{"id":"d1561e23-babb-48e2-9f16-2a35c40de400","arxiv_id":"1908.03931","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"In Al-Cu-Fe melts, chemical short-range order changes near the icosahedral quasicrystal composition, coinciding with minima in viscosity and undercooling.","lead":"Experiments and computer simulations on Al-Cu-Fe alloys show that the liquid's internal chemistry changes at the composition that forms a quasicrystal, and the liquid becomes easier to undercool there. This suggests melt viscosity and undercooling could help identify compositions for making quasicrystal materials.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Viscosity 'minimum' at i-phase stoichiometry is not statistically robust: Table 1 Arrhenius fits put xCu=20.5 within ~0.5% of xCu=25.5 at relevant isotherm temperatures, so the experimental anchor of the central claim is not established.","rationale":"The reader's conditional verdict is appropriate, and the most load-bearing point is the empirical existence of the viscosity and undercooling minima at the i-phase stoichiometry. The paper's own Table 1 permits a direct check, and that check shows the minimum is not uniquely pinned to xCu=25.5: at 1473 K the xCu=20.5 and xCu=25.5 viscosities agree to about 0.2%, far below the stated 5% total error, and the relative ordering changes with temperature. Consequently the claim that chemical-SRO changes near i-phase stoichiometry lead to these minima lacks a statistically established experimental correlate. I did not select the undercooled-representation issue as primary, because even if the AIMD at 100 K above Tm were fully representative, the experimental minima would still need to be real; the non-equilibrated undercooled AIMD is explicitly acknowledged in Section 4 and is an honest limitation rather than a hidden assumption. My proposed check would settle whether the minima survive error propagation; if they do not, the paper should be revised to temper the causal claim. This sharpens the condition needed for acceptance but does not move the reader's conditional verdict.","tokens_in":18358,"tokens_out":15690,"duration_ms":134827,"concrete_test":"Use the Table 1 Arrhenius fits to recompute the xFe=12.5 viscosity isotherms at 1373, 1473, 1573, and 1673 K, propagate the stated 5% total error, and test whether the value at xCu=25.5 is significantly lower than at xCu=20.5 and xCu=23.0, e.g., by constructing confidence intervals on the fitted Arrhenius parameters or a pairwise comparison. If the differences are within the combined error, or if the minimum location depends on temperature, the claimed viscosity minimum at i-phase stoichiometry and the inferred SRO linkage are not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central applied claim is that a chemical-SRO change near i-phase stoichiometry produces minima in viscosity and undercooling isotherms (Abstract; Section 4). The experimental anchor is a 'weakly pronounced minimum' at xCu≈25 at.% (Fig. 2a). The Methods state a total viscosity error of no more than 5%, and figure captions show no error bars. Re-evaluating the Arrhenius parameters in Table 1 for the xFe=12.5 cross-section at 1473 K gives ν(Al67Cu20.5Fe12.5)=5.47×10^-7 m2/s and ν(Al62Cu25.5Fe12.5)=5.48×10^-7 m2/s, a difference of about 0.2%; at 1573 K the difference is about 0.5%, and at 1673 K about 2%. Thus the supposed minimum at the i-phase composition is not unique: xCu=20.5 is statistically indistinguishable from the claimed minimum, and at lower temperatures the computed minimum location shifts toward xCu≈20.5. With the stated 5% error, the feature at the i-phase composition cannot be distinguished from neighboring compositions. Since the same data anchor the undercoolability and liquidus comparisons and the inferred chemical-SRO correlation, the central claim currently rests on an experimentally unestablished feature.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies Al-Cu-Fe melts along two composition cross-sections containing the icosahedral quasicrystal (i-phase) stoichiometry. The authors report experimental kinematic viscosity and undercoolability data across a wide composition range, and combine these with ab initio molecular dynamics (AIMD) simulations of six compositions at about 100 K above the liquidus. From the simulations they extract Warren-Cowley short-range order (SRO) parameters, radial and bond-angle distribution functions, Voronoi polyhedra, and bond-orientational order parameters. The central claim is that the main features of interatomic interaction in Al-Cu-Fe are the same in liquid and solid states, and that a change in chemical SRO near the i-phase stoichiometry produces minima in viscosity and undercoolability isotherms, suggesting that melt SRO and structural-sensitive properties can serve as indicators