We use molecular dynamics simulations to study osmotically driven transport of water molecules through hexagonally packed carbon nanotube membranes. Our simulation setup comprises two such semipermeable membranes separating compartments of pure water and salt solution. The osmotic force drives water flow from the pure-water to the salt-solution compartment. Monitoring the flow at molecular resolution reveals several distinct features of nanoscale flows. In particular, thermal fluctuations become significant at the nanoscopic length scales, and as a result, the flow is stochastic in nature. Further, the flow appears frictionless and is limited primarily by the barriers at the entry and exit of the nanotube pore. The observed flow rates are high (5.8 water molecules per nanosecond and nanotube), comparable to those through the transmembrane protein aquaporin-1, and are practically independent of the length of the nanotube, in contrast to predictions of macroscopic hydrodynamics. All of these distinct characteristics of nanoscopic water flow can be modeled quantitatively by a 1D continuous-time random walk. At long times, the pure-water compartment is drained, and the net flow of water is interrupted by the formation of structured solvation layers of water sandwiched between two nanotube membranes. Structural and thermodynamic aspects of confined water monolayers are studied.

We present results on the thermodynamic and structural aspects of the hydration of hydrophobic solutes in three tetramethylammonium [N(CH 3 ) 4 + ] salt solutions at various concentrations obtained from molecular dynamics simulations. Monovalent counterions of different sizessF -, Cl -, and a relatively large model ion BIsare chosen in order to cover a range of kosmotropic to chaotropic behaviors. Chemical potentials of hard-sphere solutes obtained using test particle insertions display both salting-in and salting-out effects depending on the type of salt. Water and salt-ion densities in the vicinity of hard-sphere solutes are calculated. Small and strongly hydrated Fions (kosmotropes) are excluded from the vicinity of hydrophobic solutes, leading to an increase in local water densities near hydrophobic solutes (i.e., preferential hydration). This increases the excess chemical potential of hydrophobic solutes in solution which leads to salting-out. Opposite behavior is observed for large, less favorably hydrated BIions (chaotropes) which associate strongly with hydrophobic solutes. Compressive forces due to neighboring water molecules, cations, and anions on the surface of the hard sphere solute are calculated. We find that water molecules make the most significant contribution toward the total compressive force. This explains the observed linear correlation between the extent of preferential hydration or dehydration of the solute surface and salting-out or salting-in effects. The trends in the thermodynamics of hydration of hydrophobic solutes upon addition of salts are explained in terms of the structural hydration of individual salt ions.

The well defined shape and size of carbon nanotubes (CNTs) makes them attractive candidates for theoretical and experimental studies of various nanoscopic phenomena such as protection and confinement of molecular species as well as transport of molecules through their interior pores. Here we investigate solute partitioning and transport using molecular dynamics simulations of CNTs in mixtures of hydrophobic solutes and water. The hydrophobic pores of CNTs provide a favorable environment for partitioning of hydrophobic solutes. We find that the transfer of a methane molecule from aqueous solution into the CNT interior is favored by about 16 kJ/mol of free energy. In 50 molecular dynamics simulations, we observe that methane molecules replace water molecules initially inside the nanotubes, and completely fill their interior channels over a nanosecond time scale. Once filled with methane molecules, the nanotubes are able to transport methane from one end to the other through successive methane uptake and release events at the tube ends. We estimate a net rate of transport of about 11 methane molecules per nanotube and nanosecond for a 1 mol/L methane concentration gradient. This concentration-corrected rate of methane transport even exceeds that of water through nanotubes (∼1 per nanosecond at a 1 mol/L osmotic gradient). These results have implications for the design of molecule-selective CNT devices that may act through mechanisms similar to those of biological transmembrane channels.