{"id":"21d40e6e-6d77-40cf-84a2-bf4bf3c6d54b","arxiv_id":"2608.06118","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Simulations with the Tadah!Kr3b potential predict an entropically stabilized bcc phase in krypton above about 36 GPa and reinterpret the experimental speckle melt-curve anomaly as an fcc-bcc transition.","lead":"A computer model of krypton based on a two-body potential predicts a new solid phase (bcc) near the melting line above 36 GPa, and suggests that an experimental anomaly in the melt curve is actually this phase transition, not melting. The study maps the full pressure-temperature phase diagram of this model and shows that a more complex machine-learned model performs worse.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The bcc-stability claim rests on a potential-transferability assumption that the authors' own numbers flag as doubtful; the enthalpy-vs-free-energy confusion in §3.5 needs to be resolved before the argument can be trusted.","rationale":"The paper builds a plausible phase diagram with multiple independent methods (direct coexistence, Gibbs-Helmholtz, Clapeyron slope field, Frenkel-Ladd free energies) that all point to a bcc region. However, the central claim is not just the existence of bcc for this model potential, but the suggestion that the experimental speckle anomaly at ~50 GPa is the fcc-bcc boundary. That interpretive leap requires the potential to be accurate at extreme conditions, where the paper itself admits the melt curve deviates from experiment (too steep at high pressure). The most load-bearing technical weakness is therefore the potential's transferability at high P-T, and within the paper the clearest symptom is the internally inconsistent enthalpy comparison in §3.5: an fcc-bcc enthalpy difference of ~10 eV/atom would make bcc absolutely unstable, contradicting the claimed free-energy result. If that number is a typo or uses inconsistent reference states, the paper must correct it; if it is real, the bcc-stability claim is refuted. This does not require rejecting the entire paper — the methodological framework and the coexistence simulations may still be sound — but the central predictive claim cannot be accepted as-is. The reader's verdict of CONDITIONAL is appropriate, and the needed revisions are concrete: fix or explain the §3.5 numbers, and temper the speckle-reinterpretation statement from a conclusion to a hypothesis. I do not see grounds for outright rejection; the independent Frenkel-Ladd verification and the careful mapping of the phase diagram are significant supporting evidence. My recommendation is therefore CONDITIONAL, unchanged from the reader's verdict, because the concern is addressable and the central construction is not demonstrably wrong once the internal inconsistency is resolved.","tokens_in":13718,"tokens_out":1885,"duration_ms":14138,"concrete_test":"Recompute the per-atom enthalpy of fcc and bcc Kr with the Tadah!Kr3b potential at 110 GPa and 3983 K in NPT, using identical simulation protocols and identical total-energy reference states. If the reported 14.898 eV/atom and 4.862 eV/atom values are reproduced, the paper's central bcc-stability claim is internally contradicted. If the values are found to be a typo or to use different reference states, recompute the fcc-bcc enthalpy difference and check whether the sign and magnitude are consistent with the Gibbs-Helmholtz and Frenkel-Ladd free-energy results presented in §3.2.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim is that Tadah!Kr3b predicts an entropically stabilized bcc phase above ~36 GPa/2630 K, and that the experimental speckle anomaly at ~50 GPa is actually the fcc-bcc boundary. For this to hold, the potential must reproduce the small free-energy differences between fcc and bcc at extreme conditions. Yet the paper itself gives evidence that the potential is not reliable in this regime: in §3.1, the calculated melt curve overestimates the low-pressure melting temperature by ~13 K and rises more steeply than the DAC measurements above ~30 GPa. More directly, §3.5 states that 'at 110GPa/3983K we find a bcc enthalpy of 14.898 eV/atom and fcc as 4.862 eV/atom' — a ΔH of ~10 eV/atom that would make bcc catastrophically unstable, contradicting the claimed bcc stability at 110 GPa and the paper's own assertion that bcc has moderately higher enthalpy than fcc. Unless these numbers are mislabeled (e.g., potential energies excluding different reference states, or a typo), the enthalpy comparison is internally inconsistent with the central phase-diagram claim. The reader's verdict correctly identifies the weakest assumption as potential transferability; this specific numerical inconsistency is the concrete, checkable symptom of that breakdown. Because the bcc region is narrow and entropically stabilized, even small errors in the short-range repulsion (which the authors themselves trace as the cause of bcc stability) could erase or move the bcc field substantially, undermining the reinterpretation of the speckle line.