pith. sign in

arxiv: 2606.25991 · v1 · pith:L5SXLVLInew · submitted 2026-06-24 · ❄️ cond-mat.mtrl-sci

Capturing Nuclear Quantum Effects in Hydrogen Diffusion through MoS2 via Machine-Learning-Enhanced Path-Integral Simulations

Ismail Eren , Ege Yigit Erbil , Maria-Judith Caisachana-Lozada , Hossein Mirhosseini , Thomas D. K\"uhne , Agnieszka B. Kuc This is my paper

Pith reviewed 2026-06-25 19:44 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci
keywords hydrogen diffusionMoS2nuclear quantum effectspath-integral molecular dynamicsmachine learning potentialskinetic isotope effecttwisted bilayerfree energy barriers
0
0 comments X

The pith

Nuclear quantum effects lower free-energy barriers for hydrogen diffusion in MoS2 at 300 K and produce a 35 meV kinetic isotope effect between H and D.

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

The paper runs path-integral molecular dynamics on machine-learned potentials fitted to DFT data to track hydrogen and deuterium atoms moving between layers of MoS2 in several polytypes and twisted bilayers. Treating the nuclei as quantum objects rather than point masses reduces the free-energy barriers that control diffusion at room temperature, which raises the calculated self-diffusion coefficient relative to classical-nucleus runs. The same quantum treatment creates a clear difference between hydrogen and deuterium barriers. In moiré structures the local stacking pattern further modulates the barriers from place to place.

Core claim

Simulations that combine well-tempered metadynamics with path-integral molecular dynamics show that nuclear quantum effects substantially lower free-energy barriers for hydrogen diffusion at 300 K, increasing the self-diffusion coefficient compared with classical nuclei simulations; the same calculations identify a 35 meV difference between the quantum free-energy barriers of hydrogen and deuterium, and they reveal strong spatial variations in transport inside the moiré superlattices of twisted bilayer MoS2.

What carries the argument

Machine-learning interatomic potentials trained on r2SCAN+rVV10 DFT data, used inside path-integral molecular dynamics combined with well-tempered metadynamics to compute quantum free-energy surfaces for diffusion.

If this is right

  • Accurate modeling of hydrogen transport rates through layered 2D materials at room temperature requires explicit inclusion of nuclear quantum effects.
  • Twisted bilayer MoS2 exhibits position-dependent diffusion rates set by the local stacking environments inside the moiré pattern.
  • The observed kinetic isotope effect supplies a microscopic basis for isotope-selective transport in structurally complex 2D systems.
  • The same simulation protocol can be applied to other polytypes and twist angles to map how stacking controls quantum-corrected barriers.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar path-integral studies on other van der Waals materials could test whether nuclear quantum effects remain decisive for hydrogen diffusion at room temperature.
  • The spatial modulation inside moiré superlattices suggests that twist angle could be used to engineer preferred diffusion channels.
  • Lowering temperature would likely amplify the tunneling contribution and widen the isotope difference further.

Load-bearing premise

The machine-learning potentials trained on the chosen DFT data accurately reproduce the potential-energy surface that governs hydrogen motion across the relevant MoS2 configurations and temperatures.

What would settle it

An experimental measurement at 300 K that finds the hydrogen self-diffusion coefficient in MoS2 to be indistinguishable from the value obtained in a classical-nuclei simulation, or that finds no 35 meV barrier difference between hydrogen and deuterium.

Figures

Figures reproduced from arXiv: 2606.25991 by Agnieszka B. Kuc, Ege Yigit Erbil, Hossein Mirhosseini, Ismail Eren, Maria-Judith Caisachana-Lozada, Thomas D. K\"uhne.

Figure 1
Figure 1. Figure 1: FIG. 1. Atomic structures of high-symmetry stackings of MoS [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Radial pair distribution functions, [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Free-energy surfaces (FES) for hydrogen diffusion in selected MoS [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Comparison of classical (a, b) and quantum (c, d) [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Atomic models and moiré superlattice domains [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7. Training dynamics and predictive accuracy of the neural network potential. (a) Evolution of training and validation [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: FIG. 8. Validation of the MLIP against DFT for various [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: FIG. 9. Comparison of the phonon dispersion relations calculated using Density Functional Theory (DFT) and the developed [PITH_FULL_IMAGE:figures/full_fig_p014_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: FIG. 10. Thermal equilibration and dynamical stability of MoS [PITH_FULL_IMAGE:figures/full_fig_p014_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: FIG. 11. Convergence of number of beads via well-tempered PIMD calculations for H atom in [PITH_FULL_IMAGE:figures/full_fig_p015_11.png] view at source ↗
read the original abstract

