Separation and Purification Technology · 8 = 6). The most effective heating rate that has been applied in the prep-aration of the polymer-based carbon membrane is in the range of

Separation and Purification Technology 88 (2012) 174–183

Contents lists available at SciVerse ScienceDirect

Separation and Purification Technology

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

Effects of carbonization heating rate on CO2 separation of derivedcarbon membranes

W.N.W. Salleh a,b, A.F. Ismail a,b,⇑a Advanced Membrane Technology Research Centre (AMTEC), Universiti Teknologi Malaysia, 81310 Skudai, Johor Bahru, Malaysiab Faculty of Petroleum and Renewable Energy Engineering, Universiti Teknologi Malaysia, 81310 Skudai, Johor Bahru, Malaysia

a r t i c l e i n f o a b s t r a c t

Article history:Received 9 November 2011Received in revised form 16 December 2011Accepted 16 December 2011Available online 24 December 2011

Keywords:Carbon membraneHollow fiberCarbonizationCarbon dioxide

1383-5866/$ - see front matter � 2011 Elsevier B.V. Adoi:10.1016/j.seppur.2011.12.019

⇑ Corresponding author at: Advanced Membrane(AMTEC), Universiti Teknologi Malaysia, 81310 Skuda+60 7 5535592; fax: +60 7 5581463.

E-mail addresses: [email protected], afauzi@

High performance carbon hollow fiber membranes (CHFM) for CO2 separation were prepared by manip-ulating of carbonization heating rates under a N2 atmosphere. During heat treatment process, carboniza-tion of the polymeric hollow fiber membrane was conducted up to 650 �C and in the range of 1–9 �C/minfor heating rate. The gas permeation properties were determined using a single gas permeation apparatusat room temperature. Fine-tuning of the carbonization conditions appears necessary to obtain desiredpermeation properties. The gas permeance for all examined gases (N2, CH4, CO2) and selectivity of CO2/CH4 and CO2/N2 was decreased and increased, respectively, for the PEI/PVP-based CHFMs prepared atlow heating rate. It was found that the heating rate posses significant effect on structure and gas perme-ation properties of the resultant CHFMs. Experimental results indicated that, carbonization heating rateof 3 �C/min is the best conditions in the preparation of CHFMs derived from polymer blends of PEI/PVP forthe CO2/CH4 and CO2/N2 separation.

� 2011 Elsevier B.V. All rights reserved.

1. Introduction

Today, membrane science and technology have recognized aspowerful tools in solving some global problems and developingnew industrial processes required for a sustainable industrialgrowth [1]. The advantages contributed by membrane gas separa-tion are its simplicity in operation that can be fitted easily onto thepower plant without requiring complicated integration and noneed to add chemical or to regenerate an absorbent or adsorbent.In addition, membrane processes are now being explored for CO2

capture from power plant emissions and other fossil fuel based fluegas streams due to their fundamental engineering and economicadvantages over competing separation technologies. However,one of the challenges that needed to be overcome in the currentcommercial membranes are to develop membrane materials thatcan stands high temperature process which is suitable for CO2 cap-ture [2]. According to Brunetti and his coworkers [3], the mem-brane should posses several properties in order to be useful forthe capture of CO2, namely high CO2 permeability, high CO2/N2

selectivity, thermally and chemically resistance and plasticizationresistance. Membrane process as energy saving, easy to scale-up

ll rights reserved.

Technology Research Centrei, Johor Bahru, Malaysia. Tel.:

utm.my (A.F. Ismail).

and space-saving have been accepted to be the future technologyfor CO2 separation [4].

Carbon membranes have been extensively studied by previousresearchers in order to tackle this problem. Carbon membranesare typically produced by the heat treatment of various types ofpolymers under vacuum or inert gas atmospheres at temperaturerange of 500–1000 �C [5–9]. The high thermal and chemical stabil-ity of these membranes provide hope in gas separation applica-tions. The successful implementation of carbon membranes insuch applications depends on both, the types of precursor and car-bon membrane preparations. A comprehensive review on develop-ment of the various types of carbon membranes is provided byIsmails’ group [5].

Many studies reveal that the carbonization conditions imposestrong effect on the gas permeation properties of carbon mem-branes. One of the process parameters that give a strong influenceon the structure, separation performance, and transport mecha-nism of carbon membranes is carbonization heating rate. Thisheating rate would determine the evolution rate of volatile compo-nents from polymeric membrane during carbonization. Based onliterature, carbonization heating rate is typically carried out at awide range of 1–13 �C/min. Nevertheless, it is still depends onthe types of polymeric precursor membranes used [5]. Centenoet al. [10], has investigated the effect of carbonization heating rateon carbon membrane derived from phenolic resin by performingcarbonization up to 700 �C at different heating rate. It is showedthat the carbonization heating rates of up 10 �C/min shifted the

W.N.W. Salleh, A.F. Ismail / Separation and Purification Technology 88 (2012) 174–183 175

pore size distribution towards smaller pores and was beneficial forimprovement of molecular sieve characteristics of the resultantcarbon membrane. This membrane was successfully tested for sep-aration of permanent gas pairs (O2/N2 = 5, CO2/N2 = 27, and CO2/CH4 = 23) and olefin/paraffin (C2H4/C2H6 = 3 and C3H6/C3H8 = 6).The most effective heating rate that has been applied in the prep-aration of the polymer-based carbon membrane is in the range of1–5 �C/min.

