Molecular dynamics simulations are used to investigate the relation between the self-diffusion and structural relaxation of water in cylindrical pores with varying diameters of 2–6 nm over a wide temperature range of 220–350 K. We consider amorphous silica pores and chemically neutral pores, consisting of pinned water molecules. We find that water dynamics are, on average, slower in narrower pores. This confinement effect differs little between silica and neutral pores but is stronger at lower temperatures. A comparison between the self-diffusion coefficients and structural relaxation times reveals an increasing breakdown of the Stokes–Einstein relation upon cooling, which is more pronounced in the confinements than in the bulk. For deeper insights, position-resolved analyses of water dynamics are performed. We find that the structural relaxation strongly slows down near the silica and neutral walls, where the magnitude and range of the effect increase when the temperature is decreased. This temperature-dependent spatial heterogeneity enables an understanding of the confinement effects on water dynamics. In particular, we show in the framework of a random-walk approach that knowledge about the variation in structural relaxation times across the pores allows us to nearly quantitatively explain the diameter and temperature dependence of the self-diffusion coefficient. We conclude that liquid diffusion in nanopores is strongly affected by gradients in the molecular mobility near the pore walls. This effect, on the one hand, causes a failure of the Stokes–Einstein relation when employing confinement-averaged observables and, on the other hand, provides a means to tune the transport through nanopores by chemically changing the liquid–matrix interactions.
Nisar et al. (Tue,) studied this question.