Driving forces in MD simulations of transition and 'Free' flows. Issue 5 (13th April 2017)

• Record Type: Journal Article
• Authors: Dayhoff, Guy W.; Rogers, David M.
• Is Part Of: Molecular simulation. Volume 43:Issue 5/6(2017)
• Page Start: 467
• Page End: 477
• Publication Date: 2017-04-13
• Subjects: Finite-size effects -- hydrodynamics -- channel flow -- Joule–Thomson -- local equilibrium; Molecular dynamics -- Computer simulation -- Periodicals; Statistical mechanics -- Computer simulation -- Periodicals; 539.6
• DOI: 10.1080/08927022.2016.1273524
• Languages: English
• ISSNs: 0892-7022

Abstract

Abstract: Simulations of porous gaseous flows are routinely used to investigate membrane permeation in catalytic adsorption and separation problems. Although basic continuum equations are supposed to breakdown in these nanoscale pores, many studies of force/flow relations assume flow to be linear in chemical potential or pressure differences. This work tests common assumptions using simulations of an atomistic, Lennard–Jones pore flow with distant, Langevin forcing at densities stretching through the transition and free flow regimes. Using NVE dynamics in very large boundary reservoirs, we find local equilibrium is established in the steady-state, but also identify two new finite-size effects. First, there is a steady flow of heat from the high-pressure reservoir backward to the thermostat region, and second, a significant proportion of the channel flow originates from the monolayer adsorbed to the flat outer wall. All walls are shown to obey a simple Langmuir adsorption isotherm at these low ( kPa) pressures, even in the presence of flow. Despite multi-layer formation on the inner pore walls as density increases, the current carried by atoms at the wall has the same proportion to current carried through the channel center under nearly all conditions tested (with constant pore diameter). Comparing our flow rates to Fickian and Knudsen linear relations shows that the difference in reservoir pressure is significantly more predictive than the difference in chemical potential for this size regime.