Chemical Engineering Science 104 (2013) 908–924

Contents lists available at ScienceDirect

Chemical Engineering Science

journal homepage: [Link]/locate/ces

Review

Membranes from nanoporous 1D and 2D materials: A review

of opportunities, developments, and challenges
Wun-gwi Kim, Sankar Nair n
School of Chemical & Biomolecular Engineering, Georgia Institute of Technology, Atlanta, GA 30332-0100, USA

art ic l e i nf o a b s t r a c t

Article history: Membranes utilizing nanoporous one-dimensional (1D) and two-dimensional (2D) materials are
Received 20 July 2013 emerging as attractive candidates for applications in molecular separations and related areas. Such
Received in revised form nanotubular and nanolayered materials include carbon nanotubes, metal oxide nanotubes, layered
21 September 2013
zeolites, porous layered oxides, layered aluminophosphates, and porous graphenes. By virtue of their
Accepted 24 September 2013
Available online 16 October 2013
unique shape, size, and structure, they possess transport properties that are advantageous for membrane
and thin film applications. These materials also have very different chemistry from more conventional
Keywords: porous 3D materials, due to the existence of a large, chemically active, external surface area. This
Membranes feature also necessitates the development of innovative strategies to process these materials into
Layered materials
membranes and thin films with high performance. This work provides the first comprehensive review
Nanoporous
of this emerging area. We first discuss approaches for the synthesis and structural characterization of
Nanotubes
Separations nanoporous 1D and 2D materials. Thereafter, we elucidate different approaches for fabrication of
Nanocomposites membranes and thin films from these materials, either as multiphase (composite/hybrid) or single-phase
membranes. The influence of surface chemistry and processing techniques on the membrane morphol-
ogy is highlighted. We then discuss the applications of such membranes in areas relating to molecular
transport and separation, e.g. gas and liquid-phase separations, water purification, and ion-conducting
membranes. The review concludes with a discussion of the present outlook and some of the key scientific
challenges to be addressed on the path to industrially applicable membranes containing nanoporous 1D
and 2D materials.
& 2013 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 909
1.1. Polymeric and inorganic membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 909
1.2. Composite (‘mixed matrix’) membranes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 910
1.3. Membranes incorporating 1D and 2D materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 910
2. Membrane-Forming 1D and 2D materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 910
2.1. Nanotubular (1D) materials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 910
2.2. Layered oxide (2D) materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 911
2.2.1. MCM-22, ITQ, and Nu-6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 912
2.2.2. AMH-3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 912
2.2.3. Layered MFI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 913
2.2.4. Layered titanosilicates and aluminophosphates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 913
2.3. Graphene-based materials. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 913
3. Processing of membranes containing 1D and 2D materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 915
3.1. Nanocomposite membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 915
3.1.1. Composite membranes with nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 915
3.1.2. Composite membranes with porous layered oxides. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 915
3.2. All-inorganic membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 916
3.3. Microstructural analysis of nanocomposite membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 917

n
Corresponding author. Tel.: þ 1 4048 944 826.
E-mail address: [Link]@[Link] (S. Nair).

0009-2509/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.
[Link]
 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924 909

4. Separation applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 918

4.1. 1D nanotubular membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 918
4.2. 2D layered oxide membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 919
4.3. Graphene and other carbon membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 921
5. Outlook and challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 921
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 922
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 922

1. Introduction
containing 1D and 2D materials. In the subsequent sections, we
Advanced membrane separations have generated much inter- present an overview of the main classes of 1D and 2D materials
est in the last two decades, owing to their potential for technolo- synthesized for use in membranes, and discuss the processing
gical impact in the energy, fuels, petrochemical, and renewable routes that have been developed for fabricating membranes
chemical sectors (Pandey and Chauhan, 2001; Vane, 2005; Jagur- comprising these materials. We then review the characterization
Grodzinski, 2007; Bernardo et al., 2009). Conventional separation and separation performance of such membranes, and conclude
methods, currently in heavy use for petrochemical processing, are with a discussion of the current challenges and prospects in
increasingly challenged by energy cost, performance limitations, this area.
and environmental (e.g., carbon) footprint. Other important tech-
nological problems such as large-scale potable water production,
greenhouse gas capture, and production of renewable fuels and
chemicals, also require energy-efficient, environmentally benign, 1.1. Polymeric and inorganic membranes
and lower-cost separation technologies that are unlikely to be
dominated by conventional processes such as distillation, liquid Polymers used for fabricating membranes can be semi-
absorption and stripping, crystallization, and liquid extraction. crystalline (e.g., polyvinylalcohol and cellulose acetate), glassy
At the same time, the science and technology of membrane (e.g., polysulfone and polyimides), or elastomeric (e.g., silicone-
fabrication and scale-up has made considerable progress over based polymers). These polymers show a variety of structural and
the last decade, further brightening the prospects for future dynamical behavior, leading to a large range of molecular permea-
large-scale commercial applications replacing conventional tion properties. While diffusivity and solubility both determine the
separations processes. permeability of molecules in polymers, the selectivity of glassy
Polymeric and inorganic membranes have received substantial polymers for one kind of molecule over another is mainly
research and development attention (as summarized in the following governed by diffusivity differences. Glassy polymers mainly
section), but are constrained by trade-offs in performance metrics or exploit size exclusion effects arising from differences in the kinetic
by lack of scalability of processing techniques. Nanocomposite/hybrid diameters of molecules (Stern et al., 1989; Park and Paul, 1997).
membranes, which typically contain a high-performance inorganic However, selectivity in elastomers is dominated by solubility
materials (fillers) dispersed in a polymeric matrix, have gained differences, and hence more condensible large molecules (such
ground as a means of combining the best characteristics of polymeric as hydrocarbons) are more permeable in elastomers than small
and inorganic materials while potentially overcoming their individual gases like N2. Elastomeric polymer membranes can be used to
limitations. Most nanocomposite membranes contain fillers of iso- separate organic vapors from non-condensible gases such as air
tropic or near-isotropic morphologies, with dimensions typically in (Schultz and Peinemann, 1996; Pinnau and He, 2004). This
the 100–1000 nm range. However, it has been realized for some behavior is sometimes referred to as ‘reverse selectivity’ in relation
time that membranes containing strongly anisotropic and low- to the properties of glassy polymers. Semicrystalline and glassy
dimensional fillers would offer some unique possibilities and advan- polymers find commercial applications in natural gas purification
tages. These possibilities have moved closer to realization in the last and dehydration of organic/water mixtures. However, the largest
decade, owing to key breakthroughs in the synthesis and character- obstacle in utilization of polymeric membranes in industrial
ization of anisotropic and low-dimensional separation materials and applications is their intrinsic trade-off relationship between selec-
their fabrication into functional membranes by different processing tivity and permeability, embodied in the Robeson plot (Robeson,
strategies. 2008).
This review focuses on understanding the current state of the Nanoporous inorganic materials offer much higher permeabil-
art, opportunities, and challenges in the science and engineering ities than polymers, and also high selectivities in many separations
of membranes containing one-dimensional (1D) and two- of technological interest. Nanoporous materials such as zeolites
dimensional (2D) nanoscopic materials. The 1D materials consid- and metal-organic frameworks (MOFs) can be fabricated into
ered are primarily of nanotubular structure, and it is desired to selective separation membranes that exploit their molecular-
obtain fast, selective permeation of molecules through their sieving nanopores (Davis, 2002; Snyder and Tsapatsis, 2007;
tubular nanopores. For membrane transport applications, a per- Shah et al., 2012). Membranes of various zeolites such as MFI,
pendicular (i.e., transmembrane) orientation of the 1D materials FAU and LTA are being developed for separation applications (Caro
embedded in the membrane is expected to be strongly preferable and Noack, 2008). Despite two decades of development on zeolite
over random or in-plane orientations. The 2D materials of interest membranes, and a few commercialized applications such as
are primarily those with a layer-like morphology characterized by alcohol dehydration, zeolite membranes are presently not widely
a high aspect ratio (large lateral dimensions and small thickness), utilized due to the difficulties their economical and quality-
and containing nanoporous ‘perforations’ spanning the layer controlled fabrication on a large scale. Challenging issues in zeolite
thickness through which selective molecular transport can occur. membranes, such as reproducibility of membrane manufacture
In this introductory section, we overview the state of polymeric, and control over defect formation must be overcome to realize the
inorganic, and isotropic composite membranes, and contrast these industrial applications of zeolite membranes (Choi et al., 2009a;
membrane morphologies with emerging membrane architectures Gascon et al., 2012).
910 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924

1.2. Composite (‘mixed matrix’) membranes through an appropriate fabrication process, favorable properties of
nanotubes such as fast molecular transport can be most efficiently
‘Mixed matrix’ membranes (MMMs) can be prepared by incor- exploited. The potentially tunable functionality of the nanotube
porating appropriately chosen inorganic particles in a polymeric mouths, as well as inner and outer surfaces, can be used to expand
matrix. In doing so, it is hoped (Koros, 2004) that MMMs can the range of separations applications and compatibility with
combine the advantages of polymers (such as good processibility, various types of matrix materials. Fig. 1b shows a schematic
scalable fabrication, and lower cost) with those of inorganic diagram of a membrane incorporating high-aspect ratio layered
materials (such as high permeability, selectivity, and thermal/ (2D) materials. In comparison to isotropic fillers, the high aspect
mechanical stability). Many research groups have incorporated ratio of the layers or ‘flakes’ necessitate longer and more tortuous
various inorganic materials such as zeolites, MOFs, and other paths for the larger molecule that cannot easily pass through
molecular sieves into polymer matrices. Some pioneering works the pores, and thereby effectively decrease the permeability of
on this topic include the incorporation of zeolite 4A in the the larger molecule (Ray and Okamoto, 2003). At the same time,
polyimide Matrimid (Mahajan and Koros, 2002), carbon molecular the nanoscopic thickness of the flakes allows fast permeation of
sieves in Matrimid (Vu et al., 2003), fumed-silica nanoparticles in the smaller molecule. For the above reasons, it is hypothesized
poly(4-methyl-2-pentyne) (Merkel et al., 2002), zeolite MFI or that membranes containing nanoscopically thin flakes can allow
carbon black materials in PDMS (Moermans et al., 2000; Vane high separation performance with a much lower loading of the
et al., 2008). A large number of works have followed these early flakes than required for isotropic fillers. From the viewpoint of
reports, and have been recently reviewed (Nasir et al., 2013). An membrane fabrication and scale-up, the nanoscopically thin
important theme in mixed matrix membranes is the improvement layered flakes are favorable because they can be incorporated in
of interfacial compatibility between the polymer matrix and the ultra-thin membranes such as the skin layers of hollow fiber
filler particles. For example, in the case of zeolite materials a membranes (Johnson and Koros, 2009). Beyond their utilization
number of surface-treatment methodologies have been developed in composite membranes, 2D materials could also be used to form
to improve interfacial adhesion and filler dispersion, thereby purely inorganic membranes. Direct deposition or coating of the
leading to higher separation performance (Shu et al., 2007; Bae flakes on porous supports could provide a less challenging alter-
et al., 2009). MMMs incorporating MOF materials have also been native to fabrication of inorganic membranes by hydrothermal
reported in the recent literature, and are reviewed elsewhere growth on the support.
(Jeazet et al., 2012). Overall, MMMs are making their way to
reaching performance levels viable for technological consideration
over polymeric membranes. The problem of scalable processing of 2. Membrane-Forming 1D and 2D materials
MMMs is non-trivial, but perhaps less difficult than the processing
of zeolitic membranes. Specifically, the physics of transport in 2.1. Nanotubular (1D) materials
MMMs dictates that high filler loadings (30–50 wt%) are usually
required to obtain substantial improvements over the base poly- Fig. 2a shows the molecular structure of a single-walled carbon
meric membrane (Adams et al., 2011). The scalable fabrication of nanotube (SWNT) and a multi-walled carbon nanotube (MWNT)
thin (submicron) MMMs containing large filler fractions (Ismail, 2011). These nanotubes are based upon the graphitic sheet
(Pendergast and Hoek, 2011) is challenging and is a topic of structure of hexagonally arranged carbon atoms with sp2 hybridi-
ongoing work. Furthermore, the study of transport phenomena zation. This arrangement contributes to the unique mechanical
in MMMs also reveals an intrinsic limitation of using isotropic and electronic properties of carbon nanotubes (CNTs), which are
fillers, namely the high probability of permeating molecules under intense investigation for a variety of applications. CNTs have
bypassing the filler particles (Vinh-Thang and Kaliaguine, 2013). been used to form composites with polymers, specifically for
mechanical reinforcement and improvement in electrical conduc-
tivity (Roy et al., 2012; Tkalya et al., 2012). The hollow tubular
1.3. Membranes incorporating 1D and 2D materials morphology of the CNT has also led to its suggestion as a perme-
able and mechanically strong filler material for membrane-based
Porous 1D nanotubular (Cong et al., 2007; Ismail et al., 2009; molecular separations (Goh et al., 2013). A primary advantage of
Kang et al., 2012) and 2D layered (Jeong et al., 2004; Galve et al., CNTs as membrane materials is the very high flux that can be
2011, Zornoza et al., 2011) materials have been suggested as obtained, due to the atomically smooth nature of the walls.
leading to membranes with unique transport properties, that However, the corresponding drawback is a lack of high selectivity
could overcome the limitations of isotropic inorganic fillers in for different molecules through a pristine CNT (Kim et al., 2007).
MMMs. Fig. 1 illustrates the structures of idealized membranes The electronic structure of the inner wall of the CNT is considered
containing 1D and 2D materials. In Fig. 1a, the nanotube materials essentially unreactive and difficult to functionalize with organic or
are shown as vertically aligned and completely spanning the other moieties that may impart selectivity (Chamberlain et al., 2011)
thickness of the membrane. By realizing such an ideal structure Another key challenge for CNT-based membrane materials is the

