DFT-trained MACE NNPs reproduce Mg2+ first-shell structure, diffusion, ion pairing and water-exchange kinetics within one order of magnitude of experiment but significantly underestimate solvation free energy.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
physics.chem-ph 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Can DFT-trained neural network potentials reproduce structure, solvation, and water-exchange properties in aqueous magnesium solutions?
DFT-trained MACE NNPs reproduce Mg2+ first-shell structure, diffusion, ion pairing and water-exchange kinetics within one order of magnitude of experiment but significantly underestimate solvation free energy.