Surface functionalized UiO-66/Pebax-based ultrathin composite hollow fiber gas separation membranes

Putu Doddy Sutrisna a, Jingwei Hou *ab, Muhammad Yazid Zulkifli a, Hongyu Li a, Yatao Zhang c, Weibin Liang d, Deanna M. D'Alessandro d and Vicki Chen a
aUNESCO Centre for Membrane Science and Technology, School of Chemical Engineering, University of New South Wales, Sydney, NSW 2052, Australia. E-mail: Jingwei.hou@unsw.edu.au
bDepartment of Materials Science and Metallurgy, University of Cambridge, Cambridge, CB3 0FS, UK
cSchool of Chemical Engineering and Energy, Zhengzhou University, Zhengzhou 450001, PR China
dSchool of Chemistry, The University of Sydney, Sydney, NSW 2006, Australia

Received 25th August 2017 , Accepted 1st November 2017

First published on 2nd November 2017


Pebax-based composite hollow fibre membranes are promising candidates for industrial gas separation, but their application is limited by the inherent separation performance of the polymeric materials and the poor operational stability especially under elevated pressures. The incorporation of metal–organic frameworks (MOFs) has been extensively investigated as a potential solution to these problems. However, the major challenges are to control the microvoids in the interfacial region and to improve the effective MOF loading within the selective layer. In this work, we applied a zirconium-based rigid MOF (UiO-66) to fabricate a nanocomposite hollow fibre membrane, and (–COOH) and (–NH2) modified UiO-66 were applied to investigate the effect of surface functionalization. Up to 80 wt% UiO-66 was incorporated into the thin Pebax selective layer, and both improved CO2 permeance and selectivity were obtained simultaneously with the (–NH) functionalized UiO-66. In addition, the presence of UiO-66 in Pebax significantly improved the membrane's operational stability under high pressure.


1 Introduction

Membrane-based gas separation processes are flexible, easy to operate and scale up, and require smaller footprints. They have therefore been the subject of considerable research attention, especially for CO2 capture from flue gas and natural gas sweetening.1–3 Polymeric membranes are considered as the most promising candidate for industrial application, yet their application is limited by the inherent performance of the polymeric materials and poor operational stability, especially under high pressure. To tackle these problems, different nanofillers have been incorporated into the polymeric matrix to fabricate mixed matrix membranes (MMMs). Recently, metal–organic frameworks (MOFs) have been extensively investigated for gas adsorption and separation.4,5 MOFs are heralded for their chemical and structural versatility in being able to accommodate an enormous amount of functionalities: molecular sieve pores can be effectively incorporated into polymeric materials to promote the separation performance.4,6,7 Some MMMs have their performance surpassing the Robeson upper bound which depicts the gas permeability–selectivity trade-off for pure polymeric materials.8–13

Currently, most MMMs are synthesized as dense flat sheets.14–20 These membranes, from the material's perspective, can have superior permeability. Yet this does not necessarily indicate that the membrane can readily achieve high permeation flux (permeance), which is of higher practical importance. For gas separation membranes, the thickness does matter. Ideally, hollow fiber composite membranes with large surface areas and thin selective layers are preferred.21 However, the behaviour of MOF–polymer membranes in their thin film can be different from that of their bulkier, thick counterparts. This is due to the difficulty in achieving a homogeneous MOF dispersion and a higher possibility of generating extra non-selective defects. In addition, the polymer chain rigidification effect caused by interfacial interaction can be more significant.22 Among different MOF materials, ZIF-8 has been extensively investigated due to its small pore size (3.4 Å) and reasonably good chemical stability.23,24 In one of our previous studies, the introduction of in-house synthesized ZIF-8 into a thin Pebax layer clearly generated extra defects due to the poor compatibility in the interfacial region, and the loss of gas selectivity was more significant for the thin selective layer on the hollow fiber composite membranes compared with thick mixed matrix membranes.25 To tackle this problem, polydopamine modification has been carried out to fine-tune the interfacial compatibility: compared with the pristine ZIF-8, the membranes with modified ZIF-8 exhibited improved gas selectivity, which however were still lower than that of the original pure polymer membrane.15

Recent studies have started to shed light on the mechanical properties of MOFs.26 Among different MOF materials, ZIF-8 has a low elastic modulus and hardness, indicating its good elasticity in the family of crystalline materials. We also first demonstrated that a thin, continuous ZIF-8 membrane can maintain a certain degree of flexibility: it can sustain bending and elongation without compromising its molecular sieving capability.5 The rotational and vibrational movement of the imidazole linkers in the ZIF has also been experimentally demonstrated by Terahertz FTIR with synchrotron beamlines.27 This interesting property, however, may compromise their inherent molecular sieving capability based on their crystalline framework structures.28 As a result, the incorporation of such “soft” MOFs into the thin composite membrane can potentially lead to the formation of non-selective interfacial microvoids, or its flexible framework structure can allow the passage of gas molecules which are larger than its pore aperture.29 This problem can potentially be solved by blending more rigid MOFs into the membrane. Another problem for the MOF-containing membranes is their relatively low MOF loading, as aggregation of the particles at high concentration (over 30 wt%) can create defects in membranes: for most existing nanocomposite membranes, only up to 30 wt% of MOFs can be achieved without significant deterioration in the selectivity.17–19,30

In this study, we fabricate our composite membrane with different types of MOFs, namely University of Oslo-66 (UiO-66) and zeolitic imidazolate framework-7 (ZIF-7). UiO-66 ([Zr6O4(OH)4(bdc)6] where bdc = 1,4-benzenedicarboxlate) is considered as one of the most rigid MOFs due to the high degree of coordination of the Zr–O node to organic ligands.31 Its structure is comprised of a central octahedral cage with 8 face-sharing super-tetrahedra (ST). The four vertices of the ST are each occupied by a Zr6-oxo cluster with edges formed by the bdc ligands.32 Its satisfactory thermal and chemical stabilities over a wide range of temperature and pH make it a suitable candidate for membrane application, both as a filler and a coherent selective layer.33–35 In addition, by substituting the bdc ligand with other linear dicarboxylate linkers, a series of isoreticular frameworks with different cavity sizes and functionalities can be obtained (Scheme 1). Enhancement in the gas separation performance in the presence of functional groups such as amine on UiO-66/Pebax flat sheet mixed matrix membranes has also been reported,30 but their effect on the composite membrane performance has never been investigated. Another targeted MOF is ZIF-7 ([Zn(BIm)2] where BIm = 2-benzimidazolate) which has an aperture size of ∼3.0 angstroms, smaller than that in ZIF-8 (∼3.4 angstroms). Although ZIF-7 possesses a similar structure to ZIF-8, it has higher rigidity.26 Thus, ZIF-7 is considered to be an appropriate candidate to investigate the effect of the structural rigidity of a MOF on the membrane performance, and a direct comparison can be made against ZIF-8.


image file: c7ta07512j-s1.tif
Scheme 1 Schematic diagram of the composite membrane and the chemical structure of UiO-66.

In this work, UiO-66, UiO-66-NH2, UiO-66-(COOH)2 and ZIF-7 were incorporated into the thin Pebax selective layer of composite hollow fiber membranes. The copolymer Pebax, which features a hard polyamide (PA) segment and a soft polyethylene oxide (PEO) segment, has shown good performance for CO2 gas separation and has exhibited plasticization resistance to CO2 gas. The interaction between the copolymer and MOF additives can be complex and therefore warrants detailed investigation. Membranes containing up to 80 wt% of UiO-66 and 30 wt% of ZIF-7 were examined to understand their effect on the CO2, N2 and CH4 permeances. Both pure gas and mixed gases were investigated. In addition, a complete pressurization–depressurization cycle was applied to investigate the effect of MOFs on the membrane plasticization and compaction resistance.