of quasicrystal-forming compositions.","tokens_in":1840,"tokens_out":3663,"duration_ms":81194,"significance":"If established, the proposed connection between melt chemical SRO, viscosity, undercoolability, and quasicrystal-forming ability would be practically useful for selecting alloy compositions for casting or rapid quenching. The paper contributes a substantial experimental dataset of viscosity and undercoolability for Al-Cu-Fe melts over a broader composition range than previous studies, and the AIMD analysis addresses an important question about whether icosahedral order survives melting. The finding that topological icosahedra are almost absent in the equilibrium melt while pair and angular correlations persist is interesting and appears internally consistent. However, the load-bearing experimental claim of viscosity minima at the i-phase stoichiometry is not robust to the stated measurement uncertainty, and the claim is not uniformly supported across both cross-sections. The equal-viscosity line construction also introduces a nearly tautological comparison with the liquidus. The work would be publishable if the central claims are re-evaluated and appropriately qualified.","major_comments":[{"comment":"The claimed 'weakly pronounced minimum' in viscosity at xCu ≈ 25 at.% is not statistically established by the data in Table 1. Re-evaluating the Arrhenius parameters for the xFe = 12.5 cross-section at 1473 K gives ν(Al67Cu20.5Fe12.5) = 5.49×10^-7 m²/s and ν(Al62Cu25.5Fe12.5) = 5.50×10^-7 m²/s, a difference of about 0.2%; at 1573 K the difference is about 0.5%, and at 1673 K about 1.7%. All of these differences are far smaller than the stated total error of 5%, and Fig. 2a shows no error bars. The authors themselves call the minimum weak, so the central experimental anchor of the 'minima at i-phase stoichiometry' claim is not supported by the reported measurements.","section":"§3, Fig. 2a and Table 1"},{"comment":"The blanket statement that viscosity and undercoolability isotherms develop minima at the i-phase stoichiometry is contradicted by the Fe cross-section. Along xCu = 25.5 at.%, Fig. 2b shows viscosity increasing monotonically with xFe, and Fig. 3d shows undercoolability with only a kink at xFe ≈ 12.5, not a minimum. The Conclusions state that 'All of these characteristics develop minima at concentration corresponding to i-phase stoichiometry', which overstates the reported data. The claims should be restricted to the Cu cross-section, or the text should explicitly characterize the Fe cross-section as showing monotonic increase and a kink.","section":"Abstract and §5 Conclusions, compared with Figs. 2b and 3b"},{"comment":"The lines of equal viscosity T_visc(x) are constructed using a reference viscosity of 7.5×10^-7 m²/s taken from the i-phase near its melting point, and the text states that most melts have approximately this same viscosity at their own melting points. Consequently, the observed coincidence T_visc(x) ≈ Tm(x) is largely a consequence of the construction rather than an independent physical finding. The claim that this coincidence supports structural heredity between liquid and solid is therefore over-interpreted and should be reframed as a restatement of the near-constancy of viscosity at the liquidus, not as new evidence.","section":"§3, Fig. 3"},{"comment":"The only undercooled-liquid AIMD simulation (Al52Cu25.5Fe22.5 at 1000 K) is explicitly described as not equilibrated, having been cooled from 1600 K over 10,000 fs and relaxed for only 5,000 fs. This is the sole direct simulation evidence connecting SRO to the undercooled state from which solidification begins. Since the central argument requires that SRO changes near the i-phase stoichiometry affect the initial stage of solidification, the conclusion rests on a non-equilibrium simulation of a single off-stoichiometry composition. Equilibrated undercooled-structure data, or at least a clear quantitative statement of the relaxation limitations, are needed to support that inference.","section":"§4, Fig. 7 caption and text"}],"minor_comments":[{"comment":"There are several typographical errors: 'udercoolability' should be 'undercoolability', and in the Abstract 'bong-angle distribution function' should be 'bond-angle distribution function'.","section":"Abstract and Highlights"},{"comment":"The text refers to 'Fig. 7(a-d)' twice for two different temperatures (1600 K and 1000 K); the figure actually has panels (a)-(h). The first reference should be to panels (a-d) and the second to panels (e-h).","section":"§4, Fig. 7"},{"comment":"The sentence 'A high-temperature configuration of the system at T = 1600 K was cooled down to 10,000fs forandthenrelaxedatthistemperaturefor5,000fs' is grammatically broken and should read 'cooled down to 1000 K for 10,000 fs and then relaxed