Interaction between membrane and protein properties on flux decline during sterile microfiltration
Date
2010
Authors
Cutler, Hailey, author
Wickramasinghe, Sumith Ranil, advisor
Kipper, Matt J., committee member
Venayayamoorthy, S. Karan, committee member
Journal Title
Journal ISSN
Volume Title
Abstract
Microfiltration is widely used in industry to filter out particulate matter that contaminates or slows down the performance of the membrane. In the biopharmaceutical industry in particular, bacteria, microorganisms and viruses are filtered out using sterile microfiltration. Numerous studies have been conducted to further the understanding of flux decline due to protein fouling. Many times the operating conditions, the type of membrane and type of protein all interact to have an effect on protein fouling and flux decline. Normal-flow microfiltration experiments were conducted using uncoated polytetrafluoroethylene (PTFE) and polyvinylidene fluoride (PVDF) membranes, and PTFE and PVDF membranes coated with polyvinyl alcohol (PVA). Feed streams consisted of lysozyme, β-lactoglobulin and ovalbumin. The pH values of the solution were set at the isoelectric point of each of the proteins (11.0, 5.8, and 4.7 respectively). The experiments were operated with a feed pressure of 2 or 10 psi. Each of the proteins was tested at 0.1 and 2 g/L with uncoated PTFE. No flux decline was seen using 0.1 g/L, so 2 g/L was focused on for PVA coated PTFE, PVA coated PVDF and uncoated PVDF membranes. Protein fouling of the membrane was investigated by determining the variation of permeate flux versus filtrate volume and by analysis of Attenuated Total Reflection-Fourier Transform Infrared (ATR-FTIR) spectra and Field Emission Scanning Electron Microscopy (FESEM) images of unfouled membranes and membranes after microfiltration. Results indicate that the greatest amount of fouling occurs with ovalbumin. The order of most to least fouling was found to be ovalbumin> β-lactoglobulin>lysozyme. Fouling was more severe at the higher protein concentration (2 g/L) and feed pressure (10 psi) and seen only when filtering the solution through uncoated and PVA coated PTFE. Flux decline under these conditions was analyzed using classical pore blockage models. In general, flux decline was found to be caused by complete pore blocking. In the case of ovalbumin filtered through PVA coated PTFE, the flux decline was first caused by pore blockage and then later transitioned to cake filtration. The proteins which showed significant fouling conditions were looked at more closely by pre-filtering the protein solution. The goal of pre-filtration was to decrease any protein aggregates present in solution. This pre-filtration step was conducted with 0.2, 0.45 and 1 μm diameter pore sizes. The flux decline when pre-filtering the feed solution with 1 μm pores was equivalent to the filtration experiments without pre-filtration. The only significant decrease in flux was present when pre-filtering with the 1μm pores. Additional experiments were conducted using hemoglobin (Hb) at 2 g/L and 10 psi operating conditions. Previous literature had shown that using 1 μm pre-filtration, there was severe flux decline for uncoated and PVA coated PTFE. To follow up on these experiments, Hb was pre-filtered using 0.2 and 0.45 μm pre-filtration membranes and then filtered through uncoated and PVA coated PTFE. These experiments resulted in no flux decline. The Hb experiments verified the results from the ovalbumin and β-lactoglobulin experiments. All together, these results indicate that there is an interaction among membrane properties, protein properties, operating conditions and pre-filtration characteristics that determine whether fouling occurs and to what extent.
Description
Department Head: Stuart A. Tobet.