Study of Morphology and Gas Separation Properties ofPolysulfone/Titanium Dioxide Mixed Matrix Membranes

Pourya Moradihamedani,1 Nor Azowa Ibrahim,1 Wan Md Zin Wan Yunus,2 Nor Azah Yusof1

1 Department of Chemistry, Faculty of Science, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia2 Department of Defence Science, Faculty of Defence Science and Technology, National Defence University ofMalaysia, Sungai Besi Camp, Kuala Lumpur, Malaysia

Polysulfone (PSf)-based mixed matrix membranes(MMMs) with the incorporation of titanium dioxide (TiO2)nanoparticles were prepared. Distribution and agglomera-tion of TiO2 in polymer matrix and also surface of mem-branes were observed by scanning electron microscopy,transmission electron microscopy, and energy dispersiveX-ray. Variation in surface roughness of MMMs with differ-ent TiO2 loadings was analyzed by atomic force micros-copy. Physical properties of membranes before and aftercross-linking were identified through thermal gravimetricanalysis. At low TiO2 loadings (�3 wt%), both CO2 andCH4 permeabilities decreased and consequently gasselectivity improved and reached to 36.5 at 3 bar pres-sure. Interestingly, PSf/TiO2 3 wt% membrane did notallow to CH4 molecules to pass through the membraneand this sample just had CO2 permeability at 1 bar pres-sure. Gas permeability increased considerably at high fil-ler contents (�5 wt%) and CO2 permeance reached to37.7 GPU for PSf/TiO2 7 wt% at 7 bar pressure. It wasdetected that, critical nanoparticle aggregation hasoccurred at higher filler loadings (�5 wt%), which contrib-uted to formation of macrovoids and defects in MMMs.Accordingly, MMMs with higher gas permeance and lowergas selectivity were prepared in higher TiO2 contents (�5wt%). POLYM. ENG. SCI., 00:000–000, 2014. VC 2014 Society ofPlastics Engineers

INTRODUCTION

Nowadays, membrane technology has known as an energy

efficient and inexpensive method in gas separation treatments

[1]. Since polymeric membranes have the preferred mechanical

and thermal properties as well as the flexibility to be operated

into different modules [2–4], they are presently the main materi-

als for gas separation procedures such as natural gas upgrading,

separation of CO2 from CH4, recovery and purification of

hydrogen, air and flue gas separation, etc. Despite these advan-

tages, polymeric membranes presented low CO2 permeance and

moderate CO2/CH4 selectivity and suffered from tradeoff

between permeability and selectivity where high gas permeance

and selectivity could not be attained simultaneously. However,

because of cost considerations and the ease of processing, they

remain the predominant choice for industrial applications [5–7].

