Journal of Membrane Science Effects of sulfone/ketone in poly ...

Journal of Membrane Science 330 (2009) 319–325Contents lists available at ScienceDirectJournal of Membrane Sciencejournal homepage: www.elsevier.com/locate/memsciEffects of sulfone/ketone in poly(phthalazinone ether sulfone ketone) on the gaspermeation of their derived carbon membranesTonghua Wang a,∗ , Bing Zhang a,b , Jieshan Qiu a , Yonghong Wu b , Shouhai Zhang c , Yiming Cao da State Key Lab of Fine Chemicals, Carbon Research Laboratory, Department of Material Science and Chemical Engineering, School of Chemical Engineering,Dalian University of Technology, 158 Zhongshan Road, P.O. Box 49, Dalian, Liaoning 116012, Chinab School of Petrochemical Engineering, Shenyang University of Technology, 30 Guanghua Street, Liaoyang, Liaoning 111003, Chinac Department of Polymer Science and Materials, Dalian University of Technology, 158 Zhongshan Road, Dalian, Liaoning 116012, Chinad Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, ChinaarticleinfoabstractArticle history:Received 20 July 2008Received in revised form 17 December 2008Accepted 6 January 2009Available online 14 January 2009Keywords:Molecular sievingMicroporous carbonStabilizationPyrolysisGas separationA series of copolymer, poly(phthalazinone ether sulfone ketone)s (PPESKs) with the sulfone over ketoneunit (S/K) ratio varying from 20/80, 50/50 to 80/20, were used as precursors to prepare carbon membranes.The effects of chemical structure as S/K ratio of PPESKs on the microstructure and gas separationperformance of their derived carbon membranes were mainly investigated. The properties of PPESKs weredetected in terms of density, fractional free volume, char yield, interlayer distance and glass transitiontemperature. During the formation process of carbon membranes (i.e., stabilization and pyrolysis), thechanges in functional groups, microstructural parameters and gas permeation were monitored by FTIR,X-ray diffraction, TEM and single gas permeation techniques. The results have shown that the microstructureand gas permeation of obtained carbon membranes are significantly affected by the S/K ratio inprecursor PPESKs. Carbon membranes exhibit higher selectivity and lower permeability when preparedat low pyrolytic temperature (i.e., 650 ◦ C and 800 ◦ C) and from PPESKs with S/K ratio equaling 50/50,followed with 20/80 and 80/20. As for carbon membranes prepared at high pyrolytic temperature (i.e.,950 ◦ C), the selectivity order of them is well in accordance with S/K mole ratio in precursor PPESKs:20/80 > 50/50 > 80/20, and vice versa for permeability.© 2009 Elsevier B.V. All rights reserved.1. IntroductionSince membrane-based gas separations emerged as a commercialprocess on large-scale in 1980s, much attention has focusedon them due to their superior advantages over conventional cryogenicdistillation and pressure swing adsorption in terms of lowenergy consumption and capital investments, simple and easyoperation, and compact equipment [1,2]. Today, the gas separationmembranes have widely been used in various gas separationfields, such as the separation of hydrogen from the product streamof ammonia synthesis, the oxo process, the separation of oxygenfrom air, the removal of acid gases (e.g., CO 2 ,H 2 S and H 2 O)from natural gas, the recycling of helium, the separation of olefinsand alkanes from the processing of cracking products, and theseparation of SO 2 , NO x and CO 2 from exhaust fume [3–7]. Itis reported that many kinds of materials can be used to fabricategas separation membranes. Among them, polymers are themost commonly used materials. However, with the increasing∗ Corresponding author. Tel.: +86 411 38893968; fax: +86 411 38893968.E-mail address: wangth@chem.dlut.edu.cn (T. Wang).demands for membrane materials with excellent properties suchas the higher gas separation performance and application underthe aggressive atmosphere, the conventional polymeric membranematerials seem incapable due to their intrinsic shortcomings andthe inevitable “permeability–selectivity tradeoff” [1,2,8]. Therefore,many researchers have switched their interests to inorganic molecularsieving materials with desirable gas separation performance(e.g., zeolite and carbon membranes).Carbon membranes are one of the most promising membranematerials that have been rapidly developing in the past twodecades. Carbon membranes own the outstanding gas separationperformance, as well as thermal and chemical stability, etc.Those advantages endow the carbon membranes with more potentialapplications than polymeric membranes for most industrialgas separation and catalytic reactions [2,9,10]. Carbon membranesare basically prepared by pyrolysis of polymeric precursors underinert or vacuum atmosphere. Studies have shown that the gasseparation performance of carbon membranes is subjected bythe following several aspects: (1) pyrolytic protocols, e.g., heatingrate, atmosphere, final temperature and soak time [11,12];(2) modification (i.e., pre- or post-treatment, such as stabilization,activation or oxidation, chemical vapor deposition) [13,14];(3) the choice of precursors, such as poly(furfuryl alcohol) (PFA),0376-7388/$ – see front matter © 2009 Elsevier B.V. All rights reserved.doi:10.1016/j.memsci.2009.01.006

