High-pressure CO2/CH4 separation of Zr-MOFs based mixed matrix membranes

The gas separation properties of 6FDA-DAM mixed matrix membranes (MMMs) with three types of zirconiumbased metal organic framework nanoparticles (MOF NPs, ca. 40 nm) have been investigated up to 20 bar. Both NPs preparation and MMMs development were presented in an earlier publication that reported outstanding CO2/CH4 separation performances (50:50 vol% CO2/CH4 feed at 2 bar pressure difference, 35 °C) and this subsequent study is to demonstrate its usefulness to the natural gas separation application. In the current work, CO2/CH4 separation has been investigated at high pressure (2–20 bar feed pressure) with different CO2 content in the feed (10–50 vol%) in the temperature range 35–55 °C. Moreover, the plasticization, competitive sorption effects, and separation of the acid gas hydrogen sulfide (H2S) have been investigated in a ternary feed mixture of CO2:H2S:CH4 (vol% ratio of 30:5:65) at 20 bar and 35 °C. The incorporation of the Zr-MOFs in 6FDA-DAM enhances both CO2 permeability and CO2/CH4 selectivity of this polymer. These MMMs exhibit high stability under separation conditions relevant to an actual natural gas sweetening process. The presence of H2S does not induce plasticization but increases the total acid gas permeability, acid gas/CH4 selectivity and only causes reversible competitive sorption. The overall study suggests a large potential for 6FDA-DAM Zr-MOF MMMs to be applied in natural gas sweetening, with good performance and stability under the relevant process conditions.


Introduction
The acid gas content (carbon dioxide, CO 2 ; hydrogen sulfide, H 2 S) in raw natural gas varies accordingly to the hydrocarbon geo-origins [1,2] and is commonly in the range of 25-55 mol.% for CO 2 and below 2 mol.% for H 2 S (≥5 mol.% in several regions) [3,4]. CO 2 , the most undesirable diluent aside from H 2 S, is essential to be discarded from the gas stream as it corrodes transmission pipelines in the presence of water [5,6]. Additionally, CO 2 lowers the natural gas caloric value and causes atmospheric pollution [2,3,5]. Consequently, the content of these impurities must be reduced to meet the industrial processing and pipeline distribution requirements, e.g., maximum allowable contents of 2-3 mol.% CO 2 and 0.0004-0.0005 mol.% (4.3-5.0 ppm) H 2 S (see Table S1) [7]. In the last decades, the advances in gas separation membranes have allowed the technology to increase its share of the total membrane market, comprising over 1000-1500 million US dollar per year [8] and appear to be the most viable alternative to substitute the conventional highly energy consuming processes, including the solvent-based adsorption processes [5]. However, due to challenges such as plasticization especially at high-pressure operation and degradation, membrane processes only represents < 5% of the natural gas sweetening market [9,10].
Permeation of a mixture of gases through a membrane can depend strongly on the operating parameters, for example, the feed pressure and temperature, amongst others due to the gases' non-ideal behavior [21][22][23] and their competitive sorption [21,[23][24][25]. Moreover, in a MMM system, the presence of a porous filler and the new filler-polymer interfacial phase created need to be understood as they further influence the gas mobility and sorption through the membrane. Metal organic frameworks (MOFs), formed with metal-based clusters linked by organic ligands [26] in three-dimensional crystalline frameworks with permanent porosity, are an emerging class of porous fillers [27]. They have gained substantial attention due to their high CO 2 uptake (i.e., HKUST-1 of 7.2 mmol·g −1 [28], MOF-74 of 4.9 mmol·g −1 [29], at 1 bar, 273-298 K), large surface areas up to 7000 m 2 ·g −1 [30], well-defined selective pores due to their crystallinity, amongst other features. Many researchers observed that the incorporation of a MOF into the polymer continuous phase improved not only its separation properties but also its physical properties [16,[31][32][33], due to interfacial interactions between the polymer and the MOFs. The polymer, in some cases, penetrates the MOF open pores or rigidifies and forms microvoids at the interface [34,35], thereby affecting the membrane's physical properties and gas separation performance.
