{"id":"53a7031c-da85-4721-9f47-fd39bf506050","arxiv_id":"1908.02099","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A new non-equilibrium partition coefficient is derived near the transition to diffusionless solidification, and it differs from an earlier LNM formula at high V_D/V_0.","lead":"This paper derives new boundary conditions and a new solute partition coefficient for rapid solidification of binary alloys when the interface moves near the diffusion speed. The new formula behaves differently from an earlier local nonequilibrium model when the ratio V_D/V_0 is large.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The partition coefficient (28) depends entirely on the uncalibrated O(1) coefficients alpha_A, alpha_B introduced in Eq. (15); without independent justification of that linearization, the claimed marked deviation from Eq. (1) is not a robust prediction.","rationale":"The reader and I converge on the same point: the derivation from Eq. (15) through Eq. (28) is internally consistent once the Eq. (24) sign is corrected, and the remaining question is whether the assumed interfacial linear response is physically correct. Because the result is a one-parameter family with A unspecified, the quantitative claim that Eq. (28) differs markedly from Eq. (1) is not yet established; however, the paper is transparent about the approximation, so a conditional acceptance is the right level. I see no reason to reject and no external contradiction: the interface, unlike the bulk, can admit a linear dependence on the normal diffusion current, and the positivity argument after Eq. (26) is coherent. The main weakness is that the final prediction is not parameter-free, which is exactly the kind of assumption a conditional verdict should require to be justified or measured. I therefore leave the reader's verdict unchanged.","tokens_in":7818,"tokens_out":12681,"duration_ms":135837,"concrete_test":"Fit Eq. (28) to published MD or experimental k(V) data for a rapid-solidification alloy where V_D and V_0 are independently known (e.g., Ni-Cu or Si-As from Refs. [13,14,17,34-39]), treating A = alpha_A + alpha_B c0 as the single free parameter and using only data with V/V_D between about 0.7 and 1. If the best-fit A lies outside, say, 0.1-10, or the fit cannot reproduce the near-threshold curvature, the O(1) linearization in Eq. (15) is falsified. As an independent check, derive alpha_A and alpha_B from an atomistic or phase-field interface model and verify that A is indeed O(1).","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central derivation hinges on Eq. (15), where the interfacial liquid chemical potentials are written as local-equilibrium values minus alpha_i RT/(rho V_DI M) J_D^L, with alpha_i declared dimensionless and of order unity but otherwise unspecified. Substituting this linearization into the interface driving-force condition yields Eq. (25) and then, using C_S = c0 and solute conservation, the partition coefficient Eq. (28), whose velocity dependence enters only through the combination A = alpha_A + alpha_B c0. No derivation, measurement, or microphysical estimate of A is given; the comparison in Figs. 1-2 sets alpha_A = 1, alpha_B = 0 for illustration. The claimed marked difference from the earlier coefficient Eq. (1) at larger V_D/V_0 is therefore a consequence of assuming A ≈ 1, not of the boundary-condition structure alone: if A were 0.1 or 10, the near-threshold slope of k(V) would change by an order of magnitude and the curves in Fig. 2 would shift substantially. This is a load-bearing modeling assumption, not a parameter-free prediction. The separate sign typo in Eq. (24), minus instead of plus before c0 Delta_mu*_B, does not affect Eq. (28) once corrected, but it should be fixed for the derivation to be checkable.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript derives interface boundary conditions for rapid solidification of a binary mixture in the local nonequilibrium model (LNM), focusing on the velocity range just below the diffusion speed V_D at which diffusionless solidification sets in. The authors expand the interface chemical potentials linearly in the interfacial diffusion current J_D^L (Eq. 15), combine the result with entropy production and mass conservation conditions to obtain boundary conditions (25), and derive a new nonequilibrium solute partition coefficient k(V) in Eq. (28). The new coefficient is compared with the earlier LNM expression (1), and the paper reports that the two differ markedly when the ratio V_D/V_0 is large.","tokens_in":8156,"tokens_out":6990,"duration_ms":62859,"significance":"If the central result holds, the paper provides a new thermodynamically motivated prediction for solute trapping in the high-speed regime that is potentially testable against experimental rapid-solidification data or molecular-dynamics simulations. The