320 T. Wang et al. / Journal of Membrane Science 330 (2009) 319–325phenolic resin, poly(vinylidene chloride), polyacrylonitrile (PAN),cellulose derivates [9].Of the above-mentioned polymeric materials,the thermal-setting materials are believed to be more suitableto prepare the carbon membranes. Besides, the chemical structureof precursors is considered to be another important effecton the gas separation performance of carbon membranes. It hasshown that polymers featuring with bulky and rigid groups in theirmolecular chains are inclined to form membranes with high permeabilityand selectivity due to the higher fractional free volume (FFV)[1,8]. Because the frameworks of originally carbonaceous molecularchain in polymeric precursors would be preserved in the resultantcarbon matrix to a certain degree after pyrolysis of the precursors,it might suggest that the gas permeation of carbon membranesshould be tailored by designing the chemical structure of polymericprecursor at molecular level. Park et al. have found that the gas permeabilityof carbon membranes increased with increasing the FFVin precursor polyimide by introducing the methyl substitution tomolecular chain [15]. Xiao et al. reported that the gas permeabilityof carbon membranes was improved by increasing the FFV orlowering the thermal stability of precursor [16].In past five years, we have developed a promising polymericmaterial, poly(phthalazinone ether sulfone ketone)s (PPESKs), toprepare carbon membranes, of which the effects of pyrolyticprotocols and stabilization on the microstructure and gas permeationwere investigated [17]. The PPESKs-based carbon membranesexhibit a satisfactory gas separation performance with the O 2 /N 2selectivity above 24 under the optimal preparation conditions [18].In this paper, the relationships between chemical structure ofPPESKs and the gas permeation of their derived carbon membraneswere investigated.PPESKs are a kind of fully aromatic copolymer with a rigidmolecular chain and thermal resistance. The detailed synthesisprocedure of PPESKs was described elsewhere [19]. The chemicalstructure of PPESKs, as well as a three-dimensional drawingcreated by Hyperchem 7.0 software is shown in Fig. 1. Itcanbeseen that the noticeable element of the three-dimensional chemicalstructure is the difference in sulfone and ketone groups. Thelarge spatial restriction of –SO 2 – linkage in the backbone makes itsneighboring benzene rings nearly curled perpendicularly to eachother. The steric hindrance of ketone unit –CO– contributes to thedistortion between its neighboring benzene rings. Thus, the sulfoneand ketone units provide spacers and add to the free volumein the polymer matrix. Moreover, the rigid and bulky phthalazinonestructure in sulfone or ketone repeated unit contributes additionallymuch more distortion along the molecular chain. In a word, thespatial stereo-structure feature of PPESKs helps to inhibit the intersegmentalpacking and the segmental mobility of PPESK molecules.It is expected that the gas permeation of PPESK membranes andtheir derived carbon membranes would be adjusted by changingsulfone over ketone units molar ratio (S/K). In order to better understandand insight into the effects of S/K ratio in PPESKs on themicrostructure and gas separation performance of their derivedcarbon membranes, the evolutions of microstructure and gas permeationwere monitored throughout the heat treatment.2. Experimental2.1. Materials and membrane preparationA series of copolymer PPESKs with the S/K equaling 20/80, 50/50and 80/20 were supplied by Dalian New Polymer Company of China.The three PPESKs were correspondingly dissolved in N-methyl-2-pyrrolidone (NMP) solvent to form a concentration of 15 wt.%solutions by the aid of vigorous stirring and mild heating at 50 ◦ C.After filtering and de-foaming under a reduced pressure, the solutionswere cast into film form on horizontally clean glass plates seton heating platform at 80 ◦ C for 24 h. Then, the glass plates wereplaced in a vacuum drier at 100 ◦ C for another 24 h to further removethe residual solvent in polymeric membranes. Freestanding symmetricpolymeric PPESK membranes were obtained after peelingthe membranes from glass plates. Hereafter, the polymeric membranesprepared from PPESKs with the S/K ratio of 20/80, 50/50and 80/20 were designated as PPESK(20/80), PPESK(50/50) andPPESK(80/20), respectively.In order to prevent the membranes from melting and tar formingduring subsequent pyrolysis, a stabilization process was employedat 460 ◦ C for 30 min in the air at a heating rate of 3 ◦ C/min. Thestabilized membranes were symbolized as SM(x) (x is the S/K ratioin PPESKs).Carbon membranes were made by pyrolysis of stabilized membranesin a programmable controlled tubular furnace. The pyrolyticconditions were set at the final temperature of 650 ◦ C, 800 ◦ Cor950 ◦ C for 60 min with a heating rate of 1 ◦ C/min under argon atmosphereat a flowing rate of 200 mL/min. After the furnace coolingFig. 1. Chemical structure and three-dimensional drawing of PPESKs.

