{"id":"6a450e67-fa14-453a-95f8-f5086a0d7a3f","arxiv_id":"1908.09633","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A time-dependent polymer nucleation model with quantum-chemically computed cluster energies shows monomer-only growth underpredicts large cluster formation at low temperatures, and gives a new ranking of TiO2, MgO, SiO, and Al2O3 as AGB dust precursors.","lead":"Astronomers have built a new computer model of how dust grains first form in the winds of aging stars, letting any size cluster stick to any other instead of only adding one molecule at a time. It shows that simpler monomer-only models underestimate how much dust forms at low temperatures, and it points to alumina as the likely first dust when alumina molecules are present.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The favored-precursor claim depends on the unverified assumption of homomolecular, fixed-stoichiometry cluster growth, which the paper itself flags in Sec. 6.1.1.","rationale":"The reader's weakest_assumption identified exactly this issue: homomolecular, fixed-stoichiometry growth with no heteromolecular or non-stoichiometric pathways (Sec. 3.1, Sec. 6.1.1). The paper itself states that this assumption is not established, and it also acknowledges that heteromolecular nucleation is likely needed for MgAl2O4 formation, which is abundant in presolar grains. My stress-test confirms that this is the most load-bearing concern: the central comparative claims (monomer vs. polymer, Al2O3 vs. TiO2 as first precursor) are all computed within this restricted cluster space. The concern is not that the paper is wrong internally—the calculations are consistent given the assumption—but that the astrophysical conclusion about the first condensate could change if the cluster space is expanded. The paper is transparent about the limitation, and the reader already conditioned acceptance on testing it. Therefore, I do not need to change the verdict; the conditional acceptance stands. The proposed concrete test—extending the Al2O3 network to non-stoichiometric clusters—would directly settle whether the favored precursor survives a relaxation of the fixed-stoichiometry assumption.","tokens_in":75188,"tokens_out":3140,"duration_ms":33954,"concrete_test":"Re-run the closed Al2O3 nucleation model with a network that includes all low-energy AlxOy clusters (e.g., AlO, AlO2, Al2O, Al2O2, Al2O3, Al3O4, etc., using structures from Patzer et al. 2005 and Li & Cheng 2012) up to a total of 8 Al atoms, allowing polymer growth between every pair of clusters with the same hard-sphere/detailed-balance rate prescription as in Eqs. (10)–(16). If any non-stoichiometric cluster overtakes (Al2O3)8 as the dominant large cluster at T > 1800 K, or if the apparent Al2O3 formation threshold shifts below ~1200 K, then the favored-precursor claim in Secs. 4.1.4 and 5.2 is not robust to the fixed-stoichiometry assumption. The same check should be applied to the TiO2 network with Ti/O ratios other than 1:2.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that Al2O3 is the favored first condensate in AGB winds: in closed models (Al2O3)8 forms fastest at 1800–2400 K, while in the comprehensive atomic network only TiO2 clusters form (Secs. 4.1.4, 4.2, 5.2). Both legs of this comparison rest on a network that permits growth only by integer multiples of a fixed stoichiometric monomer (Al2O3, TiO2, MgO, SiO) and never allows non-stoichiometric or heteromolecular clusters (Secs. 2.2, 3.1). The paper explicitly concedes: 'the fact that nucleation occurs via the addition of monomer-multiples with a fixed stoichiometry is not established either' (Sec. 6.1.1), and it also notes that heteromolecular nucleation is 'most likely necessary to create MgAl2O4-clusters, which are abundant in pre-solar AGB grains'. This is load-bearing because the identity and temperature threshold of the first dust precursor could change if, for instance, AlxOy clusters with x:y ≠ 2:3 (e.g., AlO, Al2O2, AlO2) attach to growing clusters, or if MgAl2O4-type mixed clusters form efficiently. The comprehensive-model result that 'only TiO2 forms' is likewise a statement about the absence of the chosen homomolecular monomers under the adopted gas-phase network; it does not rule out non-stoichiometric or heteromolecular nucleation pathways that bypass those monomers. Since the paper's headline conclusions are about which dust precursor forms first and at what temperature, this unverified structural assumption is the weakest load-bearing point in the argument.