Mixed matrix membranes (MMMs) which comprised of inor-

ganic materials distributed in polymeric phase, are based on

solid-solid system. When inorganic nanoparticles are added to

the polymeric phase, it is predictable that resulting membrane

separation properties become better than customary polymeric

membranes due to the high gas permeance and selectivity [8–

10]. Fillers are typically categorized into two main classifica-

tions namely nonporous and porous [11]. Zeolite as usual porous

filler has been widely used in many MMMs. Leo et al. [5] ele-

vated the membrane gas separation properties by introduction of

5 wt% of zeolite to PSf membrane that revealed promoted CO2/

CH4 and CO2/N2 selectivities of 25 and 22, respectively. In

another study, influence of zeolitic imidazolate frameworks on

the performance of ZIF8-poly (1,4-phenylene ether-ether-sul-

fone) hybrid membranes was considered by Luis et al. [12]. The

higher ZIF-8 content increases the permeability without touch-

ing the ideal selectivity in a significant way. Consequently, the

higher zeolite concentration the better the membrane gas trans-

port properties. Presence of nonporous silica nanoparticles in

MMMs for gas separation was investigated in many studies

[13–16]. For instance, Separation of CO2 from CH4 using poly-

sulfone/polyimide silica nanocomposite membranes was exam-

ined by Rafiq et al. [13]. They reported that, CO2 permeance

increased to 73.7 6 0.2 GPU with the addition of 5.2 wt% of

silica into the PSF/PI-20% blend. Interestingly, the highest CO2/

CH4 selectivity of 61 was observed by increasing silica content

to 15.2 wt%. Omidkhah et al. [17] investigated the effect of

inclusion of TiO2 nanoparticles on MMMs based on

Matrimd5218 prepared using solution-casting method. Their

results demonstrated that presence of TiO2 nanoparticles

increases the gas permeability of MMMs probably because of

chain packing distraction, void development at polymer–nano-

particle interface and nanoparticle accumulation. Also, their

results revealed that inclusion of TiO2 nanoparticles improves

membrane performance for CO2/CH4 separation and presents a

trade-off line with a similar slope compare to Robeson upper

insertion bound. In another work, Liang et al. [18] reported that,

CO2/CH4 selectivity improved from 24.5 for polyethersulfone-

based membrane to a maximum of 38.5 in MMMs containing 4

wt% of TiO2 and then a reduction was observed in higher TiO2

contents (e.g., 17.3 for 20 wt%). The void formation as well as

membrane defects in MMMs contributed to the high gas perme-

ability and low gas selectivity in higher TiO2 loadings.

The novelty of this paper is to explain how the synthesis of

PSf/TiO2 MMMs could significantly enhance performance of

membrane in term of CO2/CH4 separation. In addition, this is

the first time that PSf-based MMMs with the incorporation of

TiO2 nanoparticles were used for CO2/CH4 separation. Interest-

ingly, PSf/TiO2 3 wt% did not have CH4 permeability at 1 bar

feed pressure. While, the amount of CO2 permeance at same

pressure for mentioned membrane was 0.26 GPU. Accordingly,

PSf/TiO2 3 wt% membrane was able to separate CO2 from CH4

Correspondence to: Nor Azowa Ibrahim;

e-mail: norazowa@upm.edu.my

DOI 10.1002/pen.23887

Published online in Wiley Online Library (wileyonlinelibrary.com).

VC 2014 Society of Plastics Engineers

POLYMER ENGINEERING AND SCIENCE—2014

completely at 1 bar feed pressure. The small diameter of nanopar-

ticles (3 nm) as well as high specific area (500 m2/g) improves

TiO2 distribution and prevents non-selective void development in

polymer/nanoparticles matrix interface. There is no potential for

TiO2 nanoparticles to fuse together inherently and then they dis-

perse individually [19, 20]. According to the above facts, TiO2

was selected for this research. The permeability of pure CO2 and

CH4 gases as well as CO2/CH4 selectivity of prepared PSf/TiO2

MMMs were determined in different TiO2 loadings. The morphol-

ogy of resulted membranes was observed by scanning electron

microscopy (SEM) and TiO2 nanoparticles distribution on the

membrane top layer was characterized by Energy dispersive X-

ray (EDX) and transmission electron microscopy (TEM) analysis.

Variation in surface roughness of prepared membranes in different

content of TiO2 was evaluated by Atomic force microscopy

(AFM) images. Finally, the structural analysis of pure PSf and

TiO2 filled PSf was performed by FTIR spectra.

THEORY

Gas Transport Experiments

Gas transport mechanism through a nonporous polymeric mem-

brane is typically explained by the solution–diffusion phenomenon.

Based on this mechanism, the permeants dissolve into the polymer

matrix at the upstream face (high pressure), afterward diffuse

through the polymer film and lastly desorb from the downstream

side. Gases with a larger molecular diameter diffuse slower across

the prepared membrane [21–23]. Therefore, due to the larger

molecular diameter of CH4 in comparison with CO2 [23], PSf/TiO2

MMMs are able to separate these two gases by different selectiv-

ities. Stainless steel filter holder which was equipped with a back-

pressure support screen with effective area of 13.8 cm2 (Merck,

Frankfurter, Darmstadt, Germany) was used for gas permeation

experiments. Glass soap bubble flow meter (Sigma-Aldrich, St.

Louis, MO, USA) was employed for measuring rate of permeate

stream. Glass soap bubble flow meter is useful for measuring any

gas flow rate and it gives accurate measurement [22, 23]. The gases

below the surface of a soap bubble solution and the bubble moves

up the flow meter. We time the leading edge of the bubble from

one line to another. To ensure accuracy in our experiments, the gas

permeation test was repeated three times in the steady state.