at this temperature for 5,000 fs'.","section":"§4, undercooling simulation paragraph"},{"comment":"The r(Fe-Fe) values in Table 2 vary erratically (2.98, 2.28, 2.23, 2.90, 2.63, 2.34 Å) without a clear trend. The authors note low accuracy for iron-related quantities, but the table should include an explicit statement of estimated uncertainties or a footnote explaining the scatter.","section":"Table 2"}],"recommendation":"major_revision","confidential_remarks":"The paper's advertised central claim that viscosity and undercoolability minima appear exactly at the i-phase stoichiometry is not supported by the reported experimental data once the stated 5% viscosity error is considered, and it is contradicted by the monotonic behavior along the Fe cross-section. The authors should either supply a statistically rigorous treatment (error bars, confidence intervals, or a significance test of the minimum) or substantially revise the abstract, highlights, and conclusions to match what the data actually show. The equal-viscosity line construction should also be repositioned as a consequence of near-constant viscosity at the liquidus rather than as an independent confirmation. The AIMD structural results on chemical SRO are more convincing and could carry the paper if the experimental claims are appropriately qualified."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the AIMD SRO analysis is the real value here; the headline claim about viscosity and undercooling minima at the i-phase stoichiometry is not statistically supported. I re-ran the stress-test calculation on their Table 1 Arrhenius fits, and it holds: at xFe=12.5, the viscosities at xCu=20.5 and xCu=25.5 differ by roughly 0.2% at 1473 K, 0.5% at 1573 K, and 2% at 1673 K. With their stated 5% total error, the 'weakly pronounced minimum' is indistinguishable from scatter, and the minimum's location is not determined. The undercooling data show a kink rather than a clean minimum, and no error bars are given. So the abstract's causal chain – chemical SRO change leads to minima – is plausible but not established by the experiments.\n\nWhat is new and useful: two composition cross-sections of viscosity and undercooling for Al-Cu-Fe, plus AIMD at six compositions with Warren-Cowley parameters, Voronoi analysis, BADFs, and BOOPs. The Cu-Fe repulsion, the near-absence of perfect icosahedra in the equilibrium melt, and the dominance of distorted Kasper polyhedra are concrete, re-usable results. The authors are also honest: they call the minimum weak, and they explicitly say the deep-undercooled AIMD run is not equilibrated. That is the right way to report a preliminary observation.\n\nThe weaknesses are proportionate: the reference viscosity for the equal-viscosity lines is taken from the i-phase itself, so the T_visc–liquidus match is partly constructed. The AIMD is run at 100 K above Tm, not in the undercooled state, so the connection to nucleation is inferential. The Fe statistics are thin at low Fe content, as they note. None of this makes the paper worthless; it just means the central claim should be demoted from a result to a hypothesis in the abstract.\n\nThis paper deserves a serious referee, not a desk reject. The data and simulation protocols are described well enough to reproduce, and the SRO findings stand on their own. I would want the experimental minimum either confirmed with proper error bars and more compositions or removed from the abstract and conclusions. I'd cite it for the AIMD SRO numbers, not for the minima claim. Probably worth a reading group slot as a case study in how a weak experimental feature can become a headline.","headline":"Careful AIMD and viscosity study of Al-Cu-Fe melts, but the headline 'minima at the i-phase' claim is not statistically supported by the experimental data.","tokens_in":19201,"tokens_out":2668,"would_cite":true,"duration_ms":37650,"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":"The paper argues that Al-Cu-Fe melts retain the chemical short-range order of the quasicrystal solid—Cu–Fe repulsion and Fe–Al attraction—while losing its topological icosahedra, and that this chemical memory shows up as minima in…","keywords":["liquid alloy","viscosity","undercoolability","short-range order","Kasper polyhedra","structural heredity","quasicrystal","ab initio molecular dynamics"],"falsifier":"Measure the viscosity and undercoolability along the 12.5 at.% iron cross-section with per-point error bars and multiple independent runs: if the minimum near 25 at.