Pith. sign in

REVIEW 4 major objections 8 minor 88 references

Machine-Learning Potentials for sodium-potassium chloride mixtures: Predicting thermophysical properties and phase behavior of multicomponent salts

T0 review · 4 major / 8 minor · reviewed 2026-07-31 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read One first-principles machine-learning potential tracks NaCl–KCl molten-salt trends across composition and temperature, but absolute values still need experimental anchors.

desk verdict Solid multi-chemistry MTP package for NaCl–KCl/NaK with honest hybrid-modeling framing; miscibility-gap numbers rest on a weaker free-energy diagnostic than the liquidus/solidus work. read the letter →

arxiv 2607.25022 v1 pith:EGXPWQG2 submitted 2026-07-27 cond-mat.mtrl-sci

Karim Zongo , Hao Sun , Zijian Meng , Christopher Maxwell , Edmanuel Torres , Laurent Karim Béland This is my paper
classification cond-mat.mtrl-sci
keywords machine-learninginteratomicpotentialsmomenttensorpotentialmoltensaltsNaCl–KClphasediagramthermophysicalpropertiesdispersioncorrectionmoleculardynamics
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Direct quantum simulations cannot reach the sizes and times needed for transport and phase behavior in multicomponent molten salts. This paper builds one transferable moment tensor potential from DFT data on NaCl, KCl, their mixtures, and the NaK alloy, then runs large molecular-dynamics simulations of density, diffusion, structure, heat capacity, thermal conductivity, and the NaCl–KCl phase diagram. The model recovers many temperature- and composition-dependent trends, including solidus/liquidus topology and solid miscibility. Absolute numbers often sit systematically off experiment, and adding a D3 dispersion correction helps some systems while hurting others. The authors therefore argue for a hybrid workflow: use the potential for transferable trends and mechanism, and fold in experiment for quantitative engineering accuracy.

What carries the argument

The moment tensor potential (MTP): a systematically expandable many-body machine-learning interatomic potential fit to DFT energies, forces, and stresses via small-cell active learning, optionally paired in MD with a Grimme D3 correction as a related modeling variant.

What would settle it

Measure solidus, liquidus, and densities for several NaCl–KCl compositions (especially NaCl-rich) under the same conditions and check whether the reported MTP/MTP-D3 offsets shrink or grow when the potential is retrained with a different functional or with experimental targets included.

Watch

Extended reading notes

Core claim

A single level-22 moment tensor potential, trained on small-cell DFT-PBE configurations spanning NaCl, KCl, NaCl–KCl mixtures, and NaK, can reproduce the main temperature- and composition-dependent trends in liquid structure, thermophysical and transport properties, and the NaCl–KCl phase diagram, while systematic offsets in absolute properties remain and post-hoc D3 dispersion is not uniformly beneficial.

Load-bearing premise

That DFT-PBE reference data without built-in dispersion, plus ideal mixing entropy and some extrapolated free-energy crossings, are accurate enough to place absolute melting lines and miscibility limits.

Editorial extensions

If this is right

  • One potential can interpolate molten-salt properties into composition regions that experiments have not fully mapped.
  • Dispersion corrections must be treated as system- and property-dependent variants, not automatic upgrades.
  • Engineering use of such models should plan for experimental calibration of absolute melting points, densities, and diffusivities.
  • Phase-diagram topology (solidus/liquidus and miscibility gap shape) is a realistic target for first-principles-informed MLIPs even when absolute temperatures shift.
  • Downstream CALPHAD or CFD workflows should carry explicit systematic-offset corrections when ingesting MTP data.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The large NaCl melting offset relative to KCl suggests cation-specific DFT bias that may worsen in richer multicomponent chloride blends containing Na.
  • Training with a small set of experimental densities or melting points as soft targets could cut absolute errors without losing composition-trend transferability.
  • The same small-cell active-learning recipe is a natural template for next systems such as MgCl2- or fluoride-bearing coolants once end-member offsets are characterized.
  • Fixed-density RDF agreement paired with NPT density errors implies that local structure is easier for the potential than the equation of state.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 8 minor

Summary. The manuscript presents a level-22 moment tensor potential (MTP, 935 parameters) trained on DFT-PBE reference data assembled via small-cell active learning, covering NaCl, KCl, NaCl–KCl mixtures, and the NaK alloy. The authors also test an MTP-D3 variant in which Grimme's D3 correction (BJ damping, 20 Å cutoff) is added during MD only, without retraining on DFT-D3 labels. The potentials are applied to liquid densities, partial RDFs, heat capacities, thermal conductivities (Müller–Plathe), self-diffusion coefficients, end-member melting points and enthalpies of fusion (moving-interface method), the NaCl–KCl solidus/liquidus, and the solid-state miscibility gap (Frenkel–Ladd thermodynamic integration with ideal configurational entropy). The paper's central claim is that one transferable potential reproduces many temperature- and composition-dependent trends and the overall phase-diagram topology, while absolute values retain systematic, composition-dependent errors (e.g., NaCl melting overestimated by ~84 K; diffusion systematically underestimated; NaCl bulk modulus overestimated). The authors explicitly argue for a hybrid framework in which MLIP predictions provide trends and experimental anchors provide calibration.