Furthermore, great interest has been given in the preparation ofcarbon hollow fiber membrane (CHFM) for gas separation recentlybased on improved gas performance and thermal stability com-pared to those of polymeric membrane [11–18]. The hollow fibergeometry also were more preferable to other geometry such as flatsheet membranes because of the high packing density (>1000 m2/m3), self supporting characteristic, easy module construction aswell as useful for treating a large volume of gas stream [19]. Hence,an experiment was carried out by using hollow fiber geometry.

Current work was focused on the investigation of the prepara-tion and characterization of CHFMs derived from polyetherimide(PEI), which is considered as one of the most suitable polymersdue to its commercial availability, process ease and low cost. Thepotentialities and the gas permeation properties of the resultantCHFMs, giving general guidelines on some process parameters totake into account in the choice of a technology suitable for theCO2 separation. The relation between membrane properties andprocess parameter (carbonization heating rate) affecting the gaspermeation performance will be thoroughly analyzed. Further,the membrane performances for N2, CH4 and CO2 gas moleculesare compared with PEI based CHFMs.

2. Membrane preparation

2.1. Materials

A commercially available polyetherimide (PEI), Ultem 1000 waschosen as main polymer precursor in this study. This polymer wasblended with polyvinylpyrrolidone (PVP) (Fluka, K90). The poly-mer was dried overnight at 120 �C prior use. N-methyl-2-pyrroli-done (NMP) was used as the solvent of PEI/PVP to prepare thepolymeric hollow fiber membranes.

Fig. 1. Heat treatment profile.

2.2. Preparation of PEI/PVP-based carbon hollow fiber membrane(CHFM)

The polymeric hollow fiber membranes were fabricated from ahomogeneous dope solution consisting of 17 wt% PEI, 6 wt% PVPand 77 wt% NMP using the phase inversion technique. The dopesolution was extruded through a tube in orifice spinneret withnominal inside and outside diameters of 0.5 and 1.0 mm, respec-tively. Distilled water was used as the bore fluid. The externalcoagulation bath was filled with water and maintained at ambienttemperature (25 �C). The nascent hollow fiber from the spinneretpassed through a 8 cm air gap before reaching the external coagu-lant. The dope extrusion rate was 4.5 ml/min. The prepared PEI/PVP hollow fiber membranes were kept in water at room temper-ature for overnight to remove excess solvent present on the mem-branes. The membranes were then subjected to a simple solventexchange process by immersing it into ethanol and n-hexane for2 h each, followed by drying naturally for 24 h at room tempera-ture before the subsequent process. The slower evaporation rateof n-hexane solvent could prevent collapse of the pore structureand minimized the surface corrugations. After that, the membranewas stored in a clean environment for drying. At last, the polymerichollow fiber membranes were ready for heat treatment andtesting.

In the case of the CHFMs preparation, the polymeric hollow fi-ber membranes were placed at the center of Carbolite (ModelCTF 12/65/550) wire wound tube furnace with a programmableheating control systems (Eurotherm 2416CC). Experimental workwas carried out by manipulating heating rate parameters duringcarbonization step. In the first step, the membranes were stabilizedunder compressed air atmosphere (200 ml/min) up to 300 �C witha heating rate of 3 �C/min. At this stage, the membranes were heldfor 30 min to provide greater stability of the membranes to sustainhigh temperature process in the subsequent carbonization step.The stabilization step would also offer the potential to preventthe melting and fusion of the polymeric membranes and avoidexcessive volatilization of carbon element throughout the hightemperature process. As a result, the final carbon yielded fromthe precursor can be maximized [5]. Subsequently, the tempera-ture was increased to final carbonization temperature of 650 �Cwith the various heating rate (1, 3, 5, 7, and 9 �C/min) and holdingit constant for 30 min under N2 flow (200 ml/min). At last, themembrane was cooled down naturally to room temperature.

The reason why the N2 gas was used in carbonization step isthat, it could accelerates the carbonization reaction throughincreased gas-phase heat and mass transfer to form a more openporous matrix. The carbonization under inert gas flow also favoredthe volatile compounds release during carbonization and avoidedthe carbon deposition in the pores already formed [10,20]. Luaand Su [21] experienced that the carbonization under vacuum isfavorable for producing carbon membranes with high selectivitywhilst carbonization under N2 is favorable for preparing carbonmembranes with high permeability. This is in agreement withthe study reported by Song et al. [22]. The detailed heat treatmentprofile used in this study is illustrated in Fig. 1. The nomenclatureof resultant CHFMs is given in the form of CM – Heating Rate.