Zr-based MOF UiO-66 is a highly stable new material and has recently been applied as part of a MMM [31,36,37]. The synthesis of three types of Zr-MOFs, namely UiO-66 and its functionalized derivatives, UiO-66-NH 2 and UiO-66-NH-COCH 3 , as well as MMM fabrication with 6FDA-DAM have been presented earlier [31,34]. In the current paper, we present the gas separation performance of the neat 6FDA-DAM membranes and their derived Zr-MOF MMMs as a function of feed pressure between 2 and 20 bar. At the highest pressure, the effects of CO 2 content in the feed mixture on membrane performance have been investigated, at various temperatures (35-55°C). Finally, the presence of H 2 S to the separation performances has been studied.

Materials and membrane fabrications
The UiO-66 and UiO-66-NH 2 NPs (ca. 40 nm in size) were synthesized accordingly to Hou et al. [38], at 1 to 1 M ratio of zirconium (IV) chloride (ZrCl 4 , ≥99.5% trace metal basis) to 1,4-benzenedicarboxylic acid (BDC, 98%) or 2-amino-1,4-benxenedicarboxylic acid (NH 2 -BDC, 99%), in N,N-dimethylformamide (DMF, ≥99.9%), through a solvothermal process in a pre-heated oven at 120°C/24 h for UiO-66 and at 80°C/14 h for UiO-66-NH 2 . A second heating step was conducted for UiO-66-NH 2 at 100°C for 24 h. UiO-66 was activated by thermal treatment in a furnace at 300°C for 3 h, with a heating rate of 15°C·min −1 , whereas chemical activation was conducted for UiO-66-NH 2 , where the precipitated NPs were washed in an absolute ethanol bath at 60°C, three times in three days (ethanol was changed daily). After the complete cycle, the NPs were dried at room temperature. A covalent post-synthetic modification (PSM) was conducted onto UiO-66-NH 2 to produce UiO-66-NH-COCH 3 in chloroform (CHCl 3 , anhydrous ≥99%) and acetic anhydride (AcO 2 , ACS Reagent, ≥98.0%) solution, under reflux at 55°C/24 h. Once completed, the colloidal solution was centrifuged, rinsed with fresh CHCl 3 (15 mL, 3x) and dried overnight at 150°C before characterization and use. The conversion yield was determined by the percentage of amide groups present in the modified NPs using proton nuclear magnetic resonance ( 1 H NMR), and the digestion method was presented elsewhere [38,39]. All reactants applied in the NP synthesis were supplied by Sigma-Aldrich.
6FDA-DAM (Mw = 418 kDa) was purchased from Akron Polymer Systems, Inc. and dried overnight at 100°C before use. Pure polymer membranes ("neat") and MMMs were fabricated by dissolving the corresponding amount of 6FDA-DAM in chloroform, making a dope solution of 10 wt%. In the case of MMM, a priming step was conducted with 10-15 wt% of the total polymer weight that proves to improve the inorganic filler dispersion in the continuous polymer phase [40][41][42]. The final dope solutions were casted in a Petri dish and covered for controlled solvent evaporation overnight before being treated at 110°C before subsequent characterization and permeation measurements. The flat sheet membranes were in the thickness range of 100-150 μm.

Standard permeation measurement
To assess the gas separation performance of the membranes, a 25/ 25 cm 3 (STP)·min −1 CO 2 /CH 4 binary feed mixture was used at a pressure difference of 2 bar at 35°C applying He as sweep gas at 1 cm 3 (STP)·min −1 . The permeate composition was analyzed online by an Agilent 3000A micro-GC equipped with a thermal conductivity detector (TCD) at the Institute Nanoscience of Aragon (INA), University of Zaragoza. The membrane module is as described elsewhere [43]. The permeability was calculated as the penetrated gas flux, normalized for the membrane thickness and the partial pressure drop across the membrane, and presented in Barrer (1 Barrer = 10 −10 cm 3 (STP)·cm·cm −2 ·s −1 ·cmHg −1 (Eq. (1)).