Significance. If the results hold, this is a useful and honest contribution: a single transferable MTP covering NaCl, KCl, NaCl–KCl, and NaK that reproduces many temperature- and composition-dependent trends in liquid structure, density, Cp, κ, diffusion, and the NaCl–KCl phase diagram, with quantified percent deviations against experiment throughout. The systematic MTP vs MTP-D3 comparison — showing D3 helps KCl densities (15.74%→2.35%) but badly hurts NaCl (1.45%→16.78%) — is a genuinely instructive result for the molten-salt MLIP community, and the authors' explicit framing of trend-fidelity vs absolute-accuracy limitations (e.g., the ~84 K NaCl melting offset) is a strength. Reproducibility is a further strength: training data, potentials, and analysis scripts are released on a public repository, and the declared AI-use disclosure is appropriately scoped. The work does not yet deliver quantitatively reliable absolute phase boundaries, and the authors say so themselves; its value lies in the transferable trend-level capability plus the clear-eyed error budget.

major comments (4)
  1. [§II.F, Eqs. (9)–(11) and Fig. 10] The miscibility-gap construction is not, in general, the thermodynamic binodal. Eq. (11) evaluates ΔG_mix(x,T) for a homogeneous solid solution relative to the chord joining the end members, and the 'miscibility limit' is taken as the T at which ΔG_mix = 0 at fixed x (interpolated for x=0.10, linearly extrapolated for all other compositions). For a binary with a miscibility gap, the binodal is given by the common-tangent/convex-hull construction on the full G(x) curve, and the fixed-composition condition ΔG_mix=0 coincides with it only in special cases (e.g., symmetric regular solution at the critical composition). Since the authors compute G(x,T) at multiple compositions anyway, a common-tangent analysis is feasible with existing data and would either validate or revise the reported miscibility limits and the 7.11% mean-deviation figure in §III.D. As written, the Fig. 10 'MTP-D3 misc' p
  2. [§II.F] The Frenkel–Ladd TI protocol uses only 5000 steps (5 ps) each for equilibration and integration, with forward and backward switching. For solid solutions with the large Na/K size mismatch, this is short, and no TI convergence data, forward/backward hysteresis, or error bars on the absolute free energies are reported in the main text. Because the miscibility limits are extracted from small differences of large free energies (and often extrapolated), the TI statistical and systematic uncertainty is load-bearing for the miscibility-gap claim. Please report the TI hysteresis and block-averaged uncertainties, and justify (or extend) the integration length; a convergence check at one representative composition and temperature would suffice.
  3. [§II.E, Fig. 10] For intermediate compositions, the moving-interface method brackets a coexistence interval ('highest T remaining solid' vs 'lowest T fully liquid' within 1 ns), and these bounds are presented as solidus and liquidus in Fig. 10. For a binary alloy, equilibrium coexistence generally involves composition partitioning between solid and liquid (distinct tie-line endpoints); the manuscript does not report whether the solid and liquid compositions were measured during coexistence or whether the bracketed bounds correspond to peritectic-like kinetic limits rather than equilibrium boundaries. This matters most for the 90 mol% NaCl points, where the largest deviations (liquidus 5.04%, solidus 9.36%) occur and where the pure-NaCl melting offset of ~84 K already propagates. At minimum, the text should clarify what the two bracketing temperatures physically represent and quantify their sensitivity to
  4. [§III.C–E] All transport (D, κ, Cp) and phase-equilibrium results are reported only for MTP-D3, while §III.A–B and §III.E themselves demonstrate that D3 is not uniformly beneficial — most strikingly, the NaCl density deviation worsens from 1.45% (MTP) to 16.78% (MTP-D3), yet NaCl melting, κ, Cp, and D are evaluated with the D3 variant. The authors acknowledge in §III.E that 'the direct effect of D3 on these observables cannot be isolated from the present data,' but this leaves the paper's central engineering recommendation (which variant to trust for which property) under-supported for exactly the properties that matter most. A baseline-MTP calculation of, at minimum, the end-member melting points and one transport property per system would substantially strengthen the conclusions and is computationally inexpensive given the framework in hand.
minor comments (8)
  1. [Abstract] Typo: 'trained using a a DFT dataset' (also 'that arises' in §I should be 'that arise'; 'due systematic errors' in §I is missing 'to'; 'uniformly improve' in the Conclusion should be 'uniformly improves').
  2. [§I, third paragraph] Formatting artifact: 'absolutepropertyvaluesneededtovalidate andcalibratepre-dictive models' appears as a run-together string.
  3. [§II.C] Unresolved reference markers: 'obtained from crystallographic repositories? ?' — the citations are missing.
  4. [§II.C] Training/validation RMSEs for energies, forces, and stresses of the final level-22 MTP (per system and overall) are not reported in the main text; a short table would help readers judge the quality of the underlying fit independently of the downstream property comparisons.
  5. [§II.F, Eqs. (9)–(11)] Eq. (9) defines ΔG_mix with a −TΔS_mix term while Eq. (11) writes it as +RT[x ln x + (1−x) ln(1−x)]; the signs are consistent given Eq. (10), but presenting Eq. (9) with ΔS_mix already substituted would avoid confusion.
  6. [Fig. 3 caption] The text gives the absolute MTP cost (~10⁻⁷ core·h·atom⁻¹·step⁻¹) but the figure reports a normalized cost; please state the normalization convention and, if possible, the hardware, so the 20–25% D3 overhead is interpretable.
  7. [Fig. 10] Please add symbols/error bars for the predicted miscibility points and indicate explicitly which compositions were interpolated (x=0.10) versus extrapolated; the reader currently cannot tell from the figure. The 'Exp misc' source should also be cited at the figure, not only via the general reference list.
  8. [§III.B, Fig. 6] RDFs are benchmarked only at fixed experimental densities; since the density deviations under NPT are up to ~16% for some systems, a brief note on how RDFs change at the model's own equilibrium density (or at least a pointer to SI) would clarify whether the good MTP-vs-DFT agreement in Fig. 6 survives at the densities the potential actually predicts.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: DFT-trained MTP predictions are validated against external experiment, not against fitted targets.

full rationale

The derivation chain is standard and non-circular. The MTP is fit to DFT-PBE energies, forces, and stresses from small-cell active learning; predicted densities, RDFs, Cp, thermal conductivity, diffusion, melting points, solidus/liquidus, and mixing free energies are then compared to independent experimental compilations. No experimental target used in training is later relabeled as a prediction. Self-citations (prior NaCl MTP, small-cell active learning) supply methods and software practice, not the NaCl–KCl phase-diagram or thermophysical results. Optional MTP-D3 is a fixed post-hoc Grimme correction with a converged cutoff, not a fit to the reported observables. Methodological concerns about ideal-entropy ΔG_mix=0 extrapolation versus a common-tangent binodal affect correctness of the miscibility interpretation, not circularity of the claim chain. The paper is self-contained against external benchmarks.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The central claim rests on standard MLIP/MD/DFT practice plus a few modeling choices that directly affect absolute phase and density numbers: PBE without dispersion as the PES, empirical D3 only at inference, ideal mixing entropy, active-learning grade cutoffs, and MTP complexity/cutoffs. No new physical entities are postulated; free parameters are model-complexity and correction settings rather than fits to the experimental phase diagram.

