The scale-up of protein microfiltration based on laboratory studies of fundamental effects.

Membrane microfiltration has important applications in the processing of protein produced by the biotechnological industries. However, the utilisation of this technique is limited by the loss of membrane performance with time. The thesis addresses the problem of predicting membrane performance through initially at a laboratory scale and then translation to pilot scale using process feedstreams. There are three major developments. Firstly, the mechanisms responsible for the decline in membrane performance were examined using pure enzyme solutions and capillary pore membranes. Two mechanisms were identified through the analysis of filtration data and were shown to be concentration dependent. It was demonstrated that a low concentration of protein led to in-pore deposition, whereas higher concentrations of protein led to an increased tendency for the deposition to occur on the separating surface of the membrane. The use of atomic force microscopy provided visual conformation of these conclusions. Further studies were also performed to identify methods of maintaining or enhancing membrane performance. Secondly, the microfiltration of feedstreams from successive stages of a process that leads to the isolation of alcohol dehydrogenase from Baker's yeast (Saccharomyces cerevisiae) was studied. This demonstrated the impact of the feedstream complexity on microfiltration performance. It was demonstrated that after the second ammonium sulphate precipitation stage, improved recoveries were obtained. The yield of active enzyme provided the basis for a discussion on the application of microfiltration within the process. The last section of this thesis explores the development and prediction of membrane performance for the scale-up of membrane processes based on small scale tests and empirical data. The gel polarisation model and a transmembrane pressure relationship were used as a framework to predict membrane performance. These predictions were investigated using process feedstreams and pilot scale microfiltration equipment. Predictions generated using the transmembrane pressure relationship were demonstrated to successfully predict microfiltration performance, of unclarifled yeast homogenate, at a pilot plant scale