T. Wang et al. / Journal of Membrane Science 330 (2009) 319–325 321down to room temperature naturally, carbon membranes with thethickness 50 ± 2 m were obtained and stored in a dessicator toavoid the effects of H 2 O and CO 2 species in the air. Carbon membranesderived from PPESKs with the S/K ratio equaling of 20/80,50/50 and 80/20 were denoted as CM(x)-y (x and y stand for the S/Kmolar ratio and pyrolytic temperature, respectively).2.2. CharacterizationsThe thermal degradation behaviors of precursor PPESKs wereevaluated by a Mettler-Toledo TGA/SDTA851 thermogravimetricanalyzer in flowing nitrogen at a heating rate of 10 ◦ C/min from100 ◦ Cto900 ◦ C. In order to investigate the effects of stabilizationon the degradation behavior of PPESKs, flowing air was also adoptedas thermal degradation atmosphere by the same heating protocolas that in N 2 .Thermal stability indicator, glass transition temperatures (T g ),was measured for PPESK membranes and stabilized membranes bydynamic scanning calorimetry with a Mettler-Toledo DSC822 usingthe onset method. Heating runs were typically done at 10 ◦ C/minup to 500 ◦ C in nitrogen atmosphere.Density of the polymer PPESKs was determined at 25 ◦ Cbythedensity gradient column method using a density scale, PZ-B-5,made in the Shanghai Scales Factory of China.Based on van der Waals’ volumes of functional groups [20], vander Waals’ free volumes, V w of the repeated units of PPESKs withdifferent S/K ratio were calculated by the increment method offunctional groups [21]. According to the datum calculated hereinand the average molecular weight of the repeated units (M) anddensity () of the polymer PPESKs, the free volume V f and FFV werecalculated based on the following equations:V f =FFV = V fVM − 1.3 V w(1)(2)V = 1 (3)ATR-FTIR spectra of membrane samples were measured ona NEXUS TM FTIR spectrometer by an OMNIC sampler from thewavenumber of 4000–400 cm −1 . The use of ATR-FTIR provides anextra advantage by eliminating some of the problems associatedwith transmission infrared spectroscopy such as path length, concentration,and interference of water bands in the spectra due to thehygroscopic nature of KBr used for the preparation of conventionalFTIR pellets [22].X-ray diffraction (XRD) patterns were recorded using a D/Max-2400 diffractometer with CuK radiation in the range of diffractionangle 2 from 5 ◦ to 60 ◦ . The interlayer distance d 002 of sampleswere calculated by the well-known Bragg equation. The crystallitestacking height Lc and in-plane size La were calculated from the fullwidth at half maximum (FWHM) of the (0 0 2) and (1 0 0) diffractionpeaks using Scherrer’s equation, respectively [23].The microscopic morphology of carbon membranes wasobserved using a Philips TECNAI G 2 20 high-resolution transmissionelectron microscopy (HRTEM) on 200 kV. The selected-areaelectron diffraction (SAED) patterns were also taken to illustratethe general microstructure of carbon membranes.Gas permeation of membranes was tested by single componentgases through conventional variable volume–constantpressure method [24]. The detailed gas permeating test processwas described elsewhere [18]. The permeability was calculated bythe following equation (Eq. (4)) from the flux of pure gas on thepermeating side.P =Flux(4)A · p/lwhere p, A and l are the partial pressure difference of the gasestransmembrane, effective permeating area and the thickness of thetested membrane, respectively.The ideal separation factor or selectivity ˛ is calculated basedon the permeation rate of pure gases (Eq. (5)).˛ = P A(5)P Bwhere P A and P B are the permeation rate for gases A and B, respectively.To ensure good reproducibility, the replicate experiments foreach membrane were performed with more than three differentsamples prepared at the same time and the data of gas permeationappeared in this paper referred to the average results.3. Results and discussion3.1. Physical properties of PPESK polymeric membranesTable 1 shows the effects of S/K on some physical properties ofPPESKs, including density, molecular weight in repeated unit (M),FFV, glass transition temperature (T g ), inter-chain distance d spacing(d 002 ) and gas permeation. It can be found that the T g , M, FFVand d 002 of PPESKs increase with increasing their S/K ratio. A minimumdensity value of PPESKs is reached at the S/K molar ratioof 50/50. The change tendency of FFV with S/K is in good accordancewith d 002 , revealing that both FFV and d 002 can be regardedas indicators for the effective gas permeating volume in polymericmembranes. Like most other amorphous polymers, the gas permeabilityof PPESK membranes increases with the FFV rising. The T gchanges of PPESKs shown in Table 1 indicate that the thermal stabilityof molecular chains was improved with increasing the S/Kratio.Fig. 2 gives the thermally gravimetric analysis (TGA) of PPESKs.All the three thermal weight loss profiles are almost in the sametrend and are nearly superposed beneath 500 ◦ C. The 5 wt.