Fig. 1. Schematic diagrams of composite membranes incorporating (a) 1D nanotubular materials, and (b) 2-D nanoplatelets (adapted from Kim et al., 2013).
 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924 911

Fig. 2. Structures of (a) single- and multi-walled carbon nanotubes (SWNTs/MWNTs), (b) a single-walled aluminosilicate nanotube, and (c) summary of molecular-level
approaches to control the curvature and composition of single-walled metal oxide nanotubes (adapted from Yucelen et al., 2012).

unsatisfactory interface compatibility between CNTs and many manipulation of the growth mechanisms of these materials by
membrane-forming polymers, which results in defect formation in controlling the shape and identity of the molecular and nanoscale
CNT/polymer nanocomposite membranes (Ismail et al., 2009). Sev- precursors responsible for nanotube formation (Yucelen et al., 2011).
eral routes for modification of the outer surface of CNTs are available, Metal oxide nanotube diameters can now be controlled with
including covalent and non-covalent binding with desired substitu- Ångstrom-level precision (Yucelen et al., 2012). Their lengths are
ents (Vaisman et al., 2006; Zhao and Stoddart, 2009). tunable in a wide range of approximately 20–800 nm. Furthermore,
Inorganic nanotubes, in the form of metal oxides, have several the hydroxylated inner and outer surfaces are amenable to modifica-
advantages over CNTs for membrane applications (Mukherjee tion by a number of grafting reactions (Kang et al., 2011a, 2011b) to
et al., 2007; Rao and Govindaraj, 2009; Tenne and Seifert, 2009). introduce different types of functional groups. Organic-functionalized
A wide range of multi-walled inorganic nanotubes with large pore aluminosilicate nanotubes have also been produced by direct synth-
diameters in the 10–50 nm range have been synthesized (Tenne esis, thereby yielding nanotubes with organosilane groups replacing
and Seifert, 2009); however, single-walled (or few-walled) nano- the silanol groups in the inner wall (Bottero et al., 2011). Fig. 2c
tubes with pore diameters in the  1 nm or sub-nm range are summarizes different approaches available to produce single-walled
clearly much more suitable for molecular separations. Although metal oxide nanotubes with controlled nanotube curvature and
the currently available repertoire of such materials is not large, composition.
several interesting materials of this type are known. These include
single-walled aluminosilicate or aluminogermanate-based nano-
tubes which are synthetic analogs of the nanotube mineral 2.2. Layered oxide (2D) materials
imogolite. The structure of the single-walled aluminosilicate
nanotube is illustrated in Fig. 2b. The outer wall is composed of Layered materials have created recent interest as molecular
a hexagonal gibbsite-like lattice of aluminum hydroxide, whereas sieves and heterogeneous catalysts, because they can be exfo-
the inner walls are lined with an ordered array of pendant silanol liated/delaminated into single-layered (or few-layered) nanostruc-
groups (Mukherjee et al., 2005). These materials are synthesized tures, and can provide many functional sites with a large surface
under much milder conditions in comparison to CNTs, specifically area (Liu, 2007; Centi and Perathoner, 2008; Dasgupta and Torok,
by hydrothermal methods at  100 1C and mildly acidic pH. An 2008). Clays are representative nonporous layered materials.
aluminogermanate analog is also known, with the silicon atoms A single sheet of a typical clay material has a thickness of
replaced by germanium. The latter material has recently also been  1 nm, and the structure typically includes two ‘tetrahedral’
obtained in double-walled form (Thill, 2012a, 2012b). The last sheets (i.e., with elements such as Si and Al in tetrahedral
decade has seen significant progress in the understanding and coordination with oxygen) sandwiching an inner ‘octahedral’
912 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924

sheet. The valency difference between atoms such as Al and Mg in MCM-22 after the swelling process (He et al., 1998). Exfoliated
the octahedral layer imparts a net negative charge to the layer and MCM-22 (ITQ-2) is quite attractive for high performance hetero-
also creates active sites for catalytic applications (Paul and geneous catalysis, due to its large surface density of active sites.
Robeson, 2008). Clays have been used as heterogeneous catalysts Also, it is useful as a molecular sieve for H2, due to its 6-membered
by intercalating them with a metal complex (Pinnavaia, 1983). The rings perpendicular to the layers. The direct deposition of MCM-22
spacing between individual clay sheets can be expanded using flakes has been used to form prototype H2 separation membranes
appropriate swelling agents such as surfactants, and can then be (Choi and Tsapatsis, 2010). Different kinds of exfoliated materials
‘pillared’ by additional silica incorporation to create microporous/ such as ITQ-6, ITQ-18, and Nu-6(2) have been developed starting
mesoporous materials (Kloprogge, 1998). Clays can also be exfo- from layered versions of zeolites Ferrierite and Nu-6 (Corma et al.,
liated after swelling, by application of mechanical force or disper- 2000, 2001) respectively. For example, the layered zeolite Nu-6,
sion in a suitable solvent. Exfoliated clays have been incorporated which contains 8MRs with 0.33 nm pore openings, can be directly
in polymer matrices to improve the mechanical and chemical exfoliated during ion exchange of its bipyridine structure-directing
properties of engineering plastics (Usuki et al., 1993; Messersmith agent (SDA) with ions such as Na þ and CTA þ at mild pH (  9) and
and Giannelis, 1994) or to improve barrier properties (Choudalakis room temperature (Gorgojo et al., 2011).
and Gotsis, 2009). For example, the problem of methanol crossover
in sulfonated fluoropolymer (e.g., Nafions) membranes (Rhee et al.,
2005; Thomassin et al., 2006, Lin et al., 2007) can be mitigated by the 2.2.2. AMH-3
inclusion of exfoliated clay materials to create a barrier for methanol AMH-3 is the first 3D-porous layered silicate/layered zeolite,
permeation, leading to more efficient membranes for direct methanol having 8 MR pores in all three principal crystallographic directions
fuel cells (DMFCs). (Jeong et al., 2003). Fig. 3b shows the pore structure of AMH-3
materials, with the charge-balancing cations removed for clarity.
Single layers of AMH-3 of thickness 1 nm are spaced by charge-
2.2.1. MCM-22, ITQ, and Nu-6 balancing Na þ and Sr2 þ cations (not shown). AMH-3 is an
MCM-22 is a porous layered material with a 10-membered ring attractive candidate for membranes containing ‘selective flakes’,
(MR) pore structures along two lateral directions and 6MR pores because its 3-dimensional 8MR pore openings are of the appro-
along the vertical direction, as illustrated in Fig. 3a (Leonowicz priate range for various gas separations. The nominally 0.34-nm
et al., 1994). Corma (1998) reported swelling methods for MCM-22 pore openings of AMH-3 can be applied to various combinations of
using surfactant intercalation, and exfoliated the resulting material separations involving gases such as H2 (kinetic diameter 0.29 nm),
to single layers, thereby obtaining a new layered material named CO2 (0.33 nm), O2 (0.35 nm), N2 (0.36 nm), and CH4 (0.38 nm). A
ITQ-2. The material MCM-36 has been introduced by pillaring number of challenges have been addressed to apply AMH-3 in

Fig. 3. Structures of nanoporous layered silicates: (a) MCM-22 and (b) AMH-3.
 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924 913

Fig. 4. (a) Schematics of layered MFI assembly and (b) layered MFI structure visualized by high resolution-transmission electron microscopy (adapted from Choi et al.,
2009b).

