Ion-Specific Anomalous Water Diffusion in Aqueous Electrolytes: A Machine-Learned Many-Body Force Field Study with MACE
Massimo Ciacchi, Ilnur Saitov, Nico Di Fonte, Isabella Daidone, Carlo Pierleoni · 2026
Read Ion-Specific Anomalous Water Diffusion in Aqueous Electrolytes: A Machine-Learned Many-Body Force Field Study with MACE by Massimo Ciacchi, Ilnur Saitov, Nico Di Fonte, Isabella Daidone, Carlo Pierleoni (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-13659.txt.
Introduction
Ciacchi and four colleagues explain a puzzle in salt water: water moves faster in cesium iodide solutions yet slower in table-salt solutions. Using molecular dynamics with a machine-learned many-body force field trained in the MACE framework, they reproduce both anomalies. The gain over the earlier DeePMD approach is largest for table salt. Strong sodium-water binding in the first hydration shell slows water there, while a loose iodide shell lets water exchange fast in cesium iodide.
Catalog record
| CREATOR | Massimo Ciacchi, Ilnur Saitov, Nico Di Fonte, Isabella Daidone, Carlo Pierleoni |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 52619 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.13659
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Ion-Specific Anomalous Water Diffusion in Aqueous Electrolytes: A Machine-Learned Many-Body Force Field Study with MACE Massimo Ciacchi, Ilnur Saitov, Nico Di Fonte, Isabella Daidone, Carlo Pierleoni Abstract The dynamics of water in electrolyte solutions exhibits a striking, ion-specific anomaly: the diffusion coefficient of water is enhanced relative to the neat liquid in chaotropic CsI solutions, yet suppressed in kosmotropic NaCl solutions. This phenomenon, long challenging for classical force-field-based molecular dynamics, is studied here using classical molecular dynamics simulations with a many-body machine-learned force field (MLFF) trained within the MACE equivariant graph neural network framework. The force field is trained on energies, forces, and stresses computed at the density functional theory level with the revPBE-D3 exchange–correlation functional, which provides a reliable balance between accuracy and computational efficiency for aqueous systems. Simulations of NaCl and CsI aqueous solutions at ambient conditions over a concentration range of 0.89–3.56 mol/kg reproduce the experimentally observed anomalous diffusion and show a quantitative improvement over previous results obtained with the DeePMD framework, trained on the same theory, particularly for NaCl solutions. This improvement is traced to a stronger Na + –water interaction in the first hydration shell and the non-negligible retarding contribution of the second hydration shell of Na + . For CsI solutions, the water acceleration is shown to be primarily driven by the anion I - , whose diffuse and weakly structured hydration shell facilitates rapid water exchange with the bulk. These results are rationalised through a shell-decomposition analysis of time-dependent water diffusivities and ion–oxygen potentials of mean force providing a coherent microscopic picture of the acceleration–retardation mechanism in the studied aqueous electrolytes. I Introduction Water is arguably the most important solvent on Earth, playing a central role in chemistry, biology, and numerous technological applications. Despite decades of investigation, a quantitative description of its properties, both as a pure substance and in aqueous solutions, remains an ongoing challenge, owing in part to the limited structural information accessible from direct experimental measurements. 30 Among the many open problems in the physics of ionic solutions, a particularly striking and long-debated phenomenon is the ion-specific effect of dissolved salts on the translational dynamics of water. Over six decades ago, Gurney 15 introduced the notion of “structure-making” (kosmotropic) and “structure-breaking” (chaotropic) ions to rationalize how different ions perturb the hydrogen-bond network of liquid water. These ideas have since been generalized into the celebrated Hofmeister series, an empirical ordering of ions according to their ability to salt out proteins from solution, 17 ; 4 and have been applied to explain a broad range of thermodynamic and transport phenomena in electrolyte solutions. However, the microscopic origins of ion-specific behavior have remained elusive, and the relationship between structural perturbations to the hydrogen-bond network and the corresponding dynamical response of water is still an active area of research. A particularly illuminating manifestation of ion specificity is the anomalous concentration dependence of the water diffusion coefficient, D_{\mathrm{w}} , in electrolyte solutions. Nuclear Magnetic Resonance (NMR) experiments 35 have established that, below approximately 3 M salt concentration, D_{\mathrm{w}} increases with concentration in chaotropic CsI solutions, while it decreases monotonically in kosmotropic NaCl solutions. A quantitative benchmark is provided by the data of Müller and Hertz 35 : at 3 M, the diffusion coefficient of water is enhanced by approximately 23% in CsI and suppressed by approximately 19% in NaCl, relative to neat water. This contrast behavior is intimately related to the viscosity B -coefficients of the respective ions: I - and Cs + both carry negative B -coefficients indicative of mobility enhancement, while Na + has a large positive value signaling retardation. 