Computer simulations of N2 adsorption on graphite frequently use the 10-4-3 equation with Steele’s molecular parameters to describe the dispersive-repulsive interaction between a molecule and graphite. This model assumes that graphite is a uniformly homogeneous continuum solid, and its derivation implies the following assumptions: (1) the solid is built from stacked, equally spaced graphene layers, (2) there is an infinite number of layers, and (3) the carbon atom molecular parameters are invariant for all layers (collision diameter of 0.34 nm and reduced well depth of interaction energy of 28 K). Despite the fact that this model can give an acceptable description of experimental data for this system, there are experimental observations that simulation results fail to account for. First, the isotherm does not exhibit a step in the sub-monolayer coverage region at 77 K, which is attributed to a transition from the supercritical state of the adsorbate to the commensurate state, and therefore fails to reproduce the cusp and heat spike in the experimental isosteric heat curve versus loading at close to monolayer coverage. Second, the simulation results overpredict the experimental data in the multilayer region. These discrepancies suggest that (1) the absence of lateral corrugation in the 10-4-3 potential misses the commensurate to incommensurate transition and (2) the long-range solid-fluid potential, experienced by the second and higher layers onwards, is too strong. Here we examine a revised graphite potential model that incorporates three features absent from the 10-4-3 model: (1) an energetic corrugation of the potential arising from the discrete atom structure of the adsorbent, (2) the unequal spacing of the graphene layers due to the anisotropic force field acting on graphene layers at the surface, and (3) the different polarizabilities of carbon atoms in graphite, parallel and normal to the graphene surface. These features are corroborated by a number of experimental measurements and quantum-mechanical calculations: (1) the Low-Energy Electron Diffraction (LEED) and Surface-Extended X-ray Absorption Fine Structure (SEXAFS) experiments show that the first adsorbate layer is smaller than predicted by the 10-4-3 model with the traditional molecular parameters suggested by Steele, and (2) the potential well depth for atoms in graphene is stronger than for C-atoms in graphite. The simulation results using this revised graphite model give an improved description of the fine features of adsorption of N2 on graphite: the sub-step in the first layer of the isotherm, the spike in the isosteric heat curve versus loading, and the coverage at higher loadings.
Read full abstract