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a time-dependent, non-equilibrium chemical-kinetics nucleation model for AGB winds, applied to TiO2, MgO, SiO, and Al2O3. Growth rate coefficients are hard-sphere collision rates (Eq. 11) and destruction rates are obtained from detailed balance using DFT-based Gibbs free energies (Eq. 16). Two classes of models are studied: closed nucleation networks, in which the monomer is assumed present and the system contains only one nucleating species, and a comprehensive network starting from an atomic composition and including a large gas-phase reaction network. The main findings are that monomer-restricted growth underpredicts large clusters at low temperatures and overpredicts formation times; that in the closed models (Al2O3)8 forms fastest and at the highest temperatures (1800-2400 K), making Al2O3 the favoured first condensate; and that in the comprehensive atomic network only TiO2 clusters form, with Al and Mg remaining essentially atomic. The authors conclude that the absence of Al2O3 in the comprehensive model is likely due to incomplete Al reaction data and explicitly argue, based on presolar grains and observations, that Al2O3 remains the prime candidate.","tokens_in":75605,"tokens_out":6478,"duration_ms":67916,"significance":"The central methodological contribution is a transparent, parameter-free nucleation framework that replaces classical steady-state nucleation with polymer growth and quantum-chemical cluster energetics. The comparison of monomer vs polymer growth yields a clear, qualitative result: monomer-only models underestimate large-cluster abundances at low temperatures and overestimate formation times, which is directly relevant to existing AGB wind models. The model makes falsifiable predictions: in an initially atomic mixture, only TiO2 clusters form among the four candidates, while Al2O3 nucleates efficiently if its monomer is assumed present. The paper is unusually open about its limitations, and the release of homogenized thermochemical data and code, including Gibbs free energies for the clusters, is a significant strength that will allow independent checks. The two legs of the central claim rest on assumptions that are acknowledged but not quantitatively tested, which is why the work needs revision rather than acceptance as is.","major_comments":[{"comment":"The assumption that nucleation is homomolecular and proceeds by integer multiples of a fixed stoichiometric monomer is load-bearing for both parts of the central claim: the closed-model preference for Al2O3 and the comprehensive-model conclusion that only TiO2 clusters form. The paper itself states in Sec. 6.1.1 that 'the fact that nucleation occurs via the addition of monomer-multiples with a fixed stoichiometry is not established either.' If, for example, AlxOy species with x:y not equal to 2:3 (such as AlO or Al2O2) can attach to growing clusters, or if heteromolecular clusters like MgAl2O4 form efficiently, the identity and temperature threshold of the first dust precursor could change. The authors should add a sensitivity test that relaxes the fixed-stoichiometry restriction for at least one candidate (e.g., allowing AlO or Al2O2 addition to Al2O3 clusters) and report how the formation threshold and cluster distribution shift, or provide a quantitative argument for why such pathways cannot compete.","section":"2.2, 3.1, 6.1.1"},{"comment":"The growth rates used throughout the paper are geometric hard-sphere collision rates (Eq. 11). Figure 6 shows that for Al2O3 dimerization this approximation overestimates the RRKM/Lindemann rate by roughly an order of magnitude. Because the closed-model conclusion that (Al2O3)8 forms rapidly at 1800-2400 K depends directly on the magnitude of k+ in Eq. (11), an order-of-magnitude overestimate in k+ could materially change the formation-temperature boundary and the convergence times reported in Sec. 4.1.4. The paper acknowledges this qualitatively but does not propagate the uncertainty. I request a simple sensitivity run: repeat the closed polymer-nucleation model for Al2O3 with k+ reduced by a factor of 10 (or with the RRKM rate of Sharipov & Loukhovitski 2018) and report the resulting (Al2O3)8 abundance map and time evolution. This would establish whether the qualitative ranking of candidates is robust.","section":"6.1.1, Fig. 6"},{"comment":"The abstract and Sec. 5.2 present the comprehensive-model result as 'only TiO2-clusters form' while simultaneously concluding that Al2O3 is the prime candidate. The manuscript is transparent that the Al2O3 preference is based on external evidence rather than the model, but the phrasing in the