Review

Recent advances of inorganic llers in mixed matrix membrane for gas separation
P.S. Goh
a
, A.F. Ismail
a,b,
, S.M. Sanip
a,b
, B.C. Ng
a
, M. Aziz
a,c
a
Advanced Membrane Technology Research Centre (AMTEC), Universiti Teknologi Malaysia, 81310 UTM, Skudai, Johor, Malaysia
b
Faculty of Petroleum and Renewable Energy Engineering, Universiti Teknologi Malaysia, 81310 UTM, Skudai, Johor, Malaysia
c
Faculty of Science, Universiti Teknologi Malaysia, 81310 UTM, Skudai, Johor, Malaysia
a r t i c l e i n f o
Article history:
Received 8 June 2011
Received in revised form 26 July 2011
Accepted 31 July 2011
Available online 6 August 2011
Keywords:
Mixed matrix membranes
Gas separation
Inorganic llers
a b s t r a c t
Mixed matrix membrane (MMM) is a new class of membrane materials that offers the signicant poten-
tial in advancing the current membrane-based separation technology. As an attractive material that dem-
onstrates outstanding separation properties, MMM has been the subject of worldwide academic studies
conducted by many researchers especially those related to membrane technology. The past decades have
witnessed substantial progress and exciting breakthroughs in both the fundamental and application
aspect of MMM in various forms of separation, particularly in gas separation. These emerging materials
for separation have been traditionally accomplished by incorporating conventional inorganic llers such
as zeolite, carbon molecular sieve and silica nanoparticles in a polymer matrix. The recent advances have
shifted towards the introduction of new and novel materials namely carbon nanotubes, metal organic
framework and clay layered silicate as potential llers in the polymer matrix. The successful implemen-
tation of MMM depends greatly on the polymer matrix selection, the inorganic ller as well as the inter-
action between the two phases. The selection of suitable types of inorganic ller, the surface
modication, and the performance of the resulted MMM membranes were discussed and represented
the major contribution in this review. The recent efforts to tackle the underlying problems and the effects
of various kinds of modication that would eventually heighten the performance of membrane applica-
tions in gas separations were discussed. Better understanding on the improvement and optimization of
MMM process was provided by considering the possible solutions to overcome the problems encountered
during MMM preparation. This hybrid system holds signicant potential and great promise for further
investigations, development and applications. The future direction and perspective in MMM research
for gas separation was also briey outlined to further advance the materials for MMM in gas separation.
2011 Elsevier B.V. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
2. Mixed matrix membrane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
3. Inorganic fillers as dispersed phase in MMMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
4. Conventional fillers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
4.1. Zeolite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
4.2. Carbon molecular sieve. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
4.3. Silica. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
4.4. Metal oxide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
5. Alternative fillers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
5.1. Carbon nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
5.2. Layered silicate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
5.3. Metal organic framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
6. New and emerging material for MMM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
6.1. Graphene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
7. Conclusion and future direction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
1383-5866/$ - see front matter 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.seppur.2011.07.042

Corresponding author at: Advanced Membrane Technology Research Centre

(AMTEC), Universiti Teknologi Malaysia, 81310 UTM, Skudai, Johor, Malaysia. Tel.:
+60 7 5535592; fax: +60 7 5581463.
E-mail address: afauzi@utm.my (A.F. Ismail).
Separation and Purication Technology 81 (2011) 243264
Contents lists available at SciVerse ScienceDirect
Separation and Purication Technology
j our nal homepage: www. el sevi er. com/ l ocat e/ seppur
Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
1. Introduction
Membrane technology is an attractive separation approach that
has been studied extensively as a result of the fast and energy ef-
cient process without any phase changes [13]. Membrane-based
separation involves the use of membrane as a thin barrier between
miscible uids to separate a mixture. This interface may be molec-
ularly homogeneous in which it is completely uniform in composi-
tion and structure, or it may be chemically or physically
heterogeneous in which pores or layered structures are formed.
During the separation processes, a suitable force such as concen-
tration or pressure differential is applied to allow preferential
transport of one or more feed component across the membrane.
Generally, the permeation and selectivity are the two common ba-
sic performance characteristics of a membrane. In a broader con-
text of denition, the permeability is the ability of the permeates
to pass through a membrane meanwhile the ratio of permeability
of the more permeable component to that of the less permeable
is known as selectivity of the membrane. Higher permeability de-
creases the amount of membrane area required to treat a given
amount of gas, thereby decreasing the capital cost of membrane
units meanwhile higher selectivity will result in a higher purity
gas product.
Membrane is a promising candidate for separation due to the
advantages offered by the process which include high stability
and efciency, low operating cost and capital, low energy require-
ment and also ease of operation [47]. Furthermore, it is also
known for its reliability to be used in remote locations as mem-
brane has no moving parts and thus making it mechanically robust.
Since the last few decades, membrane-based separation has
emerged as a promising process for various industrial applications,
particularly in gas separation industries for air separation, hydro-
gen recovery and CO
2
removal [8]. Other well established mem-
brane processes in chemical and petrochemical industries are
listed in Table 1. One of the most important characteristics of
membranes is the ability to control the rate of permeation of differ-
ent species. The mechanism of permeation can be described using
two models, i.e. solution-diffusion model and pore-ow model [9].
Diffusion is the basis of the solution-diffusion model in which the
permeable components dissolve in the membrane materials and
then transport across by a concentration gradient. On the other
hand, the pore-ow model describes the transportation of perme-
able components by pressure-driven convection ow through tiny
pores in the membrane materials. In general, two types of gas sep-
aration processes have been encountered, i.e. gas permeation and
gas diffusion. The former process is of great industrial interest
where it is a pressure driven process in which the vapor compo-
nents pass through a non-porous membrane by a solution-diffu-
sion mechanism. Meanwhile, gas diffusion process can be done
for the microporous membranes that operate under a concentra-
tion or partial pressure gradient.
The successful application of membranes in an identied gas
separation process, in both lab and industrial scale, greatly de-
pends upon the selection of membrane materials with the appro-
priate chemical, mechanical and permeation properties. The
major issue of current membrane manufacturing to researchers is
to develop highly selective and permeable membrane materials
which also show sufcient resistance towards the feed compo-
nents as well as towards the operation condition [10]. Gas separa-
tion using polymeric membrane has achieved important success as
high separation performance membrane since the rst commer-
cial-scale membrane gas separation system which was produced
in the late 1970s [11]. Polymeric membranes, particularly those
prepared from glassy polymers, have received considerable atten-
tion because they possess advantages of mechanical properties,
reproducibility and relative economical processing capability
[12]. Polymeric membranes can be classied as: porous and non-
porous membranes. A porous membrane has rigid, highly voided
structure with randomly distributed, interconnected pores. Separa-
tion in these kinds of membranes is dependent on both the molec-
ular size and pore size distribution. Non-porous membranes or
dense membranes consist of a dense lm in which the permeate
molecules are absorbed and followed by diffusion through the
membrane matrix under the driving force of a pressure, concentra-
tion, or electrical potential gradient. The diffusivity and solubility
of the permeant molecules in the membrane material play a signif-
icant role to determine the mechanism of the gas transport. Among
the many types of polymers that exhibit gas separation properties
for gas mixtures, a few glassy polymers such as polysulfone (PSf),
polyethersulfone (PESf), polyetherimide (PEI) and polyimide (PI)
have been recognized as promising polymers with respect to their
permeability and selectivity. Despite their suitability for various
applications in research and commercialization, polymeric mem-
branes are still ineffective in meeting the requirement for the cur-
rent advanced membrane technology as these materials have
demonstrated a trade-off between the permeability and selectivity,
with an upper-bound evident as proposed by Robeson. A correla-
tion of membrane separation data offering an analysis of the limits
of polymer permeability and selectivity is often referred to as the
upper boundary in which the gas separation properties of the poly-
meric membranes follow distinct trade-off relations where more
permeable polymers are generally less selective and vice versa
[13,14]. The revisited upper bound relationship by Robeson for
O
2
/N
2
membrane separation for polymeric membranes is illus-
trated in Fig. 1 [15]. Apart from that, polymeric membranes are
also suffering in their limited solvent and poor chemical and ther-
mal resistance, as well as the occurrence of swelling phenomenon
that subsequently alter the membrane separation properties [16].
Due to the susceptibility of polymeric membranes to chemical
degradation and thermal instability, their applications have been
limited to separation processes where hot reactive gases are not
encountered. In order to combat the inherent limitations of poly-
meric membrane, research is underway for alternative membrane
materials. In this context, inorganic membranes that are formed
from metals, ceramic or pyrolyzed carbon has attracted global
Table 1
Various applications of membrane-based gas separation in industries.
Gas separation Applications
O
2
/N
2
Oxygen enrichment, inert gas generation
H
2
/hydrocarbon Renery hydrogen recovery
H
2
/N
2
Ammonia purge gas
H
2
/CO Syngas ratio adjustment
CO
2
/hydrocarbon Acid gas treatment, greenhouse gas capture
H
2
O/hydrocarbon Natural gas dehydration
H
2
S/hydrocarbon Sour gas treating
He/hydrocarbon Helium separation
He/N
2
Helium recovery
Hydrocarbon/air Hydrocarbon recovery
H
2
O/air Air dehumidication
244 P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264
interest as they offer several advantages over the polymeric mem-
branes for many gas separation processes. Inorganic membranes
are increasingly being explored due to their attractive characteris-
tics and advantages over the polymeric membranes [1720]. In
general, inorganic membranes can be categorized as dense and
porous. Porous inorganic membranes such as zeolite and carbon
molecular sieve (CMS) are favourable, rendered by their excellent
selectivity which is signicantly higher compared to that of poly-
meric membranes. Dense inorganic membranes possess very spe-
cic separation behaviours, for example metallic membranes that
are hydrogen or oxygen specic [21]. The ability to withstand high
temperature for long time and the resistance to harsh operation
environment have shown promise for application in membrane
reactors for many industrial processes. Besides having these robust
characteristics, inorganic membranes also have much higher
throughput and longer life span compared to polymeric mem-
branes. While attracting considerable attention in many industrial
processes, some disadvantages exhibited by inorganic membranes
are very critical and serious to be addressed. Although some inor-
ganic materials display properties well above the trade-off curve
for polymers, such properties are challenging to duplicate in
large-scale modules containing thousands of square meters of
membrane areas as it involves expensive capital and repairing cost.
Furthermore, the nature of inorganic membrane that is brittle and
normally with low surface-to-volume ratio has also hampered the
full optimisation of this material for industrial applications [17].
Many newattempts have been made to overcome the limitation
of polymeric and inorganic membranes. Introduction of new mem-
brane materials is still a big challenge to be researched in this fron-
tier technology, with most researchers looking at techniques to
enhance the separation processes. So called mixed matrix mem-
brane (MMM), where inorganic llers are dispersed at a nanometer
level in a polymer matrix have been identied to potentially pro-
vide a solution to the trade-off issues of the polymeric membrane,
as well as solving the inherent brittleness problems found in inor-
ganic membranes, is illustrated in Fig. 2. The choice of the polymer,
inorganic phase and ller particle loading are some of the impor-
tant parameters affecting the morphology and performance of
the MMM. The successful implementation of the further develop-
ment of MMM greatly lies on these parameters. In order to choose
the appropriate continuous polymer phase and dispersed inorganic
phase, the transport mechanisms and the gas component preferen-
tially transporting through the membranes should be rst taken
into consideration [22]. To obtain remarkably well performing
MMM that surpass the intrinsic properties of the polymer matrix,
the role of inorganic ller as the molecular sieve that favors one
component while hindering other components to pass through it
is of importance. Molecular sieves such as zeolite and CMS have
been conventionally packed in the polymer phase to form a dense
region of mixed matrix layer. These nanoporous materials possess
the shape and size selective nature and hence allow molecular
sieving discrimination by permitting smaller sized penetrates to
diffuse at higher rate than that of larger sized. Recently, alternative
nano-structured materials such as carbon nanotubes (CNTs), clay
and metal organic framework (MOF) have also found their excel-
lent sieving characteristic in the fabrication of MMM. However, de-
spite the promising application of these materials as inorganic
ller, the existence of adequate contact between the polymer
phase and dispersed ller to eliminate the interfacial defects is al-
ways a major issue to be addressed in order to achieve the desired
as well as improved performance of the resulting MMM.
The current contribution reviews recent scientic and techno-
logically advances in the application of different types of inorganic
ller materials in the fabrication of MMM. As a review of the state-
of-the-art in MMM using various kinds of llers, the progress that
has already been made so far will be explored. The membrane
properties, particularly gas separation performance of the MMM
using both conventional and alternative materials as the inorganic
phase are highlighted and discussed. This review article is meant to
stimulate the interest and enthusiasm towards the currently well-
identied as well as the under exploration research on MMM
among the membrane research community. Aside, it is also aimed
to impart greater understanding and hence assess the effectiveness
of each class of materials in tuning the intrinsic properties of the
polymer phase to form a MMM with improved and desired separa-
tion performance. The conclusion and future outlook summarize
the research activities and present future research directions.
2. Mixed matrix membrane
Since the pioneering literature by Zimmerman et al. [23], sev-
eral signicant review papers on the prospects of MMM as well
as the research reports on the capability of MMM as alternative
membrane materials for separation processes have been well pub-
lished [22,2427]. Much research into potential new materials for
Fig. 1. Revisited upper bound data for O
2
/N
2
membrane separation [15].
Dense structure
Inorganic fillers
Support layer Dense skin
(a)
(b)
Fig. 2. Mixed matrix membranes in conguration: (a) symmetric at dense mixed
matrix membrane. (b) Asymmetric hollow ber.
P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264 245
gas separation membranes is driven by the limitation and disad-
vantages which is in existence in the current available polymeric
and inorganic membranes. Few approaches have been carried out
to optimize the structure of the polymer chains in order to improve
the gas diffusivity by increasing the volume fraction with restric-
tive or selective channels communicating the voids. In addition
to that, attempts have also been made to increase the solubility
of permeate molecules by introducing certain chemical afnity
for a gas to allow adsorption of certain moieties of the polymer
[28]. As the consequences of the efforts made, MMM, a heteroge-
neous membrane consisting of inorganic ller embedded in a poly-
mer matrix, have emerged as an alternative approach in the state-
of-art membrane based separation technology.
MMMs are polymer composites that involve phase separated
membrane separation system. In this structurally engineered hy-
brid membrane, inorganic llers act to create preferential perme-
ation pathways for selective permeability while posing a barrier
for undesired permeation in order to improve separation perfor-
mance [29]. The superior permeability and selectivity of inorganic
membranes with the processability of polymeric membranes are
combined to achieve synergistic separation performance, in which
the rigid adsorptive porous type inorganic phase provides superior
separation properties and the polymeric phase enables the ideal
membrane forming hence solving the problem of fragility inherent
found in the inorganic membranes. A wide range of commercial-
ized polymer matrix choices exist within the so-called triangular
cross hatched region formed by the upper boundary such as Udel,
Matrimid and Ultem have been well known for their practical
applications and are readily available for the incorporation of
highly selective or permeable dispersed phase to further enhance
the current separation performances [23]. The incorporation of ll-
ers with molecular sieving properties in the polymer matrix is ex-
pected to lead to higher permeability, higher selectivity, or both
compared to the polymeric membranes. The incorporation of inor-
ganic llers to the polymer matrix may involve (i) spherical or low
aspect ratio llers or (ii) ller thin platelets with very high ratio
[30]. The latter case is of recent interest as it offers improved sep-
aration properties over the conventional case [14]. Although inor-
ganic llers are used in all cases of preparation, the resulting
morphology and membrane separation performance can be varied
greatly, primarily arise from different capability of these molecular
sieves, either based on shape or size, to discriminate between dif-
ferent molecules present in the feed mixture.
Typically, MMMs are easy to process and manufacture as com-
pared with the inorganic membranes. They are normally prepared
by casting a solution of polymer and dispersed inorganic phase,
and then evaporating the solvent in a controlled environment to
obtain a dense membrane. The critical reviews on the recent devel-
opment and application of MMM have revealed the viability of
these materials as an alternative route for the currently existing
polymeric and inorganic membranes in industrial applications
[22,23,27]. However, new fabrication approach is still in urgent
need to reduce materials cost and provide greater permeability
for further modications and improvements [31]. The signicant
breakthrough on MMM fabrication in useful congurations is
undoubtedly essential. Besides that, the comprehensive research
on MMM should also be devoted to the successful integration of
the inorganic llers to the polymeric phase [32].
