Pith. sign in

REVIEW 2 major objections 4 minor 1 cited by

Machine Learning the Energetics of Electrified Solid/Liquid Interfaces

T0 review · 2 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read RAZOR machine-learns the work function and Born effective charges of an electrified interface, extending machine-learned interatomic potentials to finite electrode bias and explaining the pH-dependent adsorption site of OH on Cu(100).

desk verdict RAZOR is a smart, clean way to add finite-bias response to MLIPs, but the headline site-switch result rests on a quadratic truncation that the paper's own AITD comparison cannot independently validate. read the letter →

arxiv 2505.19745 v1 pith:UK2ZAGLN submitted 2025-05-26 cond-mat.mtrl-sci

Nicolas Bergmann , Nicéphore Bonnet , Nicola Marzari , Karsten Reuter , Nicolas G. Hörmann This is my paper
classification cond-mat.mtrl-sci
keywords machine-learnedinteratomicpotentialsworkfunctionBorneffectivechargeselectrifiedinterfacesOHadsorptiononCu(100)constant-potentialmoleculardynamicsimplicitsolvation
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

The paper introduces RAZOR, a response-augmented machine-learning scheme that makes fast interatomic-potential models aware of the excess charge carried by an electrified electrode. Instead of training on many charged configurations, it learns the work function, the first-order change in energy when a bias charge $q$ is added, and uses Born effective charges, the cross-derivatives of forces with respect to $q$, to stabilize that learning. With a constant electronic capacitance for the quadratic term, the resulting surrogate is correct to second order in $q$ at a modest increase in training cost. Applied to OH adsorbed on Cu(100) in implicit water, RAZOR-based molecular dynamics finds that the preferred adsorption site switches from bridge at negative charge to hollow at positive charge, which the paper connects to the experimentally observed pH-dependence of the adsorption site at fixed electrode potential.

What carries the argument

The load-bearing object is the second-order Taylor expansion of the interface energy in excess bias charge $q$, combined with an equivalence between $q$-response and uniaxial electric-field response. For a planar metal electrode the relevant response is the scalar work function $\phi_0^\alpha=\partial E^\alpha/\partial q|_{q=0}$, and its atomic derivatives are Born effective charges $Z^*_i$, which enter force labels as $\partial F_i/\partial q$ and stabilize learning of $\phi_0^\alpha$ exactly as force labels stabilize ordinary machine-learned potentials. The Helmholtz relation $P_z=\epsilon_0 A\,\Delta\phi_0$ lets the work function be treated as an extensive sum of atomic contributions, and the electronic capacitance $C_{\mathrm{el},0}$ supplies the second-order term; because it is taken to depend only on composition, not on configuration, it drops out of fixed-composition force rankings and does not need to be learned per atom. Thermodynamic integration over $\langle\phi\rangle_q$ and a Legendre transform then convert constant-charge simulations into constant-potential free energies.

What would settle it

Run explicit constant-charge density-functional theory (not relying on the quadratic expansion) for OH on Cu(100) at several charges spanning the claimed transition, for example $q=0$, $+0.02$, $+0.05$, and $+0.1\,e$ per Cu surface atom, and compare the relaxed adsorption sites and relative free energies with the RAZOR-MLIP predictions at the same $q$. If the explicit calculations do not show the bridge-to-hollow switch, or place it at a different charge, the second-order response assumption is falsified.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central claim is that the energetics of a charged metal/electrolyte interface can be captured by learning response coefficients at the charge-free state rather than by sampling many biased states. RAZOR writes the energy as $E^\alpha(q)=E^\alpha_0+\phi^\alpha_0\,q+\tfrac12\,C^{-1}_{\mathrm{el},0}\,q^2$, where $\phi^\alpha_0$ is the work function (interfacial potential drop) and $C_{\mathrm{el},0}$ the electronic capacitance; it then machine-learns $\phi^\alpha_0$ and the atomic Born effective charges $Z^*_i=-\partial\phi^\alpha_0/\partial r_i=\partial F_i/\partial q$ from local descriptors, while treating $C_{\mathrm{el},0}$ as a composition-dependent constant. In the showcase system, 0.5 monolayer OH on Cu(100), RAZOR-MLIP molecular dynamics at 300 K shows a sharp bridge-to-hollow site switch within a narrow charge window near $q\approx0.02$--$0.05\,e$ per Cu surface atom, accompanied by a $\sim1.5$ V work-function change; after Legendre transformation to constant potential and reference-electrode alignment, the predicted pH-dependence of the preferred site matches the two experimental studies cited by the paper.

