Contents lists available at ScienceDirect

Separation and Purification Technology

journal homepage: www.elsevier.com/locate/seppur

Gas separation characteristic of mixed matrix membrane prepared by MOF-5including different metalsHulya Aykac Ozen, Bahtiyar Ozturk⁎

Ondokuz Mayis University, Environmental Eng. Dept., 55200 Samsun, Turkey

A R T I C L E I N F O

Keywords:Mixed matrix membraneMOFsGas separationH2/CO2

H2/CH4

CO2/CH4

A B S T R A C T

In this study mixed matrix membranes (MMMs) were manufactured by introducing metal organic frameworks(MOFs) within polyimide (PI) in order to investigate their separation performances for H2, CO2 and CH4 gases.MOF-5, Cu, Co and Ni doped MOF-5 and CuCo, CuNi and NiCo doped MOF-5 particles have been produced andused as fillers for the manufacturing MMMs. MOFs and membranes have been characterized by using SEM, FTIR,XRD, TGA, DTG and DSC analysis. It has been evidenced that MOF particles have the similar shape, and XRD andFTIR analysis showed all metals exist in the MOFs. MOF particles and PI showed generally a good compatibility,but MMMs were brittle and comprised some nonselective fissures in case of loading above 5w.% MOFs includingtriple metals. MMM manufactured by using Cu doped MOFs showed higher gas permeation, and Cu also madepositive effect on gas permeation of MMMs which were manufactured initially using Co doped MOF-5 or Nidoped MOF-5. Permeation rate of gases through MMMs enhanced gradually with increasing the loading rate ofMOFs (5w.%, 10w.%, 15w.%) and transmembrane pressure (100–500 kPa) at room temperature. 5w.% of MOF-5 incorporation within PI increased its H2, CO2 and CH4 permeabilities as 20.35%, 31.41% and 23.98%, re-spectively. But, introducing 5w.% of Cu doped MOF-5 (CuMOF-5) within PI increased its H2, CO2 and CH4

permeabilities as 62.86%, 94.00% and 121.27%, respectively. This higher interest of MOFs to CO2 and CH4

reduced the H2/CO2 and H2/CH4 selectivities. There was no a linear relationship between gas adsorption ca-pacities of MOFs and permeation rate of these gases through the MMMs. The more adsorbed gas (H2) permeatedslowly through MMMs. Feed pressure increased the permeability of gases and reduced selectivities of themembranes. MMMs manufactured by using MOFs can be used for H2 separation from CO2, and enrichment ofCH4 from natural or biogas.

1. Introduction

Membrane technology took a great deal of attention in gas separa-tion, because of their low energy requirements, modular state, easyscale-up or down and operation over other conventional gas separationtechnologies [1,2]. Since the first large scale application of polymericmembranes to hydrogen separation from the purge gas stream of am-monia plants by Monsanto in late seventies, permeability and se-lectivity of polymeric membranes is needed to improve to find out amarket value. However, polymeric membranes posses trade off betweenpermeability and selectivity which was first evidenced by Robeson in1991 and then updated in 2008 [3,4], illustrating the progress achievedby polymer science for the main gas separations. Inorganic membranesare alternatives to polymeric membranes and have higher permeabilityand selectivity [5] and coped with Robeson’s upper bound for gas se-paration; but these membranes have higher manufacturing costs,

difficult to synthesize and prone to breakage [6]. Recently, a new typeof hybrid membrane which is called as mixed matrix membrane (MMM)first reported by Zimmerman [7] has been introduced to overcomedisadvantages of pure polymeric and inorganic membranes. Inorganicparticles such as zeolites [8–11], carbon molecular sieves (CMSs)[6,12], carbon nanotubes (CNTs) [13], silica [14,15], fullerens [16],metal peroxides [17] and metal organic frameworks (MOFs) [18–20]are incorporated or dispersed into polymeric matrix to form MMMs.Among them, MOF materials are good candidates to overcome in-compability of filler and polymer constructed by transition metals andorganic ligands. They offer advantages to get high performance inMMMs because of its numerous attractive features such as large surfacearea, adjustable structure and pore size, uniformity and thermal-me-chanical stability [21,22]. Table 1 summarizes the results of pioneeredresearch studies about pure gas permeation and separation propertiesby MMMs.

https://doi.org/10.1016/j.seppur.2018.09.052Received 20 July 2018; Received in revised form 17 September 2018; Accepted 17 September 2018

⁎ Corresponding author.E-mail address: bozturk@omu.edu.tr (B. Ozturk).

Separation and Purification Technology 211 (2019) 514–521

Available online 08 October 20181383-5866/ © 2018 Published by Elsevier B.V.

T

MOF-5 is one of the most outstanding framework due to its excellentproperties. The structure of MOF-5 built up from [Zn4O]6+ clusterlinked with terephthalate acid ligands, to obtain three dimensionalporous cubic framewoks, resulting in high surface area, high porosityand tuneable structure [30]. However, MOF-5 framework is sensitive towater because of its weak metal-oxygen bond, and its structure can becollapsed when exposed to moisture [31]. In order to enhance hydro-stability of MOF-5 many strategies have been developed, and onemethod is to add transition metals into MOF-5 structure. To date, dif-ferent types of MOFs doped with metals are presented very few instudies. Yang et al. [32] and Botaş et al. [33] used Co to form Co-MOF-5and found that higher adsorption capacities for gases compared to pureMOF-5. Li et al. [34] and Yang et al. [35] demonstrated the metho-dology for the synthesis of Ni doped MOF-5 and they found that the Ni-MOF-5 improved hidrostability compared to the MOF-5. Bae et al.[36,37] constructed Li doped MOF-5 and observed increase in separa-tion performance.

