Pith. sign in

REVIEW 4 major objections 5 minor 81 references

A constant potential reactor framework for electrochemical reaction simulations

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A constant-potential reactor framework couples variable-charge neural network potentials to constant-potential molecular dynamics, enabling nanosecond-scale simulations of electrified gold–water interfaces that reveal how K+ cations…

desk verdict A useful constant-potential ML framework, but the headline cation mechanism leans on charge analysis that is not yet quantitatively supported. read the letter →

arxiv 2411.16330 v1 pith:K3ALDUMI submitted 2024-11-25 physics.chem-ph

Letian Chen , Yun Tian , Xu Hu , Suya Chen , Huijuan Wang , Xu Zhang , Zhen Zhou This is my paper · ORCID
classification physics.chem-ph
keywords constantpotentialmoleculardynamicsvariable-chargeneuralnetworkelectrifiedsolid-liquidinterfaceCO2electroreductionalkalimetalcationeffecthydrogenevolutionreactionelectricaldoublelayersurfacereconstruction
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

This paper claims that a new computational framework, a constant potential reactor, can simulate electrochemical reactions at electrified solid–liquid interfaces with density functional theory accuracy over nanoseconds, long enough to watch interfacial structure evolve in real time. The paper applies it to Au(111) and Au(110) surfaces in water with CO2 and K+ ions, and reports two mechanistic findings: potassium cations promote CO2 activation and suppress the hydrogen evolution reaction, and surface protrusions created by reconstruction act as electron-rich active sites. A sympathetic reader would care because the method addresses a long-standing gap: previous ab initio simulations could only afford picosecond trajectories, usually at fixed charge, while real electrochemical interfaces operate at constant applied potential. If the claims hold, the framework offers a general route to studying electrode–electrolyte phenomena that were previously inaccessible to first-principles simulation.

What carries the argument

The central object is the variable electronic neural network potential (veNNP): a graph neural network with four dual-layer output heads that predict, alongside local energies and forces, the atomic charges and the corrected Fermi level, with the total system electron count fed in as a feature. Coupled to a modified Nosé–Hoover Lagrangian, of the form $\mathcal{L} = \sum_i (\tfrac{1}{2} m_i s^2 \dot{\mathbf{r}}_i^2 - \mathcal{E}) - \tfrac{1}{2} M_s \dot{s}^2 - g k_B T \ln s + \tfrac{1}{2} M_{n_e} \dot{n}_e^2 - \Phi n_e$, the electron number $n_e$ becomes a dynamical variable that flows in and out of the system to hold the applied potential fixed. The potential itself is assigned from the predicted Fermi level via $U_{\mathrm{SHE}} = -E_{\mathrm{Fermi}}/e - 4.44$ V. The veNNP is trained iteratively through an active-learning loop in which a committee of four models flags high-uncertainty configurations for fresh DFT calculations, combining enhanced sampling with targeted first-principles refinement.

What would settle it

Redo the same constant-potential slow-growth simulations using a potential reference fitted to experimental PZCs (for example, U_SHE = 4.2 V) or the cSHE method: if spontaneous CO2 adsorption no longer appears at -0.4 V versus -0.6 V, or if K+ no longer lowers the adsorption barrier, the potential-dependent mechanism is an artifact of the 4.44 V scale. Alternatively, run the K+-containing CO2 adsorption trajectory with a more concentrated electrolyte or a different cation and look for K–O distances below 4 Å; finding direct coordination would overturn the paper's statement that cations act only through the electric field.

Watch

Extended reading notes

Core claim

The central claim is that nanosecond-scale, DFT-accurate molecular dynamics of electrified interfaces under constant electrode potential is achievable by training variable-charge neural network potentials (veNNPs) that simultaneously predict energy, forces, atomic charges, and the corrected Fermi level, and by driving them with a Nosé-type Lagrangian that treats the system's electron count as a dynamical variable coupled to a fictitious potentiostat. Using this reactor on gold–water interfaces, the paper finds that CO2 adsorption onto Au(110) occurs spontaneously only at -0.4 V and -0.6 V versus SHE in the presence of K+, and that cations stabilize the activated, bent CO2 by enhancing electron accumulation on the gold surface; the resulting CO2δ– dipole is stabilized by the cation's electric field. K+–O distances remain above 4 Å, ruling out direct coordination. The paper further claims that cations disrupt the interfacial hydrogen-bond network, impeding H+ transfer and thereby suppressing HER, and that reconstructed protruding Au sites accumulate electrons and activate CO2 at potentials where flat sites do not.

Load-bearing premise

The computed electrode potentials hinge on converting the implicit-solvent corrected Fermi level to the standard hydrogen electrode scale with the fixed 4.44 V reference, and the paper's own PZC comparison suggests the true reference may be about 0.2–0.3 V lower; if the reference shifts or is system-dependent, the potentials at which CO2 adsorption appears could be mislabeled.