Permeability across polymer matrix can be described as fol-

lows [24]:

ðP=LÞ ¼ Q=ðA3DPÞ (1)

where P is permeability, L is membrane active layer thickness, Qis gas flow (at standard pressure and temperature), A is the mem-

brane effective area in cm2, and DP is the differential partial pres-

sure through the membrane. The common unit of permeance is

GPU and 1 GPU is equal to 1 3 1026 cm3 (STP)/cm2 s cmHg.

Equation (2) can be used to calculate CO2/CH4 selectivity (a),

where Pi and Pj are CO2 and CH4 permeance, respectively [25].

a ¼ Pi=Pj (2)

EXPERIMENTAL

Materials

Polysulfone (PSf) with the number average molecular weight

(Mn) of 22,000 was used as a base polymer due to its satisfac-

tory gas permeance and selectivity purchased from Sigma-

Aldrich, St. Louis, MO, USA. PSf is an amorphous thermoplas-

tic polymer with a glass transition temperature of 190�C. Also,

PSf is a flame retardant polymer, possesses high mechanical,

thermal and oxidative stability and is soluble in common

organic solvents [25]. Titanium dioxide (TiO2) nanoparticles

used in this study from Nanoscale, Manhattan, KS, USA.

According to the provider data sheet, the specific area of nano-

particles is 500 m2/g with a density of 3.7 g/cm3. The physical

properties of TiO2 nanoparticles are shown in Table 1. Dime-

thylformamide (DMF) and isopropyl alcohol (IPA) were used as

solvent and non-solvent, respectively, from Merck, Frankfurter,

Darmstadt, Germany. CO2 and CH4 gases were provided in 40

L cylinders with a purity of 99.99%. Distilled water was used as

the second coagulation bath. The PSf resin was dried in an oven

at 80�C for 24 h before the usage.

Membrane Preparation

In this work, flat sheet PSf/TiO2 MMMs were prepared

through wet/wet phase inversion technique. In this method,

membranes are developed through contacting wet polymer film

with two non-solvent baths within the series. The primary coag-

ulation bath is utilized to get a concentrated layer of polymer at

the interface. This step produces the ultra-thin top layer. The

purpose of second bath is the particular coagulation and the ulti-

mate film formation. Casting solutions containing 25 wt% of

PSf with different TiO2 loadings (0.0, 1.0, 3.0, 5.0, and 7.0

wt%) were prepared employing DMF as solvent. Concentration

of PSf in the polymer solution was kept constant at 25 wt%

keeping concentration of DMF and TiO2 in 75 wt%. The casting

solution compositions of membranes are shown in Table 2. The

polymer solution was stirred for at least 24 h. The polymer/

nanoparticle matrix was then sonicated to ensure that TiO2 dis-

persed consistently in PSf solution then left standing for 1 day.

Thirty minutes before casting, the mixture was sonicated again

and then cast on a smooth glass plate by film casting knife to a

TABLE 1. Physical properties of TiO2 nanoparticles.

PSf (wt%) TiO2 (wt%) DMF (wt%)

25 0.0 75.0

25 1.0 74.00

25 3.0 72.00

25 5.0 70.00

25 7.0 68.00

TABLE 2. Composition of casting solutions.

Physical properties

Specific surface area (BET) 500 m2/g

Crystal size Amorphous

Average pore diameter 32 A

Total pore volume �0.4 cc/g

Bulk density 0.6 g/cc

True density 3.7 g/cc

Mean aggregate size 5 mm

Moisture content �4%

Ti content (based on metal) >99.999%

2 POLYMER ENGINEERING AND SCIENCE—2014 DOI 10.1002/pen

thickness of 350 lm. The color of wet film changed from trans-

parent to white immediately after soaking in the first coagula-

tion bath containing IPA65%. Immersion time in the first

coagulation bath was 90 s and after replacement of solvent by

non-solvent, prepared membrane dipped in the distilled water

for 24 h as a second coagulation bath. Finally, prepared mem-

brane was dried at room temperature condition for 1 day.

