Journal of Natural Gas Science and Engineering 31 (2016) 428e 436 X

Contents lists available at ScienceDirectX

Journal of Natural Gas Science and Engineering

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

PVDF/CaCO3 composite hollow fiber membrane for CO2 absorption in

gaseliquid membrane contactor

a, b a, b a, b, * a, b a,
M. Fosi-Kofal , A. Mustafa , A.F. Ismail , M. Rezaei-DashtArzhandi , T. Matsuura
c
X
Advanced Membrane Technology Research Centre (AMTEC), Universiti Teknologi Malaysia, 81310 UTM Skudai, Johor Darul Ta'zim, Malaysia

Faculty of Petroleum & Renewable Energy Engineering, Universiti Teknologi Malaysia, 81310 UTM Skudai, Johor Darul Ta'zim, Malaysia

Department of Chemical and Biological Engineering, University of Ottawa, 161 Louis Pasteur St., Ottawa, ON K1N 6N5, Canada
embedding d. The incorporation of hydrophobic nano-particles
different into the polymer network enabled the formation of
amounts of more abundant and narrower finger-like pores in the
hydrophobi composite membranes compared to plain PVDF
article info c calcium membrane. Moreover, the addition of nano-particles
carbonate enhanced the surface roughness, permeation rate,
Article history: (CaCO3) porosity and wettability resistance of the composite
nano- membranes. CO2 absorption performance of the
particles in fabricated membranes was evaluated via a
Received 19 October 2015 Received in revised form 15 March 2016
the gaseliquid membrane contactor system. The CO2
polymer
flux was improved to some extent by increasing the
Accepted 19 March 2016 Available online 21 March 2016 matrix. The
effects of mixing ratio of CaCO3. Peak absorption
nano- 3 2 1
performance of 1.52 10 mol m s at 300 ml/ min
Keywords: particle
absorbent flow rate was achieved when 20/100
loadings on
the weight ratio of CaCO3/PVDF was employed. How-
Carbon dioxide absorption Composite hollow fiber membrane Hydrophobic
morpholog ever, further increase of the ratio resulted in a
CaCO3 nano-particles Wetting resistance y, structure composite membrane with lower absorption
and performance than the other composite membranes.
Porosity performanc Moreover, a long-term stability study of the
e of the composite membrane with the best CO2 absorption
spun
flux showed no decline in performance in the initial
membranes
210 h of operation, indicating that the membrane
in
abstract possesses high potential for gaseliquid contactor
gaseliquid
applications.
contactors
were
Porous hydrophobic polyvinylidene fluoride (PVDF) composite hollow investigate
2016 Elsevier B.V. All rights reserved.
fiber membranes were fabricated via phase inversion method by
pri Masoumi et al., 2014). natural gas leading to
ma Apart from environmental customers to develop
ry rea-sons, the removal of commercial
gre CO from natural gas has specifications about
2
enh the CO2
various economic and
ous practical advantages for concentrationX
e
gas the industry. First, CO2 as
1. Introduction an acidic gas corrodes
(R
pipelines, facilitates
ahi hydrate formation and
Natural gas is a fossil fuel lauded for its
minimal carbon exhaustion during m causes pres-sure drop
combustion. The primary element of et during transmission.
natural gas is methane, however it also al., Second, the presence of
contains some serious contaminants 20 CO2 also lowers the
including carbon dioxide (CO2), the 15; overall heating value of
 ane Technology Research Centre manageable levels of various advantageous
(AMTEC), Universiti Teknologi including higher
Malaysia, 81310 UTM Skudai, concentration (
Johor Darul Tazim, Malaysia. Boributh et al., 2011; interfacial area,
better flexibility and
Li and Chen,
superior mass
E-mail addresses: 2005).X transfer rate per unit
volume over
afauzi@utm.my,
fauzi.ismail@gmail.com (A.F. For the purpose of absorption columns (
Albarracin Zaidiza et
Ismail). X CO2 removal, hollow
al., 2016). However,
fiber membrane con-
membrane contactors
tactors received great
http://dx.doi.org/10.1016/j.jngse. come with their own
2016.03.053 1875-5100/ 2016 attention recently ( intrinsic challenges.
Elsevier B.V. All rights reserved.
Zhang et al., 2013). Wetting of the
In a gaseliquid polymeric
membrane contactor, membranes deployed
CO2 diffusion occurs in gaseliquid
in a non-dispersive contactor processes
mode via a porous leads to sharp
membrane which reduction in mass
*
Cor acts only as a barrier transfer. Since
resp in natural gas. To meet all membrane wetting is
ond those practical, ( Rahim et al., partially determined
ing commercial and envi- 2014). Thus, by membrane proper-
auth ronmental demands membrane contactors ties, the membrane
or.
Adv necessitated the have no operational needs to be highly
anc development for various problems ( hydrophobic and
technologies that reduce itsX
ed
Me
Rezakazemi et al.,
mbr CO2 in natural gas to 2012) and they enjoy
M. Fosi-Kofal et al. / Journal of Natural Gas Science and Engineering 31 (2016) 428e436
429