abstract and summary can be read as a model prediction. Since the authors themselves attribute the non-formation of Al2O3 to incomplete Al reaction data, the paper should state more crisply that the comprehensive-model prediction is conditional on the adopted network, that the absence of Al2O3 is not a falsification of Al2O3 nucleation, and that the Al2O3-favouring conclusion is an inference from the closed models plus observational evidence. This distinction is essential for readers assessing the strength of the paper's main claim.","section":"4.2.4, 5.2, 6.1.2"}],"minor_comments":[{"comment":"The comparison with equilibrium abundance ratios in Sec. 4.1.5 uses Eq. (15), which is derived from the same Gibbs free energies that set the destruction rate coefficients in Eq. (16). The comparison therefore tests kinetic convergence toward the equilibrium implied by the model's own thermodynamics, not the validity of the thermodynamics. This should be stated explicitly in the text to avoid the impression of an independent equilibrium check.","section":"4.1.5, Eqs. (13)-(16)"},{"comment":"Table D2 is difficult to parse because the legend uses 'CCCBDB' both as a source name and as an entry, and several entries contain '?' without a reference (e.g., SO, NO2). The authors should either provide the missing references or mark these entries as unverified. This matters because reversed rate coefficients for the comprehensive network depend on these thermochemical data.","section":"Appendix D, Table D2"},{"comment":"In the extrapolation from (Al2O3)8 to (Al2O3)1000, the factor relating cluster sizes is 8/1000 = 125, not 100 as stated in the text. The difference is small compared to the other uncertainties, but the arithmetic should be corrected or stated as 'roughly 100' with the exact factor.","section":"6.2.2"},{"comment":"The opening statement that 'various nucleation theories exist, yet all assume chemical equilibrium, growth restricted by monomers' is too strong, since the paper itself cites works (e.g., Sarangi & Cherchneff 2015; Gobrecht et al. 2016) that abandon equilibrium. Please qualify this statement so it refers to the most commonly used nucleation prescriptions in AGB wind models.","section":"Abstract, Sec. 1"}],"recommendation":"major_revision","confidential_remarks":"The paper is borderline between major revision and rejection. The core rate-equation formalism is sound and the authors are appropriately transparent about limitations; however, the two sensitivity checks I request (relaxing fixed-stoichiometry growth and reducing the hard-sphere growth rates) are necessary to support the quantitative central claims. If the authors can show that the qualitative ranking of Al2O3 vs TiO2 survives these perturbations, the paper would be a valuable contribution. If the sensitivity tests overturn the ranking, the conclusions would need substantial reframing. The paper is within the journal's scope and the data release is a strong point in its favour."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Couple things to know before you read this one. The real result is not the candidate ranking; it's the demonstration that monomer-restricted nucleation fails at low temperatures because of monomer depletion, and that time-dependent polymer nucleation fixes that. The comparison is internally consistent and clearly presented. The headline claim that Al2O3 is the favored first condensate is conditional on assumptions the authors themselves flag as unestablished, so read the abstract with that in mind.\n\nWhat's new: the authors let clusters grow by arbitrary polymer collisions, not just monomer addition, and combine that with time-dependent kinetics and DFT-computed Gibbs free energies for TiO2, MgO, SiO, and Al2O3 clusters. As far as I know no prior astrophysical nucleation study does all three. They also ship the data and code (Zenodo, Bitbucket), which makes the thermodynamic inputs checkable. The monomer-depletion bottleneck is a real effect that steady-state classical nucleation theories miss, and they show it convincingly.