Hydrogen transport through layered two-dimensional (2D) materials is central to technologies such as hydrogen storage, fuel cells, and isotope separation. Among these materials, MoS2 exhibits tunable interlayer diffusion properties, whose accurate theoretical description requires accounting for nuclear quantum effects (NQEs), including zero-point motion and tunneling. Here, we present a machine-learning-enhanced atomistic study of hydrogen and deuterium diffusion in layered MoS2 based on interatomic potentials trained on r2SCAN+rVV10 density-functional-theory data. Combining well-tempered metadynamics with path-integral molecular dynamics, we investigate diffusion across multiple MoS2 polytypes and twisted bilayer structures while explicitly incorporating NQEs. Our simulations show that NQEs substantially lower free-energy barriers for hydrogen diffusion at 300 K, significantly increasing the hydrogen self-diffusion coefficient compared to classical nuclei simulations. We further identify a pronounced kinetic isotope effect, with a 35 meV difference between hydrogen and deuterium quantum free-energy barriers. In twisted bilayer MoS2, hydrogen transport exhibits strong spatial variations governed by the local stacking environments within the moir\'e superlattices. These results highlight the critical role of NQEs in hydrogen transport through layered materials and provide atomistic insight intoisotope-selective diffusion in structurally complex 2D systems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 2 minor

Summary. The manuscript uses machine-learning interatomic potentials trained on r2SCAN+rVV10 DFT data together with path-integral molecular dynamics and well-tempered metadynamics to quantify nuclear quantum effects on hydrogen and deuterium diffusion through MoS2 polytypes and twisted bilayers. The central results are that NQEs lower free-energy barriers at 300 K (increasing the quantum self-diffusion coefficient relative to classical nuclei), a 35 meV H/D kinetic isotope effect is obtained, and transport in twisted bilayers shows strong spatial variation tied to local stacking in the moiré superlattice.

Significance. If the ML-potential transferability and sampling convergence hold, the work supplies concrete, falsifiable predictions for NQE-driven barrier reductions and isotope selectivity in structurally complex 2D materials. The combination of ML potentials with PIMD-metadynamics enables atomistic resolution of moiré-scale variations that would be inaccessible to direct ab initio sampling, strengthening the case for including nuclear quantum effects in hydrogen-transport modeling.

minor comments (2)
  1. Abstract: the phrase 'intoisotope-selective' is missing a space.
  2. The abstract reports a specific 35 meV barrier difference and diffusion-coefficient changes without accompanying uncertainties or convergence diagnostics; the main text should make these explicit (e.g., via block averages or bootstrap estimates) even if they appear in supplementary figures.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive evaluation of our work, the recognition of its significance, and the recommendation for minor revision. No specific major comments were listed in the report.

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The paper's central results—NQEs lowering free-energy barriers by direct comparison of path-integral metadynamics versus classical nuclei simulations, plus the 35 meV H/D isotope difference—are obtained from explicit dynamical simulations on an ML potential trained on external DFT data. No equation or claim reduces the reported diffusion coefficients or barrier differences to quantities defined by the authors' own fitted parameters; the ML training step is a standard preprocessing step whose outputs are then used as input for independent quantum-classical comparisons. No self-citation chain, self-definitional loop, or fitted-input-renamed-as-prediction is present in the derivation. The transferability assumption is a validity condition, not a circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on the transferability of the ML potential trained on r2SCAN+rVV10 data, the adequacy of the path-integral discretization, and the convergence of the metadynamics sampling; these are domain-standard assumptions rather than new axioms invented for the paper.

axioms (2)
  • domain assumption r2SCAN+rVV10 DFT provides a sufficiently accurate reference for training interatomic potentials that describe hydrogen diffusion in MoS2
    Stated as the source of training data in the abstract methods summary.
  • domain assumption Path-integral molecular dynamics with the chosen number of beads and metadynamics bias sufficiently captures nuclear quantum effects at 300 K
    Implicit in the combination of techniques described in the abstract.

pith-pipeline@v0.9.1-grok · 5803 in / 1549 out tokens · 32502 ms · 2026-06-25T19:44:04.742277+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

46 extracted references

  1. [1]