Editorial extensions

If this is right

  • If the framework delivers DFT accuracy at nanosecond timescales, constant-potential simulations can replace constant-charge AIMD for reactions where interfacial restructuring, ion adsorption, or proton transfer matter, since the reactor freely exchanges electrons with a fictitious potentiostat.
  • The cation mechanism implies that alkali cations improve CO2 reduction on gold without binding to CO2; the promotion should scale with the cation's ability to stabilize a dipole, not with its coordination strength.
  • The hydrogen-bond-network disruption by cations provides a concrete selectivity origin: CO2 activation is enhanced while HER is suppressed through hindered H+ transport, matching experiments that show cations raise CO2RR selectivity.
  • Surface reconstruction becomes a first-order descriptor: reconstructed protruding sites on Au(110) activate CO2 at -0.6 V even without cations, offering a possible explanation for contradictory experimental reports on whether cations are required.

Reading between the lines

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

  • The framework's charge–structure–potential coupling is not specific to gold or CO2; it should transfer to other electrocatalytic interfaces, and the same active-learning loop could be used to build veNNPs for other metals, alloys, or single-atom catalysts with little modification.
  • A direct testable corollary of the field-mediated mechanism is that different alkali cations (Li+, Na+, Cs+) should order CO2 activation barriers by their surface-enhancing or field-stabilizing propensity, and that ordering could be measured in the same slow-growth setup.
  • Because the paper notes the 4.44 V SHE reference may be off by about 0.2–0.3 V, the exact potential window for CO2 activation (-0.4 V versus -0.6 V) is the least secure quantitative output; re-referencing with a fitted U_SHE or the cSHE method could sharpen or shift the reported thresholds.
  • The slow-growth barriers carry systematic uncertainty from anomalous desorption trajectories and non-minimum pathways; replacing them with a more converged free-energy estimator such as OPES metadynamics would test whether the cation promotion and the protruding-site promotion survive.
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 / 5 minor

Summary. The paper proposes a 'constant potential reactor' framework that combines a variable-electron-count neural network potential (veNNP) with a modified Nosé-Hoover constant-potential molecular dynamics scheme, an active learning protocol, and slow-growth free energy calculations. The method is applied to Au(111)/Au(110)-water interfaces with CO2 and K+ ions, producing nanosecond-scale trajectories and free-energy barriers for the Volmer step and CO2 adsorption. The central conclusions are that cations promote CO2 activation by enhancing electron accumulation on the Au surface and by stabilizing the bent CO2 dipole through the cation electric field, and that surface reconstruction together with cation-induced disruption of the hydrogen-bond network affects CO2RR/HER selectivity. The paper also reports validation accuracies of the veNNP (energy 0.7 meV/atom, forces 14 meV/Å, Fermi level 4 meV, Bader charges 0.05 e) and describes the methodology in detail.

Significance. If the framework performs as claimed, it represents a useful advance in simulating electrified solid-liquid interfaces at near-DFT accuracy over nanosecond timescales, a regime previously inaccessible to constant-potential AIMD. The inclusion of multi-objective learning with variable electron counts, active learning, and constant-potential dynamics is a valuable methodological combination, and the paper gives quantitative validation of the machine-learned model. The proposed cation-promotion mechanism is plausible and consistent with several experimental observations, but some of the supporting evidence remains qualitative. The paper's primary value is as a methodological demonstration; the mechanistic conclusions require additional quantitative support before they can be considered firmly established.