In this study, in order to understand the effect of metal doped MOF-5 on gas permeability and selectivity of polyimide, Cu, Co or Ni dopedMOF-5 (CuMOF-5, CoMOF-5 and NiMOF-5) and CuCo, CuNi or CoNidoped MOF-5 (CuCoMOF-5, CuNiMOF-5, CoNiMOF-5) were succesfullysynthesized and incorporated into polyimide to fabricate MMMs (MOF/PI). The prepared MOFs and membranes were characterized by scan-ning electron microscopy (SEM), Fourier transform infrared spectro-scopy (FTIR), X-ray diffraction (XRD) and thermal analysis (TGA/DTA/DTG). Membrane gas permeation measurements were conducted for H2,CO2 and CH4 gases using a constant-volume/variable pressure method.All measurements were carried out at room temperature and differentgas feed pressures (from 100 to 500 kPa).

2. Experimental

Experimental methods comprises four sections, namely synthesis ofMOF-5 and metal doped MOF-5, fabrication of pure polymeric andMMMs, MOF and membrane characterization tests and pure gas se-paration operations.

MOF-5 crystals were synthesized by solvothermal method [38,39].Typically, 3.25 mmol of Zn(NO3)2·6H2O and 1.7 mmol of H2BDC wasdissolved in 16 mL of DMF in a glass tube under strong agitation in asonicator. Then the prepared solution was placed in an oven at 378 Kfor 24 h. Glass tube was then removed from the oven and allowed to becooled down to room temperature. MOF-5 crystals were obtained byfiltering the white solution and then the white powder washed threetimes using ethanol of 5 mL. The white powder was dried for 1 d at333 K in a vacuum oven. MOF-5 powders were collected and stored in acapped vial.

Metal doped MOF-5 synthesis was performed using divalent metalions of Cu [Cu(NO3)2·3H2O], Co [Co(NO3)2·6H2O] or Ni [Ni(NO3)2·6H2O] according to previously reported method [32,34]. Inorder to obtain doped MOF-5, 0.05 mmol of metal (Cu, Co, or Ni),

0.05 mmol of Zn and 0.05 mmol H2BDC were added into 16 mL of DMFand 4 mL of ethanol with stirring. The prepared solution was placed inan oven at 378 K for 24 h to yield precipitates, then the precipitatescooled down to room temperature, washed several times with ethanol,the powder was collected by filtration, and finally the dried powder in avacuum oven at 333 K for 24 h stored in a capped vial. Accordingly,MOF-5, CuMOF-5, CoMOF-5, NiMOF-5, CuCoMOF-5, CuNiMOF-5, andCoNiMOF-5 were synthesized and used as fillers in MMMs.

2.1. Manufacturing of pure and mixed matrix membranes

Polyimide (PI) has been selected as a continues phase for membranemanufacturing because of three reasons; (1) it has higher Tg value(585 K) that is important for removing of solvents during annealing, (2)rubbery polymers introduce into pores of MOFs and it reduces the gaspermeation, (3) glassy polymers are generally brittle for manufacturingMMMs including particles.

Membrane was prepared using the solution casting method asmentioned previous study [11]. 0.5 g of the PI was dissolved in 4.5 g ofN-methylpyrrolidone (NMP) and stirred for 12 h at 343 K until a clearuniform viscous solution was acquired. Then solution was casted on aclean glass plate using a casting rod. The glass plate was placed in avacuum oven at 343 K including 50 kPa of N2 gas for 12 h to allow thesolvent to evaporate completely. The glass plate was immersed into awater bath to separate the polymeric membrane from the glass plate.After that, the membrane was annealed in a vacuum oven at 493 K for24 h. The average thickness of a membrane was determined by mea-suring the thickness at least five points using a micrometer. All themembranes used in gas separation experiments have 40 ± 2 µmthickness.

The fabrication method for the MMMs was similar to the prepara-tion of pure polymeric membrane with the additional step of compo-siting MOF crystals. Two different solutions of PI and MOF were pre-pared using NMP, than those solutions were mixed. The first solutionincludes 0.5 g of PI in 4.5 g of NMP and the second solution includes0.025 g (5w.%), 0.050 g (10w.%) or 0.075 g (15w.%) of MOF in 4.5 g ofNMP. The two solutions were bath sonicated and stirred for 3 h thenwere mixed by pouring the polymer solution into the MOF solution. Thecombined solution was stirred and bath sonicated for 5 h. Other stepsfor the fabrication of MMMs were identical to that of the pure polymermembrane preparation. Uniform films were casted on a clean glassplate using a casting rod. The glass plate was placed in a vacuum ovenat 343 K and 50 kPa for 12 h to allow the solvent to evaporate. Then themembrane separated from the glass plate by immersing the glass plateinto a water bath. The membrane was annealed in a vacuum oven at493 K for 24 h.

2.2. Characterization of PI and MOF-5 MMMs

Fourier transform infrared spectroscopy (FTIR) spectra of the MOF

Table 1Some gas separation results obtained by using mixed matrix membranes.

