Preparation and properties of DMFC membranes from polymer-brush nanoparticles

Abstract

We describe a novel approach for the preparation of proton-conducting membranes using polymer brush nanoparticles (PBNPs). PBNPs can be assembled into the membranes directly, or the membranes can be prepared by first assembling porous scaffolds from unmodified NPs followed by surface modification with polymer brushes. In both cases, proton-conducting channels are formed in the interstitial spaces between the nanoparticles. First, proton-conducting membranes were prepared using silica nanoparticles (SNPs) surface-grafted with 40–400nm sulfonated polymer brushes, poly(3-sulfopropylmethacrylate), pSPM, and poly(4-styrenesulfonic acid), pSSA, grown via surface-initiated atom transfer radical polymerization (SI-ATRP). The membranes prepared from the NPs carrying longer polymer chains possessed polymer-like characteristics, compared to the stiffer membranes made with shorter polymer brushes. All these membranes showed comparable proton conductivities, with a maximum value of ~0.06S/cm at 98°C and 70% R. H. Second, we prepared proton-conducting pore-filled membranes. In this case, nanoporous colloidal crystals were first assembled from unmodified SNPs, followed by filling the pores with pSPM or pSSA brushes covalently attached to the pore surface using SI-ATRP. The membranes could be hydrated almost completely and did not swell. The proton conductivity was similar for both polymer-filled membranes, with a maximum value of ~0.02S/cm achieved at 30°C and 94% R. H. A sigmoidal dependence of the proton conductivity on the amount of sulfonic acid groups was found for pSPM pore-filled membranes. The proton conductivity remained relatively low at low degrees of sulfonation, increased rapidly around 50% sulfonic acid group content, and did not increase significantly after reaching ca. 75% sulfonic acid group content. We found that OCV for DMFC MEAs built using these membranes reached its maximum at 65% sulfonic group and decreased after that. We attribute this effect to the increased methanol cross-over, which was confirmed by methanol uptake measurements for the membranes.