major comments (4)
  1. [Results, 'Nanosecond-scale constant potential MD simulations'; Eq. (1)] The absolute electrode potential is obtained from Eq. (1) with a fixed USHE value of 4.44 V, yet the paper reports PZC values for Au(111) and Au(110) that deviate from experiment and notes that a fitted USHE of 4.2 V (Yu et al., ref. 28) brings them closer. Because the central potential-dependent comparison is between -0.4 V and -0.6 V where CO2 adsorption is observed, a plausible ~0.24 V offset in the reference could shift the adsorption threshold and might blur the distinction between the two conditions. I request that the authors demonstrate that their qualitative conclusions (e.g., CO2 adsorption appears only at -0.4 V and -0.6 V; the Volmer barrier decreases with decreasing potential) are robust to the choice of USHE, for example by recalibrating the potential axis with a 4.2 V reference or by reporting the results as a function of the measured Fermi level with both reference values.
  2. [Methods, 'Molecular dynamics at a constant electrode potential' and 'A versatile machine learning framework' (Eqs.] The extended Nosé-Hoover Lagrangian (Eq. 8) requires that the electrode potential Φ entering the equation of motion for the electron number (Eq. 12) equal the derivative of the potential energy with respect to the electron number for the dynamics to conserve the underlying Hamiltonian. The veNNP predicts the energy (Eq. 18) and the corrected Fermi level (Eq. 27) as separate outputs of independent dual-layer networks, with a loss function (Eq. 28) that does not couple these two quantities. Unless ∂E_NN/∂ne is explicitly trained to match the predicted Fermi level, the fictitious potentiostat can exchange energy with the system in a non-Hamiltonian way, potentially biasing the canonical sampling and the reported free-energy differences. I ask the authors to verify on the test set whether the analytical derivative of the NN energy with respect to electron number reproduces the predicted Fermi level, and to discuss the magnitude of any mismatch and its effect on the MD trajectories.
  3. [Results, 'Atomic charge analysis' and Discussion (Fig. 5a-c)] The mechanistic claim that cations promote CO2 activation by enhancing electron accumulation on the Au surface is supported only by qualitative inspection of Bader charge distributions (Fig. 5a, Supplementary Videos 15-18) and the statement that 'the presence of nearby cations enhances this localized accumulation of electrons at the interface.' No quantitative measure of the cation-induced charge difference is reported. Given the veNNP's Bader charge RMSE of 0.05 |e| (Fig. 2c), the subtle, distributed charge changes expected from a K+ ion located more than 4 Å from the surface may be within the model's prediction error. The authors should provide numerical averages with statistical uncertainties (e.g., region-resolved Au charges with and without cations, trajectory-block averages) and demonstrate that the differences exceed both the model RMSE and the temporal fluctuations. Without such quantitative support, the central mechanistic conclusion is not yet established.
  4. [Results, 'Nanosecond-scale constant potential MD simulations' (Fig. 3, Supplementary Figs. 19-20)] The applied potential is reported to fluctuate by up to ±0.3 V around the target, while the target potentials in the series (0.2, 0, -0.2, -0.4, -0.6 V) are separated by only 0.2 V. This implies that the instantaneous potential distributions for adjacent conditions overlap substantially, which could compromise the potential dependence of the free-energy barriers and adsorption onsets. The authors should report the mean and standard deviation of the actual applied potential for each production and slow-growth simulation, and should ideally tighten the potential control (e.g., by adjusting the fictitious masses M_ne and M_s) so that adjacent target potentials are statistically distinguishable in the sampled distributions.
minor comments (5)
  1. [Throughout; Methods, 'A versatile machine learning framework'] The network name is spelled inconsistently: 'SpookeyNet' appears in the Methods section (e.g., 'we used features of SpookeyNet as inputs') while 'SpookyNet' is used elsewhere; please ensure consistent spelling throughout.
  2. [Methods, 'Hyperparameters and Training'] The loss-function weights are denoted a_E, a_F, a_Fermi, a_q in the text but appear as a_S, a_F, a_S_Fermi, a_q in Eq. (28) and the surrounding text; please align the notation.
  3. [Results, 'Volmer Step and CO2 Adsorption'] The sentence 'CO2 adsorption occurred spontaneously at -0.4 V and -0.6 V2' contains a misplaced citation superscript '2' that appears to be a reference to Monteiro et al.; please correct the citation placement.
  4. [Data and Code Availability] The statement that datasets and source code are available 'upon reasonable request' is restrictive for a methods-oriented paper; I recommend depositing the training data, trajectories, and code in a public repository to facilitate reproducibility and adoption by the community.
  5. [Discussion] The sentence 'It was not possible to study with previous purely explicit models' is vague about which specific finding is meant; please clarify the statement to specify the observation that could not be captured by purely explicit models.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the central predictions are emergent outputs of a DFT-trained surrogate, benchmarked externally; the single self-citation is not load-bearing.

full rationale

The derivation chain is self-contained. The veNNP is a machine-learned surrogate trained on DFT energies, forces, Fermi levels, and Bader charges, and it is validated on held-out DFT data (RMSEs of 0.7 meV/atom, 14 meV/Å, 4 meV, and 0.05 |e| for charges) and against external potential-of-zero-charge measurements. The constant-potential MD uses a standard extended-Lagrangian scheme in which the target potential is an input and the electron number is a dynamical variable; the spontaneous CO2 adsorption at -0.4 and -0.6 V, the potential-dependent free-energy barriers, and the Bader-charge-derived charge-accumulation mechanism are emergent results, not fitted targets. The potential scale is not tuned to force the conclusion: the paper uses the external IUPAC U_SHE = 4.44 V for Eq. (1), reports PZC deviations, and explicitly discusses the alternative fitted U_SHE = 4.2 V from Yu et al. The one in-house citation (ref. 49, Hu et al.) appears only as a consistency remark about hydrogen-bond stabilization of adsorbed CO2 and is not load-bearing for the cation-electric-field mechanism. Concerns about the 0.05 |e| Bader-charge RMSE or the qualitative nature of the charge-accumulation evidence are accuracy and robustness issues, not circularity, because the mechanism is not imposed as a training label or by construction.