31 The failure of classical force fields to reproduce these trends has been well documented. Non-polarizable force fields systematically predict D_{\mathrm{w}}/D_{0}<1 , with D_{0} being the diffusion coefficient in bulk water, at all concentrations and for all salts, regardless of their chaotropic or kosmotropic character. 35 ; 10 This deficiency is not specific to a particular parameterization but is common to almost all classical non-polarizable models and even to some polarizable force fields. 1 Berkowitz and co-workers 49 argued that an explicitly fluctuating-charge model with dynamical charge transfer is necessary to capture the acceleration of water in mildly chaotropic KCl solutions; yet even such models struggle to generalize across the full Hofmeister series. A decisive step forward was provided by Ding, Hassanali, and Parrinello, 10 who performed Ab Initio molecular dynamics (AIMD) simulations, based on density functional theory (DFT) with the revPBE-D3 exchange-correlational functional, for 3 M NaCl and CsI solutions. They demonstrated that explicit treatment of the electronic degrees of freedom is necessary to reproduce the qualitative experimental trends, and identified dynamical heterogeneity in the water ensemble as the key microscopic feature absent from empirical force-field simulations. Their analysis showed that the ions do not disrupt the hydrogen-bond network in any dramatic manner, but rather induce subtle, measurable changes in the tails of the dynamical distribution. Na + and Cl - participate in directed ring structures of the hydrogen-bond network in a manner analogous to water molecules, preserving much of the network topology while modulating its dynamics. Despite this conceptual breakthrough, AIMD simulations are severely limited by accessible system sizes (typically \sim 100 atoms) and simulation times (a few tens of picoseconds), making statistically converged estimates of transport properties such as diffusion coefficients and viscosities challenging. 10 A major advance in the molecular modeling of aqueous systems was achieved through the development of many-body potential energy functions, most notably the MB-pol and MB-nrg frameworks introduced by Paesani and co-workers, which explicitly incorporate many-body interactions derived from high-level electronic structure calculations. 2 ; 33 ; 44 These models have demonstrated near-quantitative accuracy for pure water from small clusters to the liquid phase, and have been successfully extended to ion–water systems, providing a consistent and predictive description of hydration structure, energetics, and spectroscopic properties. However, the computational cost associated with the explicit evaluation of many-body terms still limits their routine application to large systems, high salt concentrations, and long-time dynamical properties. Machine-learned force fields (MLFFs) trained on DFT or MB-pol data offer an attractive route to circumvent these limitations: they inherit the accuracy of the underlying electronic structure method while reducing the computational cost by several orders of magnitude, enabling nanosecond-scale trajectories for systems of hundreds to thousands of atoms. 37 ; 38 ; 47 ; 48 Several MLFF frameworks have been applied to water and aqueous ionic solutions. The Princeton group developed the Deep Potential Molecular Dynamics (DeePMD) framework 53 and applied it extensively to bulk water and ionic solutions. In a landmark study, Avula, Klein, and Balasubramanian 1 demonstrated that DeePMD force fields trained on revPBE-D3 data successfully reproduce the anomalous diffusion phenomenon in both NaCl and CsI solutions, overcoming the limitations of classical force fields. Their work established a structure–property relationship linking the diffuse, weakly structured hydration shell of Cs + to the faster water dynamics in CsI, and the tightly bound, well-defined hydration shells of Na + to the retardation in NaCl. Using a different DFT functional (SCAN), Zhang et al. 52 further clarified that the structural changes induced by salt dissolution differ qualitatively from those induced by applied external pressure. Panagiotopoulos and Yue 39 also illustrated the promise of