2 Materials and methods

2.1 Materials

Hollow fiber PVDF porous membranes were kindly supplied by Beijing OriginWater Technology Co., Ltd. (China) with a diameter of 1.0 mm, a wall thickness of 0.25 mm and a pore size of ∼0.05 μm. Poly[1-(trimethylsilyl)prop-1-yne] (PTMSP) utilized as the gutter layer was provided by Gelest, Inc., PA, USA. Polyether oxide–polyamide (PEO–PA) blocks (Pebax-1657) supplied by Arkema, France were used as the polymer matrix for the selective layer. The Pebax-1657 applied in this work contained 60% rubbery PEO phase and 40% glassy PA phase. Chemicals for ZIF-7 and UiO-66 particle synthesis were supplied by Sigma-Aldrich, including benzimidazole, zinc nitrate hexahydrate (for ZIF-7), ZrCl4 (>99.5%), and ligands (H2bdc (1,4-benzendicaroboxylic acid, 98%) for UiO-66, H2bdc-NH2 (2-amino-1,4-benzendicarboxylic acid, 99%) for UiO-66-NH2, and H2bdc-(COOH)2 (1,2,4,5-benzenetetracarboxylic acid, 96%) for UiO-66-(COOH)2). Hydrochloric acid (HCl, 32%) and N,N′-dimethylformamide (DMF, > 98%) were obtained from Merck. CO2, CH4 and N2 pure gases and mixed gases (CO2/N2, 20[thin space (1/6-em)]:[thin space (1/6-em)]80 v/v and CO2/CH4, 20[thin space (1/6-em)]:[thin space (1/6-em)]80 v/v) for the gas permeation test were purchased from Coregas. All other chemicals were used without further purification.

2.2 Synthesis of UiO-66 and ZIF-7 particles

ZIF-7 particles were synthesized at room temperature as described elsewhere with slight modification.36,37 In a typical procedure, a mixture of 0.453 g zinc nitrate hexahydrate, 1.154 g benzimidazole and 150 ml DMF was stirred for 24 h at room temperature. The milky solution was subsequently centrifuged for 15 min at 13[thin space (1/6-em)]000 rpm. ZIF-7 particles were then washed and centrifuged with DMF for another 2 cycles, before being washed with methanol for another 24 h to remove DMF. The final product of the ZIF-7 particles was recovered by centrifugation and mixed with an ethanol/water mixture (70/30 w/w or 74.7/25.3 v/v) for membrane fabrication. For the particle characterization, the particles were dried for 24 h at 85 °C.

The UiO-66 and functionalized UiO-66 particles were synthesized by microwave-assisted synthesis using an Anton Paar Monowave 300 microwave oven.38,39 A 30 ml glass microwave vial was filled with ZrCl4 and the appropriate ligand, H2bdc (1,4-benzendicaroboxylic acid) for UiO-66, H2bdc-NH2 (2-amino-1,4-benzendicarboxylic acid) for UiO-66-NH2, and H2bdc-(COOH)2 (1,2,4,5-benzene tetracarboxylic acid) for UiO-66-(COOH)2. Thereafter, concentrated HCl, formic acid and DMF were introduced into the vial. The mixture was heated with magnetic stirring (600 rpm) to 160 °C within 30 min and held at this temperature for 40 min before the final cooling to 55 °C. The precipitates were obtained by vacuum filtration and repeatedly washed with DMF (3 × 20 ml) and acetone (5 × 20 ml). The materials were subsequently solvent exchanged with methanol using a Soxhlet washing procedure for 10 h. The resulting powders were dried under vacuum. The yields for UiO-66 were as follows: UiO-66: 88.9%, UiO-66-NH2: 91.0% and UiO-66-(COOH)2: 90.3%.

2.3 Fabrication of hollow fiber nanocomposite membranes

In this work, supporting PVDF membranes with relatively large pores were selected to ensure high mass transfer efficiency. To mitigate the coating layer intrusion into the pores, the PVDF membranes were firstly soaked overnight in deionized (DI) water and then briefly wiped with a paper tissue to remove the water on the membrane surface, but leaving the membrane pores occupied with water. Both ends of the fibers were clamped with longtail clips to prevent coating solution intrusion into the lumen side during the dip coating process. The outer surface of fibers was firstly dip-coated with a highly permeable gutter layer four times using 2 wt% PTMSP in n-hexane solution. The smooth surface can facilitate the subsequent coating of a thin and continuous selective layer. The selective layer of the MOFs/Pebax-1657-based composite membrane was prepared by dispersing nanoparticles in Pebax solution (3 wt% Pebax in 70/30 w/w ethanol/water). Two coating cycles were applied for the selective layer. Finally, a pure Pebax-1657 layer was coated as a protective layer to seal any possible defects.

To ensure good MOF dispersion, the particles were first primed with the Pebax-1657 solution. Then the mixture was probe sonicated. For the dip-coating procedure, the fibers were turned upside down after each coating cycle and dried in an oven at 50 °C. More details of the dip-coating procedure can be found in our previous publication.25 For the subsequent gas permeation tests, three hollow fibers were housed in a 1/4 inch and 18 cm long stainless steel module with an effective membrane area of 17 cm2.

2.4 Characterization of the nanoparticles and membranes

ZIF-7 and UiO-66 particles were analyzed using a Transmission Electron Microscope (TEM), FEI Tecnai G2 20, for imaging purposes. The particle size was monitored using a dynamic light scattering (DLS) device (Malvern Nano DLS). For each test, at least 13 cycles of reading were performed to minimize error. The Brunauer–Emmett–Teller (BET) surface area, Barrett–Joyner–Halenda (BJH) pore volume, and N2 and CO2 sorption isotherms were recorded on an Accelerated Surface Area and Porosimetry System, ASAP 2020 (Micromeritics Instruments Inc.). Approximately 80 mg of the powdered solid was loaded into a glass analysis tube and dehumidified for 3 h under dynamic vacuum at 150 °C. N2 adsorption and desorption isotherms were measured at 77 K while CO2 adsorption and desorption isotherms were measured at 298.15 K. The chemical structure of the particles was analyzed using a Fourier Transform Infra-Red (FTIR) Alpha spectrometer from 400 to 4000 cm−1. The crystallinity of the particles was examined using a PANalytical Empyrean Thin-Film X-Ray Diffraction (XRD) instrument in the 2θ range from 4 to 36° with a 0.026° step size.

Characterization of the composite membranes was conducted using Scanning Electron Microscopy (SEM), Energy Dispersive X-ray (EDX) analysis, Differential Scanning Calorimetry (DSC), FTIR and XRD techniques. The surface and cross-sectional areas of the membranes were examined under an FEI Nova NanoSEM 450 FESEM after the membrane sample was coated with a layer of chromium. The presence of nanoparticles and the quality of dispersion in the membrane matrix were examined by EDX line scans (FEI Nova NanoSEM 450 FESEM). The samples were coated with a layer of carbon prior to the EDX tests. The degree of crystallinity of the membranes was determined by DSC analysis. The Pebax-based membranes were tested with a Mettler Toledo DSC 823e analyzer from −30 °C to 400 °C in two cycles. The crystallinity of the membranes was analyzed using a PANalytical Empyrean Thin-Film XRD instrument in the 2θ range from 4 to 36° with a 0.026° step size. The chemical structure of the membranes was analyzed using a Fourier Transform Infra-Red (FTIR) Alpha spectrometer from 400 to 4000 cm−1. Tensile strength tests were carried out with a textural analyzer (TAXT2, Stable Micro Systems). The sample length was 100 mm and the testing speed was 0.5 mm s−1.

