γ-alumina ultrafiltration membrane

γ-alumina ultrafiltration membrane

Journal of Membrane Science 550 (2018) 26–35 Contents lists available at ScienceDirect Journal of Membrane Science journal homepage: www.elsevier.co...

2MB Sizes 0 Downloads 3 Views

Journal of Membrane Science 550 (2018) 26–35

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Fabrication and in-situ fouling mitigation of a supported carbon nanotube/γalumina ultrafiltration membrane Hengyang Maoa, Minghui Qiua, Xianfu Chena, Hendrik Verweijb, Yiqun Fana,

T



a

State Key Laboratory of Materials-Oriented Chemical Engineering, College of Chemical Engineering, Nanjing Tech University, No.5 Xin Mofan Road, Nanjing 210009, PR China b Department of Materials Science and Engineering, College of Engineering, Ohio State University, 2041 N College Road, Columbus, OH 43210, United States

A R T I C L E I N F O

A B S T R A C T

Keywords: Ultrafiltration Piezoelectric Carbon nanotubes Alumina Anti-fouling membrane

A novel ultrafiltration (UF) membrane with built-in defouling capability was made, characterized and tested for its water purification performance. The asymmetric ultrafiltration membrane with pore size of ~ 8 nm is obtained by coating a carbon nanotube (CNT)/γ-alumina composite layer on a porous PbZr0.52Ti0.48O3 (PZT) piezoelectric support. The PZT support not only serves to provide mechanical strength to the membrane, but can also be used to generate ultrasound by application of an alternating voltage (AV). This ultrasound, in turn, avoids and/or removes any fouling during filtration. The conducting composite layer serves both as a size-selective membrane and an electrode. The optimum membrane composition was a 1:1 CNT to alumina weight ratio at a sintering temperature of 600 °C. PZT-supported membrane structures were poled with a 3 × 103 kV/m electric field. Filtration of poled and unpoled membranes was carried out with a 2.5 g/L dextran solution to test antifouling performance. It was found that a poled membrane with application of a 20 V AV had a stable permeance of 55.6 L m−2 h−1 bar−1. The permeance of an unpoled membrane without application of a voltage was 31.0 L m−2 h−1 bar−1. This 79% permeance increase is ascribed to the mitigation of fouling during the filtration of the dextran solution.

1. Introduction Membranes are widely used in water treatment and purification due to their high cost-efficiency and minimal environmental impact [1]. However, water purification membranes tend to foul. They are covered by or impregnated with retained particles, colloids, macromolecules and precipitates [2]. This results in significant decrease in flux, and consequently increase of operational costs and the need for off-line cleaning and premature failure of membranes. Fouling is the foremost technical challenge in membrane filtration [3]. Many research efforts have been made to develop methods for limiting or reducing membrane fouling. Examples include application of vibration [4], gas sparging [5], electrical fields [6] and ultrasound [7]. After the application of ultrasound (US) was reported for the first time, in 1980 [8], several more studies of membrane cleaning [9,10] and fouling control [11–13] followed. It was confirmed that, when intense US waves propagate through a liquid, gas bubbles form in the negative pressure waves when the liquid's local tensile strength is exceeded. These bubbles rapidly grow and subsequently collapse in the positive waves resulting in a strong localized energy release. This process is known as cavitation, often utilized in the cleaning of surfaces ⁎

[14,15]. Kobayashi et al. [16,17] investigated the effects of ultrasonic conditions during cross-flow filtration with flat sheet membranes. They found that ultrasound frequency, power density and the irradiation direction had a significant effect on flux recovery after treatment. Gondrexon et al. [18,19] investigated the effect of ultrasound on membrane fouling during ultrafiltration. They found that this application of ultrasound significantly increased the water permeance for ultrafiltration of nano-particles, natural clay and skim milk. A permeance enhancement factor of 1.6–13.5 was found. Recently, the in-situ generation of ultrasound by a piezoelectric layer inside the membrane was introduced to sidestep the requirements of external ultrasound generation and/or off-line cleaning [20–23]. We believe that this use of a built-in structure will result in a better efficiency by a sophisticated optimization of continuous, in-line fouling mitigation. In addition, we anticipate that the ultrasound, emanating from the membrane structure will disrupt the formation of laminar boundary layers. These boundary layers adversely affect salt rejection and are thought to facilitate the settling of fouling particulates. To have the piezoelectric layer emit ultrasound, AVs are applied between both sides of the PZT supported CNT/γ-alumina composite membrane. Insitu sound generation has also been reported [20–23] for membranes

Corresponding author. E-mail address: [email protected] (Y. Fan).

https://doi.org/10.1016/j.memsci.2017.12.050 Received 6 October 2017; Received in revised form 15 December 2017; Accepted 17 December 2017 Available online 18 December 2017 0376-7388/ © 2017 Elsevier B.V. All rights reserved.