MOF name Membrane material Effect of fillers on permeability Effect of fillers on selectivity Ref

H2 CH4 CO2 H2/CH4 H2/CO2 CO2/CH4

Cu-BPY-HFS Matrimid Improved Improved Improved Not improved Not improved Not improved [23]ZIF 71 6FDA-Durene Improved Improved Improved Not improved Not improved Not improved [24]ZIF-11 6FDA-DAM Improved Improved Improved Not improved Not improved Not improved [25]CuTPA PVAc – Improved Improved – – Improved [26]MIL-53(Al PVDF – Improved Improved – – Not improved [27]MIL 101(Cr) Matrimid Improved Improved Improved [28]C-MOF-5 PEI Improved Improved Improved Improved Improved Improved [18]MIL-53 Matrimid – Improved Improved – – Improved [29]MOF-5 Matrimid Improved Improved Improved Not improved Not improved Not improved [20]

H. Aykac Ozen, B. Ozturk Separation and Purification Technology 211 (2019) 514–521

515

particles were recorded at room temperature on a Perkin-ElmerMIRacle Single Reflection Diamond ATR and FTIR spectra were col-lected at wave-number between 650 cm−1 and 4000 cm−1. Scanningelectron microscopy (SEM) images of MOFs were taken using a JSM-7001F Field Emission Scanning Electron Microscope. Thermal analysesperformed to determine the weigh loss of samples by temperature usinga Shimadzu DTG-60 instrument under the nitrogen atmosphere with aheating rate of 20 °C min−1. To explain crystal characterisation ofsamples, RIGAKU, SMARTLAB X Ray Diffraction (XRD) equipment wasused with Cu Ka radiation and analysis was performed from 5 to 50 (2θ)at a scanning rate of 0.02 deg/min.

2.3. Gas separation operation

Gas permeation tests of pure membrane and MMMs were carried outusing a variable pressure/constant volume method. Experimental set uppresented in Fig. 1 equipped with a stainless steel permeation cell,temperature sensors and pressure transducer. Permeation fluxes weremonitored using Model Hobby Projects data logger version 1.20. Thetypical membrane area located at the test cell was 49.2 cm2 andpermeate side volume was 30.2 mL. Single gas permeabilitiesthroughout the membranes were determined for CO2, CH4 and H2 gasesfor 100–500 kPa pressure intervals. All measurements were performedat room temperature.

The gas permeability (P) was calculated by the slope of curve ofpermeate pressure vs. time (dp/dt) using the Daynes-Barrier time lagequation as follows [40]:

=P V TA P T P

dPdt

o

o (1)

where P is the permeability coefficient expressed in Barrer [1Barrer = 1 × 10−10 cm3 cm (STP) cm−2 s−1 cmHg−1], (dp/dt) is theslope of the straight line in the steady-state region at which permeationpressure increases with time on the downstream side, V (cm3) is thecalibrated downstream volume, ΔP (cmHg) is the transmembranepressure difference between the two sides, A (cm2) is the effective areaof the membrane, δ (cm) is the membrane thickness, T (K) is themeasure temperature, and T0 and P0 are the standard temperature andpressure, respectively. The ideal separation factor of pure gas A over B(αAB) is defined as the ratio of the permeation rates of A and B, whichcan be expressed as:

= P /PAB A B

3. Results and discussion

3.1. Characterization of MOFs and membranes

In order to determine the thermal stability, crystal structure,structural bonds and morphological characteristics of MOFs andmembranes produced TGA/DTG, XRD, FTIR and SEM analyses havebeen performed, respectively. Thermal decomposition of MMMs hasbeen started at lower temperatures compared to pure PI membrane. Forexample, thermal decompositions of PI, MOF-5/PI, CoMOF-5/PI andNiMOF-5/PI have started at 762 K, 707 K, 713 K and 717 K, respec-tively. The results showed that the decomposition temperature ofmembranes follows the order of PI > MOF-5/PI > NiMOF-5/PI > CoMOF-5/PI > CuMOF-5/PI > NiCoMOF-5/PI > NiCuMOF-5/PI > CuCoMOF-5/PI. This is due to fact that the organic linker(terephthalic acid) which was used for MOF production has lower de-composition temperature than PI. Decomposition of terephthalic acidand PI start at 605 K and 793 K, respectively.

The results of TGA analyses of PI and MMMs given in Fig. 2 showthat MMMs start to weight loss at lower temperatures compared to purePI membrane. Thermo gravimetric analysis (TGA/DTG) curves dis-played three main steps of weigh loss for MOFs. The first step of weighloss can be attributed to the lost of guest molecules (water and solvent)which are encapsulated by the framework during the synthesis ofMOFs. The release of the guest molecules from MOFs at differenttemperatures can be attributed to the differences of porous state ofthose MOFs. The second and third weigh losses correspond to the de-composition of organic linkers and vaporization of low boiling pointzinc, respectively. The weight loss between 901 and 1073 K is the thirdstage, and the low boiling point zinc metal are removed in this step; theresidue is considered to be carbon and metal oxide based species re-maining from the degradation of the ligand and the metal oxide. Theweight loss values obtained in the first and second stages of the curvesare in agreement with the values reported in the literature [41]. Table 2

Fig. 1. Gas permeation rig.

Fig. 2. TGA curves of PI amd MMMs.

Table 2DTG profiles of MOFs.

MOF The first step of weight loss (lossof water and solvent)

The second step of weight loss(removal of organic linker)

Temperature, K DTGmax Temperature, K DTGmax

MOF-5 406–582 494 671–874 781NiMOF-5 385–626 506 657–878 776CoMOF-5 445–643 616 681–832 780CuMOF-5 461–591 559 605–875 688NiCoMOF-5 455–633 578 681–830 777NiCuMOF-5 451–608 547 626–809 698CuCoMOF-5 454–600 560 620–811 694

H. Aykac Ozen, B. Ozturk Separation and Purification Technology 211 (2019) 514–521

516

summarizes the results of thermo gravimetric analyses of MOFs pro-duced in this study. As it can be seen from Table 2, MOFs including Cuhave lower stability and MOFs including Co have higher stability.

Glass transition temperature (Tg) which is the highest temperatureto which polymer withstand during annealing is an important para-meter for a membrane. Tg value of PI was determined as 585 K by DSCanalysis. Solvents which remain in the interfacial voids betweenpolymer chains and in the pores of a MOF can affect negatively the gastransport through a membrane. Therefore, at a temperature below theTg value of a polymer, solvents must be released-off during annealing ofa membrane. It can be seen from Table 2 that CoMOF-5 completed itsfirst weigh loss just above the Tg value of PI. Hence, MMMs annealedunder the vacuum in N2 atmosphere.