% weightloss of PPESKs in N 2 is ca. 490 ◦ C that is much higher than traditionalresin. It demonstrates that PPESK polymers are a family ofhigh rank of thermal resistance resins. When the thermal degradationtemperature is elevated above 500 ◦ C, the weight loss curves ofPPESKs start to disjoin and the char yields follow with the order ofPPESK(50/50) > PPESK(20/80) > PPESK(80/20). The sequence of charyields against S/K ratio is maintained up to the final temperatureof 900 ◦ C, where the final char yields are 63%, 57% and 45% forPPESK(50/50), PPESK(20/80) and PPESK(80/20), respectively. TheTable 1Structure and property parameters of original PPESK membranes.Sample codes T g ( ◦ C) M (repeated unit) Density (g/cm 3 ) FFV (Bondi) d 002 (nm) Permeability (Barrer a ) SelectivityH 2 CO 2 O 2 N 2 H 2/N 2 CO 2/N 2 O 2/N 2PPESK(20/80) 275.6 425.3 1.348 0.1034 0.447 2.29 0.82 0.07 0.03 73.8 26.5 2.2PPESK(50/50) 283.7 436.1 1.345 0.1178 0.450 3.40 1.60 0.27 0.08 41.4 19.5 3.3PPESK(80/20) 291.4 446.9 1.359 0.1196 0.452 4.09 2.06 0.42 0.09 46.3 23.3 4.8a 1 Barrer = 10 −10 cm 3 (STP)·cm·cm −2 s −1 cmHg −1 .

322 T. Wang et al. / Journal of Membrane Science 330 (2009) 319–325Fig. 2. Thermal gravimetric analysis of PPESKs.Fig. 3. ATR-FTIR spectra of (a) original PPESK(50/50), (b) stabilized, and pyrolyzedat (c) 650 ◦ C, (d) 800 ◦ C and (e) 950 ◦ C, respectively.relatively low char yields for PPESK(80/20) and PPESK(20/80) aredue to the much removal of SO 2 by the breakage of weak linkage ofC–S bond in PPESK(80/20) and the asymmetric thermal degradationof PPESK(20/80) molecular chain [25].The thermal degradation of sample PPESK(50/50) in the air wasalso shown in Fig. 2. Compared to the thermal degradation behaviorof PPESK(50/50) in N 2 , there is no obvious difference before 500 ◦ C.However, the thermal degradation of PPESKs in the air presentsappreciably lower weight loss than that in N 2 during the temperatureof 500–560 ◦ C. It suggests that the stabilization atmosphereundoubtedly affect the thermal degradation behavior of PPESKs dueto the cross-linking reaction of PPESK molecular chains with oxygeninair[13,26].When the temperature is higher than 560 ◦ C, thechar yield of PPESKs in air is drastically reduced owing to excessiveoxidation. Therefore, the air stabilization temperature should notadopt too high. Here, we chose 460 ◦ C. The effects of stabilizationon the microstructure and gas permeation property of stabilizedmembranes will be discussed in the following section.3.2. Microstructure and gas permeation of stabilized membranesAfter stabilization process, it was found that the original PPESKmembranes were changed in apparent color, solubility and thermalproperty. The color of membrane was shifted from light yellow todark yellow; and the membrane craps were quite insoluble in thesolvent of NMP even kept for several weeks with the aid of heating at80 ◦ C. Furthermore, the T g values of stabilized membranes shown inTable 2 are increased by 7–10 ◦ C in comparison with that of originalPPESK membranes. The changes of stabilized membranes suggestthe formation of cross-linking structure in the matrix [13,26].To further insight into the formation of cross-linking structure,the functional groups of PPESK membranes through heat treatment(stabilization and pyrolysis) were monitored by ATR-FTIRspectra as shown in Fig. 3. For the original PPESKs, the reflectionbands can be assigned to Ar–O–Ar (1277 cm −1 and 1488 cm −1 ), C O(1667 cm −1 ) in diphenyl ketone, backbone vibration (1591 cm −1 )of benzene ring, C–N bond stretching (1326 cm −1 ), and O S O(1152 cm −1 and 1167 cm −1 ), respectively. The intensity for thoseaforementioned bands is drastically reduced for stabilized membranesexcept for carbonyl (1669 cm −1 ) and ether (1233 cm −1 and1305 cm −1 ) formed during stabilization process