{"id":"3574dd7a-be4f-4b38-85cc-54339b92eff5","arxiv_id":"2607.27803","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"An EOS-anchored, elasticity-constrained Grüneisen function predicts finite-temperature bulk-modulus softening in four solids without fitting to thermal data.","lead":"An equation-of-state-based Grüneisen function, anchored by elastic Debye temperatures and an infinite-compression limit, predicts how bulk moduli soften with temperature using only static energy-volume data—no parameters are fit to thermal measurements. Tested on diamond, MgO, silicon, and sodium chloride with three machine-learning potentials, it reproduces the leading softening trends and exposes which static descriptions are thermodynamically transferable.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Load-bearing assumption is the unvalidated linear t(V) truncation: imposed rather than derived, and q(V) is EOS-sensitive (Fig. 7); validation against phonon-derived Grüneisen functions is needed.","rationale":"The reader's weakest assumption—the linear t(V) truncation with t0=5/2—is exactly the load-bearing point. I checked the internal algebra of the construction, including Eq. (19) as the integral of Eq. (3), and found no internal inconsistency. The issue is external validation: the paper's own Fig. 7 shows that q(V), the most sensitive input to Eq. (21), depends strongly on the analytic EOS even with fixed equilibrium anchors. Since no phonon or MD Grüneisen function is computed to certify the volume dependence, the central claim is plausible but not yet demonstrated. The lack of public data/code is a secondary auditability concern, not the scientific crux. Therefore the reader's CONDITIONAL verdict is appropriate; no change is needed.","tokens_in":22160,"tokens_out":7902,"duration_ms":78060,"concrete_test":"Using the same UMA/UPET calculators, compute mode-resolved Grüneisen parameters from phonopy at V/V0 ∈ {0.98, 1.00, 1.02, 1.06}; average them to γ_ph(V). Re-evaluate K_S(T) and dK_S/dT from Eqs. (21)–(22) with γ_ph(V) replacing γ(V) and with ΘD(V) obtained by integrating −γ_ph/V from V0. If the 300 K softening rates differ by more than ~20% from Fig. 4 for any of the four solids, the imposed t(V) form is load-bearing and the no-thermal-fitting claim is not yet established; if they agree, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that an EOS-based construction captures leading thermal softening without fitting thermal data—depends on Eq. (2), the lowest-order truncation t(V)=t0−t1(V/V0)^{1/3} with t0=5/2 fixed by an asymptotic argument from Ref. 4. The paper does not derive this form for diamond, MgO, Si, or NaCl; it states that higher-order coefficients would be non-unique and would require additional constraints. The only material-specific anchor is the equilibrium condition Eq. (4), which fixes t1 but not the functional shape away from V0. The volume dependence—and hence q(V)=d lnγ/d ln V, the quantity entering the Mie–Grüneisen bulk modulus Eq. (21)—is then inherited from the chosen EOS and the assumed linear form. Figure 7 shows that q(V) changes substantially between Vinet and AP2 even when the same γ0 and Θ0 are used, and the paper acknowledges q is more EOS-sensitive than γ or ΘD. Without an independent check of t(V) against phonon- or MD-derived Grüneisen data, the agreement in Figs. 4–6 could be a consequence of the chosen interpolation rather than a robust EOS-based prediction. This is a validation gap, not an internal inconsistency; the derivation from Eq. (1) onward is internally coherent.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops an EOS-based construction of the volume-dependent Grüneisen function γ(V) within the Burakovsky–Preston framework, using a lowest-order truncation t(V)=t0−t1(V/V0)^{1/3} with t0=5/2 fixed by an infinite-compression argument, t1 determined by an elasticity-derived equilibrium Grüneisen parameter γ0, and an EOS-dependent asymptotic shift s. Static energy–volume data from MLIPs (UMA, d-UMA, UPET) are fit to Vinet or AP2 EOSs; Debye temperatures from the Christoffel equation anchor γ0. The resulting γ(V) is used in a Mie–Grüneisen–Debye model to predict K_T(T), K_S(T), and dK/dT for diamond, MgO, Si, and NaCl, compared with experiment without fitting thermal data. The paper reports reasonable qualitative agreement and