Encapsulation of yeast for efficient 2nd generation bioethanol production: Finite element modeling of concentration profiles in encapsulated yeast

Conference poster, 2011

With the increasing energy demands and pressures on the environment, 2nd generation ethanol, produced from lignocellulosic raw materials, appears as an economically and environmentally viable alternative to oil-based fuels. The lignocellulosic materials have a recalcitrant structure, and must be pretreated before hydrolysis to fermentable sugars. Unfortunately, many inhibitory compounds are formed during this process. The inhibitory properties of lignocellulosic fermentation media usually leads to reduced rates of ethanol production or even complete absence of fermentation. Encapsulation of yeast cells in 3-4 mm large membranes of Ca-alginate and chitosan has been shown to improve the inhibitor tolerance of the yeast and lead to improved production rates. A likely explanation to this is that the outer layers of cells can detoxify some of the inhibitors, thereby protecting the cells situated deeper in the capsule. Another issue with lignocellulose materials is the sometimes high pentose content. Xylose is the most abundant pentose. Recombinant xylose-fermenting Saccharomyces cerevisiae strains have been constructed, but the xylose uptake rate is hampered by competitive inhibition by glucose. Our hypothesis is that concentration gradients in the capsules may cause favourable conditions for simultaneous uptake and metabolism of both glucose and xylose even in the presence of lignocellulosic inhibitors. To investigate this we simulated the concentration profiles for glucose, xylose, furfural and HMF by finite element modeling using COMSOL Multiphysics 4.1. An example of concentration profiles is shown in figure 1. Xylose consumption was found to always be improved by encapsulation of cells. For glucose, encapsulation may be beneficial if the combined inhibition effect is strong enough. The models formulated in this project were validated by comparison to pulished experimental values. Generally, the models fitted the experimental data reasonably well, after an adaptation of the maximum glucose consumption rate. Additional experimental work to validate these simulation results are underway.