{"id":"138e9f55-5ce5-4ffe-b559-1313c6b9c508","arxiv_id":"2501.05521","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"The new MSG model grid produces redder near-infrared spectra than observed and cannot reproduce the 10-micron silicate absorption feature seen by JWST and Spitzer, despite self-consistent microphysical clouds.","lead":"The MSG project couples the MARCS atmosphere code with the DRIFT cloud-formation code to build self-consistent models of cloudy brown dwarfs and giant planets from 1200 to 2500 K. The models still fail to reproduce the observed 10-micrometer silicate cloud feature, and the authors trace the gap to nucleation and mixing assumptions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The negative result is tested against field L dwarfs at log g≈5.2 while all MSG models use log g=4.0; the missing 10-µm silicate feature and NIR redness may be gravity-mismatch artifacts.","rationale":"The central claim is that the MSG grid with microphysical clouds cannot reproduce the observed 10-µm silicate feature and is too red in the NIR. For this claim to be robust, the models must be compared to observations with matched physical parameters that control cloud structure. The paper's only detailed model-observation comparison, Fig. 6, is against 2MASS 1507-1627 at log(g)≈5.2, while the entire grid is computed at log(g)=4.0. Because scale height, convective mixing, and gravitational settling all depend strongly on g, the cloud particle sizes and the location of τ=1 will change with gravity; low-gravity models are naturally redder in the NIR. Without models at the gravity of the comparison objects, the negative result may be a gravity-mismatch artifact. The reader's weakest assumption, the mixing-timescale treatment in detached convective zones (Section 6.3), is a real and important model-physics caveat, but the gravity mismatch is a more basic validity condition for the observational comparison. This concern does not overturn the paper; it strengthens the conditionality already assigned by the reader. The proposed log(g)=5.2 grid would settle whether the negative result is physical or an artifact of the chosen gravity.","tokens_in":38192,"tokens_out":9570,"duration_ms":97906,"concrete_test":"Compute self-consistent MSG models at Teff=1600 K (and 1500, 1700 K) with log(g)=5.2, the gravity of 2MASS 1507-1627, for both TiO2 and SiO nucleation, using the same MARCS+DRIFT setup. Compare the synthetic spectra to the Cushing et al. (2005) NIR and Suárez & Metchev (2022) MIR data with the same normalization as Fig. 6, quantifying the 9-11 µm feature depth and the NIR slope or J-K color. If the silicate feature emerges or the NIR redness is substantially reduced at log(g)=5.2, the central negative result must be qualified as specific to log(g)=4.0; if the discrepancy persists unchanged, the gravity mismatch is not the cause.","verdict_should_be":"UNCHANGED","load_bearing_attack":"All MSG models are computed at log(g)=4.0, yet the principal spectral comparison in Fig. 6 is to 2MASS 1507-1627, an L5 dwarf with Teff≈1600 K and log(g)≈5.2 (Filippazzo et al. 2015), and the 'observed population' behind the NIR-redness and Spitzer silicate statements is dominated by field dwarfs with log(g)≈5.0-5.5. Cloud structure depends strongly on gravity: the scale height (Eq. 10), the convective velocity, and the settling-mixing balance in Eqs. 1, 6, and 9 all shift with g, changing both the particle sizes and the pressure at which τ=1 is reached. A log(g)=4.0 model is therefore not a valid proxy for a log(g)=5.2 L dwarf. The missing 10-µm silicate feature and the too-red NIR continuum could be gravity-mismatch artifacts rather than intrinsic failures of the MSG microphysical cloud treatment. The paper presents no MSG spectra at the gravity of the comparison object and no direct MSG fit to the JWST MIRI spectrum of VHS 1256 b, so the central negative claim is not yet established for the observations it is meant to explain.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents the MSG grid, a new set of self-consistent 1D substellar atmosphere models coupling MARCS with the DRIFT/StaticWeather microphysical cloud model, with effective temperatures 1200–2500 K at log(g)=4.0. The central scientific claim is that this grid, despite including cloud radiative feedback, microphysical nucleation/growth/evaporation, and an updated opacity treatment, cannot reproduce the ~10 μm silicate absorption feature seen in recent JWST/Spitzer observations, and that the TiO2-nucleation models are too red in the near-infrared compared to the observed L-dwarf population. The authors explore two modifications, switching the condensation nuclei from TiO2 to SiO and reducing the mixing efficiency, and show that both reduce the NIR