2.5 Gas permeation testing

The gas permeation tests using pure gas and mixed gases to measure the initial performance of the membrane were carried out at room temperature (around 25 °C). The volumetric flow rate of the permeate line was measured using a bubble flow meter for low permeate flow rates (<1 ml min−1), while flow rates higher than 1 ml min−1 were recorded using an Agilent ADM1000 gas flow meter. The pure gas permeance was calculated using eqn (1):
 
image file: c7ta07512j-t1.tif(1)
where P/l is the gas permeance through the membrane, Q is the volumetric flow rate of the permeate line (ml s−1), Δp is the pressure difference across the membrane (cm Hg), and A is the membrane surface area (cm2).

The ideal selectivity of the membrane for a given gas pair was calculated from the ratio of the permeance of fast gas (A) to that of slow gas (B) based on eqn (2):

 
image file: c7ta07512j-t2.tif(2)

The operational stability of the composite membranes was studied under different feed pressures. The experiments were conducted by exposing the membranes to different feed pressures in a full pressurization–depressurization cycle from 2 to 15 bar. The feed gas pressure was increased and decreased stepwise. Under each pressure, the membrane was exposed to the feed gas for 1 hour for sufficient equilibration. The permeability of CO2 was firstly tested, followed by a CH4 test to understand the effect of the condensable gas on the permeation behavior of the non-condensable gas.

To study the effect of competitive sorption during gas permeation, mixed gas permeation tests were also carried out using CO2/CH4 and CO2/N2 (20/80, v/v) gas mixtures as feeds. The permeate composition was analyzed with a Shimadzu gas chromatograph (Shimadzu GC-2014) using a TCD and the mixed-gas permeability was calculated using eqn (3):

 
image file: c7ta07512j-t3.tif(3)
where px and py are the pressures of the feed and permeate, A is the membrane area, and X and Y are the concentrations in the feed and permeate sides. The selectivity of the membrane for mixed gases was calculated using eqn (4):40,41
 
image file: c7ta07512j-t4.tif(4)

3 Results and discussion

3.1 Characterization of UiO-66 and ZIF-7 particles

The morphologies of UiO-66 and ZIF-7 nanoparticles were examined by TEM (Fig. 1). Both UiO-66 and functionalized UiO-66 showed rectangular shapes with particle sizes of around 100–200 nm, which aligns with the DLS results in Table 1. This observation suggests that the different organic ligands did not change the crystal structure or particle sizes of UiO-66. This is preferable to help us understand the effect of the intrinsic MOF properties, rather than the particle size and crystalline structure, on the final composite membrane performance. In terms of the ZIF-7 nanoparticles, the morphology of such particles showed a rhombic dodecahedral shape with a particle size of around 150 nm. Such an observation is consistent with previous literature.37
image file: c7ta07512j-f1.tif
Fig. 1 TEM images of (a) UiO-66, (b) UiO-66-NH2, (c) UiO-66-(COOH)2 and (d) ZIF-7. The scale bar is 100 nm.
Table 1 Average particle size, BET surface area and BJH pore volume of various UiO-66 and ZIF-7 particles
Type of particle DLS results (nm) BET surface area (m2 g−1) BJH pore volume (cm3/g−1) Adsorption average pore diameter (nm)
UiO-66 130 ± 15 1800 ± 2.5 0.68 1.5
UiO-66-NH2 146 ± 15 1213 ± 6 0.48 1.45
UiO-66-(COOH)2 150 ± 10 400 ± 0.8 0.17 1.33
ZIF-7 171 ± 13 469 ± 2 0.39 1.12


The particles were analyzed for their BET surface area, BJH pore volume and adsorption average pore diameter, with the results presented in Table 1. The ZIF-7 synthesized in this research had a BET surface area of around 469 m2 g−1, which is in agreement with a previous study.37 The surface area of ZIF-7 was smaller than that of ZIF-8 reported previously,25 possibly due to the smaller accessible window apertures in ZIF-7 (∼3.0 angstroms) compared with ZIF-8 particles (∼3.4 angstroms). The BET surface area of UiO-66 particles was around 1800 m2 g−1. The high surface area confirmed the effectiveness of the solvent exchange process during particle synthesis. The BET surface area, BJH pore volume and pore diameter of the UiO-66 were higher than those of the functionalized UiO-66. This is consistent with the presence of functional groups on the organic ligands which partially block the pores in UiO-66.42,43 Particles with the bulkier –(COOH)2 functional group showed the smallest BET surface area and BJH pore volume among all the UiO-66 derivatives.

To further explore the effect of functional groups on the UiO-66 gas adsorption process, the adsorption–desorption isotherms were investigated for both N2 and CO2 (Fig. 2). Functionalization of the organic ligands has two effects: on the one hand, it could reduce the UiO-66 pore size and lead to a reduced BET surface area as discussed above. For the –COOH functionalized MOFs, each organic ligand contains two –COOH groups, leading to significant pore blockage and subsequently low adsorption capacity. On the other hand, the presence of polar moieties such as –NH2 can improve the adsorption capacity for CO2, resulting in the highest CO2 adsorption for UiO-66-(NH2), even though its surface area is smaller than that of UiO-66.


image file: c7ta07512j-f2.tif
Fig. 2 (a) N2 and (b) CO2 adsorption for UiO-66 derivatives and ZIF-7 particles. The filled symbols represent adsorption and the open symbols represent desorption. N2 adsorption and desorption isotherms were measured at 77 K while CO2 adsorption and desorption isotherms were measured at 298.15 K. Both measured pressures were up to 1 bar.

The chemical properties of the MOFs synthesized in this research were analyzed using FTIR analysis and the results are presented in Fig. S1. The FTIR spectra of UiO-66 and its derivatives show a weak band at 1660 cm−1, which was assigned to the stretching vibrations of C[double bond, length as m-dash]O in the carboxylic acid present in the bdc ligand. The asymmetric stretching vibration observed at around 1585 cm−1 originated from O–C–O asymmetric stretching in the bdc ligand. The vibration of C[double bond, length as m-dash]C in the benzene ring can be observed at 1506 cm−1 and the O–C–O symmetric stretching vibration in the carboxylate group of the bdc ligand was observed as a small band at 1395 cm−1.44,45 For UiO-66-NH2, the presence of primary amino functional groups (N–H) was confirmed by the small peaks at 3367 and 3475 cm−1.46 The peak at 1710 cm−1 for UiO-66-(COOH)2 is consistent with a free carboxylic acid C[double bond, length as m-dash]O stretching vibration.47 For ZIF-7 particles, signature peaks at 1455 and 750 cm−1 were observed in Fig. S1c, corresponding to the C[double bond, length as m-dash]C and C–H bonds in the benzene functional group of benzimidazole as the organic ligand.48

XRD analysis was then carried out to investigate the crystallinity of the UiO-66 and ZIF-7 particles as shown in Fig. S2. In terms of UiO-66 and its derivatives, the XRD pattern of UiO-66 in Fig. S2a shows signature peaks at 2θ = 7.28°, 8.42° and 25.63°. These were in accordance with the peak positions observed in other studies on UiO-66 particle synthesis.49 After the incorporation of the amine –(NH2) and –(COOH)2 functionalities, the positions of the peaks were largely invariant. This indicates the preservation of crystallinity of the UiO-66 particles despite the change in chemical moieties on the organic ligands. The crystallinity of ZIF-7 was confirmed from the powder X-ray diffraction patterns shown in Fig. S2b. The pattern aligned well with literature values.37

3.2 Characterization of UiO-66/Pebax-based hollow fiber composite membranes

The performance of the composite membrane is determined by the intrinsic properties of the materials as well as the interfacial interactions.50 In this work, the composite membranes consisted of a PVDF porous support, a layer of PTMSP as a gutter layer, a Pebax-1657-based selective layer (containing different amounts of MOFs) and a top pure Pebax-1657 protective layer. Each of these layers contributes to the overall gas separation performance of the membranes.
3.2.1 Morphology of the composite membranes. Firstly, we examined the membrane cross-sectional morphology. As shown in Fig. 3, the thickness of the PTMSP gutter layer was approximately 6 to 7 μm. This is the optimum thickness that can ensure even and uniform coverage of the PVDF support as has been investigated in our previous study.25 Uniform and continuous coverage of the porous support with PTMSP will prevent the intrusion of the selective layer into the pores of the porous support, ensuring an even and thin selective layer for higher permeation flux.
image file: c7ta07512j-f3.tif
Fig. 3 Cross-sectional SEM images of composite membranes with different coating layers: (a) PTMSP, (b) PTMSP and pure Pebax, (c) PTMSP and 50 wt% UiO-66 in Pebax, (d) PTMSP and 50 wt% UiO-66-NH2 in Pebax, (e) PTMSP and 50 wt% UiO-66-(COOH)2 in Pebax and (f) PTMSP and 80 wt% UiO-66 in Pebax. Membranes (b–f) have an extra top Pebax protective layer.