actual gas separation membranes. Unlike most other layered 2.2.4. Layered titanosilicates and aluminophosphates
silicate materials, the swelling of AMH-3 is impeded by the Apart from layered nanoporous silicates, other layered oxide
strongly bound Na þ and Sr2 þ cations between layers. A ‘sequen- materials such as titanosilicates and aluminophosphates (AlPOs)
tial intercalation’ method (Choi, 2008a, 2008b, 2008c) has been have also been synthesized. The layered titanosilicate JDF-L1
developed to overcome this difficulty. It involves the controlled ion (Roberts et al., 1996) contains 6MRs formed by two TiO5 square
exchange of the interlayer cations with protons from a buffered pyramids and four SiO4 tetrahedra, and whose pore openings of
solution, followed by the introduction of a primary (rather than  0.3 nm may be suited for selective hydrogen permeation. Using
quaternary) amine surfactant such as dodecylamine between the a sequential intercalation process similar to that developed for
layers. However, this process cannot avoid structural changes AMH-3, this material has been successfully exfoliated by proton
occurring in swollen AMH-3, involving intra-layer condensation exchange from a buffered histidine solution followed by intercala-
of terminal Si-OH groups (Choi, 2008a, 2008b, 2008c). More tion of nonylamine and final extraction with an HCl/ethanol/water
recently, it has been shown that the use of a primary diamine solution (Rubio et al., 2010). Layered AlPO materials also have
surfactant (such as dodecyldiamine) allows better preservation of potential applications in catalysis (Thomas et al., 2001) and in
the AMH-3 layer structure (Kim et al., 2011). The functionalization fabrication of nanocomposite materials and membranes (Jeong
of swollen AMH-3 surfaces with silane reagents containing hydro- et al., 2004). By varying the SDA, porous AlPO materials can be
carbon chains also improved its hydrophobicity and may allow synthesized in different forms such as 1D chains, 2D layers, and 3D
better compatibility with hydrophobic polymer matrices. open frameworks (Williams et al., 1997; Yu et al., 1998a, 1998b; Li
et al., 1999). The dimensionality, pore structure, and composition
of the AlPO are influenced by the size and shape of the organic
2.2.3. Layered MFI (usually amine) SDA (Li et al., 1999). The 2D (layered) AlPOs have
Layered MFI is the most recently proposed nanoporous layered negatively charged layers with Al:P ratios less than unity, in
material. This material was obtained by synthesizing a new SDA comparison with 3D AlPOs with an Al:P ratio of unity.
that is terminated with a long-chain C22 alkyl group (Choi et al., The anionic AlPO layers contain AlO4 tetrahedra and O¼PO3
2009b). The amine groups in the SDA allow the formation of an (PO4) units with corner-shared oxygen atoms, to form AlPO
MFI structure in the lateral directions (crystallographic ac-plane) materials with varying Al:P ratios (Williams et al., 1996; Yu
in a manner similar to the conventional tetrapropylammonium et al., 1998a, 1998b; Yan et al., 2000). The SDA molecules (such
SDA, but the long-chain alkyl group prevents crystal growth in the as trimethyl-, triethyl-, or isopropanolamine) interact with the
perpendicular (crystallographic b-axis) direction (Fig. 4a). Hence, a AlPO layers via non-covalent bonds, and hence the 2D layers can
layered material containing 2 nm thin MFI sheets could be be delaminated to single layers for the fabrication of nanocompo-
synthesized in a one-step hydrothermal reaction. Layered MFI site membranes. However, unlike the slab-like layered silicate
has the same pore size and local structure as conventional MFI, i.e. materials described above, the AlPO layers have more fragile
10MR pores with a nominal pore size of 0.55 nm (Diaz, 2004). This net-like morphologies of thickness  0.5 nm, leading to lower
relatively larger pore size (in comparison to MCM-22 and AMH-3) chemical stability upon swelling and exfoliation.
can be applied to the separation of bulkier molecules such as
hydrocarbons or other organics. Moreover, as shown in Fig. 4b, 2.3. Graphene-based materials
layered MFI has an already swollen structure since the MFI layers
are spaced by the long-chain alkyl groups of the SDA. This may be Graphene is a single-atom-thick 2D carbon material with a
an advantage for utilization of this material in membrane fabrica- graphitic hexagonal layer structure. It has a network of sp2
tion applications, since it avoids the need for swelling processes hybridized carbon atoms that leads to a zero bandgap and enables
involving ion exchange, surfactant intercalation, and other steps. the fast transport of electrons. Since graphene was first isolated
By controlling the Na þ content in the reaction gel, two different from graphite sheets by mechanical exfoliation methods
types of layered MFI – multilamellar and unilamellar – have been (Novoselov et al., 2004), scientific interest in its electrical, optical
synthesized. The pillaring process for layered MFI has expanded its and mechanical properties has grown extremely fast (Bunch et al.,
potential applications in heterogeneous catalysis (Na et al., 2010). 2007; Staley et al., 2007; Liu et al., 2009). High Resolution-
Layered MFI has also been combined with conventional MFI in the Transmission Electron Microscopy (HRTEM) visualizes the single
form of a hybrid BMLM (bulk MFI-layered MFI) material. This and bilayer graphene sheet as shown in Fig. 5a (Urban, 2011).
material is synthesized by epitaxial growth of layered MFI on Utilization of graphene as an additive material requires larger-
conventional MFI crystal surfaces (Kim et al., 2012a, 2012b). The scale production of graphene sheets. Single-sheet graphene was
growth of layered MFI on bulk MFI has been found to occur in all obtained from graphite oxide by a reduction process using hydra-
three principal crystallographic directions, and involves both zine (Tung et al., 2009). Apart from the developments on utilizing
homoepitaxial and heteroepitaxial growth. graphene directly grown on substrates for electronic/optical
914 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924

Fig. 5. (a) HRTEM image of single and bilayer of graphene (adapted from Urban, 2011), and (b) schematics of porous graphene structure.

Fig. 6. Fabrication processes for nanotube/polymer composite membranes using (a) solution casting method, and (b) CVD method followed by polymer coating and etching
process (adapted from Majumder et al., 2011).

applications (Allen et al., 2010; Shao et al., 2010), another important layer structure and strong mechanical properties are attractive for
potential application of bulk-processed graphene is in the form of application in molecular separation. However, as shown in Fig. 5a,
separation membranes (Blankenburg et al., 2010). The one-atom-thin graphene sheets do not have pores for molecular sieving. The
 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924 915

introduction of sub-nanometer pores is required to obtain sufficient (6FDA-4MPD/6FDA-DABA 4:1) matrix under ultrasonication,
molecular permeance and to control the permselectivity. As a proof- thereby allowing the formation of a composite membrane
of-concept, electron beam lithography has been used to create holes (Sieffert and Staudt, 2011). There is no evidence for a preferential
in suspended graphene sheets (Fischbein and Drndic. 2008). How- orientation/alignment of the MWCNTs in the membrane.
ever, this ‘top-down’ fabrication method is hard to control and often
not precise. ‘Bottom-up’ assembly processes are also being consid-
ered for the formation of porous 2D sheets of graphene (Bieri et al., 3.1.2. Composite membranes with porous layered oxides
2009). For example, as shown in the schematic of porous graphene For both nanocomposite and all-inorganic membrane fabrica-
(Fig. 5b), phenylene molecules can be used as building units for tion utilizing layered oxides, controlling the exfoliation of the
porous grapheme (Li et al., 2010). Computational studies have been layers is important. As-synthesized or surfactant-swollen layered
performed to calculate the separation properties in porous gra- materials usually comprise stacks of 100-10,000 layers. Depending
phenes, indicating interesting molecular sieving possibilities (Jiang on the degree of exfoliation, the final processed layered materials
et al., 2009; Blankenburg et al., 2010, Li et al., 2010; Schrier, 2010). (flakes) can be defined (Paul and Robeson, 2008) as either
The use of few-layer graphene (rather than monolayer graphene) (1) ‘immiscible’ (essentially unaltered from their as-synthesized
may also lead to other interesting possibilities for molecular separa- or swollen state), (2) ‘intercalated’ (having polymer chains
tion. However, this form of graphene is less commonly considered threaded between the layers which still maintain some registry
for composite film formation in the current literature (Prolongo with each other), or (3) ‘exfoliated’ (with individual layers or few-
et al., 2013). layer stacks separated from each other and dispersed). There are
many studies on polymer/clay composite materials for applica-
tions to mechanical reinforcement and enhancement of barrier
3. Processing of membranes containing 1D and 2D materials properties (Ray and Okamoto, 2003; Schaefer and Justice, 2007;
Vaia and Maguire, 2007; Pavlidou and Papaspyrides, 2008). These
After the preparation of well-defined 1D (nanotubular) and 2D studies contain considerable insight into intercalation and exfolia-
(layered) porous materials, there are several steps in the formation tion processes for clay materials, which has served as a useful
of membranes utilizing these materials. In this section, we con- starting point for work on exfoliation of porous layered materials.
sider processing methodologies for fabricating both nanocompo- Melt-blending, a technique used in the majority of studies
site and single-phase membranes with low-dimensional porous on exfoliation of clay materials in polymers, is also an attractive
materials. method for exfoliative dispersion of porous layered materials in
polymers. It involves the use of different types of screw extruders
3.1. Nanocomposite membranes to blend molten polymers and layered materials at temperatures
near 400 1C. The results depend strongly on the melt-blending
3.1.1. Composite membranes with nanotubes conditions, the affinity between the polymer and clay, and the
In nanotube/polymer composite membranes, the highest per- extruder type. Specifically, layered MFI powders (Fig. 7a) have
formance in terms of molecular selectivity and permeability can be been incorporated in polystyrene and melt-blended using a twin
attained by the simultaneous dispersion and alignment of nano- screw extruder. As a result (Fig. 7b), it was found that the single
tubes in polymer matrix, as illustrated in Fig. 1a. Several layers of layered MFI have been completely exfoliated (Varoon
approaches using physical and chemical methods have been et al., 2011). However, the technique is not directly applicable to
reported to align CNT arrays in polymer matrices. One example the fabrication of nanocomposite membranes for separations,
is the alignment of CNTs during solvent casting (Chen and Tao, since the polymers required for permselective membranes (e.g.,
2005). The relaxation and alignment of polymer chains during the polyimides, cellulose acetate, polyvinylalcohol) are usually
swelling and curing stages of solvent casting can be used for the solution-processed and not melt-processable. Previous attempts
alignment of CNT arrays, as illustrated in Fig. 6a. This process is to exfoliate porous layered materials in solution (i.e., solution-
potentially advantageous for scale-up. However, it currently suf- blending) have encountered problems such as strong aggregation
fers from several problems, such as the difficulty in controlling of the layered materials (Krikorian and Pochan, 2003; Ray and
the uniformity of the final membrane. A CVD process for aligned Bousmina, 2005; Pavlidou and Papaspyrides, 2008). Similar to the
growth of a nanotube 'forest' using iron clusters as the catalysts, illustration in Fig. 6a, the conventional solution blending process
followed by infiltration of the intertube spaces with a polymer includes the intercalation of polymer chains between layers,
(Fig. 6b), has also been suggested. Composite membranes with followed by membrane casting and solvent evaporation. An active
well-aligned CNTs have been fabricated with the above process exfoliation process can be added to the conventional solution
(Cheung et al., 2002). This method can provide high quality blending process. For example, it has been recently shown that
membranes; however, the scale-up of the CVD process is challen- high-shear mixing can be used to exfoliate swollen AMH-3 (Fig. 7d
ging. Other methods for the alignment of nanotubes employ shear and e) in solution during blending with cellulose acetate. Fig. 7f
flows or electric fields (Pujari, 2009; Mauter et al., 2010). The shows a TEM cross-section image of the membrane forming after
deposition of CNTs on substrates (that have different surface the casting step. The stacked layers of swollen AMH-3 have been
properties from the nanotubes) has been shown to result in converted to few-layer flakes dispersed in the cellulose acetate
vertical alignment (De Heer et al., 1995; Kim et al., 2007). matrix (Kim et al., 2013).
Apart from CNTs, there has been a recent demonstration of the Other potential methods for membrane fabrication with exfo-
fabrication of composite membranes using single-walled alumi- liated porous layered materials include in situ polymerization and
nosilicate nanotubes (Fig. 2b) dispersed in a hydrophilic polymer sol-gel processing. In situ polymerization was used to fabricate
(polyvinylalcohol) (Kang et al., 2012). The hydrophilic nature of the clay/Nylon-6 composite materials (Usuki et al., 1993). The concept
nanotube material allows excellent dispersion in the polymer was to polymerize the monomer in the gallery spaces of layered
solution. Since no driving force was present for nanotube align- materials and thereby expand their d-spacing, eventually resulting
ment, the resulting membranes showed a random orientation of in delamination. An advantage of this method is the capability to
the nanotubes in the polymer matrix. Similarly, hydroxyl- control the polymerization rate and exfoliation conditions. Using
functionalized multiwalled CNTs (MWNTs) have also been well ring-opening polymerization of cyclic oligomers, the degree of
dispersed via formation of covalent linkages to a copolyimide exfoliation of clay materials showed significant improvements (Lee
916 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924

Fig. 7. Exfoliation and membrane fabrication processes using (a–c) layered MFI (adapted from Varoon et al., 2011), and (d–f) SAMH-3 (adapted from Kim et al., 2013).
The initial as-made layered MFI and AMH-3 (a and d) have been delaminated to single-layer and few-layer flakes respectively (b and e); and (c and f) subsequently fabricated
into all-inorganic layered MFI and cellulose acetate/AMH-3 composite membranes respectively.