Assumptions & free parameters 5 free parameters · 7 assumptions · 0 invented entities

The central claims rest on the accuracy of the DFT reference, the representation of the electrochemical environment via VASPsol, the fixed SHE reference, and the validity of the ML surrogate and thermodynamic integration. The framework itself contributes no free parameters beyond algorithmic choices; the physical parameters such as U_SHE and the DFT functional are inherited from prior literature.

free parameters (5)
  • Loss function weights (a_E, a_F, a_Fermi, a_q) = 1, 50, 10, 1
    Weights in Eq. (28) balancing energy, force, Fermi level, and charge RMSE terms; chosen by hand, not fitted.
  • Active learning uncertainty thresholds = Energy 0.0015, force 0.15, Fermi 0.023, charge 0.1
    Supplementary Table 1; set just below maximum uncertainties of initial dataset, affecting which structures are selected for DFT training.
  • Committee size and top-k selection = 4 committee members, 100 structures per cycle
    Active learning protocol uses a committee of four veNNPs and labels the 100 most uncertain structures each cycle; hand-chosen.
  • Fictitious masses M_s and M_ne = Not reported
    Nosé-Hoover thermostat and electron-number inertia in Eq. (8); values are not provided, which affects potential control and reproducibility.
  • NNP architecture hyperparameters = 3 interaction modules, 168 features, cutoff 5 A, 130 epochs
    SpookyNet-based architecture choices in Methods; affect model accuracy and are chosen without systematic optimization.
assumptions (7)
  • domain assumption rPBE-D3 DFT with PAW accurately describes Au-water interfaces, CO2 activation, and cation effects
    All training data and validation are generated with this DFT setup (Methods); the central mechanistic conclusions inherit its accuracy.
  • domain assumption VASPsol implicit solvent plus explicit water captures the electrochemical double layer and yields reliable Fermi levels for potential calculation
    Used in Eq. (1) and throughout; the paper notes PZC deviations, indicating partial validity.
  • domain assumption The absolute electrode potential of SHE is 4.44 V
    Eq. (1) uses the IUPAC recommended value; the paper acknowledges accepted range 4.2-4.7 V and that fitted values change PZC agreement.
  • standard math The modified Nosé-Hoover Lagrangian from Bonnet et al. samples the grand canonical ensemble at constant electrode potential
    The equations of motion in Eqs. (8)-(14) are taken from ref 44; the paper implements them but does not re-derive or benchmark the ensemble.
  • domain assumption Slow-growth thermodynamic integration yields reliable free energy barriers for the Volmer step and CO2 adsorption
    Despite citing refs 57-58 noting slow-growth may not trace the MEP, the paper uses it for all barriers.
  • domain assumption Bader charge analysis is a valid descriptor of electron transfer in these systems
    The mechanism relies on Bader charges predicted by the NNP; the paper itself notes discrepancies for O and H charges.
  • domain assumption The veNNP generalizes across varying electron counts and chemical environments
    The charge-state encoding is assumed to capture charge-structure-potential coupling, validated only on the training distribution.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A constant potential reactor framework for electrochemical reaction simulations." pith.science (2026). https://pith.science/paper/K3ALDUMI

@misc{pith2026241116330,
  author       = {Pith},
  title        = {Pith review of: A constant potential reactor framework for electrochemical reaction simulations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/K3ALDUMI}},
  note         = {Machine review of arXiv:2411.16330}
}
read the original abstract

Understanding the evolution of electrified solid-liquid interfaces during electrochemical reactions is crucial. However, capturing the dynamic behavior of the interfaces with high temporal resolution and accuracy over long timescales remains a major challenge for both experimental and computational techniques. Here, we present a constant potential reactor framework that enables the simulation of electrochemical reactions with ab initio accuracy over extended timescales, allowing for real-time atomic scale observations for the electrified solid-liquid interface evolution. By implementing an enhanced sampling active learning protocol, we develop fast, accurate, and scalable neural network potentials that generalize across systems with varying electron counts, based on high-throughput density functional theory computations within an explicit-implicit hybrid solvent model. The simulation of reactions in realistic electrochemical environments uncovers the intrinsic mechanisms through which alkali metal cations promote CO2 adsorption and suppress the hydrogen evolution reaction. These findings align with previous experimental results and clarify previously elusive observations, offering valuable computational insights. Our framework lay the groundwork for future studies exploring the dynamic interplay between interfacial structure and reactivity in electrochemical environments.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

81 extracted references · 80 canonical work pages

  1. [1]

    Choi C. et al. Highly active and stable stepped Cu surface for enhanced electrochemical CO 2 reduction to C2H4. Nat. Catal. 3, 804-812 (2020)

    2020

  2. [2]