Load-bearing premise

The load-bearing premise is that the energy of the charged interface is exactly quadratic in the added electrode charge $q$, with a fixed curvature, over the whole charge range studied; if higher-order terms or a curvature that depends on atomic arrangement matter, the predicted adsorption-site switch could be an artifact.

Editorial extensions

If this is right

  • RAZOR can be added to any existing machine-learned interatomic potential, giving finite-bias energetics with only a modest premium in training data (roughly three charged-state DFT calculations per structure, reduced in practice by reusing the neutral-density SCF initialization).
  • Constant-charge molecular dynamics at applied bias becomes feasible for time scales of hundreds of picoseconds, long enough to observe adsorbate site interconversion that direct ab initio molecular dynamics cannot afford.
  • The learned decomposition of the work function into atomic contributions provides a route to local interfacial potential drops without relying on arbitrary electronic charge-density partitioning schemes.
  • For OH on Cu(100), the charge-induced site switch translates into a non-Nernstian pH-dependence of the adsorption site, matching the experimentally inferred preference for bridge sites in alkaline conditions and hollow sites in acidic conditions.

Reading between the lines

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

  • Editorial: the most transferable output may be the local work-function decomposition itself; if it proves accurate on other surfaces, it could give a direct mapping from MLIP configurations to local electrochemical driving forces, a use the paper only names as future work.
  • Editorial: the configuration-independent treatment of $C_{\mathrm{el},0}$ will likely be the first assumption challenged at low coverage or with mobile adsorbed water, where the double-layer response depends on local structure; making the capacitance a learned descriptor-dependent quantity is a natural stress test.
  • Editorial: the same response-learning logic could extend to semiconductor or two-dimensional-material electrodes, where Born effective charges remain well defined but the capacitance physics differs from that of a metallic double layer.
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

2 major / 4 minor

Summary. The paper introduces RAZOR, a framework that augments standard machine-learning interatomic potentials with the response of an electrode's total energy to excess charge q. It learns the work function phi0 as the first-order derivative of energy with respect to q, stabilizes this via Born effective charges, and treats the second-order response through a fixed, configuration-independent capacitance derived from an implicit solvent model. The resulting surrogate is used for constant-charge MD of 0.5 ML OH on Cu(100), yielding a charge-induced site switch from bridge at negative q to hollow at positive q. After a Legendre transform and CHE referencing, the authors obtain a pH-dependent preferred adsorption site that they argue matches experimental reports.

Significance. If the second-order truncation is validated, RAZOR is an elegant and practical extension of MLIPs to biased electrochemical interfaces. The learning targets are clearly defined from DFT, the additional data cost is modest (less than a doubling, by the authors' estimate), and the approach is architecture-agnostic. The MD result is a concrete, falsifiable prediction (site switch in a narrow charge window) that matches experiments, and the agreement with AITD provides a useful internal consistency check between the learned MLIP and static DFT inputs. The main reservation is that the truncation and the configuration-independence of the capacitance are not tested against explicit finite-q DFT data, so the quantitative site-switch prediction rests on an assumption that is load-bearing for the headline application.

