Carbon nanotube-based membranes: Fabrication and application to desalination

Carbon nanotube-based membranes: Fabrication and application to desalination

Journal of Industrial and Engineering Chemistry 18 (2012) 1551–1559 Contents lists available at SciVerse ScienceDirect Journal of Industrial and Eng...

1MB Sizes 5 Downloads 48 Views

Journal of Industrial and Engineering Chemistry 18 (2012) 1551–1559

Contents lists available at SciVerse ScienceDirect

Journal of Industrial and Engineering Chemistry journal homepage: www.elsevier.com/locate/jiec

Review

Carbon nanotube-based membranes: Fabrication and application to desalination Chang Hoon Ahn a, Youngbin Baek b, Changha Lee c, Sang Ouk Kim d, Suhan Kim e, Sangho Lee f, Seung-Hyun Kim g, Sang Seek Bae h, Jaebeom Park h, Jeyong Yoon b,* a

Water and Environment Team, Civil Engineering Center, Samsung C&T Corporation, Seoul 137-956, Republic of Korea World Class University (WCU) Program of Chemical Convergence for Energy & Environment (C2E2), School of Chemical and Biological Engineering, College of Engineering, Institute of Chemical Process, Seoul National University (SNU), Seoul 151-742, Republic of Korea c School of Urban and Environmental Engineering, Ulsan National Institute of Science and Technology (UNIST), Ulsan 698-805, Republic of Korea d Department of Materials Science and Engineering, KAIST, Daejeon 305-701, Republic of Korea e Department of Civil Engineering, Pukyung National University, Busan 609-735, Republic of Korea f School of Civil and Environmental Engineering, Kookmin University, Seoul 136-702, Republic of Korea g Department of Civil Engineering, Kyungnam University, Changwon 631-701, Republic of Korea h Department of Water Supply Technology, Korea Water Resources Corporation (K-water), Daejeon 306-711, Republic of Korea b

A R T I C L E I N F O

Article history: Received 25 January 2012 Accepted 2 April 2012 Available online 10 April 2012 Keywords: Carbon nanotube (CNT) Membrane Desalination

A B S T R A C T

Membranes based on carbon nanotubes (CNTs) have been highlighted as an emerging technology for water purification system applications. With their ultra high water flux and low biofouling potential, CNT membranes are believed to lack various problems encountered when using the conventional membrane separation process that requires a large amount of energy and meticulous maintenance. Although diverse types of CNT membranes have been reported, no commercialized products are available. This article reviews the proper manufacturing methods for CNT membranes and speculates on their performances. Future applications of integrated CNT membrane systems are also outlined. ß 2012 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights reserved.

Contents 1. 2. 3. 4.

5.

6.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basic CNT information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanofluidics of CNT membranes . . . . . . . . . . . . . . . . . . . . . Fabrication of CNT membranes . . . . . . . . . . . . . . . . . . . . . . Types of CNT membranes. . . . . . . . . . . . . . . . . . . . . 4.1. Vertically aligned CNT membranes . . . . . . . . . . . . . 4.2. Mixed (composite) CNT membranes . . . . . . . . . . . . 4.3. Further considerations of CNT membranes . . . . . . . . . . . . Key parameters of CNT membrane performance. . . 5.1. Production of high quality CNTs . . . . . . . . . . . . . . . 5.2. Manufacturing high-performance CNT membranes 5.3. CNT density . . . . . . . . . . . . . . . . . . . . . . . . 5.3.1. Aggregation of nanotubes . . . . . . . . . . . . . 5.3.2. Thickness of the CNT layer . . . . . . . . . . . . 5.3.3. Etching process . . . . . . . . . . . . . . . . . . . . . 5.3.4. Filler material . . . . . . . . . . . . . . . . . . . . . . 5.3.5. Functionalization of CNT tips . . . . . . . . . . . . . . . . . . 5.4. Desalination potential of CNT membranes . . . . . . . . . . . . . Major factors determining desalination potential . . 6.1. Projecting performances of CNT membranes. . . . . . 6.2. Practicable targets for CNT membranes. . . . . . . . . . 6.3.

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

* Corresponding author. Tel.: +82 2 880 8927; fax: +82 2 876 8911. E-mail address: [email protected] (J. Yoon). 1226-086X/$ – see front matter ß 2012 The Korean Society of Industrial and Engineering Chemistry. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jiec.2012.04.005

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . .

1552 1552 1552 1553 1553 1554 1554 1554 1554 1554 1554 1555 1555 1556 1556 1556 1556 1557 1557 1557 1557

C.H. Ahn et al. / Journal of Industrial and Engineering Chemistry 18 (2012) 1551–1559

1552

7.

Integrating CNT membranes into a desalination system . Assembling CNT membranes into modules . . . . . . 7.1. Configuration of a CNT membrane system . . . . . . 7.2. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

