Stochastic reconstruction of mixed-matrix membranes and evaluation of effective permeability

Reconstructing the microstructure of polyimide–silicalite mixed‐matrix membranes and their particle connectivity using FIB‐SEM tomography

“…The microstructural descriptors were extracted from primary data that consisted of 2D digital images of planar cuts through the materials. The enhanced effective permeability of three samples of the MMMs differing in filler volume fractions was likely caused by numerous contacts between silicalite particles in the MMMs (Čapek et al ., ). However, the fact that no direct microstructural information from 3D space was available limited the reliability of reconstructed 3D microstructures, particularly in relation to particle clustering.…”

“…The purpose of this work is to carry out the FIB‐SEM reconstruction of the three MMMs that were studied in the previous work (Čapek et al ., ). Since the reconstruction method yields direct 3D microstructural information, we reconsider the hypothesis on particle clustering being the cause of elevated effective permeability.…”

“…The reconstructed membranes are thoroughly characterized by evaluating microstructural descriptors and compared with those reconstructed stochastically. In addition, the effective permeability computed via simulation of a random walk is compared to its experimental counterparts and to the values predicted for stochastic replicas of the MMMs (Čapek et al ., ). To the best of our knowledge, there is no comprehensive study combining the FIB‐SEM reconstruction method with the evaluation of effective permeability in the field of MMM characterization.…”

“…Many reconstruction methods are capable of reproducing volume (3D) images of material microstructures: X‐ray computed microtomography (Wildenschild & Sheppard, ; Veselý et al ., ; Bultreys et al ., ; Latief et al ., ), focused‐ion beam scanning electron microscope tomography (FIB‐SEMt) (Wilson et al ., ; Munroe, ; Prill et al ., ; Joos et al ., ), electron tomography associated with high‐resolution transmission electron microscopes (Novák et al ., ; Ercius et al ., ), confocal laser scanning microscopy (Shah et al ., ), stochastic methods constrained by various correlation functions (Jiao et al ., ; Zachary & Torquato, ; Čapek et al ., , ), methods for spatially nonstationary disordered media (Tahmasebi & Sahimi, , , ), process‐based methods (Øren & Bakke, ) and methods based on Boolean models (Thovert et al ., ). The method used is determined by the chemical and physical nature of the phases constituting the material, the number of phases and the size of objects of interest.…”

“…When the material consists of a continuous matrix and dispersed particles, connectivity of the particulate phase depends on contacts between particles, which is a microscopic detail. In our previous work (Čapek et al ., ), we studied properties of such materials, particularly MMMs made of a polyimide matrix and silicalite‐1 particles. The materials in that study were reconstructed stochastically using simulated annealing constrained by microstructural descriptors, specifically, the two‐point probability function for the polyimide phase and the lineal‐path functions for both phases.…”