Novel Cellulose Triacetate (CTA)/Cellulose Diacetate (CDA) Blend Membranes Enhanced by Amine Functionalized ZIF-8 for CO2 Separation

Currently, cellulose acetate (CA) membranes dominate membrane-based CO2 separation for natural gas purification due to their economical and green nature. However, their lower CO2 permeability and ease of plasticization are the drawbacks. To overcome these weaknesses, we have developed high-performance mixed matrix membranes (MMMs) consisting of cellulose triacetate (CTA), cellulose diacetate (CDA), and amine functionalized zeolitic imidazolate frameworks (NH2-ZIF-8) for CO2 separation. The NH2-ZIF-8 was chosen as a filler because (1) its pore size is between the kinetic diameters of CO2 and CH4 and (2) the NH2 groups attached on the surface of NH2-ZIF-8 have good affinity with CO2 molecules. The incorporation of NH2-ZIF-8 in the CTA/CDA blend matrix improved both the gas separation performance and plasticization resistance. The optimized membrane containing 15 wt.% of NH2-ZIF-8 had a CO2 permeability of 11.33 Barrer at 35 °C under the trans-membrane pressure of 5 bar. This is 2-fold higher than the pristine membrane, while showing a superior CO2/CH4 selectivity of 33. In addition, the former had 106% higher CO2 plasticization resistance of up to about 21 bar and an impressive mixed gas CO2/CH4 selectivity of about 40. Therefore, the newly fabricated MMMs based on the CTA/CDA blend may have great potential for CO2 separation in the natural gas industry.


Introduction
Natural gas is an attractive and relatively green energy source as compared to coal, mainly because of its lower carbon footprint [1][2][3]. However, depending on the geological location, raw natural gas varies substantially in composition and may contain 50-90 mole % methane together with other undesirable components such as H 2 O, CO 2 , H 2 S, N 2 , C 2 H 6 , C 3 H 8 , and toluene. The presence of CO 2 and H 2 S can cause pipeline corrosion and reduce the caloric value of the natural gas [1,4,5]. Therefore, the demand of high-purity natural gas is increasing by the energy producers to enhance its calorific and economic values [5]. To purify raw natural gas, membrane technology has emerged as an alternative process due to its environmental and economic benefits [1,5]. In particular, polymeric membranes have led to the commercialization of membrane-based gas separations for various applications including CO 2 separation from natural gas [6][7][8]. To make the membrane-based gas separation more competitive over conventional separation technologies such as amine blends were chosen as the matrix material because they have similar chemical structures, good separation performance, and a cheap and environmentally friendly nature. NH 2 -ZIF-8 was used as a filler because its pore size is between the kinetic diameters of the separating gases (CO 2 and CH 4 ). Moreover, the NH 2 group attached on the surface of ZIF-8 has good affinity with condensable gases like CO 2 . This study may provide useful insights to design next-generation CA membranes for the purification of natural gas.

ZIF-8 Synthesis
A solvo-chemical method was employed to synthesize ZIF-8 nanoparticles as reported by Pan et al. [35]. In this process, Zn(NO 3 ) 2 ·6H 2 O (1.17 g) and 2-methylimidazole (22.70 g) were dissolved separately in deionized (DI) water. The prepared solutions were stirred for a few minutes prior to the mixing and were continuously stirred at room temperature for another 12 h. The resultant milky solution was centrifuged, followed by washing with DI water. Afterwards, the washed product was dried in an oven at 65 • C overnight.

Amine Modification of ZIF-8
Amine functionalization of ZIF-8 nanoparticles was carried out following the method reported by Nordin et al. [36], with some modifications. Briefly, NH 4 OH (28 mL) and H 2 O (10 mL) were slowly added to ZIF-8 (1 g) under constant stirring, followed by overnight sonication at 80 • C. The resulting mixture was centrifuged and washed multiple times with distilled water to remove any impurities before being dried in a vacuum oven at 70 • C for 12 h. The chemical structures of ZIF-8 and NH 2 -ZIF-8 are presented in Figure 1C,D, respectively [37].
be the first study on the fabrication of CTA/CDA-NH2-ZIF-8 blend MMMs and investigation of their CO2/CH4 separation performance as a function of NH2-ZIF-8 loading. CTA and CDA blends were chosen as the matrix material because they have similar chemical structures, good separation performance, and a cheap and environmentally friendly nature. NH2-ZIF-8 was used as a filler because its pore size is between the kinetic diameters of the separating gases (CO2 and CH4). Moreover, the NH2 group attached on the surface of ZIF-8 has good affinity with condensable gases like CO2. This study may provide useful insights to design next-generation CA membranes for the purification of natural gas.