The separation factor (α) of two competing gases was calculated using Eq. (2), considering the mole fraction (x) of gas i and j in both feed and permeate streams. The mixed gas separation performance was previously discussed [34], and the best performing MMMs are with 14-16 wt% Zr-MOF particle loadings.

High-pressure performance evaluation
The membranes were placed in a proprietary high-pressure permeation module obtained from the European Membrane Institute (EMI, The Netherlands). The membrane was supported with an S&S 589/1 black ribbon ash-less filter paper on a perforated plate to avoid membrane deformation during the high-pressure testing. The sample was sealed with an o-ring system providing for an effective membrane area of 0.78 cm 2 . Both feed and retentate sides were connected by highpressure Swagelok quick-connects whereas the permeate gas was collected using a 1/8 in. Swagelok connector.
The permeation module was placed inside a Memmert UF450 forced air circulation oven, connected to a proprietary high-pressure permeation set-up at SINTEF Materials and Chemistry, Oslo for gas separation measurement (Fig. 1). The permeation set-up is designed to withstand pressures up to 92 bar with a forced air temperature control up to 300°C. The feed (150 cm 3 (STP)·min −1 ) and permeate (10 cm 3 (STP)·min −1 ) flow rates were controlled by automated Bronkhorst High-Tech mass controllers (MFC), equipped with a back pressure controller (Bronkhorst High-Tech, P-512C equipped with an F-033C control valve, max of 92 bars) on the feed side for pressure regulation. The atmospheric-pressure permeate gas analyzed by a twochannel column (MolSieve 5A, MS5 and PoraPLOT U, PPU) Agilent 490 micro-GC, coupled with thermal conductivity detectors (TCD). The micro-GC was calibrated for low CO 2 (0-12 vol%), CH 4 (0-5 vol%) and H 2 S (0-0.5 vol%) concentrations in argon. Good correlation coefficients of R 2 = ≥0.999 were obtained for the µ-GC response as a function of CO 2 , CH 4 , and H 2 S concentration. The fluxes were calculated from the measured permeate concentrations and the calibrated flow of Ar sweep gas.
High-pressure gas permeation measurements were conducted accordingly to the following experimental sequence, and the separation performances were calculated correspondingly to Eqs. (1) and (2).
1. Pressure variation with 50:50 vol% CO 2 : CH 4 feed mixture: Preliminary measurement at 2 bar and 35°C was conducted to validate the initial membrane performances, and the pressure was subsequently increased to 5 and 10 bar. Before proceeding to 20 bar, the CO 2 feed content was decreased to 10 vol% for the second step measurements. 2. CO 2 feed content variation at the feed pressure of 20 bar: At 20 bar, the 10 vol% CO 2 feed content was subsequently increased to 20 vol %, 30 vol%, and 50 vol% with CH 4 . 3. The effect of temperature variation on the separation performance, with 30:70 vol% CO 2 :CH 4 feed mixture at 20 bar: The temperature increase was conducted by stepwise increments from 35°C to 45°C and 55°C, and followed by a reduction back to 35°C prior to the H 2 S introduction (step no. 4). 4. Investigation of separation performance in the presence of H 2 S with 30:5:65 vol% CO 2 :H 2 S:CH 4 feed mixture was conducted at 20 bar and 35°C. 5. Finally, the H 2 S in feed was removed and the system was allowed to purge before the separation efficiency was re-evaluated with 30:70 vol% CO 2 :CH 4 feed mixture, at 20 bar and 35°C.
It is important to note that the samples were allowed to reach permeation steady-state overnight, after each pressure or feed composition change. Specific attention was given to Health, Safety and Environmental (HSE) matters, and the lab was equipped with preventive safety measures which include H 2 , CO, and H 2 S detection systems, personal portable gas detectors, and separate floor level ventilation suction.