The further deposition of a selective layer and a protective layer increased the thickness of the membranes by around 1–1.5 μm. However, the distinct boundary between the gutter and selective layers cannot be observed clearly under a SEM due to the interfusion between these two layers. This has also been observed in our previous study.25 It was difficult to obtain high-resolution cross-sectional images of the nanocomposite hollow fiber membranes: due to the poor conductivity of the porous substrates and their tubular shape, a significant electron beam drifting can occur at high magnification. Based on the SEM images presented here, the presence of UiO-66 (Fig. 3c–e) can be observed in the top region of the composite membrane (as highlighted in the images). When the UiO-66 loading was at 80 wt%, a significant nanoparticle aggregation was observed on the membrane surface (Fig. 3f). It should be noted that for conventional mixed matrix and nanocomposite membranes, the loading of the MOF filler is limited due to the difficulty in maintaining a uniform and even distribution within the polymeric matrix. The MOF aggregation can potentially lead to non-selective defects and particle loss during the fabrication process. As a result, most previous work has only studied a relatively low particle loading (<30 wt%).25,30 In this work, up to 80 wt% of MOFs were blended into the Pebax dip-coating solution. At low to moderate UiO-66 loadings (10–50 wt%), the particles were evenly distributed without significant aggregation, and clear UiO-66 aggregation was only observed at 80 wt% MOF loading (Fig. S3 and 3). In addition, it should be noted that compared with the composite membrane with a pure Pebax layer, the composite membrane containing 80 wt% of UiO-66 had comparable tensile strength (elongation at break: 35% and 37% and tensile strength at break: 3.1 and 2.9 MPa, respectively), indicating a good compatibility between the MOF and Pebax.

To confirm the presence of UiO-66 particles (or those of its derivatives) inside the composite membranes, EDX scanning of Zr was conducted. Typical results for UiO-66 are presented in Fig. 4. The Zr content of UiO-66 particles can be observed near the top surface of the composite membrane with a thickness of around 1–1.5 μm for all samples. This confirms the observation from SEM that the thickness beyond the PTMSP gutter layer was around 1–1.5 μm. The Zr signal strength within the selective layer increased with a higher particle loading (from 10 wt% to 80 wt%). The EDX scan can provide a rough indication of the selective layer thickness. However, it must be noted that the exact thickness of the selective layer and the top protective layer was difficult to measure due to the interfusion of the PTMSP and Pebax layers during the repeated coating process. In addition, for a membrane with high UiO-66 loading (80 wt%), it is difficult to accurately determine its selective layer thickness due to the nanofiller aggregation. Based on the SEM image (Fig. 3), the UiO-66 aggregates can be up to 10 μm on the membrane surface.


image file: c7ta07512j-f4.tif
Fig. 4 EDX linear scanning of the Zr content of the cross section of membranes containing different UiO-66 loadings.
3.2.2 The crystallinity of the composite membrane. XRD analysis was performed to determine the crystal structure of the membranes. Four common crystal phases (α, β, γ, and δ) have been reported for PVDF.51 The crystallinity of PVDF is one of the most important factors influencing its mechanical properties. The XRD patterns of the composite membranes are shown in Fig. S4, and the observed characteristic peak at θ = 20.17° for all membranes originated from the PVDF supporting membrane (β phase).5 After the membrane coating with a PTMSP gutter layer, a broad signature peak of the PTMSP polymer at 10° was observed. The signature peak of Pebax-1657 at 24.1° for the crystalline polyamide phase and the broader peak of the polyethylene oxide phase from 17.5° to 22.5° could not be observed clearly from the XRD patterns of the composite membranes.25 This can be mainly attributed to a thin Pebax layer as well as their overlap with the peaks of PVDF. To better highlight the preservation of UiO-66 crystallinity within Pebax, we conducted XRD analysis with UiO-66/Pebax-1657 mixed matrix films (Fig. 5). It shows that after the addition of UiO-66 nanoparticles, the signature peaks of UiO-66 (Fig. S2) were clearly observed, indicating the preservation of crystallinity of particles inside the layer for the mixed matrix films.
image file: c7ta07512j-f5.tif
Fig. 5 XRD patterns of the different UiO-66/Pebax-1657 mixed matrix membranes.

DSC analysis was also carried out to study the evolution of the melting point (Tm) of the composite membrane (Table 2). The pure PVDF membrane had a Tm of around 157 °C and after the coating of PTMSP and Pebax layers, the Tm value increased to a slightly higher value. After incorporation of the UiO-66, the melting point further increased to around 165 °C, indicating a good interfacial compatibility between UiO-66 and its surrounding Pebax matrix, which could rigidify the polymer chains.

Table 2 Melting point (Tm) of the UiO-66/Pebax-1657-based composite membranes (typical error of Tm was around 0.5 °C)
Type of membrane T m (°C)
Pure PVDF 157
PTMSP coated 159
PTMSP + Pebax coated 160
50 wt% UiO-66/Pebax coated 164
50 wt% UiO-66-NH2/Pebax coated 165
50 wt% UiO-66-(COOH)2/Pebax coated 164


It is important to explore the effect of UiO-66 on the polymeric structure of the Pebax selective layer. However, it can be challenging to characterize the thin selective layer within the composite membrane. Therefore, we fabricated a series of thick mixed matrix membranes with UiO-66 and Pebax. The degree of crystallinity in both soft and hard phases of Pebax was analysed using DSC and estimated using eqn (5):

 
image file: c7ta07512j-t5.tif(5)
where ΔHm is the tested melting enthalpy of a certain polymeric section and ΔH0m is the melting enthalpy of its fully crystalline form. The melting enthalpy (ΔHm) was estimated from the area of the melting peak in the DSC curves, while the melting enthalpy of the pure crystalline phase (ΔH0m) of PEO is 166.4 J g−1 and PA is 230 J g−1.52 The degrees of crystallinity of the PA and PEO phases of pure Pebax and UiO-66/Pebax are presented in Table 3. After the incorporation of UiO-66, especially the functionalized UiO-66 nanoparticles, both PA and PEO phases showed enhancements in crystallinity. This indicates an increase in the rigidity of the polymer matrix after the incorporation of particles due to the potential formation of hydrogen bonds between polymeric chains and UiO-66.