redness but do not produce the silicate feature. They attribute the missing feature to a lack of small cloud particles at low pressures and to gas opacity hiding the cloud's silicate signature, and they identify convective transport as the main avenue for future work.","tokens_in":38481,"tokens_out":3646,"duration_ms":40207,"significance":"If the central negative result holds, the paper is valuable: it provides a new self-consistent cloudy grid with kinetic cloud formation, a documented control-theory convergence algorithm, and a clear falsifiable prediction (the absence of a 10 μm silicate feature in this modeling framework). The paper is also unusually honest: it explicitly reports where the model fails, labels the mixing-timescale scaling as breaking self-consistency, and discusses the detached-convective-zone caveat in Section 6.3. These features make the manuscript a useful benchmark for the substellar-atmosphere community. The strength of the published claim, however, depends on whether the model is compared to observations at the correct gravity and whether the mixing prescription controls the result; both need to be addressed before the negative result can be considered established for field L dwarfs.","major_comments":[{"comment":"The central observational comparison is made to 2MASS 1507-1627, an L5 dwarf with Teff≈1600 K and log(g)≈5.2, while every MSG model in the grid is computed at log(g)=4.0. The cloud scale height (Eq. 10), the mixing timescale (Eq. 9), the settling-mixing balance in the dust moment equations (Eqs. 1 and 6), and the pressure where τ=1 is reached all depend on gravity. The missing 10 μm silicate feature and the too-red near-infrared continuum could therefore be artifacts of comparing a log(g)=4.0 model to a log(g)=5.2 object. The paper should either compute at least one MSG model at the gravity of the comparison target, or compare to an object with log(g)≈4.0, before concluding that the grid as a whole cannot reproduce the observed silicate feature.","section":"§5.3, Fig. 6"},{"comment":"The treatment of detached convective zones is a load-bearing assumption for the central claim: when a detached convective zone appears (which it does at Teff≤1600 K, exactly the L-dwarf regime of interest), τmix is set to a constant value taken from the top of the radiative zone below. This neglects the convective velocity field and the possibility that convective motions drag cloud particles, as implemented in Witte et al. (2011). Since the paper's own discussion recognizes that this can change cloud structure and potentially inhibit reddening, the conclusion that the MSG grid cannot produce the silicate feature is conditional on this mixing prescription. A quantitative sensitivity test, or at least a clear statement that the result is a property of this mixing model rather than of the microphysical cloud treatment, is needed.","section":"§6.3, Eq. (12)"},{"comment":"The 1000× slower mixing models that improve the near-infrared agreement are explicitly not self-consistent, as the authors state. These non-self-consistent models are nevertheless used in Fig. 11 to support the statement that reduced mixing makes the models less red. This is acceptable as an exploratory test, but the conclusion should be framed as a property of the perturbed, non-self-consistent model, and the manuscript should state whether the perturbed models satisfy any structural convergence check beyond the quoted criteria. As written, the reader cannot tell how much of the NIR improvement is due to the intended physical effect and how much is due to the loss of self-consistency.","section":"§5.5, Figs. 10–11"},{"comment":"The optical-depth analysis correctly shows that the silicate feature is present in the cloud optical properties at 0.1–1 mbar but is hidden in the emergent spectra because the gas is optically thicker at those wavelengths. However, this analysis is performed only for log(g)=4.0 models at one effective temperature. Given that the comparison targets are predominantly field dwarfs with log(g)≈5.0–5.5, the pressure and altitude of the τ=1 surface will shift, and the conclusion that the gas always hides the silicate feature needs to be demonstrated at the gravity of the observed objects.","section":"§5.6, Figs. 12–13"}],"minor_comments":[{"comment":"The notation V^s_ℓ is used in Eq. (5) but is not defined consistently with V_ℓ in Eq. (1); please define the per-species lower volume boundary explicitly.","section":"§2, Eq. (5)"},{"comment":"The sentence beginning 'The major reasoning for testing one species against the other Comparisons of classical and non-classical nucleation theories...' is grammatically incomplete and should be rewritten.","section":"§6.4"},{"comment":"In the text, 'Fig. 8 left