UC Berkeley

Furfural produced from the biomass-derived xylose may serve as a platform molecule for sustainable fuel production. The Brønsted acid-catalyzed dehydration of xylose to furfural is plagued by side reactions that form a set of soluble and insoluble degradation products, collectively known as humins, which reduce the yield of furfural. The formation of humins can be minimized by removal of furfural, either by steam stripping or by liquid-liquid extraction (LLE). However, both these techniques are very costly. The goal of this study was to demonstrate the feasibility of using pervaporation, a membrane process, to remove furfural as it is produced. A laboratory-scale reactor/membrane system was designed, built, and tested for this purpose and its performance for furfural production was compared with that achieved by carrying out the reaction with and without furfural extraction by LLE. Furfural production assisted by pervaporation (with a commercially available membrane or a triblock copolymer membrane) or LLE produced comparable amounts of furfural, and more than could be achieved by reaction without extraction. A model of the reaction kinetics and the rate of furfural extraction was fit to the pervaporation- and LLE-assisted furfural production data and was used to predict the performances of these processes at near-complete xylose conversion. Pervaporation is shown to have two advantages over LLE: pervaporation extracts a greater fraction of the furfural produced and the furfural concentration in the permeate phase is significantly higher than that present in the extractant phase obtained by LLE. It is noted that further improvement in the separation of furfural from the aqueous phase where it is produced can be achieved by using a more-permeable, thinner pervaporation membrane of larger area, and by operating the membrane at the reaction temperature.