Table 3 The degree of crystallinity of PA and PEO phases in Pebax-based dense mixed matrix membranes
Type of membrane Integral of melting peak (J g−1) X PEO (%) X PA (%) X c (total) (%)
PEO block PA block
Pure Pebax 14.72 21.1 14.74 22.93 18.02
50 wt% UiO-66/Pebax 18.10 22.43 18.13 24.38 20.63
50 wt% UiO-66-(COOH)2/Pebax 19.46 23.06 19.49 25.07 21.72
50 wt% UiO-66-NH2/Pebax 19.89 23.19 19.92 25.22 22.04


3.2.3 Chemical properties of the UiO-66/Pebax-1657 composite membranes. To examine the chemical structure of pure PVDF and the composite membranes, FTIR analysis was conducted. The FTIR spectra of pure PVDF and typical composite membranes incorporating 50 wt% UiO-66 are depicted in Fig. 6. The peak for pure PVDF at 1403 cm−1 was attributed to CH2 vibrations, and the C–C band of PVDF was observed at 1185 cm−1. The CF2 stretching vibration and bending modes were observed as peaks at around 745–840 and 510 cm−1, respectively. The bands located at around 3022 and 2980 cm−1 corresponded to the CH2 asymmetric and symmetric vibration of PVDF.51 After deposition of the Pebax-1657 mixture, the stretching vibration of the C–O–C group in the PEO segment of Pebax appeared as a distinct peak at around 1094 cm−1. In addition, the PA segment in Pebax exhibited additional peaks at 3297 cm−1 for –N–H–, 1636 cm−1 for H–N–C[double bond, length as m-dash]O and 1730 cm−1 for O–C[double bond, length as m-dash]O groups.52,53 Previous research studies have demonstrated that MOFs can form hydrogen bonds with the glassy section of Pebax.25,30 But the N–H– peak shift for the PA segment is negligible in this work (Fig. 6b), possibly due to the difficulty in characterizing the thin composite layer.
image file: c7ta07512j-f6.tif
Fig. 6 (a) FTIR spectra of the different UiO-66/Pebax-1657 nanocomposite membranes and (b) the enlarged wavenumber range between 3000 and 3400 cm−1.

3.3 The gas separation performance of various UiO-66-based membranes

In the case of mixed matrix membranes, it is relatively difficult to incorporate a large percentage of nanofillers into the matrix due to the difficulty in obtaining a homogeneous dispersion, as well as the loss in mechanical strength. In the present work, we incorporated up to 80 wt% of UiO-66 into the thin Pebax layer and the gas separation performance was investigated with both pure gases and mixed gases. Both PDMS and PTMSP were applied as gutter layers and the results are summarized in Table S1. PTMSP clearly had higher CO2 permeance and selectivity. As a result, PTMSP was applied for membrane fabrication in the subsequent tests. A smooth gutter layer can prevent the intrusion of Pebax into supportive membrane pores, ensuring a thin and continuous selective layer. Fig. 7 presents the gas permeation performance of UiO-66/Pebax-1657-based composite membranes with various particle loadings. In terms of the composite membrane containing pure Pebax as the selective layer, the CO2 permeability was much higher than that of other gases such as N2 and CH4, which could be attributed to the rubbery PEO block in Pebax which has a strong affinity to polar gas (CO2) over non-polar gases (N2 and CH4).54,55
image file: c7ta07512j-f7.tif
Fig. 7 Gas separation performance of UiO-66/Pebax-1657-based composite membranes at different particle loadings: (a) CO2 permeance, (b) CO2/N2 gas selectivity and (c) CO2/CH4 gas selectivity (solid lines represent pure gas and dashed lines represent mixed gas).

Compared with the pure Pebax membrane, the incorporation of UiO-66 into the selective layer only slightly increased the CO2 permeance, even for the membrane containing 80 wt% of UiO-66. These results are unexpected considering the large BET surface area of UiO-66 and its relatively wide BJH pore diameter range compared with the target CO2 molecules (Table 1). Within the thin composite selective layer, the gas transport can be affected by the polymeric sections, MOF structures and MOF–polymer interfacial regions.25,33,55 The relatively unchanged CO2 permeance indicates a good interfacial compatibility; otherwise rapid gas transport can occur via the non-selective defect voids throughout the thin layer. As suggested in Table 3, the addition of UiO-66 into Pebax can effectively increase the crystallinity and thereby rigidify the polymer chains, which slows the gas transport. As a result, the CO2 permeance only slightly increased with the addition of UiO-66, even though its inherent highly porous structure can facilitate rapid gas molecule transport. In terms of the gas selectivity (Fig. 7b and c), for the membrane with UiO-66, the highest selectivity was observed with a nanofiller loading of 50 wt% for both CO2/N2 and CO2/CH4. It should be noted that the increased selectivity should mainly originate from the increased polymer chain rigidity as the pore size of UiO-66 (Table 1) is unlikely to provide effective molecular sieving for CO2 against N2 and CH4.

The functionalization of UiO-66 can have a significant effect on the composite membrane performance. Compared with UiO-66, much higher CO2 permeance results were obtained for all membranes with different UiO-66-NH2 nanofiller loadings. Based on our previous gas adsorption results, the presence of amine groups on the organic ligands appear to have improved the selective CO2 uptake for UiO-66-NH2, leading to higher CO2 solubility within the thin selective layer.

On the other hand, for the UiO-66-(COOH)2 nanofillers, the MOF pore blockage by the carboxylate functional groups limited its gas adsorption within the framework structure, such that its incorporation into the Pebax matrix had a relatively minor effect on CO2 permeance, similar to the case of the UiO-66. For both functionalized MOFs, the highest selectivities were obtained at 50 wt% loading, suggesting good compatibility between nanofillers and the polymeric matrix. Based on the gas permeation results with the functionalized UiO-66, we can further confirm that the composite membrane performance is dominated by the interactions between nanofillers and the polymer matrix: the highest gas permeance and selectivity were obtained with UiO-66-HN2, the BET surface area of which was significantly lower than that of UiO-66. In addition, even though the COOH functionalized UiO-66 had the smallest BJH pore size, the functionalized MOF pore was still too large for effective molecular sieving; therefore the shrinkage in pore size did not effectively lead to an improved selectivity.56 In this work, we also conducted the Maxwell model calculation for the composite membranes. However, the model failed to accurately predict the composite membrane performance, possibly because the effect of interfacial region was not considered by the Maxwell model. A more detailed discussion and Maxwell model results are presented in the ESI (Fig. S5).

3.4 The effects of operating pressure and plasticization on gas separation performance

Membrane plasticization caused by polymer chain swelling is a common phenomenon for gas separation membranes when exposed to aggressive feed conditions,57,58e.g., CO2 separation from CH4 in natural gas treatment. The condensable gas increases the mobility of the polymer chain segments, which further increases the diffusion coefficients of all penetrants through the polymer membrane. This eventually leads to the increase of permeability and loss of selectivity. For a polymer membrane, the plasticization pressure is defined as the pressure at which the gas permeability increases with increasing feed pressure.59

Pebax has been thoroughly investigated as a promising candidate for industrial CO2 separation.17,37,60 However, the behavior of the Pebax-based mixed matrix membrane under elevated operating pressures has not been fully elucidated, especially for the thin Pebax layer within a composite hollow fiber membrane. In this study, we exposed the UiO-66/Pebax composite membrane to different feed pressures in a complete pressurization–depressurization cycle to investigate the reversibility of the polymer matrix swelling/compaction effect. In order to understand the effect of membrane plasticization on the non-condensable gas permeation, CH4 permeation tests were also carried out upon completion of the CO2 permeation experiment under each tested pressure.61 The best-performing membranes (containing 50 wt% of UiO-66 and its derivatives) were tested here.

The gas permeation results for the UiO-66/Pebax-based composite membranes are presented in Fig. 8 and S6. The experiment was carried out at up to 15 bar, as further increasing the feed pressure can lead to the collapse of the supporting membranes. In general, the membranes showed a decrease in CO2 permeability with the increase of feed pressure. Previous studies have revealed that gas transport through pure Pebax-based membranes occurs mostly through the soft PEO block.54,55 Therefore, the effect of pressure on the gas permeability was based on the competing effects of hydrostatic pressure and plasticization.25,55 CO2 gas permeance through the pure Pebax-coated membrane at elevated pressure showed a relatively severe decrease, to almost half of its initial value at 2 bar, due to a compaction effect of the polymer matrix. The gas permeance did not return to its initial value during the depressurization process indicating that the polymer matrix was irreversibly compacted. In addition, up to 15 bar, the membranes did not show an obvious plasticization effect as observed in the glassy polymer.


image file: c7ta07512j-f8.tif
Fig. 8 The effect of feed pressure on (a) CO2 permeance, and (b) CO2/CH4 selectivity of various UiO-66/Pebax-1657-based composite membranes (solid lines represent the pressurization step and dashed lines represent the depressurization step).