free parameters (5)
  • MTP lev_max (level-22) and 935 linear/radial coefficients = lev_max=22; 935 parameters
    Model capacity and all Θ={cβ, cμ,τi,τj} are fitted to the curated DFT set (1277 train configs); complexity is a chosen hyperparameter that controls bias/variance of every downstream property.
  • Active-learning extrapolation thresholds γ_select and γ_break = γ_select=2, γ_break=10
    Hand-chosen gates (2 and 10) decide which MD configs enter DFT training and when runs abort; they shape the support of the learned PES.
  • MTP radial cutoff R_c = 7.5 Å
    Local environment cutoff fixed at 7.5 Å; truncates many-body interactions entering all energies/forces.
  • D3 cutoff and damping scheme (BJ) = 20 Å cutoff, BJ damping
    Chosen after NaCl/KCl density/energy convergence tests; applied only in MD, not in training labels, and shifts absolute densities and lattices system-dependently.
  • Plane-wave cutoff and k-mesh policy for DFT labels = 900 eV
    900 eV and size-dependent Monkhorst–Pack meshes define the reference surface the MTP imitates; not fitted to experiment but chosen by convergence.
assumptions (6)
  • domain assumption PBE-GGA ultrasoft DFT without dispersion is an adequate reference PES for training transferable molten alkali-chloride/NaK potentials.
    Stated in Ab initio calculations; all MTP accuracy ceilings inherit functional and missing-vdW errors, visible in NaCl bulk modulus and melting offsets.
  • domain assumption Small-cell active-learning configurations (N=2–36, liquid-biased ~98%) suffice for large-cell transport and solid–liquid coexistence.
    Methods §C and prior small-cell AL citations; load-bearing for claiming transferability to 1728–16000 atom property runs and 5760-atom interfaces.
  • domain assumption Ideal configurational entropy ΔS_mix = −R[x ln x+(1−x)ln(1−x)] plus end-member TI free energies locates solid miscibility limits.
    Free energy calculation §F eqs. (9)–(11); non-ideal vibrational/excess entropy neglected; most compositions use extrapolation of ΔG_mix(T) to zero.
  • domain assumption Moving-interface NPT energy-drift zeros (pure salts) or solid/liquid brackets (mixtures) equal thermodynamic solidus/liquidus.
    §E; finite 1 ns trajectories and random cation substitutions can bias coexistence widths versus true equilibrium free-energy crossings.
  • standard math Moment tensor basis with rotational/permutational invariance can represent the multicomponent PES at fixed lev_max.
    MTP formalism §A, Shapeev framework; standard MLIP assumption.
  • ad hoc to paper Adding Grimme D3 only during MD is a valid probe of dispersion without retraining on DFT-D3 labels.
    §D and §E discussion; creates MTP-D3 as a hybrid variant whose forces are not variationally consistent with the training Hamiltonian.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Machine-Learning Potentials for sodium-potassium chloride mixtures: Predicting thermophysical properties and phase behavior of multicomponent salts." pith.science (2026). https://pith.science/paper/EGXPWQG2

@misc{pith2026260725022,
  author       = {Pith},
  title        = {Pith review of: Machine-Learning Potentials for sodium-potassium chloride mixtures: Predicting thermophysical properties and phase behavior of multicomponent salts},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EGXPWQG2}},
  note         = {Machine review of arXiv:2607.25022}
}
read the original abstract

Predicting the properties of multicomponent molten salts using density functional theory (DFT) remains challenging because the spatial and temporal scales required to evaluate transport properties and phase behavior are computationally prohibitive. In this work, we develop a moment tensor potential trained using a a DFT dataset of NaCl, KCl, NaCl-KCl mixtures, and the NaK alloy, enabling large-scale molecular dynamics simulations across wide ranges of temperatures and compositions. We systematically evaluate the effect of D3 dispersion corrections and apply the resulting potential to predict liquid densities, diffusion coefficients, radial distribution functions, heat capacities, thermal conductivities, and the NaCl-KCl phase diagram. The model successfully reproduces many temperature- and composition-dependent trends. However, systematic deviations in several absolute properties persist, highlighting the importance of experimental validation and calibration. These findings support a hybrid modeling framework in which first-principles-informed machine-learning potentials provide transferable predictive capability and mechanistic insight, while experimental data incorporated during model development or subsequent engineering assessments is necessary to improve quantitative accuracy.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

88 extracted references

  1. [1]

    Molten salt for advanced energy applica- tions: A review.Annals of Nuclear Energy, 169:108924, 2022

    Robin Roper, Megan Harkema, Piyush Sabharwall, Catherine Riddle, Brandon Chisholm, Brandon Day, and Paul Marotta. Molten salt for advanced energy applica- tions: A review.Annals of Nuclear Energy, 169:108924, 2022

    2022

  2. [2]

    Evaluation of salt coolants for reactor applications.Nuclear technology, 163(3):330–343, 2008

    David F Williams and KT Clarno. Evaluation of salt coolants for reactor applications.Nuclear technology, 163(3):330–343, 2008

    2008

  3. [3]

    Thermophysical properties ofliquidchloridesfrom600to1600k: Meltpoint, enthalpy of fusion, and volumetric expansion.Journal of Molecular Liquids, 346:118147, 2022

    Stephen Scott Parker, A Long, C Lhermitte, S Vogel, M Monreal, and JM Jackson. Thermophysical properties ofliquidchloridesfrom600to1600k: Meltpoint, enthalpy of fusion, and volumetric expansion.Journal of Molecular Liquids, 346:118147, 2022

    2022

  4. [4]

    Review of commercial thermal energy storage in concentrated solar power plants: Steam vs