derivation is transparent and builds on a standard entropy-production framework, and the algebraic route from Eq. (15) to Eq. (28) is mostly consistent. The main limitations are the undetermined order-unity coefficients alpha_A and alpha_B and the dilute approximations, which leave the quantitative predictions conditional on an uncalibrated parameter.","major_comments":[{"comment":"Equation (24) states (1-c0)Delta_mu*_A - c0 Delta_mu*_B = (RT/M) ln(1-VD/V0), but insertion of Eqs. (22)-(23) into Eq. (13) and use of the reference-state condition yields Eq. (25) only if the left-hand side is (1-c0)Delta_mu*_A + c0 Delta_mu*_B. The minus sign is therefore a sign error that must be corrected; with the printed sign the derivation does not close.","section":"Section 2, Eq. (24)"},{"comment":"The partition coefficient (28) depends on the combination alpha_A + alpha_B c0 introduced through the linearized interface chemical potentials in Eq. (15). No microscopic derivation, numerical estimate, or experimental calibration of alpha_i is provided, and the figures arbitrarily set alpha_A = 1 and alpha_B = 0. Since this combination multiplies the velocity-dependent term in the denominator, the claimed marked deviation from Eq. (1) at large V_D/V_0 is not robust against order-unity variations of alpha_i; the authors should justify these coefficients or provide a sensitivity analysis.","section":"Section 2, Eq. (15) and Section 3, Eq. (28)"},{"comment":"The comparison of k1 and k2 is made for a single choice of alpha_A and alpha_B. Because the central conclusion about sufficiently different behavior rests on this choice, the paper should show curves for several values of alpha_A + alpha_B c0 (for example 0.1, 1, and 10) to demonstrate whether the qualitative conclusion persists.","section":"Section 3, Figs. 1 and 2"}],"minor_comments":[{"comment":"The sentence 'the solute transfer in the liquid bhas no time to occur' contains a typo ('bhas' should be 'has').","section":"Abstract and Introduction"},{"comment":"Reference [1] contains typos in 'Fandamentals' and '2rd ed.'; these should be corrected.","section":"Introduction, Reference [1]"},{"comment":"The phrase 'the sign of which will be discussed later' is never followed by an explicit discussion of the sign of alpha_i; please add such a discussion or remove the promise.","section":"Section 2, after Eq. (15)"},{"comment":"The condition V < V_D should be stated explicitly when Eq. (28) is introduced, along with the statement that k = 1 for V >= V_D, to avoid the apparent singularity of the logarithm for V > V_D.","section":"Section 3, Eq. (28)"},{"comment":"The caption has minor grammatical issues ('coefficients k1 ... is shown') and should be reworded.","section":"Fig. 1 caption"}],"recommendation":"major_revision","confidential_remarks":"The paper relies heavily on the authors' own prior LNM publications for the reference state and for the parameters V_D and V_DI, but this is a natural extension of a well-established framework rather than an inappropriate citation pattern. The main editorial concern is whether the undetermined alpha_i coefficients are acceptable for the target journal; if the authors supply a sensitivity analysis or an independent estimate, the paper would be substantially stronger."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a short, careful derivation of a new solute partition coefficient for rapid solidification just below the diffusionless transition. The result, Eq. (28), is genuinely new relative to the earlier LNM formula, and the boundary-condition derivation is clean. But the predicted k(V) depends on an undetermined O(1) coefficient alpha_A + alpha_B c0 from the linear-in-current expansion (15), so the claimed deviation from the prior formula is conditional on that coefficient's value, not a parameter-free prediction.\n\nWhat is new: the paper starts from entropy production and the sharp-interface driving force, expands the interfacial chemical potentials linearly in the solute diffusion current, and derives modified boundary conditions (25) and then the partition coefficient (28). That linear-in-current step is a departure from the authors' earlier work, which kept only J^2 terms. The algebra is mostly transparent and the limiting checks (k→1 at V→VD, correct dilute reduction) are sensible. No circularity: the derivation uses the kinetic relation and entropy production, not the target k(V).\n\nSoft spots: Eq. (24) has a sign typo (minus should be plus before c0 ∆µ*_B); it doesn't affect Eqs. (25)-(28) but should be fixed. The bigger issue is the alpha_i. They are declared dimensionless and O(1), with alpha_A=1, alpha_B=0 for the figures, but no microphysical estimate, MD, or experimental calibration is given. Since the velocity dependence in Eq. (28) enters only through A = alpha_A + alpha_B c0, the difference from Eq. (1) shown in Figs. 1-2 is essentially the assumption A≈1. If A were 0.1 or 10, the curves would shift substantially. That doesn't make the derivation wrong, but it makes the headline claim 'differs markedly' a conditional statement, not a robust prediction. The dilute-solution and CS≈c0 approximations are also stated but not quantified.