XRD analyses of MOFs show that intensity of MOF-5 increases whenmetal/metals inserted to the MOF-5 cluster (Figs. 3 and 4). MOF-5showed three characteristic peaks at 2Θ = 9.87, 2Θ = 17.84 and2Θ = 24.85. In addition to these three characteristic peaks, some othersmall peaks of metal doped MOF-5s can be seen from the XRD chro-matogram. These results prove that metal doped MOF-5s have the samecubic state as MOF-5, and the differences between those small peaksprove new crystal planes of doped metal/metals. These results are inaccord with the results indicated in the literature [42,43]. FTIR ana-lyses of MOFs also proved that metal doping into the MOF-5 structurehave been performed successfully (Fig. 5). FTIR spectra of MOF-5 ex-hibited characteristic peaks at 1580–1500 cm−1 while two peaksaround 1600 cm−1 proves the metal/metals doped, and the intensity of

those peaks changed depending on the type of metal/metals. SEMpictures of MOF-5 and metal doped MOF-5s showed the same cubicstate which verify that the morphology of MOF-5 has not been changedby the addition of metal/metals into the MOF-5 cluster (Fig. 6).

3.2. Gas permeation and separation performances of MOF based MMMs

Gas permeation and separation characteristics of pure polyimidemembrane have been compared with MMMs in which MOF-5 andmetal/metals doped MOF-5 particles exist for H2, CO2 and CH4 gases.The results below are given and discussed about the effects of feedpressure (ΔP) and MOF type and amount in the polymer.

Figs. 7–9 represent the relationship between pure gas permeabilities(or ideal permeability) of H2, CO2 and CH4 gases as a function of feedpressure between 100 and 500 kPa for pure PI and MMMs at298 ± 2 K. MOF loadings into PI and elevation of feed gas pressureincreased the permeabilities of gases and reduced the H2/CO2, H2/CH4

and CO2/CH4 selectivities. The highest effect of pressure on gas per-meability was seen when CuMOF-5 was introduced into PI, and it fol-lowed by NiCuMOF-5/PI, CuCoMOF-5/PI and CoMOF-5/PI mixed ma-trix membranes. A five times increase in feed gas pressure increased theH2, CO2 and CH4 permeabilities as 47.28%, 52.55% and 60.33%, andreduced the H2/CO2, H2/CH4 and CO2/CH4 selectivities as 3.44%,8.10% and 4.84%, respectively when 5w% CuMOF-5 was used inMMM. Houde et al. [44] indicated that increasing the feed pressurereduced the transport resistance against to the gaseous molecules byincreasing intersegmental spacing between polymer chains, which iscorrelated with the Tg (glass transition temperature) of the polymer.This assertion can be convincing for pure polymers having lower Tgvalues. But, in our case, there are porous particles in MMMs and theirpore sizes are quite higher (7–8 Å) than intersegmental spaces of thepolymer chain. Those MOFs with higher pore sizes encouraged thegaseous transport through the MMMs and reduced the membrane gasselectivity. The effect of pressure on gaseous transport was so obviousfor MMMs compared to pure PI membrane. Thus, both MOFs in MMMsand the trans-membrane pressure increased the permeability of biggersize of gaseous molecules such as CH4 and CO2, and it resulted in thereduction of H2/CH4 or H2/CO2 selectivities. Membranes includingmore than 5w% of MOFs produced by doping two metals into MOF-5were so brittle. The possible reason for such a brittle membrane is therigidification of PI by those MOFs. As it can be seen from the Figs. 7–9,there is no membrane including more than 5w% of MOFs which includetriple metals.

Fig. 3. XRD patterns of MOF-5 and metal/metals doped MOF-5.

Fig. 4. XRD patterns of PI and MMMs.

Fig. 5. FTIR curves of MOF-5 and metal/metals doped MOF-5.

H. Aykac Ozen, B. Ozturk Separation and Purification Technology 211 (2019) 514–521

517

Gas permeation yields of MOF-5/PI was elevated when Ni, Co or Cudoped on MOF-5 in the order of Cu > Co > Ni. Electrostatic interac-tion between the MOF particles and gas molecules is an importantfeature for selective permeation. Those MOFs have been produced byusing transition metals of two valence electrons. These electrons in-creased the affinity of gas molecules and enhanced the transport ofgases through MMMs. All metals used have almost the same electro-negativities; but, their electron affinities which are the energy of aneutral atom or molecule when an electron is added to the atom to forma negative ion are different; namely 0.0, 63.7, 112.0 and 118.4 kJ/molfor Zn, Co, Ni, and Cu, respectively. It can be said that different metals

have different affinities against gas pairs. Some Zn ions in MOF-5substituted with Co, Cu or Ni ions when Co, Cu or Ni doped into theMOF-5, and the substitution increased the interaction of gaseous mo-lecules with MOFs. The substitution of Cu ion with Ni ion when Nidoped into CuMOF-5, and this substitution reduced the affinity betweenNiCuMOF-5 and gas pairs. On the other hand, the permeabilities ofgases through NiMOF-5/PI were increased when Co or Cu doped into

Fig. 6. SEM images of MOFs.

Fig. 7. Effect of feed gas pressure on H2 permeability.

Fig. 8. Effect of feed gas pressure on CO2 permeability.

Fig. 9. Effect of feed gas pressure on CH4 permeability.

