(228b) Solvent Effects in Acid-Catalyzed Reactions of Biomass-Derived Oxygenates

Authors: 
Dumesic, J., University of Wisconsin-Madison
Mellmer, M. A., Bristol-Myers Squibb Co
Sanpitakseree, C., University of Minnesota
Demir, B., University of Wisconsin-Madison
Bai, P., University of Minnesota
Ma, K., University of Wisconsin-Madison
Walker, T., University of Wisconsin - Madison
Chew, A., University of Wisconsin
Li, H., Dalian Institute of Chemical Physics
Zhang, Z. C., Dalian Institute of Chemical Physics
Huber, G. W., University of Wisconsin-Madison
Van Lehn, R. C., University of Wisconsin-Madison
Neurock, M., University of Minnesota
We will present results for the Brønsted acid-catalysed conversions of ethyl tert-butyl ether, tert-butanol, levoglucosan, hydroxymethylfurfural (HMF), 1,2-propanediol, fructose, cellobiose, and xylitol were measured in solvent mixtures of water with three polar aprotic co-solvents: gamma-valerolactone; 1,4-dioxane; and tetrahydrofuran. We show that the rates of dehydration of reactants containing multiple hydroxyl groups are promoted in organic solvent mixtures with water, and the promotional effects for these reactions can be understood in terms of solvation effects for the acidic proton, reactants, and transition states for these catalytic reactions. Moreover, we show that the extent of promotion by the solvent increases as the number of vicinal hydroxyl or oxygen-containing groups in the reactant increases (i.e., as the reactant becomes more hydrophilic). This behaviour makes it possible to achieve high yields for the partial dehydration of highly functionalized biomass-derived reactants, such as carbohydrates. We utilize these solvent effects to achieve high yields for the production of HMF from fructose by promoting the conversion of fructose to HMF, while limiting the subsequent conversion of HMF to levulinic acid. We also present classical molecular dynamics simulations to explain these solvent effects in terms of three simulation-derived observables: (1) the extent of water enrichment in the local solvent domain of the reactant; (2) the average hydrogen bonding lifetime between water molecules and the reactant; and (3) the fraction of the reactant accessible surface area occupied by hydroxyl groups, all as a function of solvent composition. We develop a model, constituted by linear combinations of these three observables, that predicts experimentally determined rate constants as a function of solvent composition for the entire set of acid-catalyzed reactions.