Development of Highly Efficient Oil-Water Separation Carbon Nanotube Membranes with Stimuli-Switchable Fluxes.

In this work, a carbon nanotube (CNT)-based membrane [(4-((4-((11-ferroceneundecyl)oxy)phenyl)diazenyl)phenoxy)-diethylene triamine (FADETA)/polyethyleneimine (PEI)-decorated CNT membrane] with stimuli-switchable separation fluxes was developed. The multiwalled CNTs were modified by a pH-, light-, and redox stimuli-responsive surfactant FADETA initially, and then the FADETA-decorated CNTs were further cross-linked by PEI and finally coated on the polypropylene membrane. Interestingly, the particular membrane was successfully applied in emulsion systems to separate oil and water with high efficiency. First, the FADETA-/PEI-decorated CNT membrane showed highly porous microstructural characteristics owing to the overlapped and cross-linked CNTs as confirmed by the scanning electron microscopy observation. Then, it showed strong hydrophilicity to water in the air and high oleophobicity to oil underwater, thereby endowing the membrane with the potential to separate oil and water. Owing to the modified multiple stimuli-responsive FADETA on CNTs, the separation fluxes were stimuli-switchable, which could be adjusted reversibly by environmental factors including pH, light, and redox.

Introduction

Nowadays, the environmental pollution of air, water, and soil caused by daily life, agriculture, and industry has become one of the most serious threats to human beings globally. The purification of rising sewage is a big challenge, and the classic physical adsorption method is dominantly applied currently.[1−4] Development of new renewable techniques with low cost, high efficiency, and convenient operation is attractive, and the flourishing functional nanomaterials show tremendous potential and contribute lots in this field as summarized by Li and co-workers.[5] Beyond the direct degradation of chemicals by the biological and photocatalytic techniques,[6−9] membrane separation is an alternative and promising method to desalinate and reuse wastewater especially in the large-scale treatment.[10−12] Emulsified water is a typical sewage that can be renewed rapidly and efficiently by the oil–water separation membrane technique,[13] and numerous porous nanomaterials including both inorganic and organic chemicals were employed as the objective substrates successfully.[14−20] Generally speaking, membranes with the particular surface nature such as the superhydrophilic–superoleophobic or superoleophilic–superhydrophobic characteristics are essential in the efficient oil–water separation treatments.

Carbon nanotubes (CNTs), one isotope of amazing carbon, have prominent properties such as good chemical stability, high mechanical strength, excellent electric and thermal conductivity, and so forth,[21−24] thereby attracting much attention since the day of discovery. Nowadays, CNTs and their nanocomposites show a remarkable potential in wide fields including oil–water separation.[25−28] Shi et al.[29] developed a type of ultrathin single-walled CNT (SWCNT) network films, which were applied to separate oil and water efficiently from the emulsified oil and water mixtures because of their selective superwetting effect to oil drops. Similarly, Jin et al.[30] constructed a super-oil-repellent membrane based on the tannic acid-decorated multiple-walled CNT (MWCNT) with high affinity to water, endowing it with a promising oil–water separation capability. Alternatively, Hu and co-workers developed a nanoporous membrane based on the polymer cohybrid SWCNTs,[31] which was highly efficient in the oil–water separation field. More importantly, the membrane showed interesting light-response that the separation fluxes could be reversibly controlled by light irradiation. Beyond the tremendous potential in the oil–water separation of nanocomposites of CNTs, attractive separation controllability was also realized through simple modification on them, which was critically important in extending the application of CNT membranes. It is well-known that surfactants can change the surface wettability of materials significantly.[32,33] Therefore, the surface nature of the CNT membrane would certainly be improved through the modification of the surfactants on CNTs. Furthermore, novel surfactants containing some particular stimuli-responsive moieties might endow the membrane with additional controllability and synergy in oil–water separation.

According to the primary principle mentioned above, a multiple stimuli-responsive surfactant-decorated CNT membrane was developed according to Scheme . Initially, the pH-, light- and redox-responsive surfactant, (4-((4-((11-ferroceneundecyl)oxy)phenyl)diazenyl)phenoxy)-diethylene triamine (FADETA), was covalently combined with CNTs. Then, polyethyleneimine (PEI) was employed as the cross-linking reagent to favor the formation of a stable and hierarchical CNT network. Finally, the commercial polypropylene (PPE) membrane was used as the substrate to support the FADETA-/PEI-decorated CNT membrane by vacuum filtration. The microstructural and physicochemical properties of the FADETA-/PEI-decorated CNT membrane were studied in detail. By employing the sodium dodecylsulfate (SDS)-stabilized oil-in-water (O/W) emulsions as the model objects, the oil–water separation efficiency of the developed CNT membrane was studied quantitatively. Moreover, the effect of stimuli-responsive factors including pH, light, and redox on the separation efficiency was also studied. The present work not only reported a new oil–water separation membrane but also provided a new way to control the separation process through introducing multiple stimuli-responsive surfactants, which was of great importance from both fundamental and application aspects.