et al., 2005). While the in situ polymerization method involves only perpendicular to the substrate with c-axis out-of-plane orienta-
the polymerization of monomers, the sol–gel process includes both tion. After post-treatment with TEOS vapor, the membrane
the polymerization reaction and crystal growth of the inorganic displayed H2/N2 separation with high selectivity of 120. In
particles at the same time (Pavlidou and Papaspyrides, 2008). Neither another example, the exfoliated flakes of layered MFI embedded
of these techniques has yet been applied to porous layered materials. in polystyrene after melt blending (Fig. 7b) were recovered by
dissolution of the polymer in an organic solvent, and were then
3.2. All-inorganic membranes deposited as coatings on porous supports using vacuum filtration
(Fig. 7c) (Varoon et al., 2011). The initial coatings were not
Apart from their processing into nanocomposite membranes, permselective due to the presence of gaps between layers. To
nanoporous layered materials after exfoliation can be used to improve the separation properties, the coated membranes were
form inorganic membranes on porous substrates. Methods for the post-treated with a silica-containing solution by hydrothermal
fabrication of inorganic membranes using exfoliated layered growth, to close the gaps by growth of crystalline MFI. The
flakes are layer-by-layer (LBL) dip coating, spray coating, spin resulting membranes showed high selectivity for p-xylene over
coating, and vacuum filtration. There are several examples of the o-xylene (Section 4). Multilayer graphene membranes have been
formation of inorganic membranes from porous layered materi- fabricated by spin coating or spray coating of graphene oxide
als. MCM-22 flakes were exfoliated and then coated on α-alumina suspensions on copper substrates, followed by etching using
substrates using LBL dip coatings followed by TEOS vapor post- nitric acid (Nair et al., 2012). Also, a prototype single-sheet
treatment (Choi and Tsapatsis, 2010). The latter step was required porous graphene membrane has been fabricated by the combina-
in order to fill microscopic gaps (defects) between coated layers. tion of photolithography and mechanical exfoliation of graphene
Because the MCM-22 flakes have a high aspect ratio and sig- with a cellophane tape (Koenig et al., 2012). The subsequent UV-
nificant interlayer interactions, the 6MR pores of the successively oxidative etching step introduced sub-nanometer pores in the
coated MCM-22 layers were found to be approximately aligned graphene sheet.
 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924 917

3.3. Microstructural analysis of nanocomposite membranes morphological information on polymer/inorganic composite

membranes. The slope of SAXS/SANS patterns in the low momen-
An important aspect of nanocomposite membrane studies is tum transfer (Q) regions is a quantitative indication of the
the determination of the membrane microstructure. This is parti- dispersion and morphology of embedded 1D and 2D materials.
cularly challenging in the case of membranes containing 1D and The Q-dependence of single exfoliated 2D layers is expected to
2D porous materials. Microstructural characterization should have a power law exponent of  2, and the exponent is expected to
reveal the degree of exfoliation of 2D flakes, the dispersion quality become steeper as the degree of exfoliation and dispersion
of flakes or 1D nanotubular materials in the polymer matrix, their decreases (Kratky and Porod, 1949). For example, a MMT/PA66
degree of orientation, and the interface quality/adhesion between composite (whose TEM image is shown in the inset of Fig. 8a) has
the polymer and low-dimensional inorganic fillers. These char- a steeper exponent of  3 at low Q (Fig. 8a), due to the presence of
acteristics are intimately connected to the ultimate separation globular clusters of exfoliated clay flakes. Similarly, an AlPO/
performance of the nanocomposite membrane (Schaefer and polyimide composite membrane (Jeong, 2004) showed a slope of
Justice, 2007; Vaia and Maguire, 2007). Here we review some of  3.5, corresponding to multi-layer particles (10–100 nm in size).
the representative microstructural analyses on composite mem- Detailed scattering model fits quantitative morphological infor-
branes incorporating nanoporous 1D and 2D materials. mation. A variety of X-ray/neutron scattering model equations
Wide Angle X-ray Scattering (WAXS) can be used to determine have been developed based on the geometry of the inorganic
the d-spacings of layered materials dispersed in polymers, and it fillers, e.g., spheres, disks/layers, rods (Kratky and Porod, 1949;
can be a qualitative indication of the degree of exfoliation (Paul Wang et al., 2007). SAXS data from composite membranes con-
and Robeson, 2008). The crystallinity of the polymeric phase in the taining single-walled aluminosilicate nanotubes in PVA has been
composite membrane can be determined by deconvolution of the fitted in detail to a model of scattering from rod-like dispersed
WAXS patterns to determine the fractions of crystalline and objects to obtain morphological characteristics such as the average
amorphous polymer. Crystallinity changes in the polymer matrix intertubular distance and nanotubular particle dimensions (Kang
correlate with changes in the polymer chain packing, and can lead et al., 2012). This analysis, together with WAS patterns, clearly
to interesting modifications of the separation properties in addi- revealed a near-perfect dispersion of individual nanotubes in the
tion to the effect of the inorganic material (Kang, 2012). However, membrane even at high loadings (up to 40 vol%). The WAXS
the characteristic peaks seen in WAXS may depend strongly on patterns also revealed a decrease in the crystalline:amorphous
measurement conditions such as scanning time and detector type. ratio of the cellulose acetate phase with increasing nanotube
Misleading quantitative results can be obtained from WAXS loading, thereby indicating significant changes in the polymer
patterns, and they should be considered as qualitative information packing induced by the nanotube fillers. On the other hand, the
(Paul and Robeson, 2008). stacked-tactoid model is applicable to the case of composite
On the other hand, small angle X-ray and neutron scattering membranes containing 2D flakes (Ho et al., 2001; Hanley et al.,
(SAXS/SANS) measurements can give detailed microstructural 2003). The scattering model consists of the form factor and
information, and have been widely used for analyzing the micro- structure factor of the flake. The form factor includes information
structure and morphology of polymeric and composite materials. on flake dimensions and orientation, and the structure factor
To highlight a specific case of polymeric membranes, SAXS analysis describes the internal morphology of the flakes, such as the
gave detailed information on the structure (specifically, the size average number of layers in the flakes and the interlayer spacing.
and orientation of water nanochannels) of ionic polymers such as The structure factor can also be convoluted with probability
Nafions and sulfonated polyimides (Gebel and Diat, 2005; distributions such as Gaussian functions to capture the statistical
Schmidt-Rohr and Chen, 2008). The structure and morphology of variations in the number of layers and interlayer spacing in the
cellulose acetate membranes has also been studied by SANS, and sample. Fig. 8b (Kim et al., 2013) shows SAXS model fitting results
the results have compared well to transmission electron micro- from AMH-3/cellulose acetate composite membranes at different
scopy (TEM) analysis (Kulkarni et al., 1994). Beyond the character- AMH-3 loadings. The analysis demonstrated a high degree of
ization of polymer microstructure, SAXS/SANS analysis can give exfoliation of the AMH-3 flakes in relation to the initial swollen

Fig. 8. SANS analysis of (a) PA66/MMT composite material (adapted from Schaefer and Justice, 2007), and (b) SAXS model fitting to analyze the degree of exfoliation of
swollen AMH-3 in cellulose acetate (adapted from Kim et al., 2013). The inset table shows the average number of layers in the AMH-3 flakes obtained from the SAXS analysis
at different loadings of AMH-3 in the membrane.
918 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924

material (Fig. 7e). The dispersed AMH-3 flakes contain (on the investigated for gas/vapor-phase and liquid-phase separations
average) 4–8 layers. Further details of the statistical variation of (Cong et al., 2007). The incorporation of SWNTs and MWNTs in
the number of layers and the interlayer spacings were also brominated poly(2,6-biphenyl-1,4-phenylene oxide) polymeric
obtained. Similar analyses have been demonstrated for mem- membranes significantly improved the CO2 gas permeability while
branes containing clay platelets (Yoonessi et al., 2005). maintaining a high CO2/N2 selectivity (Fig. 9). Additionally, the
Other characterization techniques can also yield considerable incorporation of CNTs in the polymer matrix improved the
insight on the composite microstructure. 1H-NMR can be used to mechanical properties of the membrane, such as the tensile
trace the degree of exfoliation and alignment of layered materials modulus. The latter results were attributed to the favorable
in composite membranes. For example, Xu et al. (2009) measured interactions of the outer walls of the CNTs with the biphenyl
1
H spin-lattice relaxation times in clay/polypropylene composites groups of the polymer. Separation of water from ethanol has also
and thereby tracked the evolution of clay morphology (specifically, been investigated using CNT/PVA composite membranes (Fig. 10a)
the exfoliation and alignment of montmorillonite) induced by (Choi et al., 2007). The water flux increases while the water/
equibiaxial stretching of the composite. Polarized Raman spectro- ethanol separation factor decreases, as nanotube loading increases.
scopy has also been employed to deduce a preferential horizontal These results can be rationalized by the fact that CNTs are known
(in-plane) orientation of JDF-L1 titanosilicate layers in a copolyi- to allow very high fluxes (by virtue of their atomically smooth
mide membrane matrix (Galve et al., 2011). Raman spectra were walls) of many molecules such as water, other polar molecules,
measured with light sources polarized either parallel or perpen- and hydrocarbons (Majumder et al., 2011). This property also
dicular to the membrane plane. The comparison of vibrational renders them unselective for mixtures such as water/ethanol. On
Raman peak intensities of the polymer chains as well as the JDF-L1 the other hand, it has been shown recently that selectivity can be
layers under the two measurement conditions indicated a pre- imparted to CNTs by modifying their ends with functional groups.
ferred horizontal orientation for the JDF-L1 flakes. HRTEM is also a For example, the modification of CNT ends with zwitterions
useful method to examine the dispersion of layered materials in a (containing carboxylate and tertiary amine groups), followed by
polymer matrix. Its main drawback is that it can only image very the incorporation of the modified CNTs into a polyimide matrix,
small portions of the sample at a time and hence does not produce resulted in membranes that retain the expected high water flux
statistically valid results. Hence, it should be combined with more and also achieve approximately 99% rejection of salt ions (Chan
statistically sound techniques such as SAXS/SANS, NMR, and et al., 2013). The zwitterionic modifying groups allow water
Raman spectroscopy to develop a reliable picture of the nano- molecules to pass through but block the permeation of ions by
composite membrane microstructure. charge repulsion and steric hindrance.
The uniformity of nanotube dispersion in the polymer is also
important for performance enhancement. For example, carbon
4. Separation applications nanotube/polymer composite membranes containing aggregated

As mentioned earlier, the pore structures of nanoporous 1D and

2D materials are attractive for obtaining molecular separation
properties. A number of separations applications of membranes
containing these materials have been investigated. This section
highlights advances in gas separation, water purification, organics
purification, and proton conduction by membranes containing
nanoporous 1D and 2D materials. Throughout the following
discussion, different mechanisms of molecular separation operat-
ing in these materials are illustrated. Broadly speaking, the 1D
nanotubular materials provide molecular transport paths of
lengths greater than  10 nm, and hence display adsorption and
diffusion-based selectivity behavior similar to that observed in
nanoporous materials like zeolites and metal-organic frameworks
(MOFs). On the other hand, the  1 nm thin layers found in the 2D
nanoporous materials provide much shorter, discontinuous, mole-
cular transport paths that likely cannot be described in the same
manner as transport in zeolites and MOFs. This issue is highlighted
in Section 4.2 with several examples.