Membrane Characterization

In order to explore the prepared membranes morphology and

evaluate the distribution and agglomeration of TiO2 nanopar-

ticles in a polymer matrix, cross section photos were taken by

SEM (LEO 1455 SEM, LEO & Leica factory, Cambridge, UK)

and TEM (LEO 912AB TEM, LEO & Leica factory, Cam-

bridge, UK). AFM (Ambios Q-scope, Linthicum heights, MD,

USA) in tapping mode was used to analyze variations in surface

roughness parameters of prepared MMMs. EDX analysis was

also conducted by means of INCA instrument, (Oxford Instru-

ments, Abingdon, Oxfordshire, UK) to confirm distribution of

the TiO2 nanoparticles on the MMMs active layer. A Fourier

transform-infrared spectroscopy (Series100 PerkinElmer FT-IR

1650, Waltham, Massachusetts, USA) in the wavenumber range

280–4000 cm21 was used to assess synthesized PSf and PSf/

TiO2 membranes at room temperature. Variations in physical

properties of membranes before and after cross-linking were

identified through thermal gravimetric analysis (TGA) (Perki-

nElmar TGA7, Waltham, MA, USA) with a heating rate of

10 �C/min from room temperature up to 700�C.

RESULTS AND DISCUSSION

Morphology

Gas transport properties of MMMs are strongly dependent on

the morphology of dispersed phase [17]. The morphology and

inorganic filler distribution of MMMs were observed by SEM

and TEM, and the related images are shown in Figs. 1–3. Figure

1 shows the SEM cross section view of pure PSf membrane in

low and high magnifications. Figures 2 and 3 also present the

SEM and TEM images of PSf-based MMMs with different TiO2

loadings. According to Figs. 2 and 3, nanopaticles distribute

homogeneously in polymer matrix in 1 and 3 wt% of TiO2.

However, particle agglomeration has occurred at higher TiO2

concentrations (5 and 7 wt%). Based on Figs. 1 and 2, the mor-

phology of support layer for pure PSf membrane is similar to

MMMs with lower TiO2 loadings (1 and 3 wt%) which

FIG. 1. SEM photograph of pure PSf membrane. [Color figure can be

viewed in the online issue, which is available at wileyonlinelibrary.com.]

FIG. 2. SEM photographs of MMMs with different TiO2 contents: (a) 1 wt%, (b) 3 wt%, (c) 5 wt%, and (d) 7

wt%. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—2014 3

containing spongy and macrovoid free structure. However, the

thickness of surface layer improved gradually by addition of

TiO2 in the following order : 0 wt% (6.3 lm) < 1 wt% (8.4

lm) < 3 wt% (12.4 lm). While, as shown in Fig. 2, MMMs

with porous structures containing tear like and finger like pores

were obtained by addition of 5 and 7 wt% of TiO2 to the poly-

mer matrix. The variation in membrane morphology in different

TiO2 loadings can be justified by increasing the hydrophilicity

of wet film in higher TiO2 contents (�5 wt%) [26]. Due to the

higher affiliation of TiO2 nanopaticles with water than PSf, the

water diffusion velocity through the membrane structure

enhanced in higher TiO2 loadings (�5 wt%) during phase inver-

sion. Then, instantaneous phase inversion has occurred between

solvent and non-solvent in the coagulation bath which attributed

to the membranes with thin surface layer, large tear-like and

long finger-like pores close to the surface layer. As known, size

of pore and membrane porosity improved with non-solvent dif-

fusion speed [27]. In another word, the small quantity of TiO2

nanoparticles (�3 wt%) would reduce the precipitation velocity

and result in a denser and thicker surface layer on the sublayer.

The same observation was reported by Madaeni et al. [28] dur-

ing preparation and characterization of polyethersulfone (PES)/

TiO2 MMMs for gas separation. Conversely, further addition of

TiO2 nanoparticle (�5 wt%), leads to significant particle aggre-

gation, disruption of polymer chain packing and also extra free

spaces for polymer chain movements [18]. Accordingly, a

porous membrane structure containing tear like and finger like

pores was obtained.

Distribution of TiO2 nanoparticles in polymer matrix was

analyzed by EDX. The EDX spectra of pure PSf and PSf/TiO2

membranes with 3 and 7 wt% of TiO2 are shown in Fig. 4a–c,

respectively. As demonstrated in Fig. 4b, the bright spots corre-

sponding to the Ti elements were found homogeneously distrib-

uting on the surface of prepared MMMs containing 3 wt% of

TiO2. As shown in Fig. 4c, it was observed that higher loadings

of TiO2 nanoparticle attributed to a serious inorganic filler

agglomeration.