Scheme 1

Representative Synthetic Process of FADETA-/PEI-Decorated CNTs

Results and Discussion

Characterization of the FADETA-/PEI-Decorated CNT

In order to clarify the successful modification of FADETA and PEI on CNTs, thermogravimetric analysis (TGA) and Fourier transform infrared (FTIR) measurements were employed. TGA results (Figure a) showed the weight loss of pure CNTs, FADETA-decorated CNTs, and FADETA-/PEI-decorated CNTs. For pure CNTs, the slight decrease of mass below 100 °C was attributed to the removal of adsorbed water, whereas the apparent mass loss about 9.1% above 115 °C was caused by the loss of functional groups including carboxyl and hydroxyl groups.[34] The TGA curve of FADETA-decorated CNTs showed two typical mass loss stages: the first one between about 120 and 250 °C was caused by the residual carboxyl and hydroxyl groups in the pure CNTs and the second one between about 300 and 500 °C was caused by the modified FADETA on CNTs. The total weight loss of FADETA-decorated CNTs was 19.6% that was far larger than that of pure CNTs, indicating the successful modification of FADETA on CNTs. Alternatively, the FADETA-/PEI-decorated CNTs only showed one mass loss stage from about 200 to 500 °C with a weight loss of 30.7%, suggesting that PEI was also covalently cross-linked with CNTs successfully. TGA results showed that the mass ratios of the modified FADETA and PEI in FADETA-/PEI-decorated CNTs were 14.4 and 18.7% (Supporting Information, eq S1–S3), respectively.

Figure 1

TGA curves (a) and FTIR spectra (b) of pure CNTs, FADETA-decorated CNTs, and FADETA-/PEI-decorated CNTs.

Figure b shows the corresponding FTIR spectra of pure CNTs, FADETA-decorated CNTs, and FADETA-/PEI-decorated CNTs, and obvious difference could be observed. For pure CNTs, typical stretching vibration peaks of O–H, C=O, and C–O were observed clearly.[35] In addition, the stretching vibration peak of residual C–H in CNTs was also presented at around 2750–3000 cm–1. However, a new peak at 840 cm–1 in the FADETA-decorated CNTs was observed, which represented the out-of-plane bending mode of C–H in the aromatic moieties, suggesting the presence of FADETA on CNTs well.[36] Once FADETA-decorated CNTs were further modified by PEI, the stretching vibration band at 1403 cm–1 nearly disappeared, whereas a distinguishing C–N stretching vibration peak at 1467 cm–1 was observed instead. Moreover, there existed a strong peak at about 3500 cm–1, corresponding to the abundant stretching vibration of N–H bond in the FADETA and PEI molecules. Thus, the FTIR spectra results also supported the successful modification of FADETA and PEI on CNTs.

Figure a shows the photograph of the FADETA-/PEI-decorated CNT membrane, which showed a macrouniform and black surface clearly. In order to make a better understanding on the microstructural characteristics of the CNT membrane, scanning electron microscopy (SEM) observation was employed. Figure b shows the corresponding surface characteristic of the CNT membrane that contained large porous nanochannels with the size of about tens of nanometers, which were generally formed by the highly overlapped and cross-linked CNTs, endowing the membrane with the potential to separate fluids. The cross-sectional image of the CNT membrane (Supporting Information, Figure S3) suggested that the porous PPE membrane was completely covered by a large amount of agglomerated CNTs. Figure c shows the corresponding magnified cross-sectional appearance of the CNT membrane, and two typical characteristics were observed. The surface of the supported PPE membrane was relatively smooth, whereas that of the CNT membrane was highly rough and porous with a thickness of about several micrometers. Certainly, the SEM results evidently confirmed the successful development of the FADETA-/PEI-decorated CNT membrane.

Figure 2

(a) Photograph of the FADETA-/PEI-decorated CNT membrane and the corresponding SEM images of the surface (b) and cross section (c), photographs of water (d) and CH2Cl2 (e) drops on the CNT membrane in the air and underwater, respectively, and effect of pushing (f) and lifting (g) on the appearance of an oil drop on the CNT membrane underwater.

