Article pubs.acs.org/est
Constructing All Carbon Nanotube Hollow Fiber Membranes with Improved Performance in Separation and Antifouling for Water Treatment Gaoliang Wei, Hongtao Yu, Xie Quan,* Shuo Chen, Huimin Zhao, and Xinfei Fan Key Laboratory of Industrial Ecology and Environmental Engineering (Ministry of Education, China), School of Environmental Science and Technology, Dalian University of Technology, Dalian 116024, China S Supporting Information *
ABSTRACT: Manipulating carbon nanotubes (CNTs) through engineering into advanced membranes with superior performance for disinfection and decontamination of water shows great promise but is challenging. In this paper, a facile assembly of CNTs into novel hollow fiber membranes with tunable inner/outer diameters and structures is developed for the first time. These free-standing membranes composed entirely of CNTs feature a porosity of 86 ± 5% and a permeation flux of about 460 ± 50 L m−2 h−1 at a pressure differential of 0.04 MPa across the membrane. The randomly oriented interwoven structure of CNTs endows the membranes considerable resistance to pore blockage. Moreover, the adsorption capability of the CNT hollow fiber membranes, which is crucial in the efficient removal of small and trace contaminant molecules, is about 2 orders of magnitude higher than that of commercial polyvinylidene fluoride hollow fiber membranes. The unique advantage of the CNT hollow fiber membranes over other commercial membranes is that they can be in situ electrochemically regenerated after adsorption saturation.
■
INTRODUCTION Increasingly serious environmental pollution and the energy crisis are two of the most pervasive problems afflicting people throughout the world. Clean water scarcity and growing demand for potable water have forced us to explore new advanced water treatment technologies that can provide a safe water supply in a more energy efficient, environmentally sustainable way.1−3 Due to their easy operational control, low energy costs and potential material recovery, membrane processes are among the most effective strategies to achieve high removal of contaminants from wastewater.2,4 Nowadays hollow fiber membranes (HFMs) have attracted considerable attention in wastewater purification, owing to their many advantages, including large surface area per volume ratio,5−7 their self-supporting structure,5,6 and low energy consumption.7 The present HFM market has been dominated for decades by polymeric membranes polymeric membranes for their good flexibility, low cost, and easy preparation. Nevertheless, they usually suffer some important technical limitations regarding, for example, chlorine resistance,8,9 inherent fouling,10−13 and the compromise between permeability (water flux) and selectivity (efficient rejection) existing in all membrane separation processes,14,15 leading to low plant reliability and limiting their wide industrial application in wastewater treatment. To control and mitigate membrane fouling, many strategies have been implemented, such as pretreating feedwater, © 2014 American Chemical Society
optimizing operating conditions, and development of antifouling membranes.16 The recently emerging carbonaceous nanofiber membranes in sheet configurations exhibit greater fouling resistance than commercial polymeric membranes with vertical cylindrical water channels,17 mainly due to their interconnected open pore structure, which is quite different from that of polymeric membranes. The compromise between flux and efficient rejection, another important issue faced by polymeric HFMs, is principally controlled by characteristics of the separation layer, such as pore size and surface porosity.18 However, the phase-inversion polymeric membranes usually possess a low porosity, thereby leading to a low water flux.19 In contrast, many flat-sheet nanofiber membranes, reported by previous works, demonstrated high flux and excellent rejection properties that were attributed to their high surface porosity and pore density as a result of the random interwoven structure of nanofibers.17,18,20 Despite the progress made so far, the compromise between flux and efficient rejection is not well addressed because their surface porosity and pore density could not be infinitely increased. Membrane adsorption processes are probably among the most attractive techniques to circumvent the compromise Received: Revised: Accepted: Published: 8062
September 11, 2013 May 20, 2014 June 18, 2014 June 18, 2014 dx.doi.org/10.1021/es500506w | Environ. Sci. Technol. 2014, 48, 8062−8068
Environmental Science & Technology
Article
HFM was prepared (Figure S1c, Supporting Information). For helical and snake-like HFMs, corresponding templates were shaped by a cylindrical bar. Bichannel and trichannel HFMs were fabricated by utilizing two or three Cu wires and controlling the position and distance of one another. The methods for porosity and pore size measurements and other membrane characterization methods are shown in the Supporting Information. Membrane Performance Tests. All performance tests were performed using a dead-end membrane filtration system, and the pressure was provided by high-pressure nitrogen (schematically shown in Figure S2 (Supporting Information)). Unless otherwise specified, the outer/inner diameter of membranes tested is about 610/450 μm, and the membrane pressure difference is 0.04 MPa. The antifouling tests of CNTHFMs were performed using polystyrene microspheres with diameters of 114 and 362 nm (∼30 mg L−1) to simulate suspended particles with different diameters in wastewater. Rhodamine B (RhB, 10 mg L−1) was used as the removal target to certify the high adsorption capacity of CNT-HFMs at a pressure difference of 0.04 MPa. The concentrations of RhB in the original feed and filtrate were determined by a UV−vis spectrophotometer. To verify the regenerative capability of CNT-HFMs under electrochemical assistance after adsorption saturation, a low voltage of 2 V was applied on the membrane module (anode, CNT-HFMs; cathode, stainless steel network) and then the RhB concentrations in filtrate were determined. To accurately ascertain the regenerative mechanism, the test was performed at a low membrane flux of 4.2 mL cm−2 h−1 (mL of water per cm2 of membrane area per hour). No salts were added into the RhB solution in the whole test process. The filtrate was analyzed by a high-performance liquid chromatography (HPLC) instrument (Waters, 2698) equipped with UV− vis diode array detector using a C18 inversed-phase column. The mobile phase was methanol/water (70:30, v/v) at a flow rate of 1.0 mL/min. The N-deethylated intermediates of RhB were further detected by mass spectrometery (MS) with an Agilent 1100 HPLC-tandem and Agilent 6410 Triple Quadruple mass spectrometer.