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper computes a pressure-temperature phase diagram for krypton using the Tadah!Kr3b two-body potential, which is fitted to CCSD(T) dimer and trimer energies. The authors combine direct two-phase coexistence simulations for the fcc and bcc melt lines, a newly introduced 'Clapeyron slope field' to trace the fcc-bcc boundary from a computed triple point, Gibbs-Helmholtz integration, static and quasiharmonic calculations for low-temperature fcc-hcp boundaries, slab coexistence for the liquid-gas line, and Frenkel-Ladd free-energy integrations as an independent check. The principal findings are: (i) an entropically stabilized bcc phase appears above a fcc/bcc/liquid triple point at approximately 36 GPa and 2630 K; (ii) two narrow hcp pockets are present at low temperature; (iii) the bcc phase exhibits 'greedy snake' collective diffusion events; and (iv) on the basis of these results, the experimental laser-speckle melting anomaly near 50 GPa is reinterpreted as the fcc-bcc transition rather than the melt curve. A comparison with the MACE foundation model concludes that more flexible machine-learned potentials are not automatically more reliable.","tokens_in":14018,"tokens_out":8867,"duration_ms":72592,"significance":"The paper is technically impressive. The use of multiple independent methods (coexistence, Gibbs-Helmholtz, Clapeyron field, Frenkel-Ladd, quasiharmonic) to cross-check each boundary sets a high standard. The Clapeyron slope field is a useful methodological idea, and the uncertainty estimates from Monte Carlo resampling are a strength. The prediction of a bcc field is concrete and falsifiable: it can be tested by x-ray diffraction melting experiments or by direct free-energy calculations with many-body potentials or DFT. The reinterpretation of the speckle anomaly, if confirmed, would resolve a long-standing discrepancy. However, the physical significance of the predictions rests on the transferability of a potential fitted to two- and three-atom clusters, which the paper itself acknowledges is not directly established at the extreme conditions where bcc is predicted. The manuscript also contains an obvious numerical inconsistency in §3.5 that must be fixed before the results can be accepted.","major_comments":[{"comment":"At 110 GPa and 3983 K, the paper reports a bcc enthalpy of 14.898 eV/atom and an fcc enthalpy of 4.862 eV/atom, and then states that the bcc phase is stable. These numbers are mutually incompatible: the 10.036 eV/atom enthalpy penalty would require an entropy difference of roughly 28 k_B per atom at 3983 K to stabilize bcc, which is physically implausible and inconsistent with the claimed free-energy results. This is almost certainly a typographical error (the bcc value was likely intended to be 4.898 eV/atom, making the enthalpy difference 0.036 eV/atom), but as written it directly contradicts the central phase-stability claim. The authors must correct the numbers and re-state the enthalpy difference, and they should verify that the corrected values are consistent with the Frenkel-Ladd and Clapeyron-field results.","section":"§3.5"},{"comment":"Section 3.1 states that 'At high pressures the computed melt curve rises much more steeply than the laser-heated diamond-anvil-cell measurements, which flatten above about 30 GPa,' while the Conclusion states that 'Our calculated melt curve tracks the experiment well up to 50 GPa.' These statements cannot both be true. The behavior of the model melt curve relative to experiment is central to the paper's reinterpretation of the speckle anomaly, so the authors must reconcile this contradiction and specify quantitatively where and by how much the model and experiment diverge.","section":"§3.1 and §4"},{"comment":"The claim that the experimental speckle-disappearance line 'should be interpreted as the fcc–bcc transition' goes beyond what the simulations demonstrate. The simulations show bulk 'greedy snake' events in bcc Kr at specific (P,T) conditions, but no simulation of surface roughness or of the speckle pattern itself is presented. To make this reinterpretation credible, the authors should either perform a direct simulation of surface morphology evolution under the relevant conditions, or soften the language to 'may be' and clearly label the reinterpretation as a