Results and discussions
In the previous publication [34], we found very promising performance indicators for several 6FDA-DAM MMMs with Zr-MOFs when tested at low pressure (2 bar), with the best performance observed for membranes that contain 14-16 wt% Zr-MOF. An increase in the Zr-MOF loading shows a clear permeability-selectivity trade-off, and selectivity reductions have been observed [34,44]. Table 1 shows the re-measured gas separation performance of the duplicate membranes, at 35°C, with a pressure difference of 2 bar with an equimolar binary mixture of CO 2 and CH 4 in SINTEF facility. The permeability values are lower than the published data [34], possibly due to the aging phenomenon which may have occurred during shelf-storage at room temperature for over 250 days. However, the similar improvement trends upon Zr-MOF incorporation were observed. The presence of 14 wt% UiO-66, 16 wt% UiO-66-NH 2 and 16 wt% UiO-66-NH-COCH 3 improves the CO 2 permeability of 6FDA-DAM (P CO2 = 335 Barrer) by 165%, 56% and 37%, respectively. These enhancements are well-related to the CO 2 -philic nature of the Zr-MOFs where a stronger energetic interaction between CO 2 (higher quadrupole moment than CH 4 ) and the nanoparticle surfaces at zero coverage, and to the increments in fractional free volume (FFV) in the MMMs (Neat 6FDA-DAM, FFV = 0.238). 14 wt% UiO-66 MMM presents the highest increment value of 39%, followed by 16 wt% UiO-66-NH 2 and 16 wt% UiO-66-NH-COCH 3 with 16% and 22%, respectively. The CO 2 /CH 4 selectivity of the samples also increased by 23-32%.
At these observed optimum loadings, the Zr-MOFs addition enhances both CO 2 permeability and CO 2 /CH 4 selectivity tremendously. Besides a higher gas diffusion in the Zr-MOFs, the NPs addition improved the MMM gas diffusivity by inducing an ancillary selective interface phase [45] with additional free volume [46,47]. Agglomeration of the NPs was more prominent at the highest loadings, and the concurrent reduction of the selectivity reduction is likely due to the formation of non-selective by-pass channels in the filler agglomerates [46] and possibly micro-voids in the filler-polymer interface region [41], although such morphological features are not observed by SEM analyses. All the MMMs also presented excellent distribution and inorganic filler-polymer interface interaction (please refer to SEM images, Fig. S5 in the previous publication [34]).

Effect of feed pressure variation to mixed gas separation
Most of the fundamental studies on Zr-MOF polyimide MMMs related to Matrimid® and 6FDA-copolyimides have been conducted at low pressures where CO 2 -induced plasticization is expected to be of minor importance [31,48,49]. Here, we have investigated the gas separation performance of 6FDA-DAM and its Zr-MOF MMMs at a pressure ranging from 2 to 20 bar in a 50:50 vol% CO 2 :CH 4 feed mixture at 35°C. The obtained mixed gas permeability and CO 2 /CH 4 selectivity behavior as a function of pressure are shown in Fig. 2.
The CO 2 -induced plasticization pressure is defined to occur at the minimum observed in the CO 2 -permeability as a function of CO 2 -partial feed pressure. In the case of mixed gases, the permeation rate of all gases is affected due to swelling of the polymer matrix and the increased chain mobility caused by the high CO 2 concentration. The permeation enhancement is more pronounced for the least permeable gases, resulting in a decrease of the selectivity as a function of pressure. In contrast, for all samples in the present study, a monotone decrease in CO 2 permeability with increasing pressure is observed (Fig. 2), which does not indicate substantial plasticization [21]. The decrease in CO 2 permeability reduction is a result of competitive sorption and the concave shape of the sorption isotherm [25,50]. This constitutes a reduction in driving force for transport with increasing pressure and also gradual saturation of the material may result in lower mobility. Overall, this results is further supported by the clear decrease in permeation coefficient in the polymer matrices (see Fig. S1). The CO 2 permeability continuously decreases with increasing pressure indicating there is no apparent CO 2 -induced plasticization in the thick membrane [21], opposite to the reported single-gas CO 2 -plasticization pressure of neat 6FDA-DAM membrane between~10-20 bar, at 35°C [51,52]. The plasticization pressure differences may be attributed to different physical properties, i.e., molecular weight, density, and polymer free volume, as previously discussed [31,34].
