EUVIMEDEUVIMED entdecken
Crossref

Electron Alchemy with Machine-Learned Interatomic Potentials: Case Studies of Local Charge in Bond Dissociation Curves

Ericka Roy Miller, Vignesh Sathyaseelan iD, Dylan M Gilley, Olexandr Isayev iD, Brett M Savoie iD

Journal of Chemical Theory and Computation · 28.07.2026 · 22 (14) · 7264-7273

DOI: 10.1021/acs.jctc.6c00719 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Chemical Theory and Computation, online erschienen am 08.07.2026. Verfasst von Ericka Roy Miller und 4 weiteren Autor:innen. Beteiligte Einrichtung: Department of Chemical and Biomolecular Engineering, University of Notre Dame, Notre Dame, Indiana46556, United States. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 41 Literaturangaben. Gefördert von: Office of Naval Research.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The convergence of molecular dynamics simulations and machine-learned interatomic potentials (MLIPs) promises density functional theory (DFT) level accuracy at near-classical force-field computational costs. However, while the average fidelity to reference energies and forces approaches perfection, several failure modes limit MLIP reliability in production simulations. These include spurious bond formation, inconsistent reproduction of long-range interactions, and inconsistent spin-state references. Here, the origins of these behaviors are studied by benchmarking the UMA, ORB, MACE, and AIMNet2 models against reference DFT bond dissociation curves for an illustrative range of species. These benchmarks reveal that models without explicit atomic charge resolution predict spurious stable bonds between like-charged halide anions, effectively transmuting two Cl– ions into neutral Cl2. Models with atomic partial charge equilibration correctly predict repulsion in these systems. Conversely, several secondary limitations are exposed in these benchmarks, including inconsistent agreement with unrestricted DFT (uDFT) versus restricted DFT (rDFT) energies and inconsistent core-region treatment. This comparative analysis suggests that, while artifacts related to core repulsion and asymptotic electrostatics are readily repairable through improved physical priors and better data curation, the issue of spurious bond formation is intrinsic to the inability of global charge specification to disambiguate similar local geometries at different charge and spin states.

Autor:innen und Institutionen

  • Department of Chemical and Biomolecular Engineering, University of Notre Dame, Notre Dame, Indiana46556, United States.
  • Charles D. Davidson School of Chemical Engineering, Purdue University, West Lafayette, Indiana47907, United States.
  • Department of Chemistry, Mellon College of Science, Carnegie Mellon University, Pittsburgh, Pennsylvania15213, United States.
  • Department of Materials Science and Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania15213, United States.

ORCID-Profile: Vignesh Sathyaseelan, Olexandr Isayev, Brett M Savoie

Förderung

  • Office of Naval Research

Zitieren

APA

Roy Miller, E., Sathyaseelan, V., Gilley, D. M., Isayev, O., & Savoie, B. M. (2026). Electron Alchemy with Machine-Learned Interatomic Potentials: Case Studies of Local Charge in Bond Dissociation Curves. Journal of Chemical Theory and Computation, 22(14), 7264-7273. https://doi.org/10.1021/acs.jctc.6c00719

Vancouver

Roy Miller E, Sathyaseelan V, Gilley DM, Isayev O, Savoie BM. Electron Alchemy with Machine-Learned Interatomic Potentials: Case Studies of Local Charge in Bond Dissociation Curves. Journal of Chemical Theory and Computation. 2026;22(14):7264-7273. doi:10.1021/acs.jctc.6c00719

BibTeX
@article{RoyMiller2026,
  author = {Roy Miller, Ericka and Sathyaseelan, Vignesh and Gilley, Dylan M and Isayev, Olexandr and Savoie, Brett M},
  title = {Electron Alchemy with Machine-Learned Interatomic Potentials: Case Studies of Local Charge in Bond Dissociation Curves},
  journal = {Journal of Chemical Theory and Computation},
  year = {2026},
  volume = {22},
  number = {14},
  pages = {7264--7273},
  doi = {10.1021/acs.jctc.6c00719}
}
Kostenlos starten →