material/potential/EOS-dependent quantitative agreement, and concludes that the construction captures the leading thermal softening without thermal fitting.","tokens_in":22482,"tokens_out":3643,"duration_ms":28210,"significance":"If the claim holds, the value is methodological: a simple, inexpensive, non-thermal-fitting route from static EOS and near-equilibrium elasticity to finite-temperature bulk moduli. This is potentially useful for screening machine-learning interatomic potentials and for materials lacking thermal data. Strengths: the framework is internally coherent; the workflow is specified in sufficient detail to be reproducible (including fitting weights, convergence checks, and code-level descriptions); the no-thermal-fitting property is structurally true, as thermal data enter only as comparison targets; the paper is transparent about limitations (quasiharmonic level, EOS sensitivity, t(V) truncation) and tests four materials with multiple potentials and two EOSs. The main risk is whether the imposed t(V) form and the EOS-dependent asymptotic shift are sufficiently justified to sustain the strong conclusion that the agreement demonstrates a robust EOS-based prediction rather than a favorable interpolation.","major_comments":[{"comment":"The central load-bearing assumption is the linear-in-V^{1/3} truncation t(V)=t0−t1(V/V0)^{1/3} with t0=5/2. This form is adopted from Ref. 4's suggested expansion and its validity for diamond, MgO, Si, and NaCl is not established. The paper notes that higher-order terms would be non-unique and would require additional constraints, but no independent validation of the truncation is provided. Since q(V) is the most sensitive quantity in Eq. (21) and Fig. 7 shows substantial EOS dependence in q(V), the agreement in Figs. 4–6 could depend on the chosen interpolation rather than being a robust consequence of the EOS construction. An independent check against phonon- or MD-derived γ(V) or q(V) for at least one material, or a sensitivity analysis of the truncation, is needed to validate the central claim. This is a validation gap rather than an internal inconsistency.","section":"Sec. II.A, Eq. (2)"},{"comment":"The asymptotic shift s is EOS-dependent (s=1/3 for Vinet, s=0 for AP2) and is imposed rather than derived for these materials. The paper acknowledges that EOS forms with γ^EOS_inf substantially exceeding γ_inf require scrutiny, but the impact of this shift on finite-temperature predictions is not quantified separately from other EOS effects. Since the shift is constant in γ, it does not enter q(V), but it does change γ(V) and hence the thermal pressure and K_S in Eq. (21). The paper shows UPET(Vinet) and UPET(AP2) differ (e.g., MgO in Fig. 4), but does not isolate how much of that difference comes from the shift versus the EOS shape. A decomposition would help assess whether the asymptotic constraint is doing accepted physical work or merely adjusting the baseline.","section":"Sec. II.A, Eq. (3)"},{"comment":"The equilibrium anchoring γ0 is obtained from elastic Debye temperatures via Eq. (9). Figure 7 shows that the constructed γ values are systematically above representative experimental estimates for all four materials, with the offset largely inherited from γ0 at V0. This systematic offset in γ, combined with the unvalidated t(V) form, weakens the claim that the construction 'captures the leading thermal-softening behavior' in a predictive sense: the thermal softening in K(T) is mediated by γ(V) (Eqs. 21–22), and if γ is systematically high, the agreement in dK/dT may benefit from compensating errors (e.g., the quasiharmonic Debye model underestimating anharmonic softening). The paper notes this qualitatively, but the central claim would be strengthened by a quantitative discussion of how sensitive the K(T) predictions are to the γ0 offset and the t(V) truncation.","section":"Sec. III, Fig. 7 and Eq. (8)"},{"comment":"The Wachtman form is used to fit the calculated K(T) curves and to evaluate dK/dT analytically. This is clearly presented as an auxiliary interpolation, which is acceptable. However, in Fig. 4 the experimental comparisons in the bottom row are analytic derivatives of experimental fits to the same form, while the calculated curves are derivatives of fits to Eq. (24) of the Mie–Grüneisen points. Systematic fitting errors (e.g., the form being imperfect at low or high T) could bias the comparison. A direct numerical derivative of the raw calculated points, or a statement of fit residuals, would make the claimed agreement in dK/dT more robust.","section":"Sec. II.B, Eq. (24)"}],"minor_comments":[{"comment":"The statement 't_1^AP2 ≈ t_1^Vinet + 1 whenever the Vinet and AP2 fits yield close values of K'_0' could be derived explicitly from Eq. (4) with s_Vinet=1/3 and s_AP2=0; as written it is somewhat abrupt.","section":"Sec. II.A, after Eq. (4)"},{"comment":"The constrained quadratic fit is stated to give virtually the same γ0 as the linear fit, but no numbers or a figure showing the difference is provided. A sentence with typical γ0 differences would be helpful.","section":"Sec. II.A, Eq. (9)"},{"comment":"The caption states that zero-point-corrected experimental reference values are taken from Ref. 42, but the reader has to infer what zero-point correction is applied. A brief note in the text or caption would clarify the reference baseline.","section":"Sec. II.C, Fig. 2"},{"comment":"The caption is dense and combines many references with descriptions of curves. Consider separating the experimental parametrization description into a table or a separate paragraph, since the current caption is hard to parse.","section":"Sec. III, Fig. 4 caption"},{"comment":"The conclusion that all four present values for NaCl lie below the experimental estimate indicates an overestimate of softening, but the figure symbol legend is not explicit. Adding distinct markers or labels for 'present' vs 'literature' values would improve readability.","section":"Sec. III, Fig. 6"},{"comment":"The weighting wi = exp(−βΔEi)/Σ_j exp(−βΔEj) with β=40 eV^{−1} is described, but the choice of β is not justified beyond 'in all fits'. A brief statement that results are insensitive to β over a reasonable range would be useful.","section":"Sec. II.C, Eq. (15)"}],"recommendation":"major_revision","confidential_remarks":"The paper is generally sound and the central construction is internally coherent. The main issue is validation of the assumed t(V) truncation. The author acknowledges limitations, but the conclusion is stronger than the validation supports. A comparison with phonon-based γ(V), or at least a sensitivity analysis of the truncation, would substantially strengthen the paper. Also, the EOS-dependent shift and the γ0 offset deserve more quantitative decomposition. These are addressable within the scope of the manuscript, so major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper does what it says: it builds a Grüneisen function from static EOS data plus elasticity-derived Debye temperatures, with no parameters adjusted to thermal data, and uses it to predict finite-temperature bulk moduli. The construction is genuinely new in a modest way—prior variable-t Grüneisen forms leaned on experimental compression or melting data. Here the volume dependence is fixed from static-lattice information and an infinite-compression constraint. The derivation is clean, the workflow is careful (Monte Carlo angular sampling for Debye temperatures, weighted strain fits, separate Vinet and AP2 baselines), and the comparisons against experiment and other finite-temperature methods are fair and appropriately hedged. The authors are also honest about the material-, potential-, and EOS-dependent accuracy. The claim that the framework can serve as a screening tool for MLIP thermoelastic transferability is plausible and useful.\n\nThe soft spots are where the reader and stress-test put them. The linear t(V) truncation with t0=5/2 is imposed from Ref. 4, not derived for diamond, MgO, Si, or NaCl. The paper's own Fig. 7 shows q(V), the quantity that enters the thermal bulk-modulus formula, changes substantially between Vinet and AP2 even when γ0 and Θ0 are anchored identically. That is a real validation gap: without a cross-check against phonon- or MD-derived Grüneisen functions, the agreement in Figs. 4–6 could be partly an artifact of the chosen interpolation. However, the central claim—capturing leading thermal softening without fitting to thermal data—is worded carefully and is supported by the trends, so this is a limitation rather than a fatal flaw. The bigger practical issue is that data and code are not public, only available on request, which makes the quantitative claims hard to audit. That is a legitimate concern for a methods paper. Minor issues: only four materials, Debye-model anharmonicity is neglected (acknowledged), and the MLIP static baselines carry errors that propagate into the predictions.