2.3. Membrane characterization

The outer surface and cross section morphology of the carbonhollow fiber membrane were observed under JEOL JSM-5610LVscanning electron microscopy (SEM). The microstructure proper-ties of resulting CHFMs were examined on X’Pert PRO X-raydiffractometer (XRD) from PANalytical with the diffraction angle2h from 10 to 50�. Ni-filtered CuKa radiation with a wavelengthof k = 1.54 Å was applied in the experiments. The interplanar dis-tance (d-spacing) of the CHFMs were calculated by the well-knownBragg equation, as follows:

nk ¼ 2d sin h ð1Þ

176 W.N.W. Salleh, A.F. Ismail / Separation and Purification Technology 88 (2012) 174–183

where d is the dimension spacing (Å), h is the diffraction angle (�), kis the X-ray wavelength (Å) and n is an integral number (1,2,3, . . .).

(a)

(b)

(c)

(d)

(e)

d=0.383 nm

d=0.381 nm

d=0.379 nm

d=0.377 nm

d=0.368 nm

d=0.207 nm

d=0.206 nm

d=0.205 nm

3. Gas permeation measurement

In this study, single gas permeation measurement instead ofmixed gas was used, since there is no article has been reportedregarding to this issue. As other articles, this paper covers specificcase studies. For instance, Lagorsses’ [23] and Favvass’ [24] groupsstudied the permeation properties of the carbon membranes onvarious pure gases (He, H2, Ar, O2, N2, CO2, CH4), but did not per-form a gas mixture testing in their studies. Other researchers thatreported on carbon membranes performance using single gaspermeation measurements were Favvas et al. [14] and Yoshimuneet al. [16]. Based on literature, single gas permeation measurementwas mostly applied for all of the fabricated carbon membranes andit could provide the trends of this material. It is reported that thepermeation test using pure and mixed gas did show similar results[25,26].

The hollow fiber membranes were assembles in a stainless steelmodule as a shell casing. A few pieces of the hollow fiber mem-branes were potted with epoxy resin for both ends, with one endsof the fiber bores was fully open (permeate side) and close, respec-tively. The effective length of each hollow fiber was 10 cm, whichcorresponds to a total permeation area of 21 cm2 in the membranemodule. Since the module design in the shell side feed configura-tion is relatively simple, most of the commercial hollow fibermembrane devices have adapted this configuration in the gas sep-aration systems. This is because, the hollow fiber membranes usedin industrial scale applications are relatively long (1–3 m) and thepressure build up in the fiber lumen can be substantial when ashell side feed configuration is employed [19,27]. Thus, currentwork will focus on the preparation of (CHFM) using shell side feedconfiguration for permeation test, as shown in Fig. 2.

The feed gas was admitted to shell side of the module, and thepermeate gas exited at atmospheric pressure from the bore side ofthe membranes. The flow rate of permeate was measured by a sim-ple soap film flow meter. Before the measurement, the high puritygas is introduced to the shell side, allowed to pass through thesystem for about one hour to ensure that there is no air insidethe system. After that, several data was collected and the averagevalues were used to assess the separation performance. Each mea-surement value is the result of several different membranes andthe precision in gas permeance for each membrane was found tobe within the error less than ±10%. The error that occurred wasprobably due to the fluctuation occurred during the data collectionby simple soap film flow meter.

From gas permeation test, the performance of the CHFMs can becharacterized by two important parameters: permeance and selec-tivity. The permeance of three pure gases with different molecularsizes; N2 (3.64), CH4 (3.80) and CO2 (3.3) [14], through the resultingCHFMs was measured, after each carbonization step. The test wasperformed at room temperature and 7 bar and always follows theorder of N2, CH4 and finally CO2 to prevent the strongly adsorbinggases from disturbing the performance of the carbon membranes.

Fig. 2. Shell side feed configuration.

The permeance, P (GPU) and selectivity, a of the membranes werecalculated using the following equations:

Permeance, P (GPU):

ðP=lÞiQ i

Dp � A ¼Q

npDDPð2Þ

1 GPU = 1 � 10�6 cm (STP) cm�2 s�1 cm Hg�1

Selectivity, a:

aA=B ¼PA

PB¼ ðP=lÞAðP=lÞB

ð3Þ

where P/l is the permeance of the hollow fiber (cm3 (STP)/cm2 s cm Hg), Qi is the volumetric flow rate of gas i at standard tem-perature and pressure (cm3 (STP)/s), Dp is the pressure differencebetween the feed side and the permeation side of the membrane(cm Hg), A is the membrane surface area (cm2), n is the numberof fibers in the module, D is an outer diameter of hollow fiber(cm) and l is an effective length of hollow fiber (cm).

4. Results and discussion

4.1. Microstructure properties

Braggs’ rule has been applied in the most research to determinethe distance between side-group atoms and skeletal atoms or dis-tance between atoms in neighboring planes in the amorphouspolymers by measuring the maximum intensity in the broad bandregion obtained from XRD measurement. XRD is a useful tool forunderstanding the organization of carbon at the molecular leveland provide qualitative and comparative assessment of the degreeof packing of microstructures. The most relevant peaks whenexamining polymer-based carbon are the (002) and (100) peaksas stated by Geiszler [28]. The (002) peak is commonly used todetermine the crystal length, which can be attributed to the tur-bostratic structure with randomly oriented graphitic carbon layers[29], while the (100) peak is related to the distance between car-bon atoms on the same plane.