ZIF-8 Synthesis
A solvo-chemical method was employed to synthesize ZIF-8 nanoparticles as reported by Pan et al. [35]. In this process, Zn(NO3)2.6H2O (1.17 g) and 2-methylimidazole (22.70 g) were dissolved separately in deionized (DI) water. The prepared solutions were stirred for a few minutes prior to the mixing and were continuously stirred at room temperature for another 12 h. The resultant milky solution was centrifuged, followed by washing with DI water. Afterwards, the washed product was dried in an oven at 65 °C overnight.

Amine Modification of ZIF-8
Amine functionalization of ZIF-8 nanoparticles was carried out following the method reported by Nordin et al. [36], with some modifications. Briefly, NH4OH (28 mL) and H2O (10 mL) were slowly added to ZIF-8 (1 g) under constant stirring, followed by overnight sonication at 80 °C. The resulting mixture was centrifuged and washed multiple times with distilled water to remove any impurities before being dried in a vacuum oven at 70 °C for 12 h. The chemical structures of ZIF-8 and NH2-ZIF-8 are presented in Figures 1C and 1D, respectively [37].

Membrane Fabrication
Solution casting and solvent evaporation techniques were used to fabricate dense membranes. CDA and CTA powders were dried in a vacuum oven at 120 • C overnight to remove moisture. The blend polymer consisting of 80/20 (wt.%) CTA/CDA was dissolved in NMP followed by vigorous stirring at 40 • C overnight until the solution became homogeneous. Separately, the NH 2 -ZIF-8 and NMP suspension was sonicated for 2 h prior to mixing with the pre-prepared polymer blend solution. After mixing, the solution was stirred for 30 min and then cast on a clean glass plate by a casting blade, followed by solvent evaporation in a conventional oven at 100 • C overnight. The dried films were peeled off from the glass plate and dried in a vacuum oven at 120 • C for at least 24 h. The loading of NH 2 -ZIF-8 was calculated using Equation (1)

Characterizations
The crystal structure of NH 2 -ZIF-8 nanoparticles and the corresponding MMMs were analyzed by a Bruker wide-angle X-ray diffractometer (Bruker D8 advanced diffractometer, Bruker, Tokyo, Japan) using Cu-Kα as a radiation source at a wavelength of 1.54Ǻ. The morphologies of the samples were examined by a field emission scanning electron microscope (FESEM, JEOL, JSM-6700LV, Tokyo, Japan). Prior to the inspection, the membranes were frozen and fractured in liquid nitrogen followed by platinum coating using a platinum (Pt) sputter coater (JEOL JFC-1300, Peabody, MA, USA). NH 2 -ZIF-8 nanoparticles were directly glued on the surface of the sample holder before being coated. Chemical functionalities and interactions between polymers and nanofillers were confirmed by Fourier-transform infrared spectroscopy (FTIR, spectrum 100, Perkin Elmer, Beaconsfield, England, UK). The measurements were conducted between the wave numbers ranging from 4000 to 600 cm −1 . Thermal stabilities of pure NH 2 -ZIF-8 nanoparticles and MMMs were examined by a Shimadzu Thermal Analyzer (DTG-60AH/TA-60WS/FC-60A, Shimadzu corporation, Tokyo, Japan). All samples were heated at a heating rate of 10 • C/min from room temperature to 800 • C under air atmosphere.
2.6. Gas Permeation Measurements 2.6.1. Pure Gas Tests A constant volume variable pressure method was used to record the pure gas permeation properties. Detailed experimental setup and procedures can be found elsewhere [38]. All the membranes were tested at 35 • C under the trans-membrane pressure of 5 bar. For each membrane, at least three samples were tested and the average results were reported. The gas permeability was calculated based on the rate change of the downstream pressure increase (dp/dt) at steady state by using Equation (2) P = 273 × 10 10 760 VL AT(p 2 × 76/14.7) dp dt (2) where P is the membrane permeability in Barrer (1 Barrer = 1 × 10 −10 cm 3 (STP) cm/(cm 2 s cmHg), L is the membrane thickness (cm), V represents the volume of the downstream chamber (cm 3 ), A is the effective membrane area (cm 2 ), T signifies the operating temperature (K), and the upstream pressure is represented by p 2 (psia). The ideal selectivity (α A/B ) of two gases (A and B) was calculated according to Equation where P A and P B represent the permeability (Barrer) of gases A and B, respectively. S and D denote the solubility and diffusion coefficients of the gas, respectively. S A /S B and D A /D B are the solubility selectivity and diffusion selectivity of the gas pair, respectively. The sorption behaviors of the blend membrane and the optimized MMM were evaluated using a XEMIS microbalance. Detailed experimental procedures can be found elsewhere [39]. Equation (4) was used to estimate the solubility coefficient (S, m 3 (STP)/cm 3 cmHg) of the adsorbed gas inside the membrane at 5 bar, while the diffusivity coefficient (D, 10 −8 cm 2 /s) was calculated based upon Equation (5).
where C represents the total adsorbed gas concentration (cm 3 (STP)/cm 3 ) and p is the feed pressure (cmHg).

