Memory-kernel extraction for different molecular solutes in solvents of varying viscosity in confinement

The friction coefficient of molecular solutes depends on the solute, on the solvent, and on the solute-solvent interactions, but is typically assumed to not depend on an externally applied force that acts on the solute. In this paper we compute the friction memory function from molecular dynamics simulations and show that the friction coefficients of harmonically confined methane, water, Na^{+}, an artificial Na^{-} ion, and glycerol in water in fact increase with confinement strength. The results show that the friction increase with confinement strength is a fundamental effect that occurs for hydrophobic, hydrophilic, as well as charged molecules. We demonstrate that a parameter-free extraction of the running integral over the memory function yields the most robust results when compared to methods based on parametrization or Fourier transforms. In all systems, this friction increase is accompanied by a slowdown of the solvent dynamics in the first hydration shell of the solutes. By simulations of a confined glycerol molecule in water-glycerol mixtures, we furthermore demonstrate that the friction dependence on the confining potential is magnified in more viscous solvents, which suggests that this effect plays an important role for larger molecules in highly viscous solutions like polymer melts, in line with dynamic scaling arguments.