major comments (2)
  1. [Eq. (1), Fig. 2 bottom] The central validation of the second-order truncation is not independent. The comparison to AITD in Fig. 2 (bottom) uses the same quadratic expansion of Eq. (1) and the same configuration-independent C_el,0; agreement between RAZOR-MLIP and AITD therefore establishes internal consistency between the learned MLIP and the static DFT inputs, but it does not test whether cubic terms or site-dependent second-order terms are negligible over the charge range where the site switch occurs (0.02 to 0.05 e per Cu atom). The manuscript reports no direct finite-q DFT energies for bridge and hollow reference structures. Since the additional DFT calculations at ±q used to construct Z* are already available, the authors should use them (or new calculations at representative q values) to compare the predicted E(q) of Eq. (1) with explicit DFT energies, and promote this comparison to the main text.
  2. [Paragraph beginning 'This leaves the electronic capacitance...'] By construction, RAZOR's relative stability of bridge and hollow at fixed q is governed solely by E0 and phi0, because C_el,0 is taken to be configuration-independent and the quadratic term therefore cancels in energy differences between configurations of the same composition. This assumption is plausible for a metal electrode but is not tested. If the true second-order response (or the DL capacitance) differs between adsorption sites, the predicted switch could be an artifact. The authors should quantify the site dependence of C_el,0 (e.g. from finite-q DFT calculations for the two ordered c(2x2) structures) or otherwise bound the error this assumption introduces in the switch charge range.
minor comments (4)
  1. [Fig. 2 caption] The definition of the OH-bond angle phi and the fractional coverage shown in the top panel should be stated explicitly in the caption; the green line appears to be a smoothed or running quantity.
  2. [After Eq. (8)] The exchange P_z <-> eps0 A Delta phi0 would be clearer if it stated explicitly that P_z is an extensive quantity for the simulation cell and A is the cell surface area, to avoid confusion with intensive potential drops.
  3. [Paragraph preceding Eq. (11)] It would be helpful to state explicitly that Eq. (11) is exact under the RAZOR model's assumptions rather than an additional approximation, since this is the key step that again relies on the quadratic truncation.
  4. [Final paragraph before 'In summary'] The term 'non-Nernstian' is used without definition; a sentence explaining that the site preference changes with phi_E at fixed pH, or that the adsorption onset shifts nonlinearly with pH, would clarify the claim.

Circularity Check

1 steps flagged · score 3.0 of 10

AITD validation is self-referential; central RAZOR derivation is otherwise self-contained.

  1. self definitional [Main text, paragraph discussing Fig. 2 (bottom panel), p. 4; see also Fig. 2 caption.]
    "The comparison to ab initio thermodynamics (AITD) [9, 61, 70] – the prevalent approximate approach to such energetics at applied charge – shows excellent agreement (Fig. 2 bottom), e.g. for the site switch as derived from the crossing of the free energy curves of the static DFT reference structures. Notably, this agreement prevails over the entire wide range of bias charges, supporting the validity of RAZOR’s second-order expansion."

    AITD as used here is 'DL-corrected ab initio thermodynamics' (Fig. 2 caption), meaning it adds the same implicit-solvent capacitive correction q^2/(2C_el,0) to the same q=0 DFT energies and work functions that RAZOR also learns. The AITD free-energy curves for the reference c(2x2)-OH structures are therefore built from the same quadratic E(q) = E0 + phi0 q + q^2/(2C_el,0) model of Eq. (1) that RAZOR uses. Agreement between RAZOR and AITD can validate the MLIP's fit of E0 and phi0, but it cannot validate the second-order truncation over the applied charge range, because the validation quantity is defined from that same expansion.