1. Introduction Water is under constant pressure as a resource. Due to population growth, economic development, rapid urbanization, large-scale industrialization, and environmental concerns, water stress has emerged as a real threat [1]. Additionally, climate aberrance will significantly affect water stress, change rainfall patterns, and shrink the snow and ice covers that feed rivers. While scarcity drives us to use lower quality and unconventional water sources, membrane separation technology can meet these global climate challenges [2]. Since their inception in the early 1960s, membranes have revolutionized fluid separation processes [3]. Membranes efficiently remove specific particles and molecules from liquids, which has been difficult to achieve using conventional water treatment technologies. Membranes are widely used not only to treat surface water but also to reuse wastewater and to desalinate seawater which can alleviate the problem of water scarcity [4]. However, one of the main problems with desalination membrane technology is the high level of energy consumption [5]. Although the capital and operation costs of membrane processes have fallen dramatically over the past decade [6], the use of membranes is still an energy-intensive process. For example, the energy consumption for seawater desalination using reverse osmosis (RO) membranes has dropped from 8.0 kWh/m3 to 3.4 kWh/m3 [7,8]. Advanced energy recovery devices are expected to be available soon, and the specific energy consumption (SEC) will decrease to <2.5 kWh/m3 [9]. Nevertheless, this consumption is still higher than the theoretically limited value for seawater desalination of 1.06 kWh/m3 (assuming 35,000 mg/L of salt in seawater and a typical recovery of 50%) [4]. One key for further decreasing energy consumption of using membranes is to develop novel membrane materials with high permeability. Nevertheless, the current thin film composite RO membranes suffer from a trade-off between salt rejection rate and permeability. To overcome the limits of current polymeric membranes, new types of membrane with higher permeability and rejection rate have been invented. These membranes use carbon nanotubes (CNTs) as membrane pores [10,11]. These CNT membranes could potentially provide a solution to water shortages, as they seem to outperform existing membranes by providing higher water flux and lower energy consumption. In contrast, the feasibility of CNT membranes has not fully investigated, as they are still in the laboratory stage of development and not yet commercially available. Fabrication of CNT membranes, which have controlled geometry, porosity, and pore shapes, is also challenging [12]. This paper reviews the state-of-art approaches for fabricating CNT membranes and critically evaluates the advantages and disadvantages of these approaches. Two types of CNT membranes are compared, including (1) vertically aligned (VA) CNT membranes, and (2) mixed (composite) CNT membranes. The prospect of using CNT membranes for water purification is also discussed.

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

1558 1558 1558 1559 1559

medical devices, chemical sensors, and environmental technologies [14–16]. A variety of techniques have been explored to produce CNTs, including arc discharge [17] and laser deposition [18]. Typically, the growth of CNTs on a metal catalyst such as iron, nickel, or cobalt is employed during the chemical vapor deposition (CVD) process [19]. The cylindrical shape of a single walled nanotubes (SWNTs) can be imagined virtually by wrapping them in a layer of graphite called graphene [20]. The way graphene winds can be described by a pair of indices (n, m). The indices n and m are integers indicating the number of unit vectors along two directions of graphene (Fig. 1) [21]. The inner diameter of a nanotube can be calculated from the ‘‘rolled up’’ vector as follows [22].  a qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi din ¼ (1) ðn2 þ m2 þ nmÞ  2r c

p

din, inner diameter (I.D.) of nanotubes; a, lattice parameter of graphene (=2.46 A˚); and rc, van der Waal’s radius of a carbon atom (1.7 A˚). 3. Nanofluidics of CNT membranes The inner walls of CNTs are smooth and hydrophobic. Movement of water molecules passing through the interior a nanotube can be explained by the ballistic motion of water chains (1D wire) due to strong hydrogen bonding between water molecules and minimal interaction with the CNT inner wall [23–25] (Fig. 2). The mass movement of water molecules through a CNT does not follow conventional fluid mechanics [26]. Thus, it is necessary to introduce a plausible transport phenomenon called ‘‘nanofluidics’’. In this novel theory, it is assumed that the fluid flowing through a

2. Basic CNT information Since Iijima’s group created the first CNT synthesis protocol [13], which was originally intended to produce fullerenes, various CNT applications have been investigated, including their use in

Fig. 1. Letters (n, m) indicate the number of unit vectors in an infinite graphene sheet and Ch is a ‘rolled up’ vector. T denotes the tube axis, and a1 and a2 are the unit vectors of graphene. If m = 0, the CNTs are called ‘zigzag’. If n = m, the CNTs are called ‘armchair’. In other cases, the CNTs are ‘chiral’.

C.H. Ahn et al. / Journal of Industrial and Engineering Chemistry 18 (2012) 1551–1559

1553

Fig. 2. Movement of water molecules through a SWNT. Reprinted with permission from [25].

nano-channel has a slip length with no friction [27]. Adopting the slip-flow condition, the Hagen–Poiseuille equation can be used as follows [26]: Q slip ¼

pðd=2Þ4 þ 4ðd=2Þ3  LS ðdÞ DP  8m L

(2)

Qslip, water flux depending on slip length; d, diameter of the nanochannel; DP, pressure difference between both ends of the nanochannel; m, viscosity of water; and L, the length of nano-channel. The slip length (LS (d)) can be computed as follows: LS ðdÞ ¼ LS;1 þ

C 3

d

(3)

LS,1, slip length of the graphene surface (assumed to be 30 nm); and C is a fitting parameter. Additionally, the diffusion coefficient of water molecules is estimated as DH2 O ¼ 0:9423  109 m2 =s for a 2.1 nm diameter nanotube [28]. In this manner, the CNT membrane uses the inner and/or outer surface of nanotubes as nano-channels for transporting fluid. With this configuration, applications of the CNT membrane may not be limited to desalination processes. Additionally, it may be feasible to extend the use of CNT membranes to include separation technologies for oil and gases [29–31]. 4. Fabrication of CNT membranes 4.1. Types of CNT membranes CNTs can be classified into two major categories according to the fabrication methods; (1) vertically aligned CNT membranes, and (2) mixed (composite) CNT membranes. For VA-CNT membranes, nanotubes are arranged straight up and perpendicular to the membrane surface. In this configuration, nanotubes are bound to each other by an organic or inorganic filler material. On the other hand, mixed-CNT membranes have a structure similar to that of the thin-film composite RO membranes, where the top layer is mixed with nanotubes and a polymer such as polyamide (PA). Conceptual images of both CNT membranes are shown in Fig. 3.

