Transferability of Coarse Grained Potentials: Implicit Solvent Models for Hydrated Ions

Understanding the relation between structural and thermodynamic quantities obtained with simplified-e.g., coarse-grained (CG) or implicit-solvent models is an ongoing challenge in the field of multiscale simulation. Assessing the transferability of such models to state points that differ from the one where the model was parametrized is important if one wants to apply these models to complex systems, which, for example, exhibit spatially varying compositions. Here, we investigate the transferability of CG (in this case implicit-solvent) ion models with effective pair potentials derived at very low concentrations to different ion concentrations in aqueous solution. We evaluate both thermodynamic and structural properties of systems of NaCl in aqueous solution both in atomistic explicit-solvent and CG simulations. For the explicit solvent simulations, osmotic coefficients have been calculated at a wide range of salt concentrations and agree very well with experimental data. It had been shown previously that a concentration-dependent dielectric permittivity can be used to make effective implicit-solvent pair potentials transferable since it accounts for the effect of ion concentration on solvent properties, resulting in very good osmotic properties of these models for a certain range of salt concentrations. We investigate the explicit and implicit solvent models also in terms of structural properties, where we can show how with a concentration-dependent dielectric constant one obtains very good structural agreement at low and intermediate salt concentrations, while for larger salt concentrations, multibody ion ion correlations put a limit to straightforward transferability. We show how guided by this structural analysis the transferability of the implicit-solvent model can be improved for high ion concentrations. Doing so, we obtain transferable implicit-solvent effective pair potentials which are both structurally and thermodynamically consistent with an explicit solvent reference model.