The incorporation of UiO-66 and its derivatives helped to reduce the compaction effect as can be seen from the relatively stable gas permeance at different feed pressures. Our previous study on ZIF-8/Pebax-1657-based composite membranes showed a similar phenomenon.25 The increase in the rigidity of the polymer matrix helped to increase the compaction resistance of the membranes incorporating UiO-66 particles. The increase in the rigidity of the membranes was reflected in the increase in melting point and the degree of crystallinity as shown in Tables 2 and 3. All the membranes incorporating particles showed good compaction resistance during the experiment.

CH4 gas permeation tests were also conducted directly after the CO2 permeation test at each pressure to analyse the plasticization/compaction effect on the permeation of the non-condensable gas. The results of the CH4 permeance and CO2/CH4 gas selectivity are presented in Fig. 8b and S6. Compaction of the polymer matrix at elevated pressures also affected the CH4 gas permeance through the membranes, leading to lower permeance for the pure Pebax coated membranes. Again, the incorporation of UiO-66 particles helped to improve the compaction resistance of the membranes. In terms of gas selectivity, a relatively constant CO2/CH4 selectivity was observed in membranes incorporating UiO-66 particles. On the other hand, a decrease in gas selectivity with the increase of feed pressure was observed in the pure Pebax membrane, as this membrane experienced a relatively severe compaction effect.

3.5 ZIF-7/Pebax-1657-based hollow fiber composite membranes

Based on our results on the utilization of UiO-66 and its derivatives for composite membrane fabrication, it is very important to investigate the effect of surface functional groups and pore size on the final membrane performance, especially at elevated pressures. In addition, to understand the general applicability of the membrane fabrication method used in this work, we also fabricated ZIF-7/Pebax-1657-based composite membranes using a similar dip coating approach to that used for the UiO-66/Pebax-based membranes. ZIF-7 has a window aperture of ∼3.0 angstroms, which is smaller than the kinetic diameter of CO2 gas (∼3.4 angstroms). However, previous studies on ZIF-7 revealed the flexible nature of ZIF-7's apertures37,62 that provided an enhancement in CO2 uptake by the framework at a relatively high pressure, which is in good agreement with the results obtained in this work (Fig. 3).
3.5.1 Gas separation performance of the ZIF-7/Pebax-based hollow fiber composite membranes. In this experiment, ZIF-7/Pebax-1657-based membranes with 10–30 wt% particle loading were synthesized and tested for their gas separation performance. Further increasing the ZIF-7 content in the coating solution would lead to particle aggregation and settling. The SEM images of the composite membrane are presented in Fig. S7: at 30 wt% of ZIF-7 loading, clear nanoparticle aggregation can be observed. We further tested the membrane performance with both pure and mixed gases (Fig. 9). Both pure gas and mixed gas showed similar results, with only marginally lower gas permeance for the mixed gas due to a competitive sorption effect. The incorporation of ZIF-7 slightly increased the CO2 permeance through the membranes, which can be attributed to the small pore size of ZIF-7.
image file: c7ta07512j-f9.tif
Fig. 9 Gas separation performance of ZIF-7/Pebax-1657-based composite membranes at different particle loadings: (a) CO2 permeance, (b) CO2/N2 and CO2/CH4 gas selectivity (solid lines represent pure gas and dashed lines represent mixed gas).

Our previous study on ZIF-8/Pebax-1657-based membranes suggests the loss of gas selectivity of the nanocomposite membranes.25 The incorporation of MOF particles with flexible framework structures could allow the rapid transport of bulkier gas molecules. It could also create microvoids in the interfacial region. In comparison, the incorporation of ZIF-7 slightly increased the gas selectivity. A similar increase in gas selectivity with higher particle loading was also reported in a previous study on flat sheet membranes containing ZIF-7.37

3.5.2 Gas separation performance of the ZIF-7/Pebax-1657-based hollow fiber composite membranes at elevated pressure. The effects of elevated pressure on CO2 and CH4 gas permeances are depicted in Fig. 10 and S8. The competing effects of compaction and plasticization resulted in a decrease of gas permeance for all the membranes tested. Compared to the pure Pebax membrane, the decreases in gas permeance in the ZIF-7/Pebax-1657-based membranes were less severe. This shows an increase in polymer rigidity after the incorporation of the particles, as was observed for the UiO-66/Pebax-1657-based membranes (Section 3.4).
image file: c7ta07512j-f10.tif
Fig. 10 (a) CO2 gas permeances and (b) CO2/CH4 selectivity at different feed pressures of ZIF-7/Pebax-1657-based composite membranes (solid lines represent the pressurization step and dashed lines represent the depressurization step).

3.6 Comparison of gas separation performance and understanding the effect of MOF rigidity

Recently, there have been a number of studies on nanocomposite membranes incorporating MOFs. To better understand how the MOF functions within the thin selective layer, the results obtained in this work are compared with our previous results using ZIF-8 nanocrystals in addition to other relevant reports (Table 4). In the present work, the incorporation of UiO-66 and ZIF-7 into the composite membranes increased both CO2 permeance and gas selectivity. The highest CO2/N2 selectivity obtained with 50 wt% UiO-66 was 57 with a CO2 permeance of 338 GPU, making this membrane one of the best performing MOF-containing composite membranes (Table 4). It should be noted that by optimizing the dip-coating substrates, gutter layer materials, coating technique and coating material solution, some pure polymer composite membranes can exhibit comparable gas separation performance.21,67,68 However, as suggested earlier, the operational stability of the pure polymer membrane still limits its industrial application under harsh conditions.
Table 4 Comparison of the performance of nanocomposite membranes incorporating different particles in CO2/N2 gas separation
Support Gutter layer Selective layer Membrane type Max. particle loading Testing T/P (°C/bar) CO2 permeance (GPU) CO2/N2 selectivity Ref.
PAN PDMS Pebax-2533/soft PEG-b-PDMS nanoparticle Flat sheet 40 wt% 35/3.4 1374 12 63
PAN PDMS Pebax-2533/soft PEG-based nanoparticle Flat sheet 50 wt% 35/3.4 601 28 64
PAN PDMS Pebax-2533/PRXs soft nanoparticle Flat sheet 30 wt% 35/3.4 1670 14 65
PAN PTMSP Pebax-1657/ZIF-7 Flat sheet 34 wt% 20/3.75 39 105 37
PVDF Pebax-1657/UiO-66 Flat sheet 20 wt% 25/3 ∼130 barrer 20 30
PVDF Pebax-1657/UiO-66-NH2 Flat sheet 20 wt% 25/3 ∼125 barrer 25 30
PSf PDMS/Cu3(BTC)2 Hollow fiber 4000 ppm 25/5 109.2 33.46 66
PVDF PTMSP Pebax-1657/ZIF-8 Hollow fiber 30 wt% 25/2 345 31.7 25
PVDF PTMSP Pebax-1657/UiO-66 Hollow fiber 50 wt% 25/2 225 45 This work
PVDF PTMSP Pebax-1657/UiO-66-NH2 Hollow fiber 50 wt% 25/2 338 57
PVDF PTMSP Pebax-1657/UiO-66-(COOH)2 Hollow fiber 50 wt% 25/2 240 50
PVDF PTMSP Pebax-1657/ZIF-7 Hollow fiber 30 wt% 25/2 300 48


In our previous work, the addition of ZIF-8 into the thin Pebax layer led to the loss of its gas selectivity.25 In comparison, both improved gas permeance and selectivity were obtained with UiO-66 and ZIF-7 in this work. The addition of ZIF-8 into the Pebax polymeric layer may rigidify the polymeric chains via hydrogen bonding, and their crystalline framework structure may ensure high molecular sieving capabilities. Both of these aspects can improve the gas selectivity of the composite membranes. On the other hand, it has been demonstrated that the organic ligand of ZIF-8 can experience pore-gate-opening due to the rotational and vibrational movement of the organic ligands, which can be attributed to the bending of the N–Zn–N bond and rotation of the Zn–MeIM–Zn bond of ZIF-8.27 This can reduce the intrinsic molecular separation capability of the ZIF-8. At the same time, since ZIF-8 is considered as a “flexible” MOF due to its low metal-center-coordination number, it may experience conformational changes under external mechanical stress. All these aspects may eventually lead to the loss of selectivity.

