504.19897","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":10,"one_line_summary":"An evolutionary dust echo model with a day-timescale evaporation cavity and a wide grain-temperature distribution explains AT 2018cow's near-infrared excess and implies a larger, more distant dust shell than earlier fits.","lead":"AT 2018cow's near-infrared excess can be explained as heat re-emitted by surrounding dust as the explosion's light evaporates it. The paper models this evolving dust echo and finds the dust shell is larger and farther out than previous single-temperature fits suggested.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The derived dust mass and CSM mass scale directly with the assumed r^-3 density profile and ISM grain-size index (Eqs. 1, 23), neither of which is constrained for AT 2018cow; changing these assumptions can shift the headline mass/distance conclusions.","rationale":"The reader's weakest assumption correctly identifies the adopted r^-3 density profile and ISM grain-size distribution as the least secure link in the argument. I agree because Eq. (23) makes the scaling explicit: the dust mass, and hence the CSM mass and mass-loss-rate estimates, are proportional to n_d R*^3 and to the grain-size normalization, so the numerical conclusions inherit these priors. The same profile dependence enters the echo integral through the n(r) r^2 weighting, so the fitted outer radius is not independent of the assumed slope. This is not an internal inconsistency: the physics of time-dependent grain heating and evaporation is standard, and the authors honestly flag the ISM assumption, the z-band residual, and the need for radio constraints. The concern is therefore a condition on the quantitative claim, not a demonstration that the model is wrong. The suggested refit with a free or varied density index would settle whether the mass/distance comparison is robust or merely a consequence of the chosen parameterization. Because the reader's verdict is already CONDITIONAL and this concern reinforces that condition rather than overturning the model, the verdict should remain UNCHANGED.","tokens_in":14803,"tokens_out":16242,"duration_ms":186233,"concrete_test":"Re-run the same MCMC fit with the density profile generalized to n(r) ∝ r^{-gamma}, first with gamma as a free parameter (e.g., flat prior in [1,4]) and then with gamma fixed to 2 and to 4, keeping all other priors and the two-step fitting procedure unchanged. Compare the resulting posteriors for R_out and for the dust mass M_d from Eq. (23). If the median M_d shifts by more than a factor of ~3 across these gamma choices, or if gamma itself is essentially unconstrained by the NIR data, then the headline mass and distance conclusions are artifacts of the assumed r^-3 profile rather than measurements.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim—dust shell mass ~3e-4 M_sun and distance larger than direct single-temperature fits—depends on assumptions that are not tested by the data. The shell density is fixed to n(r) = n_d (r/R*)^-3 with R* = 10^16 cm (Eq. 1), the index -3 adopted from Bright et al. (2022) rather than measured for AT 2018cow, and the grain-size index is fixed to the ISM value alpha = 3.5 with carbonaceous grains over 0.001-10 micron. Equation (23) shows M_d ∝ n_d R*^3 (a_max/a_min)^(1-alpha) ln(R_out/R_ev), so the derived dust mass, and the inferred CSM mass (~3e-2 M_sun via the assumed 100:1 gas-to-dust ratio), inherit these choices directly. The fitted R_out is also profile-dependent because the echo integral weights shells by n(r) r^2 ∝ r^-1, so changing the slope changes which radii dominate the observed flux. The authors explicitly concede that the ISM grain properties may not apply to FBOTs. Compounding this, the NIR data are sparse—a few JHK epochs plus two private WISE points, one explicitly tentative—so the reported 0.00-0.01 dex posterior uncertainties on n_d and R_out likely reflect conditioning on the assumed profile rather than genuine constraints on the physical dust distribution. If the true density slope is steeper or shallower, or if the grain size distribution is truncated at larger sizes, the 'much larger than direct fitting' conclusion could change substantially.