panel shows the cloud composition' is misleading: the left panel of Fig. 8 shows optical depth, while the composition is shown in the middle panel. Please correct the cross-reference.","section":"§5.4, Fig. 8"},{"comment":"The control-factor percentages (10% increase, 50% decrease) are stated to have been chosen after testing on a toy model; a brief description of that toy model, or a reference to where it is described, would help the reader judge the robustness of the convergence algorithm.","section":"§4.3"}],"recommendation":"major_revision","confidential_remarks":"This is a useful and honest contribution, and the negative result on the silicate feature is likely of interest to the substellar-atmosphere community. The main obstacle is that the headline comparison is done at the wrong gravity: all models are log(g)=4.0 while the primary observed comparison object has log(g)≈5.2. In my view this is a fixable but load-bearing issue, not a fatal one; the authors could add a small number of models at higher gravity or reframe the conclusion as a property of their log(g)=4.0 grid. The mixing-scale test in §5.5 should also be presented more cautiously because it breaks self-consistency by design. I would support publication after these points are addressed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a legitimate, useful modeling paper, not a breakthrough. It builds a new self-consistent cloudy-substellar grid with MARCS+DRIFT, extends down to 1200K, adds five cloud species, and reports an honest null result: no 10µm silicate feature in any of its models. The convergence-control algorithm is a real contribution; the optical-depth and single-scattering-albedo analysis is the best part of the paper and shows why the feature is absent under these assumptions.\n\nWhat the paper does well: the coupling is described in enough detail to reproduce, the grid is systematic, and the exploration of TiO2 vs SiO nucleation in a fully self-consistent setting is new. The 1000x slowdown of mixing is clearly labeled as breaking self-consistency, and the paper is upfront about the detached-convective-zone approximation. That kind of transparency should be credited.\n\nThe soft spots: the gravity problem is the main one. Every model is at log g=4.0, but the flagship comparison (Fig. 6, Fig. 11) is to 2MASS 1507-1627, an L5 with log g≈5.2. Cloud structure, scale height, and the τ=1 pressure all depend on g, so the statement that the grid is unable to reproduce the silicate feature is only proven for log g=4.0. It is not yet proven that the microphysics fails in the objects being compared. The paper never discusses this mismatch, and the conclusions overreach. This is fixable: compute a few higher-g models or soften the language. Second, the 1000x mixing scaling is a free-knob improvement; the authors admit it breaks self-consistency, so it should be treated as a sensitivity test, which they do, but the NIR better match in Fig. 11 is close to fitting a parameter. Third, the truncated sentence in Section 6.4 is a simple copyedit issue. Fourth, no code or grid data is released, which limits reproducibility, though this is not unusual.\n\nNet: for cloud modelers and brown dwarf spectroscopists this is a worthwhile paper that deserves referee time. I would accept for review with major revision; the gravity mismatch needs to be addressed, and the conclusions should be scoped to log g=4.0 or supported with higher-g models. I would cite it if working on substellar clouds.","headline":"Honest, useful cloudy-substellar grid at log g=4.0 with a real negative result, but the headline comparison to field L dwarfs is weakened by a gravity mismatch and needs revision.","tokens_in":39061,"tokens_out":3213,"would_cite":true,"duration_ms":31909,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A new grid of cloudy brown dwarf models cannot reproduce the 10-micron silicate feature.","keywords":["brown dwarfs","substellar atmospheres","microphysical cloud formation","silicate absorption feature","atmospheric mixing timescale","self-consistent atmosphere models","DRIFT cloud model","MARCS atmosphere model"],"falsifier":"Re-running the MSG coupling with the convective-drag terms of Witte et al. (2011) included in the dust moment equations, and checking whether a $T_\\mathrm{eff}=1500$ K, $\\log(g)=4.0$ model then produces the $10\\,\\mu$m silicate feature in its synthetic spectrum, would settle whether the missing feature is intrinsic to the microphysics or an artifact of the mixing prescription.","tokens_in":37979,"feed_emoji":"☁️","tokens_out":14516,"duration_ms":114524,"temperature":0.7,"pith_summary":"The paper presents a new grid of self-consistent cloudy substellar atmosphere models, the MSG grid, produced by coupling the MARCS