Unexpectedly High Propylene/Propane Separation Performance of Asymmetric Mixed-Matrix Membranes through Additive-AssistedIn SituZIF-8 Filler Formation: Experimental and Computational Studies

Abstract

Zeolitic-imidazolate framework-8 (ZIF-8), composed of a zinc center tetrahedrally coordinated with 2methylimidazolate linkers, has garnered extensive attention as a selective filler for propylene-selective mixed-matrix membranes (MMMs). Recently, we reported an innovative and scalable MMM fabrication approach, termed "phase-inversion in sync with in situ MOF formation" (PIMOF), aimed at addressing the prevailing challenges in MMM processing. In this study, we intend to investigate the effect of additives, specifically sodium formate and 1,4-butanediol, on the modification of ZIF-8 filler formation within the polymer matrix in order to further improve the separation performance of the asymmetric MMMs prepared by the PIMOF. Remarkably, MMMs prepared with sodium formate as an additive in the coagulation bath exhibited an unprecedented C 3 H 6 /C 3 H 8 separation factor of 222.5 ± 1.8 with a C 3 H 6 permeance of 10.1 ± 0.3 GPU, surpassing that of MMMs prepared without additives (a C3 separation factor of 57.7 ± 11.2 with a C3 permeance of 22.5 ± 4.5 GPU). Our computational work complements the experimental investigation by studying the effect of ZIF-8 nanoparticle size on the specific surface interaction energy and apertures of ZIF-8. Calculations indicate that by having smaller ZIF-8 nanoparticles, stronger interactions are present with the polymer affecting the aperture of ZIF-8 nanoparticles. This reduction in aperture size is expected to improve selectivity toward propylene by reducing the permeability of propylene. These results represent a significant advancement, surpassing the performance of all previously reported propylene-selective MMMs and most high-quality polycrystalline ZIF-8 membranes. The notably enhanced separation performance primarily arises from the formation of exceedingly small ZIF-8-like particles with an amorphous or poorly crystalline structure, corroborated by our computational work.