Solving Chemical Absorption Equilibria using Free Energy and Quantum Chemistry Calculations: Methodology, Limitations, and New Open-Source Software

Abstract

We developed an open-source chemical reaction equilibrium solver in Python (CASpy, https://github.com/omoultosEthTuDelft/CASpy) to compute the concentration of species in any reactive liquid-phase absorption system. We derived an expression for a mole fraction-based equilibrium constant as a function of excess chemical potential, standard ideal gas chemical potential, temperature, and volume. As a case study, we computed the CO2 absorption isotherm and speciation in a 23 wt % N-methyldiethanolamine (MDEA)/water solution at 313.15 K, and compared the results with available data from the literature. The results show that the computed CO2 isotherms and speciations are in excellent agreement with experimental data, demonstrating the accuracy and the precision of our solver. The binary absorptions of CO2 and H2S in 50 wt % MDEA/water solutions at 323.15 K were computed and compared with available data from the literature. The computed CO2 isotherms showed good agreement with other modeling studies from the literature while the computed H2S isotherms did not agree well with experimental data. The experimental equilibrium constants used as an input were not adjusted for H2S/CO2/MDEA/water systems and need to be adjusted for this system. Using free energy calculations with two different force fields (GAFF and OPLS-AA) and quantum chemistry calculations, we computed the equilibrium constant (K) of the protonated MDEA dissociation reaction. Despite the good agreement of the OPLS-AA force field (ln[K] = −24.91) with the experiments (ln[K] = −23.04), the computed CO2 pressures were significantly underestimated. We systematically investigated the limitations of computing CO2 absorption isotherms using free energy and quantum chemistry calculations and showed that the computed values of μiex are very sensitive to the point charges used in the simulations, which limits the predictive power of this method.