Journal of Membrane Science 550 (2018) 26–35

H. Mao et al.

side electrode and size-exclusion filtration are combined in a separate membrane that is supported by the PZT layer. The conducting thin film membrane is a composite of carbon nanotube (CNT) and 15 nm γ-alumina particles. We have chosen for a composite structure containing CNTs since CNTs have a superior electrical conductivity and thermochemical stability [25,26]. Ultrafiltration membranes consisting of aligned CNT [27,28] and CNT networks [29,30] were demonstrated to have a high water permeance due to their ultrathin and fibrous morphology. Jassby et al. [31] synthesized conducting ultrafiltration membranes based on the CNT and polysulfone. Huang et al. [32] prepared ultrafiltration membrane that had a composite CNT/PVDF structure. Both membrane types adhered well to their polymeric supports (PS and PVDF) and had excellent electrical conductivity. But during our initial explorations and the literature [33], it was found that 100% CNT networks are too much prone to mechanical damage from wear and scratching. Hence it is necessary to enhance the mechanical properties of the CNT networks for practical application. Methods such as CNT modification [34,35] and adding binders [36,37] were found to improve the mechanical properties of CNT networks. In this work, we studied the use of CNT/γ-alumina composite structures for the membrane. The alumina particles form strong necks to reduce their external surface. We anticipated that the presence of hydrophilic alumina between the hydrophobic CNT would improve the adhesion with the hydrophilic PZT ceramic support, and at the same time improve scratch resistance (Scheme 1d). While the effect of CNT reinforcement on the mechanical properties of alumina is well-documented [38–42], no studies have been reported for the effect of alumina particle additions to CNT networks. In the present work, composite CNT/γ-alumina membrane layers were prepared by dip-coating CNT/γ-alumina suspensions on porous PZT substrates, followed by thermal processing. The effect of thermal processing and composition were studied by SEM, EDS, N2 sorption and nano scratch tests. The eventual optimized composite membrane was poled in an electric field of 3 × 103 kV/m at 120–140 °C to obtain permanent piezoelectricity. Water purification performance was tested with dextran solutions to evaluate the membranes’ transport properties and anti-fouling performance.

made of piezoelectric polyvinylidene fluoride (PVDF) polymer. PDVF has at least four known crystalline structures (α, β, γ and δ) and is commonly used for micro- and ultra-filtration (MF and UF). The membranes studied had a pore size of 220 nm, and the PVDF membrane, was poled in an electric field of 16.3 × 103 kV/m to transform all crystalline structures to the all-trans (β) phase which is mainly responsible for its piezoelectric properties. It was found that application of 0.5–1 kHz AVs to poled PVDF membranes during filtration resulted in an increase of the permeance and, simultaneously, a decrease of the fouling rate. After 30 min filtration, the permeance of a PVDF membrane with no voltage applied was 50 L m−2 h−1, while the permeance of a PVDF membrane with an effective AV of 10 V application was increased to 160 L m−2 h−1. In another study, lead zirconate titanate (PZT), an inorganic piezoelectric material, was used to fabricate a macro-porous membrane with an average pore size at about 300 nm [24]. It was found that application of an AV at 70 kHz, at sonic resonance, to a poled PZT membrane during dead-end filtration of a 1 µm latex dispersion resulted in complete fouling mitigation, the permeance was maintained at its initial value. The geometries of PVDF and PZT piezoelectric membranes reported in literature are shown in Scheme 1a and b. A porous steel electrode is present at the permeate-side of the membrane, another porous steel electrode is present at or above the feed-side of the membrane. A feedside electrode at mm's distance away from the membrane is referred to as a “remote electrode”. A remote electrode, the “rod electrode” in Scheme 1b, has the advantage that it does not adversely affect the membrane process at the feed side, for instance by reduction of filtration surface, promotion of localized fouling, and inhomogeneous field distribution. Remote application of a voltage to the piezoelectric layer is possible due to capacitive coupling in combination with minor ionic conductivity in the liquid [24]. However the distance between the remote electrode and the membrane surface may lead to substantial, unwanted electrical energy dissipation. In this paper, we present fabrication and characterization of a composite membrane with an asymmetric structure as shown in Scheme 1c. Disk-shaped porous PZT supports with 300 nm pore size are made by pressing and sintering and provided with a conductive layer with an average pore size of < 10 nm. In this structure, the functions of feed-

Scheme 1. a, b, c: Schematic of the cross flow membrane module used for membrane operation with in-situ ultrasound generation: 1) symmetric piezoelectric membrane, 2) permeate side mesh electrode, 3) remote rod electrode, 4) asymmetric structure with a conductive membrane layer and a piezoelectric support. d: Composite membrane fabrication process.

27

Journal of Membrane Science 550 (2018) 26–35

H. Mao et al.

2. Experimental

PROBES TECH, China). The composition of the PZT support was abtained with an X-ray fluorescence elemental analyzer (3080E3, Rigaku, Japan).

2.1. Preparation of porous PZT support

2.6. Sonic resonance characterization

Porous PbZr0.52Ti0.48O3 (PZT) disks with 30 mm diameter and 2 mm thickness were used to support the CNT/γ-alumina composite membranes and to generate ultrasound from within the membrane structure. They were made by dry pressing S42 PZT powder (500 nm, Sunnytec, China) at 10 MPa, followed by sintering at 950 °C in air for 2 h with heating and cooling rate of 2 °C/min. Supports thus obtained had an average pore size of 300 nm (Fig. S1) and a permeability of 85 ± 6 L m−2 h−1 bar−1.

Sonic resonance frequencies and emission were measured with the sample immersed in water inside a PMMA tank with 15 cm diameter and 15 cm height, see Fig. 2. The supports and supported membranes were placed between two stainless mesh electrodes, connected to a waveform generator (DG1022, Rigol, China). A sinusoidal voltage with a 20 V amplitude was applied across the membrane, while sweeping the frequency over the 1–300 kHz range. Acoustic emissions from the membrane were detected by a hydrophone (RHSM-10, HAARI, China) that was positioned ~ 3 cm above the membrane surface. The hydrophone signal was monitored with a digital oscilloscope (DS1052E, Rigol, China).