    S. Fan, X. Zou, H. Du, L. Gan, C. Xu, W. Lv, Y.-B. He, Q.-H. Yang, F. Kang, and J. Li, Theoretical investigation of the intercalation chemistry of lithium/sodium ions in transition metal dichalcogenides, The Journal of Physical Chemistry C121, 13599 (2017)

    2017

  2. [2]

    J.-Y. Ji, T. Bao, H. Wang, Y. Xu, D. Zhang, and Q.-K. Xue, Homogeneous lateral lithium intercalation into transition metal dichalcogenides via ion backgating, Nano Letters22, 7336 (2022)

    2022

  3. [3]

    Y. Shao, Y. Zhang, N. Jiang, Y. Hao, K. Qu, Y. Zeng, Z. Liu, X. Lu, Y. Li, Q. Yang, and J. Qiu, Two- dimensionalmaterialsassodium-ionbatteryanodes: The mass transfer and storage mechanisms of “fat” na+, iScience26, 108470 (2023)

    2023

  4. [4]

    Astles, J

    T. Astles, J. G. McHugh, R. Zhang, Q. Guo, M. Howe, Z. Wu, K. Indykiewicz, A. Summerfield, Z. A. H. Good- win, S. Slizovskiy, D. Domaretskiy, A. K. Geim, V. Falko, and I. V. Grigorieva, In-plane staging in lithium-ion in- tercalation of bilayer graphene, Nature Communications 15, 6933 (2024)

    2024

  5. [5]

    Liers, L

    L. Liers, L. Mereacre, H. Li, J. Mickenbecker, M. Knapp, S. Indris, M. Behrens, and S. Mangelsen, Electrochemical intercalation reaction of sodium into the layered transi- tion metal dichalcogenide zrs2 influence of the electrolyte solvent, Inorganic Chemistry64, 18045 (2025)

    2025

  6. [6]

    S. Hu, K. Gopinadhan, A. Rakowski, M. Neek-Amal, T. Heine, I. Grigorieva, S. Haigh, F. Peeters, A. Geim, and M. Lozada-Hidalgo, Transport of hydrogen isotopes through interlayer spacing in van der waals crystals, Na- ture nanotechnology13, 468 (2018)

    2018

  7. [7]

    Y. Cao, V. Fatemi, S. Fang, K. Watanabe, T. Taniguchi, E.Kaxiras,andP.Jarillo-Herrero,Unconventionalsuper- conductivity in magic-angle graphene superlattices, Na- ture556, 43 (2018)

    2018

  8. [8]

    F. M. Arnold, G. Alireza, K. Agnieszka, J. Kunstmann, and T. Heine, Relaxation effects in twisted bilayer molyb- denum disulfide: structure, stability, and electronic prop- erties, IOP 2D Materialsaccepted(2023)

    2023

  9. [9]

    M. S. Ramzan, Z. A. Goodwin, A. A. Mostofi, A. Kuc, and J. Lischner, Effect of coulomb impurities on the elec- tronic structure of magic angle twisted bilayer graphene, npj 2D Materials and Applications7, 49 (2023)

    2023

  10. [10]

    L. Yuan, B. Zheng, J. Kunstmann, T. Brumme, A. B. Kuc, C. Ma, S. Deng, D. Blach, A. Pan, and L. Huang, Twist-angle-dependent interlayer exciton dif- fusion in ws2–wse2 heterobilayers, Nature materials19, 617 (2020)

    2020

  11. [11]

    S. J. Ahn, P. Moon, T.-H. Kim, H.-W. Kim, H.-C. Shin, E. H. Kim, H. W. Cha, S.-J. Kahng, P. Kim, M. Koshino, et al., Dirac electrons in a dodecagonal graphene qua- sicrystal, Science361, 782 (2018)

    2018

  12. [12]

    G. Ni, H. Wang, B.-Y. Jiang, L. Chen, Y. Du, Z. Sun, M. Goldflam, A. Frenzel, X. Xie, M. Fogler,et al., Soliton superlattices in twisted hexagonal boron nitride, Nature communications10, 4360 (2019)

    2019

  13. [13]

    I. Eren, Y. An, and A. B. Kuc, Hydrogen transport between layers of transition metal-dichalcogenides, Ad- vanced Materials Interfaces11, 2300798 (2024)

    2024

  14. [14]

    Y. An, A. Kuc, P. Petkov, M. Lozada-Hidalgo, and T. Heine, On the chemistry and diffusion of hydrogen in the interstitial space of layered crystals h-bn, mos2, and graphite, Small15, 1901722 (2019)

    2019

  15. [15]

