We report a Grand Canonical Monte Carlo simulation study of Ar adsorption at 77 K in silica nanopores having various morphologies/topologies. Both the morphological and topological disorders are shown to significantly affect the capillary condensation phenomenon. In the case of an ellipsoidal pore, we observe that the filling mechanism is similar to that of a cylindrical pore having the same section area but with a lower condensation pressure. We show that the adsorption/ desorption hysteresis loop is asymmetrical for the pore with constrictions while it is symmetrical for the regular cylindrical pore. Moreover, the Ar adsorption isotherm for the constricted pore reproduces the main features of that for the fully disordered Vycor-like porous matrix. The results for the different pore geometries (having no direct interface with the gas reservoir) indicate that the desorption occurs through cavitation at a pressure driven by the smallest void size. We also consider the validity of the BET and BJH methods for the different porous matrices. Except for the Vycor-like matrix, the BET surface assessed from Ar adsorption isotherm at 77 K always significantly overestimates the intrinsic surface of the pore (even for a planar surface). The disagreement between the BET surface and the geometrical surface is found to increase as the confinement increases (cylindrical pore) and/or the shape of the pore becomes asymmetrical (ellipsoidal pore). Interestingly, the best agreement between the BET and the geometrical surface is found in the case of the pore with a constriction, i.e. a system that exhibits a surface with both negative and positive curvature regions. This idea is supported by the results for the Vycor-like matrix, which has a distorted surface (many negative curvature regions combined with positive curvature regions): for this disordered porous matrix, the BET surface is found to underestimate the intrinsic surface. Finally, we show that the pore size determined using the BJH method always underestimates the pore size, in agreement with previous experimental and simulation studies.
