Solvent Extraction of Levulinic Acid from Its Aqueous Solution: A Monte Carlo Simulation Study

Abstract

A GEMC–NPT simulation study of the liquid–liquid extraction of levulinic acid (LA), a keto–acid, from its aqueous solution via six organic solvents has been performed at 313.15 K and 101.325 kPa to identify the optimal solvent. Continuous fractional component Monte Carlo approach (by using Brick–CFCMC) to enable efficient sampling of dense coexisting liquid phases via particle transfer moves for chemical equilibrium has been adopted. The solvent performance indicators (SPIs) are distribution coefficient (K_D), separation factor (S), and Gibbs free energies of transfer (ΔG_trans) for LA and water from the aqueous to the organic solvent-rich phase. Based on SPIs, ethyl acetate is the optimal solvent, and benzene, toluene, and xylene are ineffective. The molecular-level structure resulting from the complex interplay of interactions present has been investigated by computing the center of mass (COM)–COM radial distribution functions (RDFs) and their corresponding number integrals (NIs) in both the coexisting phases. The NIs from these RDFs for LA–LA, LA–water, solvent–water, and water–water molecules in the organic solvent-rich phase exhibit trends that are correlated with those in the SPIs for the solvents. Extent of hydrogen bonding between the hydrogen H9 in the carboxylic acid group of LA with that of the oxygen atom of the solvent, and with the oxygen O_Wof water is investigated via site–site intermolecular RDFs. The NIs from carboxylic acid group carbonyl oxygen O7 of LA–O7 RDFs including the first two peaks agree with the trends in K_D and ΔG_trans of LA for ethyl acetate, n-octanol, and 2-heptanone. Further, NI values from H9–O_W RDFs including the first and second coordination shells show trends in agreement with ΔG_trans of water.