full rationale

The central RAZOR method is not circular: the MLIP is trained on DFT data at q=0 (energies, forces, work functions) and on Born effective charges obtained from +/-q DFT calculations, and the finite-bias site switch emerges from MD sampling without being imposed. The capacitance C_el,0 is taken from an implicit solvent model rather than fitted to the experimental or finite-bias target, and it is configuration-independent by assumption, so the site switch is governed by learned E0 and phi0 differences. The only substantive circularity is the use of AITD agreement to validate the second-order expansion: AITD is here DL-corrected with the same quadratic energy model and the same implicit-solvent capacitance, so the agreement is partly by construction. This inflates confidence in the truncation but does not invalidate the core ML derivation. Score 3 reflects one self-referential validation step while the central claim retains independent content.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the validity of a second-order Taylor expansion of the bias-charge dependence, on a local atomic decomposition of the work function, on a configuration-independent capacitance from the implicit solvent model, and on the CHE mapping to experimental pH/potential scales. No new entities are introduced; the method is a fitting of DFT targets. The MLIP hyperparameters are standard training choices and are not counted as physical free parameters.

assumptions (4)
  • domain assumption The energy E(q) is exactly quadratic in the bias charge q up to second order over the sampled range (Eq. 1).
    The MLIP architecture can only represent E0, phi0, and a fixed C_el,0, so any cubic or higher-order response to q is neglected by construction.
  • domain assumption The total dipole moment and work function change decompose into atomic contributions p_z(d_i) depending only on the local environment (Eq. 9).
    The MLIP descriptors are local; long-range electrostatic effects beyond the cutoff are assumed negligible.
  • domain assumption The electronic capacitance C_el,0 is configuration-independent and dominated by the double-layer capacitance described by the implicit solvent model.
    Used to convert constant-charge results to constant-potential via Eq. (11)-(13).
  • domain assumption The CHE referencing mu_O = mu_O^0 + k_B T ln(10)*pH + e(phi_E - phi_SHE) correctly maps the computed charge-dependent free energies to experimental pH and potential scales.
    Standard electrochemical referencing, but its applicability to the implicit-solvent model is assumed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Machine Learning the Energetics of Electrified Solid/Liquid Interfaces." pith.science (2026). https://pith.science/paper/UK2ZAGLN

@misc{pith2026250519745,
  author       = {Pith},
  title        = {Pith review of: Machine Learning the Energetics of Electrified Solid/Liquid Interfaces},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UK2ZAGLN}},
  note         = {Machine review of arXiv:2505.19745}
}
read the original abstract

We present a response-augmented machine learning (ML) approach to the energetics of electrified metal surfaces. We leverage local descriptors to learn the work function as the first-order energy change to introduced bias charges and stabilize this learning through Born effective charges. This permits the efficient extension of ML interatomic potential architectures to include finite bias effects up to second-order. Application to OH at Cu(100) rationalizes the experimentally observed pH-dependence of the preferred adsorption site in terms of a non-Nernstian charge-induced site switching.

Figures

Figures reproduced from arXiv: 2505.19745 by the authors.

Figure 1
Figure 1. FIG. 1. Schematic workflow showing how combining the ”Response Analysis in [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. RAZOR-MLIP based MD simulations for a [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Adsorption free energy ∆Ω [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. VASP Plugins: Linking the Vienna ab-initio Simulation Package with Python

    cond-mat.mtrl-sci 2026-07 accept novelty 5.5 of 10

    A C++/pybind11 shared-memory plugin layer exposes VASP SCF and ionic data as NumPy arrays so Python can modify structure, forces, local potential, and occupancies in place.

Reference graph

Works this paper leans on

79 extracted references · 74 canonical work pages · cited by 1 Pith paper

  1. [1]

    J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, T. Bligaard, and H. J´ onsson, J. Phys. Chem. B 108, 17886 (2004)

    work page 2004

  2. [2]

    Resasco, F

    J. Resasco, F. Abild-Pedersen, C. Hahn, Z. Bao, M. T. Koper, and T. F. Jaramillo, Nat. Cat. 5, 374 (2022)

    work page 2022

  3. [3]

    Groß and S

    A. Groß and S. Sakong, Chem. Rev 122, 10746 (2022)

    work page 2022

  4. [4]