Because the selective affinity to either water or oil is an essential requirement for the oil–water separation membrane, the surface nature of the CNT membrane was also studied by the contact angle measurements as shown in Figure d–g. It was noticed that the contact angle of a drop of water on the CNT membrane was only 17° in the air (Figure d) and the drop would completely spread later, whereas that of a drop of CH2Cl2 was high up to 150° underwater (Figure e). Moreover, the shape of CH2Cl2 drops changed a little through pushing or lifting (Figure f,g), suggesting a high superoleophobicity. Undoubtedly, the results evidently confirmed the superhydrophilic–superoleophobic nature of the FADETA-/PEI-decorated CNT membrane, endowing it with the attracting potential in the oil–water separation.

Application of the FADETA-/PEI-Decorated CNT Membrane in Oil–Water Separation

According to the particular surface nature of the FADETA-/PEI-decorated CNT membrane, SDS-stabilized O/W type emulsions (Supporting Information S4) were employed as the typical templates to evaluate its oil–water separation efficiency by the equipment represented by Figure a. Figure b,c shows the typical macroappearance of dodecane/water emulsions and the filtrate through flowing past the FADETA-/PEI-decorated CNT membrane. Clearly, the original emulsion containing Sudan red in the oil phase was opalescent and pink, whereas the filtrate became transparent and colorless through the CNT membrane, suggesting that the dodecane phase was completely separated from the emulsion. The corresponding laser scanning confocal microscopy (LSCM) images also supported the result that emulsion droplets containing some oil-soluble fluorochrome Nile red disappeared completely in the filtrate (Figure b,c). In order to clarify the importance of the FADETA-/PEI-decorated CNT membrane in the oil–water separation rather than the contained PPE layer, the oil–water separation efficiency of the pure PPE membrane was also studied. It was noticed that the PPE membrane showed little selective affinity to both water and oil, and nearly no apparent difference could be observed in comparison with the original emulsion and the filtrate (Supporting Information, S5). Undoubtedly, these results evidenced that the excellent oil–water separation efficiency was attributed to the FADETA-/PEI-decorated CNT membrane certainly. It was also noticed that the flux was kept nearly constant after 10 times of continuous separation processes for hexane/water emulsions (Supporting Information S6), suggesting the excellent antifouling performance of the FADETA-/PEI-decorated CNT membrane.

Figure 3

(a) Representative illustration of an oil–water separation device, (b) LSCM image of SDS-stabilized hexane/water emulsions (b) and filtrate (c). The inset images in (b,c) represent the corresponding appearance of emulsion and filtrate and the LSCM image of the filtrate obtained under bright field. Bars: 10 μm.

Figure a shows the quantitative separation fluxes of the CNT membrane from three different O/W emulsions through changing the oil phase from hexane, dodecane, to toluene. No significant difference was observed from the averaged fluxes for hexane/water and dodecane/water emulsions, which were 816 and 790 L·m–2·h–1·MPa–1, respectively. However, the flux for toluene/water emulsion was limited to 695 L·m–2·h–1·MPa–1, which was much lower than those emulsions using alkanes as the oil phase. A possible reason might come from the stronger affinity between toluene and the FADETA-/PEI-decorated CNT membrane because there existed additional interactions such as cationic-π and π–π interactions between toluene and FADETA molecules. Certainly, the present results confirmed the general efficiency of the FADETA-/PEI-decorated CNT membrane in the oil–water separation.

Figure 4

Separation fluxes of the FADETA-/PEI-decorated CNT membrane for different emulsions.

In addition, FADETA is a multiple-stimuli-responsive surfactant, which responses to light, redox, and pH through the following mechanism (Figure a). Generally speaking, UV-light irradiation induces the trans–cis isomerization of an azobenzene moiety, acidification by HCl favors the protonation of amino groups, and addition of FeCl3 oxidizes the ferrocene moiety into its oxidation state, respectively. Such transitions were observed from the corresponding UV–vis spectra of the FADETA aqueous solution (Figure b), and similar responsive properties and transition mechanisms of particular functional moieties by stimuli-responsive factors were reported widely.[37−43] It should be mentioned that the hydrophilicity of FADETA was strengthened either by UV-light irradiation, acidification, or oxidation though the alteration amplitude was slightly different.