The pressure dependence of the CH 4 permeability ( Fig. 2(b)) over the measured pressure range, however, suggests that the neat 6FDA-DAM starts to swell immediately after the first pressure increment. It can be explained by dynamic swelling of the polymer matrices upon exposure to the CO 2 at high pressure [53], where the penetrating CO 2 causes the material dilation and subsequently increases its macromolecular mobility. Several researchers have reported the thermodynamics of swollen glassy polymers by a penetrant [54,55], and a thorough discussion was recently presented by Ogieglo et al. [53] when studying the glassy polymer relaxation in this films. The phenomenon, to the function of pressure, causes extensive dilation of the matrices, influencing the penetrants' permeation. Here, the effect is more apparent in CH 4 permeability increase compared to the readily highpermeability CO 2 . In the case of UiO-66-NH 2 MMM, the high CO 2 -affinity amino functional group increases the CO 2 adsorption in the polymer matrixes and directly further influences the molecular dynamic dilation. Even though it is not the membranes' plasticization pressure, their CO 2 /CH 4 selectivity reduced by 55% and 58% respectively. This behavior also defined as swelling-induced perm-selectivity losses [34], which was observed in several other co-polyimides, such as 6FDA-APAF and TPDA-APAF, when measured with CO 2 /CH 4 binary mixture up to 25 bar feed pressure, at 35°C [56]. Heck et al. [57] also observed similar behavior in (6FDA-mPDA)-(6FDA-durene) block co-polyimide, for which they reported an increase in CH 4 permeability with pressure (up to 20 bar feed pressure), causing CO 2 /CH 4 and He/CH 4 selectivity reductions.
The continuous decrease of CH 4 permeability in both UiO-66 and UiO-66-NH-COCH 3 MMMs demonstrated the competitive sorption effect [59], where CO 2 penetrated the membranes' sorption sites which associated to the non-equilibrium free volume in glassy polymer and hindered CH 4 to permeate. Polymer plasticization was not observed in these membrane samples. Fig. S2(a-b) show the CO 2 and CH 4 permeability of the neat 6FDA-DAM and Zr-MOF MMMs, measured at 20 bar feed pressure and 35°C, with a different CO 2 feed content between 10 and 50 vol%. The significant differences in the initial CO 2 permeabilities between the membranes were discussed in the previous publication [34]; higher CO 2 permeability in the UiO-66 MMM is attributed to the easiness of CO 2 to diffuse into its frameworks, compared to the higher steric hindrance functionalized-MOFs, and also its higher FFV.

Effect of CO 2 feed composition in high-pressure separation
The CO 2 permeability in the neat 6FDA-DAM and its Zr-MOFs MMMs decreases between 9 and 22%, with the increase of CO 2 partial pressure when tested at 20 bar. The lowest reduction of 8.7% was observed for the UiO-66 MMM. The observation, however, is opposite to the previously reported CO 2 permeability relationship with CO 2 partial pressure at low-pressure measurements, i.e., 6FDA-DAM Zr-MOF MMMs (at 2 bar) [34] and PES/SAPO-34/2-hydroxyl 5-methyl aniline MMMs (at 3 bar) [60]. At the low pressure, a higher CO 2 partial pressure produced a more prominent competitive sorption effect, where an increase in CO 2 solubility and transport through the membrane medium was observed and inversely decreased the second component's ability to permeate, in this case, CH 4 .