4.1. 1D nanotubular membranes

Several molecular simulation studies have indicated that nano-

tubular materials can be attractive candidates for gas separations
(Skoulidas et al., 2002; Sokhan et al., 2004; Matranga et al., 2006)
as well as liquid separations (Sokhan et al., 2002; Konduri et al.,
2008). A macroscopic model of permeation has also been recently
developed for membranes containing 1D nanotubular fillers, and
the effects of parameters such as the nanotube orientation
distribution, aspect ratio, volume fraction, and membrane defects,
have been assessed in detail via this model (Kang, 2011a, 2011b).
The above works have indicated that nanotube-containing mem-
branes are capable of significant enhancements in molecular flux Fig. 9. CO2/N2 separation performance of composite membranes incorporating
and potentially in selectivity over polymeric membranes. Mem- (a) SWNT, and (b) MWNT in a brominated poly(biphenylphenylene oxide) polymer
branes containing carbon SWNTs and MWNTs have been matrix. (Adapted from Cong et al., 2007).
 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924 919

that of the CNT/PVA membrane. These results are also likely to be

influenced by the change in the PVA packing characteristics upon
increasing loading of the NTs, which was characterized by XRD
and NMR techniques described earlier in this review.

4.2. 2D layered oxide membranes

The first gas separation data from polymer/selective flake

composite membranes was reported for the case of porous layered
AlPO/polyimide composite membranes (Jeong et al., 2004). The
composite membrane showed improvements in O2/N2 selectivity,
attributed to molecular sieving through the porous layered AlPO.
However, SANS analysis showed that the AlPO material was not
exfoliated in the polymer. Similarly, swollen AMH-3 was incorpo-
rated in polybenzimidazole (PBI), and showed moderate improve-
ments on overall CO2/CH4 selectivity attributed to molecular
sieving effect of the AMH-3 pores (Choi et al., 2008a, 2008b,
2008c). The layered titanosilicate JDF-L1 has also been incorpo-
rated in a copolyimide, and showed significant improvements
in H2/CH4 separation attributed to molecular sieving through its
0.3 nm pore openings (Galve et al., 2011). In a recent study, (Kim et
al., 2013) swollen AMH-3 was significantly exfoliated in cellulose
acetate (Section 3.1), and the nanocomposite AMH-3/CA mem-
brane system showed significantly improved CO2 permeability
while maintaining (but not improving) CO2/CH4 selectivity even at
low loadings (2–6 wt%) of swollen AMH-3 (Fig. 11). This result is
attributed to a complex transport mechanism involving transport
through the AMH-3 pores as well as the interlayer mesopores in
the partially exfoliated stacks of 4–8 AMH-3 layers. The MFI BMLM
material (Section 2.2) has also been incorporated into a polyimide
matrix and showed superior particle/polymer adhesion properties
over the conventional MFI material (Kim et al., 2012a, 2012b).
Fig. 10. Water/ethanol pervaporation performance of PVA membranes incorporat-
ing (a) CNTs (adapted from Choi et al., 2007) (b) and aluminosilicate nanotubes Several characterization techniques revealed the infiltration of the
(adapted from Kang et al., 2012). polyimide chains between the layers of the epitaxially grown MFI
layers, thereby enhancing the adhesion between the inorganic and
polymeric phases. This resulted in superior CO2/CH4 separation
nanotubes have low molecular selectivity. To address the problem characteristics over conventional MFI particles dispersed in the same
of increasing the nanotube loading in nanocomposites while polymer. All-inorganic membranes fabricated with layered zeolites
maintaining good dispersion, a range of techniques for outer have also been applied to various separations. The c-oriented
surface modification of carbon nanotubes have been developed inorganic membranes fabricated from MCM-22 flakes (with 6MR
(Dyke and Tour, 2004; Zhao and Stoddart, 2009; Clavé et al., 2013) pore openings) have been applied to H2 separations and showed
that can enhance the nanotube compatibility with the polymeric significant improvements on H2/N2 selectivity (Choi and Tsapatsis,
matrix. Nevertheless, the highest volume fraction reported to date 2010). A selective membrane fabricated from exfoliated MWW flakes
of CNTs dispersed in a polymeric material without significant showed highly improved selectivity in He/N2 separation attributed to
nanotube aggregation is only about 20% (Jain, 2010; Chan, 2013). molecular sieving through its 6MR pores (Varoon et al., 2011).
This limitation hinders the performance enhancement that CNTs Exfoliated layered MFI has been applied to xylene separations, by
can potentially create in a composite material or membrane. utilizing its 10MR pore openings (Varoon et al., 2011). A high p-/o-
However, individual dispersion of nanotubes in polar liquids can xylene separation factor of 65 was obtained.
be achieved in the case of single-walled metal oxide nanotubes The membrane applications of porous 2D materials have been
that have polar surfaces. As mentioned earlier, single-walled expanded to ion conducting membranes, such as proton exchange
aluminosilicate nanotubes (Fig. 2b) are synthesized hydrother- membranes for direct methanol fuel cells (DMFCs). The major
mally in water, and have a high degree of dispersion in aqueous components of a DMFC are the anode, the cathode and the proton
media. These SWNTs are hypothesized to be amenable to the exchange membrane (PEM). At the anode, methanol is converted
fabrication and application of high-loading nanotube composites to electrons, protons, and CO2 as a byproduct. The protons are
with near-ideal dispersion of nanotubes. Previous studies have transported through the PEM to the cathode, where they react
suggested that aluminosilicate NTs possess extraordinarily high with oxygen and electrons to produce water. The development of
interior hydrophilicity due to their high inner surface silanol higher-performance PEMs is important to improve the overall
densities, thereby leading to a high water/alcohol selectivity DMFC performance (Jagur-Grodzinski, 2007). Specifically, the PEM
(Zang et al., 2009, 2010). Single-walled aluminosilicate NT/poly- is required to have high proton conductivity, as well as low
vinylalcohol (PVA) composite membranes have been fabricated methanol permeability (to prevent transport of the fuel through
and applied to water/ethanol separation by pervaporation (Kang the PEM). The existence of mesoscopic water channels in PEMs
et al., 2012). As described earlier in this work, SAXS and WAXS made with Nafions and other sulfonated fluoropolymers (also called
analysis showed excellent dispersion of the NTs at high loadings ionomers) is important for obtaining good proton conductivity, but
(up to 40 vol%). An increased water flux was obtained (Fig. 10b) in they also provide transport paths for methanol crossover as a side
comparison to pure PVA, whereas the water/ethanol separation effect. Several works have reported the incorporation of clay materi-
factor (35–45) showed a decrease from PVA and was similar to als in ionomers (Jung et al., 2003; Rhee et al., 2005; Thomassin et al.,
920 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924

volume fractions and length:thickness aspect ratios, showed reason-

able agreement with finite-element simulations for flakes with aspect
ratios greater than about 5. However, an important assumption of both
the analytical model as well as the finite-element simulations is that a
bulk molecular diffusivity or permeability can be defined for the flake.
While this is an excellent assumption for inorganic fillers of bulk
materials like zeolites or metal-organic frameworks, and also for flakes
that contain a substantial number of layers, its validity is not clear for
exfoliated flakes of nanoscopic thickness. Since it is not clear whether
the small (1 nm) thickness of the flake allows the establishment of a
true random walk condition of the diffusing molecule, the definitions
of diffusivity and permeability in such thin flakes are not physically
obvious. Konduri and Nair constructed a molecular model of AMH-3
layers dispersed in a polydimethylsiloxane (PDMS) matrix and calcu-
lated the effective diffusivities of several gases through such compo-
sites (Konduri and Nair, 2007). It was found that the Cussler model
greatly underpredicted the effective diffusivity. Similarly, experimental
data on AMH-3/CA membranes could not be reconciled with the
model predictions (solid line in Fig. 11b), even in the limit of the
highest possible permeability predicted by the model (Kim et al.,
2013). It has been suggested by the above authors that membranes
containing nanoscopically thin flakes might be better modeled as
nanoscale composite structures, and cannot be described by analytical
models that assume bulk permeation behavior in the flakes.

Fig. 11. (a) CO2/CH4 selectivity and CO2 permeability at different AMH-3 loadings,
and (b) CO2 permeability as a function of AMH-3 loading in AMH-3/cellulose
acetate (CA) composite membranes at a pressure differential of 65 psi. The solid
blue line is the highest possible CO2 permeability predicted by the Cussler model
(adapted from Kim et al., 2013).

2006). However, clays are non-porous layered materials that act as

barriers for both protons and methanol, resulting in a trade-off
between reduction in methanol crossover and maintenance of proton
conductivity. Porous layered oxides may be good alternative materi-
als, as they can permeate the protons through their subnanometer
pores. For example, swollen AMH-3 flakes have been incorporated in
Nafions membranes (Hudiono et al., 2009; Kim et al., 2012a, 2012b).
A significant (30–50%) decrease in methanol permeability was
observed, whereas high proton conductivity was maintained. SAXS
measurements also revealed that the incorporation of AMH-3 sub-
stantially altered the microstructure of the ionomer (Hudiono et al.,
2009). The mechanisms responsible for the improvements in perfor-
mance (such as molecular sieving through the pores of the layered
material, or alteration of the transport properties of the water
channels) have not yet been fully clarified.
To predict the effects of incorporation of layered materials on the
separation performance of composite membranes, Cussler developed
an analytical model for the calculation of permeability and selectivity
of gas molecules in composite membranes containing selective flakes
(Cussler, 1990). The limits of applicability of this model have been
evaluated in detail via finite-element simulations of membrane Fig. 12. Comparison of gas permeation rates of pristine and porous graphene
membranes: (a) maximum deflection of the membrane surface versus time before
matrices containing flake-like domains in ordered arrangements and after etching, and (b) gas permeance versus molecular size, revealing selective
(Sheffel and Tsapatsis, 2007, 2009). It was found that the Cussler permeation of H2 and CO2 over larger gas molecules after etching (adapted from
model, including modifications that allow a wider range of flake Koenig et al., 2012).
 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924 921

4.3. Graphene and other carbon membranes a very low He permeability in comparison to both a barrier polymer
like PET (currently used as a commercial packaging material) as well
The first experimental evidence of gas separation by porous as 1-mm-thick glass. However, the water permeability of the
graphene was obtained via a ‘pressurized blister’ test with an atom- membrane was found to be extremely high (Fig. 13), and was
ically thin membrane (Koenig et al., 2012). Ultra-violet oxidative described as ‘unimpeded permeation of water’. This unusual prop-
etching was used to introduce pores in a fabricated non-porous erty, observed to varying extents with molecules like water, ethanol,
graphene membrane. The deformation of the film surface due to gas and hexane, could be explained by a nanoscale capillary effect
pressure on the feed side was measured to quantify the permeability, whereby monolayers of water molecules form between graphene
with a higher deflection indicating a lower permeability due to the gas layers. This effect was also demonstrated by atomistic simulations.
molecules not being able to pass the graphene barrier. The etched While graphene-based membranes are actively being developed,
porous graphene membrane was found to be selective (Fig. 12) for H2 other carbon-based membranes have also shown potentially interest-
and CO2 over larger molecules like Ar, N2, and CH4, hence suggesting ing properties. Diamond-like carbon (DLC) porous nanosheets (10–
the presence of pores slightly larger than 0.3 nm in size. 40 nm thin) have been recently fabricated by plasma chemical vapor
Nair et al. reported very rapid permeation of water through deposition, and allowed ultrafast viscous permeation of organic
a ‘leak-tight’, non-porous multilayer graphene oxide membrane (Nair solvents (Karan et al., 2012). The sub-nanometer pore size of the
et al., 2012). The membrane was determined to be non-porous due to DLC membrane allowed selective organic solvent separation from
larger solute molecules. While carbon molecular sieve (CMS) mem-
branes produced by pyrolysis have been known for several decades
and currently have more optimal transport properties than graphene
and DLC membranes, they have some limitations such as a rather low
permeability, and fragility especially in oxygen-containing environ-
ments (Ismail et al., 2011). The new developments on graphene and
DLC membranes, while currently not as advanced as other 2D and 1D
materials in fabricating functional and cost-effective membranes, may
be promising future alternatives for applications in both gas and liquid
separations. Different mechanisms for molecular separations in these
membranes have been illustrated recently (Paul, 2012).