Variation in surface roughness of prepared membranes with

different TiO2 contents was shown in Fig. 5. Samples with a

size of 5 lm 3 5 lm surface areas were scanned at 0.8 Hz.

Roughness parameters which were determined through the AFM

analysis software (Nano scope Software Version) are recorded

in Table 3. According to the results obtained, surface roughness

increased by addition of nanoparticle to polymer matrix and

reached to the maximum for MMMs containing 5 wt% of TiO2.

FTIR spectra resulted for pure PSf and PSf/TiO2 3 wt% and

PSf/TiO2 7 wt% membranes presented in Fig. 6. As illustrated

in Fig. 6, two absorption peaks at the range 1290–1320 cm21

indicating the symmetric O@S@O stretching vibration of PSf

[29]. Symmetric and asymmetric deformation vibration of OH

can be observed at 1355 and 1485 cm21, respectively. Also,

high intensity peak at 1570 cm21 was assigned to a C@C conju-

gation system of the benzene ring. A clear peak around 3000

cm21 was noted for CH3 stretching vibration [30]. As presented

in Fig. 6b and c, a clear peak near 500 cm21 demonstrates the

stretching mode of TiAO and also proves the presence of TiO2

nanoparticle in prepared MMMs. As shown in spectra, the inten-

sity of TiAO peak increased in higher TiO2 loading (PSf/TiO2 7

wt%).

TGA and DTG curves of PSf and PSf/TiO2 membranes with

various TiO2 compositions are shown in Fig. 7. Thermograms

of each specimen indicate that there is no weight loss detected

below 400�C, which shows the whole removal of solvents from

the created membranes. A vital weight loss has occurred in the

temperature range of 450 to 550�C indicating polymer decom-

position. The amounts of weight loss for developed PSf/TiO2

MMMs reduced in the following order: pure PSf (66.94%) >PSf/TiO2 1 wt% (64.11%) > PSf/TiO2 3 wt% (54.12%) > PSf/

TiO2 5 wt% (53.83%) > PSf/TiO2 7 wt% (46.95%). So, a drop

in the amount of weight loss was detected by increasing TiO2

concentration in polymer solution. In another word, thermal sta-

bility of MMMs increased gradually with the TiO2 contents in

the matrix [18]. Improvement in thermal stability of developed

MMMs can be explained due to higher thermal stability of TiO2

compared to PSf [30].

Gas Separation Properties of PSf and PSf/TiO2 Membranes

The permeances of CO2 and CH4 at pressure range 1–7 bar

are shown in Fig. 8a and b, respectively. The error bars are cal-

culated from three separate permeability measurements for each

sample. As indicated in Fig. 8, CO2 and CH4 permeances

decreased and reached to a minimum by increasing TiO2 content

from 0 to 3 wt%. CO2 permeance was around 12.9 GPU in 1

bar feed pressure for pure PSf membrane. This number reduced

and reached to 6.57 and 0.26 GPU for PSf/TiO2 1 wt% and PSf/

TiO2 3 wt%, respectively, at same feed pressure. In the case of

CH4 permeance a similar trend was observed by increasing

TiO2 concentration from 0 to 3 wt%. CH4 permeance decreased

from 1.8 to 0.52 GPU at 1 bar pressure by addition of TiO2

FIG. 3. TEM photographs of MMMs with different TiO2 loadings: (a) 1 wt%, (b) 3 wt%, (c) 5 wt%, and (d) 7

wt%.

4 POLYMER ENGINEERING AND SCIENCE—2014 DOI 10.1002/pen

nanoparticle from 0 to 1 wt%. Interestingly, PSf/TiO2 3 wt%

did not allow to CH4 molecules to pass through the membrane

at 1 bar feed pressure. While, the amount of CO2 permeance at

1 bar pressure for mentioned membrane was 0.26 GPU. Accord-

ingly, MMM containing 3 wt% of TiO2 was able to separate

CO2 from CH4 completely at 1 bar feed pressure. As illustrated

in Fig. 8, CO2 and CH4 permeances increased significantly with

further TiO2 loadings (5 and 7 wt%). Table 4 presents CO2/CH4

selectivity of prepared membranes with different TiO2 loadings.

