Abstract
Molecular dynamics has been widely used in nanofluidics as it provides an atomistic description of fluid transport that cannot be achieved experimentally. The interactions among atoms are modeled by empirical force fields whose reliability for a specific problem must be tested against experimental evidence. Such validation is often difficult to obtain in nanofluidics due to the challenges associated with measuring nanoscale flows. Here, we systematically compare three popular force fields (CHARMM36, Amber-ff14SB, both with TIP3P water, and Amber-ff19SB with the OPC water) in terms of their electrohydrodynamical outputs. As test cases, we used CytK and MspA, two biological nanopores employed in sensing. For each nanopore, we simulated one cation-selective and one anion-selective mutant. Our results show that while, overall, the anion/cation selectivity and the direction of electroosmosis are coherent across the force fields, quantitative differences are observed. These differences cannot be simply explained by the transport properties of the solution, such as the ion conductivity and viscosity. For narrow pores such as MspA, these differences can become so significant as to yield qualitatively different outcomes; for example, one force field may predict no detectable electroosmosis while another shows a clear electroosmotic flow.