hypothesis.","section":"§3.5 and §4"},{"comment":"The bcc field is presented as a prediction for real krypton, but the Tadah!Kr3b potential is fitted only to dimer and trimer CCSD(T) data. The Frenkel-Ladd and Clapeyron-field checks verify internal consistency of the potential, not its accuracy against real krypton. The melt-curve comparison in §3.1 shows that the potential deviates from experiment at high pressures, which is a direct warning that the potential may not be reliable in the regime where bcc is predicted. To support the physical prediction, the authors should provide additional validation of the potential under extreme conditions, for example by comparing its room-temperature equation of state with the experimental data of Rosa et al. (Ref. 4) or by computing the fcc-bcc enthalpy difference with an independent many-body method (e.g., a DFT-D or a different ML potential). Without such a test, the bcc prediction should be framed explicitly as a property of the model, not of krypton.","section":"§2.2 and §3.1"}],"minor_comments":[{"comment":"The word 'hexahonal' should be 'hexagonal'.","section":"Abstract"},{"comment":"The word 'Mechaniism' should be 'mechanism'.","section":"§3.5"},{"comment":"The angle brackets in '¡111¿' should be typeset as '<111>'.","section":"§3.5"},{"comment":"The phrase 'In extremis' is used incorrectly; consider 'in the extreme case' or similar.","section":"§2.4"},{"comment":"The sentence beginning 'MACE is trained on neither krypton nor explicit physical constraints' is clear, but the following sentence is grammatically awkward; consider rewording.","section":"§3.6.1"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is fundamentally sound in its methodology, and the Clapeyron slope field is a nice contribution. However, the numerical inconsistency in §3.5 is so obvious that it will undermine the paper's credibility unless fixed. The authors should also be asked to reconcile the contradictory statements about the melt curve. The transferability question is a substantive concern, but it is not a fatal flaw if the authors are willing to temper their claims. I would be comfortable with acceptance after these revisions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nYou should know two things about this paper. It computes the full phase diagram of a specific krypton pair potential (Tadah!Kr3b) and reports an entropically stabilized bcc field near melting above ~36 GPa, with a tentative reinterpretation of the experimental speckle anomaly as an fcc-bcc boundary. That is a genuinely new and interesting result. But Section 3.5 contains an enthalpy comparison that, as written, cannot be right: bcc at 14.898 eV/atom vs fcc at 4.862 eV/atom at 110 GPa and 3983 K. For bcc to be the stable phase, it would need a lower Gibbs free energy, which would require TΔS of order 10 eV/atom — several tens of k_B per atom. No vibrational entropy difference does that. Either the numbers are mislabeled or the central claim is undermined. This has to be fixed before anything else.\n\nWhat the paper does well: the bcc phase is not a fluke of one method. Direct solid-liquid coexistence gives the fcc/bcc/liquid triple point, and independent Frenkel-Ladd free-energy calculations reproduce the same bcc band. The Clapeyron slope field is a nice generalization of Gibbs-Duhem integration, and the paper is honest about metastability, error accumulation, and where the quasiharmonic approximation is unreliable. The comparison with the MACE foundation model is a useful cautionary tale: flexible ML potentials trained on unrelated data can produce unphysical cluster stability. The paper deserves credit for that.\n\nThe soft spots beyond the enthalpy number: the speckle reinterpretation is presented too strongly. The paper's own melt curve does not reproduce the experimental flattening above 30 GPa, so the claim that the speckle-disappearance line is really the fcc-bcc transition is circumstantial. It would be better as an explicit hypothesis. And the scripts and data are only 'available on reasonable request' — for a computational phase-diagram paper, that should be a deposited package with a commit hash. The potential-transferability concern is real but not disqualifying: the model overestimates low-pressure melting by ~13 K and rises more steeply than DAC data, so its fidelity at bcc conditions is unproven. Still, the bcc field is a property of the model, and that is the actual claim.