Evidently, the continuous CO 2 permeability reduction with increasing pressure suggests that the competitive sorption effect at high pressure is less influenced by the CO 2 partial pressure (see Fig. 3). Instead, it is related to the gradual saturation of permeating gases inside the polymer micro-voids [18]. Nevertheless, a slight increase in the CH 4 permeability for the neat membrane (9%) and UiO-66-NH 2 MMM (21%) is observed, indicating the possibility of CO 2 -induced plasticization that started to take effect [61,62]. These samples exhibited the highest CO 2 /CH 4 selectivity reductions of between 28 and 33% in all the samples (shown in Fig. S2(c), relative to 2008 Robeson's upper bound [58]). Despite this CH 4 permeability increment, the behavior can be explained as swelling-induced perm-selectivity losses, an early stage in polymer plasticization [56].
With regard to the initial separation performance (with 10 vol% CO 2 ), similarly to the previous discussion, neat 6FDA-DAM showed a lower CO 2 /CH 4 selectivity than that of MMMs (UiO-66-NH 2 < UiO-66 < UiO-66-COCH 3 ). The proportional selectivity increase in MMMs to the increasing CO 2 partial pressure [63][64][65], which only observed in UiO-66 MMM at the tested feed pressure of 20 bar (3% selectivity increment) represents the membrane's extended CO 2 sorption capability due to the CO 2 -induced plasticization or swelling at constant pressure [63]. Its reduction conversely was explained based on CO 2 self-inhibition as a consequence of saturation of the filler active sites at a high CO 2 concentration in a feed mixture [60,66]. Referring to that hypothesis, a lower reduction exhibited by UiO-66-NH-COCH 3 MMM (13%) compared to UiO-66-NH 2 MMM (28%), represented by its lesser concave shape in the permeability isotherm, may be due to a higher CO 2 affinity towards acetamide functional groups, with a higher number of adsorption sites compared to UiO-66-NH 2 NPs. Moreover, constant selectivity values demonstrate no dependency of an MMM system towards the increasing CO 2 partial pressure, as also revealed in the PES/SAPO-34/HMA MMM system, measured at 3 bar [60]. This hypothesis implies that only a minor amount of the active sites is occupied at low pressure. Fig. S3(a-c) shows the CO 2 and CH 4 permeability and the CO 2 /CH 4 selectivity as a function of the operating temperature applying a 30:70 vol% CO 2 :CH 4 feed mixture at 20 bar. A minor increase in CO 2 permeability of < 6% was recorded for all samples, whereas for CH 4 Fig. 3. CO 2 and CH 4 permeabilities, and CO 2 /CH 4 selectivity of 6FDA-DAM and its Zr-MOF MMMs against CO 2 partial pressure, at 20 bar and 35°C.  (3) where P 0 is a pre-exponential factor of permeation, E a is activation energy for permeability (kJ·mol −1 ), R is the universal gas constant (8.314 J·mol −1 ), and T is the temperature in K. Using CO 2 /CH 4 selectivity expression of the permeability coefficient ratio of CO 2 over CH 4 , the gas selectivity is defined as the following: (4) Fig. 4 indicates that CH 4 permeability in the 6FDA-DAM neat membrane and its Zr-MOF MMMs followed Arrhenius rule in the temperature range of 35-55°C, while the CO 2 permeability was less influenced by the temperature. Their permeability coefficients are summarized in Table 2. The permeability dependency is a combination of the diffusion and solubility coefficients temperature dependencies, and the lower CO 2 and CH 4 activation energies in MMMs as compared to the neat polymer indicate the gas transport through filler porosity [49], and in the interfacial voids on polymer-MOF and MOF-MOF regions which may also reduce the overall permeability E a of MMMs. Regarding 6FDA-DAM, in addition to polymer matrix compression at the high pressure, the overall CO 2 activation energy trend does not show a clear correlation to the membrane FFVs (MMMs (UiO-66; 0.331 > UiO-66-COCH 3 , 0.292 > UiO-66-NH 2 ; 0.277) > neat 6FDA-DAM, 0.238). Instead, the activation energy seems profoundly influenced by the presence of Zr-MOF nanoparticles in MMMs, in the order of their group functionalities (UiO-66-NH-COCH 3 > UiO-66-NH 2 > UiO-66 > neat 6FDA-DAM). It also concludes that the CO 2 permeation is predominately influenced by its solubility (sorption) in the membrane systems, and less dependent on temperature. The higher activation energies presented by the non-polar CH 4 also indicated that its diffusion or transport was more influenced compared to CO 2 molecules, giving higher CH 4 permeability increments and consequently reduced the CO 2 /CH 4 selectivity by 22-26%. This observation is also consistent with activated diffusion of non-polar molecules in glassy polymers (related to chain mobility and polymer free volumes) [68], where the least permeable gas often possesses higher activation energy and realizes a more substantial permeability increase with increasing temperature. In any event, the activation energies (temperature-dependent) are low for both the neat polymer membrane and the MMMs, compared to the other 6FDA-based polyimides in the literature (see Table S2). This suggests a low penetrant-membrane interaction perhaps because there is a relatively large difference between the CO 2 and CH 4 kinetic diameter and the membrane controlling pore size.

Effect of operating temperature in the high-pressure separation
Lower CO 2 temperature-dependency at this high-pressure separation also indicated by its fugacity coefficient values, closing to 1.0 (ideal  gas) when temperature is increased (see Fig. S4(a) and S4(b)), proves that the molecule's non-ideal behavior is less influenced by the increasing temperature but predominantly by pressure. It is supported by the fact that CO 2 possesses lower fugacity coefficients at the tested separation conditions Table 3 (overall compressibility factor and fugacity coefficient calculated values are presented in Fig. S4). The compressibility factors were determined by an eleven-constant Dranchuk and Abou-Kassem equation of state (DAK-EOS) [69]. The detail is presented in the supporting information document. Besides that, the CH 4 permeability increase was also influenced by the increase of polymer free volume (as a function of polymer chain packing and intersegmental motion) by the effect of elevated temperature. The activated diffusion often proves to be a significant advantage in the separation of non-polar H 2 from CO 2 , giving enhanced H 2 /CO 2 selectivity at higher temperatures as demonstrated in 6FDA-mPBI [68] and PBI-ZIF8 MMMs [70].
Regardless of common polymer chemical structures, Van Krevelen [71] presented a positive slope of 1 × 10 −3 for log P 0 and E p /R plot (Eq. (5)), with Z values of −7.0 and −8.2 for rubbery and glassy polymers respectively, for permeability measurement below their glass transition temperatures. Fig. S5 indicates that the addition of Zr-MOFs into 6FDA-DAM altered CO 2 permeability-temperature dependency significantly, giving a negative E p /R slope of −0.15 × 10 −3 , while only reduced CH 4 permeability-temperature dependency by roughly 70% (CH 4 permeability E p /R slope = 0.32 × 10 −3 ).

Effect of the presence of H 2 S on membrane separation
The concentration of H 2 S in the natural gas mixture varies depending on the geo-origin and can be more than 5 vol% [4,72]. As aforementioned, besides investigating the 6FDA-DAM and its Zr-MOF MMMs performances for H 2 S separation, it is important to understand the H 2 S effect on membrane performance. We studied the gas separation performance of 6FDA-DAM and its Zr-MOF MMMs with 30:70 vol% CO 2 :CH 4 feed mixture at 20 bar and 35°C, before adding 5 vol% of H 2 S, making the feed composition to 30:5:65 vol% CO 2 :H 2 S:CH 4 . The separation performance after H 2 S exposure was also investigated and summarized in Table 4.
