{"id":"2e9a6bac-6f18-446d-8366-251a980b881c","arxiv_id":"1908.03029","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"From a stochastic order parameter model, the interface temperature in steady heat conduction is derived to deviate from the equilibrium transition temperature in proportion to the heat flux and the difference in thermal conductivities.","lead":"This paper uses a noisy order parameter model to derive a variational rule fixing where a phase boundary sits while heat flows through a material. It predicts that the boundary temperature shifts away from the equilibrium transition temperature, creating a super-heated ordered or super-cooled disordered layer at the interface.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 1/3 coefficient in (VI.40) is fixed by an ad hoc boundary condition in the phenomenological Onsager step (V.7)-(V.10); the quantitative central claim is not derived from the stochastic model.","rationale":"The reader's weakest assumption identifies exactly the same load-bearing concern: the interface-region entropy production in Sec. V A is obtained from an Onsager ansatz with uncontrolled approximations, and the coefficient 1/3 is fixed by an ad hoc boundary condition rather than derived from the stochastic model. This concern directly targets the central quantitative claim (VI.40), because the effect vanishes entirely if the interface contribution is omitted, and the coefficient changes if the Onsager coefficients or the g-fixing condition are modified. The paper is transparent about this limitation, and the qualitative sign is independently supported by global thermodynamics, so the result is plausible but not quantitatively settled. The concrete numerical test would directly measure the coefficient C in the stochastic model and thus determine whether the phenomenological step is correct. Since the reader already assigned CONDITIONAL for precisely this reason, my stress-test does not change the verdict; it reinforces it.","tokens_in":40318,"tokens_out":19456,"duration_ms":185292,"concrete_test":"Simulate the stochastic model (II.65)-(II.67) with the explicit entropy functional of Appendix A in a quasi-one-dimensional domain with the non-equilibrium adiabatic boundary condition (II.73)-(II.74). For a sequence of small η (e.g., 10^-3, 10^-4, 10^-5), measure the stationary interface temperature θ* as a function of J and extract the coefficient C in θ* − Tc = −C J (1/κ_o − 1/κ_d) X_eq(1−X_eq). If C deviates from 1/3 by more than the statistical error, the phenomenological step in Sec. V A is quantitatively wrong and Eq. (VI.40) must be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is Eq. (VI.40): θ* − Tc = −(J/3)(1/κ_o − 1/κ_d) X_eq(1−X_eq). The entire quantitative content is the coefficient 1/3. That coefficient comes from the interface contribution I_int, which is computed in Sec. V not from the stochastic model but from a phenomenological Onsager description of energy exchange between the interface and the two bulk regions, Eqs. (V.7)-(V.10), with L_o = λ_o/(g X_−) and L_d = λ_d/(g(1−X_+)). The authors explicitly state that this description 'involves uncontrolled approximations' and is 'not yet derived from the stochastic model' (Sec. V A). The value g = 1/3 is then fixed by imposing, without physical derivation, that the inverse-temperature gap β^int_+ − β^int_− vanishes as X → 0 and X → 1. If the true interface contribution differs, the coefficient in front of J(1/κ_o − 1/κ_d)X_eq(1−X_eq) changes; for instance, global thermodynamics gives 1/2. Moreover, without any interface contribution the variational equation would give no shift at all (c = 0), so the entire effect is carried by this uncontrolled step. The qualitative sign may survive because the independent global-thermodynamics framework predicts the same sign, but the quantitative claim — the stated central result — is not established to the same standard as the rest of the derivation. This is precisely the weakest link, and the authors themselves flag it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript constructs a stochastic order-parameter model for a Z2 order-disorder transition coupled to heat conduction, and imposes a non-equilibrium adiabatic boundary condition in which the boundary energy flux is fixed while total energy is conserved. From the Zubarev-McLennan representation the authors derive a stationary distribution for single-interface configurations, decompose the excess entropy production into ordered-bulk, disordered-bulk, and interface contributions, and evaluate the bulk contributions from deterministic relaxation while estimating the interface contribution with a projected Onsager description. The resulting potential function V(X;E,J) leads to the main result (VI.40), θ*−Tc = −(J/3)(1/κ_o−1/κ_d) X_eq(1−X_eq), predicting a superheated ordered interface when κ_d>κ_o and a supercooled disordered interface when κ_d<κ_o, in qualitative agreement with the authors' earlier global thermodynamics.","tokens_in":40624,"tokens_out":8565,"duration_ms":98774,"significance":"The bulk calculation is clean and the variational structure is transparent; deriving a variational principle for phase coexistence under heat flow from a stochastic model would be a significant conceptual advance. The authors are explicitly honest about the weak point: in Sec. V A they state that Eqs. (V.7)–(V.10) \"involve uncontrolled approximations\" and are \"not yet derived from the stochastic model.