H. Aykac Ozen, B. Ozturk Separation and Purification Technology 211 (2019) 514–521

518

the NiMOF-5. Permeability of gases also increased when double metaldoped MOF-5 was used as filler in PI except the NiCuMOF-5 whichreduced the permeabilities of CO2 and CH4 as 5,56% and 3,7%, re-spectively at 500 kPa. This negative effect of Ni on CuMOF-5 increasedwhen feed pressure was elevated from 100 to 500 kPa. The interactionof gases with CuMOF-5 is elevated by increasing pressure, but this in-teraction is diminished when Ni doped into the CuMOF-5. Ahmadi et al.[35] indicated that Cu-based MOFs (Cu-BTC: Copper benzene-1,3,5-tricarboxylate) offer an exceptional CO2 uptake due to their higheraffinity with polar molecules. But, another MOF including Cu (Cu-BPY-HFS:Cu-4,4′-bipyridine-hexafluorosilicate) shows strong affinity to-wards CH4 compared to CO2 and lower CO2/CH4 selectivity [23]. Thesetwo results imply that permeability and selectivity of a MMM don’tdepend on just a metal in a MOF, and also some other membranecharacteristics play important roles.

MMMs showed higher permeability for all gases used compared topure PI membrane. Because of the porous state of MOFs, diffusion ofgases through MMMs increased by increasing the amount of MOF in thepolymer, it means that the more MOF is the more pathways for gastransfer. But this increase in permeability was not equivalent for allgases used. MOF loading increased the permeability of CH4 more thanH2 and CO2; hence, H2/CH4 selectivity decreased by increasing MOFamount in a MMM (Fig. 10).

It must be noted that we could not manufactured MMMs includingmore than 5w% of MOF which has produced by using three metals suchas CuCoMOF-5, NiCoMOF-5 or CuNiMOF-5. Such a membrane was sobrittle, and contained nonselective fissures. Kusworo et al. [45] ex-plained that the increase of zeolite particles in polyimide elevated thegaseous permeability due to the disruption of polymer chain by zeoliteparticles and it caused the more free volumes for gas transfer. Thisexplanation might be true when polymer-particle compatibility is poorsuch as zeolite/PI-MMM. But, our experimental results showed thatpermeability of gases did not increase excessively by pressure, thismeans that the compatibility of PI-MOF is good. SEM image (Fig. 6i)also showed that the interfacial contact between the MOFs and polymermatrix is good. Ahmadi et al. [46] noted that the presence of organicligands in MOFs’ structure leads to better affinity and adhesion betweenMOFs and polymers. Shadid and Nijmeier [47] indicated that a physicalcrosslinking occurs between the polymer chains and the MOF particles,which results in the formation of a more rigid intermediate phase be-tween the polymer and the MOF particles. A slightly penetration ofpolymer solution into the MOFs pores can also leads to chain rigidifi-cation of the nearby polymer chains in the vicinity of the MOF particles[23,48].

Solution-diffusion mechanism plays a major role on gas permeationthrough a pure polymeric membrane while gas molecules can be

transferred through a mixed matrix membrane under different me-chanisms, such as solution-diffusion, Knudsen and surface diffusion inthe pores and on the walls of the MOF crystals. There are also differentfactors such as porosity, pore size, tortuosity, particle size and the in-terest of particles to gas pair affecting gaseous transport from a porousmaterial. The effect of MOFs on gas permeability is quite obvious forlarger molecules such as CO2 and CH4. But, this increase in perme-abilities is not consistent for all the MOF types and loadings. There canbe four main factors which affect the gas permeability through MMMs,namely: (1) pore sizes, porosity and tortuosity of MOFs, (2) compat-ibility between fillers and the polymer, (3) particle sizes of fillers, and(4) the interest of MOFs against gases. The first factor has no so mucheffectiveness on gas permeation through MMMs owing to the MOFshave the similar pore sizes because of the MOFs have been producedusing the same organic linker and the atomic sizes and volumes of themetals used are the same; permeability of gases has reasonable increaseby the increase of feed pressure means that there is no nonselectivevoids between polymer-fillers interface, that proves the compatibilitybetween polymer and the fillers is good; the third factor cannot also beeffective on gas permeation because the SEM pictures showed that theMOF particles have the same sizes. We think that the fourth factor hasmuch effectiveness on gas permeation through the MMMs owing to theinterest of MOF particles differs against gaseous molecules. An ad-sorption study has been conducted at 100 kPa and 298 ± 2 K in orderto determine the interest of MOFs against the gases used. It can be seenfrom Fig. 11 that the MOFs adsorb more H2 and it is followed by CH4

and CO2. Almost all MOFs used showed high CH4 adsorption, and theirfavourable interest to CH4 resulted in the reduction the H2/CH4 se-lectivity (Fig. 10). Zhang et al. [23] introduced Cu-BPY-HFS of 40w.%into Matrimid 5218, and they reported that the MOF showed a betteraffinity to CH4 than CO2, and enhanced CH4 transport in both pure andmixed gas tests. They also indicated that the Cu-BPY-HFS interactedmore strongly with CO2 than H2 and reduced the H2/CO2 ideal se-lectivity.

One can expect that the permeability of H2 through MMMs shouldbe higher than CO2 and CH4 because of its smaller molecular sizecompared to CO2 and CH4. But, the higher adsorption of H2 caused itsaccumulation over the pore surface of MOF particles and it has resultedin the blockage of pores for H2 transfer. But, gases are transferredthrough nonporous PI membrane by solution-diffusion mechanism andthrough chain gaps which are so small compared to molecular sizes ofgases. Fuertes [49] prepared a microporous carbon membrane by de-positing a thin phenolic resin film on the inner surface of a porousceramic tube by means of dip coating technique. He reported that themore strongly condensable gases are preferentially adsorbed on themicroporous of the membrane, and this reduced the open porosity andpermeated preferentially through the membrane. He also indicated that

Fig. 10. Effcet of MOF loading on ideal gas selectivities. Fig. 11. Effect of adsorbent type on gas adsorption.

