Functionalized KIT-6/Polysulfone Mixed Matrix Membranes for Enhanced CO2/CH4 Gas Separation

The development of mixed matrix membranes (MMMs) for effective gas separation has been gaining popularity in recent years. The current study aimed at the fabrication of MMMs incorporated with various loadings (0–4 wt%) of functionalized KIT-6 (NH2KIT-6) [KIT: Korea Advanced Institute of Science and Technology] for enhanced gas permeation and separation performance. NH2KIT-6 was characterized by field emission scanning electron microscope (FESEM), X-ray diffraction (XRD), Fourier transform infrared (FTIR), and N2 adsorption–desorption analysis. The fabricated membranes were subjected to FESEM and FTIR analyses. The effect of NH2KIT-6 loading on the CO2 permeability and ideal CO2/CH4 selectivity of the fabricated membranes were investigated in gas permeation and separation studies. The successfulness of (3-Aminopropyl) triethoxysilane (APTES) functionalization on KIT-6 was confirmed by FTIR analysis. As observed from FESEM images, MMMs with no voids in the matrix were successfully fabricated at a low NH2KIT-6 loading of 0 to 2 wt%. The CO2 permeability and ideal CO2/CH4 selectivity increased when NH2KIT-6 loading was increased from 0 to 2 wt%. However, a further increase in NH2KIT-6 loading beyond 2 wt% led to a drop in ideal CO2/CH4 selectivity. In the current study, a significant increase of about 47% in ideal CO2/CH4 selectivity was achieved by incorporating optimum 2 wt% NH2KIT-6 into the MMMs.

In view of the drawbacks of polymer membranes, which limit the application of polymer membranes in gas separation, development of mixed matrix membranes (MMMs) for gas separation applications has been gaining popularity among researchers in recent years. MMMs are commonly formed by incorporating inorganic fillers into the polymer matrix. Among the various inorganic fillers,

Preparation of the Membranes
Preparation of NH 2 KIT-6/PSF MMMs (which were incorporated with functionalized KIT-6) were conducted following previously reported procedures with some modifications [11,21]. A desired loading (0-4 wt%) of NH 2 KIT-6 was added to 10 mL of THF, followed by ultrasonication for 30 min. The PSF pellet was added and dissolved into the solution for 18 h. Then, the solution was sonicated for 30 min to remove air bubbles, and subsequently was cast on a glass plate using a casting blade with a gap of 200 µm. The glass plate was covered and left for 3 days at room temperature to ensure complete solvent evaporation. The prepared membranes were peeled off and stored in desiccators.

Characterizations of KIT-6, NH 2 KIT-6, and the Membranes
The morphology of the NH 2 KIT-6 filler was analyzed by FESEM (VPFESEM, Zeiss Supra55 VP). NH 2 KIT-6 were subjected to XRD (X'Pert 3 Powder & Empyrean, PANalytical) scanning for crystalline structure study. Functional groups in KIT-6 and NH 2 KIT-6 were determined by FTIR (Perkin Almer, Frontier). The pore characteristics of KIT-6 and NH 2 KIT-6 were analyzed using N 2 adsorption-desorption analysis (TriStar II 3020 V1.04) with liquid nitrogen at 77 K. The specific surface area of the sample was calculated by using the Brunauer-Emmett-Teller (BET) method. The mesopore size distribution was determined by using the Barrett-Joyner-Halenda (BJH) method. The morphology of the fabricated membranes was analyzed by FESEM (VPFESEM, Zeiss Supra55 VP). The membranes were also subjected to FTIR analysis (Perkin Almer, Frontier).

Gas Permeation and Separation Studies
The membrane was sealed in a membrane gas cell that was connected to the membrane gas permeation and separation test system. CO 2 or CH 4 gas with a purity of 99.99% was fed separately to the membrane gas permeation and separation system. The gas permeation was performed at 25 • C. The pressure difference was regulated to be 5 bar or 7 bar. The permeate flow was measured using a bubble flow meter. The gas permeability, P, was calculated using Equation (1): where l is the membrane thickness, N is the permeate flow, P f is the feed pressure, P p is the permeate pressure, and A is the membrane area. The ideal selectivity, α of the membrane was calculated as the ratio of CO 2 gas permeance to CH 4 gas permeance.
Each gas permeation measurement was repeated three times.