Fig. 3. Conceptual structures of CNT membranes. (a) Vertically aligned (VA) CNT membranes, and (b) mixed (composite) CNT membranes.

The features of VA-CNT membranes and mixed-CNT membranes are summarized in Table 1. One important advantage of VA-CNT membranes is that water flux should be very rapid due to the short nano-channel length and compactness of the nanotube forest. The VA-CNT membrane was initially attempted due to the nanofluidics, but fabrication made it difficult to produce the large quantities needed for commercialization. Meanwhile, the mixedCNT membrane has its own merits such as a relatively simple manufacturing procedure and similarity to existing membrane processes. Therefore, more attention must be given to mixed-CNT membranes. Table 1 Comparison of vertically aligned (VA) CNT membranes and mixed (composite) CNT membranes. VA CNT membranes

Mixed CNT membranes

 CNTs are aligned vertically

 CNTs are mixed with polymeric materials  Composite layers with PSf membrane and non-woven support  Water flux is moderately fast

 CNTs’ forest is compacted densely  Water flux is supposed to be fast drastically  Functional group can be attached at the tip of CNTs or on the membrane surface conveniently  Fabrication procedures are complicated  May need specially adjusted operating system

 Low (or anti-) fouling membrane

 Fabrication processes are conveniently simple  Operationally feasible to the conventional membrane process

1554

C.H. Ahn et al. / Journal of Industrial and Engineering Chemistry 18 (2012) 1551–1559

4.2. Vertically aligned CNT membranes The first prototype for a VA-CNT membrane was introduced by Hinds’s research group [10]. After being grown on an iron catalyst using the CVD process, MWNTs were embedded in polymeric filler composed of polystyrene (PS). Hinds and coworkers performed a series of supplementary pressure-driven flow experiments with the MWNT/PS membrane and found that water flow rates increased 4- to 5-fold over those of conventional fluid flow, which was estimated from the Hagen–Poiseuille equation [11]. Originally, the Hinds’s research group developed the CNT membrane as a chemically selective gate keeper, which could separate different sized enzymes [32]. Thus, they did not report ion selectivity, which is strongly related to desalination potential during the desalination process. Based on a moderate inner diameter (I.D. = 7 nm), we conjectured that its salt rejection capability might not be very high. In addition, Holt et al. developed a micro-electro mechanical system -compatible fabrication process for another type of VA-CNT membrane [33] which employs nanotubes with an inner diameter <2 nm (average I.D. = 1.6  0.4 nm) to enhance the nanofluidic effect. Inorganic filler (silicon nitride, Si3N4) was employed to ensure that water flowed only through the nano-channels and did not permeate the nanotube-filler matrix. Holt and his collaborators reported water fluxes that were >3-fold greater than those of non-slip hydrodynamic flow as calculated from the Hagen–Poiseuille equation [33]. The major features of the two representative VA-CNT membrane studies for water treatment are compared in Table 2. 4.3. Mixed (composite) CNT membranes An earlier model of a mixed-CNT membrane was mainly designed to upgrade a UF membrane with nanotubes. MWNTs (up to 5% by weight volume) were blended with polysulfone (PSf), and water fluxes were measured under an operating pressure of 1– 4 bar [34]. Intriguingly, the MWNT/PSf membrane revealed two pieces of conflicting data according to the molecular weight of the solute. For an aqueous solution of poly-ethyleneoxide 100,000, the solute rejection efficiency was high (>95%) and the water flux was measured at 14–17 L m2 h1 (LMH). In contrast, the solute rejection efficiency was reduced by 20–60% for aqueous solution of poly-vinylpyrrolidone (PVP) 55,000, whereas the water flux was increased to >40 LMH. Choi et al. presumed that the plugging effect between both sizes of nano-pores and the solute molecule might contribute to differences in the solute rejection efficiencies [34]. Thus, it seemed to be a dilemma to accomplish a higher permeability and rejection rate at the same time with the mixed-CNT membrane. Functionalized MWNTs blended with PSf have been prepared for UF membranes [35]. MWNTs were modified by attaching isocyanate and isophthaloyl chloride functional groups, and protein adsorption on the membrane surface was suppressed. Thus, it was anticipated that a functionalized MWNT/PSf membrane would alleviate membrane biofouling. Additionally, a mixed-CNT membrane for RO membrane has a water flux of 4.05 LMH/bar [36]. Compared to a commercialized brackish water (BW) RO membrane whose water flux ranged from 2.5 to