    Edouard González-Roubaud, David Pérez-Osorio, and Cristina Prieto. Review of commercial thermal energy storage in concentrated solar power plants: Steam vs. molten salts.Renewable and sustainable energy reviews, 80:133–148, 2017

    2017

  5. [5]

    Operating temperature windows for fusion reactor structural materials.Fusion Engineering and design, 51:55–71, 2000

    SJ Zinkle and NM Ghoniem. Operating temperature windows for fusion reactor structural materials.Fusion Engineering and design, 51:55–71, 2000

    2000

  6. [6]

    Development of robust neural-network interatomic potential for molten salt.Cell Reports Physical Science, 2(3), 2021

    Qing-Jie Li, Emine Küçükbenli, Stephen Lam, Boris Khaykovich, Efthimios Kaxiras, and Ju Li. Development of robust neural-network interatomic potential for molten salt.Cell Reports Physical Science, 2(3), 2021

    2021

  7. [7]

    The phase equilibrium dia- gram for the kcl-nacl system.Materials Research Bulletin, 2(10):935–938, 1967

    DS Coleman and PDA Lacy. The phase equilibrium dia- gram for the kcl-nacl system.Materials Research Bulletin, 2(10):935–938, 1967

    1967

  8. [8]

    Thermo- dynamics of the nacl–kcl system.Thermochimica acta, 606:25–33, 2015

    Dmitry Sergeev, D Kobertz, and M Müller. Thermo- dynamics of the nacl–kcl system.Thermochimica acta, 606:25–33, 2015

    2015

Show all 88 references
  1. [9]

    Molten salt storage for power generation.Chemie Inge- nieur Technik, 93(4):534–546, 2021

    Thomas Bauer, Christian Odenthal, and Alexander Bonk. Molten salt storage for power generation.Chemie Inge- nieur Technik, 93(4):534–546, 2021

    2021

  2. [10]

    Physical properties data compilations relevant to energy storage

    George J Janz, Carolyn B Allen, NP Bansal, RM Murphy, and R Tomkins. Physical properties data compilations relevant to energy storage. ii. molten salts: data on sin- gle and multi-component salt systems. Technical report, Rensselaer Polytechnic Inst., Troy, NY (USA). Cogswell ...

    1979

  3. [11]

    Physical properties of molten-salt reactor fuel, coolant, and flush salts

    Stanley Cantor. Physical properties of molten-salt reactor fuel, coolant, and flush salts. Technical report, Oak Ridge National Lab., Tenn., 1968

    1968

  4. [12]

    Experimental investigation and thermodynamic modeling of an innovative molten salt for thermal energy storage (tes).Applied energy, 212:516–526, 2018

    Xiang Li, Shuang Wu, Yang Wang, and Leidong Xie. Experimental investigation and thermodynamic modeling of an innovative molten salt for thermal energy storage (tes).Applied energy, 212:516–526, 2018

    2018

  5. [13]

    Fan Yuan, Ming-Jia Li, Zhao Ma, Bo Jin, and Zhanbin Liu. Experimental study on thermal performance of high- temperature molten salt cascaded latent heat thermal energy storage system.International Journal of Heat and Mass Transfer, 118:997–1011, 2018

    2018

  6. [14]

    Structure of molten nacl and the decay of the pair-correlations.The Journal of Chemical Physics, 157(9), 2022

    Anita Zeidler, Philip S Salmon, Takeshi Usuki, Shinji Kohara, Henry E Fischer, and Mark Wilson. Structure of molten nacl and the decay of the pair-correlations.The Journal of Chemical Physics, 157(9), 2022

    2022

  7. [15]

    Microstructure and thermal properties of nacl–zncl2 molten salt by molecular dynam- ics simulation and experiment.Solar Energy Materials and Solar Cells, 250:112108, 2023

    Jing Ding, Fengchun Gong, Senfeng Yang, Weilong Wang, Shule Liu, and Jianfeng Lu. Microstructure and thermal properties of nacl–zncl2 molten salt by molecular dynam- ics simulation and experiment.Solar Energy Materials and Solar Cells, 250:112108, 2023

    2023

  8. [16]

    First-principles molecular dynamics modeling of the licl–kcl molten salt system

    Amelia Bengtson, Hyo On Nam, Saumitra Saha, Ridwan Sakidja, and Dane Morgan. First-principles molecular dynamics modeling of the licl–kcl molten salt system. Computational materials science, 83:362–370, 2014

    2014

  9. [17]

    Density-functional-based molecular-dynamics simulations of molten salts.The Journal of chemical physics, 123(13), 2005

    Sandrine Hazebroucq, Gérard S Picard, Carlo Adamo, Thomas Heine, Sibylle Gemming, and Gotthard Seifert. Density-functional-based molecular-dynamics simulations of molten salts.The Journal of chemical physics, 123(13), 2005

    2005

  10. [18]

    New advances in the study of local structure of molten binary salts.Physical review letters, 78(3):460, 1997

    Andrea Di Cicco, Marco Minicucci, and Adriano Filipponi. New advances in the study of local structure of molten binary salts.Physical review letters, 78(3):460, 1997

    1997

  11. [19]

    Experimental techniques in molten fluoride chemistry

    Carlos E Bamberger. Experimental techniques in molten fluoride chemistry. InAdvances in Molten Salt Chemistry: Volume 3, pages 177–248. Springer, 1975

    1975

  12. [20]

    Phase diagrams of binary and ternary fluoride systems

    Roy E Thoma. Phase diagrams of binary and ternary fluoride systems. InAdvances in Molten Salt Chemistry: Volume 3, pages 275–455. Springer, 1975

    1975

  13. [21]

    Assessment of candidate molten salt coolants for the advanced high temperature reactor (ahtr)

    David F Williams. Assessment of candidate molten salt coolants for the advanced high temperature reactor (ahtr). 2006

    2006

  14. [22]

    Corrosion in molten salts

    K Sridharan and TR Allen. Corrosion in molten salts. In Molten salts chemistry, pages 241–267. Elsevier, 2013

    2013

  15. [23]