10 20 30 40 502 θ

d=0.210 nm

d=0.208 nm

Fig. 3. XRD spectra at different heating rate (a) 1 �C/min, (b) 3 �C/min, (c) 5 �C/min,(d) 7 �C/min, and (e) 9 �C/min.

0 2 4 6 8 10

0

2

4

6

8

10

12

14

Perm

eanc

e (G

PU)

Sample

N2

CH4

CO2

CM-PEI

Fig. 4. Effect of carbonization heating rate on the gas permeation rate through theCHFMs (Error analysis is about ±10%).

W.N.W. Salleh, A.F. Ismail / Separation and Purification Technology 88 (2012) 174–183 177

XRD spectra for the CHFMs treated at different heating rate dur-ing carbonization step are shown in Fig. 3. The peak appeared onthe spectra suggests that the CHFMs have formed rigid conjugatedaromatic graphitic planes, making the structure more ordered andbetter packed [12]. The CHFMs prepared at different heating ratewere also showed the same crystalline structural characteristicswithout any strong point for the graphite formation proof. As ex-pected for ungraphitizable carbon membranes, all the samplesshowed wide peaks that indicative of the degree of amorphousnessand appeared below the corresponding value for graphite of0.335 nm with diffraction peak appears at 26.6� [30,31]. It is notedthat CHFMs prepared at 9 �C/min presents more intense peaks,suggesting a behavior closer to graphite and a more ordered struc-ture than those prepared at lower heating rate. Concerning theplane 002, the diffraction peak appears between 23.20 and24.15� for the prepared CHFMs under heating rate of 1–9 �C/min,corresponding respectively to a d-spacing of around 0.383–0.368 nm. The decrease of the d-spacing upon heating leaded tothe molecular sieving effect [32]. Another peak correlated to thegraphite (100) plane, which represents the repeated aromatic ringin graphitic structure between 2h = 43–45�, was observed in allsamples of the resultant CHFMs. The carbonization of the CHFMsshrunk the ordered structure, reducing the d-spacing from 0.383to 0.368 nm, which corresponds to 4% contraction, for the mem-branes prepared at heating rate of 1–9 �C/min.

In overall, there are two d-spacing peaks obtained after carbon-ization step, indicates the wide pore distribution in CHFMs struc-ture. The difference between a, b, c, d and e curves (see Fig. 3) isonly at the diffraction peak appeared at (002) and (100) planes,which results in different d-spacing values. This XRD patterns isin agreement with the literatures when the carbon membrane de-rived from Kapton [20], PAN [22], PEI/PVP [29], and PFA [33] wereinvestigated. Although amorphous, carbon membranes are en-dowed with a regular nanostructure that leads to a pore networkwith narrowly distributed pore dimensions [34].

4.2. Effect of carbonization heating rate on gas permeation properties

Based on literature, carbonization heating rate would determinethe evolution rate of volatile components from polymeric mem-brane during carbonization and it has effects on the microstructureof resultant carbon membranes [35]. Fig. 4 presents permeation re-sults for carbon membranes carbonized at different heating rate.The data represent an average value obtained from at least threeCHFMs and smaller error analysis of ±10% in both permeance valueand selectivity was achieved. As shown in Fig. 4, the data plotted at‘0’ (x-axis) referred to CHFMs derived from PEI at 650 �C with heat-ing rate of 3 �C/min. While the rest of the data was representCHFMs derived from polymer blends of PEI/PVP and plottedaccording to its carbonization heating rate. The CHFMs preparedfrom PEI/PVP display a better CO2/CH4 and CO2/N2 selectivity com-pared to those prepared from pure PEI. In contrast, the permeabil-ity notably increased up to about 5–8 times and selectivity of CO2/N2 decreased about 21% with the addition of PVP for flat sheet car-bon membrane produced from Rao et al. [29]. This is due to thepreparation of PEI derived carbon membranes using hollow fiberconfiguration with defect free surface areas and sufficient mechan-ical strength for gas separation measurement was more difficultcompared to those carbon membrane prepared using polymerblends precursor membrane and flat sheet configuration.

With the increase of heating rate from 1 to 5 �C/min, the gaspermeance of N2, CO2 and CH4 decreased about 83%, 86% and61%, respectively. This behavior was attributed to a narrowing ofthe carbon pore size distribution to smaller sizes with simulta-neous densification of the carbon structure. Many reports haveconfirmed that the pore size distribution alters towards smaller

pores with the increasing of the carbonization heating rate, whichcaused more restriction in degree of rotation freedom of the gases[10,36,37]. On the other hand, Centeno et al. [10] stated that thehigher carbonization heating rate could give a more random distri-bution of smaller pores in polymer-derived carbon membranes.This is because, as a consequence of the higher removal rate of vol-atile carbon compounds as the heating rate was increased, a partialcarbon vapor deposition may take place in pores previously cre-ated. This would result in less permeable carbon membranes butmuch more selective as carbonization heating rate increased. How-ever, when the CHFM was carbonized at 7 and 9 �C/min, the gaspermeance increased remarkably due the microscopic crack andpinholes formation on the membrane surface. This membranewas also tend to deform during the heat treatment process. In con-trast, in the case of phenolic resin-based supported carbon mem-branes, high performance membranes were obtained for carbonmembranes prepared at 10 �C/min. This is because, the preparationmethod used in supported membrane is far more different com-pared to self supporting membranes. As stated by Tins’ group[38], the pore distribution in the carbon membranes are not onlysignificantly dependent upon the carbonization conditions, butare also affected by the selection of polymer precursors. Theyfound that the chemical composition of polymer precursor is alsoone of the crucial factors that determine the pore population cre-ated in the carbon matrix.