3.0 LMH/bar, the mixed-CNT membrane demonstrated 1.5-fold higher water flux. However, a number of issues should be resolved, which are addressed in the following section. 5. Further considerations of CNT membranes 5.1. Key parameters of CNT membrane performance The performance of a CNT membrane can be mainly gauged by water flux and salt (or ion) rejection efficiency. As described earlier, the smooth and hydrophobic wall of nanotubes facilitates the rapid and frictionless movement of water molecules in chains. Molecular dynamic (MD) simulations show that a narrower inner diameter of nanotubes results in faster transport of water molecules [26]. 5.2. Production of high quality CNTs Nano-pores should be evenly distributed on the surface layer to enhance the performance of CNT membranes. Above all, it is crucial to produce high quality CNTs. Nanotubes should be prepared uniformly to obtain high desalination capacity. Thus, homogeneity of nano-pores is crucial, and the inner diameter of nanotubes should be distributed within a very narrow range. The inner diameter of MWNTs can range from a few nanometers to tens of nanometers, which may result in poor desalination capacity [37,38]. Additionally, a ‘‘bamboo’’ structure may be formed inside of MWNTs, which would block the nano-channels, and the path of the water molecules [39,40]. In contrast, SWNTs (or DWNTs), whose inner diameters are <1–2 nm usually do not have the bamboo structure, but their manufacturing procedure is relatively complicated [41,42]. During the manufacture of nanotubes, the inner diameter and number of walls are determined by the size of the inorganic catalyst (e.g., Fe) [43]. Nano-patterning processes that can uniformly detect ultrafine catalysts will be essential to produce homogenous nanotubes. Therefore, selection of an appropriate catalyst for the platform should be extensively assessed beforehand. Block copolymer lithography (BCL) sustains self-assembly of molecules, which can minimize the size of a catalyst [44]. When using the BCL process, it is expected that nanotubes with an inner diameter <2 nm can be produced in large quantities (Fig. 4). Growing nanotubes vertically is another hurdle to overcome. Because nanotubes are formed by blowing substrate gases (e.g., ethylene) onto the platform, it is extremely difficult to control their tortuosity. Plasma enhanced CVD (PECVD) is a currently available technology used for straight and perpendicular growth of nanotubes [45]. As dipole moments are charged by an electric field, torque forces act on the metal catalyst of the platform and straighten nanotubes in the middle of their growth (Fig. 5). 5.3. Manufacturing high-performance CNT membranes After defect-free nanotubes are obtained, assembling them into a CNT membrane is another laborious task. Some of the issues involved in this task are presented below.

Table 2 Major features of vertically aligned (VA) CNT membranes. Research group

Hinds Group [10]

Holt Group [33]

Pros

 Polymer (i.e., polystyrene) was used as a filler material  Relatively simple fabrication procedure  Good reproducibility of empirical data

    

Cons

 Poor ion selectivity with MWNTs  Distribution of irregular pore sizes

 Complicated fabrication process  Additional procedures needed (e.g., back etching)

CNT membrane with ultrafine pores (<2 nm) Inorganic filler matrix (Si3N4) MEMS fabrication process Good permeability without free volume and leakage Enhanced ion selectivity (high desalination potential)

C.H. Ahn et al. / Journal of Industrial and Engineering Chemistry 18 (2012) 1551–1559

1555

Fig. 4. Proposed manufacturing process for vertically aligned (VA) CNT membranes using a block copolymer lithography method. Reprinted with permission from [44].

Fig. 5. Vertical growth of CNTs by using plasma enhanced chemical vapor deposition (PECVD) process. Reprinted with permission from [45].

5.3.1. CNT density As the number of nanopores (or nano-channels) increases, water flux is augmented proportionally. That is, CNT density is a key parameter determining permeability of a CNT membrane, whereas pore size is one of the main factors enhancing rejection efficiency. It is essential to augment CNT density (or areal density) to render the highest possible surface porosity. Considering previously reported values, the CNT density should be above 1011 CNT cm2 to attain water flux of 30–100 LMH [22,33]. As nanotechnology improves, the CNT density will continue to increase along with surface porosity. Falconer’s research group found a novel way of consolidating a nanotube forest and reported

a density of 2.9  1012 CNT cm2 [46]. The main components of various CNT membranes are compared in Table 3 [10,33,47,48]. Improvements in CNT membranes are mainly achieved by increasing CNT density from 2  109 to 7  1010 and by employing SWNT or DWNT instead of MWNT. 5.3.2. Aggregation of nanotubes Due to van der Waal’s forces, nanotubes tend to agglomerate together, and due to the strong agglomeration propensity, each cane of nanotubes should be sufficiently separated beforehand to form CNTs and a filler matrix, which may not be conveniently achievable.

1556

C.H. Ahn et al. / Journal of Industrial and Engineering Chemistry 18 (2012) 1551–1559

Table 3 Main components of various CNT membranes. CNT membrane

Mi group [48]

Hinds group [10]

Holt group [33]

Kim group [47]

Main structure Filler material CNTsb Average outer diameter (O.D.) (nm) Average inner diameter (I.D.) (nm) Thickness of CNT layer (mm) CNT density (cm2) Porosity (e)

Porous aluminum support Polystyrene MWNT 20 6.3 10 1.87  109 6.2  104

Free-standing (separated from quartz support) Polystyrene MWNT N.A.a 7.5  2.5 5–10 6  1010 2.7  102

Silicon wafer Silicon nitride DWNT 2 1.6  0.4 5 2.5  1011 5.0  103

PTFE filter Polysulfone SWNT N.A.a 1.2 6 (7.0  1.75)  1010 N.A.a

a b

Not available data. Single-walled nanotubes (SWNTs), double-walled nanotubes (DWNTs), and multi-walled nanotubes (MWNTs).