    Ceriotti, W

    M. Ceriotti, W. Fang, P. G. Kusalik, R. H. McKenzie, A. Michaelides, M. A. Morales, and T. E. Markland, Nu- clear quantum effects in water and aqueous systems: Ex- periment, theory, and current challenges, Chemical re- views116, 7529 (2016)

    2016

  16. [16]

    Waluk, Nuclear quantum effects in proton or hydrogen transfer, The Journal of Physical Chemistry Letters15, 598 (2024)

    J. Waluk, Nuclear quantum effects in proton or hydrogen transfer, The Journal of Physical Chemistry Letters15, 598 (2024)

    2024

  17. [17]

    Nishikawa, H

    K. Nishikawa, H. Tanaka, K. Kuwahata, M. Tachikawa, and T. Udagawa, Nuclear quantum effects on intramolec- ular hydrogen bonds and backbone structures in bi- uret analogues, PhysicalChemistry ChemicalPhysics27, 20474 (2025)

    2025

  18. [18]

    Marx and M

    D. Marx and M. Parrinello, Ab initio path integral molec- ular dynamics: Basic ideas, The Journal of chemical physics104, 4077 (1996)

    1996

  19. [19]

    R. O. Jones, Density functional theory: Its origins, rise to prominence, and future, Reviews of modern physics 87, 897 (2015)

    2015

  20. [20]

    Peverati and D

    R. Peverati and D. G. Truhlar, Exchange–correlation functional with good accuracy for both structural and energetic properties while depending only on the density and its gradient, Journal of chemical theory and compu- tation8, 2310 (2012)

    2012

  21. [21]

    R. F. Jucá, J. G. da Silva Filho, L. S. Oliveira, A. J. Ramiro de Castro, M. A. S. Silva, A. S. B. Sombra, P. B. Almeida Fechine, J. M. Soares, A. C. Honorato Barreto, P. T. C. Freire, and G. D. Saraiva, Experimental and theoretical investigations on the structural, electronic, and vibrational properties ofβ-Bi2Mo2O9 dibismuth di- molybdenum, The Journal ...

    2025

  22. [22]

    J. W. Furness, A. D. Kaplan, J. Ning, J. P. Perdew, and J. Sun, Accurate and numerically efficient r2scan meta- generalized gradient approximation, The journal of phys- ical chemistry letters11, 8208 (2020)

    2020

  23. [23]

    Zhang, A

    Y. Zhang, A. Ramasamy, K. Pokharel, M. Kothakonda, B. Xiao, J. W. Furness, J. Ning, R. Zhang, and J. Sun, Advances and challenges of scan and r2scan density func- tionals in transition-metal compounds, Wiley Interdisci- plinary Reviews: Computational Molecular Science15, e70007 (2025)

    2025

  24. [24]

    J. Ning, M. Kothakonda, J. W. Furness, A. D. Kaplan, S. Ehlert, J. G. Brandenburg, J. P. Perdew, and J. Sun, 10 Workhorse minimally empirical dispersion-corrected den- sity functional with tests for weakly bound systems: r 2 scan+ rvv 10, Physical Review B106, 075422 (2022)

    2022

  25. [25]

    Mortazavi, X

    B. Mortazavi, X. Zhuang, T. Rabczuk, and A. V. Shapeev, Atomistic modeling of the mechanical proper- ties: the rise of machine learning interatomic potentials, Materials horizons10, 1956 (2023)

    1956

  26. [26]

    Behler, Four generations of high-dimensional neural networkpotentials,ChemicalReviews121,10037(2021)

    J. Behler, Four generations of high-dimensional neural networkpotentials,ChemicalReviews121,10037(2021)

    2021

  27. [27]

    Kulichenko, B

    M. Kulichenko, B. Nebgen, N. Lubbers, J. S. Smith, K. Barros, A. E. Allen, A. Habib, E. Shinkle, N. Fedik, Y. W. Li,et al., Data generation for machine learning in- teratomic potentials and beyond, Chemical Reviews124, 13681 (2024)

    2024

  28. [28]

    Bhatia, P

    N. Bhatia, P. Rinke, and O. Krejčí, Leveraging active learning-enhanced machine-learned interatomic potential for efficient infrared spectra prediction, npj Computa- tional Materials11, 324 (2025)

    2025

  29. [29]

    Y. Liu, J. D. Morrow, C. Ertural, N. L. Fragapane, J. L. Gardner, A. A. Naik, Y. Zhou, J. George, and V. L. Deringer, An automated framework for exploring and learning potential-energy surfaces, Nature Communica- tions16, 7666 (2025)