was employed as the solvent without

further purification. Nano-
structure should be tailored so that its pore size is small
enough to prevent wetting but large enough to permit good precipitated CaCO3 hydro-
phobically enhanced by octadecyl
diffusion rate ( Mosadegh-Sedghi et al., 2014).X
Despite the numerous works done on dihydrogen phosphate was donated
PVDF/CaCO3 hybrid membranes, by NanoMaterials Technology Co.,
One attractive method of correcting the defects of Ltd (Singapore). The average
no study has investigated the
polymeric membranes is the exploitation of the benefits of particle size of hydrophobic CaCO3
application of these membranes for
mixed matrix membranes (MMMs). MMMs are widely is 40 nm with specific surface area of
acknowledged for their ability to combine the advantages CO2 absorption in membrane 2 -1
of both organic (i.e. ease of processability and availability) contactors. The principal objective 40 m g . Lithium chloride (LiCl,
and inorganic fillers (i.e. high permeability and mechanical of this study is, therefore, to SigmaeAldrich, >99%) was
employed to act as a pore-forming
stability) ( Husain and Koros, 2007; Chung et al., 2007; fabricate porous PVDF/CaCO3
composite hollow fiber membranes agent. Nitrogen (N2, Mega Mount
Ismail et al., 2008). Rezaei et al (2014a). inves-tigated the via wet phase inversion method by Industrial Gases, 99.9%) was used to
performance of PVDF/montmorillonite (MMT) hollow test gas per-meance of fabricated
incorporating hydrophobic CaCO3
fiber MMMs in absorption of CO2 in membrane contactor. nano-particles into the polymer membranes. During the spinning
Their fabricated MMMs exhibited higher hydrophobicity, matrix. Since the contact angle of process, distilled water was used as
the bore fluid and tap water as the
porosity and CO2 absorption performance compared to the CaCO particles is about 120, it
3 external coagulant. Methanol (GR
neat PVDF ( DashtArzhandi et al., 2015). Zhang and is expected that the resultant grade, 99.9%) from Merck was used
Wang, (2014) used triple-orifice spinneret to incorporate membranes will show superior for the post-treatment of the spun
wetting resistance and therefore membranes. In the mem-brane
fluorinated silica (fSiO2) nano-particles in polyetherimide
better performance in gaseliquid contactor test, pure CO2 (Mega
(PEI). They detected better wetting resistance, chemical
contacting processes than neat Mount Industrial Gases, 99.9%) and
compatibility and mechanical strength in modified PEI
PVDF. distilled water were used as the gas
than the neat PEI membrane. Similarly, blending calcium
absorbed and the liquid absorbent
carbonate (CaCO3) nano-particles into the PVDF matrix
respectively.
has been reported to enhance membrane structure, narrow
pore size distribution and increase membrane porosity (
2.2. Dope preparation
Hou et al., 2014). Zhi et al (2014). dramatically 2. Experimental
increased the hydrophilicity of PVDF by using unmodified
Before dope preparation, the PVDF
CaCO . In another work, Hou et al (2012). fabri-cated 2.1. Materials
3 pellets and CaCO3 nano-particles
PVDF composite membranes by incorporating modified were dried overnight in a vacuum
cal-cium carbonate into the spinning dope. The resultant Commercially available
membranes showed desirable characteristics such as polyvinylidene fluoride (PVDF) oven at 65 C to remove moisture
narrow pore size distri-bution, improved porosity, surface pellets (Kynar 740, Mw 156,000 content. A pre-determined amount of
PVDF and NMP was then subjected
roughness and contact angle. The addition of CaCO3 g/mol), used as the base membrane
particles reduced the pore diameter, resulting in an material, were purchased from to vigorous stirring at 70 C. After the
full dissolution of PVDF, LiCl (3 wt
improvement of membrane wetting resistance.X Arkema inc., Philadelphia, USA. N-
Methyl-2-pyrrolidone (NMP, 99.5%) %) were slowly introduced to the
mixture under continuous agitation. Meanwhile, in three The details of the spinning sputtering platinum before testing.
separate tanks, an emulsion of hydrophobic CaCO3 in conditions are given in Table 2.X The SEM images of the cross-
section, outer and inner surfaces of
NMP was prepared in various amounts. To avoid
the fabricated fibers were taken at
agglomeration of particles, the emul-sions were sonicated
different magnifications. The
at 45 C for 2 h. Afterwards, the CaCO3 emulsions were existence of calcium element as an
In order to achieve complete
added to the solution of PVDF/LiCl and subjected to evidence of the CaCO3 nano-
removal of the remaining solvent
further vigorous agitation. Later, the prepared dope
and non-solvent, the fabricated particles addition to the membranes
solutions were kept in an ultrasonic bath for 7 h at 25 C to hollow fibers were soaked in water was also conducted with SEM using
remove any unwanted or excess gas prior to spinning. The for 3 days. To prevent pore energy-dispersive x-ray
dope formulation is illustrated in Table 1. The primary collapsing while water is being spectrometry (EDX).
removed from the pores, the water
difference from the formulation used by Hou et al ( Hou et was exchanged with methanol by
al., 2012). is that PVDF concentration was 12% and immersing the hollow fibers in
DMAc was used as solvent in their work.X methanol for over 30 min. Then,
they were hung to dry at room 2.5. Atomic force microscopy
temperature. (AFM)

2.3. Hollow fiber fabrication

The hydrophobicity of membranes is
2.4. Scanning electron microscopy closely related to its sur-face
Non-solvent induced phase separation was the preferred (SEM) and energy-dispersive X-ray roughness ( Naim et al., 2013).
method of membrane fabrication in order to produce (EDX) Atomic Force Microscopy (AFM)
membranes of asymmetric structure. A spinneret was fitted
(SII Nanotechnology SPM-SPI3800
into a reservoir containing the prepared solutions and
The morphology of fabricated N) was used to scan the outer
pressurized nitrogen was used to deliver the solutions into
hollow fiber composite mem-branes surface morphology of the
the spinneret. The dope was extruded from the spinneret at
a fixed flow rate and entered the coagulation bath with no
was investigated by a Scanning membranes in an area of 5 mm 5
Electron Microscope (SEM) (TM mm. The AFM micrographs,
air-gap. As the external coagulant, tap water was used to
3000, Hitachi). The hollow fibers obtained by tapping mode, were
induce the phase inversion of nascent fibers. The take-up
were immersed in liquid nitrogen analyzed using an AFM software to
speed of the hollow fiber was adjusted according to the
and broken to obtain a clean smooth measure the mean roughness (Ra) of
dope extrusion rate from the spinneret. For the internal
cross-section. Then, the fractured
coagulation process, distilled water was selected and fed to
fibers were coated with a thin film of the membranes.X
the spinneret at a constant flow rate by a syringe pump.
Critical entry pressure of water (CEPw) is the lowest pressure necessary for water penetration into membrane pores and is a measure of the wetting resistance of
the membranes. CEPw ex-periments were conducted using hollow fiber modules similar to those employed in the permeation test. Pressurized nitrogen was used
to force-feed distilled water into the lumen side of the hollow fiber membranes. After 30 min of keeping the pressure constant, an increment of 0.5 bar intervals
was added to the original pressure. The pressure at which the first water droplet appeared on the membrane shell side was recorded as the CEPw.
To measure contact angle, the fibers were kept in an oven at 60 C for 12 h. The sessile drop method using a goniometer (model G1, Krss GmbH) was employed
to determine the contact angles at