between polymermolecular chains and oxygen in the air. A new strong and broad peakappears at 3429 cm −1 in the stabilized membranes, which is due tothe hydroxyl bond formed by excess oxygen. This clearly indicatesthat a cross-linking structure, such as ester-like or anhydride-likethree-dimensional network cross-linking structure, may be formedbetween the molecular chains in the stabilized membranes duringstabilization. The detail description about the formation of crosslinkingstructure will be given in the later paper. The cross-linkingstructure in molecular chain would effectively prevent the meltingor softening of polymeric membranes during the subsequentpyrolysis [27]. By combining the results of TGA with ATR-FTIR, itis inferred that the molecular chains of stabilized samples are atan intermediate state, during which some minor thermal decompositionoccurs and no obvious carbon structure is formed yet.In the pyrolysis, the background of ATR-FTIR spectra of stabilizedmembranes becomes broadening and the reflection intensity offunctional groups is gradually weakening due to the thermal degradationand graphite-like structure formation with the pyrolytictemperature increasing from 650 ◦ Cto950 ◦ C.The sequences of d 002 values and gas permeability of stabilizedmembranes shown in Table 2 were in the order ofSM(50/50) < SM(20/80) < SM(80/20), which is different from thatof original polymeric membranes. This also illustrates that themicrostructure has changed in the stabilized membranes by crosslinking.And the differences in S/K ratio of PPESK polymers makethe cross-linking structure formed in stabilized membranes different.The PPESKs with the S/K equaling 50/50 is more likely helpfulto form a well cross-linking structure and the sample SM(50/50)exhibits a lower permeability and higher selectivity than SM(20/80)and SM(80/20). Compared to the gas permeability and selectivityof polymeric PPESK membranes shown in Table 1, those of sta-Table 2Gas permeation and property of stabilized membranes (measured at 30 ◦ C and 0.1 MPa).Sample codes T g ( ◦ C) d 002 (nm) Permeability (Barrer a ) SelectivityH 2 CO 2 O 2 N 2 H 2/N 2 CO 2/N 2 O 2/N 2SM(20/80) 286.1 0.431 32.56 15.54 4.25 0.37 87.7 41.9 11.5SM(50/50) 291.2 0.427 20.38 14.86 2.10 0.18 113.2 82.6 11.7SM(80/20) 305.8 0.445 44.81 20.73 5.91 0.54 83.0 38.4 10.9a 1 Barrer = 10 −10 cm 3 (STP) cm cm −2 s −1 cmHg −1 .

T. Wang et al. / Journal of Membrane Science 330 (2009) 319–325 323Table 3Gas permeation of carbon membranes prepared at 650 ◦ C, 800 ◦ C and 950 ◦ C (measuredat 30 ◦ C and 0.1 MPa).Sample codes Permeability (Barrer a ) SelectivityH 2 CO 2 O 2 N 2 H 2/N 2 CO 2/N 2 O 2/N 2CM(20/80)-650 465.91 325.85 77.52 9.43 49.4 34.5 8.2CM(50/50)-650 118.40 85.00 10.80 0.68 174.1 125.0 15.9CM(80/20)-650 1059.51 722.98 153.93 26.05 40.7 27.8 5.9CM(20/80)-800 50.38 31.90 7.70 0.47 107.7 68.2 16.8CM(50/50)-800 47.90 30.90 4.43 0.18 266.1 171.7 24.6CM(80/20)-800 69.70 47.90 8.10 0.72 97.5 67.0 11.3CM(20/80)-950 2.22 0.90 0.66 0.58 3.8 1.6 1.1CM(50/50)-950 6.21 1.76 1.66 0.62 10.0 2.8 2.7CM(80/20)-950 6.35 2.16 2.44 0.59 10.8 3.7 4.1a 1 Barrer = 10 −10 cm 3 (STP) cm cm −2 s −1 cmHg −1 .bilized membranes markedly increase in magnitude due to theformation of pore structure in the stabilized membranes by crosslinkingand the volatilization of small molecule gases producedby minor thermal degradation during stabilization. It is obviousthat the heat stabilization is beneficial to the membrane to forma porous microstructure and improve the gas separation propertyof the stabilized membranes.3.3. Microstructure and gas permeation of carbon membranesThe microstructure parameters of carbon membranes againstthe pyrolytic temperature were plotted in Fig. 4a–c. The interlayerdistance d 002 value of the carbon membranes derived fromthe same PPESKs decreases with elevating the pyrolytic temperaturefrom 650 ◦ C to 950 ◦ C. Simultaneously, the microcrystal sizeL a and stacking height L c of carbon membranes monotonouslyincrease with increasing the pyrolytic temperature. It illustratesthat the carbon crystallite gradually becomes larger and carbonmatrix becomes denser with elevating the pyrolytic temperature.As shown in Table 3, the gas permeability of carbon membranesderived from the same PPESKs monotonously decreaseswith increasing pyrolytic temperature from 650 ◦ C to 950 ◦ C. Theselectivity of carbon membranes markedly