    Jeanmairet, B

    G. Jeanmairet, B. Rotenberg, and M. Salanne, Chem. Rev. 122, 10860 (2022)

    work page 2022

  5. [5]

    Schwarz and R

    K. Schwarz and R. Sundararaman, Surf. Sci. Rep. 75, 100492 (2020)

    work page 2020

  6. [6]

    I. Dabo, E. Canc` es, Y. L. Li, and N. Marzari, (2010), arXiv:0901.0096 [cond-mat.mtrl-sci]

    work page Pith review arXiv 2010

  7. [7]

    I. Dabo, Y. Li, N. Bonnet, and N. Marzari, Ab initio electrochemical properties of electrode surfaces, in Fuel Cell Science (John Wiley & Sons, Ltd, 2010) Chap. 13, pp. 415–431

    work page 2010

  8. [8]

    Nattino, M

    F. Nattino, M. Truscott, N. Marzari, and O. Andreussi, The Journal of Chemical Physics 150, 041722 (2018)

    work page 2018

Show all 79 references
  1. [9]

    Ringe, N

    S. Ringe, N. G. H¨ ormann, H. Oberhofer, and K. Reuter, Chem. Rev. 122, 10777 (2022)

    2022

  2. [10]

    Behler, Chem

    J. Behler, Chem. Rev. 121, 10037 (2021)

    2021

  3. [11]

    V. L. Deringer, A. P. Bart´ ok, N. Bernstein, D. M. Wilkins, M. Ceriotti, and G. Cs´ anyi, Chem. Rev 121, 10073 (2021)

    2021

  4. [12]

    Jacobs, D

    R. Jacobs, D. Morgan, S. Attarian, J. Meng, C. Shen, Z. Wu, C. Y. Xie, J. H. Yang, N. Artrith, B. Blaiszik, G. Ceder, K. Choudhary, G. Csanyi, E. D. Cubuk, B. Deng, R. Drautz, X. Fu, J. Godwin, V. Honavar, O. Isayev, A. Johansson, B. Kozinsky, S. Martiniani, S. P. Ong, I. Polt...

    2025

  5. [13]

    A. S. Raman and A. Selloni, J. Chem. Phys. 160, 244708 (2024). 6

    2024

  6. [14]

    Stocker, H

    S. Stocker, H. Jung, G. Cs´ anyi, C. F. Goldsmith, K. Reuter, and J. T. Margraf, J. Chem. Theory Com- put. 19, 6796 (2023)

    2023

  7. [15]

    Timmermann, F

    J. Timmermann, F. Kraushofer, N. Resch, P. Li, Y. Wang, Z. Mao, M. Riva, Y. Lee, C. Staacke, M. Schmid, C. Scheurer, G. S. Parkinson, U. Diebold, and K. Reuter, Phys. Rev. Lett. 125, 206101 (2020)

    2020

  8. [16]

    Bruix, J

    A. Bruix, J. T. Margraf, M. Andersen, and K. Reuter, Nat. Cat. 2, 659 (2019)

    2019

  9. [17]

    Omranpour, J

    A. Omranpour, J. Elsner, K. N. Lausch, and J. Behler, ACS Catal. 15, 1616 (2025)

    2025

  10. [18]

    Grisafi and M

    A. Grisafi and M. Ceriotti, J. Chem. Phys. 151, 204105 (2019)

    2019

  11. [19]

    T. W. Ko, J. A. Finkler, S. Goedecker, and J. Behler, Nat. Commun. 12, 398 (2021)

    2021

  12. [20]

    Gao and R

    A. Gao and R. C. Remsing, Nat. Commun. 13, 1572 (2022)

    2022

  13. [21]

    C. G. Staacke, S. Wengert, C. Kunkel, G. Cs´ anyi, K. Reuter, and J. T. Margraf, Mach. Learn.: Sci. Tech- nol. 3, 015032 (2022)

