Figure 3. (a)-(d) Scanning electron microscopy (SEM) surface images, (e)-(h) magnified SEM surface images, and (i)-(l) cross-sectional SEM images of the metal–organic framework membranes in this study. Here TFN-(Zr)-1 and TFN-(Zr)-2 denote thin-film nanocomposite membranes with UiO-66(Zr)-NH2 loadings of 0.01% and 0.03% (w/v) in 50 mL of n-hexane solution, respectively, and TFN-(Zr/Ti)-1 and TFN-(Zr/Ti)-2 denote thin-film nanocomposite membranes with UiO-66(Zr/Ti)-NH2 loadings of 0.01% and 0.03% (w/v) in 50 mL of n-hexane solution, respectively.
In addition, the AFM morphology images of the aforementioned TFC and TFN membranes were obtained to visualize more topographic details and the measured surface roughness values, viz. their mean (Ra) and root mean square (RMS, Rq) roughness (Figure 4, Table 1, and Supporting Information, Figure S10). We determined that the Ra values of the TFN membranes were higher than those of the pristine TFC membrane (Ra = ±25.4).35 Moreover, as the MOF nanoparticle loading increased, the Ra and Rq values of the TFN membranes decreased significantly, and that was attributed to the increase in the number of MOF nanoparticles that filled the ridge-and-valley structures. Furthermore, for the same MOF loading, the Ra and Rq values of the TFN-(Zr) membranes were relatively higher than those of the TFN-(Zr/Ti) membranes. That could be attributed to the surface of the UiO-66(Zr/Ti)-NH2 nanoparticles being relatively rougher than that of the UiO-66(Zr)-NH2 nanoparticles (Figure 1b). The increase in surface roughness after the incorporation of the MOF nanoparticles into membranes was highly beneficial and increased the exposed surface area for effective ion permeation.36-38 Thus, the precisely controlled symmetrical ridge-and-valley morphology and high surface roughness of the fabricated MOF membranes confirmed the merits of the adopted membrane fabrication strategy.