H. Aykac Ozen, B. Ozturk Separation and Purification Technology 211 (2019) 514–521

519

these membranes are suitable to separate non-condensable or weaklyadsorbable components (i.e. N2, H2, O2, etc.) from the more stronglyadsorbable components (i.e. hydrocarbons).

Fig. 12 compares the effectiveness of MMMs over the pure PI forpermeability of gases. As it can be seen from the figure that the MMMsproduced using Cu doped MOF-5 have higher permeability for all gasesused. In case of three metals, Cu has affected positively the permeationof gases through the MMMs manufactured initially using NiMOF-5 orCoMOF-5. Feijani et al. [35] indicated that introducing CuPTC orCuBDC into PVDF increased the He and CO2 permeability 43.6% and118.7% or 37.5% and 117%, respectively. They reported that the stronginteraction of CO2 gas with electrostatic fields of MOFs including Cuincreased the solubility and diffusivity of CO2. Angelis et al. [50] hasprepared a MMM by introducing fumed silica into a glassy polymer andexamined the pure gas permeability of He, H2, O2, N2, CO2 and CH4.They indicated that permeability of gases increased by adding moresilica into polymer, but this effect was higher for bigger size gas mo-lecules. They explained that SiO2 creates new pathways for big sizemolecules and it also show different interest against different gases.

An overall comparison of the gas separation performances of MMMsmanufactured for H2/CO2, H2/CH4 and CO2/CH4 mixtures are shown inFig. 13, and compared to the Robeson’s upper bond trade-off line

revised in 2008 [4]. As it can be seen from the figure, H2/CO2 se-lectivity is above the trade-off line which means that MMMs includingMOFs can be used for the separation of H2 from reforming gas mixtureproduced by steam reforming of CH4 for H2 production. MOF/PI-MMMscan be used for the enrichment of CH4 from natural or biogas streams.

4. Conclusion

Gas permeation and separation performances of pure PI and MMMsincluding MOF-5 and metal (Co, Cu, Ni) doped MOF-5 were in-vestigated for H2, CO2 and CH4 gas pairs. The following conclusions canbe emphasized from the results of this study: (1) FTIR analyses of MOFsproved that metal doping into the MOF-5 structure have been per-formed successfully, and those MOFs have the similar shape. Thermalanalysis showed that introducing MOFs within PI reduced its decom-position temperature. (2) MMMs including 5 to 15w.% MOF-5, CoMOF-5, CuMOF-5 or NiMOF-5 have been manufactured successfully, butMMMs comprise CoCuMOF-5, CoNiMOF-5 or CuNiMOF-5 above 5w.%were brittle and included some nonselective fissures. The reason forsuch a brittle membrane is the rigidification of PI by those MOFs. (3)Introducing MOFs within the PI increased its gas permeability yields,but reduced the H2/CO2, H2/CH4 and CO2/CH4 selectivities, (4) MMMmanufactured by using Cu doped MOFs showed higher gas permeation,and Cu also made positive effect on gas permeation of MMMs whichwere manufactured initially using Co doped MOF-5 or Ni doped MOF-5,(5) Permeation rate of gases through MMMs enhanced gradually withincreasing the loading rate of MOFs (5w.%, 10w.%, 15w.%) andtransmembrane pressure (100–500 kPa) at room temperature. Thehigher the feed pressure caused the more gas transfer through themembranes in the order of CH4 > CO2 > H2 and reduced the H2/CO2,H2/CH4 and CO2/CH4 selectivities. The descending effect of the feedpressure on the membrane selectivity was more evident for MMMs thanPI. 5w.% of MOF-5 incorporation within PI increased its H2, CO2 andCH4 permeabilities as 20.35%, 31.41% and 23.98%, respectively. But,introducing 5w.% of Cu doped MOF-5 (CuMOF-5) within PI increasedits H2, CO2 and CH4 permeabilities as 62.86%, 94.00% and 121.27%,respectively. This higher interest of MOFs to CO2 and CH4 reduced theH2/CO2 and H2/CH4 selectitivities, (6) There was no a linear relation-ship between gas adsorption capacities of MOFs and permeation rate ofthese gases through the MMMs. The more adsorbed gas (H2) permeatedslowly through MMMs. Feed pressure increased the permeability ofgases and reduced selectivities of the membranes, 7) MMMs manu-factured by using MOFs can be used for H2 separati(on from CO2, andenrichment of CH4 from natural or biogas. A further study can be car-ried out in order to determine the effect of moisture on gas permeationsand selectivities of these membranes.

Acknowledgments

The authors gratefully acknowledge Ondokuz Mayis University forthe financial support through the project numbered asPYO.MUH.1901.14.003. The authors are also grateful to MEMFO R&DGroup in Department of Chemical Engineering at the NorwegianUniversity of Science and Technology for their kind assistance in car-rying out membrane characterization experiments.

REFERENCES

[1] W.S.W. Ho, K.K. Sirkar, Membrane Handbook, Chapman & Hall, Kluwer AcademicPublishers, New York, Boston, 1992.

[2] R.W. Baker, Overview of Membrane Science and Technology, in: MembraneTechnology and Applications, John Wiley & Sons, Ltd., 2004, pp. 1–14.

[3] L.M. Robeson, Correlation of separation factor versus permeability for polymericmembranes, J. Membr. Sci. 62 (1991) 165–185.

[4] L.M. Robeson, The upper bound revisited, J. Membr. Sci. 320 (2008) 390–400.[5] A.F. Ismail, L.I.B. David, A review on the latest development of carbon membranes

for gas separation, J. Membr. Sci. 193 (2001) 1–18.[6] D.Q. Vu, W.J. Koros, S.J. Miller, Mixed matrix membranes using carbon molecular

Fig. 12. Effect of membrane type on gas permeability.

