The origins of microstructure in phase inversion coatings or membranes: Snapshots of the transient from time-sectioning cryo-SEM

Time-sectioning cryogenic scanning electron microscopy (cryo-SEM) is a unique method of visualizing how the microstructure of liquid coatings evolves during processing. Time-sectioning means rapidly freezing (nearly) identical specimens at successively later stages of the process; doing this requires that coating and drying be well controlled in the dry phase inversion process, and solvents exchange likewise in the wet phase inversion process. With control, frozen specimens are fractured, etched by limited sublimation, sputter-coated, and imaged at temperatures of ca −175°C.; The coatings examined were of cellulose acetate, of high and low molecular weights, and polysulfone in mixed solvents and nonsolvents: acetone and water with cellulose acetate undergoing dry phase inversion; and tetrahydrofuran, dimethylacetamide, ethanol with polysulfone undergoing dry-wet phase inversion. All coatings, cast on silicon substrates, were initially homogeneous. The initial compositions of the high and low molecular weight cellulose acetate ternary solutions were “off-critical” and “near-critical”, respectively, connoting their proximities to the critical or plait point of the phase diagram. The initial composition of the polysulfone quaternary solution was located near the binodal of the pseudo-ternary phase diagram. It appeared that as the higher molecular weight cellulose acetate coating dries, it nucleates and grows polymer-poor droplets that coalesce into a bicontinuous structure underlying a thin, dense skin. Bicontinuity of structure was verified by stereomicroscopy of the dry sample. The lower molecular weight cellulose acetate coating phase-separates, seemingly spinodally, directly into a bicontinuous structure whose polymer-rich network, stressed by frustrated in-plane shrinkage, ruptures far beneath the skin in some locales to form macrovoids. When, after partial drying, the polysulfone coating was immersed in a bath of water, a nonsolvent, it appeared to swell in thickness as it phase-separates. A dense skin, thinner than a micron, appeared to overlie a two-phase substructure that is punctuated with pear-shaped macrovoids. At early immersion times, this substructure is visibly bicontinuous or open-celled near the bath-side, and dispersion-like (droplets dispersed in a polymeric matrix) or closed-celled near the substrate-side. Moreover, in the bicontinuous regions, length-scales of the individual phases seem to increase across the coating thickness from the bath-side to the substrate-side. After prolonged immersion, the substructure, excluding the macrovoids, is entirely bicontinuous. The bicontinuity presumably results from a combination of spinodal decomposition and nucleation and growth plus coalescence. Quite strikingly, macrovoids are present exclusively in regions where phases are bicontinuous, and are absent where droplets are dispersed in the polymeric matrix. Evidence suggests that macrovoids result from an instability caused by a progressive rupture of polymer-rich links deeper and deeper beneath the skin, aggravated by stress localization in the rupturing network and a buildup of pressure in the polymer-poor phase (the pore space), as suspected by Gröbe and Meyer in 1959.