\" Because the entire numerical coefficient 1/3 in (VI.40) is fixed through this uncontrolled step and through the boundary-condition choice g=1/3, the paper at present establishes the qualitative sign of the effect but not the quantitative prediction. I regard this as a valuable framework whose headline claim needs either additional derivation or an explicit downgrading to a conjecture.","major_comments":[{"comment":"The central quantitative result (VI.40) is carried by the interface contribution I_int, whose evaluation in Sec. V A rests on the phenomenological Onsager description (V.7)–(V.10) with diagonal coefficients L_o=λ_o/(g X_-) and L_d=λ_d/(g(1−X_+)). The authors themselves state that this description involves uncontrolled approximations and has not been derived from the stochastic model. The value g=1/3 is then imposed by requiring β_+^int−β_-^int to vanish as X→0 and X→1, not derived. If the true interface contribution differs (for example, if off-diagonal Onsager terms contribute, or if g depends on X), the coefficient multiplying J(1/κ_o−1/κ_d)X(1−X) changes: with no interface contribution it is 0, and the independent global-thermodynamics estimate is 1/2. Thus the numerical coefficient, which is the quantitative content of (VI.40), is not established. I recommend either deriving the interface contribution from the stochastic model or explicitly presenting (VI.40) as a qualitative/conjectural prediction and marking the 1/3 factor as a model-dependent estimate.","section":"Sec. V A, Eqs. (V.7)–(V.10), (V.52)–(V.54), and (VI.40)"},{"comment":"The variational principle is introduced through the ansatz P(X;E,J)=exp([V(X;E,J)+O(√η)]/η^3) and the identification V(X;E,J)=max_{α_X∈C_X} \\tilde S(α_X;E,J), which the authors describe only as a \"reasonable conjecture.\" Since all subsequent equations in Sec. VI are manipulations of this V, the claim in the abstract that a variational principle is \"derived\" from the stochastic model goes beyond what is actually shown. This step should either be justified (at least to the same standard as the bulk contribution) or be stated explicitly as an unproved assumption of the framework.","section":"Sec. VI A, Eqs. (VI.1)–(VI.3)"}],"minor_comments":[{"comment":"The definition f_d(θ_X)=u_o(θ_X)−θ_X s_d(θ_X) appears to contain a typo: the disordered free-energy density should use u_d(θ_X), not u_o(θ_X). As written, the combination f_o−f_d in (VI.37) and the subsequent derivation of (VI.38)–(VI.40) are confusing.","section":"Eq. (VI.32)"},{"comment":"The figure caption states J<0 while the text states J≤0. This is harmless but should be made consistent.","section":"Caption of Fig. 1 and Sec. II E"},{"comment":"The sentence comparing the factor 1/3 with the factor 1/2 from global thermodynamics would be more useful with a specific equation reference to Ref. [17], so that readers can verify the comparison.","section":"Sec. VI E"},{"comment":"There are minor typographical errors in the affiliation and email lines, e.g., \"J apan\" and \"r nakayama@tohoku.ac.jp\"; these should be corrected in the final version.","section":"Author affiliation"}],"recommendation":"major_revision","confidential_remarks":"The paper is honest and the qualitative result is interesting. My main reservation is that the headline coefficient 1/3 is not derived from the stochastic model; the authors themselves identify the uncontrolled step. I would not reject outright because the framework and bulk derivation have value and the limitation is explicitly stated. However, if the journal requires a fully derived quantitative central claim, this manuscript is close to the rejection boundary."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Punchline: this is a real, serious paper, and the reader's conditional verdict is right. The variational principle and the explicit formula for the interface temperature are new; the model is carefully built; and the authors are refreshingly honest about the one step that carries the quantitative result.