3. Inorganic llers as dispersed phase in MMMs
The incorporation of inorganic llers into polymer phase can
signicantly alter the transport properties of one or several gases
through the MMM when compared to that of neat polymeric ma-
trix. The gas transport properties can be enhanced in several ways.
First, the incorporated ller particles could modify the properties
of the neighbour polymer phase, which will favor the overall trans-
port properties. This effect is especially signicant for particles that
are well distributed within the polymer matrix, where the frac-
tional free volume can be greatly modied [33]. It is also well
known that the presence of ller particles may alter the packing,
dynamics or conformation of polymer chain near its surface, and
impart great effect on the transport of large penetrates relative
to small ones. Therefore, the noticeable effect on selectivity of
the gas pair involving gases with signicant size difference like
CO
2
/He can be observed. Apart from that, according to Zimmerman
and Koros [23], the interface characteristics of the interface be-
tween the inorganic particles and the polymer matrix could deter-
mine the path to be passed through by one of the gases over the
others. This characteristic allows the selective transport of certain
gases hence improved selectivity is expected, or in some cases, it
can also result in increased permeability by reducing the path
length of the permeate molecules.
Cages with molecular dimensions open a room for separation to
take place, in which the shape and size of these voids can control
sorptive selectivity of the favoured permeating species. The best
established example of porous materials with these sieving charac-
teristics is the micro-scaled alumino-silicate zeolites and carbon
molecular sieve that have been traditionally featured by their
intrinsically high separation capacities. In fact, the formation and
gas separation properties of MMM using these conventional inor-
ganic llers have been well documented [22,34]. On the other
hand, another category of nano-scaled llers, such as silica, CNTs
and layered silicate clay are normally featured by intrinsically
lowseparation capacities. Nevertheless, a larger interfacial area be-
tween the llers and the polymer matrix per unit volume of the
llers can be achieved and have recently attracted considerable
attention due to their superior permeation properties [35]. Gener-
ally speaking, the variation in the structures and features of these
llers may subsequently produce a difference in the separation
performance behavior of the resulting MMM.
One of the main challenges encountered during the preparation
of MMM is to control the chemical structure, surface chemistry as
well as the particle types of the inorganic phase that would poten-
tially affect the membrane performance. The preparation of MMM
using polymer continuous phase has been reported to be compli-
cated due to the particle agglomeration during solution prepara-
tion in which mixing techniques are necessary to break up
particle agglomerates prior to the membrane processing. While
glassy polymers possess a much attractive separation property rel-
ative to rubbery polymers due to their high free volume, the poor
polymer chain mobility during the membrane formation may re-
sult in a weak interaction between the ller particles and the con-
tinuous polymer phase. The interaction between the polymer and
ller particles is of concern, as the undesirable channels may be
created between both phases if the polymer chains do not com-
pletely interact with the ller particles. Classication, which is de-
ned as llers separating out from the membrane matrix due to
incompatibility and forming separate ller phases or layers during
the formation of MMM may deteriorate the gas separation perfor-
mance of the membranes [36]. Fig. 3 illustrates the void formation
at the sieve-matrix interphase. The formation of these non-selec-
tive voids at the interface allows bypassing of gases hence deteri-
orates the selectivity of the MMM [3741]. To overcome this
problem, the presence of a bridging agent to facilitate good inter-
facial interaction is of crucial importance (Fig. 3c). In addition, ap-
proaches involving modication of surface hydrophobicity,
creating surface roughness as well as the varying of particle com-
position and pore architecture have also been investigated
[38,42]. Another challenge that could be faced during the prepara-
tion of MMM is the partial blockage of the ller pore by polymer
246 P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264
chains rendering the role of the incorporated ller [43]. In order to
minimize the adverse effect and to deal with the abovementioned
challenges, the well dispersion and incorporation of the inorganic
particles as individuals in the polymer matrix is one of the most
important tasks to be addressed. In such circumstance, surface
chemistry of the particles may need to be altered using various
treatment approaches.
4. Conventional llers
Over the past decades, in order to establish MMM with higher
gas separation performance relative to the neat polymeric and
inorganic membranes, various inorganic materials have been ex-
plored and numerous have been identied as potential ller in
membrane application. The most heavily researched type inorganic
llers are probably zeolite, CMSs and silica nanoparticles which
have been traditionally incorporated into the polymer matrix.
These llers are known for their versatility from the point of view
of chemical composition, particle shape and possibility of chemical
adaptation when embedded in polymer matrix. Particularly for
zeolite and CMS, the potentials are granted by their dened pore
structure that may contribute to enhanced gas selectivity. Prepara-
tion of MMM using these conventional llers have received signif-
icant attention as they have offered opportunities for advanced
separation that transcend upper bound without increasing the en-
ergy and processing cost. However, industry applications of these
conventional llers in the preparation of MMM is currently at
the bottleneck where the simultaneous increase in permeability
and selectivity that exceed Robesons trade off boundary is hardly
achieved due to the morphological imperfection created in the
MMM, in particular the lack of interfacial interaction between
the two different phases. To bring the effect of these llers into
play, modications through surface treatment are of urgent inter-
est. As mentioned earlier, several detailed reviews on MMM fabri-
cated using these conventional llers have been available. Thus,
the following sub-sections will focus on the recent development
of MMM using conventional zeolite, CMS and silica, with the
emphasis given to the surface modication of these llers. The re-
cent work on modication of the llers and the effects of these
modied llers on polymers segmental mobility, packing and
interchain interactions as well as the gas transport properties will
be discussed. Insight into the evolutions of the recent development
of MMM based on these conventional llers offers a clear picture
on how the attractive properties of the llers can be further ne-
tuned for excellent separation performance.
4.1. Zeolite
Zeolites are crystalline, hydrated aluminosilicate with opened
three-dimensional framework structures, regular intracrystalline
cavities and channels of molecular dimension. Zeolites possess
interesting physical and chemical properties and one of the most
remarkable properties, the sorption and diffusion properties of
zeolite, are due to the presence of different size channels and cav-
ities that are related to free space or void volume. Since the last few
decades, the improvement in MMM performance using zeolite as
the dispersed phase has resulted in the commercial alternative
over the polymeric and inorganic membrane. The use of zeolite
in the formation of MMM as potential ller for gas separation
membranes has received numerous attentions due to their thermal
stability as well as their promising separation and transport prop-
erties. Shape selectivity and specic sorption characteristics of zeo-
lite can be combined with easy processability of polymer to
provide desired properties in zeolite lled MMM. Numerous num-
bers of reports have indicated the favorable effects of employing
zeolite as the dispersed phase to improve the permeability and
selectivity of various polymers for gas separation of different gas
pairs [34,4452]. The gas transport properties of a zeolite-lled
MMM rely strongly on the intrinsic properties of the zeolite parti-
cles and the dense polymer phase. The expected improvement of
incorporating zeolite in the polymer matrix is gained from sus-
pended zeolite particles that are incorporated into the dense poly-
mer matrix. These particles are selective for one permeate
component over another. Thus, the undesired components typi-
cally travel a more tortuous path around the zeolite particles,
thereby decreasing mobility for that component and hence re-
sulted in the increase of overall selectivity for the desired species.
Early attempts in fabricating MMM using zeolite as dispersed
phase focused on the use of common zeolites such as zeolite-A.
The proper selection on the type of zeolite is vital to obtain high
performance MMM. Previous researches evidenced that the trans-
port properties of MMM is signicantly affected by the type of zeo-
lite used with pore size ranging from 4 to 10 [49,51,53]. This
wide range of zeolites can separate gas components either by size
discrimination or selective adsorption, depending on the sieving
characteristics of the particles. As an example, zeolite 4A, is more
favorable for the separation of O
2
/N
2
gas pair than other zeolites
due to its appropriate pore size of 3.8 to accurately discriminate
between the size and shape differences of spherocylindrical O
2
and
N
2
molecules [23,37]. Fig. 4 depicts the predicted zeolite 4A-PI
MMM performance based on Maxwells equation where the
shaded boundary represents the commercially attractive region
for forming a MMM with thin selective layer.
During the early stage of the development of zeolite-MMM,
incorporation of zeolite particles into rubbery polymer has shown
remarkable progress in the separation performance. The compati-
bility between inorganic ller and rubbery polymer is good due
to the high mobility of the polymeric chains capable to entirely
embed the particles [54,55]. Consequently, the promising results
exhibited have spurred considerable attention for more research
to incorporate zeolite particles into glassy polymer matrix in view
of their mechanical stability and more desirable transport proper-
(a) (b) (c)
Fig. 3. Interphase properties of MMMs: (a) void formation between the phases and (b) bridging of the ller and polymer matrix upon surface modication.
P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264 247
ties relative to rubber polymer [51]. In the preparation of MMM, a
key factor is represented by the afnity between the two phases in-
volved in the nal structure. However, unlike the good contact be-
tween rubbery polymers and zeolites at their interfaces, initial
attempts at fabricating MMM using glassy polymers and zeolites
have resulted in the presence of voids between the polymer and
zeolite. Poor adhesion at the zeolite-polymer interface can result
in sieve-in-a-cage morphology that is responsible for the non-
selective penetration of gas molecules as depicted in Fig. 5 [47],
hence reduce the apparent selectivity of the mixed matrix mem-
brane and increase the permeability. While the research during
the past few decades focused intensively on choosing an appropri-
ate molecular sieve with dimensions capable of discriminating two
penetrating gases with precisely similar dimension, the recent
development of zeolite/MMMis looking for effective solution to re-
solve the poor adhesion problem as mentioned above. A wide
range of modication involving the surface chemistry of zeolite ll-
ers in the formation of MMM has been introduced to allow
enhancement in polymer-ller interaction, thus optimization of
the transport and separation properties.
The early attempts were made through the coating of a diluted
solution of a highly permeable silicone rubber on the membrane
and later on, efforts to eliminate these unselective gaps often fo-
cused on the use of a coupling agent, which is normally a low
molecular weight compound with multifunctional groups, to intro-
duce favorable interactions between the polymer and zeolite. Low
molecular weight organic compound such as TAP has been used as
the tertiary component to link the polymer chain to the zeolite
crystal hence promote better adhesion at the interface. Alterna-
tively, the chain characteristics of the polymer matrix can be mod-
ied using low molecular weight additive that lead to
antiplasticization of polymer matrix [34]. Long aliphatic and poly-
aromatic based compounds containing polar atoms, rigid and pla-
nar structure are usually used to increase the stiffness of the
polymer matrix due to reduced rate of segmental motions in the
polymer chains which normally resulted in an increased selectivity
accompanied with decreased gas permeabilities [56,57]. Surface
modication of zeolite using silanes that enables chemical link be-
tween zeolite particles and the polymer matrix is also proven to be
an effective approach to improve both interfacial adhesion and gas
selectivity of the MMM by modifying the surface properties of zeo-
lite from hydrophilic to hydrophobic as well as increasing zeolite
afnity to the functional groups of the polymer matrix [5860].
However, the presence of common silane agent such as aminopro-
pyltriethoxysilane (APTES) or 3-aminopropyltrimethoxysilane
(APTMS) induces huge number of coupling point on the zeolite sur-
face which may result in pore blockage. In such circumstances, the
use of alternative silane agent such as APDMS could be solution to
reduce the number to coupling point hence avoid major blockage
of the zeolite pores [58]. Another way that has been reported to
eliminate the voids and solve the interfacial problem is by adding
a plasticizer to increase the exibility of the polymer matrix
[61,62]. Nevertheless, in some cases, the addition of a plasticizer
may lower the intrinsic gas separation performance of polymeric
materials. In view of the restrictions arisen from the addition of
foreign molecules to the polymer dope, the strategies to create bet-
ter adhesion have been suggested to carry out by forming a whis-
ker-like structure on the zeolite surface to provide additional
roughness for the interlocking between polymer chains and zeolite
particles. Table 2 summarizes the recent advances in the surface
modication to improve interaction between zeolite llers and
polymer matrix [34,52,6368].
While intensive efforts are still on-going to improve interaction
with the polymer materials, the recent zeolitic development has
also switched to structure and surface engineering to fabricate
new kinds of molecular sieve structure with more attractive prop-
erties. Hollow zeolite sphere (HZS), a newly developed spherical
molecular sieve through layer-by-layer assembling, is of interest
as it possesses advantages of zeolite as microporous and crystalline
molecular sieve, as well as the properties of spherical ller that
would minimize agglomeration and hence improve dispersability
and interaction with the polymer phase [6971]. The rst intention
of using HZS as inorganic ller in MMMs for gas separation has
been reported by Zornoza et al. [69], showing that the separation
performance of gas mixtures H
2
/CH
4
, CO
2
/N
2
and O
2
/N
2
was signif-
icantly improved due to the spherical shape that has contributed to
good dispersion and well ller-polymer contact as well as the hol-
low nature to allow fast ow and increase the gas permeability.
The author also pointed out that, the improvement in the perme-
ability was not only attributed by the free volume through the dis-
ruption of the polymer chains, but also related to the hollow space
Fig. 4. Predicted zeolite 4A-PI MMM performance based on Maxwells equation.
The shaded boundary represents the commercially attractive region for forming a
MMM with thin selective layer [37].
Fig. 5. Sieve-in-cage SEM morphology exhibited by zeolite particles when
incorporated in polymer matrix [47].
248 P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264
present in each HZS particles in which the molecules that selec-
tively permeate in the hollow particles are allowed to pass through
more easily with shorter path, compared to the molecules whose
diffusion or adsorption is not favoured by the zeolite llers. Fig. 6
shows the possible route transported by the molecules with differ-
ent afnity towards zeolite llers [69]. In another recent work,
Weng et al. [72] has demonstrated the feasibility of using CMS/
Al
2
O
3
as substrate to improve gas selectivity of MMM. The pres-
ence of the substrate has remarkably increased the roughness in
the channels with the poly(phenylene oxide) and SBA 15 interface
hence resulted in higher permeability of small gases such as H
2
and
CO
2
while interfering the permeability of bigger gases such as N
2
and CH
4
. Momentous improvement of H
2
/CH
4
from 17.6 to 50.9
was observed due to the existence of CMS/Al
2
O
3
substrate to facil-
itate the aggregation or merging of polymer nodules on the mem-
brane surface that is of great importance to obtain MMM with high
selectivity.
4.2. Carbon molecular sieve
While zeolites are attracting wide attentions in the fabrication
of MMM capitalizing on their size and shape selective capability,
the emergence of CMS has also offered an interesting option to
be potentially applied as the dispersed phase in MMM. CMS pro-
duced from the pyrolysis of polymer, primarily thermosetting
polymers, are carbonaceous porous solids that contains relatively
wide opening with constricted apertures that approach the molec-
ular dimensions of the diffusing gas molecules [73] and with this
manner, they are able to effectively separate the gas molecules
with very similar size. Since decades ago, CMS have been evidently
proven to be very effective for gas separation in adsorption appli-
cation due to their superior adsorptivity for some specic gases
[74,75]. Thereafter, the interest in developing CMS membranes
has been encouraging as a result of the outstanding permeabil-
ity-selectivity combination compared to that of polymeric mem-
branes [8,27,7679]. The high productivity accompanied with
excellent separation properties that exceed the limit of trade-off
boundary are mainly due to the porous nature of CMS that have
promoted their viability as the selective inorganic phase in the fab-
rication of MMM. The utilization of CMS particles within polymer
matrix is anticipated to offer several advantages over zeolites as
the molecular sieve entities. One of the major benets expected
from the former combination is that CMS particles appear to have
better afnity to glassy polymers, hence allow good adhesion at
the interface without introducing processing complexities. More-
over, the separation performances of the CMS can be tailored by
modifying the pyrolysis protocol to adapt to a specic gas separa-
tion [27].