    2022

  14. [22]

    Vondr´ ak, K

    M. Vondr´ ak, K. Reuter, and J. T. Margraf, J. Chem. Phys. 159, 054109 (2023)

    2023

  15. [23]

    Grisafi, A

    A. Grisafi, A. Bussy, M. Salanne, and R. Vuilleumier, Phys. Rev. Materials 7, 125403 (2023)

    2023

  16. [24]

    Hisama, G

    K. Hisama, G. Valadez Huerta, and M. Koyama, J. Com- put. Chem. , jcc.27487 (2024)

    2024

  17. [25]

    Monacelli and N

    L. Monacelli and N. Marzari, (2024), arXiv:2412.01642 [cond-mat.mtrl-sci]

    2024 arXiv

  18. [26]

    Zhang, M

    C. Zhang, M. F. Calegari Andrade, Z. K. Goldsmith, A. S. Raman, Y. Li, P. M. Piaggi, X. Wu, R. Car, and A. Selloni, Nat. Commun. 15, 10270 (2024)

    2024

  19. [27]

    A. S. Christensen, F. A. Faber, and O. A. Von Lilienfeld, J. Chem. Phys. 150, 064105 (2019)

    2019

  20. [28]

    Gastegger, K

    M. Gastegger, K. T. Sch¨ utt, and K.-R. M¨ uller, Chem. Sci. 12, 11473 (2021)

    2021

  21. [29]

    Gigli, M

    L. Gigli, M. Veit, M. Kotiuga, G. Pizzi, N. Marzari, and M. Ceriotti, npj Comput. Mater. 8, 209 (2022)

    2022

  22. [30]

    K. Joll, P. Schienbein, K. M. Rosso, and J. Blumberger, Nat. Commun. 15, 8192 (2024)

    2024

  23. [31]

    Falletta, A

    S. Falletta, A. Cepellotti, A. Johansson, C. W. Tan, M. L. Descoteaux, A. Musaelian, C. J. Owen, and B. Kozinsky, Nat. Commun. 16, 4031 (2025)

    2025

  24. [32]

    Schmiedmayer and G

    B. Schmiedmayer and G. Kresse, J. Chem. Phys. 161, 084703 (2024)

    2024

  25. [33]

    D. M. Patel and G. Kastlunger, Chem. Rev. 125, 3378 (2025)

    2025

  26. [34]

    A. M. Dudzinski, E. Diesen, H. H. Heenen, V. J. Bukas, and K. Reuter, ACS Catal. 13, 12074 (2023)

    2023

  27. [35]

    Bergmann, N

    N. Bergmann, N. G. H¨ ormann, and K. Reuter, J. Chem. Theory Comput. 19, 8815 (2023)

    2023

  28. [36]

    S. D. Beinlich, N. G. H¨ ormann, and K. Reuter, ACS Catal. 12, 6143 (2022)

    2022

  29. [37]

    N. G. H¨ ormann, S. D. Beinlich, and K. Reuter, J. Phys. Chem. C 128, 5524 (2024)

    2024

  30. [38]

    L. Li, K. Reuter, and N. G. H¨ ormann, ACS Electrochem. 1, 186 (2025)

    2025

  31. [39]

    Bonnet, I

    N. Bonnet, I. Dabo, and N. Marzari, Electrochim. Acta 121, 210 (2014)

    2014

  32. [40]

    Bonnet, T

    N. Bonnet, T. Morishita, O. Sugino, and M. Otani, Phys. Rev. Lett. 109, 266101 (2012)

    2012

  33. [41]

    X. Chen, M. El Khatib, P. Lindgren, A. Willard, A. J. Medford, and A. A. Peterson, npj Comput. Mater. 9, 73 (2023)

    2023

  34. [42]