Mixed Gas Tests
Mixed gas tests were carried out at 10 bar and 35 • C using a binary feed mixture of CO 2 /CH 4 (50/50 v/v). The detailed experimental description can be found elsewhere [40]. The permeability for each gas was calculated using Equation (6) P i = 273 × 10 10 760 where P i is the permeability of gas i, p 2 represents the upstream feed gas pressure (psia), p 1 is the downstream pressure (psia) of the permeate gas, x i is the molar fraction of gas i in the feed gas stream and y i is the molar fraction of gas i in the permeate, L is the membrane thickness (cm), and V signifies the volume of the downstream chamber (cm 3 ). Figure 2 shows the weight loss profiles of the pure NH 2 -ZIF-8 nanoparticles, CTA/CDA blend membrane, and fabricated MMMs as a function of temperature. The NH 2 -ZIF-8 nanoparticles exhibited a gradual weight loss of around 9 wt.% from 30 to 450 • C owing to the removal of guest molecules trapped in the nanocrystals during synthesis and post treatment steps, followed by a steep decrease from 450 to 600 • C due to the framework decomposition. The findings are in good agreement with the reported thermal behavior of ZIF-8 [41,42]. The TGA thermogram of the pure CTA/CDA blend membrane revealed three weight-loss steps which are consistent with the literature [43][44][45]. The initial weight loss from 30 to 120 • C corresponded to the removal of volatile matters and moisture adsorbed by the membrane due to the hygroscopic nature of CA. The second major weight loss in the range of 310-400 • C symbolized the thermal degradation of the polymer followed by the third step due to the carbonization of degraded products to ash. Thermograms of MMMs also exhibited these three steps of weight loss and demonstrate good thermal stability up to 310 • C. Table 1 shows a significant improvement in decomposition temperature (Td) with an increase in NH 2 -ZIF-8 loading. The improvement in Td arose from (1) good compatibility and strong interaction between the filler and the polymer matrix and (2) inherent thermal characteristics of NH 2 -ZIF-8 nanoparticles. Thus, the incorporation of NH 2 -ZIF-8 into the CTA/CDA blend membrane hindered the chain movement and raised the energy requirement to decompose its polymer structure. A similar observation has been widely reported in the literature [46][47][48].    [49]. The XRD pattern of the NH2-ZIF-8 nanocrystals was also in good agreement with the literature [50,51]. All the diffraction patterns of CTA/CDA-NH2-ZIF-8 MMMs possessed the characteristic peaks of both NH2-ZIF-8 and CTA/CDA moieties, signifying the preservation of their crystalline structures in the membranes. In addition, the intensity of the characteristic peaks of NH2-ZIF-8 improved with an increase in its loading in MMMs.    [49]. The XRD pattern of the NH 2 -ZIF-8 nanocrystals was also in good agreement with the literature [50,51]. All the diffraction patterns of CTA/CDA-NH 2 -ZIF-8 MMMs possessed the characteristic peaks of both NH 2 -ZIF-8 and CTA/CDA moieties, signifying the preservation of their crystalline structures in the membranes. In addition, the intensity of the characteristic peaks of NH 2 -ZIF-8 improved with an increase in its loading in MMMs.

