We present a general atomistic simulation framework for efficient reactive equilibrium calculations in dilute solutions, and its application to CO2 reactive absorption in aqueous alkanolamine solutions. No experimental data of any kind for the
solvents is required and no empirical adjustments are required for its implementation. This hybrid methodology involves calculating the required reaction equilibrium
constants by combining high–level quantum chemical calculations of ideal–gas standard reaction Gibbs energies (∆G0
) with conventional free energy calculations for
transfer of the molecular species from the ideal gas to infinite dilution in the solvent (i.e, their solvation free energies). For the solvation free energy calculations,
we use explicit solvent molecular dynamics simulations with the General AMBER
Force Field (GAFF). The resulting equilibrium constants are then coupled with a
macroscopic Henry–Law–based ideal solution model to calculate the solution speciation and the CO2 partial pressure, PCO2
. We show results for seven primary
amines: monoethanolamine (MEA), 2–amino–2–methylpropanol (AMP), 1–amino–2–
propanol (1–AP), 2–amino–2–methyl–1,3–propanediol (AMPD), 2–aminopropane–1,3–
diol (SAPD), 2–(2–aminoethoxy)ethanol (2–AEE) or diglycolamine (DGA), and 2–
amino–1–propanol (2–AP). Experimental speciation and PCO2 data for some of these
is available, with which we validate our methodology. We predict new results for others
in cases when such data is unavailable, and provide explanations for the experimental
inability to detect carbamate species in some cases. Our results for the pK value of
the carbamate reversion reaction are within the chemical accuracy limit of 218.506/T
in comparison with experiment when such data exist, which at 298.15 K corresponds
to 0.73 pK units. We argue that the precision of our pK predictions in general is
comparable to that which can be obtained from conventional experimental methodologies for these quantities. Our results suggest that the presented molecular simulation
methodology may provide a robust and cost–efficient tool for solvent screening in the
design of post–combustion CO2 capture processes.<br>
Neural networks-based variationally enhanced sampling