\n\nWho gets value from this: computational materials scientists working with universal MLIPs, especially anyone who wants a cheap thermoelastic diagnostic, and people doing Grüneisen-parameter modeling. It deserves a serious referee. I would send it to peer review and ask for two things in revision: a direct validation of γ(V) or q(V) against phonon or MD data for at least one of the four materials, and deposition of the code and data. If those are addressed, the paper is a solid contribution.","headline":"Solid, honest paper: the no-thermal-fit EOS-based Grüneisen construction delivers on its modest claim, but the imposed t(V) form and closed data are the real soft spots.","tokens_in":23003,"tokens_out":2627,"would_cite":true,"duration_ms":25805,"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 establishes that an EOS-based Grüneisen function, built only from static energy-volume data, near-equilibrium elastic constants, and the infinite-compression limit, predicts finite-temperature bulk-modulus softening without fittin","keywords":["Grüneisen parameter","bulk modulus thermal softening","equation of state","Mie–Grüneisen–Debye","Debye temperature","machine-learning interatomic potentials","thermoelastic transferability","thermal pressure"],"falsifier":"Compute the exact Grüneisen function γ(V) for MgO by quasiharmonic lattice dynamics over a moderate volume range, convert it to the implied t(V) via Eq. (1), and compare with the linear truncation t(V)=t0−t1(V/V0)^{1/3}. If the implied q(V)=d ln γ/d ln V differs substantially from the EOS-based q(V) under the same Vinet and AP2 fits, the truncation rather than the physics is carrying the predicted softening.","tokens_in":1550,"feed_emoji":"🌡️","tokens_out":2214,"duration_ms":81115,"temperature":0.7,"pith_summary":"This paper tries to establish that the Grüneisen function—the key link between atomic vibrations and thermal expansion or softening—can be constructed entirely from static, zero-temperature equation-of-state data plus near-equilibrium elastic constants, with no thermal measurements used as input. The resulting Grüneisen function is fed into a Mie–Grüneisen–Debye model, and for diamond, MgO, silicon, and sodium chloride it reproduces the observed thermal-softening trends of the bulk modulus and the overall scale of its temperature derivative. Absolute bulk moduli inherit errors from the underlying interatomic potential and from the choice of analytic EOS, but the leading thermal behavior is captured without a single thermal fit parameter. If correct, this turns a cheap static calculation into a thermal prediction and gives a diagnostic for spotting deficiencies in machine-learning interatomic potentials and density-functional descriptions.","feed_headline":"Grüneisen from static EOS predicts thermal softening of bulk moduli","feed_subtitle":"No thermal measurements enter the fit; diamond, MgO, Si, and NaCl all reproduce observed softening trends.","key_machinery":"The central object is the generalized EOS-based Grüneisen relation (Eq. 1), which expresses γ(V) in terms of static pressure, bulk modulus, and its pressure derivative, together with an auxiliary index t(V) that interpolates between known constant-t limits. The paper fixes t(V) through a lowest-order truncation t(V)=t0−t1(V/V0)^{1/3} with t0=5/2, anchors the equilibrium value to an elasticity-derived Debye Grüneisen parameter, and applies a constant asymptotic shift where needed. This construction makes γ(V) fully determined by static EOS data and near-equilibrium elastic constants. The quantity that carries the thermal prediction is q(V)=d ln γ/d ln V, which enters the bulk-modulus expressi","core_discovery":"The central claim is that the volume-dependent Grüneisen parameter can be constructed analytically from static-lattice information alone: the static energy–volume curve supplies pressure and bulk-modulus derivatives, near-equilibrium elastic tensors supply Debye temperatures that anchor the equilibrium Grüneisen value, and an infinite-compression constraint fixes the asymptotic limit. With