Operação ótima de reator para síntese enzimática de ampicilina com cristalização simultânea dos produtos

Abstract

Nowadays, industrial production of semi-synthetic penicillins requires low temperatures, organochloride solvents and yields a great amount of non-recyclable wastes. During the last decades, concerns about environmental impacts have increased, as well as the environmental legislation restrictions. Thus, the pursuit of cleaner routes has been encouraged. Enzymatic synthesis of these antibiotics, using penicillin G acylase (PGA) as biocatalyst, may be carried out at mild temperatures and pH, and is an environmentalfriendly route, alternative to the chemical synthesis. However, the low yields of the enzymatic process are still a drawback for their industrial implementation. An enzymatic semi-batch reactor using aqueous-precipitated medium is a promising approach to improve process efficiency. Yet, finding the optimal operation condition of the reactor is still a challenge, in order to make the enzymatic route economically competitive. This thesis addresses this issue, focusing on several aspects of the enzymatic synthesis of ampicillin. The comprehension of the reaction mechanism, still not a consensus in the literature, was improved especially with respect to the role of the beta-lactam nucleus during the formation of the acyl-enzyme intermediate, which has important consequences on the reactor operation. Diffusion in the biocatalyst pores and its influence on the pH profile within this micro-environment were assessed through computer simulations. Results indicate considerable diffusion resistances within the biocatalyst, yielding important pH profiles. The process complexity leaded to the use of simplified mechanistic or empirical kinetic models. Applying dynamic optimization (optimal control) techniques, feed profiles for the reactants were obtained. A simplified model for the integrated semi-batch reactor for enzymatic synthesis of ampicillin with product crystallization was used for this purpose. Different techniques of dynamic optimization provided qualitatively the same optimum heuristics for the process operation. Since simplified models were used in optimal open-loop control algorithms, theoretical feed policies may diverge from those that would be needed to maintain the track of the concentration profiles of the optimized reactor. Moreover, disturbances in the input variables might lead the system to a different course. Thus, on-line monitoring is essential in pilot plants or industrial reactors. Multivariate calibration using UV spectra was the basis for the development of a system of analysis via flow injection (FIA), with good results. Ampicillin synthesis assays using an industrial biocatalyst (Recordatti, Italy) were run in order to improve some simplified kinetic models. The re-estimated models were inserted in optimization algorithms, providing trajectories in accordance with the same heuristics previously obtained. Experimental results obtained after two runs in the integrated semi-continuous reactor put into evidence a mismatch between model responses and the real process. This difference may be explained by the extrapolation of the kinetic model with respect to the enzymatic reactor load. On the other side, using a biocatalyst wrapped with a secondary inert matrix might alter the microenvironment of the enzyme, especially with respect to pH. Using a model that takes into account the effect of pH on the kinetics seems to be important to allow the fine-tuning of the industrial reactor selectivity and productivity.