The typical preparation of CMS/MMM involves the pyrolysis of
polymer to obtain the CMS particles, followed by the incorporation
of CMS particles in a polymer to form MMM with desired proper-
ties. Previously, the initial attempt to incorporate commercial CMS
particles into rubbery polymer did not give rise to any or too little
improvement on the separation performance, mainly due to the
dead-end porous structure of the particles which was inevitably
inherent in their manufacturing process [80]. Nonetheless, the la-
ter prediction by Zimmerman et al. [23] has provided further in-
sight into the gas transport properties of permeable CMS in
MMM. By considering the through porosity in CMS which permits
not only sorptive cavities for adsorptive processes but also allows
gas permeation, they predicted a promising separation perfor-
mance of O
2
/N
2
that were well above the upper bound trade-off
curve when PM800 particles, obtained from the pyrolysis of PI at
800 C under vacuum, were incorporated in Udel without the pres-
ence of defects at the molecular sieve/polymer interface. Due to the
enormous potential of CMS, continual efforts to develop and im-
prove the separation performance of CMS in MMM have been
undertaken, with an ultimate target of achieving simultaneously
high permeability and selectivity while maintaining the superior
processibility of typical conventional membrane formation tech-
nology [79]. One of the most signicant works corresponding to
the incorporation of CMS within polymer matrix to evaluate their
possibilities and potential for gas separation has been conducted
by Vu and coworkers [8,78]. The CMS incorporated Ultem and
Matrimid exhibited profound improvement as much as 45% and
20% for CO
2
/CH
4
and O
2
/N
2
respectively, indicating that the molec-
ular sieving pores and channels of CMS that allow for very high size
and shape selectivity have without any doubt been responsible for
the remarkable increase in the gas separation performance.
Sharing the commonly found limitations during the processing
of zeolite/MMM, the rigid nature of the molecular sieve always
give rise to the difculty to form continuous and defect free mem-
brane. Another critical issue in the fabrication of CMS/MMM is the
attainment of good adhesion between the two phases which is re-
quired for effective and optimum transport properties. So far, few
approaches have been adopted to solve the interfacial problems
between the phases of CMS and polymer. Through the manipula-
tion and control of synthesis method, CMS nanoparticles with
well-dened micropores and are readily dispersed in various sol-
vents can be produced [81]. It has also been veried that the occur-
rence of the sieve-in-cage morphology can be avoided by using
CMS in polymer dope with relatively high viscosities [8]. Another
viable route to avoid the sieve in cage morphology is though the
sizing or priming of the CMS particles with the polymer matrix
or sizing agents [8,82]. In fact, the role of sizing agents is still very
much debated. Some have considered it as a medium to minimize
the aggregation of CMS particles at high loading and promote com-
Table 2
Recent advances in zeolite/MMMs.
MMM composition Coupling agent/Silane
agent/additive
Gas pair Separation performance Reference
Polymer Filler With modication Without modication
Poly(RTIL) SAPO-34 RTIL [emim][Tf
2
N] CO
2
/CH
4
a = 24.9, P
CO2
686 Barrer [63]
PES SAPO-34 HMA H
2
/CH
4
H
2
/CO
2
a = 175.8, P
H2
7:1 Barrer
a = 4.64, P
H2
7:06 Barrer
a = 61, P
H2
0:206 Barrer
a = 2.45, P
H2
12:57 Barrer
[64]
PSf Zeolite 3A APTMS H
2
/CO
2
a = 2.4, P
H2
14:3 GPU a = 1.7, P
H2
82 GPU [65]
PDMC SSZ-13 APDMES CO
2
/CH
4
a = 41.9, P
CO2
88:6 Barrer [66]
PES Zeolite 4A Dynasylan Ameo CO
2
/CH
4
O
2
/N
2
a = 28.7, P
CO2
6:7 Barrer
a = 2.7, P
O2
4:3 Barrer
a = 2.9, P
CO2
33:8 Barrer
a = 2.1, P
O2
9:0 Barrer
[67]
PEEK Zeolite NaA DEA O
2
/N
2
a = 4.2, P
N2
0:12 Barrer a = 4.2, P
N2
0:16 Barrer [52]
PC Zeolite 4A p-NA H
2
/CH
4
CO
2
/CH
4
a = 135.0, P
H2
13:4 Barrer
a = 51.8, P
H2
4:6 Barrer
a = 54.4, P
H2
13:4 Barrer
a = 32.5, P
H2
7:8 Barrer
[34]
PMMA Zeolite 4A TMOPMA O
2
/N
2
a = 5.4, P
O2
0:86 Barrer a = 5.1, P
O2
2:4 Barrer [68]
P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264 249
patibilization of the sieve with the polymer through the formation
of more chemical reactive site and surface area to facilitate better
adhesion without involving complicated chemical reaction and
grafting process, while some claimed that sizing agents are not in-
volve in promoting adhesion but is present to prevent the defects
and damage of the particle surfaces [8385]. Razah and Ismail
[82] reported the use of PVP K-15 as sizing agent for CMS in the
preparation of CMS/Udel MMM to bridge the sieve and matrix by
physically inducing the molecular interaction and allow the gas
transport through the phases without creating additional non-
selective and resistive layer in the interphase region hence im-
proved the O
2
/N
2
selectivity from 3.69 to 6.05.
4.3. Silica
Another conventional class of inorganic ller that has received
signicant attention throughout the development of MMM is silica
nanoparticles which can be further categorized into non-porous
silica and ordered mesoporous silica. These llers are in general
introduced in the polymeric matrix to form an heterogenous mem-
brane through sol-gel reaction in special conditions in order to de-
velop nanoscale particles of the inorganic oxide in a polymer
matrix. In this reaction, the silica precursors are hydrolyzed and
condensed into the dispersed nanoparticles in the polymer matrix.
Since the dispersion is at the molecular or nano-scale level, the
interactions between the silica and the organic part can be tailored
in order to manipulate the morphological structure at the interface
of the two phases [86,87]. This interesting nding is in contradic-
tion to the Maxwell model prediction of deterioration of the sepa-
ration performance by the addition of nonporous llers to a
polymer matrix due to the loss of polymer volume available for
sorption and decrease in diffusitivity attributed to the increased
penetrate diffusion pathway length [88,89]. The intensive research
in silica/polymer MMM system have shown that the addition of
non-porous nanosized fumed silica, which has opposed properties
with porous inorganic llers is of great potential to affect polymer
chain packing in glassy and high-free-volume polymers conse-
quently bring about alteration in the gas separation properties.
Due to the non-permeability of the nonporous silica particles, the
addition of this ller into the polymer matrix does not directly con-
tribute to the change of transport property, but it alters the molec-
ular packing of the polymer chains, resulting in an improvement of
the permeation as well as the selectivity [90]. Particularly, the
incorporation of nonporous silica into polymer matrix can give rise
to two outcomes [91]: (i) increase of polymer free volume without
creating non-selective voids which in turn results in increased gas
permeation properties and (ii) formation of free volume elements
that are large enough to permit non-selective Knudsen transport
hence resulting in a decrease in selectivity.
The separation performance of silica/polymer system in the
early development did not show signicant improvement that
transcends Robesons upper bound. In spite of that, stimulating
improvement in permeability upon the addition of silica particles
within the polymer matrix has been widely reported [9296].
The preliminary investigation of MMM incorporating silica parti-
cles, has proposed that the increased gas permeability was the re-
sult of the interaction between the residual silanol groups of the
silica domain and polar gases such as CO
2
hence resulting in im-
proved gas solubility due to the increase of the Henrys law coef-
cient and the Langmuir sorption constants in the MMM.
Additionally, the presence of silica particles have induced the mor-
phology change at the interface resulting in the increased amor-
phous region of the MMM as well as prompting the increase in
the mean distance between the polymer chains through the reduc-
tion of the polymer chain packing density at the interface between
the two phases. In such case, the polymer structure stiffness as-
cribed to the increased tortuosity and restricted segmental motion
has resulted in higher diffusivity and diffusivity selectivity by dis-
rupting inter-chained packing. An increasing polymer backbone
stiffness, also plays a more determining role in the polymer sepa-
ration performance in comparison with the solubility factor
[93,97]. The nature of the interface morphology, corresponding to
changes in the free volume concentration, free volume cage size
as well as the total frictional free volume of the MMM is strongly
inuenced by the type of silica, i.e. methyl phenyl or silanol groups
that may interact with the polymer matrix [98]. Besides that, the
silica particle loading also signicantly affects the permeability of
the gases as at high loading of the nanoparticles and with an in-
creased aggregate size, may also give rise to the nonlinear expan-
sion of the free volume [99]. In spite of very little improvement
in terms of the gas selectivity [100], the permeation enhancement
upon the addition of silica nanoparticles is still an interesting topic
as this characteristic becomes signicant in low-permeable poly-
mers like Matrimid 5218 and polybenzimidazole (PBI) that are sus-
ceptible to permeability increasing effect, thus pushes the
performance of these polymers to transcend the upper bound.
For MMM derived from the incorporation of high loading of sil-
ica nanoparticles, the permeation behaviour of the composites de-
pends greatly on the chemical compatibility of the silica particles
with the polymer matrix. Recent studies have shown that the weak
interaction between the silica particles and polymer matrix may
induce void formation during lm fabrication, which has a signi-
cant effect on the physical properties, as well as the gas transport
performance of the hybrid membrane [88,101104]. Paul and
Takahashi [101] revealed that dispersion of fumed silica nanopar-
ticles into PEI has resulted in small but very inuential voids or de-
fects formation at silica/polymer interface or within the
aggregates. However, by controlling the interaction between these
two phases, the amount of voids at the interface can be controlled
and the presence of extra voids volume is believed to contribute to
improved gas permeability. Therefore, the reduction in selectivity
Fig. 6. The white arrows represent the faster permeating molecules (H
2
, CO
2
and
O
2
) across 8 wt% HZS-PSF MMMs. While the black arrows represent the molecules
i.e. CH
4
and N
2
with less interaction with hollow zeolites [69].
250 P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264
regardless of gas pair indicated the voids created a parallel path-
way through the polymer matrix, as illustrated in Fig. 7, hence
allowing unselective transport of all gases when considering that
there is no effect of the nanoparticles on the local matrix proper-
ties. In the consecutive study where the silica particles were chem-
ically modied with silane coupling agent with unique bifunctional
groups to create greater compatibility with the polymer matrix,
they reported the reduction of the gas permeability for all gases
due to the elimination of parts of the void contents [105107]. It
has been suggested that by using a coupling agent, the hydroxyl
groups on the surface of the silica particles could be replaced by
the organofunctional groups of the coupling agent, hence facilitate
better reaction with the polymer chains [108].
While the exploratory research on silica/polymer MMM empha-
sized the ability of this hybrid system in boosting the gas perme-
ability and justied the suitability of silica particles in highly
selective but low permeability polymeric membrane to result in
gas separation performance that surpasses the trade off curve, re-
cent investigations have focused on the efforts to improve both
permeability and selectivity in order to make this material more
attractive for industry applications. Sadeghi et al. reported a simul-
taneous enhancement in all gas permeability and as much as 20
times increase in CO
2
/N
2
and CO
2
/CH
4
gas selectivity in MMM pre-
pared by incorporating high loading of silica nanoparticles within
ethylene vinyl acetate (EVA) [109] and PBI [87] via solgel method.
Fig. 8 shows the transport properties of gases through neat PBI and
PBI-silica MMM. The relative high content of silica accompanied by
the increased number of active OH group sites has given rise to an
increase in solubility and a corresponding decrease in the diffusiv-
ity of the gases through the MMMs that eventually resulted in the
enhanced gas solution of condensable CO
2
gases in the polymer
matrix and reduction in the permeability of the non-condensable
N
2
gas, hence changing the dominant gas permeation mechanism
from the diffusion to the solution diffusion. The authors also
pointed out that, the reduced diffusivity of the gases can be related
to the restricted motion of the gas molecules in the polymer phase
and formation of pathways with more tortuosity in the polymer
upon the addition of silica particles. Very recently, Xing and Ho
[110] have investigated the separation performance of MMM pre-
pared by incorporating fumed silica into crosslinked polyvinylalco-
holpolysiloxane. The addition of silica particles has resulted in
signicant effect on the separation performance in which the silica
content is the most inuential factor to manifest simultaneously
high CO
2
permeability and CO
2
/H
2
selectivity, which can be attrib-
uted to the decrease of H
2
transport within the MMM as the pack-
ing density of nonporous silica increased.
Ordered mesoporous silica with variety of particles size, shape
and pore diameter is another form of silica that has been poten-
tially used for the development of new generation of MMM. Prior
to their employment in MMM, these mesoporous materials have
already received worldwide attention as viable absorbent and
membranes in many applications [111,112]. Two of the most com-
mon members, MCM-41 and MCM 48 are distinguished from their
pore structure in which the former has a one-dimensional pore
channel structure while the latter possesses three-dimensional
interconnected cubic pore structure. Since the discovery of their
favorable effects of increasing the permeability of MMM without
decreasing its selectivity due to its good compatibility with the
polymer matrix [38], the potential application of these ordered
mesoporous materials have been expanded as inorganic ller in
the fabrication of MMM. Unlike the role of other size selective
molecular sieves, the incorporation of these mesoporous materials
in MMM may result in a selective membrane through the improve-
ment of the ller-polymer interaction [113]. Since the cross sec-
tional areas per chain of the most selective synthetic polymer are
Fig. 7. The possible distribution of spherical silica llers in MMMs that are partially
fused together (a) dispersed ller and voids and (b) agglomerated ller and voids
[101].
Fig. 8. The transport properties i.e. (a) permeability and (b) permselectivity of the gases through neat PBI and PBI silica hybrid membranes versus weight fraction of silica in
the polymer [87].
P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264 251
around 1 nm
2
or less and considering the typical ordered mesopor-
ous silica has relative large pores ranging from 2 to 50 nm, the
polymer chains are able to penetrate into the mesoporosity of
the ller to give rise to a real homogenous nanoporous composite,
hence render the good adhesion between the llers and polymer
matrix. Another interesting feature of these ordered mesoporous
silicas when incorporated in polymer matrix is mainly due to the
large surface area with abundant reactive silanol groups, which
can bridge the polymer chains through hydrogen bonding.
The increase in gas permeability as compared to those in neat
PSF was observed for MCM-48/PSF [114] and MCM-41 /PSF [115]
MMM by Kim et. al. The double increase in permeability of the
MMM was associated to an increase in diffusivity as well as solu-
bility. When the loading of mesoporous silica in the polymer ma-
trix is high enough to lead to the formation of a connected silica
inorganic phase, the transport properties of the composite can be
ruled by the inorganic phase [116]. The increase in diffusivity can
be accounted to the presence of high diffusivity tunnels and redis-
tribution of the rigid polymer chains near the pore entrance. The
continuous pathways present in the polymer matrix incorporated
with MCM-48 as well as MCM-41 silica have allowed the diffusion
of gas molecules solely through the molecular sieve phase. The
concurrent increase of the gas solubility was attributed to the high
coverage of silanol groups on the silica surface that has resulted in
higher adsorption capacity of the mesoporous silica relative to that
of PSF.
However, studies also revealed that the hydrogen-bonding
interactions between the surface silanol groups on the surfaces of
the mesoporous silica and the aryl ether groups of PSF can be det-
rimental to the permeability of gases as the silanol groups may
cause blockage near the pore entrance, leaving the inner pores
inaccessible [38]. Moreover, this material suffers from some limita-
tions concerning gas separation performance as the gas transport
can be dictated by the inorganic mesoporous and the mechanism
commonly follows the Knudsen diffusion model where the silica
llers has higher permeability than the polymer phase.
Recently, Zornoza et al. [94,113] investigated the spherical mes-
oporous MCM-41 with 24 lm pore size by taking advantage of
the narrow size distribution that facilitate the formation of homo-
geneous MMM through the minimization of agglomeration of the
silica particles. Increase in both permeability and selectivity for
H
2
/CH
4
and CO
2
/N
2
at optimized loading of 8 wt% has been re-
ported, due to the bimodal pore distribution that favors the gas dif-
fusivity as well as the disruption of polymer chain packing and
linking due to the presence of silica llers. The results displayed
in Fig. 9 indicates that with the optimum loading and testing con-
dition (temperature and pressure), the relationship between per-
meability and selectivity for the mixture are located within the
Robeson upper bound region. The study also revealed that the dif-
ferent approach use for the removal of structural agent from the bi-
modal pores of the ordered mesoporous silica spheres has an
affecting role in the separation performance where the selectivity
of chemically extracted mesoporous silica was slightly lower than
that of the calcined ones, while exhibiting relatively higher perme-
ability. The authors concluded that these chemically extracted
mesoporous silica spheres could be an energetically cheaper alter-
native for the fabrication of silica/MMM [94].
4.4. Metal oxide
The preparation of MMM using impermeable particles has been
dominated by the use of fumed silica as the dispersed phase.