2.7. Critical entry pressure of water (CEPw) and contact angle

2 1 1 1 1
where PP and PK represent gas permeance by Poiseuille and Knudsen flow respectively (mol m Pa s ), R is gas constant (8.314 J mol K ), T is absolute
1
temperature (K), M is molecular weight of gas (kg mol ), m is the viscosity of gas (Pa s), LP is effective pore length (m). Therefore, the mean pore size and the
effective surface porosity can be calculated by using the intercept
(A) and the slope (B) of P versus P plot as follows ( Lee et al., 2002; Yasuda and Tsai, 1974; Bakeri et al., 2012a):X

2.6. Porosity and pore size

The hollow fibers were subjected to nitrogen (N2) gas perme-ation test in order to obtain the membrane mean pore size and effective surface porosity. One hollow
fiber about 10 cm in length was potted to a steel fitting at one side and the other side was closed by epoxy resin. For each type of membrane fabricated, at least
three testing modules were prepared to obtain accurate re-sults. The feed gas (N2) was delivered to the shell side and the pressure was increased from 0.5 to 5 bar
(gauge) with 0.5 incre-ment each time. Soap-bubble flow meter was then employed to measure the gas permeance at each pressure.
The background theory of the gas permeation test is based on few assumptions. First, the shape of the membrane pores is simplified as straight and cylindrical.
Second, Knudsen and Pois-euille flow regimes are assumed to control the gas flow via the pores to obtain the overall gas permeance, P
2 1 1
(mol m Pa s ) as follows:
430 M. Fosi-Kofal et al. / Journal of Natural Gas Science and Engineering 31 (2016) 428e436

Table 1

Dope formulation for hollow fiber membrane spinning.

Membrane code
PVDF (wt%)
CaCO3 (wt%)
CaCO3/PVDF
LiCl (wt%)
NMP (wt%)

M0
18.0
0.0
0/100
3.0
79.0
M10
18.0
1.8
10/100
3.0
77.2
M20
18.0
3.6
20/100
3.0
75.4
M30
18.0
5.4
30/100
3.0
73.6
 pM
2
8RT
RT
0:5
rP m z
LP
r 2
1 P m z
8m

RT

LP
L
P

2
rP ;m
P A Bp

r
P;m
16 B

8RT

0:5

ten different points on the outer surface

of the hollow fibers and then the
arithmetic mean value was calculated.

2.8. Overall porosity and tortuosity

The membrane overall porosity, m,

defined as the volume of the pores
divided by the total volume of the
membrane, was quantified by
P PP PK gravimetric method formula:
m

w1
w2=rw
3 (5)
;

pM

x
. .
8mRTB w1 w2 rw w2 rp
p

m
3

where w1 is the weight of the wet membrane, w2 the weight of the dry
membrane, rw water density and rp is the polymer density. Three spun hollow
fibers were randomly chosen after the solvent exchange. Prior to measuring the 2.9. Measurement of
wet weight, an air stream was used to blow the lumen side of the membrane. The
CO2 absorption
dry weight was later measured when the membranes were adequately dried in an
oven for 2 h at 120 C.
By using m, we can calculate the tortuosity, an important The CO2 absorption
parameter for the diffusion of gases in membrane contactors. experiment was carried
(2) out to evaluate the
performance of the spun
hollow fibers in
t membrane contactor Hollow fiber spinning process specifications.
application. Five hollow
2 fibers of 175 mm in
2 m length were packed in
(6)
a contactor module with
both ends open. The
desired configuration
used during the
absorption experiment
was to flow pure CO2 in
the shell side and
distilled water in the
lumen side in a counter-
current mode. While the
gas side pressure was
m maintained at 1 bar, the External coagulant temperature (C)
liquid side pressure was
set at 1.2 bar to avoid the
formation of bubbles in the liquid stream (
Malek et al., 1997). Before taking any X
samples, the experiment was left to run
in a steady state condition for 30 min.
Titration was used to quantify the
concentration of CO2 absorbed in the
distilled water at the lumen side exit
using. The characteristics of membrane
contactor module employed in the
experiments are presented in Table 3.
X
Table 2
Dope extrusion rate (mL/min)
4.0
Bore flow rate (mL/min)
1.5
Bore composition (wt.%)
Distilled water
External coagulant
Tap water
Air gap distance (cm)
0.0
Spinneret o.d./i.d. (mm)
1.2/0.55
Spinning dope temperature (C)
25
25
M. Fosi-Kofal et al. / Journal of Natural Gas Science and Engineering 31 (2016) 428e436
431