5. Outlook and challenges

Table 1 shows a (non-exhaustive) summary of the separation

performance of membranes containing nanoporous 1D and 2D
materials. It is seen that such membranes can perform highly
selective separations in many industrially interesting applications,
and that the permeation rates can be attractively high. In parti-
Fig. 13. Permeation of water, ethanol, and hexane through a graphene oxide
cular, Table 1 shows that membranes containing different nano-
membrane (adapted from Nair et al., 2012). porous 1D and 2D materials are capable of separating a wide range

Table 1
Summary of separation performance of membranes utilizing 1D nanotubular and 2D nanoporous layered materials.

Membrane system Separation Permeability or permeance (of faster species) Selectivity 1D or 2D material wt%

Gas-phase separations
CNT/BPPOdp (Cong, 2007) CO2/N2 150 Barrer 30 4–9
AlPO/Polyimide (Jeong et al., 2004) CO2/CH4 28.3 Barrer 41 10
SAMH-3/PBI (Choi et al., 2008a, 2008b, 2008c) H2/CO2 1.5 Barrer 30 2
SAMH-3/CA (Kim et al., 2013) CO2/CH4 12 Barrer 30 2–6
JDF-L1/Copolyimide (Galve et al., 2011) H2/CH4 150 Barrer 36 10
MCM-22 (Choi and Tsapatsis, 2010) H2/N2 4  10  8 mol/m2 s Pa 120 100
Exfoliated MWW (Varoon et al., 2011) He/N2 5  10  9 mol/m2 s Pa 17 100
Porous graphene (Koenig et al., 2012) CO2/CH4 1.5  10  12 mol/m2 s Pa 10,000 100

Vapor-phase and pervaporative separations

Aluminosilicate NT/PVA (Kang et al., 2012) H2O/EtOH 160,000 Barrer 35 40
Carbon NT/PVA (Choi et al., 2007) H2O/EtOH 1.1  10  8 mol/m2 s Pa 49 5
Exfoliated Layered MFI (Varoon et al., 2011) p-/o-xylene 3  10  7 mol/m2 s Pa 70 100

Porous graphene oxide (Nair et al., 2012) H2O/He 1  10  6 Barrer 41010 100

Membrane system Separation H2O flux Salt rejection 1D material wt%

Water separation
CNT/polyamide (Chan et al., 2013) H2O/NaCl 28.7 gal/ft2-day 98.6% 20

Membrane system Separation H þ conductivity Selectivity (S-s/cm3) 2D material wt%

Selective proton conduction

AMH-3/Nafion (Hudiono et al., 2009) H þ /MeOH 4.2  10  2 S/cm 7.4  104 5
AMH-3 on Nafion (Kim et al., 2012) H þ /MeOH 3.8  10  2 S/cm 2.2  104 100
922 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924

of molecular species ranging from protons, small gas molecules, ted flakes, partially exfoliated/intercalated flakes, and broken/
water, polar organic molecules, and larger nonpolar/hydrocarbon damaged flakes or particles. Similarly, processing of nanotubes
molecules (such as xylene isomers). The foregoing discussion has may lead to products containing a mixture of well dispersed
also shown that the transport mechanisms in these materials span nanotubes, aggregated bundles, and damaged nanotubes. Optimi-
not only the typical adsorption and diffusion phenomena seen in zation of these processes will be required to produce high-quality
conventional nanoporous materials such as zeolites, but also exfoliated or dispersed 1D and 2D materials ready for membrane
reveal new features arising from the unique nanoscopic dimen- formation. The synthesis of pre-swollen or pre-exfoliated layered
sions of the present 1D and 2D materials. Finally, the collection of materials such as layered MFI, or well-dispersed aluminosilicate
materials discussed in this work already shows a wide spectrum of nanotubes in aqueous media, are initial examples described earlier
pore sizes (0.3–1 nm), which appears quite amenable to tailoring in this review.
as well as range expansion via ongoing synthesis of new or Another important challenge is the development of reliable
modified 1D and 2D nanoporous materials. structure-property relationships in these unique membrane systems
Hence, it is reasonable to hypothesize that a large array of (both by experimental characterization as well as by modeling and
industrially attractive molecular separations can be targeted by simulation), since they possess structural complexity over multiple
membranes containing such materials. Because these materials length scales. At the nanoscopic level, there is now a diverse range of
can be synthesized or post-processed to obtain nanoscopic dimen- nanotubular and selective flake materials that have different pore
sions (e.g., nanotubes as short as 10 nm, and layers/flakes as thin as structures, elemental compositions, and dimensions. There is cur-
1 nm), they can be incorporated in very thin separating membrane rently little understanding of the mechanisms of selective separa-
layers (down to 100 nm or thinner), leading to higher throughput tions by nanoporous 2D and 1D materials beyond the hypothesis of
rates in comparison to more conventional porous membrane ‘molecular sieving’, based upon comparison of the nominal pore size
materials. As described in this review, a number of challenging of the material with the kinetic diameters of permeating molecules.
synthesis and membrane processing issues have already been Additionally, the question of finding a polymeric matrix transport
addressed via innovative materials synthesis chemistry and com- properties that appropriately ‘match’ the nanoscopic materials and
posite fabrication strategies. With the exception of the AlPO gives the highest overall composite membrane performance in a
materials discussed briefly in Section 2.2, most of the 1D and 2D desired separation, remains unaddressed. Finally, the long-term
materials highlighted in this review show good stability and stability characteristics of these nanocomposite membrane architec-
retention of their nanopore structures under processing condi- tures have not yet been thoroughly evaluated, and very little data on
tions. These events, as well as the discovery of novel mechanisms this topic currently exists. Although recent examples of development
of selective permeation in 1D and 2D materials such as carbon of structure-property relations by experimental (e.g., SAXS/HRTEM/
nanotubes and graphene, create an optimistic outlook that mem- NMR) and computational (molecular modeling, finite-element simu-
branes containing 1D and 2D nanoporous materials may competi- lations of microstructure) have appeared in the literature as dis-
tively address the many application opportunities available for cussed earlier in this review, significant work remains to be done to
next-generation separation technologies. reach the stage of rationally designing or selecting nanoporous 1D
However, several challenges remain on the road to industrially and 2D materials for use in membrane separations.
scalable membranes incorporating nanoporous 1D and 2D materi-
als. Effective strategies and platforms for membrane scale-up will
be a critically important issue. Among the membrane types Acknowledgment
discussed in this review, those containing 2D layered materials
can be considered as being furthest along the path to scale-up,
This publication is based on work supported by Award no.
since their processing strategies (as reviewed in this work) are
KUS-I1-011-21, made by King Abdullah University of Science and
seen to be compatible with polymeric membrane processing
Technology (KAUST).
methods. For example, the incorporation of surface-modified
nanoplatelets of the layered silicate laponites into the skin layers
of polyimide hollow fibers has been demonstrated (Johnson and References
Koros, 2009), thereby allowing the hypothesis that porous 2D
materials can be processed in a similar manner into scalable Adams, R.T., et al., 2011. CO2–CH4 permeation in high zeolite 4A loading mixed
membranes. However, it is yet unclear whether the techniques matrix membranes. J. Membr. Sci. 367 (1–2), 197–203.
for inducing preferred orientation of 2D or 1D materials (as Allen, M.J., et al., 2010. Honeycomb carbon: A review of graphene. Chem. Rev. 110
(1), 132–145.
discussed earlier in this review) will also be compatible with Bae, T.H., et al., 2009. Facile high-yield solvothermal deposition of inorganic
membrane scale-up processes such as hollow fiber fabrication. In nanostructures on zeolite crystals for mixed matrix membrane fabrication.
conjunction with the goal of membrane scale-up, there are two J. Am. Chem. Soc. 131 (41), 14662–14663.
Bernardo, P., et al., 2009. Membrane gas separation: a review/state of the art. Ind.
other important challenges. The first is regarding the scale-up of Eng. Chem. Res. 48 (10), 4638–4663.
materials synthesis. Most of the materials described in this review Bieri, M., et al., 2009. Porous graphenes: two-dimensional polymer synthesis with
have only been synthesized reproducibly in quantities of a few atomic precision. Chem. Commun. 45, 6919–6921.
Blankenburg, S., et al., 2010. Porous graphene as an atmospheric nanofilter. Small 6
grams, a fact that prevents their testing in preliminary scale-up (20), 2266–2271.
efforts such as hollow fiber membrane fabrication (which typically Bottero, I., et al., 2011. Synthesis and characterization of hybrid organic/inorganic
requires batches of  100 g for a single experiment). However, due nanotubes of the imogolite type and their behaviour towards methane
adsorption. Phys. Chem. Chem. Phys. 13 (2), 744–750.
to the similarity of synthesis techniques of materials such as Bunch, J.S., et al., 2007. Electromechanical resonators from graphene sheets. Science
aluminosilicate nanotubes or layered zeolites in comparison to 315 (5811), 490–493.
well-known zeolite materials that are in commercial uses, there is Caro, J., Noack, M., 2008. Zeolite membranes – recent developments and progress.
Microporous and Mesoporous Mater. 115 (3), 215–233.
reason to believe that materials scale-up can be achieved in a
Centi, G., Perathoner, S., 2008. Catalysis by layered materials: a review. Microporous
similar manner to that of zeolites. The second challenge is to and Mesoporous Mater. 107 (1-2), 3–15.
achieve a high uniformity of exfoliation and other processing Chamberlain, T.W., et al., 2011. Reactions of the inner surface of carbon nanotubes
steps. In their current state, exfoliation processes such as high- and nanoprotrusion processes imaged at the atomic scale. Nat. Chem. 3 (9),
732–737.
shear mixing or melt-blending of porous layered materials result Chan, W.-F., et al., 2013. Zwitterion functionalized carbon nanotube/polyamide
in significantly polydisperse products containing well exfolia- nanocomposite membranes for water desalination. ACS Nano 7 (6), 5308–5319.
 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924 923