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a time-dependent model of infrared dust echo emission from fast blue optical transients (FBOTs), in which the transient's radiation evaporates the innermost dust, producing a dust-free cavity and a transition zone where only larger grains survive, and the surviving grains are continuously heated and cool as the FBOT luminosity evolves. The model is applied to AT 2018cow: an Arnett/magnetar-powered FBOT light curve plus a dust shell with n(r) ∝ r^-3 and a graphite power-law grain size distribution are used to fit the optical-to-NIR light curves and seven epochs of spectra with a single parameter set. The authors find a dust shell mass of about 3×10^-4 M_sun, an outer radius of about 5×10^17 cm, and an implied CSM mass of about 3×10^-2 M_sun, and argue that these quantities are substantially larger than those obtained by independent single-temperature blackbody fits to individual NIR spectra, whose parameters vary unphysically with time.","tokens_in":15103,"tokens_out":10463,"duration_ms":113569,"significance":"If the model is accepted, it is a useful advance over epoch-by-epoch spectral fitting: it provides a physically self-consistent explanation of the NIR excess of AT 2018cow, avoids time-varying dust mass and radius, and predicts a distinctive evolution of the dust temperature distribution. The model is transparent and the underlying grain heating, evaporation, and echo integrals are standard. The central quantitative conclusions, however, remain contingent on unconstrained assumptions about the density profile and grain properties, so the headline claim that the dust mass and distance are much larger than found by direct fitting should be read as conditional on those assumptions. The comparison with direct single-temperature fits is heuristically valuable but needs to be made more quantitative to be fully convincing.","major_comments":[{"comment":"In §2 the dust shell is fixed to n(r) = n_d (r/R*)^-3 with R* = 10^16 cm (Eq. 1) and to graphite grains with a^-3.5 from 0.001 to 10 microns; the authors themselves note that these ISM values may not truly be applicable to FBOTs. The headline dust mass is directly tied to these choices: Eq. (23) gives M_d ∝ n_d R*^3 a_max^3 (a_max/a_min)^(1-alpha) ln(R_out/R_ev,min), and the echo integral in Eq. (16) weights each shell by n(r) r, so a steeper or shallower density slope changes which radii dominate the observed flux. A change of the density index from -3 to -2 or -4, or a truncated grain size distribution, can plausibly shift the fitted n_d and R_out by much more than the quoted ±0.01 dex posterior uncertainties. I therefore ask for a quantitative sensitivity study over plausible density slopes (e.g., -2, -3, -4), grain-size power-law indices in the stated 2.0-4.5 range, and grain compositions, reporting how M_d, R_out, and the claimed factor-of-several increase over direct fits change. Without this, the central quantitative claim is not robust.","section":"§2 and Eq. (23)"},{"comment":"The posterior uncertainties in Table 1 (log10 n_d = -2.37^{+0.00}_{-0.01}, log10 R_out = 17.77^{+0.01}_{-0.01}) are conditional on all the fixed profile and grain choices, and the NIR data set consists of only a few JHK epochs plus two private NEOWISE points, one of which is explicitly tentative. These errors should not be presented as constraints on the physical dust distribution; they measure only the narrowest part of the model uncertainty. The paper should either state this limitation prominently in the results, or preferably quote parameter ranges obtained from the sensitivity analysis requested above. In addition, the two-step fitting procedure (FBOT parameters fixed from optical bands before the dust echo is fit to z/JHK/W) means that the dust parameters inherit any bias in the optical-only FBOT fit; a simultaneous fit or a demonstration that the optical parameters are unchanged when the echo is included would remove this concern.","section":"§3.3 and Table 1"},{"comment":"The comparison with direct empirical fits is weakened by the fact that the comparison model is deliberately a single-temperature blackbody, so the time-varying T_dust, M_dust, and R_dust in Fig. 7 are a predicted artifact of using the wrong spectral model on a multi-temperature distribution. To make the claim that the evolutionary echo model gives larger mass and distance quantitative, I ask the authors to apply their single-temperature fitting procedure to mock spectra generated from their own best-fit model and demonstrate that it reproduces the same order-of-magnitude underestimates and time evolution shown in Fig. 7. This would isolate the effect of the fitting method from the effect of the assumed density profile and would strengthen the paper's main contrast.","section":"§3.4 and Fig. 7"}],"minor_comments":[{"comment":"In Eq. (23), the first integral's radial limits are printed as R_out and R_ev,min; the lower limit should be R_ev,min.","section":"Eq. (23)"},{"comment":"The W1 and W2 photometry come from private communication, and the W1 point is tentative; please publish these measurements in a machine-readable table, including exact