Although some grades of fumed silica contain primary particles
as small as 13 nm in diameter, these particles are chemically fused
together so that it is not possible to disperse the primary particles
individually or in nanoscale aggregates. Metal oxide nanoparticles
such as MgO and TiO
2
are emerging materials due to their potential
applications for membrane-based separation purposes. The pri-
mary particles with diameter in nano-scale and high specic area
of these metal oxides allow improvement in particle distribution
and prevent non-selective void formation in nanoparticles/poly-
mer matrix interface. Therefore, these nanoparticles are not inher-
ently fused together and have potential to be dispersed
individually or in nanoscale aggregates. The incorporation of metal
oxides normally exerts similar effect as that of impermeable silica
particles in which the alteration of gas transport behaviour in such
MMM is the result of chain packing disruption and nanoscale
agglomeration of nanoparticle in polymer matrix.
It is in common consent that, the incorporation of metal oxide
nanoparticles in the optimum condition would result in signi-
cantly improved permeability while maintaining selectivity or in
certain cases, diminishing the overall diffusivity selectivity of the
MMM due to the inability of the porous structure of nanoparticles
to selectively distinguish the gas molecules based on their size dif-
ference. Several research work related to MgO embedded in MMM
have been reported [117119]. According to Hosseini et al. [117],
the enhanced gas permeability was ascribed to the effect of nano-
particles on the gas diffusivity due to the incorporation of highly
porous MgO nanoparticles which have substituted some portions
of the dense structure of polymer chains in the membrane struc-
Fig. 9. Results for (a) H
2
/CH
4
and (b) CO
2
/N
2
at different pressures and temperatures for mesoporous silica sphere incorporated into Matrimid and Udel [94].
252 P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264
ture. The increase in permeability was also seen to be the result of
the presence of microvoids at the polymerparticle interface as
well as the inherent transport properties of porous particles MgO
nanoparticles that encompass pore size that is larger than the ki-
netic diameter of gases in which all the gas molecules, regardless
of size, can pass through the particles without any resistance
against their transport. In principle, the presence of afnity and
interaction between MgO surface and some gas species, CO
2
for
example, has also prompted the enhancement in the transport of
the molecules. For instance, Matheucci [119] reported that CO
2
permeability increased from 52 barrer in the neat poly(butadiene)
to 650 barrer in MMM containing 27 vol% (nominal) MgO, ascribed
to the surface properties of MgO nanoparticles that are basic in
nature also show high afnity of physisorption towards acidic
CO
2
molecules hence adsorbed large concentration of CO
2
even at
low pressure and consequently enhanced the gas permeability.
Matteucci et al. have conducted systematic study to investigate
the inuence of TiO
2
particle surface chemistry on the gas trans-
port properties of the MMM based on both glassy [120] and rub-
bery polymer [121]. Upon the incorporation of TiO
2
into the
polymer matrix, the dispersion of these nanoparticles was varied
depending on the amount of loading. The MMMs were presumed
to be defect free at low particle loading in which the nanoparticles
were dispersed individually and in nanoscale aggregates whereas
when high loading of TiO
2
were lled into the polymer, some nano-
particles formed micron-sized aggregates, indicating the presence
of transmembrane defects at such high ller concentration. Besides
demonstrating great inuence on the dispersion properties, the
loading of TiO
2
was also found to manipulate the gas transport
properties of the MMMs. The diffusivity selectivity of gas pairs con-
sisted of CO
2
and nonpolar gas appeared to increase with increas-
ing particle loading. Similar trend was observed in a recent study
performed by Moghadam et al. on the effect of TiO
2
when embed-
ded in glassy polymer [122]. The void formation in nanoparticles
polymer matrix interface as well as agglomeration of particles and
weak interaction between polymernanoparticles at interface, par-
ticularly at high loading has led to greater permeability compared
to that predicted by Bruggemans model. In general, reduction in
MMMs selectivities relative to pure Matrimid has been observed,
due to the decrease in diffusivity selectivity. In the case of CO
2
,
the permeability enhancement of Matrimid containing 20 vol%
TiO
2
was 2.45 times higher than neat Matrimid, while CO
2
/CH
4
selectivity decreased by 33%, revealing that the use of TiO
2
nano-
particles improved membrane performance in CO
2
/CH
4
separation
yet presented a trade-off line with relatively similar slope with re-
spect to Robeson upper bound. This tendency has arisen from the
increase in free volume and voids, which in turn decrease the
size-selective nature of the membrane, so that permeability
enhancement of large molecules such as nitrogen and methane
was more signicant than that of smaller one.
5. Alternative llers
Despite the rapid advancement in research focusing on conven-
tional llers that have been traditionally applied in the preparation
of MMM, the efforts to create ideal membrane with desired sepa-
ration properties are still hampered by many undesired and
unavoidable obstacles, as identied in the above mentioned stud-
ies. For instance, the sieving characteristic of zeolites and CMS is
attractive however their size and aspect ratio is less favourable
for the production of asymmetric membrane that utilizes a thin
permselective layer. Furthermore, the rigid pore size with limited
number of possible structure and composition in the convention-
ally used zeolite and CMS are insufcient to effectively discrimi-
nate the gas molecules of similar sizes. A ller material with a
controllable cage dimension is of greater interest as this feature
may facilitate better sieving properties to the resulting MMM. An-
other point of concern is that the external surface modication of
the ller particles, particularly with silane coupling agents, has
unavoidably give rise to the pore blockage and limit the transport
of the gas molecules [39]. Apparently, the developments of MMM
using conventional llers are signicant yet inadequate to t the
increasing expectation for the practical application in industries.
The challenges faced by this MMM have urged extensive re-
searched on the use of other promising alternative materials to
help mitigate the existing problems in order to meet the desired
membrane separation performance.
It is only very recent since attention was driven to this new
class of alternative potential llers. Nowadays, the advancement
in science and technology has allowed the synthesis of nanomate-
rials with great control in respect to their composition. While there
are still much work devoted to the basic synthesis and character-
ization of the building blocks, the current challenge seems to trans-
fer the nano-scale properties of these materials into macro-scale
structures. The remarkable versatility of these alternative nano-
structured llers upon the incorporation in MMM has been recog-
nized as one of the greatest achievements in the area of membrane
based gas separation processes as this mixed matrix system has
shown tremendous promising separation results. The following
sections review and provide further details on how these recently
emerging materials, i.e. carbon nanotubes (CNT), clay and metal or-
ganic framework (MOF) have offered new and huge opportunities
to the development of MMM endowed with attractive and desired
characteristics that are comparable or even triumph over the con-
ventional inorganic llers.
5.1. Carbon nanotubes
In recent years there have been tremendous research advances
made for carbon nanotubes (CNTs) due to their unprecedented
physical and chemical properties. The remarkable increase in the
number of papers published in the fundamental research as well
as in the advanced engineering has indicated the great and unlim-
ited potential of this material to generate huge research activity in
most areas of science and engineering. CNTs are a novel and inter-
esting graphitic carbon material that is made up of graphite sheets
that have been rolled into a tube. This hollow cylinder, usually
capped at least at one end, is considered as nearly one-dimensional
structures according to the high length to diameter ratio where the
diameter of a nanotube ranges from a few to tens of nanometers,
while its length can reach up to several millimeters. Generally, as
depicted in Fig. 10 [123], CNTs can be synthesized as singular tubes
known as single walled carbon nanotubes (SWCNTs) or as a series
of shells of different diameters spaced around a common axis
called multiwalled carbon nanotubes (MWCNTs). Since the discov-
ery of this relatively new class of carbon material, researchers have
envisaged by taking advantage of their extraordinary electronic,
mechanical, thermal and optical properties that have sparked great
interest in their use in a wide spectrum of promising applications.
While the research in CNTs is continuing to discern the utility of
this nanostructured material, more possible applications are cur-
rently being explored and developed. One of the most signicant
applications of CNTs is probably as an ideal reinforcing ller in
polymer composites that are poised to exhibit exceptional and de-
sired properties through the careful and detailed control of the
processing and fabrication parameters. The initial interest of incor-
porating CNT in polymer lay in the possibility offered by the nano-
tubes to result in superb mechanical and electrical properties of
the fabricated polymer nanocomposites [124129]. In the context
of application in polymer composite, MWCNTs are of particular
interest over SWCNTs in view of their relatively low cost and avail-
P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264 253
ability in larger quantities as a result of their more advanced stage
in commercial production [130].
Since the last decade, considerable research, as summarized in
Table 3, has been focused on the incorporation of carbon nano-
tubes into polymers to prepare MMM [36,131137]. The intensive
research conducted are primarily inspired by the theoretical work
by Sholl and co-workers [138,139] that have reported about the
transport rate of small gases such as H
2
and CH
4
in close packed
bundles of SWCNTs being extremely rapid and in orders of magni-
tude faster than in zeolite and any other known inorganic nanopor-
ous materials. The pioneering work of incorporating aligned
MWCNTs into polymer matrix for transport studies conducted by
Hind et al. [140] has provided a clear indication of the potential
of the nanotube inner cores to act as a feasible channel for the
transport of N
2
and proposed the potential application of CNTs in
chemical separation and sensing. Several atomic simulation stud-
ies have suggested that CNTs can be served as an ideal candidate
for gas adsorption and separation purposes as a result of their
superior selectivity and permeability for the transport of light
gases [141,142]. The exceptional high self and transport diffusivity
is attributed to the extraordinary inherent smoothness of the po-
tential-energy surface dened as when the nanotubes are well dis-
persed in the polymeric matrix and all gas molecules pass through
the tube channels resulting in high gas permeability [136]. In the
case where the CNTs are vertically aligned to the membrane sur-
face, they would behave like pinholes that allow rapid transport
of the gases passing through the channel of the nanotubes, leading
to high gas permeability without selectivity [36]. The adsorption
selectivity of the gas pair is strongly driven by the interaction
potentials of the molecules with the graphitic CNT walls. The dif-
ference in the sorption interaction between the permeating gas
molecules and CNT walls gives rise to improved separation factor
of the gas pair in which it has been evidenced theoretically and
experimentally that CH
4
molecules showed preferential sorption
capacity over other gases like H
2
, O
2
and N
2
, resulting in enhanced
gas selectivity [136].
The expected superior separation properties of the CNTs are
hard to be realized due to the difculties to assess and achieve
morphological defect-free CNT/MMM in real experimental studies.
In order to obtain ideal membrane structure in which the spaces
between the CNTs are lled with a continuous polymer lm and
the closed ends of CNTs are etched opened, the polymer must have
high wettability with the nanotubes to allow good interaction be-
tween the two phases. Unfortunately, the effort of incorporating
CNTs into polymer matrix share the same challenge as other inor-
ganic llers where the poor adhesion of the different phases elicits
the presence of an interphase polymer layer near the nanotubes,
with properties differing dramatically from the bulk polymer.
Apart from the incompatibility of the ller and polymer matrix,
the intrinsic properties of CNTs which tend to form stabilized bun-
dles in nature due to the presence of van der Waals force are also a
hindrance to the well dispersed nanotubes in the polymer matrix
[143,144]. The nature of the dispersion problem for CNTs is rather
different from other conventional llers, such as spherical particles
and carbon bers, because of the characteristics of CNTs such as
small diameter, high aspect ratio and thus extremely large surface
area [145]. As a result from the agglomeration and entanglement of
CNTs, the defect sites are created hence drastically weakening the
effect of CNTs as the ller in the MMM. Correspondingly, to fully
utilize the advantages featured by CNTs in a practical manner,
the surface modication of CNTs using physical and chemical ap-
proaches have been adopted. There are two major issues that need
Fig. 10. (a) Different structures of CNTs and (b) TEM of aligned CNTs [123].
Table 3
Recent advanced in CNT/MMMs.
Polymer Functionalization/
modication
Optimum loading Gas pair Separation performance Reference
Neat polymer CNT/polymer
PES H
2
SO
4
/HNO
3
5 wt% MWCNTs CO
2
/N
2
a = 22.5, P
N2
0:12 Barrer a = 22.5, P
N2
0:2 Barrer [131]
PES H
2
SO
4
/HNO
3
Ru metal
5 wt% MWCNTs CO
2
/N
2
a = 21.8, P
N2
0:12 Barrer a = 22.1, P
N2
0:20 Barrer
a = 26.5, P
N2
0:13 Barrer
[132]
PEI Triton-X100 1 wt% MWCNTs O
2
/N
2
a = 21.8, P
N2
0:12 Barrer a = 21.8, P
N2
0:12 Barrer [133]
PI Chitosan 1 wt% MWCNTs CO
2
/CH
4
a = 19.1, P
CO2
8:9 Barrer a = 10.0, P
CO2
14:3 Barrer [134]
PBNPI H
2
SO
4
/HNO
3
15 wt% MWCNTs H
2
/CH
4
CO
2
/CH
4
a = 6.7, P
H2
4:7 Barrer
a = 3.7, P
CO2
2:6 Barrer
a = 6.7, P
H2
8:0 Barrer
a = 3.7, P
CO2
3:4 Barrer
[135]
BPPO
dp
HNO
3
5 wt% MWCNTs
5 wt% SWCNTs
CO
2
/N
2
CO
2
/N
2
a = 30, P
N2
2:6 Barrer a = 31, P
N2
4:7 Barrer
a = 30, P
N2
2:6 Barrer
[36]
PSf H
2
SO
4
/HNO
3
10 wt% SWCNTs O
2
/N
2
CH
4
/N
2
a = 5.1, P
N2
0:17 Barrer
a = 1.0, P
CH4
0:17 Barrer
a = 5.4, P
N2
0:23 Barrer
a = 1.2, P
CO2
0:28 Barrer
[136]
254 P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264
to be tackled during the surface modication of CNTs prior to the
insertion into the polymer matrix. First is to debundle the highly
entangled nanotubes to allow improved interfacial interaction
and second, the dispersion of the nanotubes to obtain CNTs that
are homogeneously distributed within the MMM to facilitate effec-
tive gas transport [146].
Aside from the relatively smooth CNT surfaces that render weak
interfacial bonding, the nanotubes present large area density of
graphitic walls as well as core entrances that can be functionalized
by molecules of length, hydrophobicity or chemical functionality,
depending on the application of the modied CNTs. Generally,
the surface modication of CNTs can be conducted in two manners,
i.e. by physical and chemical means. The most conventionally and
commonly used functionalized method is through the surface
treatment with strong inorganic acids [147149]. During the acid
treatment, functional groups such as hydroxyl and carboxyl groups
are introduced to the surface of the CNTs to improve their compat-
ibility with the polymer matrix. During the processing, the func-
tionalized CNTs are soluble in a wide range of organic solvents as
the hydrophobic nature of CNTs has been altered to hydrophilic
due to the attachment of polar functional groups. As the enhance-
ment in gas permeability of MMM is mainly ascribe to the dis-
turbed polymer chain packing by the llers, the intact interaction
of the polymer matrix and functionalized CNTs is expected to im-
part more intimate insertion of the nanotubes into the polymer
chains hence more effectively increase the gas permeability [36].
As mentioned earlier, CNTs show high tendency to entangle as
ropes or in bundles. Acid treatment has been known as a cutting
process to disentangle and disintegrate the nanotubes for better
dispersion in polymer matrix, as well as to open up the closed
end of the CNTs to facilitate greater diffusitivity of gases molecules
[137,150].
Weng et al. [135] studied the effect of high MWCNT loading in
PBNBI on the gas separation performance. H
2
SO
4
/HNO
3
mixed acid
treatment resulted in the disintegration of the nanotubes into
small fragments that showed high dispersion in the PBNPI matrix,
even at high loading of 15 wt%. They observed a maximumincrease
in the diffusitivities of CO
2
and CH
4
at 15 wt% MWCNT due to the
presence of high diffusitivity tunnels in the CNTs within the poly-
mer matrix as well as the introduction of more free volume of poly-
mer chains into the MMM. This result is in contradiction with the
previous observation where the high loading of CNTs may lead to
agglomerations and create nanogaps between the two phases in
which the polymer chains become discontinuous and the tortuos-
ity around the agglomerated CNTs limits further increase in the gas
permeability [36,137]. Apart from contributing to improved inter-
facial interaction, the presence of functional group from the acid
treatment has been proven to interact with the penetrating polar
gases to improve the solubility of the gas molecules in the mem-
branes [151]. Very recently, Ge et al. [131,132] reported the in-
crease of CO
2
/N
2
gas selectivity due to the greater afnity of the
carboxyl groups on MWCNTs with polar CO
2
, which can increase
the solubility of the polar gas while hindering that of non polar
N
2
gas. This feature is of great interest as the higher solubility of
CO
2
can compensate the selectivity loss caused by the increase of
free volume upon the introduction of CNTs. Similar effect has been
observed for CNTs that were modied with rubidium metal (Ru)
[132]. When the Ru modication site was well controlled and the
RU particles were mainly deposited on the external surface of
CNTs, greater gas adsorption difference between CO
2
and N
2
on
the outer surface was induced hence altering the solubility and to-
tal gas selectivity. It was found that Ru-modied sites can increase
diffusion resistance of N
2
gas due to the stronger adsorption of the
gas molecule on the Ru impregnated external wall, and conse-
quently facilitated the enhancement of the CO
2
/N
2
selectivity.
