Realistic Molecular Models for Disordered Porous Carbons

Abstract

The complex pore morphology and topology of many non-graphitizable porous carbons is not captured by the current molecular models that are used in analysis of adsorption isotherms. We present a novel constrained reverse Monte Carlo method to build models that quantitatively match carbon-carbon pair correlation functions obtained from experimental diffraction data of real nanoporous carbons. Our approach is based on reverse Monte Carlo with carefully selected constraints on the bond angles and carbon coordination numbers to describe the three-body correlations. Through successive Monte Carlo moves, using a simulated annealing scheme, the model structure is matched to the experimental diffraction data, subject to the imposed three-body constraints. We modeled a series of saccharose-based carbons and tested the resulting models against high resolution transmission electron microscopy (TEM) data. Simulated TEM images of the resulting structural models are in very good agreement with experimental ones. For the carbons studied, the pore structure is highly convoluted, and the commonly used slit pore model is not appropriate. We simulated adsorption of nitrogen and argon at 77 K using grand canonical Monte Carlo, and diffusion of argon at 300 K using canonical molecular dynamics simulations. The isosteric heats of adsorption at 77 K are in excellent agreement with experimental results. The adsorption isotherms and heats of adsorption in these models do not resemble those for fluids in slit pores having the same pore size distribution. We found that diffusion in the structural models is non-Fickian. Instead, a strong single-file character is observed, revealed by the proportionality of the mean square displacement to the square root of time at relatively long times. The single-file mode is a consequence of the small sizes of the quasi one-dimensional pores in the adsorbent models.