\n\nThis paper is for the high-pressure and ML-potential community. It deserves a serious referee, and with the enthalpy error corrected, the language softened, and the code shipped, it would be a useful contribution. I'd take it to peer review, but I would not accept it as-is.","headline":"A multi-method phase diagram for a krypton pair potential with a novel bcc prediction, undermined as written by a 10 eV/atom enthalpy inconsistency and an overreaching speckle reinterpretation.","tokens_in":14611,"tokens_out":5135,"would_cite":true,"duration_ms":39046,"reading_group":"yes","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 predicts an entropically stabilized body-centred cubic phase in krypton above about 36 GPa and 2630 K, and argues that the experimental speckle-disappearance line near 50 GPa marks this fcc-to-bcc transition rather than melting.","keywords":["krypton","phase diagram","body-centred cubic","melting curve","speckle method","machine-learned interatomic potential","molecular dynamics","greedy snake defects"],"falsifier":"A laser-heated diamond-anvil-cell experiment on krypton between 35 and 60 GPa using x-ray diffraction as the diagnostic would settle it: if no bcc reflections appear and the speckle-disappearance line coincides with the onset of liquid diffuse scattering, the proposed bcc field and the reinterpretation of the melt anomaly are wrong.","tokens_in":13477,"feed_emoji":"🧊","tokens_out":10091,"duration_ms":73254,"temperature":0.7,"pith_summary":"Using a two-body potential fitted only to coupled-cluster dimer and trimer energies, the paper maps the full phase diagram of krypton and finds an entropically stabilised body-centred cubic (bcc) phase along the melt curve. The bcc phase appears above a fcc–bcc–liquid triple point at about 36 GPa and 2630 K, where it has higher enthalpy but higher entropy than face-centred cubic. The paper argues that the anomaly in laser-speckle melting experiments near 50 GPa, where krypton seems to melt at lower temperatures than extrapolation predicts, is actually the fcc-to-bcc transition, because highly mobile 'greedy snake' defects can alter the surface speckle pattern without melting. If correct, this replaces the accepted reading of those experiments with a concrete prediction that x-ray diffraction can test. It also shows that a simple two-body potential fitted to accurate dimer/trimer data can outperform a more flexible machine-learned model that was not trained on krypton physics.","feed_headline":"Krypton gains a bcc phase that explains the 50 GPa melt anomaly","feed_subtitle":"A two-body potential predicts bcc krypton above 36 GPa, making the speckle 'melting' signal a solid-solid transition.","key_machinery":"The technical engine is the Clapeyron slope field $C(T,P)$, a generalisation of the Clausius–Clapeyron relation that evaluates $dT/dP = T\\Delta v/\\Delta h$ at every pressure and temperature from paired single-phase molecular-dynamics runs, reducing to the usual coexistence slope wherever $\\Delta G=0$. Its field lines trace phase boundaries from a known anchor; here the anchor is the fcc–bcc–liquid triple point, and the bcc–fcc boundary is verified independently by absolute free-energy calculations. The other load-bearing element is the Tadah!Kr3b potential itself, a machine-optimised two-body form with one screened-Coulomb term and seven Gaussian functions, fitted to CCSD(T) dimer and trimer data, whose short-range repulsion is softer than Lennard–Jones and is identified as the cause of bcc stability.","core_discovery":"The central claim is that the Tadah!Kr3b two-body potential, fitted to CCSD(T) dimer and trimer data, produces a krypton phase diagram with fcc, hcp, bcc, liquid, and gas regions, and that the bcc region is real physics rather than a fitting artifact. The bcc field is entropically stabilised along the melt curve above the fcc/bcc/liquid triple point at approximately 36 GPa and 2630 K: at fixed pressure and temperature near the melt, bcc has a higher enthalpy than fcc but enough extra entropy from vibrational motion to lower its Gibbs free energy. Because bcc is more plastic than fcc, especially through correlated chains of atom jumps called greedy snakes, the paper argues that the laser-speckle disappearance line observed experimentally at about 50 GPa is not the melting curve but the fcc–bcc boundary. The same potential also predicts two low-temperature hcp pockets, and the paper traces the bcc stabilisation to the potential's softer short-range repulsion compared with Lennard–Jones. A comparison with a foundation MACE model, which predicts unphysically compact tetrahedral clusters, supports the authors' point that physical form and appropriate training matter more than model flexibility.","pith_inferences":["If the bcc field is real, the same speckle-based reinterpretation should be examined for argon and xenon, whose melting curves show similar flattening anomalies and whose pair