5.3.3. Thickness of the CNT layer Water flux is inversely proportional to the thickness of the membrane [49]. For example, reducing the surface layer thickness by 50% increases the water flux by 2-fold. Up to now, the lengths of CNTs have ranged from 5 to 10 mm (Table 3). The thickness of the CNT layer needs to be <1 mm. 5.3.4. Etching process The tops of as-grown nanotubes are covered with fullerene caps and therefore cannot function as nano-channels. After growing nanotubes on a platform, an etching process, including plasma oxidation and argon milling, is required to open their fullerenecapped tips. Analogous to the VA-CNT membrane, the mixed-CNT membrane requires an effective etching method that also opens the nanopores at the active surface layer. 5.3.5. Filler material The filler material should be prudently selected to assemble high quality CNT membranes. Various properties should be tested to determine which filler material will have the best characteristics for nanotubes, such as optimal mechanical strength, flexibility, and stiffness. Either an organic filler or inorganic filler can be used to form a CNT matrix. 1. Inorganic fillers (e.g., silicon nitride) have robust resistance to high pressure and show little leakage of free volume.

Fig. 6. Functionalized end tips of vertically aligned (VA) CNT membrane. Reprinted with permission from [51].

2. Organic fillers (e.g., PA, PSf, etc.) have great flexibility for manufacturing CNT membranes in large quantities. However, bundles of nanotubes may be tilted by osmotic pressure during infiltration of a filler material, which will disturb the vertical alignment. Miscibility between nanotubes and filler material is crucial for mixed-CNT membranes. As nanotubes are hydrophobic, identifying the best fitting material will be a laborious task. The filler matrix has excellent miscibility and forms an active surface layer consisting of a mixed composite membrane. Therefore, it is necessary to modify the exterior of nanotubes in order to enhance miscibility. We expect that water flux can be enhanced by altering the ends of nanotubes from hydrophobic to hydrophilic. A variety of methods can be employed to alter nanotube hydrophobicity, including acid or base treatment, alkyl substitution, and functional group attached at the tip of the CNT [50,51]. Additionally, CNT should be continuously produced for further application [52]. 5.4. Functionalization of CNT tips A membrane’s hydrophobic surface may either reduce or increase water flux. As nanotubes are hydrophobic, the movement of water molecules is hindered in areas adjacent to the surface of a CNT membrane, leading to decreased water flux. However, once water molecules enter the membrane, the hydrophobic interior

C.H. Ahn et al. / Journal of Industrial and Engineering Chemistry 18 (2012) 1551–1559

wall is favorable, which helps the water molecules to be transported at ultra-high speed over 100 LMH [33]. Therefore, it may be beneficial to functionalize the end tips of nanotubes to change hydrophobicity of the CNT membrane and eventually enhance ion selectivity. The first developed VA-CNT membrane was functionalized to work as a bio gate keeper [53] (Fig. 6). However, a further detailed study will be necessary to develop a coating with densely charged polymers (e.g., polyelectrolytes, etc.) over the entire membrane surface as well as CNT end tips to increase salt rejection efficiency.

1557

ion valence of the solution [55]. They speculated that electrostatic interactions are more important than steric effects to govern ion rejection, and this trend agrees with the Donnan membrane equilibrium theory [59]. By elevating the surface charge of the CNT membrane, a sodium ion that is smaller than the pore diameter can be excluded. Therefore, an ultrafine pore size may not be a prerequisite for improving desalination potential of CNT membranes. It has also been reported that the ion exclusion efficiency of a mixed-CNT membrane is >99% [36], which suggests promising future applications.

6. Desalination potential of CNT membranes 6.2. Projecting performances of CNT membranes 6.1. Major factors determining desalination potential Organic substances and dissolved ions are mainly sieved by surface pores during the membrane separation process. Typically, the pore size of an RO membrane ranges from 6 to 8 A˚ (0.6–0.8 nm), whereas that of NF is >1 nm [54]. Although field applications for CNT membranes have not been demonstrated, their potential for use in desalination needs to be roughly assessed. Ion selectivity of a CNT membrane can be determined by the following major factors [55]: (1) the steric effects between nanopores and the hydrated diameter of ions and (2) the Donnan equilibrium of the membrane surface. Using molecular dynamic (MD) simulation, the desalination potential of a virtual VA-CNT membrane was estimated according to the size of the nanotubes [22,56]. As the inner diameter of nanotubes was increased from 0.32 nm to 0.49 nm and then to 0.59 nm and 0.75 nm, salt rejection efficiencies of the CNT membrane declined to 100%, 100%, 95%, and 58%, respectively (Table 4). Desalination capacity declined as pore size increased. Based on the MD simulation results, the inner diameter of nanotubes should be  0.6 nm, if the desalination capacity of a CNT membrane is to correspond with to that of an RO membrane. Currently, nanotubes can be produced with an inner diameter of 1–2 nm, but these diameters will be reduced by using available technologies. However, it is believed that there is a critical pore size of 7 A˚ (0.7 nm), beyond which the movement of water molecules and aqueous ions will be slowed by the confinement effect [22,57]. Although a narrower CNT pore size is essential for achieving enhanced rejection efficiency desalination, it may not be practical with current technology. Accordingly, Donnan effect factors should be considered to increase desalination efficiency. Desalination efficiency using the previously reported VA-CNT membrane (average pore size of 1.6 nm), was measured under pressure driven conditions [55]. In a 1 mM KCl solution, the ion exclusion efficiency of the VA-CNT membrane ranges between 40 and 60%, which is similar to the efficiencies of commercialized NF membranes [58]. However, the efficiency diminishes as aqueous ion concentration increases. Fornasiero et al. reported that the ion exclusion efficiency of the CNT membrane depends on the pH and