\n\nWhat is actually good: the bulk contribution to the excess entropy production is derived cleanly from the Zubarev-McLennan representation, and the decomposition (IV.23) into ordered, disordered, and interface regions is transparent. The modified entropy (V.57) and the final variation problem are genuine working pieces. The paper also articulates why deterministic hydrodynamics would give θ = Tc and why fluctuations can change that — that argument alone is worth something.\n\nThe soft spot, as flagged, is Sec. V A. The interface contribution I_int is computed from a phenomenological Onsager ansatz, Eqs. (V.7)-(V.10), with Onsager coefficients L_o = λ_o/(gX_-), L_d = λ_d/(g(1-X_+)). The value g = 1/3 is then forced by requiring the inverse-temperature gap to vanish at X→0,1. The authors state plainly that this 'involves uncontrolled approximations' and is 'not yet derived from the stochastic model.' That is accurate. Since I_int is the only source of the shift — no interface contribution means no shift at all — the coefficient 1/3 in Eq. (VI.40) is not established. The qualitative sign and structure are supported by the global-thermodynamics comparison; the exact factor is not.\n\nI'd nudge the reader's soundness score up slightly: the bulk derivation is tight, the limitation is in the open, and the paper doesn't hide the gap. But for the specific central claim, the concern is exactly right.\n\nThe citation pattern is fine; the prior Nakagawa-Sasa prediction is credited and used as comparison, not as an input. There is no code or numerical data, but the analytical work is substantial.\n\nWho should read it: people working on nonequilibrium steady states, phase coexistence under heat flow, and variational approaches to NESS. They will get a clear framework and a concrete conjecture to test.\n\nRecommendation: send it to peer review. It is not a desk reject. A good referee report should press hard on V.7-V.10, ask whether the Onsager coefficients can be derived or simulated, and request that the 1/3 be presented as a conjecture until then. With that framing, the paper is publishable.","headline":"A serious paper with a new variational principle; the interface-temperature formula is the valuable conjecture, but the coefficient 1/3 rests on an admittedly phenomenological step, so treat (VI.40) as conditional until the interface contribution is derived.","tokens_in":41124,"tokens_out":3152,"would_cite":true,"duration_ms":35199,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82B26","80A19","82C31","82C35"],"pacs":["05.70.-a","05.70.Ln","05.40.-a"],"model":"deepseek-v4-flash","headline":"A stochastic order-parameter model for phase coexistence in steady heat conduction predicts that the interface temperature deviates from the equilibrium transition temperature by a term proportional to the heat flux and the difference of…","keywords":["stochastic thermodynamics","phase coexistence","heat conduction","order parameter dynamics","variational principle","interface temperature","Zubarev-McLennan distribution","superheated ordered state"],"falsifier":"A direct test is a numerical simulation of the stochastic model (II.65)\\u2013(II.67) under the non-equilibrium adiabatic condition for a first-order order-disorder transition with unequal bulk conductivities: measure the steady interface temperature $\\theta^*$ and check whether $\\theta^*-T_c$ equals $-(J/3)(1/\\kappa_o-1/\\kappa_d)X_{eq}(1-X_{eq})$, in particular whether the deviation is linear in $J$, vanishes when $\\kappa_o=\\kappa_d$, and changes sign when the conductivity ordering is reversed. An independent check is a molecular-dynamics simulation of phase coexistence under a heat flux measuring the local temperature profile near the interface to see a superheated ordered or supercooled disordered layer of the predicted magnitude.","tokens_in":40037,"feed_emoji":"🔥","tokens_out":8932,"duration_ms":90205,"temperature":0.7,"pith_summary":"This paper tries to establish that when two phases of a material coexist while heat steadily flows through them, the temperature of the interface between the phases is not the equilibrium transition temperature $T_c$. From a stochastic order parameter model of an order\\u2013disorder transition with a flux-controlled boundary condition that conserves total energy, the authors derive a variational principle for the non-equilibrium steady state. Solving it gives $\\theta^* - T_c = -(J/3)(1/\\kappa_o - 1/\\kappa_d)X_{eq}(1-X_{eq})$ to linear order in the heat flux $J$: the interface is hotter than $T_c$, with a superheated ordered state nearby, when the ordered phase conducts heat worse than the disordered phase, and colder, with a supercooled disordered state, in the opposite case. The result matters because it turns an earlier thermodynamic prediction into a concrete, testable statement about a stochastic microscopic model, and it shows that deterministic interface thermodynamics misses the effect entirely.","feed_headline":"Heat flow shifts the temperature of a phase interface","feed_subtitle":"Interface runs