According to Hou et al (2012), is predicted to have high

permeability and wetting
3. Results and discussion the addition of CaCO3 nano- resistance which is favorable
particles into the polymer for gaseliquid contacting
3.1. Membrane morphological study dopes acts both as nucleating processes. However, when the
and anti-nucleating agents mixing ratio for CaCO3/PVDF
depending on the amount of
Porous asymmetric hollow fiber membranes were nano-particles employed. passed a threshold value, the
prepared by wet phase inversion method for CO2 During the phase inversion anti-nucleation property of
absorption via membrane contactor system. With process, the demixing be- CaCO3 prevailed and
the aim of enhancing the structure and haviors of a homogeneous consequently the rate of phase
performance of the membranes, different amounts polymer depends on the nature inversion was inhibited. This is
of hydrophobic CaCO3 nano-particles were of a certain layer of polymer reflected in the morphology of
nuclei. The formation of finger- M30 where there is evidence of
dispersed in spinning solutions. It is well-
like mac-rovoids occurs when less number of finger-like
recognized that the addition of inorganic particles
nuclei expand due to the higher pores. The phenomenon is
into poly-mer solutions affects
rate of solvent diffusing in than further consoli-dated by the
kinetic/thermodynamic relations that govern phase
coagulant diffusing out. inner surface morphology
inversion process. Subsequently, any shift in
Similarly, sponge-like whereby all membranes have
kinetic or thermodynamic stability induces
structures are formed when the visible pores except M30. The
membrane morphological and structural changes. propagation of nuclei is
higher amount of CaCO3 used
Rezaei et al (2014b). investigated the effect of
arrestedX in M30 enhanced its
nano-clay filler on the structure and performance hydrophobicity suppressing the
of PVDF mem-branes in gaseliquid contacting solvent/non-solvent exchange.
processes. They reported that the dispersion of As a result, the polymer
nano-clay filler altered the kinetic factor of PVDF concentration on the inner
solutions. Also, the surface modified nano-clay surface is eliminated leading to
particles weakened the interaction between Table 3
pore-less thin skin layer.
polymer and filler leading to disruption of the Variations could be observed
thermodynamic stability of the solution. Likewise, Specification of gaseliquid membrane between the wall thickness of
the disper-sion of CaCO3 nano-particles in the contactor module.
com-posite membranes and
polymer solutions is expected to accelerate the pure PVDF membrane. The
precipitation rate of the polymer-rich phase of the Module I.D. (mm) higher wall thickness of
solution and subsequently influence the 14 composite membranes could be
morphology and proper-ties of the fabricated attributed to the dispersion of
membranes.X CaCO3 nano-particles in the
Module length (mm) polymer matrix contributing to
270 higher particle accumulation.
Fiber O.D. (mm)
850e970 Also, the presence of CaCO3
The morphology of prepared membranes were Fiber I.D. (mm) nano-particles restricts the
investigated by SEM images. Fig. 1 presents the 630e690 degree of nascent fibers defor-
Effective fiber length (mm) mation during phase inversion
SEM micrographs of the cross-section and inner 175
surface of spun fibers. There is a striking simi- 2 process inducing the formation
Gas liquid contact area (mm )
larity between the SEM micrographs of composite 173e190 of thicker walls. It should be
mentioned that membranes
fibers showing the success of CaCO3 incorporation Number of fibers
5 with thicker walls are preferred
into the polymer matrix. From the cross-sectional in industrial applications for
SEM images, all membranes possessed an ultra their ability to withstand
thin layer at both ends and a sponge-like layer hydraulic pressure better than
sandwiched be-tween two finger-like macrovoid
layers originating from inner and outer surfaces. thin-walled membranes ( Feng
As can be seen, the plain PVDF, M0, showed the ( Smolders et al., 1992). In this et al., 2004).X
widest and the least number of finger-like pores. case, when proper quantity of
To some extent, the finger-like macrovoids became
more numerous and narrower with the progressive CaCO3 nano-particles were
incorporated homogenously in
addition of CaCO3 as observed in composite fibers
the polymer casting solution,
M10 and M20. However, M30 exhibited less the exchange rate of solvent
finger-like pores with larger diameter than other from the polymer solution and
composite membranes. The pres-ence of calcium non-solvent from the
element on the membrane outer surface was coagulation bath was acceler-
examined by EDX. The results show that the ated. Rapid solvent/non-solvent
amount of calcium element in the membrane outer exchange together with the 3.2. Membrane surface analysis
surface increased proportionately with the increase reduced thermodynamic
of the addition of CaCO3 nano-particles (see stability of dope resulted in the The 3D AFM micrographs of
Table 4). Also, the negligible amount of chlorine formation finger-like structures the outer surface of the pure
element is an ev-idence to the successful washing and porous thin skin layer. PVDF membrane and
Hence, a large number of composite membranes are
away of LiCl from the membrane structure.X
finger-like macrovoids with
displayed in Fig. 2. In the
reduced pore diameter were
suspended from both images, the peaks represent
membrane surfaces of M10 and nodule aggregates and the
M20. This membrane structure troughs the membrane pores.
Randomly chosen points of the
micrographs were analyzed by the AFM software ( Hou et al., 2014). Therefore, different amounts of
to quantify the mean roughness (Ra). From Fig. 4,since roughness contributes to hydrophobic CaCO3 nano-
it can be deduced that the surface roughness the contact angle according to particles blended in their
increases with the addition of CaCO3 nano- matrix were prepared and
the modified Young's equation (
characterized in terms of
particles. In terms of roughness, pure PVDF
Rafat et al., 2006), the contact angle, critical entry
membrane, M0, shows the smoothest surface (Ra
addition of CaCO3 nano- pressure of water (CEPw), gas
15.08). The order of the roughness of the
permeation and overall
composite mem-branes matches the order of the particles is a suitable way of
porosity. The results of
content of the CaCO3 nano-particles added in the increasing the membrane characterization tests are
polymer dope. M30 has the highest mean hydrophobicity.X
presented in Table 5.
roughness (Ra 22.10), followed by M20 (Ra
Generally, the con-tact angles
21.08) and then M10 (Ra 19.44). In fact, the
of the fabricated membranes
surface roughness of spun membranes relies on the
increased proportionally with
amount of CaCO3 nano-particles employed. This
the addition of CaCO3 nano-
can be explained by the nucleating property of 3.3. Membrane structural study
particles in the spinning
CaCO3 nano-particles to induce faster
dopesX
crystallization rate during phase separation process A series of membranes with
432 M. Fosi-Kofal et al. / Journal of Natural Gas Science and Engineering 31 (2016) 428e436

Fig. 1. SEM Images showing (1) cross-section and (2) inner membrane morphology.
EDX results of fabricated membranes. M20
M30

Table 4
M0
M10
F
49.45
49.50
49.51
49.47
C
50.51
49.83
49.32
49.22
Ca
0.00
0.63
1.14
1.28
Cl
0.04
0.04
0.03
0.03

The N2 permeance
2
(mol/m .s.Pa) versus mean
pressure (kPa) for

the spun hollow fiber

with M30 showing a maximum of about 101.53
membranes was plotted
according to Eq. (2) and
shown in Fig. 3. Best-fit
straight lines were drawn to
correlate the data. The slope
and intercept of the straight
lines were employed to
quantify the mean pore size
and effective surface porosity
by Eqs (3) and (4) and the
results are also shown in Table
5. From the gas permeation
test, neat PVDF, M0, exhibited
lower N2 permeance compared
to the composite membranes
which can be correlated to the
membrane morphology. For
composite mem-branes, the gas
permeance and the effective
surface porosities initially
increased with the increment of
CaCO3 nano-particles. As
previously discussed, the
number of finger-like pores
suspended from inner and outer
surfaces of composite
membranes are significantly
higher than that of plain PVDF
causing composite membranes
to have better permeability and
surface porosity. However, it
should be mentioned that M30
exhibited the lowest permeance
among composite membranes
which might be explained by
the anti-nucleation property of
CaCO3 at certain quantities. In
contrast, the mean pore size
kept decreasing with the
increase of the increment of
CaCO3 into the polymer dope.
This could be ascribed to the
nucleating role of CaCO3
nano-particlesX
M. Fosi-Kofal et al. / Journal of Natural Gas Science and Engineering 31 (2016) 428e436
433

Fig. 2. 3D AFM images of fabricated membranes.

