(27c) The Stability of Levulinic Acid During Acid-Catalyzed Glucose Dehydration

Glucose, 5-hydroxymethylfurfural (HMF) and levulinic acid are sequentially formed during acid-catalyzed hydrolysis of cellulose. Similarly, the acid-catalyzed conversion of glucose, fructose or HMF leads to the formation of levulinic acid. Apparently levulinic acid is stable at reaction conditions. It is produced in a 1:1 ratio with formic acid, as would be expected from stoichiometry. Concurrent with the generation of levulinic acid, the acid-catalyzed conversion of cellulose, cellobiose, glucose, fructose or HMF also produces undesired solids known as humins. The aldol addition and condensation of 2,5-dioxo-6-hydroxyhexanal (DHH, an intermediate produced from HMF) with aldehydes and ketones that are available in the reacting solution is a predominant pathway for humin formation. DHH is presumed to be highly reactive because it is present in very low concentrations. In support of this proposed pathway for humin formation, it has been shown that when HMF is used as the reactant, its furan ring is incorporated into the humins. Similarly, if benzaldehyde is added to the reacting solution, its benzene ring is incorporated in the humins.

Given that DHH is highly reactive toward aldehydes and ketones, one might expect it to similarly react with the ketone group of levulinic acid, leading to the incorporation of levulinic acid in the humins that form. Experiments were conducted show that this does not occur, even when excess levulinic acid is added at the very start of the reaction. Similarly, self addition/condensation was not observed when levulinic acid alone was processed with sulfuric acid. The first step in either of these aldol addition processes would be the protonation of the ketone group of levulinic acid. Ab-initio G3(MP2, CCSD(T)) calculations were used to study that protonation step. The results show that a cyclic cation is much more favorable than a straight chain cation. This may explain why levulinic acid is stable relative to aldol addition/condensation processes.

Being acids, both levulinic acid and its co-product, formic acid, can also form esters with available hydroxyl groups. G3 calculations for the formation of esters by each of the acids with HMF, DHH and glucose indicate free energy changes of the order of 20 kJ/mol. These reactions are expected to be reversible, suggesting that the large excess of water may limit the amount of ester that forms. HPLC analysis of the products of separate experiments using levulinic acid as both the solvent and the acid did reveal the presence of additional products. These were not identified, but are believed to be esters of levulinic acid.

Finally, it is known that levulinic acid can isomerize and dehydrate forming products that include 5-hydroxy gamma-valerolactone, alpha-, beta-, and gamma-angelica lactones, enols and diols. Previous work suggests that enols are the intermediate precursor to the formation of the lactones. Our G3 calculations do not support this mechanistic suggestion, at least not in aqueous solution. Instead, it appears that 5-hydroxy gamma-valerolactone will form much more easily via the cyclic cation of levulinic acid. The free energy changes for formation of the three angelica lactones are much smaller when they form via dehycration of 5-hydroxy gamma-valerolactone than via enols. Extended experimental studies do provide evidence for the formation of these intermediates, but the rates are very much slower than the rate of dehydration of glucose or the rate of hydration of HMF. This then explains the observed stability of levulinic acid at cellulose hydrolysis conditions.