Biofouling is one of the nuisances during the operation of a membrane process and leads to a decline in membrane permeability and salt rejection efficiency [60–63]. Biofouling aggravates the concentration polarization within a biofilm matrix, eventually resulting in augmentation of process operation costs. Therefore, continuous care should be taken to maintain membrane performance. Along with nano-sized particles, numerous studies have reported that CNTs trigger inactivation of bacterial cells by attacking cell walls [64,65]. Compared to MWNTs, SWNTs (or DWNTs) are more often fatal to bacteria, and suppress biofilm formation on deposited surfaces [66,67]. Two trials have been conducted to reduce biofouling using nanotubes [68]. Therefore, CNT membranes have a low biofouling surface that may reduce required maintenance, compared to that of commercialized NF and RO membranes. The CNT membrane process should require low energy consumption due to nanofluidics. Unlike the NF or RO processes, the CNT membrane process can run without a high-pressure pump, which consumes a tremendous amount of power. Additionally, the specific surface area of the membrane element attaining the targeted permeate volume should diminish. Consequently, the CNT membrane process will require less energy consumption than that of other membrane processes. 6.3. Practicable targets for CNT membranes The technically achievable performance of various types of CNT membranes are summarized in Table 5. By accepting outputs from MD simulation and adopting previously reported numbers, CNT membrane water fluxes are approximately 70–270 LMH [22]. Thus, it is plausible that the water flux of a VA-CNT membrane could reach 10–15 LMH/bar, which is 5-fold higher than that of BWRO. Similarly, water flux of the mixed-CNT membrane (DWNTs/PA) is 4.05 LMH/bar, which is 1.5-fold higher than that of BWRO [36]. Although extremely high water flux (150–220 LMH/ bar) of a MWNT/PSf membrane was reported, the water transport mechanism has not been identified [34].

Table 4 Salt rejection efficiencies and flow rates of the vertically aligned (VA) CNT membranes based on molecular dynamic simulation.a Rolled up vector

(5,5) (6,6) (7,7) (8,8)

Inner diameter (nm)

0.32 0.49 0.59 0.75

Salt rejection (%)

100 100 95 58

Vertically aligned CNT membraneb Flow ratec (LMH)

Enhancementd (e)

66.7 112.5 175.0 270.8

2.42 4.21 6.39 9.76

Reprinted with permission from [22]. a Assuming an operating pressure of 5.5 MPa and allow for an osmotic pressure of 2.4 MPa. b CNT density of the membrane was assumed to be 2.5  1011 CNT cm2 [33]. c LMH = L m2 h1. d Enhancement ratios are estimated relatively to the published values for a FILMTECH SW30H4-380 commercial reverse osmosis membrane.

C.H. Ahn et al. / Journal of Industrial and Engineering Chemistry 18 (2012) 1551–1559

1558

Table 5 Technically achievable performances according to types of CNT membranes. CNT membrane a

Water flux (LMH/bar) Salt rejection Antifoulingc Membrane element type

Vertically aligned

Mixed (composite)

10–15 BWRO levelb N.A. Plate and frame (1 m3)

4–6 BWRO levelb 30% Spiral wound (8 in.)

a Assuming that nanotubes of inner diameter 1 nm are employed and CNT density is 2.5  1011 CNT cm2 for both CNT membranes [33]. b Salt rejection efficiency 93% at an applied pressure of 0.5–3.0 MPa. c Reduction of biofouling brings to high energy efficiency.

It seems that VA-CNT membranes will have salt rejection efficiencies similar to those of commercialized NF membranes. Ultrafine nanotubes (inner diameter 1 nm) should be employed and their surface charge should be elevated to enhance desalination capacity to the level of an RO membrane. By upgrading nanotubes with various functional groups and surface modifications, the salt rejection efficiency of a VA-CNT membrane should reach that of a BWRO membrane. The active surface layer of a mixed CNT membrane is composed of a mixture of nanotubes and polymers (e.g., PA). By employing techniques analogous to those described above, it may not be a big challenge to increase the salt rejection efficiency of mixed CNT membranes to correspond to that of a BWRO membrane. 7. Integrating CNT membranes into a desalination system 7.1. Assembling CNT membranes into modules A prototype manufacturing process for a VA-CNT membrane can be described as follows. Fully grown nanotubes are harvested after the PECVD process, and filler material infiltrates into the CNT forest. Then, small pieces of CNT membrane would be patched into plank or plate shapes. Judging from the manufacturing procedures, conventional membrane modules may not be applicable for use with VA-CNT membranes. In this manner, we postulate that a plate-and-frame type will be suitable as a VA-CNT membrane module, which can be adapted from the disc-tube module of the Pall Corp. (Fig. 7) [69]. Plate shaped CNT membranes would be

laminated in a cylindrical tank and permeates would be collected from a drainage pipeline in the core. In contrast, we expect that a module configuration of a mixedCNT membrane may not deviate much from the conventional types of PA-RO membranes. Currently, the spiral wound type is employed worldwide in the RO process. A typical manufacturing process for a PA-RO membrane can be described as follows [70]. First, a membrane film is formed by applying a PSf layer on a polyester support during the ‘‘casting’’ process, which can then be used as an UF membrane. Second, PA is coated by interfacial polymerization as a skin layer on top of the PSf layer, which is called the ‘‘coating’’ process. Subsequently, the membrane sheet is scrolled in a leaf form, producing a spiral-wound type module, which is called the ‘‘rolling’’ process. Assuming an analogous conveyor line can be used for the mixed CNT membrane, only the coating process would be retrofitted. That is, nanotubes are mixed with a polymer (e.g., PA) solution and their mixture forms an active layer on the membrane surface. Alternately, additional procedures may be required for mixing nanotubes and polymer in advance, and the mixture would be coated on the PSf support layer. In either case, mixed (composite) CNT membranes would be scrolled in the rolling process and a spiral-would type module would be manufactured. 7.2. Configuration of a CNT membrane system In advance of on-site applications, the manufactured membrane module should be assessed for short-term and long-term use. Short-term examinations include measurements of water flux and salt rejection efficiency, whereas long-period evaluations involve specific energy consumption rate and module stability. Other operating parameters should be considered, such as disinfectant and chemical cleaning. Ultra-high water flux should be considered to integrate the CNT membrane into water treatment processes. If the water flux is extremely high, dissolved salts will be left over and crystallize, leading to a failure of the system. Additionally, various types of membrane modules should be considered in the system configuration. By maneuvering the newly developed design program, performances of a CNT membrane could be predicted including the

Fig. 7. Cross-sectional view of a disc-tube membrane [67].

