Monitoramento da hidrólise enzimática de bagaço de cana-de-açúcar por métodos espectrofotométricos

Abstract

The production process of 2nd generation ethanol (2G), using sugarcane bagasse, is a promising way to increase ethanol production without increasing the cultivation area. However, the process has been studied to become economically viable. This fact stems from several factors, among them: extraction of fermentative compounds from sugarcane bagasse, separation of enzymatic inhibitors and the great energy demand for the agitation of the mixture. In this process, sugarcane bagasse is used as a substrate, and must undergo the pretreatment process to expose the fibers and facilitate the hydrolysis of the cellulose into glucose. Although there are several ways of performing the hydrolysis of the pretreated bagasse, the enzymatic route reduces the amount of inhibitors produced by the traditional chemical routes, which adversely affect the fermentation process. On the other hand, the enzyme complex is expensive, thus increasing the cost of the process. Thus, the design of the reaction system and the definition of the process conditions must be analyzed to maximize the specific productivity of the enzyme. A non-conventional reactors system has been built in the Bioprocess Automation and Development Laboratory (LADaBio – UFSCar) to work around the issues discussed earlier. The reactionary system is complex and the attempt to use phenomenological models to describe it would make the problem intractable. Therefore, simplified kinetic models have been developed for this system. Due to the expected deviation between model and reaction, a monitoring system using spectroscopy in the near infrared and visible ultraviolet region, combined with the method of partial least squares and cross validation, was studied to monitor the reaction. Data from four assays performed were used to adjust mathematical models for predicting the concentrations of cellobiosis, glucose and xylose. As it is a complex system, it was concluded that the models generated were able to predict carbohydrate concentrations with great precision.