those inputs, the resulting γ(V) has no free thermal parameters, and the Mie–Grüneisen–Debye thermal pressure then yields adiabatic and isothermal bulk moduli as predictions. For the four test solids, the computed bulk moduli soften with temperature in the same way as experiment, and the room-temperature s","pith_inferences":["Editorial extension: because the construction needs only static EOS data and near-equilibrium elastic tensors, it should transfer directly to DFT-based workflows, making it a practical low-cost screening tool for newly synthesized or hypothetical materials where thermal data do not exist.","Editorial extension: experimental determinations of q (from ultrasonic or shock-compression data) would provide a sharper test of the Vinet-versus-AP2 choice than bulk-modulus curves alone, since the paper shows q is the most EOS-sensitive quantity.","Editorial extension: a natural stress-test is to apply the same construction to a strongly anharmonic solid, such as a BCC transition metal, where the quasiharmonic Mie–Grüneisen–Debye cap is expected to break down; the failure mode would define the practical validity boundary of the framework.","Editorial extension: the mismatch between predicted and measured softening could be reinterpreted as a fingerprint of static-description errors and used to guide iterative retraining of universal machine-learning potentials, going beyond the paper's stated diagnostic goal."],"forward_implications":["Finite-temperature bulk moduli can be predicted from static energy–volume data and elastic tensors at a few near-equilibrium volumes, without thermal data, AIMD sampling, or phonon calculations.","The room-temperature softening ratio KT(RT)/K0 for diamond, MgO, Si, and NaCl falls on the same scale as values from AIMD, quasiharmonic, and experimental references, confirming that the leading thermal softening is captured.","The effective equilibrium Grüneisen index varies strongly across materials and can fall outside the conventional constant-t range, so no single constant-t prescription can be universal.","The Vinet-versus-AP2 choice affects q(V) noticeably even when γ(V) curves look similar, so the analytic EOS form is a genuine source of uncertainty in thermoelastic predictions.","The framework can serve as a diagnostic: when a machine-learning potential or DFT functional gives poor static EOS or elastic data, the predicted thermal softening exposes it before expensive finite-temperature simulations are run."],"fun_headline_variants":["EOS-only Grüneisen captures thermal softening","No thermal data: EOS-Grüneisen softens moduli","Static EOS predicts Grüneisen, thermal softening","Grüneisen from static EOS alone softens bulk moduli"],"cache_read_input_tokens":24192,"weakest_assumption_plain":"The load-bearing assumption is that the auxiliary Grüneisen variable t(V) follows a simple linear-in-(V/V0)^{1/3} form pinned to t0=5/2 at infinite compression; the paper does not derive this shape for diamond, MgO, Si, or NaCl, and its own results show that q(V) is highly sensitive to the analytic EOS, so a real material whose t(V) curves differently would break the thermal-softening prediction.","fun_headline_variants_meta":{"raw":{"variants":["EOS-only Grüneisen captures thermal softening","No thermal data: EOS-Grüneisen softens moduli","Static EOS predicts Grüneisen, thermal softening","Grüneisen from static EOS alone softens bulk moduli"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00177,"raw_usage":{"total_tokens":6851,"prompt_tokens":805,"completion_tokens":6046,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":549,"completion_tokens_details":{"reasoning_tokens":5984}},"tokens_in":549,"tokens_out":6046,"duration_ms":37610,"temperature":1.0,"reasoning_tokens":5984,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T01:04:10.742695+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the exact Grüneisen function γ(V) for MgO by quasiharmonic lattice dynamics over a moderate volume range, convert it to the implied t(V) via Eq. (1), and compare with the linear truncation t(V)=t0−t1(V/V0)^{1/3}. If the implied q(V)=d ln γ/d ln V differs substantially from the EOS-based q(V) under the same Vinet and AP2 fits, the truncation rather than the physics is carrying the predicted softening.","supporting_citations":[],"review_version":1}