2006 AIChE Annual Meeting
(211g) Mesoporous Ceramic Membranes for Non-Aqueous Separations: Surface Modification and Solvent Permeability
Authors
Borrowing from silane grafting techniques developed for gas separation membranes, commercial 1 kDa titania (Sterlitech) and 5 nm alumina (Pall Fluid Dynamics) membranes were functionalized with octyltrichlorosilane (Cl3SiC8H17) and its fluorinated analog, trichloro(1H,1H,2H,2H-perfluorooctyl)-silane (Cl3SiC2H4C6F13). Using this approach, the surface thickness of silane is similar (~1.1 nm) while the tail chemical composition is different. Fluoroalkylsilane and alkylsilane modified membranes are referred to as C8F and C8H, respectively. Modified membranes were characterized by energy dispersive x-ray spectroscopy (EDS) and carbon dioxide and nitrogen permeability at 298 K and 0.1 bar permeate pressure. In the C8H membranes, gas permeability measurements revealed pore opening with increasing pressure, consistent with tail mobility. Comparatively, gas permeability in the C8F analog was an order of magnitude lower due to the bulkier (more rigid) fluorinated tails. These results are consistent with recent gas separation studies.
Unique liquid permeability behavior was observed in the modified membranes, depicting the interplay between solvent, membrane, and surface-modifier properties. As expected, (i) the C8F membranes were impermeable to water, (ii) hexane had the greatest permeability through the C8H membranes and (iii) liquid carbon dioxide, which is fluorophilic, had the greatest permeability in the C8F membranes. However, the permeability of all solvents were similar in the 5 nm C8H membrane (km = 1.5 to 1.8x10-15 m). This is in contrast to the native 5 nm alumina membrane where, for instance, water permeability was an order of magnitude greater than liquid carbon dioxide. Similar results were observed for 1 kDa titania. We hypothesize that grafting the ceramic surfaces shields s the strong solvent-surface interactions that influence solvent-dependent transport behavior in the native membranes. For instance, based on complete silane surface coverage (79% by area assuming a flat surface), calculated Hamaker constants were reduced from approximately 14.9x10-20 J for native alumina to 8.1x10-20 and 5.8x10-20 for C8H and C8F, respectively. This reduction in the Hamaker constants denotes a less attractive solvent-membrane interaction in the surface modified membranes, which aids permeability. However, a complete depiction of the effects of surface modification on permeability must also include solvent-silane interactions, tail mobility, the effective silane thickness, and the extent of surface modification, which are dependent on ceramic and solvent properties, and the pressure driving force. Ultimately, the ability to manipulate and enhance solvent permeability through membrane selection and surface modification may be used to increase liquid-phase separation performance and/or target specific flow characteristics.