    Monteiro M. C. O., Dattila F., Hagedoorn B., García- Muelas R., López N. & Koper M. T. M. Absence of CO2 electroreduction on copper, gold and silver electrodes without metal cations in solution. Nat. Catal. 4, 654-662 (2021)

    work page 2021

  3. [3]

    & Bard A

    Jin Z. & Bard A. J. Surface Interrogation of Electrodeposited MnO(x) and CaMnO(3) Perovskites by Scanning Electrochemical Microscopy: Probing Active Sites and Kinetics for the Oxygen Evolution Reaction. Angew Chem Int Ed Engl 60, 794-799 (2021)

    work page 2021

  4. [4]

    of Electrochemi stry C. S. The Top Ten Scientific Questions in Electrochemistry. Journal of Electrochemistry 30, 2024121 (2024)

    work page 2024

  5. [5]

    Hansen J. N. et al. Is There Anything Better than Pt for HER? ACS Energy Lett. 6, 1175-1180 (2021)

    work page 2021

  6. [6]

    Li P. et al. Hydrogen bond network connectivity in the electric double layer dominates the kinetic pH effect in hydrogen electrocatalysis on Pt. Nat. Catal. 5, 900-911 (2022)

    work page 2022

  7. [7]

    & Gasteiger H

    Durst J., Simon C., Hasché F. & Gasteiger H. A. Hydrogen Oxidation and Evolution Reaction Kinetics on Carbon Supported Pt, Ir, Rh, and Pd Electrocatalysts in Acidic Media. J. Electrochem. Soc. 162, F190-F203 (2014)

    work page 2014

  8. [8]

    Pan B., Wang Y . & Li Y . Understanding and leveraging the effect of cations in the electrical double layer for electrochemical CO 2 reduction. Chem Catal. 2, 1267-1276 (2022)

    work page 2022

Show all 81 references
  1. [9]

    Levell Z. et al. Emerging Atomistic Modeling Methods for Heterogeneous Electrocatalysis. Chem. Rev. 124, 8620-8656 (2024)

    2024

  2. [10]

    J., Hsu Y .-S., Agrawal N., Janik M

    Ovalle V . J., Hsu Y .-S., Agrawal N., Janik M. J. & Waegele M. M. Correlating hydration free energy and specific adsorption of alkali metal cations during CO 2 electroreduction on Au. Nat. Catal. 5, 624-632 (2022)

    2022

  3. [11]

    Gu J., Liu S., Ni W., Ren W., Haussener S. & Hu X. Modulating electric field distribution by alkali cations for CO2 electroreduction in strongly acidic medium. Nat. Catal. 5, 268-276 (2022)

    2022

  4. [12]

    & Yang W

    Li Z., Wang L., Sun L. & Yang W. Dynamic Cation Enrichment during Pulsed CO2 Electrolysis and the Cation-Promoted Multicarbon Formation. J. Am. Chem. Soc., (2024)

    2024

  5. [13]

    Shin S. J. et al. A unifying mechanism for cation effect modulating C1 and C2 productions from CO2 electroreduction. Nat. Commun. 13, 5482 (2022)

    2022

  6. [14]

    Wang Y . H. et al. In situ Raman spectroscopy reveals the structure and dissociation of interfacial water. Nature 600, 81-85 (2021)

    2021

  7. [15]

    S., Rathnayake S

    Hsu Y . S., Rathnayake S. T. & Waegele M. M. Cation effects in hydrogen evolution and CO2- to-CO conversion: A critical perspective. J. Chem. Phys. 160, (2024)

    2024

  8. [16]

    Zhang Z.-M. et al. Probing electrolyte effects on cation -enhanced CO2 reduction on copper in acidic media. Nat. Catal. 7, 807-817 (2024)

    2024

  9. [17]

    L i C . Y. et al. In situ probing electrified interfacial water structures at atomically flat surfaces. Nat. Mater. 18, 697-701 (2019)

    2019

  10. [18]

    & Gogotsi Y

    Zhang D., Wang R., Wang X. & Gogotsi Y . In situ monitoring redox processes in energy storage using UV–Vis spectroscopy. Nat. Energy 8, 567-576 (2023)

    2023

  11. [19]

    & De Feyter S

    Cao H., Waghray D., Knoppe S., Dehaen W., Verbiest T. & De Feyter S. Tailoring atomic layer growth at the liquid-metal interface. Nat. Commun. 9, 4889 (2018)

    2018

  12. [20]

    Du X. et al. Machine-learning-accelerated simulations to enable automatic surface reconstruction. Nat Comput. Sci. 3, 1034-1044 (2023)

    2023

  13. [21]

    Zhang Q. et al. Atomic dynamics of electrified solid-liquid interfaces in liquid-cell TEM. Nature 630, 643-647 (2024)

    2024

  14. [22]

    & Norskov J

    Chan K. & Norskov J. K. Electrochemical Barriers Made Simple. J. Phys. Chem. Lett. 6, 2663- 2668 (2015)