According to this table, gas selectivity was around 7 at 1 bar

pressure for pure PSf membrane. CO2/CH4 selectivity improved

and reached to 12.6 (1 bar) by addition of 1 wt% of TiO2 to the

casting solution. As mentioned earlier, PSf/TiO2 3 wt% mem-

brane did not allow to CH4 molecules to pass through mem-

brane and this membrane just had CO2 permeability at 1 bar

feed pressure. This phenomenon shows that, membrane contain-

ing 3 wt% of TiO2 was able to separate CO2 from CH4 com-

pletely. Also, the amount of gas selectivity for this membrane

was 36.5 at 3 bar pressure. However, CO2/CH4 selectivity

declined considerably by further incorporation of TiO2 nanopar-

ticle to the polymer solution (�5 wt%).

Based on the above gas permeation and selectivity perform-

ances, the interface morphology for PSf/TiO2 MMMs may be

divided into: Case 1 (ideal morphology) for �3 wt% loadings

and Case 2 (interface voids) for �05 wt% loadings. As dis-

cussed in Figs 1 and 2, MMMs containing �3 wt% of TiO2

have a similar sublayer morphology. However, thickness of sur-

face layer increased by incorporation of TiO2 to the casting

FIG. 4. EDX analyses of membranes top layer in different TiO2 loadings: (a) Pure PSf, (b) PSf/TiO2 3 wt%, and

(c) PSf/TiO2 7 wt%. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—2014 5

solution and reached to a maximum for PSf/TiO2 3 wt%. At 3

wt% of TiO2 incorporation, there existed a fair attachment of

polymer on the nonporous TiO2 particle surface. It may consid-

ered from the gas permeability variation in Fig. 8, interaction of

CH4 with TiO2 nanoparticles was stronger than CO2 with TiO2

when 1 and 3 wt% TiO2 was introduced in the MMMs. Similar

trends have been reported in the literatures using the polyether-

sulfone/TiO2 and polyimide/ TiO2 MMMs [6, 18]. By further

TiO2 loadings (�5 wt%), nanoparticle aggregation became more

critical (Fig. 3) and the macrovoids which were created by the

detachment of more hydrophobic part (polymer chains) from the

hydrophilic part (nanoparticle surface), attributed to higher gas

permeance and lower selectivity. Also it may consider that, the

membrane with a thicker skin layer provides higher resistance

against passing gas leading to perrmeance diminishment [25].

The gas permeation mechanism through the polymeric mem-

brane is a solution-diffusion and gases with a larger molecular

diameter diffuse slower through membrane structure [21–23]

and also due to the larger molecular diameter of CH4 in compar-

ison with CO2 [23, 31, 32], CO2 permeability is higher than

CH4. Hence, because of the thick top layer and a macrovoid

free support layer, a high selective CO2/CH4 separation has

occurred by PSf/TiO2 3 wt%.

The CO2/CH4 selectivity results of current research work are

compared with the some of available studies in Table 5. The

FIG. 5. Three-dimensional AFM images of MMMs surface layer with different TiO2 contents; (a) 0 wt%, (b) 1

wt%, (c) 3 wt%, and (d) 5 wt%. [Color figure can be viewed in the online issue, which is available at wileyonlineli-

brary.com.]

TABLE 3. Surface roughness parameters with different TiO2 contents.

MMMs type Rms rough (nm) Mean rough (nm)

Pure PSf 09.9 07.8

1 wt% 11.4 09.3

3 wt% 13.6 10.7

5 wt% 16.3 13.0

FIG. 6. FT-IR spectra of PSf membranes with different TiO2 loadings: (a)

Pure PSf, (b) PSf/TiO2 3 wt%, and (c) PSf/TiO2 7 wt%. [Color figure can

be viewed in the online issue, which is available at wileyonlinelibrary.com.]

6 POLYMER ENGINEERING AND SCIENCE—2014 DOI 10.1002/pen

selection is based on the gas separation similarity to this

research and also nanoparticle used for MMMs preparation. So,

different MMMs such as matrimid/TiO2, polybutadiene/ TiO2

and PDMS-PES/TiO2 are selected for comparison.