increases from 650 ◦ Cto800 ◦ C and reaches a maximum value at 800 ◦ C, then abruptly droptat 950 ◦ C. This change tendency was also found in the literature[28–30]. However, no one had given a detail interpretation aboutit. Here, we introduce a “Buiel’s model” to explain the evolution ofporous structure during the pyrolysis.Buiel et al. [31] divided the evolution of porous structure for acarbon material into three stages as shown in Fig. 5. The first ispore formation, which usually happened in the pyrolytic temperaturerange of 460–600 ◦ C. During this stage, porous structure isformed by the thermal degradation of non-carbon elements andvolatilization in the form of small gases or volatiles, such as H 2 O,O 2 ,CO 2 ,N 2 , HCN, SO 2 , etc. as shown in Fig. 5 for Type 1 porousstructure, which would lead to the increment of gas permeabilityfor carbon membranes. As the pyrolytic temperature increases from600 ◦ Cto800 ◦ C, another thermal reaction, thermal condensationreaction, becomes more dominant instead of the thermal degradationreaction. During this stage, the aromatic condensed structurewith the large area is formed by the combination of remaining carbonaceousdebris produced in the first stage in carbon membranematrix. The carbon structure change leads to the pore wall shrinkingand the formation of the Type 2 porous structure in carbonmembrane matrix. This is the second stage of the porous structureevolution. As a result, the gas permeability of carbon membranesis reduced with decreasing the inner diameters of pores caused bythe pore wall shrinking. In the third stage of the porous structureevolution, the pyrolytic temperature is further elevated to above800 ◦ C and the effects of pore wall shrinking are more obvious by theFig. 4. Effects of pyrolytic temperature and S/K ratio in precursors PPESKs of carbonmembranes on their structural parameters: (a) microcrystal height L c, (b) microcrystalsize L a and (c) interlayer distance d 002.drive of lowering potential energy of carbon structure as the thermalcondensation reaction occurs continuously. This causes somepores in the carbon matrix to be closed and forms the Type 3 porousstructure. Thus, the gas permeability of carbon membranes furtherdecreases with increasing the temperature from 800 ◦ Cto950 ◦ C.It is worth noting that the selectivity of carbon membranes alsodecreased as the pyrolytic temperature reaches up to 950 ◦ C. It canbe explained with the help of a “falling cards” model as shownin Fig. 6 [32]. The carbon membranes have a graphite-like turbostraticcarbon structure, which can be proved by HRTEM (Fig. 7).

324 T. Wang et al. / Journal of Membrane Science 330 (2009) 319–325Fig. 5. Evolution of pore structure of PPESK-based carbon membranes with pyrolytic temperature.Fig. 6. Falling cards phenomenon of carbon structure. Solid lines represent graphite-like sheets.The porous structure is composed of the aperture or gap formed bythe disordered stacking of graphite-like sheets or clusters. With theincrease of pyrolytic temperature up to higher temperature such asabove 950 ◦ C, the aperture or gap might be closed or collapsed andthe graphite-like sheet maybe falling down in parallel with anotheradjacent one due to the cause of lowering potential energy. As theresults, the original two small pores become one larger pore (dotline in Fig. 6) and the selectivity of carbon membranes is reducedas listed in Table 3. The complex porous structure evolution of carbonmaterials as shown in Figs. 5 and 6 is commonly ascribed tothe rearrangement and recombination of carbon structure drivenby the force of minimizing surface potential energy.Fig. 7. HRTEM with local enhanced images of carbon membranes CM(50/50)-800.The SAED pattern inserted in the TEM image show diffuse diffraction rings 1 and 2corresponding to graphite (0 0 2) and (1 0 0), respectively.For carbon membranes prepared from PPESKs with differentS/K ratio at lower pyrolytic temperature (i.e., 650 ◦ C and800 ◦ C), the d 002 and gas permeability follow the order of CM(50/50) < CM(20/80) < CM(80/20) (Fig. 4c and Table 3), and the selectivityis in reverse order for carbon membranes of CM(x)-650and CM(x)-800. This order is in good accord with the weightloss of PPESKs measured by TGA, i.e., PPESK(50/50) < PPESK(20/80) < PPESK(80/20). From the results of d 002 , thermal weight lossand gas permeability, it can be concluded that the carbon