Results and Discussion
3.1. Characterizations of NH 2 KIT-6 Figure 1 shows the FESEM image of NH 2 KIT-6. The FESEM image of NH 2 KIT-6 revealed the typical rock-like morphology. Figure 2 shows the XRD pattern of NH 2 KIT-6, which exhibits peak diffractions of about 1.1 • at 2θ. This shows that the samples had the ordered mesostructure with a three-dimensional cubic Ia3d symmetry [22]. The XRD pattern of the NH 2 KIT-6 sample in the current project is in agreement with the XRD pattern for KIT-6 reported by Ayad et al. [22].
The FTIR spectra of KIT-6 and NH 2 KIT-6 are shown in Figure 3. The characteristic band at 3464 cm −1 indicates the stretching vibration of hydrogen bonding from silanol group y(≡Si−OH) [23]. Furthermore, the characteristic band at 1640 cm −1 indicates the O−H bending vibration mode. The anti-symmetric and symmetric stretching vibrations of Si−O−Si groups are observed at characteristic bands at 1083 cm −1 and 804 cm −1 , respectively [24]. It is interesting to observe the characteristic band at 1459 cm −1 , which is present in the FTIR spectra of NH 2 KIT-6 but is absent in the FTIR spectra of KIT-6. This characteristic band at 1459 cm −1 indicates the appearance of N−H bonds in NH 2 KIT-6 and hence prove the successful amine-functionalization of KIT-6 [20,23,25]. In addition, the characteristic band at around 2926 cm −1 , which indicates the C−H stretching vibration of the organosilane, is observed in the FTIR spectra of NH 2 KIT-6 but is not found in the spectra of KIT-6 [23]. This further proves the effective functionalization of KIT-6.  The FTIR spectra of KIT-6 and NH2KIT-6 are shown in Figure 3. The characteristic band at 3464 cm −1 indicates the stretching vibration of hydrogen bonding from silanol group y(≡Si−OH) [23]. Furthermore, the characteristic band at 1640 cm −1 indicates the O−H bending vibration mode. The anti-symmetric and symmetric stretching vibrations of Si−O−Si groups are observed at characteristic bands at 1083 cm −1 and 804 cm −1 , respectively [24]. It is interesting to observe the characteristic band at 1459 cm −1 , which is present in the FTIR spectra of NH2KIT-6 but is absent in the FTIR spectra of KIT-6. This characteristic band at 1459 cm −1 indicates the appearance of N−H bonds in NH2KIT-6 and hence prove the successful amine-functionalization of KIT-6 [20,23,25]. In addition, the characteristic band at around 2926 cm −1 , which indicates the C−H stretching vibration of the organosilane, is observed in the FTIR spectra of NH2KIT-6 but is not found in the spectra of KIT-6 [23]. This further proves the effective functionalization of KIT-6.   The FTIR spectra of KIT-6 and NH2KIT-6 are shown in Figure 3. The characteristic band at 3464 cm −1 indicates the stretching vibration of hydrogen bonding from silanol group y(≡Si−OH) [23]. Furthermore, the characteristic band at 1640 cm −1 indicates the O−H bending vibration mode. The anti-symmetric and symmetric stretching vibrations of Si−O−Si groups are observed at characteristic bands at 1083 cm −1 and 804 cm −1 , respectively [24]. It is interesting to observe the characteristic band at 1459 cm −1 , which is present in the FTIR spectra of NH2KIT-6 but is absent in the FTIR spectra of KIT-6. This characteristic band at 1459 cm −1 indicates the appearance of N−H bonds in NH2KIT-6 and hence prove the successful amine-functionalization of KIT-6 [20,23,25]. In addition, the characteristic band at around 2926 cm −1 , which indicates the C−H stretching vibration of the organosilane, is observed in the FTIR spectra of NH2KIT-6 but is not found in the spectra of KIT-6 [23]. This further proves the effective functionalization of KIT-6.  