Capillary Condensation: A Molecular Simulation Study

Abstract

The capillary condensation of a Lennard-Jones fluid confined to an adsorbing slit-like pore is studied using the grand canonical Monte Carlo method of molecular simulation for several values of the slit width, H. For each slit, we calculate the adsorbate density within the slit as a function of the chemical potential, or pressure, P, of the fluid—that is, the adsorption isotherm. Capillary condensation is the jump in density from a low, vapour-like value to a high, liquid-like value at some undersaturation P/P⁰ < 1 (where P⁰ is the saturation pressure). For large H, the transition is associated with metastable states and the system grand potential must be calculated to identify the point at which the two states are in equilibrium. As H is made smaller, the point of transition shifts to lower undersaturations while the metastable region shrinks and disappears at a critical width H _c. For H < H _c, the isotherms are continuous and exhibit steep (but not infinitely steep) risers connecting branches of low and high density. The length of the low density branch (i.e. the pressure at which the pore is completely filled) goes to—essentially—zero when the pore can accommodate just two adsorbed layers. Consideration of the structure of the adsorbed phase reveals that the density jump at capillary condensation is localised to the central part of the pore space. The density of the layers that build up at the walls is insensitive to whether the overall density is vapour or liquid-like. The breakdown of the predictions of the Kelvin equation for the capillary condensation pressure is illustrated with reference to the simulation results.