    Inhomogeneous elec- tron gas.Physical review, 136(3B):B864, 1964

    Pierre Hohenberg and Walter Kohn. Inhomogeneous elec- tron gas.Physical review, 136(3B):B864, 1964

    1964

  16. [24]

    The structure of molten flinak.Journal of Nuclear Materials, 537:152219, 2020

    Benjamin A Frandsen, Stella D Nickerson, Austin D Clark, Andrew Solano, Raju Baral, Johnny Williams, Jörg Neue- feind, and Matthew Memmott. The structure of molten flinak.Journal of Nuclear Materials, 537:152219, 2020

    2020

  17. [25]

    First-principles-derived transport properties of molten chloride salts.Journal of Nuclear Materials, 585:154601, 2023

    Kai Duemmler, Michael Woods, Toni Karlsson, Ruchi Gakhar, and Benjamin Beeler. First-principles-derived transport properties of molten chloride salts.Journal of Nuclear Materials, 585:154601, 2023

    2023

  18. [26]

    Interionic potentials in alkali halides and their use in simulations of the molten salts

    MJL Sangster and M Dixon. Interionic potentials in alkali halides and their use in simulations of the molten salts. Advances in Physics, 25(3):247–342, 1976

    1976

  19. [27]

    Polarization effects in ionic systems from first principles.Journal of Physics: Con- densed Matter, 5(17):2687–2706, 1993

    M Wilson and PA Madden. Polarization effects in ionic systems from first principles.Journal of Physics: Con- densed Matter, 5(17):2687–2706, 1993

    1993

  20. [28]

    Reaxff reactive force field for molecular dynam- 14 ics simulations of hydrocarbon oxidation.The Journal of Physical Chemistry A, 112(5):1040–1053, 2008

    Kimberly Chenoweth, Adri CT Van Duin, and William A Goddard. Reaxff reactive force field for molecular dynam- 14 ics simulations of hydrocarbon oxidation.The Journal of Physical Chemistry A, 112(5):1040–1053, 2008

    2008

  21. [29]

    Tao Liang, Tzu-Ray Shan, Yu-Ting Cheng, Bryce D Devine, Mark Noordhoek, Yangzhong Li, Zhize Lu, Si- mon R Phillpot, and Susan B Sinnott. Classical atomistic simulations of surfaces and heterogeneous interfaces with the charge-optimized many body (comb) potentials.Mate- rials S...

    2013

  22. [30]

    Haimeng Wang, Ryan S DeFever, Yong Zhang, Fei Wu, Santanu Roy, Vyacheslav S Bryantsev, Claudio J Mar- gulis, and Edward J Maginn. Comparison of fixed charge and polarizable models for predicting the structural, ther- modynamic, and transport properties of molten alkali chlorid...

    2020

  23. [31]

    Adri CT Van Duin, Vyacheslav S Bryantsev, Mamadou S Diallo, William A Goddard, Obaidur Rahaman, Dou- glas J Doren, David Raymand, and Kersti Hermansson. Development and validation of a reaxff reactive force field for cu cation/water interactions and copper metal/metal oxide/me...

    2010

  24. [32]

    Reaxffmgh reactive force field for magnesium hydride systems.The Journal of Physical Chemistry A, 109(5):851–859, 2005

    Sam Cheung, Wei-Qiao Deng, Adri CT Van Duin, and William A Goddard. Reaxffmgh reactive force field for magnesium hydride systems.The Journal of Physical Chemistry A, 109(5):851–859, 2005

    2005

  25. [33]

    Lianchi Liu, Yi Liu, Sergey V Zybin, Huai Sun, and William A Goddard III. Reaxff-lg: Correction of the reaxff reactive force field for london dispersion, with applications to the equations of state for energetic materials.The Journal of Physical Chemistry A, 115(40):11016–11022, 2011

    2011

  26. [34]

    Machine learning interatomic potential: Bridge the gap between small-scale models and realistic device-scale simulations.Iscience, 27(5), 2024

    Guanjie Wang, Changrui Wang, Xuanguang Zhang, Ze- feng Li, Jian Zhou, and Zhimei Sun. Machine learning interatomic potential: Bridge the gap between small-scale models and realistic device-scale simulations.Iscience, 27(5), 2024

    2024

  27. [35]

    Computing chemical potentials with machine- learning-acceleratedsimulationstoaccuratelypredictther- modynamic properties of molten salts.Chemical Science, 16(7):3078–3091, 2025

    Luke D Gibson, Rajni Chahal, and Vyacheslav S Bryant- sev. Computing chemical potentials with machine- learning-acceleratedsimulationstoaccuratelypredictther- modynamic properties of molten salts.Chemical Science, 16(7):3078–3091, 2025

    2025

  28. [36]

    Development of deep potentials of molten mgcl2–nacl and mgcl2–kcl salts driven by machine learning.ACS Applied Materials & Interfaces, 15(11):14184–14195, 2023

    Tingrui Xu, Xuejiao Li, Yang Wang, and Zhongfeng Tang. Development of deep potentials of molten mgcl2–nacl and mgcl2–kcl salts driven by machine learning.ACS Applied Materials & Interfaces, 15(11):14184–14195, 2023

    2023

  29. [37]

    Theoretical prediction on the local structure and transport properties of molten alkali chlorides by deep potentials.Journal of Materials Science & Technology, 75:78–85, 2021

    Wenshuo Liang, Guimin Lu, and Jianguo Yu. Theoretical prediction on the local structure and transport properties of molten alkali chlorides by deep potentials.Journal of Materials Science & Technology, 75:78–85, 2021

    2021

  30. [38]

    Molecular dynamics simulations of molten magnesium chloride using machine-learning-based deep potential.Advanced Theory and Simulations, 3(12):2000180, 2020

    Wenshuo Liang, Guimin Lu, and Jianguo Yu. Molecular dynamics simulations of molten magnesium chloride using machine-learning-based deep potential.Advanced Theory and Simulations, 3(12):2000180, 2020

    2020

  31. [39]

    Mol- ecular dynamics simulation of molten strontium chloride based on deep potential.Journal of Molecular Liquids, 348:118380, 2022