Figure 4 shows the nitrogen adsorption-desorption isotherms of KIT-6 and NH2KIT-6. Type IV with a hysteresis loop from the nitrogen adsorption isotherms for KIT-6 and NH2KIT-6 is observed, which represents the characteristic of a mesoporous material [4,5,20,26]. According to the IUPAC classification, a type H1 hysteresis loop can be observed from the isotherm; this indicates the characteristic of mesoporous materials and characteristic of material with interconnected large   Figure 4 shows the nitrogen adsorption-desorption isotherms of KIT-6 and NH 2 KIT-6. Type IV with a hysteresis loop from the nitrogen adsorption isotherms for KIT-6 and NH 2 KIT-6 is observed, which represents the characteristic of a mesoporous material [4,5,20,26]. According to the IUPAC classification, a type H1 hysteresis loop can be observed from the isotherm; this indicates the characteristic of mesoporous materials and characteristic of material with interconnected large cylindrical pore geometry and a high degree of pore size uniformity [26]. In the current study, NH 2 KIT-6 sample possessed a specific surface area of 108 m 2 /g, a pore volume of 0.18 cm 3 /g, and a pore diameter of 6.50 nm.  Figure 4 shows the nitrogen adsorption-desorption isotherms of KIT-6 and NH2KIT-6. Type IV with a hysteresis loop from the nitrogen adsorption isotherms for KIT-6 and NH2KIT-6 is observed, which represents the characteristic of a mesoporous material [4,5,20,26]. According to the IUPAC classification, a type H1 hysteresis loop can be observed from the isotherm; this indicates the characteristic of mesoporous materials and characteristic of material with interconnected large cylindrical pore geometry and a high degree of pore size uniformity [26]. In the current study, NH2KIT-6 sample possessed a specific surface area of 108 m 2 /g, a pore volume of 0.18 cm 3 /g, and a pore diameter of 6.50 nm.  Figure 5 shows the top view of FESEM images of NH2KIT-6/PSF MMMs with different loadings of NH2KIT-6. The particles observed indicate that NH2KIT-6 was successfully incorporated into the PSF matrix [26]. Figure 6 shows the cross-sectional FESEM images of the NH2KIT-6/PSF MMMs. Incorporating up to 2 wt% NH2KIT-6 into the MMMs produced MMMs with no void in the matrix. After the functionalization of KIT-6 silica with APTES, the silane backbone of APTES promoted good silica-polymer interactions via van der Waals forces in the MMMs [25,27,28]. However, filler  Figure 5 shows the top view of FESEM images of NH 2 KIT-6/PSF MMMs with different loadings of NH 2 KIT-6. The particles observed indicate that NH 2 KIT-6 was successfully incorporated into the PSF matrix [26]. Figure 6 shows the cross-sectional FESEM images of the NH 2 KIT-6/PSF MMMs. Incorporating up to 2 wt% NH 2 KIT-6 into the MMMs produced MMMs with no void in the matrix. After the functionalization of KIT-6 silica with APTES, the silane backbone of APTES promoted good silica-polymer interactions via van der Waals forces in the MMMs [25,27,28]. However, filler agglomeration was observed when NH2KIT-6 loading in the MMMs was 4 wt%, as highlighted in Figure 6d.

Characterizations of the Membranes
The FTIR spectra of pristine PSF membrane and NH2KIT-6/PSF MMMs are shown in Figure 7. As observed from Figure 7a, the pristine PSF membrane shows C-H rocking at the characteristic band at 831 cm −1 . The C−C stretching is indicated by characteristic bands at 1012 cm −1 and 1103 cm −1 . The Ar−SO 2 −Ar symmetric stretching is observed at a characteristic band at 1147 cm −1 , where Ar corresponds to aromatic. On the other hand, the Ar−O−Ar stretching is observed at 1235 cm −1 and S=O symmetric stretching is observed at 1294 cm −1 [29]. Besides, the characteristic band at 2926 cm −1 indicates the C-H stretching vibration [1]. It is interesting to find about the interaction between NH 2 KIT-6 filler and the PSF polymer phase where the characteristic bands of the PSF were retained in the FTIR spectra of the MMMs. This is in agreement with the FTIR analysis reported by Khdary and Abdelsalam [30]. As observed from Figure 7b-e, a small characteristic band at 3464 cm −1 , which is assigned to the Si-OH group, was observed for the FTIR spectra of four NH 2 KIT-6/PSF MMMs [23]. This indicates the presence of NH 2 KIT-6 in the MMMs. The small characteristic band at 3464 cm −1 , which is due to the Si-OH group of the NH 2 KIT-6, was not observed in the FTIR spectra of the pristine PSF membrane.