% copper disappears within the 5% scatter or shifts with cooling rate, the claimed composition-structure link fails. Independently, in situ diffraction on levitated undercooled droplets at 20–80 K undercooling could test whether the Warren-Cowley parameter changes predicted from equilibrium melts are actually present in the liquid that nucleates.","tokens_in":18169,"feed_emoji":"","tokens_out":7854,"duration_ms":76499,"temperature":0.7,"pith_summary":"The paper sets out to test whether a liquid carries structural memory of the quasicrystal it freezes into, using Al-Cu-Fe as a model system. By combining viscosity and undercoolability measurements with ab initio molecular dynamics, it argues that the main chemical interaction features of the solid—strong Cu–Fe repulsion and strong Fe–Al attraction—survive in the melt, while the solid's topological icosahedra do not. It further claims that at the icosahedral-phase stoichiometry the chemical short-range order changes qualitatively, and that this change shows up as minima in the viscosity and undercoolability isotherms. If correct, the result makes melt viscosity and undercoolability practical indicators for finding or processing quasicrystal-forming compositions.","feed_headline":"Al-Cu-Fe melts keep chemical order after icosahedra vanish","feed_subtitle":"Viscosity and undercooling dip at the quasicrystal composition, tracking chemical order that survives melting.","key_machinery":"The argument turns on the Warren-Cowley parameters $\\alpha_{i-j}=1-\\frac{Z_{i-j}}{Z_{i,\\text{tot}}\\,x_j}$, which measure whether pairs of species attract (negative $\\alpha$) or repel (positive $\\alpha$) relative to a random mixture. Computed from ab initio partial radial distribution functions and coordination numbers, these parameters supply the concentration-resolved chemical short-range order that is then compared with experimental viscosity, equal-viscosity lines, and undercoolability. Bond-angle distribution functions around Fe—a sharp Al–Fe–Al peak and a weak Cu–Fe–Cu peak—back up the same interaction picture, and Voronoi tessellation identifies the polytetrahedral Kasper-polyhedra motif that replaces perfect icosahedra in the melt.","core_discovery":"The central claim is that structural heredity in Al-Cu-Fe is chemical rather than topological. Melt snapshots from ab initio molecular dynamics contain almost no perfect $\\langle 0,0,12,0\\rangle$ icosahedra even at the icosahedral-phase composition; the local order is polytetrahedral, dominated by distorted Kasper polyhedra such as $\\langle 0,3,6,4\\rangle$. Yet Warren-Cowley parameters extracted from partial radial distribution functions show the same interaction fingerprints as the solid: Cu and Fe avoid each other, Fe and Al bond strongly, and these features are nearly concentration-independent. In the composition window of the i-phase, the chemical short-range order changes qualitatively—Al–Cu and Al–Fe interactions flip sign, Cu–Fe repulsion is minimal—and the same window shows minima in viscosity isotherms, equal-viscosity temperature lines, and undercoolability. The paper concludes that the chemical short-range order of the melt, rather than pre-formed icosahedra, is what links melt properties to quasicrystal formation.","pith_inferences":["The same recipe—measuring a structural-sensitive liquid property and computing Warren-Cowley parameters across a composition cross-section—could be applied to other quasicrystal-forming or high-entropy alloy families, treating viscosity minima as a fast screening signal before full phase-diagram work.","AIMD's non-equilibrated undercooled run hints that substantial icosahedral order appears only on undercooling; if that tendency survives equilibration, the melt may develop topological icosahedra just before nucleation even though the equilibrium melt has none—an extension the paper explicitly leaves to future semi-empirical potentials.","A direct testable extension would be to compare these melt indicators with nucleation-rate measurements, for example from droplet dispersion or fluxing, to see whether the viscosity and undercoolability minima correspond to lower nucleation barriers or simply to liquidus-shape effects."],"forward_implications":["Concentration curves of viscosity, equal-viscosity temperatures, and undercoolability can be used as fast experimental screens for the icosahedral-phase stoichiometry in Al-Cu-Fe, without costly structural probes.","Because topological icosahedra are absent in the equilibrium melt, the initial stage of solidification is better interpreted through chemical short-range order and its concentration changes than through pre-existing icosahedral clusters.","The near concentration-independence of Cu–Fe repulsion and Fe–Al attraction means these interaction fingerprints are robust signatures of the system, useful for validating interatomic potentials in simulation.","At i-phase stoichiometry, minimal chemical interaction—near-random Al–Cu and Cu–Cu bonding, sign flips in Al–Fe—coincides with the highest viscosity near the liquidus; this can guide composition selection for