    Di Guo, Jia Zhao, Wenshuo Liang, and Guimin Lu. Mol- ecular dynamics simulation of molten strontium chloride based on deep potential.Journal of Molecular Liquids, 348:118380, 2022

    2022

  32. [40]

    A dft accurate machine learning description of molten zncl2 and its mixtures: 1

    Gechuanqi Pan, Pin Chen, Hui Yan, and Yutong Lu. A dft accurate machine learning description of molten zncl2 and its mixtures: 1. potential development and properties prediction of molten zncl2.Computational Materials Science, 185:109955, 2020

    2020

  33. [41]

    Machine learning accelerates molten salt simulations: Thermal conductivity of mgcl2-nacl eutectic.Advanced Theory and Simulations, 5(8):2200206, 2022

    Wenshuo Liang, Guimin Lu, and Jianguo Yu. Machine learning accelerates molten salt simulations: Thermal conductivity of mgcl2-nacl eutectic.Advanced Theory and Simulations, 5(8):2200206, 2022

    2022

  34. [42]

    Machine- learning-driven simulations on microstructure and ther- mophysical properties of mgcl2–kcl eutectic.ACS applied materials & interfaces, 13(3):4034–4042, 2021

    Wenshuo Liang, Guimin Lu, and Jianguo Yu. Machine- learning-driven simulations on microstructure and ther- mophysical properties of mgcl2–kcl eutectic.ACS applied materials & interfaces, 13(3):4034–4042, 2021

    2021

  35. [43]

    Theoretical prediction on the redox potentials of rare-earth ions by deep potentials.Ionics, 27(5):2079–2088, 2021

    Jia Zhao, Wenshuo Liang, and Guimin Lu. Theoretical prediction on the redox potentials of rare-earth ions by deep potentials.Ionics, 27(5):2079–2088, 2021

  36. [44]

    A dft accurate machine learning description of molten zncl2 and its mixtures: 2

    Gechuanqi Pan, Jing Ding, Yunfei Du, Duu-Jong Lee, and Yutong Lu. A dft accurate machine learning description of molten zncl2 and its mixtures: 2. potential development and properties prediction of zncl2-nacl-kcl ternary salt for csp.Computational Materials Science, 187:110055, 2021

    2021

  37. [45]

    Transferable deep learning potential reveals intermediate- range ordering effects in lif–naf–zrf4 molten salt.Jacs Au, 2(12):2693–2702, 2022

    Rajni Chahal, Santanu Roy, Martin Brehm, Shubhojit Banerjee, Vyacheslav Bryantsev, and Stephen T Lam. Transferable deep learning potential reveals intermediate- range ordering effects in lif–naf–zrf4 molten salt.Jacs Au, 2(12):2693–2702, 2022

    2022

  38. [46]

    Moment tensor potentials: A class of systematically improvable interatomic potentials

    Alexander V Shapeev. Moment tensor potentials: A class of systematically improvable interatomic potentials. Multiscale Modeling & Simulation, 14(3):1153–1173, 2016

    2016

  39. [47]

    Interatomic potential for sodium and chlorine in both neutral and ionic states.Physical Review B, 109(17):174113, 2024

    Hao Sun, Christopher Maxwell, Edmanuel Torres, and Laurent Karim Béland. Interatomic potential for sodium and chlorine in both neutral and ionic states.Physical Review B, 109(17):174113, 2024

    2024

  40. [48]

    Dft accurate interatomic potential for molten nacl from machine learning.The Journal of Physical Chemistry C, 124(47):25760–25768, 2020

    Samuel Tovey, Anand Narayanan Krishnamoorthy, Ganesh Sivaraman, Jicheng Guo, Chris Benmore, Andreas Heuer, and Christian Holm. Dft accurate interatomic potential for molten nacl from machine learning.The Journal of Physical Chemistry C, 124(47):25760–25768, 2020

    2020

  41. [49]

    Solubility of sodium in sodium chloride: a density func- tional theory molecular dynamics study.Journal of The Electrochemical Society, 161(8):E3042–E3048, 2014

    Maoyuan Liu, Patrick Masset, and Angus Gray-Weale. Solubility of sodium in sodium chloride: a density func- tional theory molecular dynamics study.Journal of The Electrochemical Society, 161(8):E3042–E3048, 2014

    2014

  42. [50]

    New benchmark set of transition-metal coordination reactions for the assessment of density func- tionals.Journal of chemical theory and computation, 10(8):3092–3103, 2014

    Thomas Weymuth, Erik PA Couzijn, Peter Chen, and Markus Reiher. New benchmark set of transition-metal coordination reactions for the assessment of density func- tionals.Journal of chemical theory and computation, 10(8):3092–3103, 2014

    2014

  43. [51]

    Vojtech Kostal, Philip E Mason, Hector Martinez-Seara, and Pavel Jungwirth. Common cations are not polarizable: Effects of dispersion correction on hydration structures from ab initio molecular dynamics.The Journal of Phys- ical Chemistry Letters, 14(19):4403–4408, 2023

    2023

  44. [52]

    Accelerating training of mlips through small-cell training.Journal of Materials Research, 38(24):5095–5105, 2023

    Jason A Meziere, Yu Luo, Yi Xia, Laurent Karim Bé- land, Mark R Daymond, and Gus LW Hart. Accelerating training of mlips through small-cell training.Journal of Materials Research, 38(24):5095–5105, 2023

    2023

  45. [53]

    A set of moment tensor potentials for zirconium with increasing complexity.Journal of Chemical Theory and Computation, 19(19):6848–6856, 2023

    Yu Luo, Jason A Meziere, German D Samolyuk, Gus LW Hart, Mark R Daymond, and Laurent Karim Béland. A set of moment tensor potentials for zirconium with increasing complexity.Journal of Chemical Theory and Computation, 19(19):6848–6856, 2023

    2023

  46. [54]

    Small-cell-based fast active learning of machine learning interatomic potentials.Computational Materials Science, 256:113919, 2025