membraneexhibits a higher gas permeability when its carbon structurebecomes more loose with larger d 002 values or higher thermalweight loss. However, the sequences of d 002 values and gas permeabilityfor carbon membranes prepared at higher temperature (i.e.,950 ◦ C) are CM(20/80) < CM(50/50) < CM(80/20) and vice versa forselectivity. The microcrystal size L a and stacking height L c for carbonmembranes derived from different S/K ratio are in the order ofCM(50/50) > CM(20/80) > CM(80/20) throughout the pyrolytic temperaturefrom 650 ◦ C to 950 ◦ C. Those differences in microcrystalstructure and gas permeation clearly reflect the considerable effectsof the S/K ratio in PPESK precursor on the properties of carbonmembranes derived.It is interesting that the sequences of d 002 values, gas permeabilityand selectivity of carbon membranes prepared at 650 ◦ Cand 800 ◦ C are in good agreement with those of the stabilizedmembranes. And the sequences of the d 002 values and gas permeabilityof carbon membranes prepared at 950 ◦ C are differentfrom them. This shows that the cross-linking structure of stabilizedmembranes formed during stabilization is inherited to CM(x)-650and CM(x)-800 and eventually disappears in CM(x)-950 due to therearrangement and recombination of carbon matrix as the pyrolytictemperature increases up to 950 ◦ C. This suggests that the stabilizationand carbonization of precursor PPESKs would significantlyaffect the microstructure and gas separation property of carbonmembranes derived by the different S/K. The detail stabilizationmechanism about the PPESKs with different S/K is under study byour research group.Fig. 7 shows the HRTEM image of carbon membrane CM(50/50)-800, in which the SAED pattern and local enhanced image are

T. Wang et al. / Journal of Membrane Science 330 (2009) 319–325 325embedded. In the HRTEM image, the entangled “black” stripes areturbostratic carbon sheets or clusters that are believed to form thecomplex porous structure for gas separation. In the SAED pattern,there are two concentric diffraction rings: the inner clear ring correspondsto the diffraction of (0 0 2) plane, while the outer blurredring corresponds to the diffraction of (1 0 0) plane. It suggests thatthe carbon is poorly crystalline and low regularity in the carbonmembranes due to its very weak diffraction (1 0 0) ring in the SAEDpattern. The average structural parameter d 002 of CM(50/50)-800 isaround 0.371 nm obtained from SAED patterns and HRTEM images,which is in good accordance with the result of XRD.4. ConclusionsA series of copolymer PPESKs were used as precursors to preparecarbon membranes. It was found that the gas permeationand microstructure of carbon membranes could be well tunedby varying the chemical structure of precursors (i.e., sulfone overketone ratio (S/K)) and membrane preparation conditions. The initialpolymeric membranes and carbon membranes prepared athigh temperature (950 ◦ C) from the high S/K ratio of 80/20 exhibitgood gas separation performance. Stabilized membranes and carbonmembranes prepared at low temperature (650 ◦ C and 800 ◦ C)from the S/K ratio of 50/50 show the best selectivity due to its denseand regular microstructure. Oxidative stabilization prior to pyrolysiscan obviously improve the gas permeability and selectivityof initial polymeric membranes by forming cross-linking structure.This cross-linking structure also influences the gas separationperformance of carbon membranes prepared at lower pyrolytictemperature (650–800 ◦ C). The promising results presented hereimply that the gas permeation of carbon membranes could be welltailored by designing the chemical structure of initial polymericprecursors.AcknowledgementsThis work was supported by the National Natural Science Foundationof China (Nos. 20276008 and 20776024) and the VisitingScholar Foundation of State Key Laboratory of Fine Chemicals inDalian University of Technology.References[1] G. Maier, Gas separation with polymer membranes, Angew. Chem. Int. Ed. 37(1998) 2961.[2] W.J. Koros, R. Mahajan, Pushing the limits on possibilities for large scale gasseparation: which strategies? J. Membr. Sci. 175 (2000) 181.[3] C.E. Powell, G.G. Qiao, Polymeric CO 2/N 2 gas separation membranes for thecapture of carbon dioxide from power plant flue gases, J. Membr. Sci. 279 (2006)1.[4] M. Freemantle, Membranes for gas separation, Chem. Eng. News 83 (2005) 49.