    2015

  15. [23]

    & Norskov J

    Chan K. & Norskov J. K. Potential Dependence of Electrochemical Barriers from ab Initio Calculations. J. Phys. Chem. Lett. 7, 1686-1690 (2016)

    2016

  16. [24]

    & Cheng J

    Le J., Iannuzzi M., Cuesta A. & Cheng J. Determining Potentials of Zero Charge of Metal Electrodes versus the Standard Hydrogen Electrode from Density -Functional-Theory-Based Molecular Dynamics. Phys. Rev. Lett. 119, 016801 (2017)

    2017

  17. [25]

    H., Yu S

    Zhao X., Levell Z. H., Yu S. & Liu Y . Atomistic Understanding of Two -dimensional Electrocatalysts from First Principles. Chem. Rev. 122, 10675-10709 (2022)

    2022

  18. [26]

    & Cheng J

    Le J.-B., Fan Q.-Y ., Li J.-Q. & Cheng J. Molecular origin of negative component of Helmholtz capacitance at electrified Pt (111)/water interface. Sci. Adv. 6, eabb1219 (2020)

    2020

  19. [27]

    Zhao X. & Liu Y . Origin of Selective Production of Hydrogen Peroxide by Electrochemical Oxygen Reduction. J. Am. Chem. Soc., (2021)

    2021

  20. [28]

    Yu S., Levell Z., Jiang Z., Zhao X. & Liu Y . What Is the Rate -Limiting Step of Oxygen Reduction Reaction on Fe-N-C Catalysts? J. Am. Chem. Soc. 145, 25352-25356 (2023)

    2023

  21. [29]

    & Parrinello M

    Behler J. & Parrinello M. Generalized neural -network representation of high -dimensional potential-energy surfaces. Phys. Rev. Lett. 98, 146401 (2007)

    2007

  22. [30]

    Four Generations of High -Dimensional Neural Network Potentials

    Behler J. Four Generations of High -Dimensional Neural Network Potentials. Chem. Rev. 121, 10037-10072 (2021)

    2021

  23. [31]

    & López N

    Lian Z., Dattila F. & López N. Stability and lifetime of diffusion -trapped oxygen in oxide - derived copper CO 2 reduction electrocatalysts. Nat. Catal. 7, 401-411 (2024)

    2024

  24. [32]

    & Reed E

    Freitas R. & Reed E. J. Uncovering the effects of interface-induced ordering of liquid on crystal growth using machine learning. Nat. Commun. 11, 3260 (2020)

    2020

  25. [33]

    Zhu J. et al. Superdiffusive Rotation of Interfacial Water on Noble Metal Surface. J. Am. Chem. Soc. 146, 16281-16294 (2024)

    2024

  26. [34]

    & Hansen H

    Yang X., Bhowmik A., Vegge T. & Hansen H. A. Neural network potentials for accelerated metadynamics of oxygen reduction kinetics at Au -water interfaces. Chem. Sci. 14, 3913-3922 (2023)

    2023

  27. [35]

    Zeng Z. et al. Mechanistic insight on water dissociation on pristine low -index TiO2 surfaces from machine learning molecular dynamics simulations. Nat. Commun. 14, 6131 (2023)

    2023

  28. [36]

    S., Liu Z

    Rice P. S., Liu Z. P. & Hu P. Hydrogen Coupling on Platinum Using Artificial Neural Network Potentials and DFT. J. Phys. Chem. Lett. 12, 10637-10645 (2021)

    2021

  29. [37]

    -Y ., Luo L.-H., Shang C

    Hu Z. -Y ., Luo L.-H., Shang C. & Liu Z. -P. Free Energy Pathway Exploration of Catalytic Formic Acid Decomposition on Pt -Group Metals in Aqueous Surroundings. ACS Catal. 14, 7684-7695 (2024)

    2024

  30. [38]

    Amirbeigiarab R. et al. Atomic-scale surface restructuring of copper electrodes under CO 2 electroreduction conditions. Nat. Catal. 6, 837-846 (2023)

    2023

  31. [39]

    & Parrinello M

    Iannuzzi M., Laio A. & Parrinello M. Efficient exploration of reactive potential energy surfaces using Car-Parrinello molecular dynamics. Phys. Rev. Lett. 90, 238302 (2003)

    2003

  32. [40]

    A., Ciccotti G., Hynes J

    Carter E. A., Ciccotti G., Hynes J. T. & Kapral R. Constrained reaction coordinate dynamics for the simulation of rare events. Chem. Phys. Lett. 156, 472-477 (1989)

    1989

  33. [41]

    K., Margl P

    Woo T. K., Margl P. M., Blöchl P. E. & Ziegler T. A combined Car −Parrinello QM/MM implementation for ab initio molecular dynamics simulations of extended systems: application to transition metal catalysis. J. Phys. Chem. B 101, 7877-7880 (1997)

