Non-equilibrium molecular dynamics of steady-state fluid transport through a 2D membrane driven by a concentration gradient

We use a novel non-equilibrium algorithm to simulate steady-state fluid transport through a two-dimensional (2D) membrane due to a concentration gradient by molecular dynamics (MD) for the first time. We confirm that, as required by the Onsager reciprocal relations in the linear-response regime, the solution flux obtained using this algorithm agrees with the excess solute flux obtained from an established non-equilibrium MD algorithm for pressure-driven flow. In addition, we show that the concentration-gradient-driven solution flux in this regime is quantified far more efficiently by explicitly applying a transmembrane concentration difference using our algorithm than by applying Onsager reciprocity to pressure-driven flow. The simulated fluid fluxes are captured with reasonable quantitative accuracy by our previously derived continuum theory of concentration-gradient-driven fluid transport through a 2D membrane [D. J. Rankin, L. Bocquet, and D. M. Huang, J. Chem. Phys. 151, 044705 (2019)] for a wide range of solution and membrane parameters, even though the simulated pore sizes are only several times the size of the fluid particles. The simulations deviate from the theory for strong solute–membrane interactions relative to thermal energy, for which the theoretical approximations breakdown. Our findings will be beneficial for a molecular-level understanding of fluid transport driven by concentration gradients through membranes made from 2D materials, which have diverse applications in energy harvesting, molecular separations, and biosensing.