radiative-convective equilibrium code with the kinetic, non-equilibrium cloud formation model DRIFT. The purpose is to test whether microphysical cloud formation, with cloud opacity feeding back on the atmospheric structure, can reproduce the silicate absorption feature at about $10\\,\\mu$m that JWST and Spitzer observe in many brown dwarfs and in the planetary-mass companion VHS 1256 b. The result is negative: across effective temperatures $T_\\mathrm{eff}=1200$–$2500$ K at $\\log(g)=4.0$, none of the models produces the feature, and the models with TiO$_2$ nucleation are redder in the near-infrared than the observed population. Switching the nucleation species to SiO, or slowing the mixing timescale by up to a factor of 1000, reduces the near-infrared redness but does not bring back the silicate feature. The authors trace the missing feature to a shortage of small ($0.1$–$1\\,\\mu$m) cloud particles at low pressures and argue that the treatment of convection in cloud-forming regions is the likely missing ingredient.","feed_headline":"Cloudy brown dwarf models still miss the 10-micron silicate feature","feed_subtitle":"JWST sees the silicate feature; these models do not, pointing to missing convection or nucleation physics.","key_machinery":"The load-bearing machinery is the iterative MSG coupling: the MARCS atmosphere code supplies pressure-temperature structure, gas density, scale height, and convective velocity to DRIFT, whose dust moment equations describe nucleation, growth, evaporation, settling, and gas-phase element depletion, with a mixing timescale $\\tau_\\mathrm{mix}$ parameterising convective replenishment and capped by $\\beta_\\mathrm{cr}=2.2$. DRIFT returns cloud particle sizes, condensate volume fractions, and depleted abundances, from which cloud opacity is computed using effective medium theory and Mie theory, and a control factor $f$ limits the change in cloud opacity and element abundances between iterations so that the radiative-convective equilibrium solution converges. This loop allows the authors to attribute the missing silicate feature to the microphysical state of the cloud at observable altitudes rather than to numerical non-convergence.","core_discovery":"The central claim is that a microphysically self-consistent cloud model, in which the cloud opacity and gas-phase element depletion are iterated to convergence with the atmospheric structure, still cannot reproduce the observed silicate absorption band near $10\\,\\mu$m in substellar atmospheres. The MSG grid couples MARCS to DRIFT via a control-factor algorithm that damps oscillations in cloud opacity and depleted abundances, and it includes twelve condensate species; nonetheless the resulting spectra show a strong cloud continuum that is too red in the near-infrared when TiO$_2$ seeds the clouds. Using SiO as condensation nuclei, or scaling the mixing timescale upward by up to $1000\\times$, produces less red near-infrared spectra, but the $10\\,\\mu$m feature remains absent. The paper argues that the model does not form enough small particles of order $0.1$–$1\\,\\mu$m at pressures near $0.1$–$1$ mbar, where the silicate opacity would be visible against the gas, and that the mixing prescription, which ignores convective drag on cloud particles inside detached convective zones, is the most plausible source of the discrepancy.","pith_inferences":["If the missing feature is an artifact of the constant-$\\tau_\\mathrm{mix}$ assumption in detached convective zones, including convective drag on cloud particles in the dust moment equations should restore the 10 $\\mu$m band, turning it into a probe of convective cloud recycling.","The paper's reasoning predicts that any change that raises the number of condensation nuclei at the top of the atmosphere (higher nucleation rate, stronger mixing, or a bimodal particle size distribution) will strengthen the silicate feature; this can be tested by varying the cluster size $N_\\ell$ or the size distribution in the same MARCS-DRIFT loop.","Because reduced-mixing models improve the near-infrared but worsen the mid-infrared match, a patchy-cloud retrieval that averages two converged MSG spectra with different cloud fractions is a natural next step, and the grid supplies the end-member spectra for it.","Applying the same MARCS-DRIFT coupling to hot Jupiter atmospheres, where $10\\,\\mu$m silicate features have already been detected, would test the same microphysics outside the brown dwarf regime."],"forward_implications":["The 10 $\\mu$m silicate feature becomes a sharp observational test of cloud microphysics: a model must get particle sizes, number densities, and altitudes simultaneously right to show it.","TiO$_2$ nucleation, a standard choice in DRIFT-based grids, yields near-infrared