\n\nSoft spots, in proportion. The hard-sphere collision rates overestimate the Al2O3 dimerization rate by about an order of magnitude compared to the RRKM/Lindemann calculation they plot (Fig. 6). That is a known limitation and they say so, but it does affect absolute formation times. More important: the whole growth scheme is homomolecular and fixed-stoichiometry—clusters only add integer multiples of Al2O3, TiO2, etc. The stress-test note is right that this is load-bearing: if AlxOy with x:y ≠ 2:3 or mixed Mg-Al oxides participate, the favored precursor and its formation temperature could change. The paper explicitly concedes the fixed-stoichiometry assumption is not established (Sec. 6.1.1) and that heteromolecular nucleation is likely needed for MgAl2O4. So the 'Al2O3 if monomers exist, TiO2 only in the atomic mixture' result is best read as the behavior of the chosen network, not as a robust astrophysical prediction. The comprehensive model's failure to form any Al2O3 is really a statement about missing Al reaction data; the authors argue that case themselves.\n\nThe central methodological argument holds up. This is a serious contribution for dust nucleation modelers and for anyone who uses monomer-based nucleation prescriptions in AGB winds. It deserves a serious referee; a good referee will push them to separate the robust comparison (monomer vs polymer) from the conditional candidate ranking, and to propagate DFT energy uncertainties. I would accept it for review and cite the monomer-vs-polymer result.","headline":"A genuine methods advance in dust nucleation modeling: the monomer-vs-polymer comparison is solid and reproducible, but the favored-precursor ranking depends on assumptions the authors themselves flag as unverified.","tokens_in":76055,"tokens_out":2667,"would_cite":true,"duration_ms":28710,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"By dropping equilibrium, monomer-only growth, and bulk-material energies from dust nucleation, this paper finds that monomer restriction underpredicts large clusters and delays formation, and that Al2O3 — not TiO2 — seeds dust first…","keywords":["AGB stars","dust nucleation","polymer nucleation","non-equilibrium chemistry","aluminium oxide clusters","titanium dioxide clusters","stellar wind driving","density functional theory"],"falsifier":"Measure or recompute the rate coefficients for the two reactions that are the only available paths to the Al2O3 monomer, AlO + AlO + M -> Al2O2 + M and Al2O2 + O + M -> Al2O3 + M, at 1500-2500 K. If experiment or higher-level theory places these rates orders of magnitude above the combustion-derived values used here, the comprehensive model's failure to form Al2O3 is the data gap the paper claims; if the rates are right, the claim that Al2O3 is the first condensate in AGB winds fails, because its monomer cannot arise from an atomic gas and, by the model's own logic, only TiO2 clusters would seed dust.","tokens_in":75035,"feed_emoji":"⭐","tokens_out":15583,"duration_ms":139048,"temperature":0.7,"pith_summary":"Asymptotic giant branch (AGB) stars return heavy elements to the interstellar medium, and their mass loss is believed to be launched by radiation pressure on dust that forms in the wind. This paper argues that the three shortcuts in current nucleation models — chemical equilibrium, growth by adding one monomer at a time, and using bulk-solid energies for clusters of a few molecules — each distort the predicted dust. Replacing them with time-dependent kinetics in which any two clusters of the same molecule can merge, driven by quantum-mechanically computed Gibbs free energies, the paper finds that monomer-restricted growth underpredicts large clusters at low temperatures and overpredicts their formation times. When monomers are assumed present, (Al2O3)8 forms fastest and at the highest temperatures (1800-2400 K), making Al2O3 the favored first condensate; starting from a purely atomic gas, only (TiO2)10 forms, a contradiction the authors attribute to missing Al-oxide reaction data rather than to Al2O3 being a poor nucleator. If the results are right, current monomer-only equilibrium wind models underestimate dust abundance and can delay or even suppress the wind-driving that the paper is ultimately trying to explain.","feed_headline":"Monomer-only nucleation models undercount dust and delay the wind","feed_subtitle":"Polymer, equilibrium-free kinetics make Al2O3 the hot inner-wind seed; from pure atoms only TiO2 forms.","key_machinery":"The carrying mechanism is a homomolecular polymer nucleation network. Each formula unit (TiO2, MgO, SiO, Al2O3) defines a ladder of cluster species, and any two clusters $C_N$ and $C_M$ can merge into $C_{N+M}$ with a growth rate from hard-sphere collision theory, $k^+_{N,M} = \\pi(r_N + r_M)^2 \\sqrt{8 k_B T / \\pi \\mu_{N,M}}$, while the reverse dissociation rate is set by detailed balance using the cluster's Gibbs free energy at standard pressure, $k^-_{N,M} = k^+_{N,M} \\frac{P^\\circ}{k_B T} \\exp\\left(\\frac{G^\\circ_{N+M} - G^\\circ_M - G^\\circ_N}{k_B T}\\right)$. The Gibbs free energies come from B3LYP density functional