In addition, rapid carbonization also could disturb the forma-tion of micropores in the CHFMs structure, which effecting theaverage porosity and pore size distribution of the resulting CHFMs.Such defect probably occurred when a short carbonization timewas performed. It is indicated that the formation of gaseous byproducts must occur at a sufficiently slow rate to avoid the forma-tion of blisters or other defects which can cause significant loss ofselectivity, since polymeric materials generally have carbon yieldsof 50% or less. Basically, the by product would react before they dif-fused out of the membranes. Shortening the carbonization timemay force out the byproducts to diffuse before they have a chanceto react [6]. The result is in agreement with study reported byHatori et al. [39].

CM-3 (PEI) CM-1 CM-3 CM-5 CM-7 CM-90

10

20

30

40

50

60

Sele

ctiv

ity

Sample

CO2/CH4

CO2/N2

Fig. 5. A trend on CO2/CH4 and CO2/N2 selectivity as a function of the carbonizationheating rate (Error analysis is about ±10%).

178 W.N.W. Salleh, A.F. Ismail / Separation and Purification Technology 88 (2012) 174–183

As can be seen in Fig. 4, a low gas permeance values were ob-tained, especially for N2 and CH4 gases. The limit of the membranearea that can be used in the hollow fiber test module could be oneof the reasons. Although these membranes have low permeancevalue, it is still capable of being tested in the module for gas per-meation testing in the laboratory scale. As shown in Fig. 5, theselectivity of CO2/CH4 and CO2/N2 significantly higher for CHFMsprepared at heating rate of 3 to 5 �C/min compared to those of 7and 9 �C/min. It is suggested that the narrower pore size distribu-tion could be achieved at longer contact with inert gas flow duringcarbonization step. As the heating rate of carbonization process

Fig. 6. SEM microphotographs of (a) outer surface an

raised from 5 to 9 �C/min, the selectivity of CO2/CH4 and CO2/N2 re-duced about 70% and 57%, respectively. Similar trends also can beobserved for carbon membrane prepared from polyimide [40].

4.3. Effect of carbonization heating rate on morphological structure

Figs. 6–11 depict the outer surface and cross section micropho-tographs of the precursor and carbon membrane prepared atdifferent carbonization heating rate (1–9 �C/min). It can be seenthat the PEI/PVP precursor membrane has some microporousstructure and few closed pores on the outer surface of the mem-branes. However, the size of microporous structure became smal-ler and invisible when the heat treatment process proceed up to650 �C and smooth surface with almost defect free were obtainedfor all the resultant carbon membrane except for carbon mem-brane prepared at 7 �C/min, which few defects was detected onthe surface of the membranes. It is indicated that the precursormembranes passes through a plastic stage during carbonizationprocess and the membrane surface became denser.

As can be seen in Figs. 6–11b, two dense layers, both outer andinner layer with porous sub-layer in between was obtained for pre-cursor and all the resultant carbon membranes. This structure wasgenerated during the dry/wet spinning process as the result of thephase inversion between polymer solution and coagulation liquid.This is revealed that any arrangement of the physical structure onprecursor membrane does not occur during the heat treatmentprocess. But, there is structure arrangement in the chemical pointof view since it forms amorphous carbon structure after carboniza-tion due to the breakdown of C–H bond. In addition, the diameterof precursor membrane was significantly reduced and the thick-ness of the skin layer increased as the heat treatment processproceeds. The porous sub-layer of the membranes also becomesmaller and shorter when high process temperature was applied.This is probably due to the effect of physical shrinking of the

d (b) cross section of the precursor membrane.

Fig. 7. SEM microphotographs of (a) outer surface and (b) cross section of the carbon membrane prepared at 1 �C/min.

Fig. 8. SEM microphotographs of (a) outer surface and (b) cross section of the carbon membrane prepared at 3 �C/min.

W.N.W. Salleh, A.F. Ismail / Separation and Purification Technology 88 (2012) 174–183 179

membranes caused by the decomposition and the evolution of thecompounds during the heat treatment process. Moreover, there

was a cross section deformation and irregularities were observedfor carbon membranes prepared at carbonization heating rate of

Fig. 9. SEM microphotographs of (a) outer surface and (b) cross section of the carbon membrane prepared at 5 �C/min.

Fig. 10. SEM microphotographs of (a) outer surface and (b) cross section of the carbon membrane prepared at 7 �C/min.