While chemical modication via acid treatment has demon-
strated viable route to modify the surface of CNTs, many efforts
have been put forward to developing alternative physical approach
through non-covalent functionalization to overcome the draw-
backs of chemical modication particularly the damage to the
CNT structures that may eventually lead to deterioration of the
intrinsic properties. The most commonly applied non-covalent
functionalization to tailor the interfacial characteristics of CNTs
are polymer wrapping and surfactant dispersion. The former ap-
proach is achieved through the van der Waals interaction and p
p stacking between CNTs and polymer chains containing aromatic
rings to form supermolecular complexes of CNTs [134] meanwhile
the latter involves the dispersion of CNTs in the presence of surfac-
tants with different charge to exert physical adsorption of the sur-
factant molecules on the surface of CNTs hence lowers the surface
tension to effectively prevent the agglomeration and facilitate the
dispersion of CNTs [133]. Aroon et al. [134] developed MMM incor-
porated with 1 wt% of chitosan wrapped CNTs. A signicant
improvement of CO
2
and CH
4
permeability of 10.47 and 2.26 Bar-
rer, respectively and increase in CO
2
/CH
4
selectivity as much as
51% compared to neat PI were observed. The enhancement of the
separation performance was observed due to the wrapping of
chitosan molecules around the MWCNTs hence promoting very
good segmental bridging of the nanotubes and PI matrix to prevent
formation of voids and therefore improved the selectivity. On the
other hand, the open-ended chitosan functionalized CNTs acted
as tunnels within the polymer matrix to render high gas perme-
ation. Similarly, Sanip et al. [152] employed beta cyclodextrin, a
toroid with an inner cavity of several angstroms in diameters, as
wrapping polymer to non-covalently functionalize MWCNTs. The
attached hydroxyl groups upon the grinding of beta cyclodextrin
with the nanotubes played a signicant role towards the enhance-
ment of the CO
2
gas permeability by increasing the solubility of the
gas molecules through the strong interaction while passing
through the channels of CNTs. The proposed transport pathway
of the gases through MMMs with different cyclodextrin treated
MWCNT loading is depicted in Fig. 11. At high loading, it is found
that regardless of the role of cyclodextrin as the dispersing agent,
the agglomeration and saturation of CNT particles in the polymer
Fig. 11. Proposed gas transport pathway through MMMs with different loading of MWCNTs: (a) 0.5%, (b) 0.7% and (c) 1.0%. At optimum loading of 0.7%, the oriented CNT
channels provided smooth ow and penetration of gas through the polymer matrix [152].
P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264 255
matrix hindered the fast transport of gases due to increased resis-
tance through the nanotube cavities. Another non-covalent ap-
proach of CNT surface modication prior to the incorporation in
MMMs has been developed recently through surfactant dispersion
[133]. The presence of non-ionic surfactant, Triton X100 on the sur-
face of MWCNTs has reduced the agglomeration hence improved
the distribution and dispersion of the nanotubes within the PEI
matrix. As such, the Triton X100 treated MWCNT/PEI MMM has
exhibited increase in permeance of O
2
and N
2
as much as 88%
and 120% respectively without much sacricing the selectivity.
The exploration of CNTs has undoubtedly given a vivid impres-
sion on the potential of this nanostructured material as promising
ller in MMM. Therefore, CNT/MMM is expected to serve as a valu-
able mean for upgrading the separation performance of the cur-
rently existing polymeric membrane. Presently, research efforts
have been actively pursued to utilize the CNTs lled polymer ma-
trix in the eld of membrane separation, particularly in gas separa-
tion. The main interest of CNT/polymer system concerns the
smoothness of their interior channels that is the main characteris-
tics for the rapid transport of gases within the nanotubes. Despite
the viability and merits demonstrated by CNT, one point of concern
is that the success implementation of the hybrid matrix still relies
vastly on the characteristics of the CNTs as well as the interaction
with the polymer matrix. For instance, CNTs as ller has been se-
verely restricted due to the difculties with the dispersion of
entangled CNTs during processing and poor interfacial interaction
between the nanotubes and polymer matrix. Additionally, it is also
noted that the dispersion of the agglomerates is not complete due
to a stronger interaction between the primary nanoparticles than
with the polymer. The state of the CNTs dispersion in a polymer
matrix plays a decisive role for the further application of the
MMM. Therefore, a greater understanding of these issues is in
wanting to further boost the separation performance to a higher le-
vel. Another drawback of CNTs as a nanoller is their higher pro-
duction cost [153]. Therefore, the mass production of CNT-based
functional composite materials becomes very challenging.
5.2. Layered silicate
As illustrated in Fig. 12, clay is a naturally occurring layered alu-
minium silicate that present itself in the form of perfectly crystal-
line aluminium octahedral and silica tetrahedral sheets of one to a
few nanometers thick and hundreds to thousands nanometers in
extent, depending on the particular layered silicate [154]. The clay
mineral, which can be further categorized into ve main groups,
namely smectite, illite, kaolinite, chlorite or sepiolite, is a member
of the 2:1 phyllosilicates structural family, in which the central
octahedral alumina sheet is sandwiched between two tetrahedral
silica to form sheets that are piled in parallel to each other and
are bonded by local van der Waals and electrostatic forces [155].
The effort of incorporating layered silicates into polymer matrix
has been known for few decades stimulated by pronounced
improvement in the mechanical, dimensional, thermal, and barrier
performance properties, even at very low silicate layer loading
[156,157]. One of the attractive properties is the ability of the sili-
cate particles to disperse into individual layers to obtain an aspect
ratio as high as 1000 for fully dispersed individual layers, which is
hard to be realized in other poorly dispersed particles [156]. It is
generally known that layered silicates can exist in two types of
structure: tetrahedral-substituted and octahedral-substituted.
The polymer matrix can interact more readily with these than with
octahedrally-substituted material as for the latter structure, the
negative charge is located on the surface of the silicate layers.
Montmorillonite, a member of the smectite group, consists of
two fused silica tetrahedral sheets sandwiching an edge-shared
octahedral sheet of aluminium is probably the most commonly
used among the available clay minerals. Depending on the polarity
of the ller, the silicate layers can be coupled through relatively
weak dipolar and van der Waals forces or in some cases, by strong
ionic forces [158]. Montmorillonite is applied in the formation of
hybrid membrane due to their attractive properties of relatively
high cationic exchange capacity, high aspect ratio and ease of
expansion, which could facilitate the intercalation of this material
into a wide range of organic species, besides presenting relative
competitive low cost and natural abundance [157,159,160]. Along
with montmorillonite, hectorite and saponite are the layered sili-
cates that have attracted great attention in the fabrication of nano-
composites [161,162]. Upon the addition of layered silicate ller
into the polymer matrix, two idealised and somewhat simplied
polymer layered silicate nanoscale morphologies are anticipated:
(i) the intercalated structure resulted from one or two extended
polymer chains that are inserted into interlayer spaces of the lay-
Fig. 12. Structure of layered silicates [154].
256 P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264
ered clay and (ii) the exfoliated structure in which clay layers are
dispersed into the polymer matrix. In both cases, clay llers are
able to create a stable layer to the polymer and exhibit unique
properties not seen for conventionally lled polymers [163,164].
The resulting unique properties possessed by the layered sili-
cates render the nanocomposites ideal materials for products par-
ticularly for high-barrier packaging for food and beverage
packaging Generally low oxygen permeability is required for oxy-
gen sensitive food product, while for some food products certain
amount of oxygen is required [158,168]. The incorporation of lay-
ered silicate into polymer matrix exhibited remarkable molecular
barrier properties as a result of the hindered diffusion pathway
through the nanocomposite membranes. The studies on gas trans-
port properties of clay/polymer revealed that an appreciable reduc-
tion of both diffusion and solubility coefcients are responsible for
the permeability reduction of O
2
through the membrane, accred-
ited to the tortuous path towards the diffusing gas molecules and
the reduced molecular mobility of polymer chains due in the pres-
ence of layered silicate ller particles. The aspect ratio of exfoliated
silicate platelets also has a critical role in controlling the micro-
structure of polymerclay nanocomposites and their gas barrier
performances [169]. The presence of llers decreases the available
free volume in the polymer and the interactions along the clay and
polymer matrix interface thus affecting the sorption process of
more condensable gases in the polymer membrane which will pro-
vide less sorption sites for gas molecules [158]. Fig. 13 depicts the
types of mixed matrix that can be obtained when clay is embedded
in the polymer matrix. Depending on the interaction between the
two phases, three types of composites can be typically obtained,
i.e. (i) phase separation, (ii) intercalated and (iii) exfoliated [170].
Recently, a new application for clay minerals as the ller in
MMM has been demonstrated. The interesting ndings of separa-
tion performance have spurred considerate interest in the system-
atic transport study of these polymerlayered silicate
nanocomposite materials. The easily tailored pore size and compo-
sitional variability available of the layered silicate when embedded
into polymer network have shed light in applications that span
from material development to membrane separations [160].
Defontaine et al. [171] reported a drastic increment up to 145%
for CO
2
/CH
4
selectivity by sacricing the gas permeability com-
pared to the neat PDMS membrane using MMM incorporated with
Sepiolite clay minerals. Sepeolite, a layered silicate with tunnel-
like structure of approximately 3.710.6 cavity size character-
ized by the inversion of one every six SiOSi bonds in their tetra-
hedral layer to form interrupted octahedral layers. These layers
were found to be covalently embedded in the polymer networks
through the formation of SiOSi bonds between the Sepiolite par-
ticles and the polymer chains due to the presence of abundant edge
silanol groups in the Sepiolite structure thus producing a hybrid
polymeric material with inorganic and organic parts.
Despite the potential exhibited by these layered silicate mate-
rial for the application in MMM, layered silicates are usually hydro-
philic and their interactions with non-polar polymers are not
favourable. As a result, these silicate compounds must be compat-
ibilized with the polymer matrix through various forms of surface
modications prior to the incorporation to improve the miscibility
of clay with the polymer, thereby achieving good dispersion of
clays in the polymer matrix. Another urgent interest for the surface
modication is to expand the interlayer space of the clay, in order
to allow large polymer molecules to enter into the interlayer space
[167]. Modications of the nanolayer surface are generally
achieved either by surfactant cation exchange where the hydrated
cations within the clay galleries have been replaced by proper cat-
ionic surfactants (e.g. alkylammonium) by a cation exchange reac-
tion. The structure and extent of modication are found depending
greatly on the type and amount of surfactant used. For instance,
the size of the cationic head group of the surfactant, the length
and the number of alkyl chains present in the surfactant molecules
have been previously reported to play a role in the degree of exfo-
liation and intercalation [172,173]. However, these determining
parameters might be varied from one system to another depending
on the type of materials as well as the processing methods, thus
unambiguous relationship could not be established among them
[174]. One of the leading demonstrations of the gas transport prop-
erties of clay/polymer MMM has been conducted on surfactant
modied Montmorillonite KSF embedded in polyamide where
the increasing amounts of clay ller has induced depletion of O
2
and N
2
permeability but increase of the O
2
/N
2
selectivity. This is
Fig. 13. Possible layered silicate distribution arising from the interaction of the ller and polymers: (a) phase-separated, (b) interacted and (c) exfoliated [170].
P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264 257
possibly due to the barrier properties of the dispersed nanolayer in
the polyamide nanocomposite membrane that resulted in a tortu-
ous diffusion pathway for the O
2
and N
2
[175].
While surfactants have been widely used to render the clay sur-
face hydrophobic, this modication is accompanied with an unde-
sired shortcoming where it generally reduces or impedes
delamination of the clay mineral particles into thinner or single
layers [176]. Furthermore, it has also been pointed out that only
the ionic part of the surfactant could interact in a favorable way
with the charged surface of the sheet-like clay particles whereas
the long alkyl tail only shows limited compatibility with the poly-
mer chains [177]. As a consequent, the research direction has
moved towards seeking for a better compatibilization that is ex-
pected interact with the solid particle surface and with the matrix
polymer and thus cater for more promising interfacial properties
between the phases. The edge modications through the introduc-
tion of coupling agent can be accomplished by anion exchange or
silane treatment [165,166]. A later study conducted by Hashemif-
ard et al. [178] has demonstrated more interesting CO
2
/CH
4
sepa-
ration performance for both permeability and selectivity. The
incorporation of organic modied montmorillonite, Cloisite 15A
into PEI polymer matrix was found to hinder the formation of voids
by providing suitable adhesion between the clay ller and polymer
matrix. The increased selectivity of 28% relative to the neat PEI
exhibited by the clay/MMM was attributed to the high aspect ratio
of clay particles combined with the intercalation due to the pres-
ence of tallow in Cloisite 15A to render an increase of tortuosity
along the gas penetrant path across the dense skin layer. In another
study, the authors reported the 0.5% loading of silylated-halloysite
nanotubes (HCT) has resulted in the 27% CO
2
permeability incre-
ment and 8% enhancement in the CO
2
/CH
4
selectivity [179]. Hal-
loysite, a naturally occurring hydrated polymorph of kaolinite
with tubular nanostructure, was incorporated into PEI with the
aid of N-b-(aminoethyl)-c-aminopropyltrimethoxy silane
(AEAPTMS) to facilitate the tubular dispersion throughout the
MMMs and enhancing the polymer-ller adhesion. The functional-
ization of clay minerals with organosilane has been explored as a
way to improve clay dispersion in the polymer matrix, thus
increasing the intrinsic properties of the resulting clay/polymer
nanocomposites. Aside from higher degree of tortuosity that is be-
lieved to yield higher selectivity, the increase in selectivity of HNT/
PEI was also ascribed to the presence of silane agent attached to
the HNTs in the polymer matrix. Surface modication using silane
agents with amine groups is able to effectively enhance the solubil-
ity of CO
2
solubility which in turn leads to increase in CO
2
perme-
ability and selectivity. Therefore, in this study, the selectivity is
possibly contributed by the shape or size discrimination passing
between the intercalated clay layers. Another attractive layered
material that has been recently explored is JDF-L1 sheet particles
with promising characteristic for separation of H
2
containing gas
pair based on selective sieving pores size across the silicate layers
[180]. The pore size of approximately 3 has effectively offered a
favourable transport for H
2
while rendered an important barrier
action for CH
4
, O
2
and N
2
with relatively larger kinetic diameter
(>3 ), particularly when a parallel preferential orientation of the
JDF-L1 sheet particles the polymer matrix is expected.
Layered silicate clay mineral has been long used as the reinforc-
ing phase due to its easy availability, low cost and more impor-
tantly environmentally friendly. However, the application of clay
mineral is still new in MMM. The transport properties of clay/poly-
mer MMM have been scarcely reported and only a few scientic
publications are available. Therefore, the corresponding knowledge
about this process is still incomplete. However, the potential of this
material is undeniable and this predestines the clay minerals as
promising ller in MMM, which demand a simultaneous improve-
ment in both selectivity and permeability for gas separation.
Remarkably, these primary results have suggested that there lies
great potential for further improvement through more feasible
structural modication and optimization of the clay mineral. Nev-
ertheless, more research remains to be carried out in order to fully
understand the relationship between the clay structures and trans-
port interaction.