2.2. Preparation of CNTs-alumina dispersion Homogeneous multi-wall carbon nanotube (CNTs, XFM19, XFNANO, China) dispersions were prepared as described in [43]. 200 mg CNTs were added to a 100 ml aqueous solution of 200 mg block copolymer (BCP, Sigma-Aldrich). The mixture was sonified with a horn at 500 W for 10 min while cooling with an ice bath. A 2 wt% Boehmite sol with 15 nm particle size (Fig. S2) was synthesized as described in [44]. The as-prepared CNT-BCP dispersion was mixed with the Boehmite (alumina precursor) sol at different ratios followed by ultrasonic horn dispersion as described before.

2.7. Permeance and dextran rejection testing Pure water permeance and the retention of dextran were measured under trans-membrane pressure of 1.5 bar and a linear crossflow velocity of 2 m/s at the operation temperature of 25 ± 2 °C. After the 12 h measurements with the dextran solution were complete, the membrane was cleaned by distilled water and the clean water permeance was obtained again to evaluate the effect of in-situ ultrasound on membrane fouling [45]. Filtration experiments were carried out using a homemade setup, shown in Fig. S5. It included a 6 mm Ø stainless steel mesh electrode in contact with the permeate side of the membrane, and a copper wire in contact with the feed side of the membrane. The dextran concentrations were 2.5 g/L for 10,000 Da, 1 g/L for 40,000 and 70,000 Da, and 2 g/L for 500,000 Da. Analysis of the feed and permeate solutions was conducted by gel permeation chromatography (GPC, Waters, America).

2.3. Preparation of PZT supported CNTs-alumina composite membranes Pure CNT membranes and CNT/γ-alumina composite membranes were prepared on the porous PZT support by dip coating. The dipcoated supports were dried at room temperature for 12 h and further dried for 12 h at 110 °C. Subsequently, the specimens were sintered in a tube furnace in Ar atmosphere with 2 °C/min ramp and 2 h hold time at 500, 600 or 700 °C. Unsupported membrane material was obtained by pouring the corresponding dispersion into Petri dishes followed by drying and sintering at the same temperatures as were used for the PZT supported membrane. Thus obtained samples were crushed to fine powders prior to characterization.

3. Results and discussion 3.1. Effect of thermal processing on PZT support properties

2.4. Membrane poling Since PZT may react with CNTs in a neutral or reducing atmosphere, the stability of porous PZT supports covered with CNTs was tested by heating for 2 h in Ar. Fig. 1a shows the surface microstructure of PZT supports after this treatment. No apparent changes in surface microstructure were observed after heating at 500–700 °C, but heating to 800 and 900 °C resulted in changes of the microstructure. The Pb element distributions were characterized by energy dispersive X-ray spectroscopy (EDX) as shown in Fig. 1b. The element distributions revealed that the Pb concentration in the PZT surface had increased after treatment at 800 and 900 °C. This is ascribed to a reduction of lead oxide in the PZT supports by carbon. The reduction led to the formation of small pores in the PZT grains and coverage of metallic Pb which affected the piezoelectric properties of the PZT supports. Consequently the treatment temperature of the composite membranes was kept at 500–700 °C. To verify the overall composition of the porous PZT supports after reaction, fragments were ground into powder and analyzed by X-ray fluorescence (XRF). As shown in Table 1, the Pb, Zr and Ti concentrations did not change after treatment at 500 and 600 °C. With higher treatment temperatures, the Pb content had decreased while the Zr and Ti content increased. This result further narrowed the maximum treatment temperatures to 500–600 °C. The water permeance of the PZT supports, heat treated as described, is also shown in Table 1. It was found that the original value of 85 L m−2 h−1 bar−1 was not affected after thermal treatment at 500–700 °C, in agreement with the surface micro-structure characterization in Fig. 1a. PZT supports after treatment at 500–700 °C were

The PZT-supported CNT/γ-alumina composite membranes and the uncovered porous PZT supports were poled by application of a strong direct electric field (~3×103 kV/m) at 140 °C. This made that the orientation of ferro-electric domains changed from random to aligned, with a net polarization vertical to the circular membrane deposition surface. During poling the membranes were mounted between two flat electrodes and placed on a hot plate with a glass cover as shown in Fig. S3 in the Supporting information. 2.5. Support and membrane characterization A field emission scanning electron microscope (FESEM) (S-4800, Hitachi, Japan) with a 5 kV accelerating voltage, coupled with an X-ray energy dispersive spectrometer was used to characterize the morphology and surface composition of the supports and membranes. The mechanical durability of the membranes was tested by immersing samples in 50 ml deionized water and treating them with an ultrasonic cleaner (SK5210L, Kudos, China) at 80 W for 1 min. The turbidity of the water after this treatment was then obtained (2100 N, Hach, China) and used as an indication for mechanical durability. The volumetric shrinkage of unsupported composite layers were measured by a thermal dilatometer (DIL, Netzsch, Germany). The scratch resistance of the membranes were measured using a nano-scratch tester (NanoTest, MML, England). The hydrophobicity of membrane was derived from the contact angle with water (A-100P, MAIST, China). The conductivity of the composite layers was obatined with a four-probe meter (RTS-8, 4 28

Journal of Membrane Science 550 (2018) 26–35

H. Mao et al.

Fig. 1. a: Surface microstructure. b: Surface distribution of Pb for PZT supports vs treatment temperature.