epochs, uncertainties, and upper limits, so that the fit can be reproduced.","section":"§3.3 and footnote 2"},{"comment":"The direct spectral fitting method is not fully specified; please state explicitly whether each spectrum is fit with a two-component model (a blue photospheric blackbody plus a single-temperature dust blackbody), and how the blue component is treated, so that the comparison with previous literature is reproducible.","section":"§3.4"},{"comment":"The z-band residual noted in the text is not quantified; a residual panel or a root-mean-square deviation for each band would help the reader judge the quality of the multi-band fit.","section":"§3.3 and Fig. 2"},{"comment":"The text around Eq. (8) discusses the absorption depth of the dust shell before dust destruction; please clarify that this estimate applies to the unperturbed shell and that after evaporation the effective inner edge is R_ev, not the original R_in, which the model then treats as irrelevant.","section":"§2 and Eq. (8)"}],"recommendation":"major_revision","confidential_remarks":"The paper is an honest forward model and the numerical machinery is a useful contribution, but the headline quantitative claims need to be reframed as conditional on assumed dust and CSM properties. I would not reject because the model framework can be salvaged with a sensitivity analysis and a more careful presentation of the comparison with direct fits. The authors' own admission that ISM grain properties may not apply to FBOTs should prompt them to explore at least a limited set of alternatives before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the paper is worth taking seriously. It builds an evolutionary dust echo model for FBOTs with three genuinely new pieces—time-dependent grain evaporation, a size-dependent transition zone, and persistent heating by the late transient emission—and shows it can explain the JHK+WISE excess of AT 2018cow with one parameter set. That is a real advance over the first-light echo models in Metzger & Perley (2023) and Tuna et al. (2025). The comparison in Fig. 7 is the most useful part: it shows clearly that single-temperature blackbody fits to the NIR spectra force the dust mass and radius to vary wildly with time, which is unphysical, while the evolutionary model holds them constant.\n\nThe grain physics is standard and the echo integral is correct. The authors are honest about the residuals and about what they did not test. I buy the qualitative story: the NIR excess is thermal dust echo from a shell with a dust-free cavity created by the flash.\n\nThe soft spots are real but not fatal. The r^-3 density profile (Eq. 1) is adopted from Bright et al. (2022), not measured for AT 2018cow, and Eq. (23) makes M_d ∝ n_d R*^3 ln(R_out/R_ev) given the fixed grain-size index alpha=3.5. The authors concede this. The reported 0.00–0.01 dex uncertainties on n_d and R_out look far too sharp for the sparse data—a handful of JHK epochs plus two private WISE points, one of which is tentative and only ~2.5 sigma. Those errors are conditioned on the fixed profile and grain properties, not on the real freedom in the model. So the headline numbers (dust mass ~3e-4 M_sun, shell radius ~5e17 cm, CSM mass ~3e-2 M_sun) should be read as order-of-magnitude estimates under specific assumptions, not precise measurements.\n\nA referee should ask for a sensitivity test over the density slope and grain-size index, and for a posterior that includes model uncertainty. Those changes would not overturn the core result, but they would make the quantitative claims defensible. The audience is transient astronomers working on IR echoes and CSM diagnostics; it is also a good case study for forward-modeling with honest limitations. I would send it to peer review and bring it to my reading group. Recommend: accept with minor-to-moderate revisions, with the assumption-dependence and error bars addressed.","headline":"A solid evolutionary echo model for AT 2018cow's NIR excess whose headline dust mass and distance inherit an assumed density profile and grain-size distribution, making the quantitative claims softer than the reported posteriors suggest.","tokens_in":15733,"tokens_out":3581,"would_cite":true,"duration_ms":33976,"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":"The near-infrared excess of the fast blue optical transient AT 2018cow is a thermal dust echo from a pre-existing circumstellar shell that the flash largely evaporates within about a day.","keywords":["astrophysical dust processes","circumstellar dust","carbonaceous grains","light curves","infrared astronomy","supernovae","fast blue optical transients","dust echo"],"falsifier":"Take continued JHK and