(605c) Combinatorial and High Throughput Membrane Synthesis and Testing Conference: AIChE Annual MeetingYear: 2009Proceeding: 2009 AIChE Annual MeetingGroup: Separations DivisionSession: Characterization and Simulation of Novel Membranes and Separations Time: Thursday, November 12, 2009 - 3:57pm-4:18pm Authors: Zhou, M., Rensselaer Polytechnic Institute Krein, M., Rensselaer Polytechnic Institute Morkowchuk, L., Rensselaer Polytechnic Institute Breneman, C., Rensselaer Polytechnic Institute Kilduff, J., Rensselaer Polytechnic Institute Belfort, G., Rensselaer Polytechnic Institute Membrane filtration is widely used in the pharmaceutical and biotechnology industries for applications such as protein recovery or removing bacteria and other impurities, improving product yields and reducing maintenance costs. During membrane filtration processes, membrane fouling is inevitable, which significantly reduces membrane performance (lowers permeate production rates), decreases design reliability, and increases capital and operating costs. Hence, an efficient method for developing and selecting low fouling synthetic membranes that are easily cleaned is necessary. We have developed a new method for synthesis and screening of customized foulant-resistant surfaces by combining a high throughput platform (HTP) approach together with our patented photo-induced graft polymerization (PGP) method, to allow facile modification of commercial poly(aryl sulfone) membranes. Using a library of 66 monomers, the HTP-PGP method quickly created candidate surfaces for six different feed solutions. New membrane surfaces that performed better than the as-received membranes for protein filtration and water purification were identified. Available data were then used in concert with traditional and electron density-based transferable atom equivalent (TAE) molecular descriptors to develop quantitative structure-efficacy relationship (QSER) models of membrane modification efficacy. Descriptor selection and model building were accomplished using part of the monomer-protein data, through a genetic algorithm/partial least squares approach. The remaining data served as a test set for the QSER models. This technique can bring new mechanistic insights to the performance of surfaces synthesized from the available library, and predict the behaviors of monomers not included. The combination of the HTP-PGP approach and QSER models has the potential to improve the design of new successful customized foulant-resistant surfaces, by combining the prediction power and the rapid, inexpensive, fast, simple, reproducible screening of the two methods.