Due to increasing population, excessive water usage and changing weather patterns, many regions in the world will face significant shortages in fresh water supply. One of the main ways to address this issue, at least in the coastal areas, is by constructing desalination facilities for producing drinking water by filtering out primarily salt and other dissolved and dispersed contaminants. Membranes, particularly reverse osmosis (RO) membranes are used in this process, and the filtration mechanism depends primarily on the morphology and the chemical structure of such semi-permeable membranes. Since desalination process is an energy intensive process, it is of the utmost importance to design such membranes to maximize selectivity and throughput simultaneously, to improve efficiency. This has been proven to be a difficult task to address using purely experimental methods, particularly due to the large number of potential membrane materials to be tested for maximum selectivity and throughput. Similar issues were also found in the use of other membranes, such as Nafion, used as a proton exchange membrane (PEM) for fuel cells. Specifically designed membranes can also be used for separating chiral molecules, highly useful in the pharmaceutical industry. Thus, a fundamental computational/theoretical model elucidating the basic mechanisms and events within the membranes would significantly help in the development of new membranes with enhanced efficiency.
Diffusion and transport through membranes is affected by a large number of parameters: the chemical moieties present in the membrane, the connectivity of polymer molecules in the membranes and how this produces free volume cavities with proper connectivity to facilitate molecular diffusion, the pressure/temperature the membrane is subjected to, surface properties of membranes, etc. These combine to create a truly multi-scale modeling problem of significant interest in different disciplines. In this work, by combining important elements from materials and computational sciences, we plan to address the following areas, although other relevant issues will be considered as they arise: (1) molecular architecture of the membrane as a function of component characteristics (such as chemical moieties present in the polymer chains, molecular weight distributions, and cross-link density), (2) free volume distribution arising from the molecular arrangement, (3) transport pathways in the membrane, for pure membranes as well as those with inclusions (nano-particles, zeolites, chitosan, etc.) as those inclusions modify the molecular architecture in a hierarchical scale, (4) calculation of diffusion parameters, and (5) optimization of membrane morphology for maximizing efficiency.
Our first step is to design and validate models for membrane filters. We will focus on coarse-grained models suitable for understanding phenomena relevant to membrane filter performance. Data for developing and validating these models come from laboratory experiments and molecular dynamics simulations. We will implement the models in our existing simulation codes, as well as extend our codes to handle features of the newly-developed models. Simulations will be used to determine the characteristics of very large numbers of potential membrane materials, and advanced data analysis techniques will be applied to understand the results and also to steer the search for new improved materials. Our goal is to use these models, simulations, and data analysis techniques to efficiently design new, highly selective and high throughput membrane for a variety of conditions and applications.