of 500, 600, 700 °C, are shown in Figs. 3 and 4. The extent of membrane delamination after sonication in water for 1 min with sonifier power set at 80 W (Fig. S4) was obtained by visual inspection and derived from the turbidity of the water that surrounded the sample. It was found that delamination decreased with a increasing γ-alumina concentration and treatment temperature. But without alumina present, increasing the temperature had no obvious effect. The volumetric shrinkage of unsupported composite layers during thermal treatment was measured with a dilatometer. As shown in Fig. 3d, the volumetric shrinkage increased with treatment temperature and alumina content. Structures consisting of just CNT have an interconnected rigid structure that is not affected by the treatment temperatures that we used. However the alumina matrix shrinks due to dehydration of the boehmite precursor, followed by conversion into γalumina and further sintering of that phase. In the CNT/γ-alumina structure this results in volumetric shrinkage, and an increase of mechanical durability in agreement with observations.

poled in air at 140 °C for 1 h, followed by sonic resonance testing as shown in Fig. 2a. The resonance frequency and intensity of supports that were poled after treatment at 500 and 600 °C were the same within experimental error as those of the original support. The amplitude for the support poled after treatment at 700 °C decreased by 50% which is ascribed to the reaction between PZT and the CNTs. This further supported the use only of treatments temperatures of < 700 °C. 3.2. Effect of CNT to alumina ratios on mechanical durability and adhesion of composite membranes As indicated, ultrasound generated by a piezoelectric membrane support can be very beneficial in continuous fouling mitigation during filtration. But the generated ultrasound also acts on the CNT/γ-alumina membrane top-layer and can potentially damage its composite structure. Testing results for CNT/γ-alumina composite layers with CNT to alumina weight ratios of 1:0, 3:1, 1:1, 1:3 and treatment temperatures Table 1 Effect of thermal treatment temperature on properties of PZT support.

PbO/% ZrO2/% TiO2/% Permeance/L m−2 h−1 bar−1

Original

500 °C

600 °C

700 °C

800 °C

900 °C

64.51 19.58 11.59 85 ± 4.6

64.6 Δ 19.58 Δ 11.54 Δ 84.7 ± 3.8

64.54 Δ 19.63 Δ 11.53 Δ 85.6 ± 4.3

64.33 ↓ 19.61 Δ 11.65 ↑ 83.6 ± 4.9

63.49 ↓ 20.01 ↑ 12.1 ↑ N/A

60.38 ↓ 20.53 ↑ 14.12 ↑ N/A

29

Journal of Membrane Science 550 (2018) 26–35

H. Mao et al.

Fig. 2. a: Schematic of the resonance testing setup (not to scale). b: Resonance curves of poled PZT supports vs thermal treatment temperature.

Fig. 3. Mechanical strength tests. a: Test system. b: Surface morphology of treated composite membranes. c: Turbidity of water in the bath after treatment. d: Volumetric shrinkage of unsupported membrane material heated in Ar.

the failure point, in the scratch profile, see Fig. 4b:

To fabricate sufficiently strong membrane composites while preserving the piezoelectric effect of the supports, a thermal treatment temperature of 600 °C was chosen for further studies. The interfacial adhesion between the CNT/γ-alumina composite layer and the porous PZT support was characterized by a nano-scratching method [46,47]. The Nano-Test system used in this work was capable of high resolution mechanical property analysis in the nano and micro meter regimes and provided a critical load (Lc) for the composite layer-ceramic interface. As shown in Fig. 4a, Lc can be obtain from the position of the first peak,

(i) At 190 µm scan displacement (Lc = 52 mN, d = 1.9 µm) of composite with 75 wt% alumina (R = 1:3). (ii) At 182 µm scan displacement (Lc = 47 mN, d = 2.2 µm) of composite with 50 wt% alumina (R = 1:1). (iii) At 140 µm scan displacement (Lc = 35 mN, d = 2.5 µm) of composite with 25 wt% alumina (R = 3:1).

30

Journal of Membrane Science 550 (2018) 26–35

H. Mao et al.

Fig. 4. Nano-scratch test results for CNT/γ-alumina composite membranes on porous PZT. a: Scratch profile. b: Load profile. c-e: SEM surface images of scratches. f-h: Cross sections of the supported membranes.

membranes increased with the alumina addition indeed. The molecular weight cut-off (MWCO) values were of 65.5–117.2 kDa, 30.6–39.8 kDa and 25.6–36.7 kDa for CNT/γ-alumina composite membranes with CNT to alumina weight ratios of 3:1, 1:1 and 1:3, respectively. The diameter of dextran coils obtained from the MWCO values was calculated to be 11–14 nm, 8–9 nm and 7–8 nm. These diameters were estimated as the Stokes-diameter obtained with (1) [48]:

The critical load (Lc) was taken as a measure of the adhesive strength between the composite layer and the PZT support. After Lc was reached, a complete film removal occurred with substantial scratch damage at the end of trace. The SEM images in Fig. 4c-e show a clear delamination of membranes with a low alumina content. The thickness of the composite layer obtained by SEM, see Fig. 4f-h, was in agreement with the scratch depth. The adhesion between the composite membrane and the PZT support was found to increase with the alumina content. This is ascribed to the affinity of hydrophyllic γ-alumina for the likewise PZT surface. In addition, the small alumina particle size, ~ 15 nm, is beneficial for improving the actual contact area. As shown in Figs. 3 and 4, the mechanical stability and adhesion of the membranes increased with the alumina ratio, and the composite layers with alumina concentration of 50 wt% and 75 wt% have similar mechanical properties. Hence we concluded that a concentration of 50 wt% alumina is sufficient.