    1997

  34. [42]

    H., Nie W., Boehme A

    Hicks M. H., Nie W., Boehme A. E., Atwater H. A., Agapie T. & Peters J. C . Electrochemical CO 2 Reduction in Acidic Electrolytes: Spectroscopic Evidence for Local pH Gradients. J. Am. Chem. Soc. 146, 25282-25289 (2024)

    2024

  35. [43]

    Fan M. et al. Cationic-group-functionalized electrocatalysts enable stable acidic CO2 electrolysis. Nat. Catal. 6, 763-772 (2023)

    2023

  36. [44]

    & Otani M

    Bonnet N., Morishita T., Sugino O. & Otani M. First -principles molecular dynamics at a constant electrode potential. Phys. Rev. Lett. 109, 266101 (2012)

    2012

  37. [45]

    & Trinkle D

    Yu M. & Trinkle D. R. Accurate and efficient algorithm for Bader charge integration. J. Chem. Phys. 134, 064111 (2011)

    2011

  38. [46]

    The absolute electrode potential: an explanatory note (Recommendations 1986)

    Trasatti S. The absolute electrode potential: an explanatory note (Recommendations 1986). Pure Appl. Chem. 58, 955-966 (1986)

    1986

  39. [47]

    & Schneider J

    Kolb D. & Schneider J. Surface reconstruction in electrochemistry: Au (100-(5× 20), Au (111)- (1× 23) and Au (110)-(1× 2). Electrochim. Acta 31, 929-936 (1986)

    1986

  40. [48]

    & Cheng J

    Li L., Liu Y .-P., Le J.-B. & Cheng J. Unraveling molecular structures and ion effects of electric double layers at metal water interfaces. Cell Reports Physical Science 3, (2022)

    2022

  41. [50]

    & Hansen H

    Qin X., Vegge T. & Hansen H. A. Cation-Coordinated Inner-Sphere CO2 Electroreduction at Au-Water Interfaces. J. Am. Chem. Soc. 145, 1897-1905 (2023)

    2023

  42. [51]

    & Hansen H

    Qin X., Vegge T. & Hansen H. A. CO 2 activation at Au(110) -water interfaces: An ab initio molecular dynamics study. J. Chem. Phys. 155, 134703 (2021)

    2021

  43. [52]

    A., Honkala K

    Qin X., Hansen H. A., Honkala K. & Melander M. M. Cation-induced changes in the inner- and outer-sphere mechanisms of electrocatalytic CO2 reduction. Nat. Commun. 14, 7607 (2023)

    2023

  44. [53]

    & Alexandrova A

    Zhang Z., Zandkarimi B. & Alexandrova A. N. Ensem bles of Metastable States Govern Heterogeneous Catalysis on Dynamic Interfaces. Acc. Chem. Res. 53, 447-458 (2020)

    2020

  45. [54]

    Wang Y . Q. et al. Alkali Metal Cations Induce Structural Evolution on Au(111) During Cathodic Polarization. J. Am. Chem. Soc. 146, 27713-27724 (2024)

    2024

  46. [55]

    J., Xie Y ., Johansson A., Sun L

    Owen C. J., Xie Y ., Johansson A., Sun L. & Kozinsky B. Low -index mesoscopic surface reconstructions of Au surfaces using Bayesian force fields. Nat. Commun. 15, 3790 (2024)

    2024

  47. [56]

    Bonati L. et al. The role of dynamics in heterogeneous catalysis: Surface diffusivity and N 2 decomposition on Fe(111). Proc. Natl. Acad. Sci. 120, e2313023120 (2023)

    2023

  48. [57]

    M., Fukuhara S., Shibuta Y

    Bal K. M., Fukuhara S., Shibuta Y . & Neyts E. C. Free energy barriers from biased molecular dynamics simulations. J. Chem. Phys. 153, 114118 (2020)

    2020

  49. [58]

    & Ziegler T

    Fleurat-Lessard P. & Ziegler T. Tracing the minimum-energy path on the free-energy surface. J. Chem. Phys. 123, 084101 (2005)

    2005

  50. [59]

    Bender J. T. et al. Understanding Cation Effects on the Hydrogen Evolutio n Reaction. ACS Energy Lett. 8, 657-665 (2022)

    2022

  51. [60]

    Ringe S. et al. Double layer charging driven carbon dioxide adsorption limits the rate of electrochemical carbon dioxide reduction on Gold. Nat. Commun. 11, 33 (2020)

    2020

  52. [61]

    Zhang Z. et al. Molecular understanding of the critical role of alkali metal cations in initiating CO2 electroreduction on Cu(100) surface. Nat. Commun. 15, 612 (2024)

    2024

  53. [62]

    Zhong W. et al. Electronic Spin Moment As a Catalytic Descriptor for Fe Single-Atom Catalysts Supported on C2N. J. Am. Chem. Soc. 143, 4405-4413 (2021)