Membrane structural characteristics. 166
208
189
CEPw (bar)
enabling rapid phase inversion and reduced 5.20
thermodynamic sta-bility to form finger-like M0 7.25
M10 8.15
macrovoids of restricted size. M20 8.60
M30 Overall Porosity (%)
According to the Laplace equation, which 77.80
79.55
measures membrane wetting resistance, the CEPw 80.91
depends on the contact angle and the structure of 79.64
3
the membrane ( Bougie and Iliuta, 2013). The Mass transfer resistance (10 ms-1)
7.30
CEPw increased with the addition of CaCO3. 3.76
Higher contact angles together with smaller pore 2.55
5.48
sizes of composite membranes contributed to the Water contact angle () Tortuosity
increase of the minimum pressure required for 88.31 1.92
water to penetrate into the membrane pores. 99.49 1.82
Hence, the composite membranes have higher 100.66 1.75
101.53 1.82
CEPw than the plain PVDF membrane.X Mean pore size (mm)
0.036
0.029
0.027
0.026
Table 5 1
Effective surface porosity (m )
73
 0.0000005 M30
morphologies.X

Since diffusion is
M0 the sole mass
0.0000007
transfer
mechanism of
0.0000002
solute gas in
contactor systems,
membrane
tortuosity is one
parameter that
assumes extra
importance.
Tortuosity
determines the
length of diffusion
pathways; the less
tortuous the
.s.pa) membrane is, the
0.0000006
better its CO2
absorption
performance (
120
140 Bakeri et al.,
160 2011). It can be
180
seen from Table 4
200
220 that the composite
240 membranes have
260 lower tortuosity
Permeance 280 than pure PVDF
membrane which
can be related to
the ability of
CaCO3 nano-
M10 Mean Pressure particles to induce
(kpa) the formation of a
larger number of
pores. It is also
Fig. 3. Measured N2
permeance versus
noteworthy that
mean pressure for the effects of
prepared membranes. tortuosity in
2 membrane
0.0000003
contactors are
Finally, the
more pronounced
porosities of the
in chemical
prepared
0.0000004 absorption process
membranes,
calculated ac- ( Zhang et al.,
cording to Eq. 2015).X
2
M20 (5), are in the
range of
77.80e80.91%.
The addition of 3.4. CO2
CaCO3 nano- absorption
particles had performance
positive impact on
N the overall
To assess the
porosity of
performance of
composite
the fabricated
membranes and
membranes in
can be explained
gaseliquid
by the abundancy
contactors,
of finger-like
modules
pores in
containing 5
(mol/m membrane
hollow fibers each
434 M. Fosi-Kofal et al. / Journal of Natural Gas Science and Engineering 31 (2016) 428e436

1.8x10-3 (mol/m

4.0x10-4

1.6x10-3
M0
1.0x10-3
2.0x10-4

1.4x10-3
M10 Flux
-4

0
50
100
150
200
250
300

.s)
2
8.0x10
M20
Absorbent Flow Rate
(ml/min)

Fig. 4. CO2 absorption performance of

the fabricated hollow fiber membranes
(pure CO2 in shell side, distilled water
in lumen side, Qg 100 ml/min, Pg
1 bar, Pl 1.2 bar, T 25 C).

CO
6.0x10-4 were prepared and distilled
M30 water was used as the liquid
absorbent. In physical CO2
absorption, the mass transfer
resistance of the gas side is
neglected since pure CO2 was
used as feed gas. Therefore, the
rate of mass transfer depends
primarily on the liquid phase
2 boundary and the membrane (
1.2x10-3 Atchariyawut et al., 2007). The
CO2 absorption flux was
obtained at different absorbent
flow rates in the shell side and
the results are shown in Fig.
4.X
From the figure, it can be observed that the CO2 membranes such as PVDF, its
absorption performance in contactors increased Since the absorption CO2 absorption perfor-mance
with the nano-particle loading. M20 shows the performance of the M20 of the membrane was not as
3
maximum absorption flux of CO2 flux of 1.52 10 membrane proved to be the high as M20.X
2 1 highest among the prepared
mol m s achieved at 300 ml/min absorbent flow membranes in this study, its
rate. This can be attributed to the superior surface performance in gaseliquid
hydrophobicity and porosity possessed by the contactors was compared to
composite membranes compared to plain PVDF other membranes covered in the
membrane. The higher surface hydrophobicity
together with the numerous and narrower finger- literature as shown in Table 6.
like pores with reduced mean pore size exhibited At liquid flow rate of 300 3.5. The effect of CaCO3 on
by composite membranes enable good pore ml/min, M20 (20/100 long-term stability of
wetting resistance, therefore, better performance in CaCO3/PVDF) showed the membranes
con-tactors. However, the flux dropped from M20
highest CO2 absorption flux of
to M30, with the latter membrane showing only 3 2 1 Long-term stability of
slightly higher flux than that of M0. As stated 1.52 10 mol m s . Polymeric
membrane applied in contactor
earlier, the 30/100 ratio of CaCO3/PVDF membranes fabricated with
applica-tions is a crucial
different pore-forming additives
employed in M30 is the threshold value of requirement for commercial
nanoparticle loading beyond which CaCO3 such as PVDF with LiCl ( viability. Basically, the
displayed its anti-nucleating property, leading to Mansourizadeh and Ismail, membrane performance should
2011) and PSf with glycerol ( not deteriorate for an adequate
reduction in permeability ( Fig. 4) and hence period of time before being
Mansourizadeh and Ismail,
inferior CO2 absorption among composite replaced. Therefore, the
2010) showed lower CO2 ab- membrane fabricated in this
membranes.X
sorption performance than M20. work with the highest CO2
Bakeri et al (2012b). modified absorption flux among the
As previously mentioned, the presence of the surface of PEI by the PVDF/CaCO3 composite
membrane itself in contacting processes exerts dispersion of SMM into the
additional resistance to the overall mass transfer polymer dope. Due to the large membranes (M20) was
process and has adverse effects on the long-term pore size and high porosity of subjected to long-term
performance of membranes. Hence, the Wilson contactor test and compared
the resultant hollow fiber
with that of pure PVDF
plot, 1/KOL versus membranes, their performance in
membrane. All the operating
membrane contactors was high.
parameters in the log-term
However, this type of membrane
Table 6 contactor test were maintained
has poor wetting resistance
Comparison of various membranes used in membrane as same as the short-term test
owing to low CEPw and contact
contactors for CO2 absorption. except feed gas and liquid
angle causing inevitable
pressures which were raised to
deterioration of absorption
1.5 and 1.7 bar, respec-tively.
V , was attempted to evaluate the mass transfer performance in the long-term. A
0.93
similar observation of low The CO2 absorption fluxes of
resistances of the fabricated membranes and the
-1 CEPw was reported by Rahbari- M0 and M20 over a prolonged
results at liquid velocity of 0.5 ms were chosen Sisakht et al. for the surface
period of time are presented in
for comparison and displayed in Table 5. Initially, modified PVDF membrane with
Fig. 5. M20 sustained a CO2
the mass transfer resistance of the membranes SMM. Zhang et al ( Zhang and
ab-sorption flux of nearly 3.75
decreased with the addition of CaCO3 and then Wang, 2013) reported the 3 2 1
fabrication of organic-inorganic 10 mol m s in the initial 210
increased when 30/100 CaCO3/PVDF mixing ratio com-posite membranes by h of operation after which the
was used. As can be seen from Table 5, M20 incorporating a layer of CO2 flux declined by 20%
3 fluorinated silica on a PEI within the next 40 h.
exhibited the lowest mass transfer of 2.55 10 ms substrate. Even tough, the
-1 Comparatively, the pure PVDF
. Also, M20 had the highest gas permeation rate composite membranes showed membrane could only maintain
among the spun membranes making it the most better hydrophobicity and
mechanical strength than stable CO2 flux in the initial 35
suitable for contacting processes.X
conventional polymeric h of operation and sufferedX
6.5 3 In-house made
4 4
Membrane type
6.00 10
In-house made
8.00 10
In-house made
( Zhang et al., 2015) X
Pore size (nm) PSf glycerol
Porosity (%) ( Rahbari-Sisakht et al., ( Mansourizadeh and 10.5
LEPw (bar) 72
2 1
CO2 flux (mol m s ) 2012) X Ismail 2011)X 4
4
Manufacturer PVDF PEI SMM 7.5 10
Ref. NA 280 In-house made
NA 81
M20 (PVDF CaCO3)
27
NA 1.5 ( Bougie and Iliuta, 2013) X
4 3 PP
81 8.00 10 3.20 10
Tianjin Motian In-house made 40
8.15
40
1.52 10
3