180 W.N.W. Salleh, A.F. Ismail / Separation and Purification Technology 88 (2012) 174–183

Fig. 11. SEM microphotographs of (a) outer surface and (b) cross section of the carbon membrane prepared at 9 �C/min.

Table 1Carbon membrane prepared by previous researchers.

Precursor Configuration CO2/CH4 CO2/N2 Ref.

PFA Flat sheet 7.5 [33]Sulfonated phenolic resin Tubular 54 [44]Matrimid Hollow fiber 20.86 23.6 [14]P84 co-polyimide Hollow fiber 38.9 42.8 [15]Cellulose acetate Hollow fiber 37.5 [18]Phenolic resin Tubular 101 39 [45]Kapton Flat sheet > 100 [20]PEI Flat sheet 17.5 [29]PEI/PVP Flat sheet 13.7 [29]BPDA-pp’ODA Tubular 40 [25]PEI Tubular 12.5 [41]PFR/PEG Tubular 20.2 [46]PI/PVP Flat sheet 30.00 – 38.00 [47]PEI/PVP Hollow fiber 55.33 41.50 Current work

W.N.W. Salleh, A.F. Ismail / Separation and Purification Technology 88 (2012) 174–183 181

5, 7, and 9 �C/min. This is in agreement with the studies reportedby Linkov et al. [17], Rao et al. [29], Coutinho et al. [11], Sedighet al. [41], and He et al. [18].

Based on the promising gas permeation properties and morpho-logical structure showed above, it is revealed that the most effec-tive carbonization heating rate that suitable to be applied for PEI/PVP-based CHFMS was at 3 �C/min. The selectivity of 55.33 and41.50 for CO2/CH4 and CO2/N2, respectively, were obtained. Thisresult is very good relative to current attractive polymericmembranes, which have CO2/CH4 and CO2/N2 selectivity typicallyranging from 30 to 50 and 25 to 45, respectively [42,43].

Furthermore, one of the drawbacks that have placed the carbonmembranes on the brink of commercialization is their brittleness,which means that they require careful handling. In this study, themechanical strength of the resultant CHFMs were not scientificallyanalyses but only observed during the handling (module construction)

of the membrane for gas permeation measurement. The CHFMs pre-pared from low carbonization heating rate (1 and 3 �C/min) wasfound to be easily mount into a module compared to those preparedfrom high carbonization heating rate (5, 7, and 9 �C/min). It is due tothe cross section deformation and irregularities of the membranesthat occur after heat treatment process. Although these membraneshave low mechanical properties, it is still capable of being con-structed into the module for gas permeation testing in the laboratoryscale and manifold studies on the gas permeation performance of thecarbon membrane have been reported in the literature. Coutinho etal. [11] have stated that the membrane fragility can also be indicatedby the existence of the cracks and defects on the external surface ofthe resultant carbon membranes. Based on the authors’ knowledge,details on the mechanical properties of the carbon membrane arestill lacking in the literature. The detailed explanation on this issuecan be found Ref. [5].

182 W.N.W. Salleh, A.F. Ismail / Separation and Purification Technology 88 (2012) 174–183

The experimental data obtained by this work and other researchgroups in the field of polymer-based carbon membranes are sum-marized in Table 1 for comparison purposes. The improvementsobserved in comparison with previous data are quite noticeable,since both the selectivity of CO2/CH4 and CO2/N2 has increased.As illustrated in Table 1, a broad range of polymeric materials havebeen investigated and different separation properties wereachieved by the time. It must be noted that the separation perfor-mance may differ considerably depending on the types of precur-sor, membrane configuration and synthesis procedures [34]. Theresearch on optimizing the carbon membranes for one gas systemmight also not be ideally suited for another system.

5. Conclusions

The appropriate carbonization conditions used during heattreatment process is critical to fabricate CHFMs with desired sepa-ration performance. The effects of heating rate during carboniza-tion on the morphological structure, microstructure properties aswell as gas transport properties of resulting CHFMs have beenstudied. Improvements in both CO2/CH4 and CO2/N2 selectivitywere observed upon changes in carbonization heating rate duringheat treatment process. CHFMs prepared at lower heating rate ofcarbonization were found to have better separation properties thanthose obtained from carbonization at higher heating rate. Based onoverall properties, the PEI/PVP-based CHFM prepared at 650 �Cwith heating rate of 3 �C/min was exhibited the best results withCO2/CH4 and CO2/N2 selectivity of 55.33 and 41.50, respectively.The use of PEI/PVP in a self supporting form was successfully fab-ricated even a low gas permeance was obtained. Research into thecontrol of the carbonization conditions in the formation of CHFMssuggests that these materials can be engineered to possess a widerange of properties.

Acknowledgements

One of the authors, W.N.W. Salleh, gratefully acknowledged thefinancial support under National Science Fellowship (NSF) from theMinistry of Science, Technology and Environment of Malaysia.

References

[1] P. Bernardo, E. Drioli, G. Golemme, Membrane gas separation: a review/state ofthe art, Ind. Eng. Chem. Res. 48 (2009) 4638–4663.

