1 Food Production Engineering, Department of Systems Biology, Technical University of Denmark2 Department of Systems Biology, Technical University of Denmark3 CHEC Research Centre, Department of Chemical and Biochemical Engineering, Technical University of Denmark4 Department of Chemical and Biochemical Engineering, Technical University of Denmark5 National Food Institute, Technical University of Denmark6 Technical University of Denmark7 Danisco AS8 Danisco AS
The aim of this study was to optimize production of MAG by lipase-catalyzed glycerolysis in a tert-pentanol system. Twenty-nine batch reactions consisting of glycerol, sunflower oil, tert-pentanol, and commercially available lipase (Novozym®435) were carried out, with four process parameters being varied: Enzyme load, reaction time, substrate ratio of glycerol to oil, and solvent amount. Response surface methodology was applied to optimize the reaction system based on the experimental data achieved. MAG, DAG, and TAG contents, measured after a selected reaction time, were used as model responses. Well-fitting quadratic models were obtained for MAG, DAG, and TAG contents as a function of the process parameters with determination coefficients (R2) of 0.89, 0.88, and 0.92, respectively. Of the main effects examined, only enzyme load and reaction time significantly influenced MAG, DAG, and TAG contents. Both enzyme amount and reaction time showed a surprisingly nonlinear relationship between factors (process parameters) and responses, indicating a local maximum. The substrate ratio of glycerol to oil did not significantly affect the MAG and TAG contents; however, it had a significant influence on DAG content. Contour plots were used to evaluate the optimal conditions for the complex interactions between the reaction parameters and responses. The optimal conditions established for MAG yield were: enzyme load, 18% (w/w of oil); glycerol/oil ratio, 7∶1 (mol/mol); solvent amount, 500% (vol/wt of oil); and reaction time, 115 min. Under these conditions, a MAG content of 76% (w/w of lipid phase) was predicted. Verification experiments under optimized reaction conditions were conducted, and the results agreed well with the range of predictions.
Journal of the American Oil Chemists' Society, 2006, Vol 83, Issue 1, p. 27-33