The absorptions and separations problematic compounds (PCs) in aqueous phase are vital for life-support systems in International Space Station. The main PCs are monomethylsilanetriol (MMST), dimethylsilanediol (DMSD), and trimethylsilanol (TMS) from hydroxylation of polydimethylsiloxane compounds. Another PC is dimethylsulfone (DMSO<sub>2</sub>), which is primarily generated in the Urine Processor Assembly. Herein, we abstracted 50959 zeolites with identical elements composition, stoichiometry, and similar volumes from hypothetic database. We choose 500 zeolites from them randomly. Then, grand canonical Monte Carlo (GCMC) simulations in Sorption module of Materials Studio 8.0 were performed for absorption of four PCs in 500 zeolites. While simulations, we calculated the average adsorption loading (/mol) and isosteric heats (kcal/mol) of them, where larger adsorption loadings and isosteric heats indicate better adsorption performances.<br/>We further used the results of 2000 absorption systems (500 × 4) as templates to train the models. Some key factors which affect adsorption performances, for instance, average bond numbers, bond lengths, and pore sizes were derived during the training process, thereby establishing solid foundations for the rational design of zeolites adsorbents. We plan to employ Support Vector Machines, Decision Trees, or Linear Regression methodology to train the machine, as implemented in scikit-learn. Finally, we used the well-trained model to make a high-throughput screening for the remaining structures in 50595 zeolites structures, to effectively predict the absorptions and separations performance, and select high-performance zeolites-based adsorbents as the potential candidates for future experimentalists.

Computational screening of zeolites for effective removal of problematic compounds in the International Space Station

2019 Spring National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.