5.3. Metal organic framework
Metal organic framework (MOF), a hybrid of organic and inor-
ganic porous materials, is essentially coordination polymer formed
in the most elementary sense by connecting metal ions with organ-
ic linkers to obtain fascinating tuneable pore geometries and ex-
ible frameworks that opens a new avenue to be efciently applied
in many potential applications [181]. Generally, MOFs are built up
of inorganic sub-units in the forms of clusters, chains, layers or
three dimensional arrangements that are connected to organic
linkers possessing complexing groups such as carboxylates, phos-
phonates, N-containing compounds by strong ionocovalent or da-
tive bonds [182]. MOFs can be made up from dimers, trimers,
tetramers or polyhedral chains in which the linkers are readily
modied with various organic groups while maintaining the same
structure type in order to preserve the desired properties. Fig. 14
illustrates a few commons MOFs with different structure architec-
tures that could lead to various forms of frameworks and porosity
[182]. MOFs are generally classied into rigid and dynamic. The
former do not exhibit any exibility behaviour with small changes
in volume reecting the content of their pores while the latter dis-
plays great exibility and their applications depend not only on the
size of the aperture, but also on adsorption conditions. The pores in
MOFs have a very uniform distribution and unlike the spherical or
slit-shaped pores that are usually observed in zeolites or activated
carbons, MOFs incorporate pores with crystallographically well-
dened shapes including square which may lead to different siev-
ing properties [183186]. Like other porous molecular sieves, the
selective gas adsorption in MOFs is achieved mainly based on the
following principle mechanisms which are capable of working
independently as well as cooperatively in both rigid and exible
MOFs: (i) adsorbatesurface interactions that involves the chemi-
cal and/or physical interaction between the adsorbent and the
adsorbate and (ii) size-exclusion that depends on the dimension
and shape of the framework pores [187].
The motivation for incorporating MOFs in MMM could be
numerous. Some of the exhibited characteristics are high surface
areas which contribute to high sorption capacity, controlled poros-
ity, high afnity for certain gases and high afnity with the poly-
mer chains due to the presence of the organic linkers in MOFs, as
well as the high exibility in terms of chemical composition which
allow the changes with systematically varying pore size, function-
alities and chemical properties of the MOFs through tailored syn-
thesis approaches in order to facilitate the selective binding or
adsorption of target gases [188,189]. MOFs with ultra large pores
(>1 nm) might allow penetration of polymer chains to facilitate
the adhesion of polymer chain to the embedded MOF particles.
Molecular dynamics simulations conducted on self-diffusion and
transport diffusion of small molecules in MOFs pointed out that
diffusivities in MOFs are similar to those in zeolites and the knowl-
edge of molecular diffusivities in MOFs might be used to screen
MOFs for use in an identied process [190]. It is found that descrip-
tion based on Knudsen diffusion is insufcient to describe diffusion
in MOFs with relative larger pores as the large open volume inside
MOFs frameworks are only lled with molecules at very high pres-
sure. Thus, it is more appropriate to conclude that diffusion of mol-
ecules in MOFs is dominated by motions where the adsorbed
species remain in close contact with the surfaces dened by the
pore structure throughout their diffusion.
258 P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264
Aside from the preliminary understanding towards the charac-
teristic and transport behavior of MOFs, the study and the interest
for these materials are still at an initial stage and answers are
highly desirable for many unanswered questions. One of funda-
mental challenges to fabricate MOF/polymer MMMs that are tech-
nologically attractive is to select appropriate MOF/polymer pairs as
well as the MOF that are selective towards certain gases. These cri-
teria appear essential for high performance MMM as there are
enormous numbers of MOFs with a broad range of sieving proper-
ties that could potentially be used as the dispersed ller phase in a
polymer matrix. Keskin and Sholl [191] adopted Maxwell model to
predict mixture permeation for CO
2
/CH
4
mixtures in MOF/Matri-
mid MMM using molecular simulations and mixing theories. The
prediction showed that the incorporation of MOF which is either
a permeable but unselective MOF like MOF-5 or the high selective
one such as Cu(hpbb)(H
2
hpbb)
0.5
can impart very inuential im-
pact on the separation performance of the resulting MOF/MMMs.
Specically, for polymer that is very permeable but less selective,
identity of high selective MOF as ller is of great importance to ob-
tain a MMM that surpass the trade-off curve for a relevant gas sep-
aration. The trend shown in Fig. 15 provides further prediction of
different combinations of MOF and polymer that would offer indis-
pensable new knowledge towards the rational identication of the
most appropriate MOF/polymer pairs to achieve CO
2
/CH
4
separa-
tion with high gas selectivity and permeability.
The pioneer attempt of incorporating MOF copper (II) biphenyl
dicarboxylate-triethylenediamine as ller in MMMs to facilitate
methane transport was reported by Yehia et al. [192] in which
the CH
4
selectivity has shown encouraging improvement that fur-
ther implied that MOFs can be served as a suitable candidate as the
disperse phase for gas separation due to their unique sieving prop-
erties. MOF-5, a cubic three-dimensional structure made from
Zn
4
O clusters linked by three 1,4-benzenedicarboxylate molecules
have been assembled into thin lm membrane to promote its
Fig. 14. Few commons MOFs with different structure architectures that could lead to various forms of frameworks and porosity: (a) HKUST-1, (b) MOF5, (c) Sodalite-ZMOF
and (d) zeolite-rho ZMOF [182].
Fig. 15. Prediction of different combinations of MOF and polymer to achieve CO
2
/
CH
4
separation with high gas selectivity and permeability [191].
P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264 259
usage in gas applications and proven to offer unique opportunities
with a potential to achieve superior performance [193,194]. Perez
et al. [43] studied the gas separation performance of MOF-5/Matri-
mid and observed increased permeability up to 120% while the
ideal selectivity remained constant compared to that of neat poly-
mer at 30% MOF-5 loading. Nevertheless, Matrimid incorporated
with MOF-5 has exhibited a remarkable increase in selectivity for
CH
4
in gas mixture of CO
2
/CH
4
and N
2
/CH
4
due to the coupling ef-
fect of favourable MOF-5 afnity towards CH
4
and larger solubility
of CO
2
and N
2
in Matrimid. The transport mechanism is not solely
based on solubility and diffusitivity in the polymer but also the
Knudsen and surface diffusion as facilitated by the MOF-5 porosity
as well as by the uniformity of the surface of its wall. Another type
of MOF that shows high afnity towards CH
4
and favours the per-
meation of CH
4
is Cu-4,4
0
-bipyridine-hexauorosilicate (Cu-BPY-
HFS) which is built up from a square network of copper bipyridyl
complexes that are pillared by SiF
6
2
ions [195]. Zhang et al.
[196] reported a profound increment of CH
4
/N
2
gas mixture selec-
tivity from 0.95 to 1.7 attributed to the high surface area and afn-
ity of Cu-BPY-HFS in the MMM towards CH
4
that resulted in the
increased solubility and hence selectivity toward CH
4
.
Recently, a new class of MOF with exceptional chemical stabil-
ity, zeolitic imidazolate framework (ZIF) with tetrahedral network
and sodalite cagelike structure that resemble the structure type of
zeolite has been identied as an attractive molecular sieve for
small gas molecules such as H
2
and CO
2
[67,182,197]. The addition
of ZIF in MMM is suitable to accomplish such purpose as the pore
window size of these MOFs can be precisely tuned for a desired gas
pair separation. ZIF-8 and ZIF 90, with pore apertures of 3.4 and
3.5 respectively can readily exclude gas molecules with larger
size from small H
2
and CO
2
hence demonstrating promising results
in H
2
/CH
4
and CO
2
/CH
4
gas pair separation. In addition, the pres-
ence of imidazole linker in these ZIF members is also a contributing
factor to the enhanced selectivity as this ligand has been found to
favorably interact chemically with CO
2
through non-covalent
bonding [198]. Bae et al. [188] have reported the performance of
ZIF 90/6FDA-DAM MMM that surpassed the trade-off of polymeric
membrane performance in which the enhancement in both perme-
ability and selectivity was mainly due to the selective sorption and
diffusion of CO
2
in the ZIF-90 crystal. In another similar study con-
ducted by Ordonez et. al. [197] using ZIF-8/Matrimid MMM, signif-
icant increase in the gas permeability was observed at ZIF-8
loading lower than 50 w/w% due to the presence of MOF llers that
has increased the distance between polymer chains and disrupted
chain packing in the polymer thus showing an increase in the free
volume. A remarkably enhanced selectivity for gas pairs containing
small gas molecules was reported at 50% ZIF loading. The incre-
ment of 213% and 290% for CO
2
/CH
4
and H
2
/CH
4
respectively was
ascribed to the molecular sieving effect of the small pore aperture
of ZIF-8 that favors the diffusion of small gas while hindering the
easy diffusion of large gas molecules like CH
4
. As such, ZIFs facili-
tate the transport of small molecules through the aperture and into
the pore, competitive adsorption for gas pair contains both small
molecules cannot be realized by these ZIFs. In such circumstances
where separation of small molecules like H
2
/CO
2
is desired, ZIF
with narrower pore size of about 3 , which is between the kinetic
diameter of H
2
(2.9 ) and CO
2
(3.3 ), is of particular interest as it
is expected to show a clear cut-off between these two gases and
thus can be served as an attractive ller candidate for MOF based
MMMs. Li et al. [67] investigated the potential application of ZIF-
7 incorporated polyetherimide MMM for H
2
separation and notice-
ably improved separation performance that transcends the trade-
off curve was achieved primarily due to the precise size-selective
molecular sieving abilities exhibited by ZIF-7.
The gas transport properties of MOF/MMMs have so far been
found to be technologically attractive. MOFs are very promising
candidates that stands out for selective gas adsorption, which
can lead to gas separation due to its remarkable selectivity and
tuneable properties. Furthermore, MOFs can be easily synthesized
and variation of MOF compositions and structures including high
aspect ratio MOFs is achievable. Organic linkages provide useful
platform for chemical reaction to take place that may improve
adhesion. Unfavourably, the main constraints of MOF when
embedded in polymer matrix are lower thermal and chemical sta-
bility due to the presence of the organic linkers. In this aspect, they
are still far to compete with traditional llers such as zeolites and
CMS. Furthermore, thermal stability of the polymers can be ob-
tained when high temperature is applied to the matrix during sol-
vent removal. These drawbacks must be balanced against the
rational design process for tailoring structures to specic proper-
ties. However, one should realize that, the true strength of MOF
springs directly from the modular process by which they are con-
structed as they allow greater chemical alteration on a periodic
scale, since the methodology for organic transformations is well-
established [199]. So far, the studies performed have underlined
the key role played by the structure of MOFs that are able to offer
selective separation based on their sorption capability and pore
window size. Such studies have also added further insight into
the dependency of the separation performance on the structure
architecture in terms of the pore size and rigidity where emphasis
can be given during the construction of MOF based MMMs.
6. New and emerging material for MMM
6.1. Graphene
Graphene, as new class of carbon nanomaterials, is found to be
economical and has novel properties similar to CNTs. Fig. 16 com-
pares the existing low-dimension carbon allotrope i.e. fullerene,
CNT and grapheme [200]. Since the early 1940s, the experimental
investigation of graphene properties, as a standalone object, has
been almost inexistent until recent years due the difculty to iden-
tify and univocally characterize the single-atomthick sheet. The di-
rect observations of isolated graphene monolayer recently have
sparked an exponentially growing interest in this two-dimensional
material. Recent studies showed that graphene could be used as a
viable and inexpensive ller substitute for CNTs in nanocomposites
owing to the excellent in-plane mechanical, structural, thermal and
electrical properties of graphite [201]. Nowadays, graphene and
graphene-based nanocomposites are being studied in nearly every
eld of science and engineering and possess much importance due
to their exceptional electronic and mechanical properties [202
205]. Recent progress has shown that the graphene based materials
can have a profound impact on the gas transport. The investigation
and tailoring of its transport properties frommacroscopic to molec-
ular scales have now captured the attention of many researchers.
Graphite can be a good candidate as ller in MMMs due to its
high aspect ratio [206,207]. Permeation rate of gas molecules dif-
fusing through membranes can be decreased by embedding high
aspect ratio, impermeable particles that provide tortuous paths
and reduce the cross sectional area available for permeation
[208]. N
2
and He permeation through PC lms reinforced with
graphite and functionalized graphene sheets (FGS) were studied
and the results showed that FGS appears to be slightly better for
blocking the diffusion of N
2
[209]. On the FGS surfaces, there can
be atomistic perforations that CO
2
evolution left during pyrolysis
treatments, which more likely allow penetration of smaller mole-
cules such as He than larger ones [210]. Furthermore, increased
free volume near the graphene-PC interface may selectively expe-
dite transmission of smaller permeants [211]. As such, N
2
and He
permeability could be reduced by incorporation of graphite and
260 P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264
FGS. Interestingly, FGS turns out to be a more effective barrier
against diffusing molecules with a larger kinetic diameter implying
its potential for materials for gas separation.
The use of graphene in different applications is rapidly emerg-
ing in current scientic research. Cost effective production of
graphene sheets and the potential in gas transport provides a plat-
form for the development of commercially feasible MMMs using
this novel material as ller. However, the effective use and exper-
imental demonstration of pure grapheme in MMM for gas separa-
tion still remain a challenge. Like CNT and other nanomaterials, the
key challenge in synthesis and processing of bulk-quantity graph-
ene sheets is the formation of aggregates that hinder complete and
homogeneous dispersion of individual graphene sheets in various
solvents. By solving this issue, the new concepts of graphene lled
MMMs for desired gas separation can soon be realized.
7. Conclusion and future direction
At present, the existing membrane materials are inadequate at
fully exploiting new opportunities in the gas separation applica-
tion. The hybrid system in MMM have the potential to address
these requirements by synergistically combining the processing
versatility of polymers with the separation characteristics of inor-
ganic molecular sieves and has been the subject of numerous re-
searches. The interplay between basic science and engineering
has resulted in some of the most exciting accomplishments in
the development of MMM. It is believed that this emerging ap-
proach is feasible to attain desired features from both phases and
overcome the bottlenecks in each medium.
In the last decades, the studies have asserted the unprecedented
rapid growth in the eld of MMM. MMM for gas separation are
conventionally comprised of zeolite, CMS and silica particles
embedded in polymer matrix with the ultimate goal of surpassing
the Robeson trade-off boundary by overcoming the drawbacks of
the existing organic and inorganic membrane materials. The past
decades have witnessed an intense research effort on the applica-
tion of these llers in various kinds of glassy and rubbery poly-
meric materials to perform promising gas separation
performance. Unlike the transport mechanism in polymeric mem-
branes which is depending largely on the transient intersegmental
gaps to effectively separate the gas molecules, inorganic llers
with molecular sieving properties are known to be beneted by
the specic pore diameters that fall in between the kinetic diame-
ters of the target gas molecules, resulting in a large diffusive selec-
tivity to the overall selectivity. The typical llers that behave in this
attractive manner are zeolites and CMS that have been unsurpris-
ingly granted with tremendous attention since the rst exploration
of MMM in the area of membrane based separation. Nonporous sil-
ica is a class of impermeable ller that have attracted great interest
rooted from their great potential to affect polymer chain packing in
polymers to alter the gas separation properties. Metal oxides such
as MgO and TiO
2
are nanoparticles that have been studied based on
their potential to be used as llers in MMM. These nanoparticles
normally exhibit high afnity towards certain gas molecules based
on their surface interactions, hence wield the advantages of facili-
tated transport for gas separation through the adoption of appro-
priate strategy for design and fabrication of the MMM. While the
exploration of MMM incorporated with these llers have brought
about some forms of delighting breakthrough to be potentially
used in separation processes, most of the contemporary studies
have concurrently pointed out that there is an obstacle to the suc-
cessful introduction of inorganic molecular sieve materials into the
polymer matrix mainly due to poor compatibility between the ll-
ers and polymer matrix, especially when rigid glassy polymer is
used as the continuous phase. The interfacial region, which is a
transition phase between the polymer and inorganic ller, is of
particular importance to crucially determine the separation perfor-
mance of the MMM. Of importance is the issues related to weak
interactions between a glassy polymer matrix and inorganic
molecular sieves that may lead to the formation of nonselective
voids and resulting in undesired Knudsen ow [39]. Therefore, in-
stead of seeking perfect matching of the inorganic llers and poly-
mer matrix to battle the weakness of MMM in term of adhesion
and interaction between the two phases as broadly focused during
the early studies, recent efforts have pinpointed the necessity to
modify the surface of llers to enhance the interfacial properties.
Modication of llers and polymer matrix has become an expand-
ing eld of research and different approaches have been proposed
to account for the limitations that originate from the poor interac-
tions between both polymers and llers.
Although early studies were mainly focused on the incorpora-
tion of conventional inorganic llers, particularly zeolites with
Fig. 16. Comparison of carbon allotropes with different dimension: fullerene (0-dimension), CNT (1-dimension) and graphene (2-dimension) [200].