On the other hand, both UiO-66 and ZIF-7 are considered more rigid. For example, UiO-66 has an ultrahigh shear modulus of ∼14 GPa (compared with ∼1 GPa for flexible ZIF-8).31 The high Zr–O coordination restricts the bending of the bonds, which can effectively preserve its intrinsic molecular sieving capability within a polymeric matrix (Fig. 11). As a result, the incorporation of rigid UiO-66 can simultaneously improve gas permeance and selectivity of the composite membranes. This may also explain why the incorporation of ZIF-7 showed an improved membrane gas selectivity: ZIF-7 and ZIF-8 have similar crystalline structures, but ZIF-7 has a more rigid framework structure.26 Still, in order to fully understand the effect of MOF rigidity on the membrane performance, more characterization should be carried out to investigate the MOF structure within the polymeric matrix. However, this is beyond the scope of this work.


image file: c7ta07512j-f11.tif
Fig. 11 Schematic diagram of the UiO-66, ZIF-7 and ZIF-8 nanocomposite membranes.

4 Conclusions

In this work, UiO-66 and ZIF-7/Pebax based hollow fiber composite membranes were fabricated. Functionalized UiO-66 was applied to understand the effect of surface modification on the final membrane performance. The addition of nanofillers can effectively promote the gas permeance. Due to good interfacial compatibility, the Pebax thin layer can host 50 wt% UiO-66 without introducing extra defects and further increase the UiO-66 loading to 80 wt% with only slightly reduced gas selectivity. This work also investigated the operational stability of the nanocomposite membranes, and the results indicate that the addition of nanofillers can rigidify the Pebax polymer chains via hydrogen bonds, which subsequently enhances the plasticization and compaction resistance of both UiO-66 and ZIF-7 composite membranes. Finally, we discussed the effect of MOF rigidity on the final membrane performance and suggested that more rigid MOFs are preferred to maintain their intrinsic molecular sieving capability within the thin composite membranes.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the Science and Industry Endowment Fund (SIEF Grant ID RP02-035, CO2MOF) project. Putu Doddy Sutrisna would like to acknowledge support from Australia Awards Scholarship (AAS) and Australia Leadership Awards Scholarship (ALAS) for scholarships provided. The authors would like to acknowledge Dr Peter Southon of the School of Chemistry, University of Sydney and Dr Jason Scott of the School of Chemical Engineering, UNSW for their help and discussions in conducting BET and gas sorption experiments.