potentials are close relatives of the krypton one.","The greedy-snake mechanism may be a general feature of bcc phases near melting in van der Waals solids, implying that their high-temperature plastic flow is governed by correlated chain jumps rather than single-atom diffusion.","A direct test of the soft-repulsion hypothesis would be to compute the same phase diagram with pair potentials fitted to CCSD(T) data for argon and xenon: if all three develop bcc fields, the phenomenon is generic to accurate rare-gas two-body potentials.","The paper's MACE comparison suggests that foundation machine-learned models should be benchmarked on phase diagrams, not just forces and energies, before being used for high-pressure predictions in systems outside their training domain."],"forward_implications":["Above about 36 GPa and 2630 K, the stable solid against melting is bcc, not fcc, and the fcc–bcc coexistence line extends metastably into the liquid region.","The laser-speckle 'melting' anomaly near 50 GPa should be read as an fcc–bcc transition, so the true krypton melt curve is steeper than speckle experiments suggest.","The bcc phase's greedy-snake defects provide a plasticity mechanism that can change surface morphology rapidly, which explains why a surface-sensitive speckle diagnostic could mistake a solid–solid transition for melting.","Krypton's bcc field is caused by a softer short-range repulsion than Lennard–Jones supplies, suggesting that accurate two-body fits can capture phase behaviour that simple generic potentials miss.","The fcc–hcp enthalpy differences are below 0.1 meV per atom, so entropy and zero-point motion decide the low-temperature stacking; fcc wins because it has higher entropy, closing the hcp pockets with temperature."],"supporting_citations":[{"why":"supplies the CCSD(T) dimer and trimer interaction energies that constitute the entire training data for the Tadah!Kr3b potential.","marker":"[16]"},{"why":"defines the Tadah!Kr3b potential used for every simulation in the paper.","marker":"[14]"},{"why":"provides the laser-speckle melting measurements whose ~50 GPa anomaly is reinterpreted as the fcc-bcc transition.","marker":"[5]"},{"why":"reported bcc in xenon and reinterpreted its speckle melt curve, the direct template for the krypton claim.","marker":"[27]"},{"why":"provides the absolute free-energy integration used as the independent check that bcc is truly lower in free energy.","marker":"[9]"},{"why":"supplies the structural diagnostic used to identify fcc, bcc, hcp and other local environments in the molecular-dynamics snapshots.","marker":"[41]"},{"why":"describes the greedy-snake/crowdion diffusion mechanism in bcc that the paper identifies as the plasticity source.","marker":"[54]"},{"why":"introduces Gibbs–Duhem integration that the Clapeyron slope field generalises for tracing phase boundaries.","marker":"[31]"}],"fun_headline_variants":["Krypton's bcc phase may explain 50 GPa melt anomaly","Bcc krypton: a new twist on the melt curve","Krypton's plastic bcc phase challenges melt-line interpretation","Greedy snakes: krypton's bcc defect explains melting mystery","Krypton phase diagram reveals bcc region and plastic defects"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The prediction rests on assuming that a two-body potential fitted only to dimer and trimer energies correctly captures how the Gibbs free energies of fcc and bcc krypton differ at 36 GPa and 2600 K, a condition the paper's own melt-curve comparison does not directly test.","fun_headline_variants_meta":{"raw":{"variants":["Krypton's bcc phase may explain 50 GPa melt anomaly","Bcc krypton: a new twist on the melt curve","Krypton's plastic bcc phase challenges melt-line interpretation","Greedy snakes: krypton's bcc defect explains melting mystery","Krypton phase diagram reveals bcc region and plastic defects"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000212,"raw_usage":{"total_tokens":1442,"prompt_tokens":991,"completion_tokens":451,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":607,"completion_tokens_details":{"reasoning_tokens":358}},"tokens_in":607,"tokens_out":451,"duration_ms":3669,"temperature":1.0,"reasoning_tokens":358,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T14:31:32.964095+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A laser-heated diamond-anvil-cell experiment on krypton between 35 and 60 GPa using x-ray diffraction as the diagnostic would settle it: if no bcc reflections appear and the speckle-disappearance line coincides with the onset of liquid diffuse scattering, the proposed bcc field and the reinterpretation of the melt anomaly are wrong.","supporting_citations":[],"review_version":1}