(57c) the Development of Data Mining-Molecular Modeling Approach to Screen Physical Solvents for Gas Separation

Integrating the design of both process and solvent for gas separation involves simultaneously optimizing the process structures, operating conditions and solvent molecules. This task is challenging partly due to the large number of non-linear equations for process models, the large number of variables (such as the operating conditions) and the large solvent chemical space to be explored. One way to alleviate this complicated problem is through sub-optimizations by restricting the search space in each step of process design and solvent development. In this work, we focus on solvent design to physically and selectively absorb one gas from the other in the gas mixture. Specifically, a hybrid computational approach was developed to systematically screen solvents to separate CO₂ from a fuel gas mixture which contains CO₂, H₂ and H₂O. The objective of this work is to search for solvents which have desired properties such as hydrophobicity, high CO₂ loading (volume based), high CO₂/H₂ solubility selectivity, non-foaming, low viscosity, low heat capacity, good thermal stability, no environmental and safety issues, and cheap price. The computational approach involves searching the NIST database for pure organic compounds, the in-house computational database development, and atomistic simulations. By using this computational approach, several promising solvents have been identified. For example, we have identified one super-hydrophobic solvent, which has a high boiling point of ~275 °C, large CO₂ loading of 1.06 mol/MPa/L at 25 °C, CO₂ heat of absorption of 10.6 kJ/mol (not high), large CO₂/H₂ solubility selectivity of 50 at 25 °C, low viscous (~4 cP) at ambient condition, heat capacity of 1.64-1.87 J/g°C at 25-100 °C (not high and much less than the heat capacity value for water), non-foaming, and reasonably cheap. The predictions are being confirmed by the experiments at NETL. In addition, we will present the largest CO₂ solubility and CO₂/H₂ solubility selectivity in any organic compound, which would be useful to implement the techno-economic analysis in process modeling to estimate the minimum operating and capital costs.