WISE photometry of AT 2018cow out to 100 to 300 days after explosion: the model predicts a light-crossing plateau followed by a roughly $t^{-3}$ decline with smoothly cooling dust temperatures, so any observed infrared light curve that stays flat, rebrightens, or shows spectral features inconsistent with cooling carbonaceous grains would contradict the echo interpretation.","tokens_in":14486,"feed_emoji":"🌌","tokens_out":11404,"duration_ms":111700,"temperature":0.7,"pith_summary":"The paper's central claim is that the near-infrared excess of AT 2018cow is thermal emission from a pre-existing dusty circumstellar shell that is heated and partially destroyed by the transient's own flash. The authors build a time-dependent model in which the flash evaporates a dust-free cavity within about a day, leaving behind only larger grains close to the cavity and a size distribution that varies with radius, and in which every surviving grain is continuously heated by the fading transient. With a single parameter set, this evolutionary echo model reproduces the multi-band light curves and spectra from about 3 to 38 days, something the authors argue a direct single-temperature fit of the infrared spectra cannot do consistently. The corrected dust shell has a mass of roughly $3\\times10^{-4}\\,M_\\odot$ and lies at distances up to about $5\\times10^{17}$ cm, much larger than empirical fits suggest, with all grain temperatures broadly distributed and evolving in time. A sympathetic reader would care because the result turns the infrared excess into a probe of the progenitor's mass-loss history and circumstellar environment.","feed_headline":"Dust echo explains AT 2018cow's near-infrared excess","feed_subtitle":"A day-long flash clears a dust cavity, and the surviving grains' evolving glow matches the transient's multi-wavelength light curves.","key_machinery":"The paper's central mechanism is a time-dependent dust-heating and evaporation calculation. Each grain balances the heating from the evolving photospheric radiation field against cooling by thermal emission and mass loss by sublimation, with evaporation rate $\\dot a=-\\zeta\\exp(-E_b/k_\\mathrm{B}T_d)$; this sets the dust-free cavity radius by energy conservation rather than by an optical-depth criterion. Outside the cavity, the minimum surviving grain size grows with radius because small grains are destroyed first, and the echo luminosity at observer time $t$ is the light-travel-delayed integral over that radius- and size-dependent population. The calculation turns a single evolving input light curve into a self-consistent, multi-temperature, time-dependent dust echo.","core_discovery":"On the paper's own terms, the near-infrared excess of AT 2018cow is an echo of the transient's radiation reprocessed by a pre-existing dusty shell. The flash evaporates every grain within roughly $2\\times10^{16}$ cm in less than a day, creating a dust-free cavity; further out, smaller grains are destroyed while larger ones survive, so the minimum surviving grain size grows with distance from the source. Each surviving grain is heated and cooled continuously, and because grains of different sizes sit at different temperatures, the echo is a multi-temperature blackbody that evolves as the transient fades. With one parameter set, the model reproduces the optical-to-NIR light curves and the spectra from 3 to 38 days, whereas independent single-temperature fits require dust mass and radius that change with time in ways the authors argue are unphysical. The fitted shell has a mass of roughly $3\\times10^{-4}\\,M_\\odot$, extends to about $5\\times10^{17}$ cm, and implies a total CSM mass of order $3\\times10^{-2}\\,M_\\odot$ and a mass-loss rate near $10^{-6}$ to $10^{-4}\\,M_\\odot\\,\\mathrm{yr}^{-1}$.","pith_inferences":["If the evolutionary echo picture is right, the first few days of NIR colour evolution in a future FBOT should show a systematic reddening as the smallest, hottest grains near the cavity are destroyed first; rapid-cadence JHK photometry could test this directly.","The high derived dust mass implies a dense CSM that should also absorb or scatter radio emission at early times; existing radio upper limits for AT 2018cow could therefore already provide an independent check on the fitted dust density.","The same evaporation-plus-echo calculation should apply to other luminous transients that show infrared echoes, such as tidal disruption events and superluminous supernovae, potentially revealing that their dust masses are systematically underestimated by single-temperature fits."],"forward_implications":["The near-infrared excess of AT 2018cow can be explained without invoking nonthermal emission: it