above or below T_c; the ordered side is superheated or the disordered side supercooled depending on conductivities.","key_machinery":"The central object is a stochastic extension of phase-field dynamics: fields $m$ (order parameter), $v$ (its momentum), and $\\varphi$ (energy density), with an entropy functional $S=\\int (s(u,m) - (d_s/2)|\\nabla m|^2)$ and noise obeying local detailed balance; the small parameter $\\eta$, the ratio of microscopic to system length, controls the separation of scales and makes interface motion singularly slow. The load-bearing identity is the stationary-distribution representation (the Zubarev\\u2013McLennan form) $P_{ss}\\propto e^{\\tilde S/\\eta^3}$, where the modified entropy $\\tilde S = S + J I$ contains the time-integrated excess entropy production $I$ during relaxation of an interface configuration. The paper evaluates $I$ by splitting it into bulk and interface contributions; the interface contribution uses a phenomenological Onsager law for energy exchange between the interface and the two bulk regions with coefficients $L_o = \\lambda_o/(g X_-)$ and $L_d = \\lambda_d/(g(1-X_+))$, and the condition that the inverse-temperature gap vanish as the interface approaches either boundary fixes $g=1/3$. Maximizing $\\tilde S$ over interface position $X$ produces the variational equation whose solution is the interface temperature formula.","core_discovery":"The central claim is that interface fluctuations, not deterministic bulk transport, set the coexistence temperature in steady heat conduction. The paper derives, within the linear response regime, the formula $\\theta^* - T_c = -(J/3)(1/\\kappa_o - 1/\\kappa_d)X_{eq}(1-X_{eq})$, with $X_{eq}$ the equilibrium interface position, $\\kappa_o$ and $\\kappa_d$ the thermal conductivities of the ordered and disordered phases, and $J$ the heat flux ($J<0$ in the setup). Equivalently, for either sign of $J$, $\\theta^* - T_c = |J|/3(1/\\kappa_o - 1/\\kappa_d)X_{eq}(1-X_{eq})$. The derivation proceeds by writing the stationary distribution of interface configurations through a modified entropy that includes the excess entropy produced during relaxation, decomposing that excess entropy into ordered-region, disordered-region, and interface parts, and maximizing the resulting potential over the interface position. The qualitative phenomenon\\u2014superheated ordered or supercooled disordered states near the interface\\u2014was predicted earlier by an extended thermodynamics framework; here it is obtained as the variational solution of a stochastic order parameter model, with a numerical prefactor $1/3$ instead of the earlier $1/2$.","pith_inferences":["Because the authors conjecture the interface temperature is independent of the boundary condition for fixed energy and flux, the formula should also apply to the more standard setup of heat baths at the two ends; testing this in simulations would strengthen or falsify the conjecture.","The numerical factor $1/3$, fixed by requiring the temperature gap to vanish at the boundaries, is likely the least secure part of the argument; a microscopic derivation of the Onsager coefficients $L_o,L_d$ from the stochastic model would either confirm $1/3$ or replace it, while the qualitative sign of the interface-temperature shift would probably survive.","The same mechanism\\u2014latent heat released by a fluctuating interface acting as a local heat source\\u2014should appear in other first-order transitions such as liquid\\u2013gas or nematic\\u2013isotropic, with a conserved density adding an extra contribution; the paper mentions liquid\\u2013gas as a natural next case.","A direct numerical check of the stochastic model could measure the inverse-temperature gap during interface relaxation and compare it with the prediction $\\beta_+^{int}-\\beta_-^{int} = (1/3)(1/\\lambda_o-1/\\lambda_d)(dX/dt)X(1-X)q_X$, isolating the coefficient $1/3$."],"forward_implications":["For a first-order order\\u2013disorder transition under steady heat flux, the coexistence line is not at $T_c$: the interface temperature shifts linearly with $J$, so measured phase diagrams in a thermal gradient should show a shifted apparent transition temperature at the interface.","The sign of the shift is controlled by which phase conducts better: with the ordered phase less conductive, the ordered side is superheated; with the ordered phase more conductive, the disordered side is supercooled.","The deviation vanishes when the two bulk conductivities are equal, so unequal conductivities are necessary for the effect; it also vanishes as the interface approaches either boundary, consistent with the boundary fixing of $g=1/3$.","The deterministic, noiseless limit of the model predicts $\\theta^*=T_c$, so the effect is carried by interface fluctuations: any coarse-grained description that omits the