Computing Mixture Adsorption in Porous Materials through Flat Histogram Monte Carlo Methods

Mixture adsorption properties of
porous materials are
critical
to determine their potential as adsorbents in separation applications.
Toward the discovery of optimal adsorbents, in silico screening studies typically employ the grand canonical Monte Carlo
(GCMC) technique to compute adsorption properties of gas mixtures
in materials of interest at a given condition (i.e., composition,
total pressure, and temperature) or to compute their adsorption properties
for each component, followed by utilizing methods to predict mixture
adsorption isotherms. However, the former approach results in the
need for repeated calculations when different conditions such as compositions
are considered. For the latter, the predictions may involve uncertainties,
sometimes originating from the fitting quality to the pure component
isotherms, and repeated simulations may also be needed for different
temperatures. To this end, this study demonstrates the potential of
flat histogram Monte Carlo methods in addressing the abovementioned
shortfalls. Specifically, the so-called NVT + W method,
first reported by Smit and co-workers, is extended herein to determine
the macrostate probability distribution (MPD) of binary mixtures in
porous materials. The obtained MPD can be reweighted to any conditions,
yielding accurate adsorption isotherms of any desired compositions
and temperatures. This approach, denoted as 2D NVT + W,
is also compared with the widely adopted ideal adsorbed solution theory
(IAST) method, and the former is found to offer more reliable predictions.
Overall, the 2D NVT + W approach represents an efficient
and effective alternative to compute mixture adsorption isotherms
for porous materials, and the obtained MPD can be conveniently reused
by peer researchers. A user-friendly Python code is also provided
along with this article to employ this method.