In-house made
( Atchariyawut et al., X
( Bakeri et al., 2011) NA
PEI Fluorinated silica 1.40 104
This work 2007) X 40 Celgard Inc.
PVDF LiCl PVDF SMM 81
233
NA
385 NA ( Wang et al., 2005) X
85 8.68 10
4
M. Fosi-Kofal et al. / Journal of Natural Gas Science and Engineering 31 (2016) 428e436
435

4.5x10-3
M0
-3
2.5x10

4.0x10-3

Flux M20
-3

5.0x10-4

.s)
3.5x10-3
2
2.0x10

0
50
100
150
200
250
300

CO
1.5x10-3

Time (hours)

Fig. 5. Long-term CO2 Absorption

performance of the fabricated hollow
fiber mem-branes (pure CO2 in shell
side, distilled water in lumen side, Ql
200 ml/min, Qg 100 ml/min, Pg
1.5 bar, Pl 1.7 bar, T 25 C).
2
-3
3.0x10
a performance reduction of
around 40% at 75 h. The
performance reduction of
polymeric membranes in
contactor applications for CO2
absorption is most likely
related to the wetting of
membrane pores as reported by
many researchers ( Korminouri
et al., 2015; Mavroudi et al.,
(mol/m
2006). Mansourizadeh and
Ismail (2011) devel-oped an
1.0x10-3 asymmetric PVDF membrane
and observed that the
membrane experienced CO2
absorption flux downturn of
roughly 25% after the initial 26
h of operation. Likewise, surface modified PSf was enhanced by the addition
membrane was reported to suffer about 18% of nano-particles to some
reduction in CO2 flux in the initial 50 h ( Rahbari- extent. Peak absorption
performance was ach-ieved
Sisakht et al., 2012).X when the mixing ratio of 2
total gas permeance (mol m )
20/100 for CaCO3/PVDF was
As known, the presence of water either in liquid or used for the dope preparation.
vapor form in membrane pores causes wetting However, further increase of P
hindering mass transfer. On one hand, membranes CaCO3 loading deterioriated
with large pore sizes are vulnerable to both direct the performance of the
penetration and capillary condensation of water composite membranes due to
into mem-brane pores, as in the case of the pure the anti-nucleation property of
PVDF membrane. On the other hand, membranes the particles beyond threshold
with small pore sizes and high hydro-phobicity, value. In addition, the long- PK
like M20, are less susceptible to water penetration; term stability test of the gas permeance by Knudsen
consequently enabling M20 to be more wetting flow regime
resistant according to Laplace equation. Therefore, membrane with the best CO2
the superior long-term performance of M20 both absorption results (M20)
showed that the per-formance
surface modified PVDF ( Rahbari-Sisakht et al., of the membrane remained
2012) and PSf ( Rahbari-Sisakht et al., 2012) is constant in the initial 210 h of
attributed to its higher wetting resistance. operation owing to its high
2 1 1
However, further research should be conducted on wetting resistance. (mol m Pa s )
methods to reduce the partial pore wetting of
membranes with small pore radius.X
PP
gas permeance by Poiseuille
flow regime
In both the short- and long-term contactor Acknowledgment
experiments, distilled water was used as the
absorbent liquid. Although the CO2 sorption
The authors gratefully
capacity of water is lower than chemical acknowledge the financial
absorbents, its usage in membrane contactors for support from Ministry of
2 1 1
physical absorption of CO2 is a common Education Malaysia (MoE) (mol m Pa s )
through the Malaysian In-
occurrence ( Atchariyawut et al., 2006; Ren et al.,
ternational Scholarship (MIS)
2006; Zhang et al., 2011). Also, since the scheme. We are also grateful rp
Nano-Materials Technology pore radius (m)
membranes fabricated in this work were
impregnated with hydrophobic CaCO3 nano- Co., Ltd (Singapore) for
providing the nano-precipitated rp,m
particles, there is a high possibility for the
calcium carbonate (NPCC). mean pore radius (m)
composite membranes to cope well in harsh
chemical operating conditions. Hence, the stability
of PVDF/CaCO3 composite membranes during R
chemical absorption will be the focus of research Nomenclature universal gas constant (8.134 J
1 1
in our forthcoming communications.X mol K )