Upon the addition of 5 vol% H 2 S in the mixed gas, P CO2 in all samples decreased by an average of 28-34%, accordingly to their functionality order: MMMs (UiO-66-NH-COCH 3 > UiO-66-NH 2 > UiO-66) > neat 6FDA-DAM. 6FDA-DAM MMMs showed a higher CO 2 permeability reduction in the presence of H 2 S, compared to the neat membrane. The observation exhibited the influence of Zr-MOFs in the MMMs, of which their active metal sites also preferentially adsorb H 2 S and thus reduce their CO 2 adsorption capacity. P H2S values are in the range of 137-352 Barrer, slightly lower than those of P CO2 , contributing to the total acid gas permeability of between 304 and 737 Barrer. The increments directly presented the acid gas selectivity over CH 4 of 16.4 for the neat 6FDA-DAM and in the range of 18.1-34.4 for its MMMs. Besides the competitive sorption of a two-component gas mixture, the presence of a third component intensifies the gas mixtures non-ideal behavior and influences each penetrant permeation rate, especially at elevated pressures [21]. Based on gas permeability values, the observed adsorption preference trend is in the order of CO 2 > H 2 S > CH 4 , well-agreed to the gasses' isosteric adsorption heat in UiO-66 (CO 2 ; 25.7 kJ·mol −1 > H 2 S; 23.8 kJ·mol −1 > CH 4 ; 18.8 kJ·mol −1 , reported at 30°C [36]). Functionalized UiO-66 derivatives presented higher values, in the same order. The gas physical properties; dipole moment (Debye), quadrupole moment (au) and polarizability (a 0 3 ), also greatly contributed to the competitive sorption outcomes and H 2 S high polarizability explained its higher permeability despite its relatively low content in the feed mixture compared to CO 2 ;  [73]. Hence, the observed α CO2/CH4 reduction can be explained by a larger competitive sorption effect induced by H 2 S (its solubility is larger than that of CH 4 ) in the membrane systems. In addition to H 2 S competitive sorption effect, the reduced CO 2 /CH 4 selectivity may also be contributed by the fact that CH 4 partial pressure in binary mixed gas (70 vol% in feed) is higher than that in ternary system (65 vol% in feed). As a higher CH 4 partial pressure will result in its higher permeability, subsequently lowers the CO 2 /CH 4 selectivity and its competitive sorption effect towards H 2 S and CO 2 permeability may also not be the same.
In the presence of H 2 S, all MMMs presented higher CO 2 , H 2 S and acid gas selectivities compared to the neat 6FDA-DAM (α CO2/CH4 = 9.1; Table 3 The compressibility Z factors for CO 2 and CH 4 , calculated using Dranchuk and Abou-Kassem equation of state (DAK -EOS) [69], presented at 35°C, 45°C and 55°C, at 20 bar.

Conclusion
6FDA-DAM polyimide offers an attractive opportunity in gas separation application, and the incorporation of the highly stable zirconium-based UiO-66 and its functionalized derivatives as MMM further enhanced the separation properties. The membranes possessed excellent CO 2 /CH 4 separation performance and presented high-performance stability at conditions relevant to actual gas processing (pressure, CO 2 content, temperature). The Zr-MOFs improved not only 6FDA-DAM gas separation properties but also deterred CO 2 -induced plasticization and swelling. Additionally, in the presence of high H 2 S content (50,000 ppm in feed mixture) at high total pressure, both CO 2and H 2 S-induced plasticization were suppressed, and only reversible competitive sorption effect was observed. This successful high-pressure testing of 6FDA-DAM MMMs with Zr-MOFs is encouraging and Abbreviation: 6FDA: 2,2-bis(3,4-dicarboxyphenyl)hexafluoropropane diandydride; DAM: 2,4,6-trimethyl-1,3-diaminobenzene; PEBAX: polyether block amide; PU: polyurethane; PIM: polymers of intrinsic microporosity; DABA: 3,5-diaminobenzoic acid. industrially relevant for natural gas sweetening at high pressure. Nevertheless, the separation understanding in the presence of water vapor and condensable hydrocarbons needs to be addressed beforehand. These impurities are not only suspected to reduce the separation performance but could also deteriorate the physical integrity of a membrane system.