[2] M.T. Ho, G.W. Allinson, D.E. Wiley, Reducing the cost of CO2 capture from fluegases using membrane technology, Ind. Eng. Chem. Res. 47 (2008) 1562–1568.

[3] A. Brunetti, F. Scura, G. Barbieri, E. Drioli, Membrane technologies for CO2

separation, J. Membr. Sci. 359 (2010) 115–125.[4] H. Yang, Z. Xu, M. Fan, R. Gupta, R.B. Slimane, A.E. Bland, I. Wright, Progress in

carbon dioxide separation and capture: a review, J. Environ. Sci. 20 (2008) 14–27.[5] W.N.W. Salleh, A.F. Ismail, T. Matsuura, M.S. Abdullah, Precursor selection and

process conditions in the preparation of carbon membrane for gas separation:a review, Sep. Purif. Rev. 40 (2011) 261–311.

[6] S.M. Saufi, A.F. Ismail, Fabrication of carbon membranes for gas separation – areview, Carbon 42 (2004) 241–259.

[7] A.F. Ismail, L.I.B. David, A review on the latest development of carbonmembranes for gas separation, J. Membr. Sci. 193 (2001) 1–18.

[8] C. Nguyen, D.D. Do, K. Haraya, K. Wang, The structural characterization ofcarbon molecular sieve membrane (CMSM) via gas adsorption, J. Membr. Sci.220 (2003) 177–182.

[9] J. Chen, L.S. Loo, K. Wang, D.D. Do, The structural characterization of a CMSmembrane using Ar sorption and permeation, J. Membr. Sci. 335 (2009) 1–4.

[10] T.A. Centeno, J.L. Vilas, A.B. Fuertes, Effects of phenolic resin pyrolysisconditions on carbon membrane performance for gas separation, J. Membr.Sci. 228 (2004) 45–54.

[11] E.B. Coutinho, V.M.M. Salim, C.P. Borges, Preparation of carbon hollow fibermembranes by pyrolysis of polyetherimide, Carbon 41 (2003) 1707–1714.

[12] D.Q. Vu, W.J. Koros, S.J. Miller, High pressure CO2/CH4 separation using carbonmolecular sieve hollow fiber membranes, Ind. Eng. Chem. Res. 41 (2002) 367–380.

[13] L.I.B. David, A.F. Ismail, Short communication: influence of thethermastabilization process and soak time during pyrolysis process on the

polyacrylonitrile carbon membranes for O2/N2 separation, J. Membr. Sci. 213(2003) 285–291.

[14] E.P. Favvas, G.C. Kapantaidakis, J.W. Nolan, A.C. Mitropoulos, N.K.Kanellopoulos, Preparation, characterization and gas permeation propertiesof carbon hollow fiber membranes based on matrimid 5218 precursor, J.Mater. Process. Tech. 186 (2007) 102–110.

[15] E.P. Favvas, E.P. Kouvelos, G.E. Romanos, G.I. Pilatos, A.C. Mitropoulos, N.K.Kanellopoulos, Characterization of highly selective microporous carbon hollowfiber membranes prepared from a commercial co-polyimide precursor, J.Porous Mat. 15 (2008) 625–633.

[16] M. Yoshimune, I. Fujiwara, K. Haraya, Carbon molecular sieve membranesderived from trimethylsilyl substituted poly (phenylene oxide) for gasseparation, Carbon 45 (2007) 553–560.

[17] V.M. Linkov, R.D. Sanderson, E.P. Jacobs, Carbon membranes fromprecursors containing low-carbon residual polymers, Polym. Int. 35 (1994)239–242.

[18] X. He, J.A. Lie, E. Sheridan, M.B. Hagg, Preparation and characterization ofhollow fiber carbon membranes from cellulose acetate precursors, Ind. Eng.Chem. Res. 50 (2011) 2080–2087.

[19] L. Liu, A. Chakma, X. Feng, CO2/N2 separation by poly(ether block amide) thinfilm hollow fiber composite membranes, Ind. Eng. Chem. Res. 44 (2005) 6874–6882.

[20] J. Su, A.C. Lua, Effects of carbonization atmosphere on the structuralcharacteristics and transport properties of carbon membranes prepared fromKapton polyimide, J. Membr. Sci. 305 (2007) 263–270.

[21] A.C. Lua, J. Su, Effects of carbonization on pore evolution and gas permeationproperties of carbon membranes from kapton polyimide, Carbon 44 (2006)2964–2972.

[22] C. Song, T. Wang, Y. Qiu, J. Qiu, H. Cheng, Effect of carbonization atmosphere onthe structure changes of PAN carbon membranes, J. Porous. Mater. 16 (2009)197–203.

[23] S. Lagorsse, F.D. Magalhaes, A. Mendes, Carbon molecular sieve membranessorption, kinetic and structural characterization, J. Membr. Sci. 241 (2004)275–287.

[24] E.P. Favvas, G.E. Romanos, S.K. Papageorgiou, F.K. Katsaros, A.Ch. Mitropoulos,N.K. Kanellopoulos, A methodology for the morphological andphysicochemical characterization of asymmetric carbon hollow fibermembranes, J. Membr. Sci. 375 (2011) 113–123.