CONCLUSIONS

PSf/TiO2 MMMs at different nanoparticle weight fractions

were prepared and employed for gas separation. SEM images

revealed enhancement in thickness of surface layers by increas-

ing TiO2 content up to 3 wt%. MMMs containing macrovoids

were prepared by further loadings of inorganic nanoparticle (�5

wt%). EDX and TEM analysis proved that nanoparticles distrib-

uted homogeneously on the surface and substructure of MMMs

in lower loadings of TiO2 (�3 wt%). However, inorganic filler

agglomeration became more critical in higher filler contents (�5

wt%). AFM characterization shows that surface roughness of

membranes increased from 9.935 for pure PSf membrane to

16.3 for MMMs containing 5 wt% of TiO2. According to TGA

and DTG results, the amount of weight loss decreased in higher

inorganic filler contents and MMMs with higher thermal stabil-

ity were prepared. In the case of gas permeability, both CO2

FIG. 7. TGA and DTG curves of pure PSf and PSf/TiO2 membranes with

various TiO2 compositions. [Color figure can be viewed in the online issue,

which is available at wileyonlinelibrary.com.] FIG. 8. CO2 and CH4 permeances of PSf and PSf/TiO2 membranes with

different TiO2 contents. [Color figure can be viewed in the online issue,

which is available at wileyonlinelibrary.com.]

TABLE 4. CO2/CH4 selectivity of membranes with different TiO2

loadings.

CO2/CH4 selectivity (60.10)

TiO2 (wt%)/pressure 1 3 5 7

0 7.1 5.4 3.3 2.1

1 12.6 7.5 5.4 3.3

3 1 36.5 16.3 5.6

5 4.1 2.1 1.9 1.8

7 3.5 1.8 1.4 1.2

TABLE 5. CO2/CH4 selectivity comparison of the present research work with previous studies.

Sample CO2/CH4 selectivity/nanoparticle content Pressure

This work 7.10/(0 wt%) 12.60/(1 wt%) 1/(3 wt%) 4.10/(5 wt%) 1 bar

This work 5.40/(0 wt%) 7.50/(1 wt%) 36.50/(3 wt%) 2.10/(5 wt%) 3 bar

Matrimid/TiO2 [17] 20.50/(0 vol%) 16.90/(5 vol%) 14.80/(10 vol%) 13.80/(15 vol%) 2 bar

Polybutadiene/TiO2 [20] 6.80/(0 wt%) 7.50/(10.1 wt%) 7.60/(15.2 wt%) 6.70/(20.1 wt%) 2 bar

PDMS-PES/TiO2 [28] 2.00/(0 wt%) 2.50/(3 wt%) 2.7/(5 wt%) 3.2/(7 wt%) 5 bar

DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—2014 7

and CH4 permeances decreased in lower TiO2 contents (�3

wt%) and then increased in higher loadings of TiO2 (�5 wt%).

A different trend was observed for gas selectivity and an

improvement for CO2/CH4 selectivity was detected by incorpo-

ration of TiO2 until 3 wt%. Interestingly, PSf/TiO2 3 wt% was

able to separate CO2 from CH4 completely at 1 bar pressure and

selectivity value for this membrane was 36.5 at 3 bar feed pres-

sure. A significant reduction in gas selectivity value was

observed in higher nanoparticle contents (�5 wt%). According

to the resulted gas separation performances, two interface mor-

phology for PSf/TiO2 MMMs can be considered: Case 1 (ideal

morphology) for �3 wt% loadings and Case 2 (interface voids)

for �5 wt% loadings.

REFERENCES

1. D. Bastani, N. Esmaeili, and M. Asadollahi, J. Ind. Eng. Chem.,19, 375 (2013).

2. A. Arabi, A. Kargari, M. Bahrami, S. Laki, and H. Ajami,

J. Ind. Eng. Chem., 19, 677 (2013).

3. A. Ebadi, H. Sanaeepur, A. Kargari, and A. Moghadassi, Sep.Purif. Technol., 82, 102 (2012).

4. M. TakhtRavanchi, T. Kaghazchi, and A. Kargari, Desalination,

235, 199 (2009).

5. M.U.M. Junaidi, C.P. Leo, S.N.M. Kamal, A.L. Ahmad, and T.

L., Fuel. Process. Technol., 112, 1 (2013).