C.H. Ahn et al. / Journal of Industrial and Engineering Chemistry 18 (2012) 1551–1559

qualities of permeates and concentrates. Eventually, a configuration for the whole system will be designed (e.g., arrangement of vessels). Along with the operating configuration, the safety of the CNT membrane is also an important issue to be resolved. Concerns will be raised about the environmental and health impacts of CNTs that may be released from CNT membranes during operation [71,72]. Considering the multiple layers and supplementary appurtenances of CNT membranes, it appears that there is little chance that CNTs are able to pass through a wholly integrated membrane element. Even with numerous efforts for regulating nano-materials, no safety guidelines for use of CNT membranes in the water purification process have been adopted. Thus, an interdisciplinary approach should be used to establish a policy for CNT membrane technology. In the near future, CNT membrane related enterprises are expected to flourish as safety rules are properly enforced.

[26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40]

Acknowledgements This research was supported by the WCU (World Class University) program through a Korea Science and Engineering Foundation grant funded by the Ministry of Education, Science, and Technology (400-2008-0230) and the National Research Foundation of Korea Grant funded by the Korean Government (NRF-2010C1AAA01-0029061) and the K-water Research & Business Project (K_RBP-1). We thank John Wiley and Sons, the American Chemical Society and IOP Publishing for the permission to publish figures and tables.

[41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51]

References [52] [1] C.J. Vo¨ro¨smarty, P. Green, J. Salisbury, R.B. Lammers, Science 289 (2000) 284. [2] M.A. Shannon, P.W. Bohn, M. Elimelech, J.G. Georgiadis, B.J. Marinas, A.M. Mayes, Nature 452 (2008) 301. [3] J. Glater, Desalination 117 (1998) 297. [4] M. Elimelech, W.A. Phillip, Science 333 (2011) 712. [5] H. Ludwig, Desalination and Water Treatment 13 (2010) 13. [6] GWI, Desalination Markets 2005–2015, GWI, Oxford, 2004. [7] C. Fritzmann, J. Lo¨wenberg, T. Wintgens, T. Melin, Desalination 216 (2007) 1. [8] R. Semiat, Environmental Science & Technology 42 (2008) 8193. [9] O.M. Al-Hawaj, Desalination and Water Treatment 8 (2009) 131. [10] B.J. Hinds, N. Chopra, T. Rantell, R. Andrews, V. Gavalas, L.G. Bachas, Science 303 (2004) 62. [11] M. Majumder, N. Chopra, R. Andrews, B.J. Hinds, Nature 438 (2005) 44. [12] M. Majumder, P. Ajayan, Comprehensive Membrane Science and Engineering 1 (2010) 291. [13] S. Iijima, Nature 354 (1991) 56. [14] A.M. Popov, Y.E. Lozovik, S. Fiorito, L.H. Yahia, International Journal of Nanomedicine 2 (2007) 361. [15] J. Wang, Electroanalysis 17 (2005) 7. [16] M.S. Mauter, M. Elimelech, Environmental Science & Technology 42 (2008) 5843. [17] T. Ebbesen, P. Ajayan, Nature 358 (1992) 220. [18] T. Guo, P. Nikolaev, A.G. Rinzler, D. Tomanek, D.T. Colbert, R.E. Smalley, The Journal of Physical Chemistry 99 (1995) 10694. [19] M. Jose Yacaman, M. Miki Yoshida, L. Rendon, J. Santiesteban, Applied Physics Letters 62 (1993) 202. [20] A. Thess, R. Lee, P. Nikolaev, H. Dai, P. Petit, J. Robert, C. Xu, Y.H. Lee, S.G. Kim, A.G. Rinzler, Science 273 (1996) 483. [21] en.wikipedia.org/wiki/File:CNTnames.png. [22] B. Corry, The Journal of Physical Chemistry B 112 (2008) 1427. [23] G. Hummer, J.C. Rasaiah, J.P. Noworyta, Nature 414 (2001). [24] J. Ko¨finger, G. Hummer, C. Dellago, Proceedings of the National Academy of Sciences 105 (2008) 13218. [25] T.A. Hilder, D. Gordon, S.H. Chung, Small 5 (2009) 2183.