stochastic coupling of the interface to the bulk energy reservoirs would miss the shift.","The variational potential $V(X)$ provides a way to compute other non-equilibrium thermodynamic properties of coexistence, such as the most probable interface position, once the stationary distribution of interface configurations is known."],"supporting_citations":[{"why":"Provides the earlier thermodynamic prediction, from an extended framework, that the interface temperature deviates from $T_c$ in steady heat conduction; the present stochastic calculation reproduces it qualitatively.","marker":"[16]"},{"why":"Supplies the global-thermodynamics framework whose quantitative interface-temperature formula (factor $1/2$) is compared against the factor $1/3$ derived here.","marker":"[17]"},{"why":"The Zubarev\\u2013McLennan representation expresses the stationary distribution through time-integrated excess entropy production, the starting point for the modified-entropy potential.","marker":"[30\\u201334]"},{"why":"Reports the experimentally observed temperature gap close to a liquid-vapour interface, the physical effect the interface-region entropy production is meant to capture.","marker":"[35]"},{"why":"Thermodynamically consistent phase-field model that the authors extend with noise and energy conservation to define the stochastic order-parameter dynamics.","marker":"[44]"}],"fun_headline_variants":["Heat flow superheats one phase or supercools the other","Phase interface temperature shifts with heat flux","Heat flux tilts coexistence temperature at interface","Superheated order or supercooled disorder at phase interface","Interface runs above or below transition temperature under heat flux"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the energy exchange between the thin interface region and the two bulk regions is described by the phenomenological Onsager law (V.7)\\u2013(V.10) with coefficients $L_o = \\lambda_o/(g X_-)$ and $L_d = \\lambda_d/(g(1-X_+))$, an uncontrolled approximation not yet derived from the stochastic model, and the factor $g=1/3$ is fixed by requiring the inverse-temperature gap to vanish at the boundaries.","fun_headline_variants_meta":{"raw":{"variants":["Heat flow superheats one phase or supercools the other","Phase interface temperature shifts with heat flux","Heat flux tilts coexistence temperature at interface","Superheated order or supercooled disorder at phase interface","Interface runs above or below transition temperature under heat flux"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000758,"raw_usage":{"total_tokens":3380,"prompt_tokens":968,"completion_tokens":2412,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":584,"completion_tokens_details":{"reasoning_tokens":2337}},"tokens_in":584,"tokens_out":2412,"duration_ms":19760,"temperature":1.0,"reasoning_tokens":2337,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:26:00.630638+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct test is a numerical simulation of the stochastic model (II.65)\\u2013(II.67) under the non-equilibrium adiabatic condition for a first-order order-disorder transition with unequal bulk conductivities: measure the steady interface temperature $\\theta^*$ and check whether $\\theta^*-T_c$ equals $-(J/3)(1/\\kappa_o-1/\\kappa_d)X_{eq}(1-X_{eq})$, in particular whether the deviation is linear in $J$, vanishes when $\\kappa_o=\\kappa_d$, and changes sign when the conductivity ordering is reversed. An independent check is a molecular-dynamics simulation of phase coexistence under a heat flux measuring the local temperature profile near the interface to see a superheated ordered or supercooled disordered layer of the predicted magnitude.","supporting_citations":[{"cited_title":"Bedeaux, E","cited_arxiv_id":null,"evidence_quote":"Provides the earlier thermodynamic prediction, from an extended framework, that the interface temperature deviates from $T_c$ in steady heat conduction; the present stochastic calculation reproduces it qualitatively."},{"cited_title":"Onuki, Dynamic van der Waals theory, Phys","cited_arxiv_id":null,"evidence_quote":"Supplies the global-thermodynamics framework whose quantitative interface-temperature formula (factor $1/2$) is compared against the factor $1/3$ derived here."},{"cited_title":"Sasa, Collective dynamics from stochastic therm o- dynamics, New Journal of Physics 17, 045024 (2015)","cited_arxiv_id":null,"evidence_quote":"Reports the experimentally observed temperature gap close to a liquid-vapour interface, the physical effect the interface-region entropy production is meant to capture."},{"cited_title":"Maes and K","cited_arxiv_id":null,"evidence_quote":"Thermodynamically consistent phase-field model that the authors extend with noise and energy conservation to define the stochastic order-parameter dynamics."}],"review_version":1}