is thermal radiation from dust heated by the transient, so the excess becomes a direct probe of the circumstellar environment.","Previous single-epoch spectral fits underestimated the dust mass and its distance; the self-consistent echo fit places the shell at about $5\\times10^{17}$ cm with a mass near $3\\times10^{-4}\\,M_\\odot$.","The dust temperature in the echo is neither unique nor constant: at any single epoch small grains near the cavity are hotter, and all temperatures decline as the FBOT fades.","The inferred CSM mass ($\\sim3\\times10^{-2}\\,M_\\odot$) and mass-loss rate ($10^{-6}$ to $10^{-4}\\,M_\\odot\\,\\mathrm{yr}^{-1}$) favour an evolved massive progenitor, possibly an ultra-stripped star in a binary, and motivate combined dust and radio modelling.","Dust echoes in future nearby FBOTs can be used to map the geometry and mass-loss history of their progenitors, provided the infrared light curves are followed for months."],"supporting_citations":[{"why":"Supplies the multi-wavelength photometry and spectra of AT 2018cow that the model fits.","marker":"Perley et al. 2019"},{"why":"Argues the near-infrared excess is dust emission near the sublimation temperature and provides the single-temperature baseline the paper corrects.","marker":"Metzger & Perley 2023"},{"why":"Provides the sublimation rate formula used to compute the evaporation-cavity sizes.","marker":"Waxman & Draine 2000"},{"why":"Provides the grain absorption coefficients used in the heating and emission calculations.","marker":"Draine & Lee 1984"},{"why":"Supplies the analytic light-curve model used to define the incident FBOT radiation field.","marker":"Arnett 1982"},{"why":"Gives the power-law density profile that the dust shell is assumed to follow.","marker":"Bright et al. 2022"},{"why":"Previous dust-echo calculation for a toroidal shell, which this work extends by including late-time heating and size-dependent evaporation.","marker":"Tuna et al. 2025"},{"why":"Contributes the W1 and W2 photometry that anchors the longest-wavelength echo data.","marker":"Jiang et al. 2021"}],"fun_headline_variants":["AT 2018cow's glow traced to dusty echo","Flash vaporizes dust, then echoes warm","Dust echo fits AT 2018cow's observed light","FBOT flash carves cavity, dust echo explains NIR","Cavity carved in a day, surviving dust echoes AT 2018cow"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The pre-existing dust shell is assumed to have a density that falls off as the cube of the distance and to be made of carbon grains with the standard interstellar size distribution, with values taken from other astronomical contexts rather than measured for AT 2018cow; the fitted dust mass, CSM mass, and mass-loss rate scale directly with those assumptions.","fun_headline_variants_meta":{"raw":{"variants":["AT 2018cow's glow traced to dusty echo","Flash vaporizes dust, then echoes warm","Dust echo fits AT 2018cow's observed light","FBOT flash carves cavity, dust echo explains NIR","Cavity carved in a day, surviving dust echoes AT 2018cow"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000778,"raw_usage":{"total_tokens":3486,"prompt_tokens":1035,"completion_tokens":2451,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":651,"completion_tokens_details":{"reasoning_tokens":2363}},"tokens_in":651,"tokens_out":2451,"duration_ms":16103,"temperature":1.0,"reasoning_tokens":2363,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T05:41:08.975947+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take continued JHK and WISE photometry of AT 2018cow out to 100 to 300 days after explosion: the model predicts a light-crossing plateau followed by a roughly $t^{-3}$ decline with smoothly cooling dust temperatures, so any observed infrared light curve that stays flat, rebrightens, or shows spectral features inconsistent with cooling carbonaceous grains would contradict the echo interpretation.","supporting_citations":[{"cited_title":"A., Mazzali, P","cited_arxiv_id":null,"evidence_quote":"Supplies the multi-wavelength photometry and spectra of AT 2018cow that the model fits."},{"cited_title":"D., & Perley, D","cited_arxiv_id":null,"evidence_quote":"Argues the near-infrared excess is dust emission near the sublimation temperature and provides the single-temperature baseline the paper corrects."},{"cited_title":"2000, The Astrophysical Journal, 537, 796","cited_arxiv_id":null,"evidence_quote":"Provides the sublimation rate formula used to compute the evaporation-cavity sizes."},{"cited_title":"2021, The Astrophysical Journal, 911, 31","cited_arxiv_id":null,"evidence_quote":"Contributes the W1 and W2 photometry that anchors the longest-wavelength echo data."}],"review_version":1}