    S. D. Beinlich, G. Kastlunger, K. Reuter, and N. G. H¨ ormann, J. Chem. Theory Comput.19, 8323 (2023)

    2023

  35. [43]

    Kastlunger, S

    G. Kastlunger, S. Vijay, X. Chen, S. Sharma, and A. Pe- terson, ChemPhysChem 25, e202300950 (2024)

    2024

  36. [44]

    Binninger, Phys

    T. Binninger, Phys. Rev. B 103, L161403 (2021)

    2021

  37. [45]

    S. D. Beinlich, G. Kastlunger, K. Reuter, and N. G. H¨ ormann, (2023), arXiv:2307.09817 [physics]

    2023

  38. [46]

    A. P. Bart´ ok, M. C. Payne, R. Kondor, and G. Cs´ anyi, Phys. Rev. Lett. 104, 136403 (2010)

    2010

  39. [47]

    J. S. Smith, O. Isayev, and A. E. Roitberg, Chem. Sci. 8, 3192 (2017)

    2017

  40. [48]

    Batatia, D

    I. Batatia, D. P. Kov´ acs, G. N. C. Simm, C. Ortner, and G. Cs´ anyi, Adv. Neural Inf. Process Syst. (2022)

    2022

  41. [49]

    Batzner, A

    S. Batzner, A. Musaelian, L. Sun, M. Geiger, J. P. Mailoa, M. Kornbluth, N. Molinari, T. E. Smidt, and B. Kozinsky, Nat. Commun. 13, 2453 (2022)

    2022

  42. [50]

    Zhang, H

    L. Zhang, H. Wang, M. C. Muniz, A. Z. Panagiotopoulos, R. Car, and W. E, J. Chem. Phys. 156, 124107 (2022)

    2022

  43. [51]

    Stocco, C

    E. Stocco, C. Carbogno, and M. Rossi, (2025), arXiv:2502.02413 [cond-mat]

    2025 arXiv

  44. [52]

    C.-Y. Wang, S. Sharma, E. K. U. Gross, and J. K. De- whurst, Phys. Rev. B 106, L180303 (2022)

    2022

  45. [53]

    Smoluchowski, Phys

    R. Smoluchowski, Phys. Rev. 60, 661 (1941)

    1941

  46. [54]

    W.-X. Li, C. Stampfl, and M. Scheffler, Phys. Rev. B 65, 075407 (2002)

    2002

  47. [55]

    Ibach, Physics of Surfaces and Interfaces, Vol

    H. Ibach, Physics of Surfaces and Interfaces, Vol. 12 (Springer, 2006)

    2006

  48. [56]

    Oschinski, N

    H. Oschinski, N. G. H¨ ormann, and K. Reuter, J. Chem. Phys. 160, 214706 (2024)

    2024

  49. [57]

    Scalfi, T

    L. Scalfi, T. Dufils, K. G. Reeves, B. Rotenberg, and M. Salanne, J. Chem. Phys. 153, 174704 (2020)

    2020

  50. [58]

    Ntim and M

    S. Ntim and M. Sulpizi, Phys. Chem. Chem. Phys. 25, 22619 (2023)

    2023

  51. [59]

    Andreussi and N

    O. Andreussi and N. Marzari, Phys. Rev. B 90, 245101 (2014)

    2014

  52. [60]

    Kastlunger, P

    G. Kastlunger, P. Lindgren, and A. A. Peterson, J. Phys. Chem. C 122, 12771 (2018)

    2018

  53. [61]

    N. G. H¨ ormann, N. Marzari, and K. Reuter, npj Comput. Mater. 6, 136 (2020)

    2020

  54. [62]

    N. G. H¨ ormann and K. Reuter, J. Chem. Theory Com- put. 17, 1782 (2021)

    2021

  55. [63]

    Kunze, V

    J. Kunze, V. Maurice, L. H. Klein, H.-H. Strehblow, and P. Marcus, J. Electroanal. Chem. 554-555, 113 (2003)

    2003

  56. [64]