6. M.A. Aroon and A.F. Ismail, Sep. Purif. Technol., 75, 229

(2010).

7. A. Shariati, M. Omidkhah, and M.Z. Pedram, Chem. Eng. Res.Des., 90, 563 (2011).

8. R.D. Noble, J. Membr. Sci., 378, 393 (2011).

9. S. Basu, A. Cano-odena, and I.F.J. Vankelecom, Sep. Purif.Technol., 81, 31 (2011).

10. A.F. Bushell, M.P. Attfield, C.R. Mason, P.M. Budd, Y.

Yampolskii, and L. Starannikova, J. Membr. Sci., 427, 48

(2013).

11. T.S. Chung, L.Y. Jiang, Y. Li, and S. Kulprathipanja, Prog.Polym. Sci., 32, 483 (2007).

12. M. L�opez-gonz�alez, F. Luis, and E. Riande, J. Membr. Sci.,383, 206 (2011).

13. S. Rafiq, Z. Man, A. Maulud, N. Muhammad, and S. Maitra,

Sep. Purif. Technol., 90, 162 (2012).

14. Y. Shen and A.C. Lua, Chem. Eng. J., 188, 199 (2012).

15. M. Naghsh, M. Sadeghi, A. Moheb, and M. Pourafshari, J.Membr. Sci., 423–424, 97 (2012).

16. J. Ahn, W.J. Chung, I. Pinnau, and M.D. Guiver, J. Membr.Sci., 314, 123 (2008).

17. F. Moghadam, M.R. Omidkhah, M.Z. Pedram, and F. Dorosti,

Sep. Purif. Technol., 77, 128 (2011).

18. C. Liang, P. Uchytil, R. Petrychkovych, Y. Lai, K. Friess, and

M. Sipek, Sep. Purif. Technol., 92, 57 (2012).

19. S. Matteucci, V.A. Kusuma, D. Sanders, S. Swinnea, and B.D.

Freeman, J. Membr. Sci., 307, 196 (2008).

20. S. Matteucci, V.A. Kusuma, S. Swinnea, and B.D. Freeman,

Polymer 49, 757 (2008).

21. F. Dorosti, M.R. Omidkhah, M.Z. Pedram, and F. Moghadam,

Chem. Eng. J., 171, 1469 (2011).

22. M.A. Aroon, A.F. Ismail, M.M. Montazer-rahmati, and T.

Matsuura, J. Membr. Sci., 364, 309 (2010)

23. C.A. Scholes, S.E. Kentish, and G.W. Stevens, Recent. Pat.Chem. Eng., 1, 52 (2008).

24. A.F. Ismail and P.Y. Lai, Sep. Purif. Technol., 40, 191 (2004).

25. S.S. Madaeni and P. Moradi, J. Appl. Polym. Sci., 121, 2157

(2011).

26. J.F. Blanco, J. Sublet, Q.T. Nguyen, and P. Schaetzel,

J. Membr. Sci., 283, 27 (2006).

27. P. Moradihamedani, N.A. Ibrahim, W.M.Z.W. Yunus, and N.A.

Yusof, J. Appl. Polym. Sci., 130, 1139 (2013).

28. S.S. Madaeni, M.M.S. Badieh, V. Vatanpour, and N. Ghaemi,

Polym. Eng. Sci., 52, 2664 (2012).

29. S. Balta, A. Sotto, P. Luis, L. Benea, B.V.D. Bruggen, and

J. Kim, J. Membr. Sci., 389, 155 (2012).

30. C.P. Leo, W.P.C. Lee, A.L. Ahmad, and A.W. Mohammad,

Sep. Purif. Technol., 89, 51 (2012).

31. P. Moradihamedani, N.A. Ibrahim, W.M.Z.W. Yunus, and N.A.

Yusof, Polym. Eng. Sci., 1 (2013); doi:10.1002/pen.23706.

32. P. Moradihamedani, N.A. Ibrahim, D. Ramimoghadam,

W.M.Z.W. Yunus, and N.A. Yusof, J. Appl. Polym. Sci., 1

(2013); doi:10.1002/app.39745.

8 POLYMER ENGINEERING AND SCIENCE—2014 DOI 10.1002/pen