Can DFT-Trained Neural Network Potentials Reproduce Structure, Solvation, and Water-Exchange Properties in Aqueous Magnesium Solutions?
Journal of Chemical Theory and Computation · 10.11.2026 · 22 (21) · 11432-11441
DOI: 10.1021/acs.jctc.6c01270 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Chemical Theory and Computation, online erschienen am 24.10.2026. Verfasst von Sebastian Falkner und 3 weiteren Autor:innen. Beteiligte Einrichtung: Institute of Physics, University of Augsburg, Universitätsstraße 1, Augsburg86159, Germany. Die Arbeit nennt 86 Literaturangaben. Gefördert von: Augsburg University, Austrian Science Fund.
Abstract (Originalsprache)
Magnesium ions play an essential role in many biological processes but remain challenging to model in biomolecular simulations. Despite considerable scientific effort, classical force fields fail to simultaneously reproduce key structural, thermodynamic, and kinetic solution properties, likely due to their inability to explicitly account for quantum many-body effects. Here, we develop and systematically benchmark MACE neural network potentials (NNPs) for aqueous MgCl2 solutions, trained on revPBE-D3/zd and revPBE0-D3/zd density functional theory reference data, and assess their ability to reproduce a broad range of experimental solution properties including the structure of the first hydration shell, diffusion coefficient, activity derivative, water-exchange rate, and mechanism as well as solvation free energy. Both NNPs accurately reproduce the octahedral structure of the first hydration shell, ion-pairing properties, and diffusion coefficients. Combining the NNPs with transition path sampling and other enhanced sampling techniques allows us to capture the rare event of water exchange in the first hydration shell of Mg2+, revealing a dissociative exchange mechanism. Transition interface sampling yields exchange rates within 1 order of magnitude of experiment, representing a substantial improvement over classical dissociative force fields. In contrast, the NNP-derived solvation free energy significantly underestimates the experimental value, revealing a limitation of the present local NNP architecture for describing ion solvation thermodynamics. Our results demonstrate that DFT-trained NNPs can accurately describe Mg2+ hydration structure, diffusion, ion pairing, and exchange kinetics while highlighting limitations of local NNP architectures and the importance of long-range electrostatic contributions to achieve quantitative agreement with experimental ion solvation free energies.
Autor:innen und Institutionen
- Institute of Physics, University of Augsburg, Universitätsstraße 1, Augsburg86159, Germany.
- Faculty of Physics, University of Vienna, Vienna1090, Austria.
- Research Platform on Accelerating Photoreaction Discovery (ViRAPID), University of Vienna, Vienna1090, Austria.
ORCID-Profile: Sebastian Falkner, Pablo Montero de Hijes, Christoph Dellago, Nadine Schwierz
Förderung
- Augsburg University
- Austrian Science Fund
Zitieren
APA
Falkner, S., Montero de Hijes, P., Dellago, C., & Schwierz, N. (2026). Can DFT-Trained Neural Network Potentials Reproduce Structure, Solvation, and Water-Exchange Properties in Aqueous Magnesium Solutions?. Journal of Chemical Theory and Computation, 22(21), 11432-11441. https://doi.org/10.1021/acs.jctc.6c01270
Vancouver
Falkner S, Montero de Hijes P, Dellago C, Schwierz N. Can DFT-Trained Neural Network Potentials Reproduce Structure, Solvation, and Water-Exchange Properties in Aqueous Magnesium Solutions?. Journal of Chemical Theory and Computation. 2026;22(21):11432-11441. doi:10.1021/acs.jctc.6c01270
BibTeX
@article{Falkner2026,
author = {Falkner, Sebastian and Montero de Hijes, Pablo and Dellago, Christoph and Schwierz, Nadine},
title = {Can DFT-Trained Neural Network Potentials Reproduce Structure, Solvation, and Water-Exchange Properties in Aqueous Magnesium Solutions?},
journal = {Journal of Chemical Theory and Computation},
year = {2026},
volume = {22},
number = {21},
pages = {11432--11441},
doi = {10.1021/acs.jctc.6c01270}
}
EUVIMED