Graphene oxide membranes: on the absence of "graphene capillaries", "ultrafast flow" rate and "ultraprecise sieving"

Abstract

The data by Q.Yang et al suggest absence of "ultrafast flow" of solvent across graphene oxide (GO) membranes. The "ultrafast flow" is result of using unrealistic geometrical model with close packed hole-free micrometer sized GO flakes providing only 0.1 % of area in each layer available for permeation. The data by Q.Yang et all demonstrate that at least 3-5 % of total layer area are available for permeation in real GO membranes due to holes between irregularly shaped flakes. At least 2-3 percent of area also needs to be added to the permeation cross section due to holes and cracks in GO flakes, especially abundant due to prolonged sonication of dispersions. Permeation of solutions mostly through pinholes penetrating tens of GO layers suggests that "graphene capillaries" are not required to explain water flow across the membrane. Taking into account realistic packing of GO flakes with holes between the flakes and across the flakes, 2-3 orders of magnitude shorter permeation path should be formed. Considering shorter zigzag permeation path between GO sheets and 5-10 % of area in each layer available for permeation due to holes across and between GO flakes, permeation rates across GO membranes can be explained by trivial diffusion. In absence of "graphene capillaries", the "ultra-precise sieving" related to "cutoff" value of 4.5 \AA{} (hydration diameter) provided by "graphene capillaries" has little meaning. As it is well known from earlier studies and now demonstrated by Q.Yang et al using their own samples the swelling of GO membranes is different for different solutions, depends on concentration of solutes and results in delamination in many solutions. Q.Yang et al provide controversial data for stability of their GO membranes in water citing use of surfactant to prevent dissolving. Oxidation state of GO membranes remains to be unknown due to incorrect analysis of XPS spectra.