Integral-skin formation in hollow fiber membranes for gas separations
Abstract
The morphologies of polymeric integrally-skinned asymmetric gas separation membranes are typically visualized as a thin selective skin region supported by a low resistance porous structure. Improvements in scanning electron microscopy (SEM) now allow for combinatorial analysis of this visualization with gas permeation measurements for previously reported ultra-thin defect-free hollow fiber membranes. The fibers were formed via a dry-jet, wet quench process with a spinning solution comprised of Matrimid polyimide and components of varying volatility. Depending on the formation conditions, the fibers displayed either defect-free skin layers or lower selectivity nodular skin morphologies. Under ideal conditions, defect-free skin thicknesses of 130 nm were characterized by O2, N2 and He permeation in conjunction with SEM studies. A fiber forming technique has allowed for the quick characterization of the skin layer via SEM analysis. The fiber forming technique, high-resolution SEM analysis and gas permeation measurements have allowed for a more complete understanding of defect-free skin formation. Typical solvent exchange techniques did not have a significant influence on the formation of the defect-free skin layer, although a critical point drying method was able to produce membranes with initial gas permeances twice those of conventionally dehydrated hollow fibers. Skin formation was found to primarily be influenced by the evaporation of volatile components from the nascent skin layer in the air gap. Phase separation of the polymer solution in the nascent skin layer was found to be detrimental to skin formation. The one-phase nascent skin layer is suggested to be kinetically hindered from phase separating due to the relative immobility of the polymer chains before immersion into the quench bath. The polymer chain mobility for different potential membrane forming dopes are compared using polymer physics models. Qualitative evidence suggests that lower molecular weight Matrimid samples may be formed into defect-free membranes if the initial dope is 40% wt. polymer, as suggested by the scaling of the previous defect-free membrane system by the Rouse model.