\n\nNet: a solid, modest modeling contribution. It deserves peer review, with requested fixes: correct the typo, discuss sensitivity to alpha, and ideally provide a way to estimate or fit alpha. I'd send it to a rapid-solidification venue. I wouldn't build on Eq. (28) in my own work until A is pinned down.","headline":"New partition coefficient near diffusionless solidification, but the claimed deviation from the earlier LNM formula rests on an uncalibrated O(1) coefficient.","tokens_in":8631,"tokens_out":3943,"would_cite":false,"duration_ms":37243,"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":"New interface boundary conditions give a solute partition coefficient that rises sharply near the transition to diffusionless solidification.","keywords":["solute trapping","solute partition coefficient","local nonequilibrium model","diffusionless solidification","rapid solidification","interface boundary conditions","binary alloys","Maxwell-Cattaneo equation"],"falsifier":"For a dilute alloy with $V_D/V_0\\approx 0.5$, measure the partition coefficient at $V=0.9V_D$ and $V=0.99V_D$. Equation (28) predicts a sharp rise toward $k=1$ across that interval, whereas the earlier coefficient (1) predicts a gradual rise; observing the gradual rise would falsify the linearization behind Eq. (28).","tokens_in":7647,"feed_emoji":"🧪","tokens_out":8734,"duration_ms":78352,"temperature":0.7,"pith_summary":"The paper derives boundary conditions for a solid-liquid interface moving just below the diffusion speed $V_D$, the velocity at which the local nonequilibrium model predicts complete solute trapping and diffusionless solidification. It obtains a new non-equilibrium solute partition coefficient $k(V)$, Eq. (28), that gives the ratio of solid to liquid solute concentrations at the interface. In the high-speed region the new coefficient behaves very differently from the earlier LNM coefficient when $V_D/V_0$ is large, rising sharply to $k=1$ at $V=V_D$. Because the partition coefficient controls the formation of supersaturated solid solutions during rapid solidification, getting this high-speed branch right matters for predicting the resulting microstructures.","feed_headline":"Solute-trapping law shifts sharply just below the diffusion speed","feed_subtitle":"Near the speed of diffusion, the new interface conditions make the partition coefficient rise steeply to 1.","key_machinery":"The load-bearing object is the linear constitutive ansatz (15), $$\\mu_i^L(C_L,J_D^L)=\\mu_{\\mathrm{leq},i}^L(C_L)-\\alpha_i\\frac{RT}{\\rho V_{DI} M}J_D^L,$$ with dimensionless coefficients $\\alpha_i$ of order unity and $V_{DI}$ the atomistic diffusive speed at the interface. This ansatz converts the entropy-production balance into boundary condition (25). Setting $C_S=c_0$ and using solute conservation (14) then yields Eq. (28) for $k(V)$. The argument's force is that the unknown interface kinetics enter only through the single combination $\\alpha_A+\\alpha_B c_0$, so the sharp rise of $k(V)$ near $V_D$ is a direct consequence once that parameter is fixed.","core_discovery":"Starting from the entropy production at a sharp interface and the local nonequilibrium premise that bulk liquid diffusion is out of local equilibrium, the paper expands the interfacial chemical potentials linearly in the solute diffusion current $J_D^L$. Combining the resulting Gibbs-energy balance with solute conservation gives the boundary condition (25) and, in the dilute limit $C_S\\approx c_0$, the partition coefficient $$k(V)=\\frac{c_0(\\alpha_A+\\alpha_B c_0)V/V_{DI}}{c_0(\\alpha_A+\\alpha_B c_0)V/V_{DI}+\\ln\\left(\\frac{1-V/V_0}{1-V_D/V_0}\\right)}.$$ Near $V=V_D$ this coefficient rises steeply to unity; compared with the partition coefficient (1) from the earlier 2007 work, it differs substantially as $V_D/V_0$ grows, while the two agree at small $V_D/V_0$. The paper's claim is that in the high-speed region the partition coefficient is controlled by macroscopic boundary conditions at the interface rather than by the details of atomic attachment kinetics.","pith_inferences":["Because only the sum $\\alpha_A+\\alpha_B c_0$ enters the final formula, the model has effectively one free parameter; fitting it to a single measured $k(V)$ curve would determine the entire high-speed branch.","Atomistic simulations that compute interfacial chemical potentials under an imposed solute flux could test the linear ansatz (15); a nonlinear response would require higher-order terms.","Carrying the expansion to second order in $J_D^L$ would recover the $J_D^2$ term used in earlier local nonequilibrium treatments and give corrections to Eq. (28) at intermediate velocities.","The derivation's dilute-solution and $C_S\\approx c_0$ assumptions suggest a natural extension to concentrated alloys, where the logarithmic terms in Eqs. (20)-(21) and the $C_S$ dependence in Eq. (25) must be kept."],"forward_implications":["If Eq. (28) is correct, the partition coefficient reaches 1 at the finite speed $V_D$, so complete solute trapping is a finite-velocity transition rather than an asymptotic limit.","At large $V_D/V_0$, the new coefficient rises far more steeply than the earlier 2007 coefficient near $V_D$, changing predicted solid compositions in rapid resolidification.","The high-speed partition coefficient should be nearly independent of the atomic attachment mechanism, because it follows from macroscopic boundary conditions.","At small $V_D/V_0$ the new and old coefficients nearly coincide, so experimental and atomistic-simulation data consistent with the older coefficient remain consistent in that regime."],"supporting_citations":[{"why":"Supplies the kinetic relation $V=V_0(1-e^{M\\Delta G_m/RT})$ used to fix the nonlinear function $f(V)$ in the interface Gibbs-energy balance.","marker":"[42]"},{"why":"Provides the sharp-interface entropy production expression used to set up the thermodynamic boundary conditions.","marker":"[40,41]"},{"why":"Gives the earlier local nonequilibrium partition coefficient (1) that the new coefficient (28) is compared against in the high-speed region.","marker":"[29]"},{"why":"Introduced the local nonequilibrium partition coefficient and the finite-velocity transition to complete trapping at $V_D$.","marker":"[22,23]"},{"why":"Earlier treatment of $\\Delta G_m$ depending on $J_D^2$, which the present linear-current expansion supplements.","marker":"[4]"},{"why":"Establishes the local nonequilibrium model for rapid solidification with $V_D$ as the transition speed.","marker":"[25]"},{"why":"One of the experimental works showing complete solute trapping at finite velocity, motivating the high-speed analysis.","marker":"[17]"}],"fun_headline_variants":["Partition coefficient climbs sharply as speed nears diffusion limit","Interface conditions steepen partition coefficient near V_D","Solute trapping rises sharply just below diffusion speed","New boundary conditions shift solute partition near diffusionless transition"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The prediction stands or falls with the assumption that the chemical potential at the interface changes in direct proportion to the solute diffusion current, with two unknown coefficients of order one; if that proportionality or its size is wrong, the partition coefficient changes.","fun_headline_variants_meta":{"raw":{"variants":["Partition coefficient climbs sharply as speed nears diffusion limit","Interface conditions steepen partition coefficient near V_D","Solute trapping rises sharply just below diffusion speed","New boundary conditions shift solute partition near diffusionless transition"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000692,"raw_usage":{"total_tokens":3114,"prompt_tokens":909,"completion_tokens":2205,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":525,"completion_tokens_details":{"reasoning_tokens":2143}},"tokens_in":525,"tokens_out":2205,"duration_ms":30265,"temperature":1.0,"reasoning_tokens":2143,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:54:05.234948+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a dilute alloy with $V_D/V_0\\approx 0.5$, measure the partition coefficient at $V=0.9V_D$ and $V=0.99V_D$. Equation (28) predicts a sharp rise toward $k=1$ across that interval, whereas the earlier coefficient (1) predicts a gradual rise; observing the gradual rise would falsify the linearization behind Eq. (28).","supporting_citations":[{"cited_title":"Turnbull, J","cited_arxiv_id":null,"evidence_quote":"Supplies the kinetic relation $V=V_0(1-e^{M\\Delta G_m/RT})$ used to fix the nonlinear function $f(V)$ in the interface Gibbs-energy balance."},{"cited_title":"Galenko , Phys","cited_arxiv_id":null,"evidence_quote":"Gives the earlier local nonequilibrium partition coefficient (1) that the new coefficient (28) is compared against in the high-speed region."},{"cited_title":"Galenko, V","cited_arxiv_id":null,"evidence_quote":"Earlier treatment of $\\Delta G_m$ depending on $J_D^2$, which the present linear-current expansion supplements."},{"cited_title":"Galenko and S","cited_arxiv_id":null,"evidence_quote":"Establishes the local nonequilibrium model for rapid solidification with $V_D$ as the transition speed."},{"cited_title":"Eckler, R","cited_arxiv_id":null,"evidence_quote":"One of the experimental works showing complete solute trapping at finite velocity, motivating the high-speed analysis."}],"review_version":1}