    2021

  54. [63]

    & Parrinello M

    Invernizzi M. & Parrinello M. Rethinking Metadynamics: From Bias Potentials to Probability Distributions. J. Phys. Chem. Lett. 11, 2731-2736 (2020)

    2020

  55. [64]

    Blöchl P. E. Projector augmented-wave method. Phys. Rev. B 50, 17953 (1994)

    1994

  56. [65]

    & Furthmüller J

    Kresse G. & Furthmüller J. Efficiency of ab -initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comp. Mater. Sci. 6, 15-50 (1996)

    1996

  57. [66]

    & Hafner J

    Kresse G. & Hafner J. Ab initio molecular dynamics for liquid metals. Phys. Rev. B 47, 558-561 (1993)

    1993

  58. [67]

    Hammer B., Hansen L. B. & Nørskov J. K. Improved adsorption energetics within density- functional theory using revised Perdew -Burke-Ernzerhof functionals. Phys. Rev. B 59, 7413 (1999)

    1999

  59. [68]

    & Goerigk L

    Grimme S., Ehrlich S. & Goerigk L. Effect of the damping function in dispersion corrected density functional theory. J. Comput. Chem. 32, 1456-1465 (2011)

    2011

  60. [69]

    Mathew K., Kolluru V . S. C., Mula S., Steinmann S. N. & Hennig R. G. Implicit self-consistent electrolyte model in plane-wave density-functional theory. J. Chem. Phys. 151, 234101 (2019)

    2019

  61. [70]

    Mathew K., Sundararaman R., Letchworth -Weaver K., Arias T. A. & Hennig R. G. Implicit solvation model for density-functional study of nanocrystal surfaces and reaction pathways. J. Chem. Phys. 140, 084106 (2014)

    2014

  62. [71]

    & Berendsen H

    Ryckaert J.-P., Ciccotti G. & Berendsen H. J. Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. J. Comput. Phys. 23, 327-341 (1977)

    1977

  63. [72]

    Hjorth Larsen A. et al. The atomic simulation environment -a Python library for working with atoms. J. Phys. Condens. Matter 29, 273002 (2017)

    2017

  64. [73]

    W., Finkler J

    Ko T. W., Finkler J. A., Goedecker S. & Behler J. A fourth-generation high-dimensional neural network potential with accurate electrostatics including non-local charge transfer. Nat. Commun. 12, 398 (2021)

    2021

  65. [74]

    T., Chmiela S., Gastegger M., Schutt K

    Unke O. T., Chmiela S., Gastegger M., Schutt K. T., Sauceda H. E. & Muller K. R. SpookyNet: Learning force fields with electronic degrees of freedom and nonlocal effects. Nat. Commun. 12, 7273 (2021)

    2021

  66. [75]

    & Bengio Y

    Glorot X. & Bengio Y . Understanding the difficulty of training deep feedforward neural networks. In: Proceedings of the thirteenth international conference on artificial intelligence and statistics). JMLR Workshop and Conference Proceedings (2010)

    2010

  67. [76]

    J., Kale S

    Reddi S. J., Kale S. & Kumar S. On the convergence of Adam and beyond. International Conference on Learning Representations (2018)

    2018

  68. [77]

    Harris C. R. et al. Array programming with NumPy. Nature 585, 357-362 (2020)

    2020

  69. [78]

    A constant-potential reactor framework for electrochemical reaction simulations

    Paszke A. et al. Pytorch: An imperative style, high- performance deep learning library. Adv. Neural Inf. Process. Syst. 32, 8024–8035 (2019). Supplementary information of “A constant-potential reactor framework for electrochemical reaction simulations” Letian Chen1, Yun Tian2,...

    2019

  70. [79]

    & Yang B

    Yang K., Liu J. & Yang B. Mechanism and Active Species in NH 3 Dehydrogenation under an Electrochemical Environment: An Ab Initio Molecular Dynamics Study. ACS Catal. 11, 4310- 4318 (2021)

    2021

  71. [80]

    Hu X. et al. Understanding the role of axial O in CO 2 electroreduction on NiN 4 single-atom catalysts via simulations in realistic electrochemical environment. J. Mater. Chem. A 9, 23515- 23521 (2021)

    2021

  72. [81]

    Zhao X. & Liu Y . Unveiling the Active Structure of Single Nickel Atom Catalysis: Critical Roles of Charge Capacity and Hydrogen Bonding. J. Am. Chem. Soc. 142, 5773-5777 (2020)

    2020

  73. [82]

    & Hansen H

    Qin X., Vegge T. & Hansen H. A. Cation- Coordinated Inner-Sphere CO2 Electroreduction at Au-Water Interfaces. J. Am. Chem. Soc. 145, 1897-1905 (2023)

    2023

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.