because of two essential reasons. First, membrane adsorption can efficiently trap small molecules (18 MΩ cm−1, Laikie Instrument Co., Ltd., Shanghai, China) was used in all experiments. Surface Functionalization of CNTs. The purified multiwalled carbon nanotubes (the detailed purification method is shown in the Supporting Information) were added to a HNO3/ H2SO4 (1:3, v/v) concentrated solution and heated to 60 °C for 4 h. The surface-functionalized carbon nanotubes were recovered by filtration, followed by being washed with water until pH of the filtrate was nearly neutral. The surfacefunctionalized CNTs were dried at 60 °C for 6 h and stored in a desiccator. Preparation of CNT Membranes. The membranes mentioned in this work are fabricated based on an electrophoretic deposition (EPD) process utilizing Cu wires with different diameters and shapes as templates (schematically shown in Figure S1a (Supporting Information)). In a typical experiment, 100 mg of the surface-functionalized CNTs was dispersed in 100 mL of anhydrous 2-propanol, assisted by sonication. Mg(NO3)2·7H2O was then dissolved to the homogeneous CNT dispersion under stirring conditions to produce positively charged CNTs. The applied voltage was maintained at 160 V. After a set period of time, the Cu wire was pulled out from the EPD reactor (Figure S1b, Supporting Information), after which a densely packed CNT layer surrounding the Cu wire formed upon drying. The process was repeated accordingly to obtain the desired thickness of the CNT layer. Subsequently, the Cu wire wrapped with CNTs was calcined at 600 °C for 1 h in the argon flow. After Cu wire was etched in an aqueous 2.5 M FeCl3/0.5 M HCl solution and thoroughly washing with distilled water, a free-standing CNT-
■
RESULTS AND DISCUSSION Membrane Characterization. CNT-HFMs were fabricated via a template-based EPD method. Various characterizations are performed to illustrate the prepared membranes. Morphological Characterization. As observed from the scanning electron microscopy (SEM) images in Figure 1a, the prepared CNT-HFMs feature a crack-free surface. Closer examination reveals a random pore structure, characterized by an interwoven network of CNTs (Figure 1b). Compared with polymeric HFMs synthesized from dry/wet spinning processes,27,28 the current prepared CNT-HFMs display more uniform inner/outer diameters (Figure 1c). The use of Cu templates generates CNT-HFMs with a patterned and smoother inner surface relative to the texture of the outer surface (Figure 1d and inset) and tailors the inner diameter of the CNT-HFMs, as exemplified in Figures 2 and S3 (Supporting Information) showing the as-fabricated CNTHFMs with inner diameters of 150 and 450 μm, respectively. Structural Regulation. Theoretically, the thickness of CNTHFMs can be precisely controlled by changing the duration of the EPD process. However, it was noted that a rapid decrease in the current from 24 to 8 mA in the first 20 s, followed by a gradual decrease to 0 mA over 120 s (Figure S4a, Supporting 8063
dx.doi.org/10.1021/es500506w | Environ. Sci. Technol. 2014, 48, 8062−8068
Environmental Science & Technology
Article
assembly strategy also affords CNT-HFMs with a highly homogeneous structure in the cross section (Figure S5, Supporting Information). Benefiting from the easy manipulation of Cu wire templates, these CNT-HFMs can be conveniently shaped into specific geometries with designed morphologies. As shown in Figure 3a,b, bichannel and trichannel CNT-HFMs were easily
Figure 1. SEM images of CNT-HFMs. (a) Two CNT-HFMs with the same inner diameter of 150 μm but different outer diameters. (b) Enlarged view of the marked area in panel a. (c) Cross-section perpendicular to axis of CNT-HFM (246 μm in outer diameter and 150 μm in inner diameter). (d) Cross-section parallel to axis of CNTHFM. The inset image shows the inner surface of CNT-HFM.