Characterizations
FTIR spectra of NH 2 -ZIF-8 nanoparticles and all fabricated membranes are compared in Figure 4. The CTA/CDA spectrum was characterized by major peaks at 3483, 2950, and 1750 cm −1 corresponding to O-H, C-H, and C=O groups, respectively [52]. In contrast, the FTIR spectrum of NH 2 -ZIF-8 nanoparticles was in good agreement with the reported literature [52,53]. The amine functionalization of ZIF-8 was also confirmed by the absorption peaks at~1309 cm −1 and 690 cm −1 due to the NH 2 bonding on ZIF-8 molecules [54].  FTIR spectra of NH2-ZIF-8 nanoparticles and all fabricated membranes are compared in Figure 4. The CTA/CDA spectrum was characterized by major peaks at 3483, 2950, and 1750 cm −1 corresponding to O-H, C-H, and C=O groups, respectively [52]. In contrast, the FTIR spectrum of NH2-ZIF-8 nanoparticles was in good agreement with the reported literature [52,53]. The amine functionalization of ZIF-8 was also confirmed by the absorption peaks at ~1309 cm −1 and 690 cm −1 due to the NH2 bonding on ZIF-8 molecules [54]. In summary, the FTIR spectra reconfirmed the existence of CTA/CDA and NH2-ZIF-8 nanoparticles in all fabricated MMMs.  This signifies the good compatibility between the NH 2 -ZIF-8 nanoparticles and CTA/CDA matrix because of (1) the small size of NH 2 -ZIF-8 nanoparticles and (2) the formation of hydrogen bonds between the −NH 2 groups of NH 2 -ZIF-8 nanoparticles and the −OH groups of the polymer matrix [5]. The uniform dispersion of NH 2 -ZIF-8 nanoparticles was also confirmed by the energy-dispersive X-ray spectroscopy (EDX). Figure 6 shows the mapping of Zn particles in the MMMs with variable loadings. Since the polymer matrix did not contain any Zn element, the uniform dispersion of Zn element verified the uniform dispersion of NH 2 -ZIF-8 nanocrystals in the polymer matrix.

Gas Transport Properties
Pure gas separation performances of the CTA/CDA blend membrane and CTA/CDA-NH 2 -ZIF-8 MMMs are shown in Figure 7. Table 1 tabulates the detailed results. Both CO 2 and CH 4 permeabilities exhibited noticeable increases as a function of NH 2 -ZIF-8 loading. However, the CO 2 /CH 4 selectivity exhibited an up and down trend. A similar trend was observed for polyimide-ZIF-8 MMMs in the literature [55]. Generally, the incorporation of nanofillers in the polymer matrix disrupts the packing of polymeric chains, which may result in additional free volume and diffusion paths for gas transport [56]. Therefore, the CTA/CDA-15 wt.% NH 2 -ZIF-8 membrane had a 50% increase in CO 2 permeability from 7.56 to 11.33 Barrer and a 23% increase in CO 2 /CH 4 selectivity from 27.0 to 33.3 compared to the neat CTA/CDA blend membrane. The higher selectivity may have resulted from the enhanced molecular sieving capability provided by NH 2 -ZIF-8 as its aperture size (3.4 Å) is between the kinetic diameters of CO 2 (3.3 Å) and CH 4 (3.8 Å) [26], while the larger permeability may have arisen from strong interaction among CO 2 , imidazole linkers, and -NH 2 groups of NH 2 -ZIF-8 that facilitated gas transport across the membrane [57]. However, a further increment in NH 2 -ZIF-8 loading to 20 wt.% resulted in a lower CO 2 /CH 4 selectivity of 24.4 but a higher CO 2 permeability of 13.2 Barrer. This is because a higher loading may generate nonselective voids. As a consequence, the CTA/CDA-15 wt.% NH 2 -ZIF-8 blend MMM was selected for subsequent investigations.   nanoparticles had a size between 50 and 80 nm. Meanwhile, they were well dispersed in the CTA/CDA matrix with no clear evidence of interfacial gaps and phase separation. This signifies the good compatibility between the NH2-ZIF-8 nanoparticles and CTA/CDA matrix because of (1) the small size of NH2-ZIF-8 nanoparticles and (2) the formation of hydrogen bonds between the −NH2 groups of NH2-ZIF-8 nanoparticles and the −OH groups of the polymer matrix [5]. The uniform dispersion of NH2-ZIF-8 nanoparticles was also confirmed by the energy-dispersive X-ray spectroscopy (EDX). Figure 6 shows the mapping of Zn particles in the MMMs with variable loadings. Since the polymer matrix did not contain any Zn element, the uniform dispersion of Zn element verified the uniform dispersion of NH2-ZIF-8 nanocrystals in the polymer matrix.    (2) the formation of hydrogen bonds between the −NH2 groups of NH2-ZIF-8 nanoparticles and the −OH groups of the polymer matrix [5]. The uniform dispersion of NH2-ZIF-8 nanoparticles was also confirmed by the energy-dispersive X-ray spectroscopy (EDX). Figure 6 shows the mapping of Zn particles in the MMMs with variable loadings. Since the polymer matrix did not contain any Zn element, the uniform dispersion of Zn element verified the uniform dispersion of NH2-ZIF-8 nanocrystals in the polymer matrix.

