(442e) Production of Gamma-Valerolactone From Corn Stover by Selective Extraction and Hydrogenation of Levulinic Acid

Martin Alonso, D. - Presenter, University of Wisconsin-Madison
Wettstein, S. G. - Presenter, University of Wisconsin-Madison
Bond, J. Q. - Presenter, University of Wisconsin-Madison
Dumesic, J. A. - Presenter, University of Wisconsin-Madison

Economic and technically viable conversion of lignocellulosic biomass into chemicals and fuels is an important challenge and several routes have been proposed using intermediate molecules, such as levulinic acid (LA) to maximize the yields. One of the main problems for these routes to be implemented at commercial scale is the high cost of purifying/recovering these intermediates, which are normally produced in dilute solutions of mineral acids. In this presentation we demonstrate the use of a family of high boiling point hydrophobic solvents, to extract LA selectively from water solutions of dilute acids, such as sulfuric acid, and to produce gamma-valerolactone (GVL) over bimetallic catalysts without the necessity to separate/purify the LA. The GVL produced can be used as chemical or fuel additive or it can be decarboxylated to produce biomass derived butene.

 The approach described here has 4 main advantages. First, the hydrophobic solvents do not extract sulfuric acid, which remains in the aqueous phase, and can be recycled. Second, LA can be hydrogenated to GVL in presence of the solvent over a bimetallic catalyst, and thus purification of LA is not necessary. Third, the high boiling point of the solvent allows for product recovery at the top of a distillation column without evaporating the solvent. Finally, the organic/aqueous partition coefficient for GVL is higher than for LA, and it is thus possible to increase the final concentration of GVL in the solvent by successive extraction-hydrogenation cycles.

 We have carried out this strategy at lab scale for 4 cycles. No sulfuric acid was detected in the organic phase which was recycled to the cellulose deconstruction reactor without further purification with minimal lost in LA yields. The final GVL concentration was increased from 0.4 M in the first cycle to 1.4 M after the 4th. Aspen simulations demonstrate that this increment in GVL concentration reduces the size of the distillation column and the energy required for distillation. Importantly, we demonstrate the aforementioned ideas using a feed produced from real biomass (corn stover) reporting stability for the catalyst during more than 300 h on stream. Solvent impurities in the GVL did not affect negatively the decarboxylation reaction over silica/alumina to produce butene.