Membrane based solvent extraction of vanillinin hollow fiber contactors

Vanillin (4-hydroxy-3-methoxybenzaldehyde), the major component of natural vanilla, is widely used in food, beverage, cosmetic and pharmaceutical industries. Since natural vanillin, extracted from the pods of the tropical vanilla orchid, features both high price and limited supply, chemically synthesized vanillin occupies more than 99% of the market share. The increasing concern for health and nutrition stimulates a worldwide demand for “natural” vanillin. The bioconversion of various substrates (eugenol, isoeugenol, ferulic acid) into vanillin becomes more and more attractive [1]. Indeed, according to current legislation, the products of microbial and enzymatic conversion can be labelled as “natural”. The biotransformation of ferulic acid into vanillin was extensively studied in our research group by using a Pseudomonas sp. strain, grown on a rich medium and successively incubated in saline buffer in the presence of ferulic acid. The maximum concentration of vanillin obtained from 2000 mg/L ferulic acid in the bioconversion broth was 1280 mg/L, corresponding to a molar yield of 88%. However, the productivity is limited both by product inhibition, due to the high toxicity of vanillin on bacterial cells, and by the oxidation or reduction of vanillin to the corresponding acid or alcohol performed by unspecific dehydrogenases. Both effects could be avoided by in-situ removal of vanillin. Adsorption as well as pervaporation have been proposed for this purpose. In this work membrane based solvent extraction [2] was investigated. Various solvents were tested measuring the partition coefficients for vanillin and other relevant compounds. Good properties were shown by ethyl and n-butyl acetate. These compounds showed high affinity for vanillin (partition coefficients 17 and 21 respectively) and low affinity for ferulic acid (partition coefficients 0.2-0.4). Extraction experiments were performed by a hollow fibre module, the feed flowed through the lumen while the solvent flowed in the shell side so that, owing to the membrane hydrophobicity, the interface between the two phases was immobilized at the inner pore mouth. The kinetics of mass transfer was investigated in batch experiments, measuring the feed and extract concentration of vanillin and ferulic acid vs. time. The overall mass transfer coefficients were calculated from the data and compared with theoretical values. In any case the feed side mass transfer resistance represents the major components of the overall resistance. Moreover, the coupling of the membrane contactor with the bioreactor is discussed.