4. Conclusion Variables T
temperature (K)

Porous PVDF hollow fiber composite membranes intercept

w
were fabri-cated via wet spinning method by 1
dispersing various amounts of hydrophobic wet weight (g)
slope
CaCO3 in the polymer matrix with LiCl as a pore-
Lp w
2
forming agent. Membranes were characterized to effective pore length (m)
investigate the effect of the incorporated nano- dry weight (g)
particles on membrane morphology, porosity,
mean pore size, hydrophobicity and CEPw. The M Greek Letters
1
addition of hydrophobic nano-particles facilitated molecular weight (g mol )
the fabrica-tion of membranes with more abundant
and narrower finger-like pores in composite m
membranes compared to neat PVDF mem-brane. overall porosity
From AFM micrographs, the membrane surface
roughness also increased with nano-particles
z
loading. Characterization test results revealed that
surface porosity
the contact angle and CEPw increased pro- mean pressure (Pa)
gressively with the addition of nanoparticles.
Conversely, an in-crease to the amount of nano- m
particle used resulted in smaller mean pore sizes of p gas viscosity (Pa s)
the composite membranes. The CO2 absorption
performance of the fabricated membranes was also
evaluated using a gaseliquid membrane contactor rw
system. The CO2 flux in the membrane absorption density of water
 Albarracin Zaidiza, D., Wilson, S.G.,
r Belaissaoui, B., Rode, S., Castel, C.,
p Roizard, D., Favre, E., 2016.
References
density of polymer Rigorous modelling of adiabatic
multicomponent CO2 post- X
436 M. Fosi-Kofal et al. / Journal of Natural Gas Science and Engineering 31 (2016) 428e436
in a membrane contactor. RSC Adv.
using the fractional rejection of
5, 21916e 21924. X nonionic and charged
combustion capture using hollow fibre membrane contactors. macromolecules. J. Membr. Sci.
Chem. Eng. Sci. 145, 45e 58. X 201, 191e 201. X
Feng, C., Shi, B., Li, G., Wu, Y.,
Atchariyawut, S., Feng, C., Wang, R., Jiraratananon, R., 2004. Preparation and properties of
Liang, D.T., 2006. Effect of membrane structure on mass- microporous membrane from Li, J.-L., Chen, B.-H., 2005.
poly(vinylidene fluoride-co- Review of CO2 absorption using
transfer in the membrane gase liquid contacting process using
tetrafluoroethylene) (F2.4) for chemical solvents in hollow fiber
microporous PVDF hollow fibers. J. Membr. Sci. 285, 272e
membrane distillation. J. Membr. membrane contactors. Sep. Purif.
281. X Sci. 237, 15e 24. X Technol. 41, 109e 122. X
Malek, A., Li, K., Teo, W.K., 1997.
Atchariyawut, S., Jiraratananon, R., Wang, R., 2007. Modeling of microporous hollow
Hou, D., Wang, J., Sun, X., Ji, Z.,
Luan, Z., 2012. Preparation and fiber membrane modules operated
Separation of CO2 from CH4 by using gase liquid
under partially wetted conditions.
properties of PVDF composite
membrane contacting process. J. Membr. Sci. 304 (1e 2),
hollow fiber membranes for Industrial Eng. Chem. Res. 36,
163e 172. X desalination through direct contact 784e 793. X
membrane distillation. J. Membr.

Bakeri, G., Matsuura, T., Ismail, A.F., 2011. The effect of Sci. 405e 406, 185e 200. X
Mansourizadeh, A., Ismail, A.F.,
phase inversion promoters on the structure and performance 2010. Effect of additives on the

of polyetherimide hollow fiber membrane using in gase liquid Hou, D., Dai, G., Fan, H., Wang,
J., Zhao, C., Huang, H., 2014.
contacting process. J. Membr. Sci. 383 (1e2), 159e 169. X
Effects of calcium car- bonate
structure and per- formance of
polysulfone hollow fiber
membranes for CO2 absorption. J.

nano-particles on the properties of Membr. Sci. 348, 260e 267. X

Bakeri, G., Ismail, A.F., Rana, D., Matsuura, T., Shariaty, M., PVDF/nonwoven fabric flat-sheet
composite membranes for direct
2012. Investigation on the effects of fabrication parameters on contact membrane distillation. Mansourizadeh, A., Ismail, A.F.,
2011. A developed asymmetric
the structure and properties of surface- modified membranes Desalination 347, 25e 33. X
PVDF hollow fiber membrane
using response surface methodology. J. Appl. Polym. Sci. 123,
structure for CO2 absorption. Int.
2812e 2827. X Husain, S., Koros, W.J., 2007.
Mixed matrix hollow fiber J. Greenh. Gas Control 5 (2), 374e

membranes made with modified 380. X

Bakeri, G., Matsuura, T., Ismail, A.F., Rana, D., 2012. A novel HSSZ-13 zeolite in polyetherimide

surface modified poly- etherimide hollow fiber membrane for polymer matrix for gas separation.
Masoumi, S., Keshavarz, P.,
gase liquid contacting processes. Sep. Purif. Technol. 89, J. Membr. Sci. 288, 195e 207. X Rastgoo, Z., 2014. Theoretical
160e 170. X investigation on CO2 ab- sorption
into DEAB solution using hollow
Boributh, S., Assabumrungrat, S., Laosiripojana, N., Ismail, A.F., Rahim, R.A.,
Rahman, W.A.W.A., 2008. fiber membrane contactors. J. Nat.
Jiraratananon, R., 2011. Effect of membrane module
Characterization of poly- Gas Sci. Eng. 18, 23e 30. X
arrangement of gase liquid membrane contacting process on
ethersulfone/Matrimid 5218
CO2 absorption performance: a modeling study. J. Membr. Sci. miscible blend mixed matrix
372, 75e 86. X membranes for O2/ N2 gas Mavroudi, M., Kaldis, S.P.,
separation. Sep. Purif. Technol. 63, Sakellaropoulos, G.P., 2006. A

