(520e) Molecular Model Development with Reliable Charge Distributions for Gaseous Adsorption in Nanoporous Materials

Nanoporous adsorbent materials possess significant potential in gas separation and storage. Advances in their development have remained challenging owing to difficulties in identifying promising ones among a myriad of possible candidates. Therefore, to effectively and efficiently facilitate the materials search, molecular simulations can play an important role. For a reliable employment of molecular simulations for predicting gaseous adsorption in nanoporous materials, an accurate description of the gas molecule of interest is critically important. To this end, we present herein a systematic and robust scheme for the model development of small gaseous molecules. In this method, all the model parameters are extensively and efficiently optimized. To ensure a reliable representation of the electrostatic potential (ESP) surrounding the molecule of interest, the number and location of pseudo-sites and the corresponding charge distribution are fitted to the ESP determined by ab initio density functional theory (DFT) calculations. Subsequently, the van der Waals interaction parameters are fitted to reproduce the experimental vapor-liquid equilibrium. This scheme ensures two critical features for the model development; (1) accuracy of the model by reproducing the DFT-computed ESP and (2) computational efficiency for the model development thanks to the decoupling of model parameters in optimization. While this systematic method should be applicable to general gas molecules, as a proof of concept, hydrogen sulfide (H₂S) molecular model is developed using this method, as its emission has serious impacts on the environment and humankind’s health. Our results indicate that the H₂S models developed from this work can adequately present the ESP and provide accurate predictions of adsorptive properties in nanoporous materials, including all-silica zeolites and metal-organic frameworks. In addition to the adsorbed phase, we note that these models are also proven to reproduce the experimental bulk liquid properties such as radial distribution function reasonably well. We anticipate this method can largely facilitate the development of accurate molecular models, which in turn can expedite the computational discovery of optimum adsorbent materials for various applications.