Polymers 2020, 12, x FOR PEER REVIEW 6 of 11 agglomeration was observed when NH2KIT-6 loading in the MMMs was 4 wt%, as highlighted in Figure 6d.  The FTIR spectra of pristine PSF membrane and NH2KIT-6/PSF MMMs are shown in Figure 7. As observed from Figure 7a, the pristine PSF membrane shows C-H rocking at the characteristic band at 831 cm −1 . The C−C stretching is indicated by characteristic bands at 1012 cm −1 and 1103 cm −1 . The Ar−SO2−Ar symmetric stretching is observed at a characteristic band at 1147 cm −1 , where Ar corresponds to aromatic. On the other hand, the Ar−O−Ar stretching is observed at 1235 cm −1 and S=O symmetric stretching is observed at 1294 cm −1 [29]. Besides, the characteristic band at 2926 cm −1 indicates the C-H stretching vibration [1]. It is interesting to find about the interaction between NH2KIT-6 filler and the PSF polymer phase where the characteristic bands of the PSF were retained agglomeration was observed when NH2KIT-6 loading in the MMMs was 4 wt%, as highlighted in Figure 6d.  The FTIR spectra of pristine PSF membrane and NH2KIT-6/PSF MMMs are shown in Figure 7. As observed from Figure 7a, the pristine PSF membrane shows C-H rocking at the characteristic band at 831 cm −1 . The C−C stretching is indicated by characteristic bands at 1012 cm −1 and 1103 cm −1 . The Ar−SO2−Ar symmetric stretching is observed at a characteristic band at 1147 cm −1 , where Ar corresponds to aromatic. On the other hand, the Ar−O−Ar stretching is observed at 1235 cm −1 and S=O symmetric stretching is observed at 1294 cm −1 [29]. Besides, the characteristic band at 2926 cm −1 indicates the C-H stretching vibration [1]. It is interesting to find about the interaction between NH2KIT-6 filler and the PSF polymer phase where the characteristic bands of the PSF were retained in the FTIR spectra of the MMMs. This is in agreement with the FTIR analysis reported by Khdary and Abdelsalam [30]. As observed from Figure 7b-e, a small characteristic band at 3464 cm −1 , which is assigned to the Si-OH group, was observed for the FTIR spectra of four NH2KIT-6/PSF MMMs [23]. This indicates the presence of NH2KIT-6 in the MMMs. The small characteristic band at 3464 cm −1 , which is due to the Si-OH group of the NH2KIT-6, was not observed in the FTIR spectra of the pristine PSF membrane.  Figures 8 and 9 show the CO2 permeability and CH4 permeability of the membranes incorporated with different NH2KIT-6 loadings. The error for the CO2 permeability and CH4 permeability was ±5%. As observed in Figure 8, the CO2 permeability increased when NH2KIT-6 loading in the MMMs was increased. At a pressure difference of 5 bar, an increase of about 47% in ideal CO2/CH4 selectivity was achieved by incorporating 2 wt% NH2KIT-6 into the MMMs. As observed in Figures 8 and 9, the CO2 permeability and CH4 permeability decreased when the pressure difference increased from 5 to 7 bar. This is a common behavior of glassy polymers [31,32]. Biondo et al. [33] reported a decrease in CO2 permeability when the pressure increased for PSF and it was explained that the behavior was typical for a dual-mode model. Figure 10 shows the ideal CO2/CH4 selectivity of the membranes incorporated with different NH2KIT-6 loadings. After the functionalization of KIT-6, the presence of amine groups on NH2KIT-6 in the MMMs enhanced the affinity of MMMs toward CO2 and hence increased the ideal CO2/CH4 selectivity of the MMMs [14]. However, further increasing the NH2KIT-6 loading from 2 to 4 wt% caused a drop in the ideal CO2/CH4 selectivity, as observed in Figure 10. This might be due to the filler agglomeration in the MMMs, which started to occur at higher NH2KIT-6 loading. In addition, the gas permeability and ideal CO2/CH4 selectivity of the MMMs incorporated with pristine (unfunctionalized) KIT-6 were reported in our earlier work [34]. The MMMs incorporated with NH2KIT-6 in the current study displayed higher ideal CO2/CH4 selectivity compared to MMMs incorporated with the same loadings of pristine KIT-6 in our earlier work [34]. However, the gas permeation and separation performance of the NH2KIT-6/PSF MMMs in the current study fall below the Robeson upper bound [35]. Hence, future