colors redder than the observed L dwarf population, so grids that assume it may bias inferred effective temperatures and gravities.","SiO nucleation and reduced mixing efficiency both make the near-infrared less red, so the two knobs are degenerate and require independent constraints from other observables.","Below $T_\\mathrm{eff}=1600$ K the cloudy models develop detached convective zones, and the paper identifies the constant-$\\tau_\\mathrm{mix}$ treatment there as the prime suspect for the over-red colors and missing silicate feature.","The control-factor convergence scheme makes it practical to run self-consistent cloudy models down to $1200$ K, enabling JWST-era comparisons across the L/T transition."],"supporting_citations":[{"why":"presents the DRIFT-PHOENIX grid, the self-consistent cloudy predecessor this work extends and compares against","marker":"Helling et al. (2008b)"},{"why":"derives the DRIFT dust moment equations and the mixing timescale parameterisation used throughout","marker":"Woitke & Helling (2004)"},{"why":"adds convective motion to the dust moment equations and is cited as the missing physics behind the missing silicate feature","marker":"Witte et al. (2011)"},{"why":"reports the JWST MIRI detection of the 9–11 µm silicate feature in VHS 1256 b, the key observation the grid fails to reproduce","marker":"Miles et al. (2023)"},{"why":"provides the Spitzer survey showing the silicate feature's prevalence among L dwarfs, used as the mid-infrared benchmark","marker":"Suárez & Metchev (2022)"},{"why":"supplies the near-infrared spectrum of 2MASS 1507-1627 used for direct model-observation comparison","marker":"Cushing et al. (2005)"},{"why":"compares TiO2 and SiO nucleation with DRIFT and establishes that SiO nucleates more efficiently","marker":"Lee et al. (2015)"},{"why":"extends MARCS to cool substellar temperatures, enabling the 1200 K cloudy models in this grid","marker":"Jørgensen et al. (2024)"},{"why":"shows five forward models fail on the VHS 1256 b silicate feature, motivating an updated self-consistent grid","marker":"Petrus et al. (2024)"}],"fun_headline_variants":["Self-consistent clouds fail to match JWST silicate spectra","Microphysical cloud grid still can't produce 10-micron bump","MSG model misses silicate feature despite cloud microphysics","Cloudy substellar models still lack the 10 micron silicate feature","No silicate bump in self-consistent cloud models"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central negative result depends on the parameterised mixing timescale, in particular the assumption that inside detached convective zones $\\tau_\\mathrm{mix}$ stays fixed at its value at the top of the radiative zone below, which ignores the possibility that convection drags cloud particles and disrupts cloud layers.","fun_headline_variants_meta":{"raw":{"variants":["Self-consistent clouds fail to match JWST silicate spectra","Microphysical cloud grid still can't produce 10-micron bump","MSG model misses silicate feature despite cloud microphysics","Cloudy substellar models still lack the 10 micron silicate feature","No silicate bump in self-consistent cloud models"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000515,"raw_usage":{"total_tokens":2595,"prompt_tokens":1135,"completion_tokens":1460,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":751,"completion_tokens_details":{"reasoning_tokens":1378}},"tokens_in":751,"tokens_out":1460,"duration_ms":10078,"temperature":1.0,"reasoning_tokens":1378,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:14:14.507188+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-running the MSG coupling with the convective-drag terms of Witte et al. (2011) included in the dust moment equations, and checking whether a $T_\\mathrm{eff}=1500$ K, $\\log(g)=4.0$ model then produces the $10\\,\\mu$m silicate feature in its synthetic spectrum, would settle whether the missing feature is intrinsic to the microphysics or an artifact of the mixing prescription.","supporting_citations":[{"cited_title":"& Helling, C","cited_arxiv_id":null,"evidence_quote":"derives the DRIFT dust moment equations and the mixing timescale parameterisation used throughout"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"adds convective motion to the dust moment equations and is cited as the missing physics behind the missing silicate feature"},{"cited_title":"E., Biller, B","cited_arxiv_id":null,"evidence_quote":"reports the JWST MIRI detection of the 9–11 µm silicate feature in VHS 1256 b, the key observation the grid fails to reproduce"},{"cited_title":"2015, A&A, 580, A12","cited_arxiv_id":null,"evidence_quote":"compares TiO2 and SiO nucleation with DRIFT and establishes that SiO nucleates more efficiently"}],"review_version":1}