    Zijian Meng, Hao Sun, Edmanuel Torres, Christopher Maxwell, Ryan Eric Grant, and Laurent Karim Béland. Small-cell-based fast active learning of machine learning interatomic potentials.Computational Materials Science, 256:113919, 2025

    2025

  47. [55]

    The mlip 15 package: moment tensor potentials with mpi and ac- tive learning.Machine Learning: Science and Technology, 2(2):025002, 2021

    Ivan S Novikov, Konstantin Gubaev, Evgeny V Podryabinkin, and Alexander V Shapeev. The mlip 15 package: moment tensor potentials with mpi and ac- tive learning.Machine Learning: Science and Technology, 2(2):025002, 2021

    2021

  48. [56]

    Zuo and et al

    Y. Zuo and et al. Performance and cost assessment of machine learning interatomic potentials.The Journal of Physical Chemistry A, 124:731–745, 2020

    2020

  49. [57]

    Quantum espresso: a modular and open-source software project for quantum simulations of materials.Journal of physics: Condensed matter, 21(39):395502, 2009

    Paolo Giannozzi, Stefano Baroni, Nicola Bonini, Matteo Calandra, Roberto Car, Carlo Cavazzoni, Davide Ceresoli, GuidoLChiarotti, MatteoCococcioni, IsmailaDabo, etal. Quantum espresso: a modular and open-source software project for quantum simulations of materials.Journal of ph...

    2009

  50. [58]

    Advanced capabilities for materials modelling with quantum espresso.Journal of physics: Condensed matter, 29(46):465901, 2017

    Paolo Giannozzi, Oliviero Andreussi, Thomas Brumme, Oana Bunau, M Buongiorno Nardelli, Matteo Calandra, Roberto Car, Carlo Cavazzoni, Davide Ceresoli, Matteo Cococcioni, et al. Advanced capabilities for materials modelling with quantum espresso.Journal of physics: Condensed ma...

    2017

  51. [59]

    From ultra- soft pseudopotentials to the projector augmented-wave method.Physical review b, 59(3):1758, 1999

    Georg Kresse and Daniel Joubert. From ultra- soft pseudopotentials to the projector augmented-wave method.Physical review b, 59(3):1758, 1999

    1999

  52. [60]

    Generalized gradient approximation made simple.Physi- cal review letters, 77(18):3865, 1996

    John P Perdew, Kieron Burke, and Matthias Ernzerhof. Generalized gradient approximation made simple.Physi- cal review letters, 77(18):3865, 1996

    1996

  53. [61]

    Assess- ment of the perdew–burke–ernzerhof exchange-correlation functional.The Journal of chemical physics, 110(11):5029– 5036, 1999

    Matthias Ernzerhof and Gustavo E Scuseria. Assess- ment of the perdew–burke–ernzerhof exchange-correlation functional.The Journal of chemical physics, 110(11):5029– 5036, 1999

    1999

  54. [62]

    Special points for brillouin-zone integrations.Physical review B, 13(12):5188, 1976

    Hendrik J Monkhorst and James D Pack. Special points for brillouin-zone integrations.Physical review B, 13(12):5188, 1976

    1976

  55. [63]

    Lammps-a flexible simula- tion tool for particle-based materials modeling at the atomic, meso, and continuum scales.Computer physics communications, 271:108171, 2022

    Aidan P Thompson, H Metin Aktulga, Richard Berger, Dan S Bolintineanu, W Michael Brown, Paul S Crozier, Pieter J In’t Veld, Axel Kohlmeyer, Stan G Moore, Trung Dac Nguyen, et al. Lammps-a flexible simula- tion tool for particle-based materials modeling at the atomic, meso, and...

    2022

  56. [64]

    The nose-hoover thermostat

    DJ Evans and BL Holian. The nose-hoover thermostat. Journal of Chemical Physics, 83(8):4069–4074, 1985

    1985

  57. [65]

    Polymorphic transitions in single crystals: A new molecular dynamics method.Journal of Applied physics, 52(12):7182–7190, 1981

    Michele Parrinello and Aneesur Rahman. Polymorphic transitions in single crystals: A new molecular dynamics method.Journal of Applied physics, 52(12):7182–7190, 1981

    1981

  58. [66]

    Active learning of linearly parametrized interatomic potentials

    Evgeny V Podryabinkin and Alexander V Shapeev. Active learning of linearly parametrized interatomic potentials. Computational Materials Science, 140:171–180, 2017

    2017

  59. [67]

    Mlip-3: Active learning on atomic environments with moment tensor potentials

    Evgeny Podryabinkin, Kamil Garifullin, Alexander Shapeev, and Ivan Novikov. Mlip-3: Active learning on atomic environments with moment tensor potentials. The Journal of Chemical Physics, 159(8), 2023

    2023

  60. [68]

    Accelerating crystal structure prediction by machine-learning inter- atomic potentials with active learning.Physical Review B, 99(6):064114, 2019

    Evgeny V Podryabinkin, Evgeny V Tikhonov, Alexan- der V Shapeev, and Artem R Oganov. Accelerating crystal structure prediction by machine-learning inter- atomic potentials with active learning.Physical Review B, 99(6):064114, 2019

    2019

  61. [69]

    Ephemeral data derived potentials for random structure search.Physical Review B, 106(1):014102, 2022

    Chris J Pickard. Ephemeral data derived potentials for random structure search.Physical Review B, 106(1):014102, 2022

    2022

  62. [70]

    Crystal structure prediction using ab initio evolutionary techniques: Princi- ples and applications.The Journal of chemical physics, 124(24), 2006

    Artem R Oganov and Colin W Glass. Crystal structure prediction using ab initio evolutionary techniques: Princi- ples and applications.The Journal of chemical physics, 124(24), 2006

    2006

  63. [71]

    How evolutionary crystal structure prediction works—and why.Accounts of chemical research, 44(3):227–237, 2011

    Artem R Oganov, Andriy O Lyakhov, and Mario Valle. How evolutionary crystal structure prediction works—and why.Accounts of chemical research, 44(3):227–237, 2011

    2011

  64. [72]

    New developments in evolutionary struc- ture prediction algorithm uspex.Computer Physics Com- munications, 184(4):1172–1182, 2013

