The fluid flow in a macroscopic channel is typically determined assuming no-slip boundary
conditions at the walls. Such an assumption no longer holds for nanoscale flow, and the
finite flow slippage at the walls is a crucial determinant of a channel's permeability. However,
there is to date no predictive theory of the solid-liquid friction coefficient. Indeed, existing
models rely on the idea of surface roughness, which is no longer relevant for atomically
smooth surfaces that occur in nanofluidic devices. Particularly, one has to account for the …