r = 0.33 × (Mw )0.46 ,

(1)

in which r is the molecular radius (Å), and Mw is the molecular weight (Da). SEM surface images of composite membranes are shown in Fig. 6. These images confirm that the alumina particles filled the space between the CNT networks, resulting in the decrease of the effective pore size of composite membranes. Inspection of SEM images of the support/ membrane interfacial area, shown in Fig. S6, revealed that the CNT network fully blocked the infiltration of the 15 nm Boehmite precursor particles in the 300 nm pores of the PZT support. This conclusion is in agreement with a similar effect, found in previous work: In dip coating 40 nm titania particles on supports with 2–3 µm pores, infiltration was effectively blocked by an intermediate layer of titania fibers with 200–400 nm diameter and 5–10 µm in length [49]. The pure water permeance results are shown in Fig. 7a. With a thermal treatment temperature of 600 °C, the permeance decreased from 75 to 68 L m−2 h−1 bar−1 with increasing alumina content in the membranes. This slight decrease is ascribed to occupation of the space between the CNT by alumina nanoparticles. This effect is not very outspoken indicating that the composite membrane's overall resistance is not effected very much by the clean top-layers. Surface wettability results for the composite membranes are shown in Fig. 7b. It was found that pure water permeates into the membrane structure, driven by capillary forces. This resulted in a gradual decrease of the effective composite contact angle (CA), as shown in Fig. 7c. The pertaining capillaries are likely the pores between the hydrophyllic alumina particles. The initial CA is thought to be related to the effective hydrophilicity of the membrane material while the rates of CA decrease are mainly connected with the structure of composite membrane.

3.3. Effect of the CNT to alumina ratios on the pore size of composite membrane structures The pore structure of unsupported CNT/γ-alumina material was investigated by recording gas adsorption/desorption isotherms, as shown in Fig. 5a. The isotherms of three membrane compositions were of type IV, which indicates a normal mesoporous structure (pore diameter between 2 and 50 nm). Pore size distributions were obtained by using a non-local density functional theory (NLDFT). As shown in Fig. 5b, a relatively small average pore size and a narrow pore size distribution were observed for composites with a higher alumina content. With increasing alumina content, the average pore diameter of composites decreased from 15 nm to 7 nm which is ascribed to an increasing occupation by alumina particles of space between the CNT networks. Membranes that consist of just γ-alumina have a typical pore size of ~ 4 nm [26]. The smaller pore sizes are favorable for obtaining a better size exclusion separation selectivity. Aqueous dextran solution retention results are shown in Figs. 5c-e, 4 membrane samples with the same CNT to alumina ratio were tested for each figure. It was found that the rejection selectivity of the composite 31

Journal of Membrane Science 550 (2018) 26–35

H. Mao et al.

Fig. 5. Effect of CNT to γ-alumina weight ratios on a: N2 adsorption/desorption isotherms. b: Derived pore size distributions of unsupported composite membrane material. c-e: Dextran retention for optimized composite membranes.

In summary, the membranes’ pore size and permeance decreased with the alumina concentration, the membranes with alumina concentration of 50 wt% and 75 wt% had a similar pore size of ~ 8 nm. Hence, 50 wt% alumina was considered sufficient for the best possible size selectivity.

Clearly, composite membranes with higher alumina content had a lower initial CA and a smoother CA decrease. These results confirm that increasing the alumina content of the membrane top-layers results in an increased hydrophilicity and a decreased pore size. This is, once more, in agreement with trends in the observed water permeance.

Fig. 6. SEM micrographs of the composite membranes vs different ratios of CNTs to alumina (heat-treated at 600 °C in Ar).

32

Journal of Membrane Science 550 (2018) 26–35

H. Mao et al.

Fig. 7. a: Permeance of composite membranes. b: Time dependence of water contact angles on composite membranes. c: Spreading and penetration behavior of water droplets.

Fig. 8. Membrane performance with and without alternating voltage for poled and unpoled composite membranes. a: Sheet conductivity of composite membranes. b: Resonance curves of poled membranes. c: Water permeance. d: MWCO of dextran solutions.

33

Journal of Membrane Science 550 (2018) 26–35

H. Mao et al.

piezoceramic substrate and conductive membrane top-layer that is used as the feed-side electrode. The membrane layer is a composite of carbon nanotubes (CNTs) and γ-alumina. It is deposited by dip coating of CNT+Boehmite dispersions, followed by thermal processing with an preferred consolidation temperature of 600 °C. The optimum membrane composition has a CNT to alumina weight ratio of 1:1 which results in sufficient conductivity, a small pore size and an excellent mechanical durability. Poled membrane structures, activated with 20 V alternating voltage at a frequency of 190 kHz have a stable water permeance of 55.6 L m−2 h−1 bar−1 at room temperature, and a dextran molecular weight cutoff of 30–40 kDa (ultrafiltration). The use of unpoled structures and membrane operation without application of an electric field resulted in rapid flux decline due to fouling. It was found that the presence of γ-alumina in the membrane structure was particularly beneficial for mechanical durability and pore size reduction. The pore size reduction is ascribed to the fact that the γ-alumina particles effectively filled the space between the CNTs. More work will be needed for further optimization of the multi-layer structure, long-term stability, more environment friendly (lead-free) piezoelectric compositions, scale-up and application on tubular geometries and cost-effectiveness.