Chen, W., Tao, X., 2005. Self‐organizing alignment of carbon nanotubes in thermo- Johnson, J.R., Koros, W.J., 2009. Utilization of nanoplatelets in organic-inorganic
plastic polyurethane. Macromol. Rapid Commun. 26 (22), 1763–1767. hybrid separation materials: Separation advantages and formation challenges.
Cheung, C.L., et al., 2002. Diameter-controlled synthesis of carbon nanotubes. J. Taiwan Inst. Chem. Eng. 40 (3), 268–275.
J. Phys. Chem. B 106 (10), 2429–2433. Jung, D.H., et al., 2003. Preparation and performance of a Nafion (R)/montmor-
Choi, J., et al., 2009a. Grain boundary defect elimination in a zeolite membrane by illonite nanocomposite membrane for direct methanol fuel cell. J. Power Sour.
rapid thermal processing. Science 325 (5940), 590–593. 118 (1–2), 205–211.
Choi, J., Tsapatsis, M., 2010. MCM-22/silica selective flake nanocomposite mem- Kang, D.-Y., et al., 2011a. Single-walled aluminosilicate nanotubes with organic-
branes for hydrogen separations. J. Am. Chem. Soc. 132 (2), 448–449. modified interiors. J. Phys. Chem. C 115 (15), 7676–7685.
Choi, J.H., et al., 2007. Modification of performances of various membranes using Kang, D.Y., et al., 2011b. Modeling molecular transport in composite membranes
MWNTs as a modifier. Macromol. Symposia 249, 610–617. with tubular fillers. J. Membr. Sci. 381 (1–2), 50–63.
Choi, M., et al., 2009b. Stable single-unit-cell nanosheets of zeolite MFI as active Kang, D.Y., et al., 2012. Single-walled aluminosilicate nanotube/poly(vinyl alcohol)
and long-lived catalysts. Nature 461, 246–249. nanocomposite membranes. ACS Appl. Mater. Interfaces 4 (2), 965–976.
Choi, S., et al., 2008a. Layered silicates by swelling of AMH-3 and nanocomposite Karan, S., et al., 2012. Ultrafast viscous permeation of organic solvents through
membranes. Angew. Chem.-Int. Edition 47 (3), 552–555. diamond-like carbon nanosheets. Science 335 (6067), 444–447.
Choi, S., et al., 2008b. Layered silicate by proton exchange and swelling of AMH-3. Kim, J., et al., 2012a. Delamination of microporous layered silicate by acid-
Microporous Mesoporous Mater. 115 (1-2), 75–84. hydrothermal treatment and its use for reduction of methanol crossover in
Choi, S.H., et al., 2008c. Fabrication and gas separation properties of polybenzimi- DMFC. Microporous Mesoporous Mater. 168, 148–154.
dazole (PBI)/nanoporous silicates hybrid membranes. J. Membr. Sci. 316 (1-2), Kim, S., et al., 2007. Scalable fabrication of carbon nanotube/polymer nanocompo-
145–152. site membranes for high flux gas transport. Nano Lett. 7 (9), 2806–2811.
Choudalakis, G., Gotsis, A., 2009. Permeability of polymer/clay nanocomposites: a Kim, W.-G., et al., 2013. Nanoporous layered silicate AMH-3/cellulose acetate
review. Eur. Polym. J. 45 (4), 967–984. nanocomposite membranes for gas separations. J. Membr. Sci. 441, 129–136.
Clavé, G., et al., 2013. Functionalization of carbon nanotubes through polymeriza- Kim, W., et al., 2012b. Epitaxially grown layered MFI-bulk MFI hybrid zeolitic
tion in micelles: a Bridge between the covalent and non-covalent methods. materials. ACS Nano 6 (11), 9978–9988.
Chem. Mater. 25 (13), 2700–2707. Kim, W.G., et al., 2011. Swelling, functionalization, and structural changes of
Cong, H.L., et al., 2007. Carbon nanotube composite membranes of brominated poly the nanoporous layered silicates AMH-3 and MCM-22. Langmuir 27 (12),
(2,6-diphenyl-1,4-phenylene oxide) for gas separation. J. Membr. Sci. 294 (1-2), 7892–7901.
178–185. Kloprogge, J.T., 1998. Synthesis of smectites and porous pillared clay catalysts: a
Corma, A., et al., 2000. New aluminosilicate and titanosilicate delaminated review. J. Porous Mater. 5 (1), 5–41.
materials active for acid catalysis, and oxidation reactions using H2O2. Koenig, S.P., et al., 2012. Selective molecular sieving through porous graphene.
J. Am. Chem. Soc. 122 (12), 2804–2809. Nature Nanotechnol. 7 (11), 728–732.
Corma, A., et al., 2001. ITQ-18 a new delaminated stable zeolite. Chem. Commun. Konduri, S., Nair, S., 2007. A computational study of gas molecule transport in a
24, 2642–2643. polymer/nanoporous layered silicate nanocomposite membrane material.
Corma, A., et al., 1998. Delaminated zeolite precursors as selective acidic catalysts. J. Phys. Chem. C 111 (5), 2017–2024.
Nature 396 (6709), 353–356. Konduri, S., et al., 2008. Water in single-walled aluminosilicate nanotubes: diffu-
sion and adsorption properties. J. Phys. Chem. C 112 (39), 15367–15374.
Cussler, E.L., 1990. Membranes containing selective flakes. J. Membr. Sci. 52 (3),
Koros, W.J., 2004. Evotving beyond the thermat age of separation processes:
275–288.
membranes can lead the way. AIChE J. 50 (10), 2326–2334.
Dasgupta, S., Torok, B., 2008. Application of clay catalysts in organic synthesis.
Kratky, O., Porod, G., 1949. Diffuse small-angle scattering of X-rays in colloid
A review. Org. Prep. Proc. Int. 40 (1), 1–65.
systems. J. Colloid Sci. 4 (1), 35–70.
Davis, M.E., 2002. Ordered porous materials for emerging applications. Nature 417
Krikorian, V., Pochan, D.J., 2003. Poly (L-lactic acid)/layered silicate nanocomposite:
(6891), 813–821.
fabrication, characterization, and properties. Chem. Mater. 15 (22), 4317–4324.
De Heer, W.A., et al., 1995. Aligned carbon nanotube films – production and optical
Kulkarni, S., et al., 1994. Investigation of the pore structure and morphology of
and electronic-properties. Science 268 (5212), 845–847.
cellulose-acetate membranes using small-angle neutron-scattering. 2. Ultrafil-
Diaz, I., et al., 2004. Surface structure of zeolite (MFI) crystals. Chem. Mater. 16 (25),
tration and reverse-osmosis membranes. Macromolecules 27 (23), 6785–6790.
5226–5232.
Lee, S.S., et al., 2005. Exfoliation of layered silicate facilitated by ring-opening
Dyke, C.A., Tour, J.M., 2004. Covalent functionalization of single-walled carbon
reaction of cyclic oligomers in PET-clay nanocomposites. Polymer 46 (7),
nanotubes for materials applications. J. Phys. Chem. A 108 (51), 11151–11159.
2201–2210.
Fischbein, M.D., Drndic, M., 2008. Electron beam nanosculpting of suspended
Leonowicz, M.E., et al., 1994. Mcm-22 – a molecular-sieve with 2 independent
graphene sheets. Appl. Phys. Lett. 93 (11), 113107. (: art. #).
multidimensional channel systems. Science 264 (5167), 1910–1913.
Galve, A., et al., 2011. Copolyimide mixed matrix membranes with oriented
Li, J.Y., et al., 1999. Structures and templating effect in the formation of 2D layered
microporous titanosilicate JDF-L1 sheet particles. J. Membr. Sci. 370 (1–2),
aluminophosphates with Al3P4O163-stoichiometry. Chem. Mater. 11 (9),
131–140.
2600–2606.
Gascon, J., et al., 2012. Practical approach to zeolitic membranes and coatings: state Li, Y.F., et al., 2010. Two-dimensional polyphenylene: experimentally available
of the art, opportunities, barriers, and future perspectives. Chem. Mater. 24 porous graphene as a hydrogen purification membrane. Chem. Commun. 46
(15), 2829–2844. (21), 3672–3674.
Gebel, G., Diat, O., 2005. Neutron and x-ray scattering: suitable tools for studying Lin, Y.F., et al., 2007. A novel composite membranes based on sulfonated mon-
ionomer membranes. Fuel Cells 5 (2), 261–276. tmorillonite modified Nafion((R)) for DMFCs. J. Power Sources 168 (1), 162–166.
Goh, P., et al., 2013. Carbon nanotubes for desalination: performance evaluation and Liu, H.T., et al., 2009. Photochemical reactivity of graphene. J. Am. Chem. Soc. 131
current hurdles. Desalination 308, 2–14. (47), 17099–17100.
Gorgojo, P., et al., 2011. Direct exfoliation of layered zeolite Nu-6(1). Microporous Liu, P., 2007. Polymer modified clay minerals: a review. Appl. Clay Sci. 38 (1–2), 64–76.
Mesoporous Mater. 142 (1), 122–129. Mahajan, R., Koros, W.J., 2002. Mixed matrix membrane materials with glassy
Hanley, H.J.M., et al., 2003. A small-angle neutron scattering study of a commercial polymers. Part 1. Polym. Eng. Sci.42 (7), 1420–1431.
organoclay dispersion. Langmuir 19 (14), 5575–5580. Majumder, M., et al., 2011. Mass transport through carbon nanotube membranes in
He, Y.J., et al., 1998. Synthesis, characterization and catalytic activity of the pillared three different regimes: ionic diffusion and gas and liquid flow. ACS Nano 5 (5),
molecular sieve MCM-36. Microporous Mesoporous Mater. 25 (1–3), 207–224. 3867–3877.
Ho, D.L., et al., 2001. Characterization of organically modified clays using scattering Matranga, C., et al., 2006. Raman spectroscopic investigation of gas interactions with
and microscopy techniques. Chem. Mater. 13 (5), 1923–1931. an aligned multiwalled carbon nanotube membrane. Langmuir 22 (3), 1235–1240.
Hudiono, Y., et al., 2009. Porous layered oxide/Nafions nanocomposite membranes Mauter, M.S., et al., 2010. Nanocomposites of vertically aligned single-walled
for direct methanol fuel cell applications. Microporous Mesoporous Mater. 118 carbon nanotubes by magnetic alignment and polymerization of a lyotropic
(1-3), 427–434. precursor. ACS Nano 4 (11), 6651–6658.
Ismail, A., et al., 2009. Transport and separation properties of carbon nanotube- Merkel, T.C., et al., 2002. Ultrapermeable, reverse-selective nanocomposite mem-
mixed matrix membrane. Sep. Purif. Technol. 70 (1), 12–26. branes. Science 296 (5567), 519–522.
Ismail, A.F., et al., 2011. Carbon-Based Membranes for Separation Processes. Messersmith, P.B., Giannelis, E.P., 1994. Synthesis and characterization of layered
Springer, New York. silicate-epoxy nanocomposites. Chem. Mater. 6 (10), 1719–1725.
Jagur-Grodzinski, J., 2007. Polymeric materials for fuel cells: concise review of Moermans, B., et al., 2000. Incorporation of nano-sized zeolites in membranes.
recent studies. Polym. Adv. Technol. 18 (10), 785–799. Chem. Commun. 24, 2467–2468.
Jain, R., et al., 2010. Processing, structure, and properties of PAN/MWNT composite Mukherjee, S., et al., 2005. Phenomenology of the growth of single-walled
fibers. Macromol. Mater. Eng. 295 (8), 742–749. aluminosilicate and aluminogermanate nanotubes of precise dimensions.
Jeazet, H.B.T., et al., 2012. Metal–organic frameworks in mixed-matrix membranes Chem. Mater. 17 (20), 4900–4909.
for gas separation. Dalton Trans. 41 (46), 14003–14027. Mukherjee, S., et al., 2007. Short, highly ordered, single-walled mixed-oxide
Jeong, H.K., et al., 2004. Fabrication of polymer/selective-flake nanocomposite nanotubes assemble from amorphous nanoparticles. J. Am. Chem. Soc. 129
membranes and their use in gas separation. Chem. Mater. 16 (20), 3838–3845. (21), 6820–6826.
Jeong, H.K., et al., 2003. A highly crystalline layered silicate with three- Na, K., et al., 2010. Pillared MFI zeolite nanosheets of a single-unit-cell thickness.
dimensionally microporous layers. Nat. Mater. 2 (1), 53–58. J. Am. Chem. Soc. 132 (12), 4169–4177.
Jiang, D.E., et al., 2009. Porous graphene as the ultimate membrane for gas Nair, R.R., et al., 2012. Unimpeded permeation of water through helium-leak-tight
separation. Nano Lett. 9 (12), 4019–4024. graphene-based membranes. Science 335 (6067), 442–444.
924 W. Kim, S. Nair / Chemical Engineering Science 104 (2013) 908–924