casting."],"supporting_citations":[{"why":"Earlier study of phase selection and solidification conditions in quasicrystal-forming Al-Cu-Fe; supplies the undercooling behavior this paper extends and compares with.","marker":"[24]"},{"why":"Viscosity measurement of Al63Cu25Fe12 near the i-phase; the single previously studied composition anchoring the new wider survey.","marker":"[44]"},{"why":"Viscosity of binary Al-Fe melts in the Al-rich range; the baseline against which the sharp rise of Al-Cu-Fe viscosity with iron content is judged.","marker":"[47]"},{"why":"Introduces the Warren-Cowley parameters used here to quantify chemical short-range order from partial coordination numbers.","marker":"[59]"},{"why":"Characterizes phonon density of states of the icosahedral phase Al62Cu25.5Fe12.5; supplies the solid-state interaction picture this paper compares with the melt.","marker":"[60]"},{"why":"Shows the relationship between structural and dynamic properties in Al-rich Al-Cu melts, including the short-range-order change near 25 at.% Cu; grounds the viscosity mechanism invoked for the ternary.","marker":"[79]"},{"why":"Reports the pre-peak in the liquid Al-Fe structure factor; supports the claim that structural ordering changes at the iron concentrations studied.","marker":"[80]"}],"fun_headline_variants":["No icosahedra in melt, but chemical order persists","Quasicrystal melt: chemical order outlives icosahedra","Melt keeps chemical bonds when icosahedra don't form","Al-Cu-Fe: chemical order survives melting, icosahedra don't","Chemical order, not icosahedra, links melt to quasicrystal"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole link rests on the claim that the weak viscosity minimum near 25 at.% copper is a real feature of the alloy rather than scatter within the stated 5% measurement error, and that melt structure measured 100 K above the melting point tells us what the undercooled liquid looks like when solidification begins.","fun_headline_variants_meta":{"raw":{"variants":["No icosahedra in melt, but chemical order persists","Quasicrystal melt: chemical order outlives icosahedra","Melt keeps chemical bonds when icosahedra don't form","Al-Cu-Fe: chemical order survives melting, icosahedra don't","Chemical order, not icosahedra, links melt to quasicrystal"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000709,"raw_usage":{"total_tokens":3252,"prompt_tokens":1063,"completion_tokens":2189,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":679,"completion_tokens_details":{"reasoning_tokens":2092}},"tokens_in":679,"tokens_out":2189,"duration_ms":14191,"temperature":1.0,"reasoning_tokens":2092,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:58:13.787006+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the viscosity and undercoolability along the 12.5 at.% iron cross-section with per-point error bars and multiple independent runs: if the minimum near 25 at.% copper disappears within the 5% scatter or shifts with cooling rate, the claimed composition-structure link fails. Independently, in situ diffraction on levitated undercooled droplets at 20–80 K undercooling could test whether the Warren-Cowley parameter changes predicted from equilibrium melts are actually present in the liquid that nucleates.","supporting_citations":[{"cited_title":"Holland-Moritz, J","cited_arxiv_id":null,"evidence_quote":"Earlier study of phase selection and solidification conditions in quasicrystal-forming Al-Cu-Fe; supplies the undercooling behavior this paper extends and compares with."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Viscosity measurement of Al63Cu25Fe12 near the i-phase; the single previously studied composition anchoring the new wider survey."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Viscosity of binary Al-Fe melts in the Al-rich range; the baseline against which the sharp rise of Al-Cu-Fe viscosity with iron content is judged."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the Warren-Cowley parameters used here to quantify chemical short-range order from partial coordination numbers."},{"cited_title":"Brand, G","cited_arxiv_id":null,"evidence_quote":"Characterizes phonon density of states of the icosahedral phase Al62Cu25.5Fe12.5; supplies the solid-state interaction picture this paper compares with the melt."},{"cited_title":"Jakse, A","cited_arxiv_id":null,"evidence_quote":"Shows the relationship between structural and dynamic properties in Al-rich Al-Cu melts, including the short-range-order change near 25 at.% Cu; grounds the viscosity mechanism invoked for the ternary."},{"cited_title":"Jingyu, B","cited_arxiv_id":null,"evidence_quote":"Reports the pre-peak in the liquid Al-Fe structure factor; supports the claim that structural ordering changes at the iron concentrations studied."}],"review_version":1}