3.4. Membrane poling and anti-fouling tests As shown in Fig. 8a, the electrical conductivity of composite membranes increased slightly with alumina content at low concentrations but it decreased substantially at higher alumina content. As shown in Fig. 3d, the CNT network is compressed by small amounts of alumina. This implies in a denser CNT packing and possibly a better electrical contact between the CNTs. However at higher alumina contents, the electrically insulating alumina starts to separate the individual CNTs which results in a decrease of conductivity. The occurrence of a maximum indicates that the conductivity of packed CNTs structures can be optimized with the concentration and morphology of nano-particle additions. Supported membranes with a CNT to alumina weight ratio of 1:1 and a treatment temperature of 600 °C were poled at temperatures of 120–140 °C in a high electrostatic voltage field of 3 × 103 kV/m. The ultrasound generated by electrical activation of poled composite membranes is shown in Fig. 2a. All poled membranes generated a detectable signal in the ultrasonic range (~ 190 kHz). The emission signal of composite membranes poled at different temperatures is shown in Fig. 8b. The maximum response of 6 mV at 20 V activating voltage was obtained after poling at 140 °C. This result was in agreement with Fig. 2b, and shows that the presence of a CNT/γ-alumina composite layer does not affect the piezoelectricity of the PZT support. Composite membranes poled at 140 °C were also used for permeation measurements. A total of four filtration tests, see Fig. 8c-d, with 2.5 g/L dextran solutions were performed as follows: (1) Poled composite membrane without voltage application; (2) Poled composite membrane with the application of a 20 V voltage at 190 kHz; (3) Unpoled composite membrane without voltage application; (4) Unpoled composite membrane with the application of a 20 V voltage at 190 kHz. The membrane performance measurements were carried out in two stages. The first was 12 h ultrafiltration of a dextran solution and the second was a pure water flux measurement of the cleaned membranes. The results are shown in Fig. 8c, the initial permeance values for the dextran solution were 70 ± 3 L m−2 h−1 bar−1. With application of an alternating voltage, the permeance of the poled composite membrane decreased by nearly 11% within 1 h and remained stable after that. In the other three tests the permeance decreased over 37% within 1 h while it gradually decreased further to 30.5 ± 0.6 L m−2 h−1 bar−1 within 12 h. This clearly demonstrates the in-situ fouling mitigation of ultrasound, generated from within the membrane structure. The recovered water permeance of the membrane which coupled with ultrasound is 60.6 L m−2 h−1 bar−1, also higher than the recovered permeance of other membranes (52.9 ± 0.8 L m−2 h−1 bar−1). This indicate that in-situ ultrasound can mitigate not only reversible fouling such as concentration polarization and pollutant deposition but also irreversible fouling which caused by infiltration of pollutant into membrane pores [50] effectively. The dextran retention did not vary during the four filtration tests as shown in Fig. 8d. And since this indicates that the effective pore size did not change, we concluded that the microstructure of the composite membrane was not affected by the ultrasonic excitation and water permeation. Cavitation has not been positively identified in the resonance detected by the hydrophone. Therefore, microstreaming or vibration resulting in diminished concentration polarization, absence cake formation and less foulants deposition is assumed to be the predominant anti-fouling mechanism [51–53].

Acknowledgements This study was financially supported by the National Natural Science Foundation of China (21506093, 91534108), the Natural Science Foundation of Jiangsu Province (BK20150947), the National key R&D plan (2016YFC0205700) and the National High Technology Research and Development Program of China (2012AA03A606). The authors thank Xiangli Kong, Sihan Li, Wenhuai Li, Liqun Liu, Huilin Wang and Enyu Zhang from the chemical engineering institute at Nanjing Tech University for preparation of schematic diagrams and PZT substrates. Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.memsci.2017.12.050. References [1] M.M. Pendergast, E.M.V. Hoek, A review of water treatment membrane nanotechnologies, Energy Environ. Sci. 4 (2011) 1946–1971. [2] Y. Lv, C. Zhang, A. He, S.-J. Yang, G.-P. Wu, S.B. Darling, Z.-K. Xu, Photocatalytic nanofiltration membranes with self-cleaning property for wastewater treatment, Adv. Funct. Mater. 27 (2017) 27–36. [3] W.Z. Yu, Y.J. Yang, N. Graham, Evaluation of ferrate as a coagulant aid/oxidant pretreatment for mitigating submerged ultrafiltration membrane fouling in drinking water treatment, Chem. Eng. J. 298 (2016) 234–242. [4] M.Y. Jaffrin, Dynamic shear-enhanced membrane filtration: a review of rotating disks, rotating membranes and vibrating systems, J. Membr. Sci. 324 (2008) 7–25. [5] A. Ceron-Vivas, J.M. Morgan-Sagastume, A. Noyola, Intermittent filtration and gas bubbling for fouling reduction in anaerobic membrane bioreactors, J. Membr. Sci. 423 (2012) 136–142. [6] X. Fan, H. Zhao, X. Quan, Y. Liu, S. Chen, Nanocarbon-based membrane filtration integrated with electric field driving for effective membrane fouling mitigation, Water Res. 88 (2016) 285–292. [7] M. Xu, X. Wen, X. Huang, Z. Yu, M. Zhu, Mechanisms of membrane fouling controlled by online ultrasound in an anaerobic membrane bioreactor for digestion of waste activated sludge, J. Membr. Sci. 445 (2013) 119–126. [8] Y. Okahata, H. Noguchi, Ultrasound-responsive permeability control of bilayercoated capsule membranes, Chem. Lett. 10 (1983) 1517–1520. [9] E. Alventosa-deLara, S. Barredo-Damas, M.I. Alcaina-Miranda, M.I. Iborra-Clar, Study and optimization of the ultrasound-enhanced cleaning of an ultrafiltration ceramic membrane through a combined experimental-statistical approach, Ultrason. Sonochem. 21 (2014) 1222–1234. [10] C.C. Kan, D.A.D. Genuino, K.K.P. Rivera, M.D.G. de Luna, Ultrasonic cleaning of polytetrafluoroethylene membrane fouled by natural organic matter, J. Membr. Sci. 497 (2016) 450–457. [11] M. Cai, S. Wang, H.-h. Liang, Optimization of ultrasound-assisted ultrafiltration of radix astragalus extracts with hollow fiber membrane using response surface methodology, Sep. Purif. Technol. 100 (2012) 74–81. [12] L.L.A. Koh, N. Hanh Thi Hong, J. Chandrapala, B. Zisu, M. Ashokkumar,