Nasir, R., et al., 2013. Material advancements in fabrication of mixed-matrix Stern, S.A., et al., 1989. Structure permeability relationships of polyimide mem-
membranes. Chem. Eng. & Technol. 36 (5), 717–727. branes – applications to the separation of gas-mixtures. J. Polym. Sci. Part B-
Novoselov, K.S., et al., 2004. Electric field effect in atomically thin carbon films. Polym. Phys. 27 (9), 1887–1909.
Science 306 (5696), 666–669. Tenne, R., Seifert, G., 2009. Recent progress in the study of inorganic nanotubes and
Pandey, P., Chauhan, R.S., 2001. Membranes for gas separation. Prog. Polym. Sci. 26 fullerene-like structures. Ann. Rev. Mater. Res. 39, 387–413.
(6), 853–893. Thill, A., et al., 2012a. How the diameter and structure of (OH) 3Al2O3SixGe1–xOH
Park, J.Y., Paul, D.R., 1997. Correlation and prediction of gas permeability in glassy imogolite nanotubes are controlled by an adhesion versus curvature competi-
polymer membrane materials via a modified free volume based group tion. J. Phys. Chem. C 116 (51), 26841–26849.
contribution method. J. Membr. Sci. 125 (1), 23–39. Thill, A., et al., 2012b. Physico-chemical control over the single-or double-wall
Paul, D.R., 2012. Creating new types of carbon-based membranes. Science 335 structure of aluminogermanate imogolite-like nanotubes. J. Am. Chem. Soc. 134
(6067), 413–414. (8), 3780–3786.
Paul, D.R., Robeson, L.M., 2008. Polymer nanotechnology: nanocomposites. Polymer Thomas, J.M., et al., 2001. Molecular sieve catalysts for the regioselective and
49 (15), 3187–3204. shape-selective oxyfunctionalization of alkanes in air. Acc. Chem. Res. 34 (3),
Pavlidou, S., Papaspyrides, C.D., 2008. A review on polymer-layered silicate 191–200.
nanocomposites. Prog. Polym. Sci. 33 (12), 1119–1198. Thomassin, J.M., et al., 2006. Improvement of the barrier properties of nafion (R) by
Pendergast, M.M., Hoek, E.M., 2011. A review of water treatment membrane fluoro-modified montmorillonite. Solid State Ionics 177 (13–14), 1137–1144.
nanotechnologies. Energy Environ. Sci. 4 (6), 1946–1971. Tkalya, E.E., et al., 2012. The use of surfactants for dispersing carbon nanotubes and
Pinnau, I., He, Z.J., 2004. Pure- and mixed-gas permeation properties of poly- graphene to make conductive nanocomposites. Curr. Opin. Colloid Interface Sci.
dimethylsiloxane for hydrocarbon/methane and hydrocarbon/hydrogen separa- 17 (4), 225–232.
tion. J. Membr. Sci. 244 (1–2), 227–233. Tung, V.C., et al., 2009. High-throughput solution processing of large-scale
Pinnavaia, T.J., 1983. Intercalated clay catalysts. Science 220 (4595), 365–371. graphene. Nat. Nanotechnol. 4 (1), 25–29.
Prolongo, S.G., et al., 2013. In situ processing of epoxy composites reinforced with Urban, K.W., 2011. Electron microscopy: the challenges of graphene. Nat. Mater.
graphene nanoplatelets. Composites Sci. Technol. 86, 185–191. 10 (3), 165–166.
Pujari, S., et al., 2009. Orientation dynamics in multiwalled carbon nanotube Usuki, A., et al., 1993. Synthesis of nylon 6-clay hybrid. J. Mater. Res. 8 (5),
dispersions under shear flow. J. Chem. Phys.130 (21), 214903. 1179–1184.
Rao, C.N.R., Govindaraj, A., 2009. Synthesis of inorganic nanotubes. Adv. Mater. Vaia, R.A., Maguire, J.F., 2007. Polymer nanocomposites with prescribed morphol-
21 (42), 4208–4233. ogy: going beyond nanoparticle-filled polymers. Chem. Mater. 19 (11),
Ray, S.S., Bousmina, M., 2005. Biodegradable polymers and their layered silicate 2736–2751.
nano composites: In greening the 21st century materials world. Prog. Mater. Vaisman, L., et al., 2006. The role of surfactants in dispersion of carbon nanotubes.
Sci. 50 (8), 962–1079. Adv. Colloid Interface Sci. 128, 37–46.
Ray, S.S., Okamoto, M., 2003. Polymer/layered silicate nanocomposites: a review Vane, L.M., 2005. A review of pervaporation for product recovery from biomass
from preparation to processing. Prog. Polym. Sci. 28 (11), 1539–1641. fermentation processes. J. Chem. Technol. Biotechnol. 80 (6), 603–629.
Rhee, C.H., et al., 2005. Nafion/sulfonated montmorillonite composite: a new Vane, L.M., et al., 2008. Hydrophobic zeolite-silicone rubber mixed matrix mem-
concept electrolyte membrane for direct methanol fuel cells. Chem. Mater. branes for ethanol–water separation: effect of zeolite and silicone component
17 (7), 1691–1697. selection on pervaporation performance. J. Membr. Sci. 308 (1-2), 230–241.
Roberts, M.A., et al., 1996. Synthesis and structure of a layered titanosilicate catalyst Varoon, K., et al., 2011. Dispersible exfoliated zeolite nanosheets and their
with five-coordinate titanium. Nature 381, 401–404. application as a selective membrane. Science 333 (6052), 72–75.
Robeson, L.M., 2008. The upper bound revisited. J. Membr. Sci. 320 (1-2), 390–400. Vinh-Thang, H., Kaliaguine, S., 2013. Predictive models for mixed-matrix membrane
Roy, N., et al., 2012. Modifications of carbon for polymer composites and performance: a review. Chem. Rev. 113 (7), 4980–5028.
nanocomposites. Prog. Polym. Sci. 37 (6), 781–819. Vu, D.Q., et al., 2003. Mixed matrix membranes using carbon molecular sieves – I.
Rubio, C., et al., 2010. Exfoliated titanosilicate material UZAR-S1 obtained from preparation and experimental results. J. Membr. Sci. 211 (2), 311–334.
JDF-L1. Eur. J. Inorg. Chem., 159–163. Wang, B.N., et al., 2007. Characterizing the morphologies of mechanically manipu-
Schaefer, D.W., Justice, R.S., 2007. How nano are nanocomposites? Macromolecules lated multiwall carbon nanotube films by small-angle X-ray scattering. J. Phys.
40 (24), 8501–8517. Chem. C 111 (48), 17933–17940.
Schmidt-Rohr, K., Chen, Q., 2008. Parallel cylindrical water nanochannels in nafion Williams, I.D., et al., 1996. Organo-template control of inorganic structures: a low-
fuel-cell membranes. Nat. Mater. 7 (1), 75–83. symmetry two-dimensional sheet aluminophosphate 3[NH3CHMeCH2NH3]
Schrier, J., 2010. Helium separation using porous graphene membranes. J. Phys. [Al6P8O32]. H2O. Chem. Commun. 15, 1781–1782.
Chem. Lett. 1 (15), 2284–2287. Williams, I.D., et al., 1997. New chain architecture for a one-dimensional alumino-
Schultz, J., Peinemann, K.V., 1996. Membranes for separation of higher hydrocar- phosphate, [H3NCH2CH2NH3][AlP2O8H]. Chem. Commun. 14, 1273–1274.
bons from methane. J. Membr. Sci. 110 (1), 37–45. Xu, B., et al., 2009. Evolution of clay morphology in polypropylene/montmorillonite
Shah, M., et al., 2012. Current status of metal–organic framework membranes for nanocomposites upon equibiaxial stretching: a solid-state NMR and TEM
gas separations: promises and challenges. Ind. Eng. Chem. Res. 51 (5), approach. Macromolecules 42 (22), 8959–8968.
2179–2199. Yan, W.F., et al., 2000. A novel open-framework aluminophosphate [AlP2O6(OH)2]
Shao, Y.Y., et al., 2010. Graphene based electrochemical sensors and biosensors: a [H2O] containing propeller-like chiral motifs. Chem. Commun. 15, 1431–1432.
review. Electroanalysis 22 (10), 1027–1036. Yoonessi, M., et al., 2005. Clay delamination in clay/poly(dicyclopentadiene)
Sheffel, J.A., Tsapatsis, M., 2007. A model for the performance of microporous mixed nanocomposites quantified by small angle neutron scattering and high-
matrix membranes with oriented selective flakes. J. Membr. Sci. 295, 50–70. resolution transmission electron microscopy. Macromolecules 38 (3), 818–831.
Sheffel, J.A., Tsapatsis, M., 2009. A semi-empirical approach for predicting the Yu, J., et al., 1998a. Al16P20O80H4.4C6H18N2: A new microporous aluminophosphate
performance of mixed matrix membranes containing selective flakes. J. Membr. containing intersecting 12- and 8-membered ring channels. Chem. Mater. 10
Sci. 326, 595–607. (5), 1208–1211.
Shu, S., et al., 2007. A general strategy for adhesion enhancement in polymeric Yu, J., et al., 1998b. Solvothermal synthesis and characterization of new alumino-
composites by formation of nanostructured particle surfaces. J. Phys. Chem. C phosphate layers templated by imidazolium ions. Supramol. Sci. 5 (3–4),
111 (2), 652–657. 297–302.
Sieffert, D., Staudt, C., 2011. Preparation of hybrid materials containing copolyi- Yucelen, G.I., et al., 2011. Formation of single-walled aluminosilicate nanotubes
mides covalently linked with carbon nanotubes. Sep. Purif. Technol. 77 (1), from molecular precursors and curved nanoscale intermediates. J. Am. Chem.
99–103. Soc. 133 (14), 5397–5412.
Skoulidas, A.I., et al., 2002. Rapid transport of gases in carbon nanotubes. Phys. Rev. Yucelen, G.I., et al., 2012. Shaping single-walled metal oxide nanotubes from
Lett. 89 (18), 185901. precursors of controlled curvature. Nano Lett. 12 (2), 827–832.
Snyder, M.A., Tsapatsis, M., 2007. Hierarchical nanomanufacturing: from shaped Zang, J., et al., 2010. Flexibility of ordered surface hydroxyls influences the
zeolite nanoparticles to high-performance separation membranes. Angew. adsorption of molecules in single-walled aluminosilicate nanotubes. J. Phys.
Chem.-Int. Edition 46 (40), 7560–7573. Chem. Lett. 1 (8), 1235–1240.
Sokhan, V.P., et al., 2002. Fluid flow in nanopores: accurate boundary conditions for Zang, J., et al., 2009. Self-diffusion of water and simple alcohols in single-walled
carbon nanotubes. J. Chem. Phys. 117 (18), 8531–8539. aluminosilicate nanotubes. ACS Nano 3 (6), 1548–1556.
Sokhan, V.P., et al., 2004. Transport properties of nitrogen in single walled carbon Zhao, Y.-L., Stoddart, J.F., 2009. Noncovalent functionalization of single-walled
nanotubes. J. Chem. Phys. 120 (8), 3855–3863. carbon nanotubes. Accounts Chem. Res. 42 (8), 1161–1171.
Staley, N., et al., 2007. Lithography-free fabrication of graphene devices. Appl. Phys. Zornoza, B., et al., 2011. Mixed matrix membranes for gas separation with special
Lett. 90, 14. nanoporous fillers. Desalination Water Treat. 27 (1–3), 42–47.