Bougie, F., Iliuta, M.C., 2013. Analysis of Laplacee Young

200e 206. X study of mass transfer resistance in

membrane gase liquid contacting

equation parameters and their influence on efficient CO2
Korminouri, F., Rahbari-Sisakht, processes. J. Membr. Sci. 272,
capture in membrane contactors. Sep. Purif. Technol. 118 (0),
M., Matsuura, T., Ismail, A.F., 103e 115. X
806e 815. X 2015. Surface modi- fication of
polysulfone hollow fiber membrane
Mosadegh-Sedghi, S., Rodrigue,
Chung, T.-S., Jiang, L.Y., Li, Y., Kulprathipanja, S., 2007. spun under different air-gap
D., Brisson, J., Iliuta, M.C., 2014.
lengths for carbon dioxide
Mixed matrix membranes (MMMs) comprising organic absorption in membrane contactor Wetting phenom- enon in
polymers with dispersed inorganic fillers for gas separation. system. Chem. Eng. J. 264, 453e membrane contactors e causes and
Prog. Polym. Sci. 32, 483e 507. X 461.X prevention. J. Membr. Sci. 452,

332e 353. X
DashtArzhandi, M.R., Ismail, A.F., Matsuura, T., 2015. Lee, S., Park, G., Amy, G., Hong,
S.-K., Moon, S.-H., Lee, D.-H.,
Carbon dioxide stripping through water by porous Naim, R., Khulbe, K.C., Ismail,
Cho, J., 2002. Deter- mination of
A.F., Matsuura, T., 2013.
PVDF/montmorillonite hollow fiber mixed matrix membranes membrane pore size distribution
Characterization of PVDF hollow fiber membrane for CO2 1992. Microstructures in phase- Zhi, S.-H., Wan, L.-S., Xu, Z.-K.,
inversion membranes. Part 1. 2014. Poly(vinylidene
stripping by atomic force microscopy analysis. Sep. Purif. Formation of macrovoids. J.
fluoride)/poly(acrylic acid)/
Technol. 109, 98e 106. X Membr. Sci. 73, 259e 275. X calcium carbonate composite
membranes via mineralization. J.

Rafat, M., De, D., Khulbe, K.C., Nguyen, T., Matsuura, T.,
Membr. Sci. 454, 144e 154. X
Wang, R., Zhang, H.Y., Feron,
P.H.M., Liang, D.T., 2005.
2006. Surface character- ization of hollow fiber membranes
Influence of membrane wetting on
used in artificial kidney. J. Appl. Polym. Sci. 101, 4386e
CO2 capture in microporous
4400. X hollow fiber membrane contactors.
Sep. Purif. Technol. 46 (1e2),
Rahbari-Sisakht, M., Ismail, A.F., Rana, D., Matsuura, T., 33e 40. X
2012. Effect of novel surface modifying macromolecules on

morphology and performance of Polysulfone hollow fiber Yasuda, H., Tsai, J.T., 1974. Pore
membrane contactor for CO2 absorption. Sep. Purif. Technol. size of microporous polymer

99 (0), 61e 68. X membranes. J. Appl. Polym. Sci.

18, 805e 819. X

Zhang, Y., Wang, R., 2013.
Rahim, N.A., Ghasem, N., Al-Marzouqi, M., 2014. Stripping
Fabrication of novel
of CO2 from different aqueous solvents using PVDF hollow
polyetherimide-fluorinated silica
fiber membrane contacting process. J. Nat. Gas Sci. Eng. 21,
organice inorganic composite
886e 893. X hollow fiber membranes intended
for membrane contactor
application. J. Membr. Sci. 443,
Rahim, N.A., Ghasem, N., Al-Marzouqi, M., 2015.
170e 180. X
Absorption of CO2 from natural gas using different amino
Zhang, Y., Wang, R., 2014. Novel
acid salt solutions and regeneration using hollow fiber method for incorporating

membrane contactors. J. Nat. Gas Sci. Eng. 26, 108e 117. X hydrophobic silica nanoparticles
on polyetherimide hollow fiber
membranes for CO2 absorption in
Ren, J., Wang, R., Zhang, H.-Y., Li, Z., Liang, D.T., Tay, J.H.,
a gase liquid membrane contactor.
2006. Effect of PVDF dope rheology on the structure of
J. Membr. Sci. 452, 379e 389. X
hollow fiber membranes used for CO2 capture. J. Membr.

Sci. 281, 334e 344. X Zhang, Y., Wang, R., Yi, S.,
Setiawan, L., Hu, X., Fane, A.G.,
Rezaei, M., Ismail, A.F., Hashemifard, S.A., Bakeri, G., 2011. Novel chemical surface
modification to enhance
Matsuura, T., 2014. Experi- mental study on the performance hydrophobicity of polyamide-imide

and long-term stability of PVDF/montmo- rillonite hollow (PAI) hollow fiber membranes. J.
fiber mixed matrix membranes for CO2 separation process. Int. Membr. Sci. 380, 241e 250. X
J. Greenh. Gas Control 26, 147e 157. X
Zhang, Y., Sunarso, J., Liu, S.,
Wang, R., 2013. Current status and
Rezaei, M., Ismail, A.F., Hashemifard, S.A., Matsuura, T.,
development of membranes for
2014. Preparation and characterization of PVDF-
CO2/CH4 separation: a review. Int.
montmorillonite mixed matrix hollow fiber mem- brane for J. Greenh. Gas Control 12, 84e
gas-liquid contacting process. Chem. Eng. Res. Des. 92, 2449e 107. X
2460. X
Zhang, L., Qu, Z.Y., Yan, Y.F., Ju,
Rezakazemi, M., Shirazian, S., Ashrafizadeh, S.N., 2012. S.X., Zhang, Z.E., 2015. Numerical
investigation of the effects of
Simulation of ammonia removal from industrial wastewater
polypropylene hollow fibre
streams by means of a hollow-fiber membrane contactor. membrane structure on the

Desalination 285, 383e 392. X performance of CO2 removal

from flue gas. RSC Adv. 5 (1), 424e

Smolders, C.A., Reuvers, A.J., Boom, R.M., Wienk, I.M., 433.X