[53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63]

[64] [65] [66] [67] [68] [69] [70] [71] [72]

1559

J.A. Thomas, A.J.H. McGaughey, Nano Letters 8 (2008) 2788. J.K. Holt, Advanced Materials 21 (2009) 3542. A. Striolo, Nano Letters 6 (2006) 633. A. Ismail, P.S. Goh, S. Sanip, M. Aziz, Separation and Purification Technology 70 (2009) 12. C. Lee, S. Baik, Carbon 48 (2010) 2192. K. Sears, L. Dume´e, J. Schu¨tz, M. She, C. Huynh, S. Hawkins, M. Duke, S. Gray, Materials 3 (2010) 127. M. Majumder, N. Chopra, B.J. Hinds, ACS Nano 5 (2011) 3867. J.K. Holt, H.G. Park, Y. Wang, M. Stadermann, A.B. Artyukhin, C.P. Grigoropoulos, A. Noy, O. Bakajin, Science 312 (2006) 1034. J.H. Choi, J. Jegal, W.N. Kim, Journal of Membrane Science 284 (2006) 406. S. Qiu, L. Wu, X. Pan, L. Zhang, H. Chen, C. Gao, Journal of Membrane Science 342 (2009) 165. T.V. Ratto, J.K. Holt, A.W. Szmodis, Membranes with embedded nanotubes for selective permeability, Google Patents, 2011. C.H. Kiang, M. Endo, P. Ajayan, G. Dresselhaus, M. Dresselhaus, Physical Review Letters 81 (1998) 1869. P.X. Hou, S.T. Xu, Z. Ying, Q.H. Yang, C. Liu, H.M. Cheng, Carbon 41 (2003) 2471. E.T. Thostenson, Z. Ren, T.W. Chou, Composites Science and Technology 61 (2001) 1899. M. Glerup, M. Castignolles, M. Holzinger, G. Hug, A. Loiseau, P. Bernier, Chemical Communications (2003) 2542. D.H. Lee, W.J. Lee, S.O. Kim, Nano Letters 9 (2009) 1427. D.H. Lee, W.J. Lee, S.O. Kim, Chemistry of Materials 21 (2009) 1368. E. Kukovitsky, S. L’vov, N. Sainov, V. Shustov, L. Chernozatonskii, Chemical Physics Letters 355 (2002) 497. D.H. Lee, D.O. Shin, W.J. Lee, S.O. Kim, Advanced Materials 20 (2008) 2480. M. Meyyappan, L. Delzeit, A. Cassell, D. Hash, Plasma Sources Science and Technology 12 (2003) 205. M. Yu, H.H. Funke, J.L. Falconer, R.D. Noble, Nano Letters 9 (2009) 225. S. Kim, J.R. Jinschek, H. Chen, D.S. Sholl, E. Marand, Nano Letters 7 (2007) 2806. W. Mi, Y. Lin, Y. Li, Journal of Membrane Science 304 (2007) 1. A. Zhu, P.D. Christofides, Y. Cohen, Industrial & Engineering Chemistry Research 48 (2009) 6010. M. Majumder, N. Chopra, B.J. Hinds, Journal of the American Chemical Society 127 (2005) 9062. J.M. Lee, S.J. Kim, J.W. Kim, P.H. Kang, Y.C. Nho, Y.S. Lee, Journal of Industrial and Engineering Chemistry 15 (2009) 66. L.S. Ying, M.A.M. Salleh, S.B.A. Rashid, Journal of Industrial and Engineering Chemistry 17 (2011) 367. N. Chopra, M. Majumder, B.J. Hinds, Advanced Functional Materials 15 (2005) 858. S.H. Kim, S.Y. Kwak, T. Suzuki, Environmental Science & Technology 39 (2005) 1764. F. Fornasiero, H.G. Park, J.K. Holt, M. Stadermann, C.P. Grigoropoulos, A. Noy, O. Bakajin, Proceedings of the National Academy of Sciences 105 (2008) 17250. B. Corry, Energy and Environmental Science 4 (2011) 751. M.E. Selvan, D. Keffer, S. Cui, S. Paddison, Molecular Simulation 36 (2010) 568. N. Hilal, H. Al-Zoubi, A. Mohammad, N. Darwish, Desalination 184 (2005) 315. F. Donnan, Journal of Membrane Science 100 (1995) 45. H. Ridgway, H. Flemming, Membrane Biofouling, McGraw-Hill, Washington, DC, 1996. J. Patching, G. Fleming, Biofilms in Medicine, Industry and Environmental Biotechnology, IWA Publishing, UK, 2003, p. 568. M. Herzberg, M. Elimelech, Journal of Membrane Science 295 (2007) 11. J. Vrouwenvelder, S. Manolarakis, J. Van der Hoek, J. Van Paassen, W. Van der Meer, J. Van Agtmaal, H. Prummel, J. Kruithof, M. Van Loosdrecht, Water Research 42 (2008) 4856. S. Kang, M. Pinault, L.D. Pfefferle, M. Elimelech, Langmuir 23 (2007) 8670. Q. Li, S. Mahendra, D.Y. Lyon, L. Brunet, M.V. Liga, D. Li, P.J.J. Alvarez, Water Research 42 (2008) 4591. G. Jia, H. Wang, L. Yan, X. Wang, R. Pei, T. Yan, Y. Zhao, X. Guo, Environmental Science & Technology 39 (2005) 1378. D.F. Rodrigues, M. Elimelech, Environmental Science & Technology 44 (2010) 4583. P.B. Messersmith, M. Textor, Nature Nanotechnology 2 (2007) 138. http://www.pall.com. R.F. Fibiger, J. Koo, D.J. Forgach, R.J. Petersen, D.L. Schmidt, R.A. Wessling, T.F. Stocker, Novel polyamide reverse osmosis membranes, Google Patents, 1988. C. Lam, J.T. James, R. McCluskey, S. Arepalli, R.L. Hunter, CRC Critical Reviews in Toxicology 36 (2006) 189. K. Kostarelos, Nature Biotechnology 26 (2008) 774.