Gas Transport Properties
Pure gas separation performances of the CTA/CDA blend membrane and CTA/CDA-NH2-ZIF-8 MMMs are shown in Figure 7. Table 1 tabulates the detailed results. Both CO2 and CH4 permeabilities exhibited noticeable increases as a function of NH2-ZIF-8 loading. However, the CO2/CH4 selectivity exhibited an up and down trend. A similar trend was observed for polyimide-ZIF-8 MMMs in the literature [55]. Generally, the incorporation of nanofillers in the polymer matrix disrupts the packing of polymeric chains, which may result in additional free volume and diffusion paths for gas transport [56]. Therefore, the CTA/CDA-15 wt.% NH2-ZIF-8 membrane had a 50% increase in CO2 permeability from 7.56 to 11.33 Barrer and a 23% increase in CO2/CH4 selectivity from 27.0 to 33.3 compared to the neat CTA/CDA blend membrane. The higher selectivity may have resulted from the enhanced molecular sieving capability provided by NH2-ZIF-8 as its aperture size (3.4Å) is between the kinetic diameters of CO2 (3.3Å) and CH4 (3.8Å) [26], while the larger permeability may have arisen from strong interaction among CO2, imidazole linkers, and -NH2 groups of NH2-ZIF-8 that facilitated gas transport across the membrane [57]. However, a further increment in NH2-ZIF-8 loading to 20 wt.% resulted in a lower CO2/CH4 selectivity of 24.4 but a higher CO2 permeability of 13.2 Barrer. This is because a higher loading may generate nonselective voids. As a consequence, the CTA/CDA-15 wt.% NH2-ZIF-8 blend MMM was selected for subsequent investigations.  3.2.1. Sorption Behavior of CO2 and CH4 Figure 8 presents the dual-mode Langmuir-Henry sorption behavior for both CO2 and CH4 in the pristine polymer blend and the optimized MMM. Both membranes showed higher CO2 adsorption than CH4 because of the inherently higher sorption affinity between the CTA/CDA polymer and CO2 [58]. Table 2 summarizes their permeability, solubility, diffusivity coefficients, and selectivity. As expected, the CO2 diffusivity and solubility coefficients showed 44% and 4% increases, respectively, when 15 wt.% NH2-ZIF-8 was incorporated into the CTA/CDA blend matrix. Consequently, the overall improvement in gas separation performance came from two factors; namely, the additional diffusive pathways and free volume provided by NH2-ZIF-8. However, comparing the per- 3.2.1. Sorption Behavior of CO 2 and CH 4 Figure 8 presents the dual-mode Langmuir-Henry sorption behavior for both CO 2 and CH 4 in the pristine polymer blend and the optimized MMM. Both membranes showed higher CO 2 adsorption than CH 4 because of the inherently higher sorption affinity between the CTA/CDA polymer and CO 2 [58]. Table 2 summarizes their permeability, solubility, diffusivity coefficients, and selectivity. As expected, the CO 2 diffusivity and solubility coefficients showed 44% and 4% increases, respectively, when 15 wt.% NH 2 -ZIF-8 was incorporated into the CTA/CDA blend matrix. Consequently, the overall improvement in gas separation performance came from two factors; namely, the additional diffusive pathways and free volume provided by NH 2 -ZIF-8. However, comparing the percentages of their increments, one can easily conclude that the former played a more important role than the latter to enhance the molecular sieving capability of the CTA/CDA blend membrane for CO 2 /CH 4 separation, as observed in the literature [59].