Information), thereby indicating the completion of the layer development. These findings suggest that thick CNT-HFMs could not be easily obtained in a one-step EPD process. The decline in the current is related to the increased resistance of the EPD system because of the weak contact among the asdeposited CNTs. Upon drying, the CNT layer compresses into a dense membrane structure under evaporation-induced capillary forces. Meanwhile, because of the excellent intrinsic electrical property of the CNTs, the high conductivity of the CNT/Cu electrode can be recovered for the EPD process to resume. Hence, the thickness of the CNT-HFMs can be controlled either by changing the duration of the EPD process or by performing multiple EPD steps, as exemplified in Figures 2 and S3 (Supporting Information) showing the as-fabricated CNT-HFMs with various thicknesses. The curve, displaying the dependence of membrane thickness on the number of EPD steps performed (expressed as y = 15.996x), is shown in Figure S4b (Supporting Information). The EPD-based layer-by-layer
Figure 3. SEM images of CNT-HFMs with controlled morphologies and structures: (a) bichannel CNT-HFMs; (b) trichannel CNTHFMs; (c) helical CNT-HFMs; (d) snake-like CNT-HFMs. Digital photographs of a helical CNT-HFM: (e) original state; (f) being compressed; (g) being extended. (f) Digital photograph of a snake-like CNT-HFM.
fabricated by using the appropriate number of Cu wires and controlling the position and distance of one another. Each of the independently addressable channels is separated by CNT
Figure 2. SEM images of CNT-HFMs with the same inner diameter of ∼150 μm but different outer diameters. Profile and cross-section of CNTHFMs with an outer diameter of 238 μm (a, d), 276 μm (b, e), and 390 μm (c, f), respectively. 8064
dx.doi.org/10.1021/es500506w | Environ. Sci. Technol. 2014, 48, 8062−8068
Environmental Science & Technology
Article
applied during the filtration process. As demonstrated in Figure S7b (Supporting Information), a 5 mg CNT-HFM can survive the high load imposed by a 50 g weight. The draw test of CNTHFMs also evidence their tensile strength, as shown in Figure S7c (Supporting Information), a 20 g weight linked with the CNT-HFM is successfully lifted. The typical stress−strain curve shows the measured tensile strength is about 6.0 MPa (Figure S8, Supporting Information), which is enough for CNT-HFMs to be handled. These test results present a promising possibility for the advancement of their practical application. Membrane Performance Evaluation. Determination of Permeation Flux. To evaluate the permeation flux of the prepared CNT-HFMs, they were compared with commercial commonly used polyvinylidene fluoride HFMs (PVDF-HFMs), featuring an average pore size of 100 nm (the datum was provided by the manufacturer, and their detailed information is summarized in Table 1). As shown in Table 1, CNT-HFMs achieve a flux (J) of 460 ± 50 L m−2 h−1, which is 1.8 times higher than those achieved by PVDF-HFMs. Two major factors are likely to contribute to the high flux of the CNT-HFMs. First, CNT-HFMs possess a higher porosity of 86 ± 5% compared with that of PVDF-HFMs (76 ± 4%) (Table 1). The SEM pictures in Figure S9 (Supporting Information) also demonstrate that the CNT-HFMs are much porous than PVDF-HFMs. Higher porosity indicates that a higher number of pores and channels are available for water transportation. Finally, unlike PVDF-HFMs, whose pores generally derived from phase inversion, CNT-HFMs have no dead-end pores, featured by nanofiber membranes.30 Antifouling of CNT-HFMs. The fouling resistance of the CNT-HFMs was assessed by using polystyrene microspheres with diameters of 114 and 362 nm to simulate suspended particles in wastewater. As illustrated in Figure 4a, the PVDFHFMs display a rejection of 87% of 114 nm polystyrene microspheres at the beginning, and the value achieves 100% when the permeate volume reaches 6 mL cm−2. In contrast, CNT-HFMs can achieve a 100% polystyrene microspheres rejection throughout the whole process. Although PVDFHFMs shows a 100% rejection of 362 nm polystyrene microspheres, their flux declines rapidly to only 61% of the original value when the permeate volume reaches 10.8 mL cm−2 (Figure 4b). More obviously, the flux of PVDF-HFMs decreased sharply to 44% of original value after a permeate volume of 10.8 mL cm−2 when 114 nm polystyrene microspheres were chosen as removal targets. Contrarily, after reaching a permeate volume of 12 mL cm−2, CNT-HFMs only showed a drop of 29% in the permeate flux, much less than that
partitions but not isolated absolutely from others owing to pores in CNT partitions. The new multichannel membranes open the immense potentials for sustainable water production via vacuum membrane distillation.29 Interestingly, in addition to straight CNT-HFMs, helical and snake-like CNT-HFMs (Figure 3c−h) could be prepared from the corresponding templates, as shaped by a cylindrical bar. The template-based synthesis of CNT-HFMs presents a versatile route for the development of multifunctional and integrative membrane systems, which are promising for efficient and smart separation processes, and other applications such as sensing, microfluidic devices, and microelectronics. Physical Properties of CNT-HFMs. The prepared CNTHFMs feature a porosity of 86 ± 5%, and it is much higher than that of the majority of commercial membranes with approximately the same pore size (generally
