Adsorption and structure of argon in activated porous carbons

Abstract

Molecular simulations are used to investigate the adsorption and structure of argon in ordered and disordered models of porous carbons. The ordered porous carbon (model A) is an assembly of regular slit pores of different sizes, while the disordered porous carbon (model B) is a structural model that reproduces the complex pore shape and pore connectivity of saccharose-based porous carbons. The same pore size distribution is used for models A and B so that we are able to estimate, for similar confinement effects, how the disorder of the porous material affects the adsorption and structure of the confined fluid. Adsorption of argon at 77.4 K in the two models is studied using Grand Canonical Monte Carlo simulations. The structure of the confined fluid is analyzed using crystalline bond order parameters and positional or bond orientational pair correlation functions. The filling pressure for the assembly of slit pores is much lower than that for the disordered porous carbon. It is also found that the isosteric heat of adsorption for the ordered porous model overestimates that for the disordered porous model. The results suggest that the agreement between models A and B would be improved if the same density of carbon atoms were used in these two models. Strong layering of Ar is observed at all pressures for model A. The confined phase is composed of liquid-like layers at low-pressures, which crystallize into well-defined hexagonal 2D crystals after complete filling of the pores. The structure of argon in the disordered porous carbon strongly departs from that in the slit pore model. Although its structure remains liquid-like overall, argon confined in model B is composed of both crystalline clusters and amorphous (solid or liquid) nano-domains.

Keywords:

• Adsorption
• Structure
• Porous carbons
• Nano-domains

Acknowledgements

We are grateful to Dr Roland Pellenq (CRMC-N, CNRS-Marseille, France) for very fruitful discussions. We thank the US National Science Foundation for support of this research (grant no. CTS-0211792). This research was performed using supercomputing resources from San Diego Supercomputer Center (NSF/MRAC-CHE050047S), the High Performance Computing Center at North Carolina State University, and the Centre Informatique National de l'Enseignement Superieur at Montpellier (CMC 2017).