[25] K. Kusakabe, M. Yamamoto, S. Morooka, Gas permeation and microporestructure of carbon molecular sieving membranes modified by oxidation, J.Membr. Sci. 149 (1998) 59–67.

[26] B. Zhang, G. Shen, Y. Wu, T. Wang, J. Qiu, T. Xu et al., Preparation andcharacterization of carbon membranes derived from poly(phthalazinone ethersulfone) for gas separation, Ind. Eng. Chem. Res. 48 (2009) 2886–2890.

[27] M.B. Hagg, J.A. Lie, A. Lindbrathen, Carbon molecular sieve membranes apromising alternative for selected industrial applications, Ann. N.Y. Acad. Sci.984 (2003) 329–345.

[28] V.C. Geiszler, Polyimde precursor for carbon molecular sieve membranes, TheUniversity of Texas, Ph.D. Thesis, 1997.

[29] P.S. Rao, M.Y. Wey, H.H. Tseng, I.A. Kumar, T.H. Weng, A comparison of carbon/nanotube molecular sieve membranes with polymer blend carbon molecularsieve membranes for the gas permeation application, Microp. Mesop. Mat. 113(2008) 499–510.

[30] R. Hussain, R. Qadeer, M. Ahmad, M. Saleem, X-ray diffraction study of heat-treated graphitized and ungraphitized carbon, Turk. J. Chem. 24 (2000) 177–183.

[31] M.C. Campo, F.D. Magalhaes, A. Mendes, Comparative study between a CMSmembrane and a CMS adsorbent: Part I – Morphology, adsorption equilibriumand kinetics, J. Membr. Sci. 346 (2010) 15–25.

[32] H. Suda, K. Haraya, Molecular sieving effect of carbonized Kapton polyimidemembrane, J. Chem. Soc. Chem. Commun. (1995) 1179–1180.

[33] C.J. Anderson, S.J. Pas, G. Arora, S.E. Kentish, A.J. Hill, S.I. Sandler, G.W. Stevens,Effect of pyrolysis temperature and operating temperature on the performanceof nanoporous carbon membranes, J. Membr. Sci. 322 (2008) 19–27.

[34] M.B. Shiflett, H.C. Foley, Reproducible production of nanoporous carbonmembranes, Carbon 39 (2001) 1421–1446.

[35] A. Soffer, D. Rosen, S. Saguee, J. Koresh, Carbon membranes, GB 2, 207, 666(1989).

[36] H. Suda, K. Haraya, Gas permeation through micropores of carbon molecularsieve membranes derived from kapton polyimide, J. Phys. Chem. B. 101 (1997)3988–3994.

[37] 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) 18741–18748.

[38] P.S. Tin, Y. Xiao, T.S. Chung, Polyimide-carbonized membranes for gasseparation: structural, composition, and morphological control of precursors,Sep. Purif. Rev. 35 (2006) 285–318.

[39] H. Hatori, Y. Yamada, M. Shiraishi, M. Yoshihara, T. Kimura, The mechanism ofpolyimide pyrolysis in the early stage, Carbon 34 (2) (1996) 201–208.

[40] N. Nishiyama, W. Momose, Y. Egashira, K. Ueyama, Partially carbonizedpolyimide membranes with high permeability for air separation, J. Chem. Eng.Jpn. 36 (2003) 603–608.

[41] M.G. Sedigh, M. Jahangiri, P.K.T. Liu, M. Sahimi, T.T. Tsotsis, Structuralcharacterization of polyetherimide-based carbon molecular sievemembranes, AIChE J. 46 (2000) 2245–2255.

W.N.W. Salleh, A.F. Ismail / Separation and Purification Technology 88 (2012) 174–183 183

[42] G.C. Kapantaidakis, G.H. Koops, M. Wessling, S.P. Kaldis, G.P. Sakellaropoulos,CO2 plasticization of polyethersulfone/polyimide gas separation membranes,AIChE J. 49 (2003) 1702–1711.

[43] P.S. Tin, T.S. Chung, A.J. Hill, Advanced fabrication of carbon molecular sievemembranes by nonsolvent pretreatment of precursor polymers, Ind. Eng.Chem. Res. 43 (2004) 6476–6483.

[44] W. Zhou, M. Yoshino, H. Kita, K. Okamoto, Preparation and gas permeationproperties of carbon molecular sieve membranes based on sulfonated phenolicresin, J. Membr. Sci. 217 (2003) 55–67.

[45] A.B. Fuertes, Effect of air oxidation on gas separation properties of adsorption-selective carbon membranes, Carbon 39 (2001) 697–706.

[46] X. Zhang, H. Hu, Y. Zhu, S. Zhu, Carbon molecular sieve membranes derivedfrom phenol formaldehyde novolac resin blended with poly (ethylene glycol),J. Membr. Sci. 289 (2007) 86–91.

[47] Y.K. Kim, H.B. Park, Y.M. Lee, Gas separation properties of carbon molecularsieve membranes derived from polyimide/polyvinylpyrrolidone blends: Effectof the molecular weight of polyvinylpyrrolidone, J. Membr. Sci. 251 (2005)159–167.