research work is still needed in order to further enhance the gas permeation and separation performance of the NH2KIT-6/PSF. Table 1 shows the comparison of CO2 permeability and ideal CO2/CH4 selectivity between MMMs from the current study and MMMs incorporated with functionalized silica reported in the literature. The CO2 permeability obtained in the current study is quite comparable with the CO2 permeability reported in several other studies. As compared with the  Figures 8 and 9 show the CO 2 permeability and CH 4 permeability of the membranes incorporated with different NH 2 KIT-6 loadings. The error for the CO 2 permeability and CH 4 permeability was ±5%. As observed in Figure 8, the CO 2 permeability increased when NH 2 KIT-6 loading in the MMMs was increased. At a pressure difference of 5 bar, an increase of about 47% in ideal CO 2 /CH 4 selectivity was achieved by incorporating 2 wt% NH 2 KIT-6 into the MMMs. As observed in Figures 8 and 9, the CO 2 permeability and CH 4 permeability decreased when the pressure difference increased from 5 to 7 bar. This is a common behavior of glassy polymers [31,32]. Biondo et al. [33] reported a decrease in CO 2 permeability when the pressure increased for PSF and it was explained that the behavior was typical for a dual-mode model. Figure 10 shows the ideal CO 2 /CH 4 selectivity of the membranes incorporated with different NH 2 KIT-6 loadings. After the functionalization of KIT-6, the presence of amine groups on NH 2 KIT-6 in the MMMs enhanced the affinity of MMMs toward CO 2 and hence increased the ideal CO 2 /CH 4 selectivity of the MMMs [14]. However, further increasing the NH 2 KIT-6 loading from 2 to 4 wt% caused a drop in the ideal CO 2 /CH 4 selectivity, as observed in Figure 10. This might be due to the filler agglomeration in the MMMs, which started to occur at higher NH 2 KIT-6 loading. In addition, the gas permeability and ideal CO 2 /CH 4 selectivity of the MMMs incorporated with pristine (unfunctionalized) KIT-6 were reported in our earlier work [34]. The MMMs incorporated with NH 2 KIT-6 in the current study displayed higher ideal CO 2 /CH 4 selectivity compared to MMMs incorporated with the same loadings of pristine KIT-6 in our earlier work [34]. However, the gas permeation and separation performance of the NH 2 KIT-6/PSF MMMs in the current study fall below the Robeson upper bound [35]. Hence, future research work is still needed in order to further enhance the gas permeation and separation performance of the NH 2 KIT-6/PSF. Table 1 shows the comparison of CO 2 permeability and ideal CO 2 /CH 4 selectivity between MMMs from the current study and MMMs incorporated with functionalized silica reported in the literature. The CO 2 permeability obtained in the current study is quite comparable with the CO 2 permeability reported in several other studies. As compared with the research works reported in Table 1, a relatively higher ideal CO 2 /CH 4 selectivity was achieved by incorporating 2 wt% NH 2 KIT-6 filler into the MMMs in the current study. research works reported in Table 1, a relatively higher ideal CO2/CH4 selectivity was achieved by incorporating 2 wt% NH2KIT-6 filler into the MMMs in the current study.   research works reported in Table 1, a relatively higher ideal CO2/CH4 selectivity was achieved by incorporating 2 wt% NH2KIT-6 filler into the MMMs in the current study.

Conclusions
In the current study, functionalized KIT-6 (NH2KIT-6) was incorporated into a PSF matrix to form MMMs. The presence of particles in the MMMs as observed in the FESEM images indicated the successful incorporation of NH2KIT-6 into the MMMs. The CO2 permeability and CH4 permeability decreased when the pressure difference increased from 5 to 7 bar. MMMs with no void in the matrix were successfully fabricated by incorporating up to 2 wt% NH2KIT-6 into the MMMs. Subsequently, an increase of about 47% in the ideal CO2/CH4 selectivity was achieved by incorporating 2 wt% NH2KIT-6 into the MMMs. However, filler agglomeration started to occur in the MMMs when a higher NH2KIT-6 loading was incorporated into the MMMs. The ideal CO2/CH4 selectivity dropped when the NH2KIT-6 loading was increased from 2 to 4 wt%.