    2025

  30. [30]

    Chmiela, H

    S. Chmiela, H. E. Sauceda, K.-R. Müller, and A. Tkatchenko, Towards exact molecular dynamics sim- ulations with machine-learned force fields, Nature com- munications9, 3887 (2018)

    2018

  31. [31]

    Hsing, D.-L

    C.-R. Hsing, D.-L. Nguyen, and C.-M. Wei, Exploring diffusion behavior of superionic materials using machine- learning interatomic potentials, Physical Review Materi- als8, 043806 (2024)

    2024

  32. [32]

    Batatia, D

    I. Batatia, D. P. Kovacs, G. Simm, C. Ortner, and G. Csányi, Mace: Higher order equivariant message pass- ing neural networks for fast and accurate force fields, Advances in neural information processing systems35, 11423 (2022)

    2022

  33. [33]

    D.MarxandM.Parrinello,Abinitiopath-integralmolec- ular dynamics, Zeitschrift für Physik B Condensed Mat- ter95, 143 (1994)

    1994

  34. [34]

    Barducci, G

    A. Barducci, G. Bussi, and M. Parrinello, Well-tempered metadynamics: a smoothly converging and tunable free- energy method, Physical review letters100, 020603 (2008)

    2008

  35. [35]

    Ceriotti, J

    M. Ceriotti, J. More, and D. E. Manolopoulos, i-pi: A python interface for ab initio path integral molecular dy- namics simulations, Computer Physics Communications 185, 1019 (2014)

    2014

  36. [36]

    Promoting transparency and reproducibility in enhanced molecular simulations, Nature methods16, 670 (2019)

    2019

  37. [37]

    G. A. Tribello, M. Bonomi, D. Branduardi, C. Camilloni, and G. Bussi, Plumed 2: New feathers for an old bird, Computer physics communications185, 604 (2014)

    2014

  38. [38]

    Beyer and H

    W. Beyer and H. Wagner, Determination of the hydrogen diffusion coefficient in hydrogenated amorphous silicon from hydrogen effusion experiments, Journal of Applied Physics53, 8745 (1982)

    1982

  39. [39]

    The nature of vibrational spectroscopy, https://www2.chemistry.msu.edu/faculty/reusch/virt- txtjml/spectrpy/infrared/irspec1.htm, accessed: 2022- 10-23

    2022

  40. [40]

    T. D. Kühne, M. Iannuzzi, M. Del Ben, V. V. Ry- bkin, P. Seewald, F. Stein, T. Laino, R. Z. Khaliullin, O. Schütt, F. Schiffmann,et al., Cp2k: An electronic structure and molecular dynamics software package- quickstep: Efficient and accurate electronic structure cal- culations, The Journal of Chemical Physics152(2020)

    2020

  41. [41]

    swiss army knife

    S. Grimme, A. Hansen, S. Ehlert, and J.-M. Mewes, r2scan-3c: A “swiss army knife” composite electronic- structure method, The Journal of Chemical Physics154 (2021)

    2021

  42. [42]

    Klimeš, D

    J. Klimeš, D. R. Bowler, and A. Michaelides, Chemical accuracyforthevanderwaalsdensityfunctional,Journal of Physics: Condensed Matter22, 022201 (2010)

    2010

  43. [43]

    Goedecker, M

    S. Goedecker, M. Teter, and J. Hutter, Separable dual- space gaussian pseudopotentials, Physical Review B54, 1703 (1996)

    1996

  44. [44]

    J.-U. Lee, K. Kim, S. Han, G. H. Ryu, Z. Lee, and H. Cheong, Raman signatures of polytypism in molyb- denum disulfide, ACS nano10, 1948 (2016)

    1948

  45. [45]

    Molina-Sanchez and L

    A. Molina-Sanchez and L. Wirtz, Phonons in single- layer and few-layer mos 2 and ws 2, Physical Review B—Condensed Matter and Materials Physics84, 155413 (2011)

    2011

  46. [46]

    Ataca, M

    C. Ataca, M. Topsakal, E. Akturk, and S. Ciraci, A comparative study of lattice dynamics of three-and two- dimensional mos2, The Journal of Physical Chemistry C 115, 16354 (2011). SUPPOR TING INFORMA TION Methodology Details Training setup In this study, we utilized the MACE (Message-Passing Atomic Cluster Expansion) machine learning model to develop an e...

    2011