Shear viscosities away from the melting line: A comparison of equilibrium and nonequilibrium molecular dynamics

Abstract

Doubts about the validity of the nonequilibrium molecular dynamics (NEMD) methods of computing shear viscosity have persisted, partly because of the apparent disagreement (∼25%) between NEMD and equilibrium Green–Kubo (GK) results for the Lennard‐Jones system near its triple point. This region of the phase diagram near the melting line is the so‐called ‘‘molasses’’ regime where the tail of the shear‐stress autocorrelation function is quite large, deviating from ‘‘exponential’’ decay at a level of about 10%. In order to see whether the effects of the ‘‘molasses tail’’ might be obscuring a more profound difference between NEMD and GK results, we have carried out independent NEMD and GK calculations for a state in the LJ fluid far away from this troublesome molasses region, namely at a temperature twice critical and a density between the triple and critical points. We find the NEMD and GK results for the linear shear viscosity to be in good agreement.