P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264 261
molecular sieving properties; the research trend in the last ve
years has shifted towards the utilization of various potential nano-
structured ller particles such as CNTs, MOFs and clay. These new-
ly explored llers have shown to be interesting materials as a
competitive alternative for MMM. The need to further explore
the development of MMM by using alternative technologies has
pushed these nanomaterials to the forefront of applications due
to their outstanding gas separation properties. On the other hand,
graphene, a one-atom-thick planar sheet of sp
2
-bonded carbon
atoms densely packed in a honeycomb crystal lattice have gener-
ated enormous interest for its possible implementation in the
preparation of MMM. There is no doubt that graphene has shed
some light for a new research dimension in searching for new
materials.
Even after decades of extensive research, the potential of
employing these inorganic llers have not reached the expectation
of current gas separation performance. This may be due to several
reasons, such as membrane defects and related processing issues,
as well as the unfavourable orientation of llers in the MMM. It
has become clear that improving the interaction between llers
and polymer matrix is the most important criterion towards prac-
tical usage of MMM in gas separation. The introduction of func-
tional groups can improve dispersion of llers and change
chemical afnities of penetrants in nanocomposite membranes.
This enhancement is proposed to originate from the strong interac-
tions between the llers and the polymer matrix. The previous
studies have also revealed that, the transport properties in terms
of the permeability and selectivity can be further optimized
through the detail governing of some inuential key parameters.
For this reason, the development of ller materials with controlled
properties through various forms of surface functionalization has
been a subject of great research challenge to membrane-based sep-
aration processes.
MMM consist of inorganic llers incorporated into a polymer
matrix have the potential to extend or even surpass the separation
benets exhibited by the polymeric membrane platform. The un-
ique behaviour found in this hybrid system will allow us to apply
it in new and different applications. The membrane technology re-
search community has made great progress in the last decade, yet
the currently gained knowledge in respect to the application po-
tential of these llers in the fabrication of MMM may just the tip
of the iceberg. Further understanding of the fundamentals of this
material, including structure and dynamics, will allow us to pro-
duce materials avoiding the trade-offs of conventional systems.
The main concern of future research should be dedicated to the uti-
lization of specic molecular interactions to control structure and
morphology and prediction from nanostructural data and molecu-
lar dynamics simulations. The future of the eld is indeed very
bright and breakthroughs in the near future are possible with the
continual development of these applications. Through the predic-
tion of the materials properties or maximum theoretical perfor-
mance for different classes of llers, many new types of llers
are expected to emerge as the research topic becomes more and
more popular and there are still multitudes of structures waiting
to be discovered. Hybrid materials with organicinorganic or
bioinorganic character represent not only a new eld of basic re-
search but also, via their remarkable new properties and multi-
functional nature, offer great prospects for many new
applications in extremely diverse elds.
Acknowledgement
The authors would like to express gratitude to Ministry of Sci-
ence, Technology and Innovation Malaysia for the nancial support
on the work undertaken in this research centre.
References
[1] R. Baker, Membr. Technol. 138 (2001) 510.
[2] D.W. Wallace, C. Staudt-Bickel, W.J. Koros, J. Membr. Sci. 278 (2006) 92104.
[3] A. Javaid, Chem. Eng. J. 112 (2005) 219226.
[4] W.S. Winston Ho, K.K. Sirkar, Chapter 1: overview, in: W.S. Winston Ho, K.K.
Sirkar (Eds.), Membrane Handbook, Chapman & Hall, New York, London,
1992, pp. 315.
[5] M. Mulder, Chapter 1: introduction, in: M. Mulder (Ed.), Basic Principle of
Membrane Technology, Kluwer Academic Publisher, Dordrecht/Boston/
London, 1991, pp. 115.
[6] W. Eykamp, Chapter 1: Microltration and ultraltration, in: R.D. Noble, S.A.
Stern (Eds.), Membrane Science and Technology Series, 1, Membrane
Separations Technology: Principles and Applications, Elsevier Science B.V.,
1995, pp. 140.
[7] R.W. Baker, Chapter 1: overview of membrane science and technology, in:
R.W. Baker (Ed.), Membrane Technology and Applications, 2nd Ed., John Wiley
& Sons Ltd., 2004, pp. 114.
[8] D.Q. Vu, W.J. Koros, S.J. Miller, J. Membr. Sci. 211 (2003) 311334.
[9] P. Pandey, R.S. Chauhan, Prog. Polym. Sc. 26 (2001) 853893.
[10] D. Sieffert, C. Staudt, Sep. Purif. Technol. 77 (2011) 99103.
[11] D. Wang, K. Li, W.K. Teo, J. Membr. Sci. 204 (2002) 247256.
[12] V.V. Namboodiri, L.M. Vane, J. Membr. Sci. 306 (2007) 209215.
[13] C.C. Periera, R. Nobrega, K.-V. Peinemann, C.P. Borges, J. Membr. Sci. 226
(2003) 3550.
[14] S. Clemenson, E. Espuche, L. David, D. Leonard, J. Membr. Sci. 361 (2010) 167
175.
[15] L.M. Robeson, J. Membr. Sci. 320 (2008) 390400.
[16] H.M. van Veen, P. van Male, M.A.G. Vorstman, J.T.F. Keurentjes, Sep. Purif.
Technol. 2223 (2001) 689695.
[17] N. Widjojo, T.-S. Chung, Ind. Eng. Chem. Res. 45 (2006) 76187626.
[18] T. Pietra, MRS Bull. 31 (2006) 765769.
[19] Y. Li, H. Zhou, G. Zhu, J. Liu, W. Yang, J. Membr. Sci. 297 (2007) 1015.
[20] K. Li, Ceramic Membrane for Separation and Reaction, Wiley, Chichester,
England, 2007.
[21] S. Husain, W.J. Koros, Ind. Eng. Chem. Res. 48 (2009) 23722379.
[22] T.-S. Chung, L.Y. Jiang, Y. Li, S. Kulprathipanja, Prog. Polym Sci 32 (2007) 483
507.
[23] C.M. Zimmerman, A. Singh, W.J. Koros, J. Membr. Sci. 137 (1997) 145154.
[24] H. Cong, M. Radosz, B.F. Towler, Y. Shen, Sep. Purif. Technol. 55 (2007) 281
291.
[25] M.A. Aroon, A.F. Ismail, T. Matsuura, M.M. Montazer-Rahmati, Sep. Purif.
Technol. 75 (2010) 229242.
[26] R. Pal, J. Colloid Interface Sci. 317 (2008) 191198.
[27] T.T. Moore, R. Mahajan, D.Q. Vu, W.J. Koros, AiChE J. 50 (2) (2004) 311321.
[28] A. Tena, L. Fernandez, M. Sanchez, L. Palacio, A.E. Lozana, A. Herhandez, P.
Pradanos, Chem. Eng. Sci. 65 (2010) 22272235.
[29] G.L. Jadav, P.S. Singh, J. Membr. Sci. 328 (2009) 257267.
[30] J.R. Johnson, W.J. Koros, J. Taiwan, Inst. Chem. Eng. 40 (2009) 268275.
[31] L.Y. Jiang, T.S. Chung, S. Kulprathipanja, J. Membr. Sci 276 (2006) 113125.
[32] N. Widjojo, T.-S. Chung, S. Kulprathipanja, J. Membr. Sci. 325 (2008) 326.
[33] T.-S. Chung, S.S. Chan, R. Wang, Z. Lu, C. He, J. Membr. Sci. 211 (2003) 9199.
[34] D. Sen, H. Kalipcilar, L. Yilmaz, J. Membr. Sci. 303 (2007) 194203.
[35] S.A. Hashemifard, A.F. Ismail, T. Matsuura, J. Membr. Sci. 359 (2010) 259268.
[36] H. Cong, J. Zhang, M. Radosx, Y. Shen, J. Membr. Sci. 294 (2007) 178185.
[37] R. Mahajan, W.J. Koros, Ind. Eng. Chem. Res. 39 (2000) 26922696.
[38] B.D. Reid, A. Ruiz-Trevino, I.H. Musselman, K.J. Balkusss, J.P. Ferraris, Chem.
Mater. 13 (2001) 23662373.
[39] R. Mahajan, W.J. Koros, Polym. Eng. Sci. 42 (7) (2002) 14201431.
[40] R. Mahajan, W.J. Koros, Polym. Eng. Sci. 42 (7) (2002) 14321441.
[41] S. Shu, S. Husain, W. Koros, J. Phys. Chem. C. 111 (2007) 652657.
[42] K.J. Balkus Jr., K. Cattanach, I.H. Musselsman, J.P. Ferraris, Mater. Res. Soc.
Symp. Proc. 752 (2003) 9196.
[43] E.V. Perez, K.J. Balkus Jr., J.P. Ferraris, I.H. Musselman, J. Membr. Sci. 328
(2009) 165173.
[44] D. Sen, H. Kalipcilar, L. Yilmaz, Desalination 200 (2006) 222224.
[45] S.B. Tantekin-Ersolmaz, C. Atalay-Oral, M. Tather, A. Erdem-Senatalar, B.
Schoeman, J. Sterte, J. Membr. Sci. 175 (2000) 285288.
[46] T.W. Pechar, S. Kim, B. Vaughan, E. Marand, M. Tsapatsis, H.K. Jeong, C.J.
Cornelius, J. Membr. Sci. 277 (2006) 195202.
[47] S. Husain, W.J. Koros, J. Membr. Sci. 288 (2007) 195207.
[48] P. Gorgojo, B. Zornoza, S. Uriel, C. Telelz, J. Coronas, Stud. Surf. Sci. Catal. 174
(2008) 653656.
[49] T.W. Pechar, S. Kim, B. Vaughan, E. Marand, V. Baranauskas, J. Rife, H.K.
Jeong, M. Tsapatsis, J. Membr. Sci. 277 (2006) 210218.
[50] C. Zeng, L. Zhang, X. Cheng, H. Wang, N. Xu, Sep. Purif. Technol. 63 (2008)
628633.
[51] Y. Li, T.-S. Chung, C. Cao, S. Kulprathipanja, J. Membr. Sci. 260 (2005) 4555.
[52] G. Clarizia, C. Algieri, A. Regina, E. Drioli, Microporous Mesoporous Mater. 115
(2008) 6774.
[53] H. Wang, B.A. Holmberg, Y. Yan, J. Mater. Chem. 12 (2002) 36403643.
[54] M. Jia, K.V. Peinemann, R.D. Behling, J. Membrane, Sci. 57 (2005) 289292.
[55] G. Clarizia, C. Algieri, E. Drioli, Polymer 45 (2004) 56715681.
[56] F.A. Ruiz-Trevino, D.R. Paul, J. Appl. Polym. Sci. 66 (1997) 19251941.
[57] N.M. Larocca, L.A. Pessan, J. Membr. Sci. 218 (2003) 6992.
262 P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264
[58] Y. Li, H.-M. Guan, T.-S. Chung, S. Kulprathipanja, J. Membr. Sci. 275 (2006) 17
28.
[59] D.H. Flinn, D.A. Guzonas, R.H. Yoon, Colloids Surf. A: Physicochem. Eng.
Aspects 87 (1994) 163176.
[60] K.C. Vrancken, K. Possemiers, P. Van Der Voort, E.F. Vansant, Colloids Surf, A:
Physicochem. Eng. Aspects 98 (1995) 235241.
[61] S. Zanardi, A. Alberti, G. Cruciani, A. Corma, V. Forns, M. Brunelli, Angew.
Chem. Int. Ed. 43 (2004) 49334937.
[62] T.T. Moore, W.J. Koros, J. Molec. Struct. 739 (2005) 8798.
[63] Y.C. Hudiono, T.K. Carlisle, A.L. LaFrate, D.L. Gin, R.D. Noble, J. Membr. Sci. 370
(2011) 141148.
[64] E. Karatay, H. Kalipcilar, L. Yilmaz, J. Membr. Sci. 364 (2010) 7581.
[65] A.L. Khan, A. Cano-Odena, B. Gutierrez, C. Minguillon, I.F.J. Vankelecom, J.
Membr. Sci. 350 (2010) 340346.
[66] A.M.W. Hillock, S.J. Miller, W.J. Koros, J. Membr. Sci. 314 (2008) 193199.
[67] Y. Li, F. Liang, H. Bux, W. Yang, J. Caro, J. Membr. Sci. 354 (2010) 4854.
[68] C.-C. Hu, T.-C. Liu, K.-R. Lee, R.-C. Ruan, J.-Y. Lai, Desalination 193 (2006) 14
24.
[69] B. Zorrnoza, O. Esekhile, W.J. Koros, C. Tellez, J. Coronas, Sep. Purif. Technol. 77
(2011) 137145.
[70] K. Vanherck, A. Aerts, J. Martens, I. Vankelecom, Chem. Commun. 46 (2010)
24922494.
[71] N. Navascues, C. Tellez, J. Coronas, Micropor. Mesopor. Mater. 112 (2008)
561572.
[72] T.-H. Weng, H.-H. Tseng, M.-Y. Wey, Int. J. Hydrogen Energy 35 (2010) 6971
6983.
[73] A.F. Ismail, L.I.B. David, J. Membr, Sci. 193 (2001) 118.
[74] J.E. Keresh, A. Soffer, J. Chem. Soc. Faraday Trans. I 76 (1980) 24572471.
[75] J.E. Keresh, A. Soffer, J. Chem. Soc. Faraday Trans. I 76 (1980) 24722485.
[76] D.Q. Vu, W.J. Koros, S.J. Miller, J. Membr. Sci. 221 (2003) 233239.
[77] M. Das, J.D. Perry, W.J. Koros, Ind. Eng. Chem. Res. 49 (2010) 93109321.
[78] D.Q. Vu, W.J. Koros, S.J. Miller, J. Membr. Sci. 211 (2003) 335348.
[79] B. Shimekit, H. Muktar, S. Maitra, J. Appl. Sci. 10 (12) (2010) 12041211.
[80] J.M. Duval, B. Folkers, M.H.V. Mulder, G. Desgrandchamps, C.A. Smolders, J.
Membr. Sci. 80 (1) (1993) 189198.
[81] J. Liu, H. Wang, L. Zhang, Chem. Mater. 16 (2004) 42054207.
[82] W.W.W. Razah, A.F. Ismail, J. Membr. Sci. 307 (2008) 5361.
[83] R.B. Guan, Y.G. Yang, J.T. Zhen, New Carbon Mater. 17 (2002) 4951.
[84] U. Zielke, K.J. Httinger, W.P. Hoffman, Carbon 32 (1996) 10151026.
[85] A. Bismarck, D. Richter, C. Wuertz, Colloids Surf. A 159 (1999) 341350.
[86] C. Joly, S. Goiset, J.C. Schrotter, J. Sanchez, M. Escoubes, J. Membr. Sci. 130
(1997) 6374.
[87] M. Sadeghi, M.A. Semsarzadeh, H. Moadel, J. Membr. Sci. 331 (2009) 2130.
[88] J. Ann, W.-J. Chung, I. Pinnau, M.D. Guiver, J. Membr. Sci. 314 (2008) 123133.
[89] H. Cong, X. Hu, M. Radosz, Y. Shen, Ind. Eng. Chem. Res. 46 (2007) 25672575.
[90] M.C. Ferrari, M. Galizia, M.G. De Angelis, G.C. Sarti, Ind. Eng. Chem. Res. 49
(2010) 1192022935.
[91] D. Gomes, S.P. Nunes, K.-V. Peinemann, J. Membr. Sci. 246 (2005) 1325.
[92] K. Kusakabe, K. Ichiki, J.-I. Hayashi, H. Maeda, S. Morroka, J. Membr. Sci. 115
(1996) 6575.
[93] M. Moadded, W.J. Koros, J. Membr. Sci. 111 (1996) 283290.
[94] B. Zornoza, C. Tellex, J. Coronas, J. Membr. Sci. 368 (2011) 100109.
[95] T.C. Merkel, Z. He, I. Pinnau, Macromolecules 36 (2003) 68446855.
[96] T.C. Merkel, B.D. Freeman, R.J. Spontak, Z. He., I. Pinnau, P. Meakin, A. J. Hill.
Chem. Mater. 15 (2003) 109123.
[97] J.H. Kim, Y.M. Lee, J. Membr. Sci. 193 (2001) 209225.
[98] C.J. Cornelius, E. Marand, J. Membr. Sci. 202 (2002) 97118.
