Kinetic modeling of the enzymatic hydrolysis of cellulose integrating the evolution of the substrate morphology.
Résumé
The biological production of ethanol from lignocellulose has been widely investigated for decades. During the SHF process (Separate Hydrolysis and Fermentation), the biomass, which is composed of cellulose, hemicelluloses and lignin, is converted into bioethanol following four main steps. After an appropriate pretreatment (step 1), the cellulose is hydrolyzed into glucose in the presence of three types of cellulolytic enzymes acting in synergy (step 2): endoglucanases (EG) which attack randomly the cellulose molecules and create new chain ends, cellobiohydrolases (CBH) which hydrolyse chain ends into cellobiose and β-glucosidases (BG) which convert cellobiose into glucose. The glucose is then converted into ethanol during a fermentation step (step 3). The alcohol is recovered by distillation (step 4) and incorporated into fossil fuel. Currently, the enzymatic hydrolysis of biomass is one of the main bottlenecks of the process, as it implies very complex mechanisms: long hydrolysis times (until 1 week) with enzyme deactivation, heterogeneity and modification of the substrate during the reaction, production of inhibitive coproducts, presence of lignin... In order to optimize the process at the industrial scale and reduce the enzyme loading, a model must be developed to predict the behaviour of the lignocellulosic substrate during enzymatic hydrolysis. However, even if a plethora of models has already been published, only few of them take into account the morphology of the cellulosic substrate and its evolution during the saccharification step. The morphological aspect of the cellulosic matrix is yet strongly linked to the global performances of the enzymes. As it is drastically changed during the reaction, it has to be integrated into an heterogeneous predictive model. The model should also distinguish the action of the different types of enzymes, although several authors consider only a global effect of the cocktail on the cellulose hydrolysis. In the model proposed in this article, the substrate is only composed of many chains of cellulose gathered into fibers. The fibers have a cylindrical shape and comprise both external and internal areas. The endoglucanases and the cellobiohydrolases are adsorbed onto the accessible area and produce cellobiose by shrinking the fiber, while the b-glucosidases form a complex with the soluble cellobiose to produce glucose. The individual action of each enzyme, the synergy between the enzymes and the product inhibitions are also taken into account. The evolution of the concentrations of glucose, cellobiose and free enzymes calculated with the model agree well with the experimental data on short times (30 min). It has still to be validated on much longer periods. In this case the deactivation of the enzymes with time will be added. This model can help to understand the mechanisms implied in enzymatic hydrolysis of cellulose, and will be next extended to real lignocellulosic substrates containing lignin. Lignin has indeed two main impacts on the hydrolysis: its presence decreases the accessibility of cellulases to cellulose and reduces the quantity of efficient enzymes because a part of them is adsorbed on lignin in a non productive way. These influences will be further studied.