[5] A. Basu, J. Akhtar, M.H. Rahman, M.R. Islam, A review of separation of gasesusing membrane systems, Pet. Sci. Technol. 22 (2004) 1343.[6] R.W. Baker, Future directions of membrane gas separation technology, Ind. Eng.Chem. Res. 41 (2002) 1393.[7] P. Pandey, R.S. Chauhan, Membranes for gas separation, Prog. Polym. Sci. 26(2001) 853.[8] L.M. Robeson, Polymer membranes for gas separation, Curr. Opin. Solid StateMater. Sci. 4 (1999) 549.[9] S.M. Saufi, A.F. Ismail, Fabrication of carbon membranes for gas separation—areview, Carbon 42 (2004) 241.[10] N. Itoh, K. Haraya, A carbon membrane reactor, Catal. Today 56 (2000) 103.[11] J. Hayashi, M. Yamamoto, K. Kusakabe, S. Morooka, Effect of oxidation on gaspermeation of carbon molecular sieving membranes based on BPDA-pp ′ ODApolyimide, Ind. Eng. Chem. Res. 36 (1997) 2134.[12] V.C. Geiszler, W.J. Koros, Effects of polyimide pyrolysis conditions on carbonmolecular sieve membrane properties, Ind. Eng. Chem. Res. 35 (1996) 2999.[13] A. Bos, I.G.M. Punt, M. Wessling, H. Strathmann, Plasticization-resistant glassypolyimide membranes for CO 2/CH 4 separations, Sep. Purif. Technol. 14 (1998)27.[14] J.N. Barsema, S.D. Klijnstra, J.H. Balster, N.F.A. van der Vegt, G.H. Koops, M.Wessling, Intermediate polymer to carbon gas separation membranes basedon Matrimid PI, J. Membr. Sci. 238 (2004) 93.[15] H.B. Park, Y.K. Kim, J.M. Lee, S.Y. Lee, Y.M. Lee, Relationship between chemicalstructure of aromatic polyimides and gas permeation properties of their carbonmolecular sieve membranes, J. Membr. Sci. 229 (2004) 117.[16] Y. Xiao, T.-S. Chung, M.L. Chng, S. Tamai, A. Yamaguchi, Structure and propertiesrelationships for aromatic polyimides and their derived carbon membranes:experimental and simulation approaches, J. Phys. Chem. B 109 (2005) 18741.[17] B. Zhang, T. Wang, S. Liu, S. Zhang, J. Qiu, Z. Chen, H. Cheng, Structureand morphology of microporous carbon membrane materials derived frompoly(phthalazinone ether sulfone ketone), Micropor. Mesopor. Mater. 96 (2006)79.[18] B. Zhang, T. Wang, S. Zhang, J. Qiu, X. Jian, Preparation and characterizationof carbon membranes made from poly(phthalazinone ether sulfone ketone),Carbon 44 (2006) 2764.[19] X.G. Jian, Y. Dai, L. Zeng, R.X. Xu, Application of poly(phthalazinone ether sulfoneketone)s to gas membrane separation, J. Appl. Polym. Sci. 71 (1999) 2385.[20] D.W. Van krvelen, P.J. Hoftyzer, Properties of Polymers, 2nd ed., Elsevier ScientificPublishing Company, New York, 1976 (Chapter 4).[21] A. Bondi, Van der Waals volumes and radii, J. Phys. Chem. 68 (1964) 441.[22] T. Tanaka, S. Nagao, H. Ogawa, Attenuated total reflection fourier transforminfrared (ATR-FTIR) spectroscopy of functional groups of humic acid dissolvingin aqueous solution, Anal. Sci. 17 (2001) i1081.[23] B.E. Warren, X-ray diffraction in Random layer lattices, Phys. Rev. 59 (1941)693.[24] S.A. Stern, P.J. Gareis, T.F. Sinclair, P.H. Mohr, Performance of a versatile variablevolumepermeability cell. Comparison of gas permeability measurements bythe variable-volume and variable-pressure methods, J. Appl. Polym. Sci. 7 (1963)2035.[25] B. Nandan, L.D. Kandpal, G.N. Mathur, Poly(ether ketone)/poly(aryl ether sulphone)blends: thermal degradation behaviour, Eur. Polym. J. 39 (2003) 193.[26] S.-I. Kuroda, K. Mita, Degradation of aromatic polymers—II. The crosslinkingduring thermal and thermo-oxidative degradation of a polyimide, Eur. Polym.J. 25 (1989) 611.[27] J. Drbohlav, W.T.K. Stevenson, The oxidative stabilization and carbonization ofa synthetic mesophase pitch. Part I. The oxidative stabilization process, Carbon33 (1995) 693.[28] T.A. Centeno, J.L. Vilas, A.B. Fuertes, Effects of phenolic resin pyrolysis conditionson carbon membrane performance for gas separation, J. Membr. Sci. 228 (2004)45.[29] H. Kita, M. Yoshino, K. Tanaka, K.-i. Okamoto, Gas permselectivity of carbonizedpolypyrrolone membrane, Chem. Comm. (1997) 1051.[30] Y. Kusuki, H. Shimazaki, N. Tanihara, S. Nakanishi, T. Yoshinaga, Gas permeationproperties and characterization of asymmetric carbon membranes prepared bypyrolyzing asymmetric polyimide hollow fiber membrane, J. Membr. Sci. 134(1997) 245.[31] E.R. Buiel, A.E. George, J.R. Dahn, Model of micropore closure in hard carbonprepared from sucrose, Carbon 37 (1999) 1399.[32] J.R. Dahn, W. Xing, Y. Gao, The “falling cards model” for the structure of microporouscarbons, Carbon 35 (1997) 825.