4. Conclusion Membrane structures that include a piezoelectric layer can be activated with an alternating electrical field to generate ultrasound from within. This in-situ ultrasound emission can be very effective in avoiding or removing fouling during water filtration. In this study, a new self-cleaning membrane was developed, consisting of a porous 34

Journal of Membrane Science 550 (2018) 26–35

H. Mao et al.

[13] [14] [15]

[16]

[17]

[18]

[19]

[20]

[21]

[22] [23]

[24]

[25] [26]

[27] [28]

[29]

[30]

[31]

[32]

[33]

S.E. Kentish, The use of ultrasonic feed pre-treatment to reduce membrane fouling in whey ultrafiltration, J. Membr. Sci. 453 (2014) 230–239. X. Li, J. Yu, A.G.A. Nnanna, Fouling mitigation for hollow-fiber UF membrane by sonication, Desalination 281 (2011) 23–29. J. Jordens, B. Bamps, B. Gielen, L. Braeken, T. Van Gerven, The effects of ultrasound on micromixing, Ultrason. Sonochem. 32 (2016) 68–78. F. Mekki-Berrada, T. Combriat, P. Thibault, P. Marmottant, Interactions enhance the acoustic streaming around flattened microfluidic bubbles, J. Fluid Mech. 797 (2016) 851–873. T. Kobayashi, T. Kobayashi, Y. Hosaka, N. Fujii, Ultrasound-enhanced membranecleaning processes applied water treatments: influence of sonic frequency on filtration treatments, Ultrasonics 41 (2003) 185–190. A. Maskooki, T. Kobayashi, S.A. Mortazavi, A. Maskooki, Effect of low frequencies and mixed wave of ultrasound and EDTA on flux recovery and cleaning of microfiltration membranes, Sep. Purif. Technol. 59 (2008) 67–73. N. Gondrexon, L. Cheze, Y. Jin, M. Legay, Q. Tissot, N. Hengl, S. Baup, P. Boldo, F. Pignon, E. Talansier, Intensification of heat and mass transfer by ultrasound: application to heat exchangers and membrane separation processes, Ultrason. Sonochem. 25 (2015) 40–50. Y. Jin, N. Hengl, S. Baup, F. Pignon, N. Gondrexon, M. Sztucki, G. Gesan-Guiziou, A. Magnin, M. Abyana, M. Karrouch, D. Bleses, Effects of ultrasound on cross-flow ultrafiltration of skim milk: characterization from macro-scale to nano-scale, J. Membr. Sci. 470 (2014) 205–218. M.T. Darestani, T.C. Chilcott, H.G.L. Coster, Separation performance of PVDF membranes poled in intense electric fields, Sep. Purif. Technol. 118 (2013) 604–611. M.T. Darestani, H.G.L. Coster, T.C. Chilcott, Piezoelectric membranes for separation processes: operating conditions and filtration performance, J. Membr. Sci. 435 (2013) 226–232. H.G.L. Coster, T.D. Farahani, T.C. Chilcott, Production and characterization of piezo-electric membranes, Desalination 283 (2011) 52–57. M.T. Darestani, H.G.L. Coster, T.C. Chilcott, S. Fleming, V. Nagarajan, H. An, Piezoelectric membranes for separation processes: fabrication and piezoelectric properties, J. Membr. Sci. 434 (2013) 184–192. J.K. Krinks, M. Qiu, I.A. Mergos, L.K. Weavers, P.J. Mouser, H. Verweij, Piezoceramic membrane with built-in ultrasonic defouling, J. Membr. Sci. 494 (2015) 130–135. G. Lota, K. Fic, E. Frackowiak, Carbon nanotubes and their composites in electrochemical applications, Energy Environ. Sci. 4 (2011) 1592–1605. D.N. Futaba, K. Hata, T. Yamada, T. Hiraoka, Y. Hayamizu, Y. Kakudate, O. Tanaike, H. Hatori, M. Yumura, S. Iijima, Shape-engineerable and highly densely packed single-walled carbon nanotubes and their application as super-capacitor electrodes, Nat. Mater. 5 (2006) 987–994. B. Lee, Y. Baek, M. Lee, D.H. Jeong, H.H. Lee, J. Yoon, Y.H. Kim, A carbon nanotube wall membrane for water treatment, Nat. Commun. 6 (2015) 7109–7115. J.K. Holt, H.G. Park, Y.M. Wang, M. Stadermann, A.B. Artyukhin, C.P. Grigoropoulos, A. Noy, O. Bakajin, Fast mass transport through sub-2-nanometer carbon nanotubes, Science 312 (2006) 1034–1037. L. Hu, S. Gao, X. Ding, D. Wang, J. Jiang, J. Jin, L. Jiang, Photothermal-responsive single-walled carbon nanotube-based ultrathin membranes for on/off switchable separation of oil-in-water nanoemulsions, ACS Nano 9 (2015) 4835–4842. Z. Shi, W. Zhang, F. Zhang, X. Liu, D. Wang, J. Jin, L. Jiang, Ultrafast separation of emulsified oil/water mixtures by ultrathin free-standing single-walled carbon nanotube network films, Adv. Mater. 25 (2013) 2422–2427. A.V. Dudchenko, J. Rolf, K. Russell, W. Duan, D. Jassby, Organic fouling inhibition on electrically conducting carbon nanotube-polyvinyl alcohol composite ultrafiltration membranes, J. Membr. Sci. 468 (2014) 1–10. S. Wang, S. Liang, P. Liang, X. Zhang, J. Sun, S. Wu, X. Huang, In-situ combined dual-layer CNT/PVDF membrane for electrically-enhanced fouling resistance, J. Membr. Sci. 491 (2015) 37–44. B.S. Shim, J. Zhu, E. Jan, K. Critchley, N.A. Kotov, Transparent conductors from