Figure 5

pH-, light-, and redox-induced structural transitions of FADETA (a) and effect of acidification and (b) UV light and oxidation on the UV–vis spectra of 0.025 mM FADETA aqueous solution.

Contact angle measurements showed that a little change could be observed from the particular surface wettability of the FADETA-/PEI-decorated CNT membrane when it was disposed by UV-light irradiation, oxidation by FeCl3, and acidification by HCl. The contact angle results of water in the air and CH2Cl2 underwater (Figure a–f) were similar to those of the original sample (Figure d,e), respectively. In other words, its primary surface characteristics of the high hydrophilicity to water in the air and superoleophobicity to oil underwater of the CNT membrane were retained. In order to make a better understanding, the effects of stimuli-responsive factors on the oil–water separation efficiency of the FADETA-/PEI-decorated CNT membrane were also studied by employing the SDS-stabilized hexane/water emulsions as the model. Generally speaking, all membranes showed excellent oil–water separation efficiency, and the milky-like emulsions were transformed into transparent phase through fluxing the corresponding membranes. However, the measured separation fluxes were changed significantly when the membranes were dealt by UV-light irradiation, oxidation by FeCl3, and acidification by HCl, which were 832, 679, and 641 L·m–2·h–1·MPa–1, respectively. The current results supported the controllability of stimuli-responsive factors certainly. Moreover, the fluxes of the FADETA-/PEI-decorated CNT membrane for hexane/water emulsions returned to its original state after the UV–vis light irradiation, oxidation–reduction, and acid–base treatment cycles (Supporting Information S7), suggesting that the separation flux could be controlled reversibly.

Figure 6

Appearance of a drop of water (left) and CH2Cl2 (right) on the FADETA-/PEI-decorated CNT membrane in the air and underwater, respectively. Images (a–f) represent the corresponding appearance after UV-light irradiation (a,b), oxidation (c,d), and acidification (e,f). (g) Stimuli-switchable separation fluxes of the CNT membrane.

In comparison with the original FADETA-/PEI-decorated CNT membrane, oxidation by FeCl3 and acidification by HCl affected the separation fluxes significantly that the fluxes were decreased about 20%, whereas UV-light irradiation affected a little. The tendency might be related to the distinguishing transition mechanisms of different stimuli-responsive factors though all factors resulted in the increase of the hydrophilicity as mentioned above. Trans–cis isomerization of the azobenzene moiety induced by UV-light irradiation changed the dipole moment a little from about 0 to 3 debye.[44,45] As a result, the hydrophilicity of FADETA was increased slightly,[46,47] which limitedly affected the microstructure and surface nature of the FADETA-/PEI-decorated CNT membrane. In contrast, the protonation of amino groups of both FADETA and PEI molecules by acidification strongly increased the hydrophilicity and oleophobicity of the CNT membrane in the air and underwater, respectively, which benefited to enhance the oil–water separation efficiency. However, the electrostatic interactions between decorated molecules were highly strengthened, which were dominantly disadvantaged to the separation efficiency.[35,48,49] Similarly, the oxidized FADETA is a bola-type surfactant after oxidation that also strengthens the electrostatic interactions of the CNT membrane, resulting in the decrease of separation flux. Simultaneously, the oxidized FADETA molecules prefer to adopt the extended configuration because of the electrostatic interactions between two headgroups within the oxidized FADETA. Correspondingly, the steric hindrance of the pore in the CNT membrane should be enhanced, which is also disadvantaged to the separation.

Conclusions

In summary, a new type of the FADETA-/PEI-decorated CNT membrane was developed, which was composed of overlapped and cross-linked CNTs with highly porous microstructural characteristics. Furthermore, the membrane showed apparent hydrophilicity to water in the air whereas superoleophobicity to oil underwater, thereby endowing it with interesting potential in the oil–water separation field. It was noticed that no significant difference was observed from the separation efficiency regardless of the oil type, whereas the oil type of emulsions affected the separation flux strongly. For example, the separation fluxes of the SDS-stabilized hexane/water and dodecane/water emulsions were 816 and 790 L·m–2·h–1·MPa–1, respectively, and that of toluene/water emulsion was limited to 695 L·m–2·h–1·MPa–1. Although the particular surface wettability of the FADETA-/PEI-decorated CNT membrane was mainly attributed to the hydrophilic amino groups of the cross-linking reagent PEI, similar works were reported previously.[35] However, the introduced multiple responsive surfactant FADETA in this work endowed the surface nature of the CNT membrane with additional controllability through stimulating environmental factors such as pH, light, and redox, thereby controlling the separation flux reversibly. The oil–water separation results showed that the fluxes were changed to 832, 679, and 641 L·m–2·h–1·MPa–1 for hexane/water emulsions once the membrane was irradiated by UV-light, redoxed by FeCl3, and acidified by HCl, respectively. Undoubtedly, the current results evidently confirmed the application of the FADETA-/PEI-decorated CNT membrane in the controlled oil–water separation well. Owing to its particular surface wettability of the FADETA-/PEI-decorated CNT membrane, it might also shed some potential in antifouling, the generation of electronic devices and textiles, and so forth.