    Andriy O Lyakhov, Artem R Oganov, Harold T Stokes, and Qiang Zhu. New developments in evolutionary struc- ture prediction algorithm uspex.Computer Physics Com- munications, 184(4):1172–1182, 2013

    2013

  65. [73]

    Randspg: an open-source program for generating atomistic crystal structures with specific spacegroups.Computer Physics Communications, 213:208–216, 2017

    Patrick Avery and Eva Zurek. Randspg: an open-source program for generating atomistic crystal structures with specific spacegroups.Computer Physics Communications, 213:208–216, 2017

    2017

  66. [74]

    A consistent and accurate ab initio parametriza- tion of density functional dispersion correction (dft-d) for the 94 elements h-pu.The Journal of chemical physics, 132(15), 2010

    Stefan Grimme, Jens Antony, Stephan Ehrlich, and Helge Krieg. A consistent and accurate ab initio parametriza- tion of density functional dispersion correction (dft-d) for the 94 elements h-pu.The Journal of chemical physics, 132(15), 2010

    2010

  67. [75]

    Effect of the damping function in dispersion corrected density functional theory.Journal of computational chemistry, 32(7):1456–1465, 2011

    Stefan Grimme, Stephan Ehrlich, and Lars Goerigk. Effect of the damping function in dispersion corrected density functional theory.Journal of computational chemistry, 32(7):1456–1465, 2011

    2011

  68. [76]

    Atomicclusterexpansion for quantum-accurate large-scale simulations of carbon

    Minaam Qamar, Matous Mrovec, Yury Lysogorskiy, An- tonBochkarev, andRalfDrautz. Atomicclusterexpansion for quantum-accurate large-scale simulations of carbon. Journal of chemical theory and computation, 19(15):5151– 5167, 2023

    2023

  69. [77]

    A comparison of methods for melting point calculation using molecular dynamics simulations.The Journal of chemical physics, 136(14), 2012

    Yong Zhang and Edward J Maginn. A comparison of methods for melting point calculation using molecular dynamics simulations.The Journal of chemical physics, 136(14), 2012

    2012

  70. [78]

    Nonequilibrium free-energy calculation of solids using lammps.Computational Materials Science, 112:333–341, 2016

    Rodrigo Freitas, Mark Asta, and Maurice De Koning. Nonequilibrium free-energy calculation of solids using lammps.Computational Materials Science, 112:333–341, 2016

    2016

  71. [79]

    Vegard’s law.Phys- ical review A, 43(6):3161, 1991

    Alan R Denton and Neil W Ashcroft. Vegard’s law.Phys- ical review A, 43(6):3161, 1991

    1991

  72. [80]

    Structural properties of nacl and kcl under pressure.Journal of Physics C: Solid State Physics, 19(15):2623–2632, 1986

    Sverre Froyen and Marvin L Cohen. Structural properties of nacl and kcl under pressure.Journal of Physics C: Solid State Physics, 19(15):2623–2632, 1986

    1986

  73. [81]

    Studies of nacl-kcl solid solutions

    WT Barrett and WE Wallace. Studies of nacl-kcl solid solutions. i. heats of formation, lattice spacings, densities, schottky defects and mutual solubilities1, 2.Journal of the American Chemical Society, 76(2):366–369, 1954

    1954

  74. [82]

    AD Kirshenbaum, JA Cahill, PJ McGonigal, and AV Grosse. The density of liquid nacl and kcl and an estimate of their critical constants together with those of the other alkali halides.Journal of Inorganic and Nuclear chemistry, 24(10):1287–1296, 1962

    1962

  75. [83]

    The density of liq- uid sodium–potassium eutectic.High temperature, 41(3):340–345, 2003

    Alexandr Gerasimovich Mozgovoi, Vladimir Vladimirovich Roshchupkin, Sergei Niko- laevich Skovorod’ko, Mikhail Alexandrovich Pokrasin, and Aleksei Ivanovich Chernov. The density of liq- uid sodium–potassium eutectic.High temperature, 41(3):340–345, 2003

    2003

  76. [84]

    Experi- mental determination of the thermal diffusivity of molten alkali halides by the forced rayleigh scattering method

    Y Nagasaka, N Nakazawa, and A Nagashima. Experi- mental determination of the thermal diffusivity of molten alkali halides by the forced rayleigh scattering method. i. molten licl, nacl, kcl, rbcl, and cscl.International journal of thermophysics, 13(4):555–574, 1992

    1992

  77. [85]

    Thermodynamic investigation of the nacl- kcl salt system from 25 to 950°c.Journal of Molecular Liquids, 391:122591, 2023

    Jason Lonergan, Vitaliy Goncharov, Michaella Swinhart, Kyle Makovsky, Mark Rollog, Bruce McNamara, Richard Clark, Derek Cutforth, Christopher Armstrong, Xiaofeng Guo, et al. Thermodynamic investigation of the nacl- kcl salt system from 25 to 950°c.Journal of Molecular Liquids,...

    2023

  78. [86]

    Thermal conductivity in molten alkali halides: composition dependence in mixtures of (na–k) cl

    Keiichi Takase, Yukiyo Matsumoto, Keisuke Sato, and Norikazu Ohtori. Thermal conductivity in molten alkali halides: composition dependence in mixtures of (na–k) cl. Molecular Simulation, 38(5):432–436, 2012

    2012

  79. [87]

    Nist-janaf thermochemical tables

    Malcolm W Chase et al. Nist-janaf thermochemical tables. Journal of physical and chemical reference data, 28:1951, 1998

    1951

  80. [88]

    Heat capacities and thermal co- efficients of sodium’s and eutectic sodium–potassium’s coolants for nuclear reactors.Applied Sciences, 15(13):7566, 2025

    Nikolay E Dubinin. Heat capacities and thermal co- efficients of sodium’s and eutectic sodium–potassium’s coolants for nuclear reactors.Applied Sciences, 15(13):7566, 2025

    2025

Pith tools

Reviewed July 31, 2026 · model on record in the stance chip above.