A Computational Model for High Speed Screening of Polymer Microstructures Kegang Wang, Martin E. Glicksman, Krishna Rajan* Department of Materials Science and Engineering, Rensselaer Polytechnic Institute, Troy, NY 12180-3590, USA Fax: (þ1) 518-276-8554; E-mail: rajank@rpi.edu Received: October 7, 2003; Revised: November 13, 2003; Accepted: November 24, 2003; DOI: 10.1002/marc.200300174 Keywords: microstructure; modeling; separation techniques; simulations; structure-property relations Introduction The development of reverse osmosis and ultrafiltration membranes is a classical example of how fundamental polymer chemistry was linked to processing in order to design membranes with specific functionalities. The key to the operation of such membranes was to control both the physico-chemical nature of the membrane as well as the pore size and more specifically the pore size distribution at the active surface of the membrane. Nearly three decades of work in this area and especially that pioneered by Sourirajan and co-workers [1] was driven by the discovery of a vast array of combinations of chemistries and processing conditions for numerous applications of membranes. The work done over the years in this field of the materials science of membrane development has identified numerous optimum parameters for separation ranging from desalina- tion to protein separation. In retrospect, it is interesting to note that combinatorial methods were implicitly used in the development of these membranes. Hence in our re- search program, we are now beginning to reexamine the wealth of data from these experiments to see if we can apply combinatorial and informatics strategies to help us accele- rate future development of new membrane materials. In this paper we explore one particular aspect of this initia- tive, namely the issue of pore size distribution in memb- ranes. The measurement of pore size distributions and its causes associated with the chemistry of the membrane material itself, and the processing parameters and their effect on solute separation and permeation rates has and continues to be a slow process. There are at present no established high-throughput/combinatorial experimenta- tion methods for the development of new membrane mater- ials. Hence, as a first step, we began to explore the use of computational strategies, which may aid in this high- throughput screening. The modeling of the physics of how an individual pore size evolves is itself not sufficient, but rather the modeling of how hundreds Summary: In this paper we outline a computational strategy that permits the modeling of simultaneous interactions between very large numbers of polymer aggregates. Using a multiparticle diffusion method, we have simulated the dynamics of phase coarsening which can serve as the basis of conducting ‘‘virtual’’ combinatorial experiments and high- throughput screening of pore size distributions for ultrafiltra- tion membranes. Particle size distributions for various volume fractions of aggregates calculated for two thousand polymer aggregates. Macromol. Rapid Commun. 2004, 25, 377–381 DOI: 10.1002/marc.200300174 ß 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Communication 377