Plasticization Behavior and Mixed Gas Tests
To investigate the CO2-induced plasticization phenomenon, the testing pressure of CO2 was intermittently ramped from 5 to 25 bar at 35 °C. Figure 9 shows

Plasticization Behavior and Mixed Gas Tests
To investigate the CO 2 -induced plasticization phenomenon, the testing pressure of CO 2 was intermittently ramped from 5 to 25 bar at 35 • C. Figure 9 shows  Table 3 compares the gas separation performance of the pristine CTA/CDA blend membrane and the optimized MMM under pure and mixed gas tests where a binary mixture of CO2/CH4 (50/50, v/v) was used as the mixed gas feed. Consistent with the literature, the mixed gas tests showed lower permeabilities for both CO2 and CH4 gases than the pure gas ones owing to the competitive diffusion and sorption between CO2 and CH4 molecules [61,62]. Since the percentage of permeability drop for CH4 in the former was higher than that in the latter, this results in a higher CO2/CH4 selectivity in the mixed gas tests.  Table 4 benchmarks the pure-gas separation performance of the newly developed membranes with other CA membranes reported in the literature for CO2 and CH4 separation. The CTA/CDA-15 wt.% NH₂-ZIF-8 membrane showed higher separation performance because its polymer matrix was made of a CTA/CDA blend that had inherently  Table 3 compares the gas separation performance of the pristine CTA/CDA blend membrane and the optimized MMM under pure and mixed gas tests where a binary mixture of CO 2 /CH 4 (50/50, v/v) was used as the mixed gas feed. Consistent with the literature, the mixed gas tests showed lower permeabilities for both CO 2 and CH 4 gases than the pure gas ones owing to the competitive diffusion and sorption between CO 2 and CH 4 molecules [61,62]. Since the percentage of permeability drop for CH 4 in the former was higher than that in the latter, this results in a higher CO 2 /CH 4 selectivity in the mixed gas tests.  Table 4 benchmarks the pure-gas separation performance of the newly developed membranes with other CA membranes reported in the literature for CO 2 and CH 4 separation. The CTA/CDA-15 wt.% NH 2 -ZIF-8 membrane showed higher separation perfor-mance because its polymer matrix was made of a CTA/CDA blend that had inherently high gas separation performance and (2) it had 15 wt.% NH 2 -ZIF-8 nanoparticles to enhance its gas diffusivity and molecular sieving characteristics.

Conclusions
We fabricated high-performance CTA/CDA-NH 2 -ZIF-8 MMMs for CO 2 /CH 4 separation. A solvo-chemical method was employed to synthesize NH 2 -ZIF-8 nanocrystals with a particle size of 50-80 nm. Gas separation performance of the fabricated membranes have been investigated as a function of NH 2 -ZIF-8 loading. The following conclusions can be drawn:

1.
There is a linear relationship between permeability and NH 2 -ZIF-8 loading. However, the relationship changes to a-shape between CO 2 /CH 4 selectivity and NH 2 -ZIF-8 loading due to void formation. Thus, there is an optimum loading to fabricate CTA/CDA based MMMs with high gas separation performance.

2.
The optimized MMM contained 15 wt.% of NH 2 -ZIF-8 nanoparticles and had a CO 2 permeability of 11.33 Barrer which was 50% higher than the pristine CTA/CDA membrane. In addition, the former had a superior CO 2 /CH 4 selectivity of 33.33 to the latter of 27 under pure gas tests.

3.
The enhanced molecular sieving capability and additional free volume provided by NH 2 -ZIF-8 nanoparticles improved the CO 2 diffusivity and solubility coefficients by 44% and 4%, respectively, under pure gas tests when 15 wt.% NH 2 -ZIF-8 was incorporated into the CTA/CDA membrane. 4.
The CO 2 /CH 4 selectivity can be further increased to 41 under mixed gas tests due to the competitive diffusion and sorption between CO 2 and CH 4 molecules.

5.
A notable improvement in CO 2 -induced plasticization pressure from 10.4 to 21.47 bar was observed owing to the good compatibility and chain interactions between CTA/CDA and NH 2 -ZIF-8 molecules.
Therefore, the newly fabricated polymer blend MMMs may have great potential for CO 2 separation in the natural gas industry. Funding: The authors would like to thank the Singapore Energy Centre (SgEC) under the project entitled "Design of novel mixed matrix membranes and thin-film composite membranes for ultra-efficient hydrogen separation and CO 2 capture" (Grant number: R-261-517-004-592) for funding this re-search.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.