Notes and references

  1. W. J. Koros and G. K. Fleming, J. Membr. Sci., 1993, 83, 1–80 CrossRef CAS.
  2. P. Bernardo, E. Drioli and G. Golemme, Ind. Eng. Chem. Res., 2009, 48, 4638–4663 CrossRef CAS.
  3. S. Zhao, P. H. M. Feron, L. Deng, E. Favre, E. Chabanon, S. Yan, J. Hou, V. Chen and H. Qi, J. Membr. Sci., 2016, 511, 180–206 CrossRef CAS.
  4. D. M. D'Alessandro, B. Smit and J. R. Long, Angew. Chem., Int. Ed., 2010, 49, 6058–6082 CrossRef PubMed.
  5. J. Hou, P. D. Sutrisna, Y. Zhang and V. Chen, Angew. Chem., 2016, 128, 4015–4019 CrossRef.
  6. J.-R. Li, R. J. Kuppler and H.-C. Zhou, Chem. Soc. Rev., 2009, 38, 1477–1504 RSC.
  7. Y. Zhang, H. Wang, J. Liu, J. Hou and Y. Zhang, J. Mater. Chem. A, 2017, 5, 19954–19962 CAS.
  8. H. B. Tanh Jeazet, S. Sorribas, J. M. Román-Marín, B. Zornoza, C. Téllez, J. Coronas and C. Janiak, Eur. J. Inorg. Chem., 2016, 2016, 4363–4367 CrossRef CAS.
  9. E. M. Mahdi and J.-C. Tan, J. Membr. Sci., 2016, 498, 276–290 CrossRef CAS.
  10. T.-S. Chung, L. Y. Jiang, Y. Li and S. Kulprathipanja, Prog. Polym. Sci., 2007, 32, 483–507 CrossRef CAS.
  11. B. Zornoza, C. Tellez, J. Coronas, J. Gascon and F. Kapteijn, Microporous Mesoporous Mater., 2013, 166, 67–78 CrossRef CAS.
  12. G. Dong, H. Li and V. Chen, J. Mater. Chem. A, 2013, 1, 4610–4630 CAS.
  13. L. M. Robeson, J. Membr. Sci., 2008, 320, 390–400 CrossRef CAS.
  14. R. Lin, L. Ge, L. Hou, E. Strounina, V. Rudolph and Z. Zhu, ACS Appl. Mater. Interfaces, 2014, 6, 5609–5618 CAS.
  15. Z. Wang, D. Wang, S. Zhang, L. Hu and J. Jin, Adv. Mater., 2016, 28, 3399–3405 CrossRef CAS PubMed.
  16. C. Zhang, Y. Dai, J. R. Johnson, O. Karvan and W. J. Koros, J. Membr. Sci., 2012, 389, 34–42 CrossRef CAS.
  17. V. Nafisi and M.-B. Hägg, J. Membr. Sci., 2014, 459, 244–255 CrossRef CAS.
  18. S. Shahid and K. Nijmeijer, J. Membr. Sci., 2014, 470, 166–177 CrossRef CAS.
  19. S. Shahid and K. Nijmeijer, J. Membr. Sci., 2014, 459, 33–44 CrossRef CAS.
  20. B. A. Al-Maythalony, A. M. Alloush, M. Faizan, H. Dafallah, M. A. A. Elgzoly, A. A. A. Seliman, A. Al-Ahmed, Z. H. Yamani, M. A. M. Habib, K. E. Cordova and O. M. Yaghi, ACS Appl. Mater. Interfaces, 2017, 9, 33401–33407 CAS.
  21. H. Z. Chen, Z. Thong, P. Li and T.-S. Chung, Int. J. Hydrogen Energy, 2014, 39, 5043–5053 CrossRef CAS.
  22. M. Benzaqui, R. Semino, N. Menguy, F. Carn, T. Kundu, J.-M. Guigner, N. B. McKeown, K. J. Msayib, M. Carta, R. Malpass-Evans, C. Le Guillouzer, G. Clet, N. A. Ramsahye, C. Serre, G. Maurin and N. Steunou, ACS Appl. Mater. Interfaces, 2016, 8, 27311–27321 CAS.
  23. K. Eum, C. Ma, A. Rownaghi, C. W. Jones and S. Nair, ACS Appl. Mater. Interfaces, 2016, 8, 25337–25342 CAS.
  24. P. Neelakanda, E. Barankova and K.-V. Peinemann, Microporous Mesoporous Mater., 2016, 220, 215–219 CrossRef CAS.
  25. P. D. Sutrisna, J. Hou, H. Li, Y. Zhang and V. Chen, J. Membr. Sci., 2017, 524, 266–279 CrossRef CAS.
  26. J. C. Tan and A. K. Cheetham, Chem. Soc. Rev., 2011, 40, 1059–1080 RSC.
  27. M. R. Ryder, B. Civalleri, T. D. Bennett, S. Henke, S. Rudić, G. Cinque, F. Fernandez-Alonso and J.-C. Tan, Phys. Rev. Lett., 2014, 113, 215502 CrossRef PubMed.
  28. J.-C. Tan, B. Civalleri, C.-C. Lin, L. Valenzano, R. Galvelis, P.-F. Chen, T. D. Bennett, C. Mellot-Draznieks, C. M. Zicovich-Wilson and A. K. Cheetham, Phys. Rev. Lett., 2012, 108, 095502 CrossRef PubMed.
  29. C. Zhang, R. P. Lively, K. Zhang, J. R. Johnson, O. Karvan and W. J. Koros, J. Phys. Chem. Lett., 2012, 3, 2130–2134 CrossRef CAS PubMed.
  30. J. Shen, G. Liu, K. Huang, Q. Li, K. Guan, Y. Li and W. Jin, J. Membr. Sci., 2016, 513, 155–165 CrossRef CAS.
  31. H. Wu, T. Yildirim and W. Zhou, J. Phys. Chem. Lett., 2013, 4, 925–930 CrossRef CAS PubMed.
  32. W. Liang, R. Babarao, M. J. Murphy and D. M. D'Alessandro, Dalton Trans., 2014, 44, 1516–1519 RSC.
  33. X. Liu, N. K. Demir, Z. Wu and K. Li, J. Am. Chem. Soc., 2015, 137, 6999–7002 CrossRef CAS PubMed.
  34. X. Liu, C. Wang, B. Wang and K. Li, Adv. Funct. Mater., 2017, 27, 1604311 CrossRef.
  35. C. Wang, M. Lee, X. Liu, B. Wang, J. Paul Chen and K. Li, Chem. Commun., 2016, 52, 8869–8872 RSC.
  36. Y.-S. Li, F.-Y. Liang, H. Bux, A. Feldhoff, W.-S. Yang and J. Caro, Angew. Chem., Int. Ed., 2010, 49, 548–551 CrossRef CAS PubMed.
  37. T. Li, Y. Pan, K.-V. Peinemann and Z. Lai, J. Membr. Sci., 2013, 425–426, 235–242 CrossRef CAS.
  38. W. Liang and D. M. D'Alessandro, Chem. Commun., 2013, 49, 3706–37088 RSC.
  39. W. Liang, C. J. Coghlan, F. Ragon, M. Rubio-Martinez, D. M. D'Alessandro and R. Babarao, Dalton Trans., 2016, 45, 4496–4500 RSC.
  40. T. Hu, G. Dong, H. Li and V. Chen, J. Membr. Sci., 2014, 468, 107–117 CrossRef CAS.
  41. T. Hu, G. Dong, H. Li and V. Chen, J. Membr. Sci., 2013, 432, 13–24 CrossRef CAS.
  42. Z. Hu, M. Khurana, Y. H. Seah, M. Zhang, Z. Guo and D. Zhao, Chem. Eng. Sci., 2015, 124, 61–69 CrossRef CAS.
  43. G. E. Cmarik, M. Kim, S. M. Cohen and K. S. Walton, Langmuir, 2012, 28, 15606–15613 CrossRef CAS PubMed.
  44. Y. Cao, Y. Zhao, Z. Lv, F. Song and Q. Zhong, J. Ind. Eng. Chem., 2015, 27, 102–107 CrossRef CAS.
  45. A. M. Ebrahim and T. J. Bandosz, ACS Appl. Mater. Interfaces, 2013, 5, 10565–10573 CAS.
  46. Z. H. Rada, H. R. Abid, H. Sun and S. Wang, J. Chem. Eng. Data, 2015, 60, 2152–2161 CrossRef CAS.
  47. Y. Luan, Y. Qi, Z. Jin, X. Peng, H. Gao and G. Wang, RSC Adv., 2015, 5, 19273–19278 RSC.
  48. Y. Ying, Y. Xiao, J. Ma, X. Guo, H. Huang, Q. Yang, D. Liu and C. Zhong, RSC Adv., 2015, 5, 28394–28400 RSC.
  49. J. H. Cavka, S. Jakobsen, U. Olsbye, N. Guillou, C. Lamberti, S. Bordiga and K. P. Lillerud, J. Am. Chem. Soc., 2008, 130, 13850–13851 CrossRef PubMed.
  50. H.-C. Yang, J. Hou, V. Chen and Z.-K. Xu, J. Mater. Chem. A, 2016, 4, 9716–9729 CAS.
  51. S. Lanceros-Méndez, J. F. Mano, A. M. Costa and V. H. Schmidt, J. Macromol. Sci., Part B: Phys., 2001, 40, 517–527 CrossRef.
  52. A. Ghadimi, T. Mohammadi and N. Kasiri, Int. J. Hydrogen Energy, 2015, 40, 9723–9732 CrossRef CAS.
  53. H. Rabiee, A. Ghadimi and T. Mohammadi, J. Membr. Sci., 2015, 476, 286–302 CrossRef CAS.
  54. V. I. Bondar, B. D. Freeman and I. Pinnau, J. Polym. Sci., Part B: Polym. Phys., 2000, 38, 2051–2062 CrossRef CAS.
  55. Y. Wang, H. Li, G. Dong, C. Scholes and V. Chen, Ind. Eng. Chem. Res., 2015, 54, 7273–7283 CrossRef CAS.
  56. S. Friebe, B. Geppert, F. Steinbach and J. Caro, ACS Appl. Mater. Interfaces, 2017, 9, 12878–12885 CAS.
  57. A. Bos, I. G. M. Pünt, M. Wessling and H. Strathmann, J. Membr. Sci., 1999, 155, 67–78 CrossRef CAS.
  58. A. F. Ismail and W. Lorna, Sep. Purif. Technol., 2002, 27, 173–194 CrossRef CAS.
  59. W. Qiu, C.-C. Chen, L. Xu, L. Cui, D. R. Paul and W. J. Koros, Macromolecules, 2011, 44, 6046–6056 CrossRef CAS.
  60. A. Jomekian, R. M. Behbahani, T. Mohammadi and A. Kargari, J. Nat. Gas Sci. Eng., 2016, 31, 562–574 CrossRef CAS.
  61. G. Dong, H. Li and V. Chen, J. Membr. Sci., 2011, 369, 206–220 CrossRef CAS.
  62. F. Li, Q. Li, X. Bao, J. Gui and X. Yu, Korean Chem. Eng. Res., 2014, 52, 340–346 CrossRef CAS.
  63. A. Halim, Q. Fu, Q. Yong, P. A. Gurr, S. E. Kentish and G. G. Qiao, J. Mater. Chem. A, 2014, 2, 4999–5009 CAS.
  64. Q. Fu, E. H. H. Wong, J. Kim, J. M. P. Scofield, P. A. Gurr, S. E. Kentish and G. G. Qiao, J. Mater. Chem. A, 2014, 2, 17751–17756 CAS.
  65. S. Tan, Q. Fu, J. M. P. Scofield, J. Kim, P. A. Gurr, K. Ladewig, A. Blencowe and G. G. Qiao, J. Mater. Chem. A, 2015, 3, 14876–14886 CAS.
  66. A. K. Zulhairun, Z. G. Fachrurrazi, M. Nur Izwanne and A. F. Ismail, Sep. Purif. Technol., 2015, 146, 85–93 CrossRef CAS.
  67. S. Li, Z. Wang, C. Zhang, M. Wang, F. Yuan, J. Wang and S. Wang, J. Membr. Sci., 2013, 436, 121–131 CrossRef CAS.
  68. Y. Wang, T. Hu, H. Li, G. Dong, W. Wong and V. Chen, Energy Procedia, 2014, 63, 202–209 CrossRef CAS.

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c7ta07512j

This journal is © The Royal Society of Chemistry 2018