Experimental Section

Materials

Chemicals including carboxylic MWCNTs (outer diameter: 8–15 nm, length: 50 μm, 95%), 4-hydroxyaminobenzene (98%), phenol (99%), diethylenetriamine (99%), SDS (99%), and 1,2-dibromoethane (99%) were obtained from J&K Chemicals. 11-Bromoundecanoic acid (95%) and PEI with a molecular weight of 10 000 were purchased from Aladdin Industrial Corporation. PPE membrane (mean pore size 0.22 mm) was obtained from Xiya Reagent. Ferrocene (98%) and all other analytical grade reagents including methanol, dichloromethane, hydrochloric acid, sodium hydroxide, and FeCl3 were obtained from Sinopharm Chemical Reagent Co, Ltd. All chemicals were used as received without further purification. The surfactant FADETA was synthesized according to the procedure as mentioned in Scheme S1 in the Supporting Information. Water was Millipore Milli-Q-grade.

Preparation of the FADETA-/PEI-Decorated CNT Membrane

The FADETA-/PEI-decorated CNT membrane was prepared according to Scheme , and the typical synthetic detail was described as follows:

The mixture of carboxylic CNTs (100 mg) and thionyl chloride (50 mL) was added into a round-bottom flask and refluxed for 3 h at 70 °C under stirring. Excess thionyl chloride was removed under reduced pressure to give the acyl chloride CNTs, which were further washed with dichloromethane until all thionyl chloride was removed. The acyl chloride CNTs were dispersed in 200 mL dichloromethane and cooled to 0 °C. Then, 40 mL dichloromethane solution containing FADETA (50 mg, 0.07 mmol) and a catalytic amount of trimethylamine was added slowly. The mixture was warmed to room temperature and stirred for additional 5 h and then filtered through a poly(tetrafluoroethylene) membrane and washed with dichloromethane to give the FADETA-decorated CNTs, which were further dried in a vacuum oven for 24 h at 50 °C. Next, 50 mg FADETA-decorated CNTs were added into PEI aqueous solution (2 g in 50 mL water), which was sonicated for additional 30 min. The mixture was centrifuged at 5000 rpm for 10 min to give the FADETA-/PEI-decorated CNTs. Finally, the commercial PPE membrane was used as the substrate to support the FADETA-/PEI-decorated CNT membrane by vacuum filtration.

Preparation of O/W Emulsions

The surfactant-stabilized O/W emulsions including hexane/water, toluene/water, and dodecane/water emulsions were prepared by the same procedure. Briefly, 2 mL oil and 0.2 g sodium dodecyl sulfate were added into 100 mL deionized water, and the mixture was sonicated for 30 min.

Characterization

FTIR were recorded on a Nicolet iS10 spectrophotometer (Thermo, USA). The samples were prepared as KBr pellets, and the spectra were calculated from a total of 16 scans. SEM images were taken on a field-emission microscope (Zeiss Sigma) operated at an accelerating voltage of 5 kV. Recorded SEM images allowed studying the morphology of the underwater superoleophobic coating. TGA was conducted on TG 209 F1 (NETZSCH, Germany) with a temperature increase of 10 °C·min–1 at the nitrogen atmosphere. LSCM observation of dodecane/water emulsion and the filtrate was performed on a dual-disk fast live-cell confocal imaging system (PerkinElmer UltraVIEW VoX). During the process, a small amount of Nile red was solubilized in the oil phase, and the excitation wavelength of 561 nm was used. Contact angle measurements were conducted by using a DSA-100 instrument (Kruss).

O/W Separation

O/W emulsion separation experiments were performed using a homemade filtration device. The flux (JF) was determined by calculating the volume of the filtrate within 1 min using the following equationwhere V is the volume of emulsion, A is the valid area of the membrane (12.56 cm2), and Δt and Δp represent the testing time and the membrane pressure, respectively.