Tunable Surface Hydrophobicity and Fluid Transport through Nanoporous Membranes

There are more than three billion people across the globe that struggle to obtain clean drinkable water. One of the most promising avenues for generating potable water is through reverse osmosis and nanofiltration. Both solutions require a semipermeable membrane that prohibits passage of unwanted solute particles but allows passage of the solvent. Atomically thin two-dimensional membranes based on porous graphene show great promise as semipermeable materials, but modeling fluid flow on length scales between the microscopic (nanometer and smaller) and macroscopic (micron and larger) regimes presents formidable challenges. This thesis explores both equilibrium and nonequilibrium aspects of this problem and develops new methodology for simulating systems away from thermal equilibrium. First, we hypothesize that there is a wetting penalty for water as it tries to breach a sheet of graphene that should be naturally hydrophobic. By using equilibrium molecular dynamics simulations, we show that the hydrophobicity depends sensitively on the degree of electrical doping, offering an opportunity to tune the hydrophobic effect of graphene using small amounts of doping. The wetting contact angle, a measure of hydrophobicity, changes dramatically with the voltage applied to single layer graphene. We find that the sensitivity of the hydrophobic effect to voltage depends not on hydrogen bonding motifs at the interface between graphene and water, but instead on a phenomenon known as electrowetting. The theory of electrowetting predicts that the difference in surface tensions that defines the contact angle is quartic in the voltage, rather than quadratic, as it would be in bilayer graphene or in a two-dimensional metal. To explore the nonequilibrium aspects of fluid passage through atomically thin membranes, we developed a molecular dynamics methodology for simulating fluid flow at constant flux based on Gauss's principle of least constraint. This method develops microscopic equations of motion that satisfy specified constraints on the kinetic temperature and total mass flux. As a proof of principle, we simulate the flow of a simple monoatomic fluid and observe emergent and collective behaviors consistent with both known hydrodynamic solutions and expectations for velocity distributions from statistical mechanics. We compare results from the Gauss method simulations with that of a method commonly used in the literature. By computing the relationship between the pressure drop across a pipe-like region and the fluid current through it, we find that these two methods agree quantitatively with one another and comment on the advantages and disadvantages for both methods