Fig. 13. H2/CO2, H2/CH4 and CO2/CH4 separation performances of MMMs.

H. Aykac Ozen, B. Ozturk Separation and Purification Technology 211 (2019) 514–521

520

sieves: I. Preparation and experimental results, J. Membr. Sci. 211 (2003) 311–334.[7] C.M. Zimmerman, Advanced gas separation membrane materials: hyper rigid

polymers and molecular sieve-polymer mixed matrices, 1998.[8] Z. Huang, Y. Li, R. Wen, M.M. Teoh, S. Kulprathipanja, Enhanced gas separation

properties by using nanostructured PES-zeolite 4A mixed matrix membranes, J.Appl. Polym. Sci. 101 (2006) 3800–3805.

[9] Y. Li, T.-S. Chung, Z. Huang, S. Kulprathipanja, Dual-layer polyethersulfone (PES)/BTDA-TDI/MDI co-polyimide (P84) hollow fiber membranes with a submicronPES–zeolite beta mixed matrix dense-selective layer for gas separation, J. Membr.Sci. 277 (2006).

[10] T.W. Pechar, S. Kim, B. Vaughan, E. Marand, M. Tsapatsis, H.K. Jeong,C.J. Cornelius, Fabrication and characterization of polyimide–zeolite L mixed ma-trix membranes for gas separations, J. Membr. Sci. 277 (2006) 195–202, https://doi.org/10.1016/j.memsci.2005.10.029.

[11] O. Bahtiyar, F. Demirciyeva, Comparison of biogas upgrading performances ofdifferent mixed matrix membranes, Chem. Eng. J. 222 (2013) 209–217.

[12] M. Anson, J. Marchese, E. Garis, N. Ochoa, C. Pagliero, ABS copolymer-activatedcarbon mixed matrix membranes for CO2/CH4 separation, J. Membr. Sci. 243(2004) 19–28.

[13] X. He, Y. Chu, A. Lindbrathen, M. Hillestad, M.-B. Hagg, Carbon molecular sievemembranes for biogas upgrading: techno-economic feasibility analysis, J. Clean.Prod. 194 (2018) 584–593.

[14] D. Gomes, S.P. Nunes, K.-V. Peinemann, Membranes for gas separation based onpoly(1-trimethylsilyl-1-propyne)–silica nanocomposites, J. Membr. Sci. 246 (2005)13–25.

[15] M.Z. Rong, M.Q. Zhang, Y.X. Zheng, H.M. Zeng, R. Walter, K. Friedrich,Structure–property relationships of irradiation grafted nano-inorganic particle filledpolypropylene composites, Polymer 42 (2001) 167–183.

[16] T.-S. Chung, S.S. Chan, R. Wang, Z. Lu, C. He, Characterization of permeability andsorption in Matrimid/C60 mixed matrix membranes, J. Membr. Sci. 211 (2003)91–99.

[17] M. Sariam, M.D. Patil, R.S. Veerapur, S.A. Patil, T.M. Aminabhavi, Novel dense Poly(vinyl alcohol)TiO2 mixed matrix membranes for pervaporation separation of waterisopropanol mixtures at 30°C, J. Membr. Sci. 281 (1–2) (2006) 95–102.

[18] M. Arjmandi, M. Pakizeh, Mixed matrix membranes incorporated with cubic-MOF-5for improved polyetherimide gas separation membranes: theory and experiment, J.Ind. Eng. Chem. 20 (2014) 3857–3868.

[19] S. Basu, A. Cano-Odena, I.F. Vankelecom, MOF-containing mixed-matrix mem-branes for CO2/CH4 and CO2/N2 binary gas mixture separations, Sep. Purif.Technol. 81 (2011) 31–40.

[20] E.V. Perez, K.J. Balkus, J.P. Ferraris, I.H. Musselman, Mixed-matrix membranescontaining MOF-5 for gas separations, J. Membr. Sci. 328 (2009) 165–173.

[21] H.B. Tanh Jeazet, C. Staudt, C. Janiak, Metal-organic frameworks in mixed-matrixmembranes for gas separation, Dalton Trans. 41 (2012) 14003–14027.

[22] B. Zornoza, C. Tellez, J. Coronas, J. Gascon, F. Kapteijn, Metal organic frameworkbased mixed matrix membranes: an increasingly important field of research with alarge application potential, Micropor. Mesopor. Mater. 166 (2013) 67–78.

[23] Y. Zhang, I.H. Musselman, J.P. Ferraris, K.J. Balkus, Gas permeability properties ofMatrimid® membranes containing the metal-organic framework Cu–BPY–HFS, J.Membr. Sci. 313 (2008) 170–181.

[24] S. Japip, H. Wang, Y. Xiao, T. Shung Chung, Highly permeable zeolitic imidazolateframework (ZIF)-71 nano-particles enhanced polyimide membranes for gas se-paration, J. Membr. Sci. 467 (2014) 162–174.

[25] M. Safak Boroglu, A.B. Yumru, Gas separation performance of 6FDA-DAM-ZIF-11mixed-matrix membranes for H2/CH4 and CO2/CH4 separation, Sep. Purif. Technol.173 (2017) 269–279.

[26] R. Adams, C. Carson, J. Ward, R. Tannenbaum, W. Koros, Metal organic frameworkmixed matrix membranes for gas separations, Micropor. Mesopor. Mater. 131(2010) 13–20.

[27] E. Ahmadi Feijani, H. Mahdavi, A. Tavasoli, Poly(vinylidene fluoride) based mixedmatrix membranes comprising metal organic frameworks for gas separation appli-cations, Chem. Eng. Res. Des. 96 (2015) 87–102.