    G. H. Simon, C. S. Kley, and B. Roldan Cuenya, Angew. Chem. Int. Ed. 60, 2561 (2021)

    2021

  57. [65]

    M. Sun, A. Staykov, and M. Yamauchi, ACS Catal. 12, 14856 (2022)

    2022

  58. [66]

    Giannozzi, S

    P. Giannozzi, S. Baroni, N. Bonini, M. Calandra, R. Car, C. Cavazzoni, D. Ceresoli, G. L. Chiarotti, M. Cococ- cioni, I. Dabo, and et al., J. Condens. Matter Phys. 21, 395502 (2009)

    2009

  59. [67]

    Giannozzi, O

    P. Giannozzi, O. Andreussi, T. Brumme, O. Bunau, M. Buongiorno Nardelli, M. Calandra, R. Car, C. Cavaz- zoni, D. Ceresoli, M. Cococcioni, and et al., J. Condens. Matter Phys. 29, 465901 (2017)

    2017

  60. [68]

    Andreussi, I

    O. Andreussi, I. Dabo, and N. Marzari, J. Chem. Phys. 136, 064102 (2012)

    2012

  61. [69]

    Tiwari, H

    A. Tiwari, H. H. Heenen, A. S. Bjørnlund, T. Maagaard, E. Cho, I. Chorkendorff, H. H. Kristoffersen, K. Chan, and S. Horch, J. Phys. Chem. Lett. 11, 1450 (2020)

    2020

  62. [70]

    Rogal and K

    J. Rogal and K. Reuter, Experiment, Modeling and Sim- ulation of Gas- Surface Interactions for Reactive Flows in Hypersonic Flights 14, 2.1 (2007)

    2007

  63. [71]

    Cruickshank, D

    B. Cruickshank, D. D. Sneddon, and A. A. Gewirth, Surf. Sci. 281, L308 (1993). 7

    1993

  64. [72]

    B. Deng, P. Zhong, K. Jun, J. Riebesell, K. Han, C. J. Bartel, and G. Ceder, Nat. Mach. Intell. 5, 1031 (2023)

    2023

  65. [73]

    Batatia, P

    I. Batatia, P. Benner, Y. Chiang, A. M. Elena, D. P. Kov´ acs, J. Riebesell, X. R. Advincula, M. Asta, M. Avay- lon, W. J. Baldwin, and et al., (2024), arXiv:2401.00096 [physics.chem-ph]

    2024 arXiv

  66. [74]

    A. D. Kaplan, R. Liu, J. Qi, T. W. Ko, B. Deng, J. Riebe- sell, G. Ceder, K. A. Persson, and S. P. Ong, (2025), arXiv:2503.04070

    2025

  67. [75]

    Kabylda, J

    A. Kabylda, J. T. Frank, S. S. Dou, A. Khabibrakhmanov, L. M. Sandonas, O. T. Unke, S. Chmiela, K.-R. M¨ uller, and A. Tkatchenko, (2025)

    2025

  68. [76]

    Mazitov, F

    A. Mazitov, F. Bigi, M. Kellner, P. Pegolo, D. Tisi, G. Fraux, S. Pozdnyakov, P. Loche, and M. Ceriotti, (2025), arXiv:2503.14118 [cond-mat.mtrl-sci]

    2025 arXiv

  69. [77]

    Vijay, G

    S. Vijay, G. Kastlunger, J. A. Gauthier, A. Patel, and K. Chan, J. Phys. Chem. Letters 13, 5719 (2022)

    2022

  70. [78]

    X. Chen, G. Kastlunger, and A. A. Peterson, Phys. Rev. Lett. 131, 238003 (2023)

    2023

  71. [79]

    Melitz, J

    W. Melitz, J. Shen, A. C. Kummel, and S. Lee, Surf. Sci. Rep. 66, 1 (2011)

    2011

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.