[34]

[35]

[36] [37] [38]

[39] [40] [41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

[50] [51]

[52]

[53]

35

layer-by-layer assembled SWNT films: importance of mechanical properties and a new figure of merit, ACS Nano 4 (2010) 3725–3734. U. Dettlaff-Weglikowska, V. Skakalova, R. Graupner, S.H. Jhang, B.H. Kim, H.J. Lee, L. Ley, Y.W. Park, S. Berber, D. Tomanek, S. Roth, Effect of SOCl2 treatment on electrical and mechanical properties of single-wall carbon nanotube networks, J. Am. Chem. Soc. 127 (2005) 5125–5131. A. Kaskela, J. Koskinen, H. Jiang, Y. Tian, X. Liu, T. Susi, M. Kaukonen, A.G. Nasibulin, E.I. Kauppinen, Improvement of the mechanical properties of singlewalled carbon nanotube networks by carbon plasma coatings, Carbon 53 (2013) 50–61. C. Wang, L. Wang, Z. Xu, Enhanced mechanical properties of carbon nanotube networks by mobile and discrete binders, Carbon 64 (2013) 237–244. K.H. Kim, Y. Oh, M.F. Islam, Graphene coating makes carbon nanotube aerogels superelastic and resistant to fatigue, Nat. Nanotechnol. 7 (2012) 562–566. I. Ahmad, M. Unwin, H. Cao, H. Chen, H. Zhao, A. Kennedy, Y.Q. Zhu, Multi-walled carbon nanotubes reinforced Al2O3 nanocomposites: mechanical properties and interfacial investigations, Compos. Sci. Technol. 70 (2010) 1199–1206. C.N. He, F. Tian, S.J. Liu, A carbon nanotube/alumina network structure for fabricating alumina matrix composites, J. Alloy. Compd. 478 (2009) 816–819. K. Ahmad, W. Pan, Microstructure-toughening relation in alumina based multiwall carbon nanotube ceramic composites, J. Eur. Ceram. Soc. 35 (2015) 663–671. Y. Feng, K. Wang, C.H.J. Davies, H. Wang, Carbon nanotube/alumina/polyethersulfone hybrid hollow fiber membranes with enhanced mechanical and antifouling properties, Nanomaterials 5 (2015) 1366–1378. G.D. Zhan, J.D. Kuntz, J.L. Wan, A.K. Mukherjee, Single-wall carbon nanotubes as attractive toughening agents in alumina-based nanocomposites, Nat. Mater. 2 (2003) 38–42. X. Chen, M. Qiu, H. Ding, K. Fu, Y. Fan, A reduced graphene oxide nanofiltration membrane intercalated by well-dispersed carbon nanotubes for drinking water purification, Nanoscale 8 (2016) 5696–5705. X. Chen, W. Zhang, Y. Lin, Y. Cai, M. Qiu, Y. Fan, Preparation of high-flux gammaalumina nanofiltration membranes by using a modified sol-gel method, Microporous Mesoporous Mater. 214 (2015) 195–203. H. Liu, G. Zhang, C. Zhao, J. Liu, F. Yang, Hydraulic power and electric field combined antifouling effect of a novel conductive poly(aminoanthraquinone)/reduced graphene oxide nanohybrid blended PVDF ultrafiltration membrane, J. Mater. Chem. A 3 (2015) 20277–20287. Y. Hang, G. Liu, K. Huang, W. Jin, Mechanical properties and interfacial adhesion of composite membranes probed by in-situ nano-indentation/scratch technique, J. Membr. Sci. 494 (2015) 205–215. W. Wei, S. Xia, G. Liu, X. Gu, W. Jin, N. Xu, Interfacial adhesion between polymer separation layer and ceramic support for composite membrane, AIChE J. 56 (2010) 1584–1592. D. Zou, M. Qiu, X. Chen, Y. Fan, One-step preparation of high-performance bilayer alpha-alumina ultrafiltration membranes via co-sintering process, J. Membr. Sci. 524 (2017) 141–150. M. Qiu, S. Fan, Y. Cai, Y. Fan, N. Xu, Co-sintering synthesis of bi-layer titania ultrafiltration membranes with intermediate layer of sol-coated nanofibers, J. Membr. Sci. 365 (2010) 225–231. Z. Wang, J. Ma, C.Y. Tang, K. Kimura, Q. Wang, X. Han, Membrane cleaning in membrane bioreactors: a review, J. Membr. Sci. 468 (2014) 276–307. M.O. Lamminen, H.W. Walker, L.K. Weavers, Mechanisms and factors influencing the ultrasonic cleaning of particle-fouled ceramic membranes, J. Membr. Sci. 237 (2004) 213–223. M.J. Lujan-Facundo, J.A. Mendoza-Roca, B. Cuartas-Uribe, S. Alvarez-Blanco, Cleaning efficiency enhancement by ultrasounds for membranes used in dairy industries, Ultrason. Sonochem. 33 (2016) 18–25. D. Chen, L.K. Weavers, H.W. Walker, Ultrasonic control of ceramic membrane fouling by particles: effect of ultrasonic factors, Ultrason. Sonochem. 13 (2006) 379–387.