[28] M. Naseri, S.F. Mousavi, T. Mohammadi, O. Bakhtiari, Synthesis and gas transport

performance of MIL-101/Matrimid mixed matrix membranes, J. Ind. Eng. Chem. 29(2015) 249–256.

[29] F. Dorosti, M. Omidkhah, R. Abedini, Fabrication and characterization of Matrimid/MIL-53 mixed matrix membrane for CO2/CH4 separation, Chem. Eng. Res. Des. 92(2014) 2439–2448.

[30] N.L. Rosi, J. Eckert, M. Eddaoudi, D.T. Vodak, J. Kim, M. O'Keeffe, O.M. Yaghi,Hydrogen storage in microporous metal-organic frameworks, Science 300 (2003)1127–1129.

[31] J.A. Greathouse, M.D. Allendorf, The interaction of water with MOF-5 simulated bymolecular dynamics, J. Am. Chem. Soc. 128 (2006) 10678–10679.

[32] J.-M. Yang, Q. Liu, W.-Y. Sun, Co(II)-doped MOF-5 nano/microcrystals: solvato-chromic behaviour, sensing solvent molecules and gas sorption property, J. SolidState Chem. 218 (2014) 50–55.

[33] J.A. Botas, G. Calleja, M. Sánchez-Sánchez, M.G. Orcajo, Cobalt doping of the MOF-5 framework and its effect on gas-adsorption properties, Langmuir 26 (2010)5300–5303.

[34] H. Li, W. Shi, K. Zhao, H. Li, Y. Bing, P. Cheng, Enhanced hydrostability in Ni-dopedMOF-5, Inorg. Chem. 51 (2012) 9200–9207.

[35] J.-M. Yang, Q. Liu, W.-Y. Sun, Shape and size control and gas adsorption of Ni(II)-doped MOF-5 nano/microcrystals, Micropor. Mesopor. Mater. 190 (2014) 26–31.

[36] Y.-S. Bae, B.G. Hauser, O.K. Farha, J.T. Hupp, R.Q. Snurr, Enhancement of CO2/CH4selectivity in metal-organic frameworks containing lithium cations, Micropor.Mesopor. Mater. 141 (2011) 231–235.

[37] Q. Xu, D. Liu, Q. Yang, C. Zhong, J. Mi, Li-modified metal-organic frameworks forCO2/CH4 separation: a route to achieving high adsorption selectivity, J. Mater.Chem. 20 (2010) 706–714.

[38] L. Huang, H. Wang, J. Chen, Z. Wang, J. Sun, D. Zhao, Y. Yan Synthesis, mor-phology control, and properties of porous metal-oragnic coordination polymers,Mic. Meso Mater. 58 (2) (2003) 105–114.

[39] B. Panella, M. Hirscher, H.PütterU. Müller, Hydrogen adsorption in metal-organicframeworks: Cu-MOFs and Zn-MOFs compared, Adv. Funct. Mater. 6 (4) (2006)520–524.

[40] R. Lin, MOFs-based Mixed Matrix Membranes for Gas Separation, Master thesis, TheUniversity of Queensland, 2016.

[41] Y. Liu, Z. Ng, E.A. Khan, H.K. Jeong, C. Ching, Z. Lai, Synthesis of continuous MOF-5 membranes on porous α-alumina substrates, Mic. Meso Mater. 118 (1) (2009)296–301.

[42] D. Saha, B.T. Low, T.S. Chung, A.R. Greenberg, Polymeric membranes for the hy-drogen economy: contemporary approaches and prospects for the future, J. Membr.Sci. 327 (1) (2009) 18–31.

[43] S. Ullah, M. Bustam, S.D. Mohd, A. Elkhalifah, G. Murshid, N. Riaz, Synthesis, CO2adsorption study of modified MOF-5 with multi-wall carbon nanotubes and ex-pandable graphite, AIP Conf. Proc. 1621 (2014) 34–39.

[44] A.Y. Houde, S.S. Kulkarni, M.G. Kulkarni, Permeation and plasticization behavior ofglassy polymers: a WAXD interpretation, J. Membr. Sci. 71 (1992) 117–128.

[45] T.D. Kusworo, A.F. Ismail, Budiyono, I.N. Widiasa, J. Johari, Sunarso, The uses ofcarbon nanotubes mixed matrix membranes (MMM) for biogas purification, Int. J.Waste Resour. 2 (2012) 5–10.

[46] M. Ahmadi, S. Janakiram, Z. Dai, L. Ansaloni, L. Deng, Performances of mixedmatrix membranes containing porous two-dimensional (2D) and three-dimensional(3D) fillers for CO2 separation: a review, Membranes 8 (2018) 1–48.

[47] S. Shadid, K. Nijmeier, Performance and plasticization behavior of polymer–MOFmembranes for gas separation at elevated pressures, J. Membr. Sci. 420 (2014)166–177.

[48] J. Ploegmakers, S. Japip, K. Nijmeijer, Mixed matrix membranes containing MOFsfor ethylene/ethane separation Part A: membrane preparation and characterization,J. Membr. Sci. 428 (2013) 445–453.

[49] A.B. Fuertes, Adsorption-selective carbon membrane for gas separation, J. Membr.Sci. 177 (2000) 9–16.

[50] M.G. De Angelis, R. Gaddoni, G.C. Sarti, Gas solubility, diffusivity, permeability,and selectivity in mixed matrix membranes based on PIM–1 and fumed silica, Ind.Eng. Chem. Rest. 52 (2013) 10506–10520.

H. Aykac Ozen, B. Ozturk Separation and Purification Technology 211 (2019) 514–521

521