theory with vibrational analysis instead of bulk-extrapolated surface energies, which is what produces non-monotonic stability patterns such as (MgO)9 being more abundant than (MgO)10. Two network setups bracket the chemistry: closed networks that assume every metal atom starts in the monomer, isolating the intrinsic nucleation efficiency of each candidate, and a comprehensive network grafted onto a reduced AGB wind chemical network — extended with Ti, Al, and Mg reactions — that starts from atoms and decides whether the monomers can form at all.","core_discovery":"On the paper's own terms, the discovery is a set of consequences of removing the equilibrium and monomer restrictions from AGB dust nucleation. Monomer-restricted growth depletes the monomer pool as soon as small clusters form, quenching further growth at low temperatures, whereas polymer nucleation lets clusters keep merging, so the monomer description underpredicts large clusters and overpredicts formation times; the difference is large for (MgO)9 (roughly 180 days versus a few hours to converge) and (TiO2)10 (60 days versus under 20 days). In the closed models, where the monomer is assumed present, (Al2O3)8 forms in about 5-10 hours at 1800-2400 K (more than 90 per cent of available Al at the most favorable conditions), ahead of (MgO)9 at 1500-1700 K and (TiO2)10 at 1000-1200 K, while SiO clusters do not form at all in the relevant range. In the comprehensive network starting from atoms, all Mg stays atomic and Al2O3 cannot assemble its monomer, so only (TiO2)10 forms; most Al remains atomic with at most 1 per cent in AlO, AlH, Al(OH)2, and Al(OH)3. The authors nonetheless conclude that Al2O3 is the prime candidate, citing presolar corundum grains, dust observed near the star at 1500-2000 K where TiO2 clusters cannot yet exist, and their judgment that the available Al-oxide rate coefficients, taken from combustion chemistry and dominated by high barriers, are the missing link.","pith_inferences":["A direct test of the paper's own logic: if new Al-oxide rate data lets the comprehensive network form (Al2O3)8 at 1800-2400 K, the Al2O3-first ranking is confirmed; if it still fails, the presolar and hot-dust evidence would instead point toward the heteromolecular pathways (such as MgAl2O4 spinel) that this paper deliberately sets aside.","Because the paper shows the hard-sphere collision rates exceed RRKM rates for Al2O3 dimerization by roughly an order of magnitude, the reported formation temperatures and abundances are probably optimistic upper bounds; the qualitative ordering — Al2O3 hottest, then MgO, then TiO2 — is more robust than the exact thresholds.","The monomer-depletion bottleneck is generic: any environment where condensation outruns monomer resupply — brown dwarf atmospheres, supernova ejecta, combustion — will misbehave under monomer-only nucleation, so the polymer treatment transfers well beyond AGB winds.","Coupling this nucleation network into a hydrodynamical wind trajectory is the natural next step; the paper's grid is in temperature and density, and its own extrapolation suggests seed abundances would be ample, but only a full dynamical coupling can test whether the Al2O3-versus-TiO2 outcome survives the true cooling and shock history of the outflow."],"forward_implications":["Current AGB wind models that keep monomer-only equilibrium nucleation will underestimate the abundance of large clusters and overestimate their formation times, because monomer depletion quenches growth once small clusters form; allowing polymer collisions raises dust yields and shortens formation times.","If Al2O3 monomers are present, dust seeds appear within hours at 1800-2400 K, which is high enough to explain dust observed at 1.5-2 stellar radii in oxygen-rich winds, where temperatures exceed the 1000-1200 K at which TiO2 clusters can form.","From an initially atomic mixture, TiO2 is the only viable first condensate among the four candidates, since magnesium never leaves the atomic phase and Al2O3's monomer cannot be assembled with currently available rate data.","Equilibrium abundance ratios are not reached within one year across much of the temperature range, so steady-state nucleation rates are generally invalid and time-dependent descriptions are necessary.","Rough extrapolation of the largest clusters to seed-sized particles yields normalized seed abundances of order $3\\times10^{-11}$ to $8\\times10^{-11}$ per hydrogen, well above the $10^{-16}$ threshold that dynamical models say is needed to drive an AGB wind."],"supporting_citations":[{"why":"Baseline non-classical