[99] J. Ahn, W.-J. Chung, I. Pinnau, J. Song, N. Du, G.P. Robertson, M.D. Guiver, J.
Membr. Sci. 346 (2010) 280287.
[100] M. Moaddeb, W.J. Koros, J. Membr. Sci. 125 (1997) 143163.
[101] S. Takahashi, D.R. Paul, Polymer 47 (2006) 75357547.
[102] S. Takahashi, D.R. Paul, Polymer 47 (2006) 75197534.
[103] J. Joseph, C.-Y. Tseng, C.-J. Pan, H.-M. Chen, C.-W. Lin, K.S. Pillai, B.-J. Hwang,
Polymer 51 (2010) 56635668.
[104] Y.-Y. Yu, W.-C. Chien, T.-W. Tsai, Poly. Test. 29 (2010) 3340.
[105] H. Kim, H.-G. Kim, S. Kim, S.S. Kim, J. Membr. Sci. 344 (2009) 211218.
[106] W. Posthumus, P.C.M. Magusin, J.C.M. Brokken-Zijp, A.H.A. Tinnemans, A.
vander Linde, J. Colloid Interface Sci. 269 (2004) 109116.
[107] Y. Kwon, H. Im, J. Kim, Sep. Purif. Technol. 78 (2011) 281289.
[108] Y. Sun, Z. Zhang, C.P. Wong, J. Colloid, Interface Sci 292 (2005) 436444.
[109] M. Sadeghi, G. Khanbabaei, A.H.S. Dehaghani, M. Sadeghi, M.A. Aravand M.A,
Kbarzade, S. Khatti, J. Membr. Sci. 322 (2008) 423428.
[110] R. Xing, W.S.W. Ho, J. Membr. Sci. 367 (2011) 91102.
[111] T.-L. Chew, A.L. Ahmad, S. Bhatia, Adv. Colloid Interface Sci. 153 (2010) 4357.
[112] R. Brandy, B. Woonton, M.L. Gee, A.J. OConnor, Innovative Food Sci. Emerg.
Technol. 9 (2008) 243248.
[113] B. Zornoza, S. Irusta, C. Tellez, J. Coronas, Langmuir 25 (10) (2009) 5903
5909.
[114] S. Kim, E. Marand, Chem. Mater. 18 (2006) 11491155.
[115] S. Kim, E. Marand, Mircoporous Mesoporous Mater. 114 (2008)
129136.
[116] N. Nishiyama, D.H. Park, A. Koide, Y. Egashira, K. Ueyama, J. Membr. Sci. 182
(2001) 235.
[117] S.S. Hosseini, Y. Li, T.-S. Chung, Y. Liu, J. Membr. Sci. 302 (2007) 207217.
[118] S. Matteucci, V.A. Kusuma, S.D. Kelman, B.D. Freeman, Polymer 49 (2008)
16591675.
[119] S. Matteucci, R.D. Raharjo, V.A. Kusuma, S. Swinnea, B.D. Freeman,
Macromolecules 41 (2008) 21442156.
[120] S. Matteucci, V.A. Kusuma, D. Sanders, S. Swinnea, B.D. Freeman, J. Membr.
Sci. 307 (2008) 196217.
[121] S. Matterucci, V.A. Kusuma, S. Swinnea, B.D. Freeman, Polymer 49 (2008)
757773.
[122] F. Moghada, M.R. Omidkhah, E. Vasheghani-Farahani, M.Z. Pedram, F. Dorosti,
Sep. Purif. Technol. 77 (2011) 128136.
[123] A. Aqel et al., Carbon nanotubes, science and technology part (I) structure,
synthesis and characterisation. Arabian Journal of Chemistry (2010),
doi:10.1016/j.arabjc.2010.08.022.
[124] S. Bredeau, S. Peeterbroack, D. Bonduel, M. Alexandre, P. Dubois, Polym. Int.
57 (2008) 547553.
[125] A.V. Desai, M.A. Haque, Thin-walled Struct. 43 (2005) 17871803.
[126] S.V. Ahir, Y.Y. Huang, E.M. Terrentjev, Polymer 49 (2008) 38413854.
[127] C. McClory, S.J. Chin, T. McNally, Aust. J. Chem 62 (2009) 762785.
[128] R. Andrews, M.C. Weisenberger, Curr. Opin. Solid State Mater. Sci. 8 (2004)
3137.
[129] M. Moniruzzaman, K.I. Winey, Macromolecules 39 (2006) 51945205.
[130] S. Qiu, L. Wu, G. Shi, L. Zhang, H. Chen, C. Gao, Ind. Eng. Chem. Res. 49 (2010)
1166711675.
[131] L. Ge, Z. Zhu, V. Rudolph, Sep. Purif. Technol. 78 (2011) 7682.
[132] L. Ge, Z. Zhu, F. Li, S. Liu, L. Wang, X. Tang, V. Rudolph, J. Phys. Chem. C. 115
(2011) 66616670.
[133] P.S. Goh, B.C. Ng, A.F. Ismail, S.M. Sanip, M. Aziz, M.A. Kassim, Sep. Sci.
Technol. 46 (8) (2011) 12501261.
[134] M.A. Aroon, A.F. Ismail, M.M. Montazer-Rahmati, T. Matsuura, J. Membr. Sci.
364 (2010) 309317.
[135] T.-H. Weng, H.-H. Tseng, M.-Y. Wey, Int. J. Hydrogen Energy 34 (2009) 8707
8715.
[136] S. Kim, L. Chen, J.K. Johnson, E. Marand, J. Membr. Sci. 294 (2007) 147158.
[137] S. Kim, T.W. Pechar, E. Marand, Desalination 192 (2006) 330339.
[138] H. Chen, D.S. Sholl, J. Membr. Sci. 269 (2006) 152160.
[139] A.I. Skoulidas, D.S. Sholl, J.K. Johnson, J. Chem. Phys. 124 (2004) 054708.
[140] B.J. Hinds, N. Chopra, T. Rantell, R. Andrews, V. Gavalas, L.G. Bachas, Science
303 (2004) 6265.
[141] V.P. Sokhan, D. Nicholson, N. Quirke, J. Chem. Phys. 120 (8) (2004) 3855
3863.
[142] C. Matranga, B. Bockrath, N. Chopra, B.J. Jinds, R. Andrews, Langmuir 22
(2006) 12351240.
[143] H. Wang, Curr. Opin. Colloid Interface Sci. 14 (2009) 364371.
[144] B. Krause, M. Mende, P. Potschke, G. Petzold, Carbon 48 (2010) 27462754.
[145] P.-C. Ma, N.A. Siddiqui, G. Marom, J.-K. Kim, Compos. Part A 41 (2010) 1345
1367.
[146] A.F. Ismail, P.S. Goh, S.M. Sanip, M. Aziz, Sep. Purif. Technol. 70 (2009) 1226.
[147] A.F. Ismail, P.S. Goh, J.C. Tee, S.M. Sanip, M. Aziz, NANO 3 (3) (2008) 127143.
[148] K. MacJenzie, O. Dunens, A.T. Harris, Sep. Purif. Technol. 66 (2009) 209222.
[149] P.-X. Hou, C. Liu, H.-M. Cheng, Carbon 46 (2008) 20032025.
[150] X.-L. Xie, Y.-W. Mai, X.-P. Zhou, Mater. Sci. Eng R 49 (2005) 89112.
[151] Q. Hu, E. Marand, S. Dhingra, D. Fritsch, J. Wen, G. Wilkes, J. Membr. Sci. 135
(1997) 6579.
[152] S.M. Sanip, A.F. Ismail, P.S. Goh, T. Siga, M. Tanemura, H. Yasuhiko, Sep. Purif.
Technol. 78 (2011) 208213.
[153] N. Liu, F. Luo, H. Wu, Y. Liu, C. Zhang, Chen J. Adv. Funct. Mater 18 (2008)
15181525.
[154] R.H.A. Ras, Y. Umemura, C.T. Johnston, A. Yamagishi, R.A. Schoonheydt, Phys.
Chem. Chem. Phys. 9 (2007) 918932.
[155] G.M. Reeves, I. Sims, J.C. Cripps, Clay Materials used in Construction,
Geological Society of London, 2006.
[156] S. Pavlidou, C.D. Papaspyride, Prog. Polym. Sci. 33 (2008) 11191198.
[157] Y.W. Chen-Yang, Y.K. Lee, Y.T. Chen, J.C. Wu, Polymer 48 (2007) 29692979.
[158] J.P.G. Villaluenga, M. Khayet, M.A. Lopez-Manchado, J.L. Valentin, B. Seoane,
J.I. Mengual, Eur. Polymer J. 43 (2007) 11321143.
[159] J.W. Cho, D.R. Paul, Nylon 6 Nanocomposites by Melt Compounding, Polymer
42 (2001) 10831094.
[160] A. Leszczynska, J. Njuguna, K. Pielichowski, J.R. Banerjee, Thermochim. Acta
454 (2007) 122.
[161] J. Hrobarikova, J.-L. Robert, C. Calberg, R. Jerome, J. Grandjean, Langmuir 20
(2004) 98289833.
[162] S.S. Ray, M. Bousmina, Prog. Mater. Sci. 50 (2005) 9621079.
[163] S.G. Adoor, M. Sairam, L.S. Manjeshwar, K.V.S.N. Raju, T.M. Aminabhavi, J.
Membr. Sci. 285 (2006) 182195.
[164] T. Mcnally, W.R. Murphy, C.Y. Lew, R.J. Turner, G.P. Brennan, Polymer 44
(2003) 27612772.
[165] H. He, Y. Ma, J. Zhu, P. Yuan, Y. Qing, Appl. Clay Sci. 48 (2010) 6772.
[166] F. Bergaya, G. Lagaly, Appl. Clay Sci. 19 (2001) 13.
[167] F. Zhao, X. Bao, A.R. McLauchlin, J. Gu, C. Wan, B. Kandasubramanian, Appl.
Clay Sci. 47 (2010) 249256.
[168] J.M. Herrera-Alonso, E. Marand, J.C. Little, S.S. Cox, J. Membr. Sci. 337 (2009)
208214.
[169] R.K. Bharadwaj, Macromolecule 34 (2001) 91899192.
[170] M. Alexendre, P. Dubois, Mater. Sci. Eng. 28 (2000) 163.
[171] G. Defontaine, A. Barichard, S. Letaief, C. Feng, T. Matsuura, C. Detellier, J.
Colloid. Interface Sci. 343 (2010) 622627.
[172] T.D. Fornes, P.J. Yoon, D.L. Hunter, H. Keskkula, D.R. Paul, Polymer 43 (22)
(2002) 59155933.
P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264 263
[173] R.A. Vaia, E.P. Giannelis, Macromolecules 30 (25) (1997) 799079999.
[174] E. Naveau, Z. Dominkovics, C. Detrembleur, C. Jrme, J. Hri, K. Renner, M.
Alexandre, B. Puknszky, Eur. Polymer J. 47 (2011) 515.
[175] Y.-C. Wang, S.-C. Fan, K.-R. Lee, C.-L. Li, S.-H. Huang, H.-A. Tsai, J.-Y. Lai, J.
Membr. Sci. 239 (2004) 219226.
[176] F. Annabi-Bergaya, Microporous Mesoporous Mater. 107 (2008) 141148.
[177] H. Fischer, Mater. Sci. Eng. C 23 (2003) 763772.
[178] S.A. Hashemifard, A.F. Ismail, T. Matsuura, Chem. Eng. J. 170 (2011) 316325.
[179] S.A. Hashemifard, A.F. Ismail, T. Matsuura, J. Colloid. Interface Sci. 359 (2011)
359370.
[180] A. Galve, D. Sieffert, E. Vispe, C. Tellez, J. Coronas, C. Staudt, J. Membr. Sci. 370
(2011) 131140.
[181] R.J. Keuupler, D.J. Timmons, Q.-R. Fang, J.-R. Li, T.A. Makal, M.D. Young, D.
Yuan, D. Zhao, W. Zhuang, H.-C. Zhou, Coord. Chem. Rev. 253 (2009) 3042
3066.
[182] T. Devic, C. Serre, Porous Metal Organic Frameworks: From Synthesis to
Applications. In Ordered Porous Solids, 2009 (Chapter 4).
[183] A.J. Fletcher, K.M. Thomas, M.J. Rosseinsky, J. Solid State Chem. 178 (2005)
24912510.
[184] A.P. Malanoski, F. van Swol, Phys. Rev. E 66 (2002) 041602.
[185] F.A.A. Paz, J. Klinowski, Inorg. Chem. 43 (2004) 38823893.
[186] R. Cao, Q. Shi, D. Sun, M. Hong, W. Bi, Y. Zhao, Inorg. Chem. 41 (2002) 6161
6168.
[187] J.-R. Li, R.J. Kuppler, H.-C. Zhou, Chem. Soc. Rev. 38 (2009) 14771504.
[188] T.-H. Bae, J.S. Lee, W. Qiu, W.J. Koros, C.W. Jones, S. Nair, Angew. Chem. Int. Ed.
49 (2010) 98639866.
[189] D. Tanaka, S. Kitagawa, Captured Molecules in coordination frameworks, MRS
Bull. 32 (2007) 540543.
[190] A.I. Shoulidas, D.S. Sholl, J. Phys. Chem. B 109 (2005) 1576015768.
[191] S. Keshin, D.S. Sholl, Energy Environ. Sci. 3 (2010) 343351.
[192] H. Yehia, T.J. Pisklak, J.P. Ferraris, K.J. Balkus Jr., I.H. Musselman, Polymer
Prepr. 45 (2004) 3536.
[193] Y. Liu, Z. Ny, E.A. Khan, H.-K. Jeong, C.-B. Ching, Z. Lai, Microporous
Mesoporous Mater. 118 (2009) 296301.
[194] Y. Yoo, Z. Lai, H.-K. Jeong, Microporous Mesoporous Mater. 123 (2009) 100
106.
[195] S.I. Noro, S. Kitagawa, M. Kondo, K. Seki, Angew. Chem. Int. Ed. 39 (2000)
20812084.
[196] Y. Zhang, I.H. Musselman, J.P. Ferraris, K.J. Balkus Jr., J. Membr. Sci. 313 (2008)
170181.
[197] M.J.C. Ordonez, K.J. Balkus Jr., P. Ferraris, I.H. Messelman, J. Membr. Sci. 361
(2010) 2837.
[198] P. Raveendran, Y. Ikushima, S.L. Wallen, Acc. Chem. Res. 38 (2005) 478485.
[199] J.L.C. Rowsell, O.M. Taghi, Microporous Mesoporous Mater. 73 (2004) 314.
[200] C. Soldano, A. Mahmood, E. Dujardin, Carbon 48 (2010) 21272150.
[201] V. Singh, D. Joung, L. Zhai, S. Das, S.I. Khondaker, S. Seal, Graphene Based
Materials: Past, Present and Future, Prog. Mater. Sci. (2011), doi:10.1016/
j.pmatsci.2011.03.003.
[202] A.K. Geim, K.S. Novoselov, Nat. Mater. 6 (2007) 183191.
[203] S. Gilje, H. Song, M. Wang, K.L. Wang, R.B. Kaner, Nano Lett. 7 (2007) 3394.
[204] F. Schedin, A.K. Geim, S.V. Morozov, E.W. Hill, P. Blake, M.I. Katsnelson, K.S.
Novoselov, Nat. Mater. 6 (2007) 652655.
[205] Z. Liu, J.T. Robinson, X. Sun, H. Dai, J. Am. Chem. Soc. 130 (33) (2008) 10876
10877.
[206] K. Kalaitzidou, H. Fukushima, L.T. Drzal, Carbon 45 (2007) 14461452.
[207] H. Kim, C.W. Macosko, Macromolecules 41 (2008) 33173327.
[208] E.L. Cussler, S.E. Hughes, W.J. Ward, R. Aris, J. Membr. Sci 38 (1988) 161174.
[209] T. Kuilla, S. Bhadra, D. Yao, N.H. Kim, S. Bose, J.H. Lee, Prog. Polym. Sci. 35
(2010) 13501375.
[210] E.L. Cussler, Diffusion: mass transfer in uid systems, 2nd ed., Cambridge
University Press, Cambridge, 1997.
[211] M.A. Osman, V. Mittal, M. Morbidelli, U.W. Suter, Macromolecules 36 (2003)
98519858.
264 P.S. Goh et al. / Separation and Purication Technology 81 (2011) 243264