Macrovoid Defect Growth during Evaporative Casting of Polymeric Membranes (original) (raw)
Macrovoid (MV) formation is a significant problem in evaporatively cast polymeric membranes. MVs are large, elongated or teardrop-shaped pores (~10−50 µm) that can impair membrane structural integrity. Although MVs have been extensively studied, there is no general agreement on the mechanisms governing MV growth. Recently, our research group has formulated the solutocapillary convection (SC) hypothesis, which contends that MV growth involves three principal forces: a Marangoni force generated by surfacetension gradients within the MV interface, a viscous drag force, and a gravitationally induced body force. Two sets of complementary experiments were conducted to test the SC hypothesis. Ground-based videomicroscopy flow-visualization (VMFV) was utilized to measure the flow velocities at the MV-casting solution interface and deep within the casting solution. The measurements were performed with casting solutions containing 10 wt% cellulose acetate (CA), 30 wt% H 2 O, 60 wt% acetone, and 200ppm TiO 2 particles for flow visualization, and the surface tension was controlled by surfactant addition. Qualitatively, the experiments indicated that MV growth occurs in three distinct phases: (1) a very rapid initial growth period, (2) a much slower growth phase, and (3) absorption of selected MVs into the expanding demixed region. The presence of tracer particles inside the MVs suggests the presence of a convective flow, which transfers the particles from the bulk solution to the MV interior. Although the VMFV experiments did not establish any surfactant effect on the interfacial velocities, a statistically significant effect on the MV number density was observed. In the second set of experiments, membranes were cast aboard a KC-135 aircraft under 0-g and 2-g conditions. Despite careful attention to the design and fabrication of the membrane casting apparatus (MCA), several problems were encountered, the most significant of which was the contamination of the casting solution by the activated carbon particles used for solvent absorption. Despite these difficulties, SEM analysis of uncontaminated membrane samples indicated that the MV morphology was strongly influenced by the solvent-nonsolvent ratio. However, dependence of MV size and number density on the magnitude of the buoyancy force could not be established since (in many cases) the MVs penetrated through the entire thickness of the cast membrane. Based upon the insights obtained from these experiments, a new MCA has been designed, which incorporates wider casting wells, deeper recesses for casting thicker membranes, and better isolation of the activated carbon. The new MCAs will be used in an upcoming KC-135 flight, and should enable complete quantitative evaluation of the SC hypothesis.