monomer nucleation using quantum-mechanically derived cluster energies; the approach this paper generalizes and the source of the preferred-cluster-size effect.","marker":"Köhler et al. (1997)"},{"why":"Represents the monomer-nucleation-with-quantum-properties models that this paper argues are invalid by dropping the monomer restriction.","marker":"Lee et al. (2015)"},{"why":"Supplies the SiO cluster global-minimum structures and energies used for the Gibbs free energies, and another monomer-restricted baseline.","marker":"Bromley et al. (2016)"},{"why":"Prior polymer-nucleation study limited to small clusters whose Al-oxide rate coefficients differ from this work by 2-10 orders of magnitude, the key comparison explaining the divergent Al2O3 prediction.","marker":"Gobrecht et al. (2016)"},{"why":"Companion paper providing the reduced AGB wind chemical network onto which the nucleation networks are grafted in the comprehensive model.","marker":"Boulangier et al. (2019)"},{"why":"RRKM/Lindemann rate coefficients for Al2O3 dimerization used to show the rigid-sphere collision approximation overestimates nucleation efficiency by about an order of magnitude.","marker":"Sharipov & Loukhovitski (2018)"},{"why":"Steady-state nucleation model claiming TiO2 as the only efficient AGB nucleator; its 1000 K cut-off is reproduced but its steady-state assumption is shown to be invalid.","marker":"Jeong et al. (2003)"},{"why":"Dynamical wind models whose seed-particle threshold (ns/nH ~ 10^-16) lets the paper argue the predicted seed numbers are sufficient to drive the wind.","marker":"Höfner et al. (2016)"},{"why":"Provides the 1 solar-mass, Z=0.02 AGB evolution model whose time-averaged mass fractions set the grid's initial abundances.","marker":"Karakas (2010)"},{"why":"Supplies the TiO2 cluster global-minimum structures used for the DFT Gibbs free energies of the (TiO2)n ladder.","marker":"Lamiel-Garcia et al. (2017)"}],"fun_headline_variants":["Polymer nucleation speeds dust seeds, flips Al2O3 timing","Atomic mix yields only TiO2 dust; Al2O3 still argued as seed","Monomer restriction overpredicts dust formation times","Non-equilibrium polymer model picks Al2O3 as hot seed"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Every cluster in the model grows by combining whole copies of one fixed molecule (Al2O3, TiO2, MgO, or SiO) through barrierless hard-sphere collisions, and the paper itself concedes that nucleation by monomer-multiples of a fixed stoichiometry \"is not established either\" (Sec. 6.1.1); if real dust forms through mixed-species or non-stoichiometric steps, or through reactions with activation barriers, the identity of the first condensate and the temperature at which it appears could both change.","fun_headline_variants_meta":{"raw":{"variants":["Polymer nucleation speeds dust seeds, flips Al2O3 timing","Atomic mix yields only TiO2 dust; Al2O3 still argued as seed","Monomer restriction overpredicts dust formation times","Non-equilibrium polymer model picks Al2O3 as hot seed"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00095,"raw_usage":{"total_tokens":4176,"prompt_tokens":1192,"completion_tokens":2984,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":808,"completion_tokens_details":{"reasoning_tokens":2909}},"tokens_in":808,"tokens_out":2984,"duration_ms":21537,"temperature":1.0,"reasoning_tokens":2909,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:06:33.709341+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure or recompute the rate coefficients for the two reactions that are the only available paths to the Al2O3 monomer, AlO + AlO + M -> Al2O2 + M and Al2O2 + O + M -> Al2O3 + M, at 1500-2500 K. If experiment or higher-level theory places these rates orders of magnitude above the combustion-derived values used here, the comprehensive model's failure to form Al2O3 is the data gap the paper claims; if the rates are right, the claim that Al2O3 is the first condensate in AGB winds fails, because its monomer cannot arise from an atomic gas and, by the model's own logic, only TiO2 clusters would seed dust.","supporting_citations":[{"cited_title":"S., Winters J","cited_arxiv_id":null,"evidence_quote":"Steady-state nucleation model claiming TiO2 as the only efficient AGB nucleator; its 1000 K cut-off is reproduced but its steady-state assumption is shown to be invalid."},{"cited_title":"T., 2017, @doi [Nanoscale] 10.1039/C6NR05788H , 9, 1049","cited_arxiv_id":null,"evidence_quote":"Supplies the TiO2 cluster global-minimum structures used for the DFT Gibbs free energies of the (TiO2)n ladder."}],"review_version":1}